308 files changed, 4600 insertions, 3798 deletions

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index dd0bfc291a68..fe11fccf7e41 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -856,6 +856,49 @@ running once the system is up.
 
         mga=            [HW,DRM]
 
+        migration_cost=
+                        [KNL,SMP] debug: override scheduler migration costs
+                        Format: <level-1-usecs>,<level-2-usecs>,...
+                        This debugging option can be used to override the
+                        default scheduler migration cost matrix. The numbers
+                        are indexed by 'CPU domain distance'.
+                        E.g. migration_cost=1000,2000,3000 on an SMT NUMA
+                        box will set up an intra-core migration cost of
+                        1 msec, an inter-core migration cost of 2 msecs,
+                        and an inter-node migration cost of 3 msecs.
+
+                        WARNING: using the wrong values here can break
+                        scheduler performance, so it's only for scheduler
+                        development purposes, not production environments.
+
+        migration_debug=
+                        [KNL,SMP] migration cost auto-detect verbosity
+                        Format=<0|1|2>
+                        If a system's migration matrix reported at bootup
+                        seems erroneous then this option can be used to
+                        increase verbosity of the detection process.
+                        We default to 0 (no extra messages), 1 will print
+                        some more information, and 2 will be really
+                        verbose (probably only useful if you also have a
+                        serial console attached to the system).
+
+        migration_factor=
+                        [KNL,SMP] multiply/divide migration costs by a factor
+                        Format=<percent>
+                        This debug option can be used to proportionally
+                        increase or decrease the auto-detected migration
+                        costs for all entries of the migration matrix.
+                        E.g. migration_factor=150 will increase migration
+                        costs by 50%. (and thus the scheduler will be less
+                        eager migrating cache-hot tasks)
+                        migration_factor=80 will decrease migration costs
+                        by 20%. (thus the scheduler will be more eager to
+                        migrate tasks)
+
+                        WARNING: using the wrong values here can break
+                        scheduler performance, so it's only for scheduler
+                        development purposes, not production environments.
+
         mousedev.tap_time=
                         [MOUSE] Maximum time between finger touching and
                         leaving touchpad surface for touch to be considered
diff --git a/arch/alpha/kernel/process.c b/arch/alpha/kernel/process.c
index abb739b88ed1..9924fd07743a 100644
--- a/arch/alpha/kernel/process.c
+++ b/arch/alpha/kernel/process.c
@@ -276,7 +276,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 {
         extern void ret_from_fork(void);
 
-        struct thread_info *childti = p->thread_info;
+        struct thread_info *childti = task_thread_info(p);
         struct pt_regs * childregs;
         struct switch_stack * childstack, *stack;
         unsigned long stack_offset, settls;
@@ -285,7 +285,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
         if (!(regs->ps & 8))
                 stack_offset = (PAGE_SIZE-1) & (unsigned long) regs;
         childregs = (struct pt_regs *)
-          (stack_offset + PAGE_SIZE + (long) childti);
+          (stack_offset + PAGE_SIZE + task_stack_page(p));
                 
         *childregs = *regs;
         settls = regs->r20;
@@ -428,30 +428,15 @@ dump_elf_thread(elf_greg_t *dest, struct pt_regs *pt, struct thread_info *ti)
 int
 dump_elf_task(elf_greg_t *dest, struct task_struct *task)
 {
-        struct thread_info *ti;
-        struct pt_regs *pt;
-
-        ti = task->thread_info;
-        pt = (struct pt_regs *)((unsigned long)ti + 2*PAGE_SIZE) - 1;
-
-        dump_elf_thread(dest, pt, ti);
-
+        dump_elf_thread(dest, task_pt_regs(task), task_thread_info(task));
         return 1;
 }
 
 int
 dump_elf_task_fp(elf_fpreg_t *dest, struct task_struct *task)
 {
-        struct thread_info *ti;
-        struct pt_regs *pt;
-        struct switch_stack *sw;
-
-        ti = task->thread_info;
-        pt = (struct pt_regs *)((unsigned long)ti + 2*PAGE_SIZE) - 1;
-        sw = (struct switch_stack *)pt - 1;
-
+        struct switch_stack *sw = (struct switch_stack *)task_pt_regs(task) - 1;
         memcpy(dest, sw->fp, 32 * 8);
-
         return 1;
 }
 
@@ -492,8 +477,8 @@ out:
 unsigned long
 thread_saved_pc(task_t *t)
 {
-        unsigned long base = (unsigned long)t->thread_info;
-        unsigned long fp, sp = t->thread_info->pcb.ksp;
+        unsigned long base = (unsigned long)task_stack_page(t);
+        unsigned long fp, sp = task_thread_info(t)->pcb.ksp;
 
         if (sp > base && sp+6*8 < base + 16*1024) {
                 fp = ((unsigned long*)sp)[6];
@@ -523,7 +508,7 @@ get_wchan(struct task_struct *p)
 
         pc = thread_saved_pc(p);
         if (in_sched_functions(pc)) {
-                schedule_frame = ((unsigned long *)p->thread_info->pcb.ksp)[6];
+                schedule_frame = ((unsigned long *)task_thread_info(p)->pcb.ksp)[6];
                 return ((unsigned long *)schedule_frame)[12];
         }
         return pc;
diff --git a/arch/alpha/kernel/ptrace.c b/arch/alpha/kernel/ptrace.c
index 9969d212e94d..0cd060598f9a 100644
--- a/arch/alpha/kernel/ptrace.c
+++ b/arch/alpha/kernel/ptrace.c
@@ -72,6 +72,13 @@ enum {
         REG_R0 = 0, REG_F0 = 32, REG_FPCR = 63, REG_PC = 64
 };
 
+#define PT_REG(reg) \
+  (PAGE_SIZE*2 - sizeof(struct pt_regs) + offsetof(struct pt_regs, reg))
+
+#define SW_REG(reg) \
+ (PAGE_SIZE*2 - sizeof(struct pt_regs) - sizeof(struct switch_stack) \
+  + offsetof(struct switch_stack, reg))
+
 static int regoff[] = {
         PT_REG(    r0), PT_REG(    r1), PT_REG(    r2), PT_REG(   r3),
         PT_REG(    r4), PT_REG(    r5), PT_REG(    r6), PT_REG(   r7),
@@ -103,14 +110,14 @@ get_reg_addr(struct task_struct * task, unsigned long regno)
         unsigned long *addr;
 
         if (regno == 30) {
-                addr = &task->thread_info->pcb.usp;
+                addr = &task_thread_info(task)->pcb.usp;
         } else if (regno == 65) {
-                addr = &task->thread_info->pcb.unique;
+                addr = &task_thread_info(task)->pcb.unique;
         } else if (regno == 31 || regno > 65) {
                 zero = 0;
                 addr = &zero;
         } else {
-                addr = (void *)task->thread_info + regoff[regno];
+                addr = task_stack_page(task) + regoff[regno];
         }
         return addr;
 }
@@ -125,7 +132,7 @@ get_reg(struct task_struct * task, unsigned long regno)
         if (regno == 63) {
                 unsigned long fpcr = *get_reg_addr(task, regno);
                 unsigned long swcr
-                  = task->thread_info->ieee_state & IEEE_SW_MASK;
+                  = task_thread_info(task)->ieee_state & IEEE_SW_MASK;
                 swcr = swcr_update_status(swcr, fpcr);
                 return fpcr | swcr;
         }
@@ -139,8 +146,8 @@ static int
 put_reg(struct task_struct *task, unsigned long regno, unsigned long data)
 {
         if (regno == 63) {
-                task->thread_info->ieee_state
-                  = ((task->thread_info->ieee_state & ~IEEE_SW_MASK)
+                task_thread_info(task)->ieee_state
+                  = ((task_thread_info(task)->ieee_state & ~IEEE_SW_MASK)
                      | (data & IEEE_SW_MASK));
                 data = (data & FPCR_DYN_MASK) | ieee_swcr_to_fpcr(data);
         }
@@ -188,35 +195,35 @@ ptrace_set_bpt(struct task_struct * child)
                  * branch (emulation can be tricky for fp branches).
                  */
                 displ = ((s32)(insn << 11)) >> 9;
-                child->thread_info->bpt_addr[nsaved++] = pc + 4;
+                task_thread_info(child)->bpt_addr[nsaved++] = pc + 4;
                 if (displ)              /* guard against unoptimized code */
-                        child->thread_info->bpt_addr[nsaved++]
+                        task_thread_info(child)->bpt_addr[nsaved++]
                           = pc + 4 + displ;
                 DBG(DBG_BPT, ("execing branch\n"));
         } else if (op_code == 0x1a) {
                 reg_b = (insn >> 16) & 0x1f;
-                child->thread_info->bpt_addr[nsaved++] = get_reg(child, reg_b);
+                task_thread_info(child)->bpt_addr[nsaved++] = get_reg(child, reg_b);
                 DBG(DBG_BPT, ("execing jump\n"));
         } else {
-                child->thread_info->bpt_addr[nsaved++] = pc + 4;
+                task_thread_info(child)->bpt_addr[nsaved++] = pc + 4;
                 DBG(DBG_BPT, ("execing normal insn\n"));
         }
 
         /* install breakpoints: */
         for (i = 0; i < nsaved; ++i) {
-                res = read_int(child, child->thread_info->bpt_addr[i],
+                res = read_int(child, task_thread_info(child)->bpt_addr[i],
                                (int *) &insn);
                 if (res < 0)
                         return res;
-                child->thread_info->bpt_insn[i] = insn;
+                task_thread_info(child)->bpt_insn[i] = insn;
                 DBG(DBG_BPT, ("    -> next_pc=%lx\n",
-                              child->thread_info->bpt_addr[i]));
-                res = write_int(child, child->thread_info->bpt_addr[i],
+                              task_thread_info(child)->bpt_addr[i]));
+                res = write_int(child, task_thread_info(child)->bpt_addr[i],
                                 BREAKINST);
                 if (res < 0)
                         return res;
         }
-        child->thread_info->bpt_nsaved = nsaved;
+        task_thread_info(child)->bpt_nsaved = nsaved;
         return 0;
 }
 
@@ -227,9 +234,9 @@ ptrace_set_bpt(struct task_struct * child)
 int
 ptrace_cancel_bpt(struct task_struct * child)
 {
-        int i, nsaved = child->thread_info->bpt_nsaved;
+        int i, nsaved = task_thread_info(child)->bpt_nsaved;
 
-        child->thread_info->bpt_nsaved = 0;
+        task_thread_info(child)->bpt_nsaved = 0;
 
         if (nsaved > 2) {
                 printk("ptrace_cancel_bpt: bogus nsaved: %d!\n", nsaved);
@@ -237,8 +244,8 @@ ptrace_cancel_bpt(struct task_struct * child)
         }
 
         for (i = 0; i < nsaved; ++i) {
-                write_int(child, child->thread_info->bpt_addr[i],
-                          child->thread_info->bpt_insn[i]);
+                write_int(child, task_thread_info(child)->bpt_addr[i],
+                          task_thread_info(child)->bpt_insn[i]);
         }
         return (nsaved != 0);
 }
@@ -355,7 +362,7 @@ do_sys_ptrace(long request, long pid, long addr, long data,
                 if (!valid_signal(data))
                         break;
                 /* Mark single stepping.  */
-                child->thread_info->bpt_nsaved = -1;
+                task_thread_info(child)->bpt_nsaved = -1;
                 clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
                 child->exit_code = data;
                 wake_up_process(child);
diff --git a/arch/alpha/kernel/smp.c b/arch/alpha/kernel/smp.c
index da0be3465791..4b873527ce1c 100644
--- a/arch/alpha/kernel/smp.c
+++ b/arch/alpha/kernel/smp.c
@@ -302,7 +302,7 @@ secondary_cpu_start(int cpuid, struct task_struct *idle)
                  + hwrpb->processor_offset
                  + cpuid * hwrpb->processor_size);
         hwpcb = (struct pcb_struct *) cpu->hwpcb;
-        ipcb = &idle->thread_info->pcb;
+        ipcb = &task_thread_info(idle)->pcb;
 
         /* Initialize the CPU's HWPCB to something just good enough for
            us to get started.  Immediately after starting, we'll swpctx
diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 54a21bdcba5c..4b4e4cf79c80 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -342,10 +342,10 @@ void flush_thread(void)
 void release_thread(struct task_struct *dead_task)
 {
 #if defined(CONFIG_VFP)
-        vfp_release_thread(&dead_task->thread_info->vfpstate);
+        vfp_release_thread(&task_thread_info(dead_task)->vfpstate);
 #endif
 #if defined(CONFIG_IWMMXT)
-        iwmmxt_task_release(dead_task->thread_info);
+        iwmmxt_task_release(task_thread_info(dead_task));
 #endif
 }
 
@@ -355,10 +355,9 @@ int
 copy_thread(int nr, unsigned long clone_flags, unsigned long stack_start,
             unsigned long stk_sz, struct task_struct *p, struct pt_regs *regs)
 {
-        struct thread_info *thread = p->thread_info;
-        struct pt_regs *childregs;
+        struct thread_info *thread = task_thread_info(p);
+        struct pt_regs *childregs = task_pt_regs(p);
 
-        childregs = (void *)thread + THREAD_START_SP - sizeof(*regs);
         *childregs = *regs;
         childregs->ARM_r0 = 0;
         childregs->ARM_sp = stack_start;
@@ -460,8 +459,8 @@ unsigned long get_wchan(struct task_struct *p)
         if (!p || p == current || p->state == TASK_RUNNING)
                 return 0;
 
-        stack_start = (unsigned long)(p->thread_info + 1);
-        stack_end = ((unsigned long)p->thread_info) + THREAD_SIZE;
+        stack_start = (unsigned long)end_of_stack(p);
+        stack_end = (unsigned long)task_stack_page(p) + THREAD_SIZE;
 
         fp = thread_saved_fp(p);
         do {
diff --git a/arch/arm/kernel/ptrace.c b/arch/arm/kernel/ptrace.c
index 2b84f78d7b0f..e591f72bcdeb 100644
--- a/arch/arm/kernel/ptrace.c
+++ b/arch/arm/kernel/ptrace.c
@@ -55,23 +55,6 @@
 #endif
 
 /*
- * Get the address of the live pt_regs for the specified task.
- * These are saved onto the top kernel stack when the process
- * is not running.
- *
- * Note: if a user thread is execve'd from kernel space, the
- * kernel stack will not be empty on entry to the kernel, so
- * ptracing these tasks will fail.
- */
-static inline struct pt_regs *
-get_user_regs(struct task_struct *task)
-{
-        return (struct pt_regs *)
-                ((unsigned long)task->thread_info + THREAD_SIZE -
-                                 8 - sizeof(struct pt_regs));
-}
-
-/*
  * this routine will get a word off of the processes privileged stack.
  * the offset is how far from the base addr as stored in the THREAD.
  * this routine assumes that all the privileged stacks are in our
@@ -79,7 +62,7 @@ get_user_regs(struct task_struct *task)
  */
 static inline long get_user_reg(struct task_struct *task, int offset)
 {
-        return get_user_regs(task)->uregs[offset];
+        return task_pt_regs(task)->uregs[offset];
 }
 
 /*
@@ -91,7 +74,7 @@ static inline long get_user_reg(struct task_struct *task, int offset)
 static inline int
 put_user_reg(struct task_struct *task, int offset, long data)
 {
-        struct pt_regs newregs, *regs = get_user_regs(task);
+        struct pt_regs newregs, *regs = task_pt_regs(task);
         int ret = -EINVAL;
 
         newregs = *regs;
@@ -421,7 +404,7 @@ void ptrace_set_bpt(struct task_struct *child)
         u32 insn;
         int res;
 
-        regs = get_user_regs(child);
+        regs = task_pt_regs(child);
         pc = instruction_pointer(regs);
 
         if (thumb_mode(regs)) {
@@ -572,7 +555,7 @@ static int ptrace_write_user(struct task_struct *tsk, unsigned long off,
  */
 static int ptrace_getregs(struct task_struct *tsk, void __user *uregs)
 {
-        struct pt_regs *regs = get_user_regs(tsk);
+        struct pt_regs *regs = task_pt_regs(tsk);
 
         return copy_to_user(uregs, regs, sizeof(struct pt_regs)) ? -EFAULT : 0;
 }
@@ -587,7 +570,7 @@ static int ptrace_setregs(struct task_struct *tsk, void __user *uregs)
 
         ret = -EFAULT;
         if (copy_from_user(&newregs, uregs, sizeof(struct pt_regs)) == 0) {
-                struct pt_regs *regs = get_user_regs(tsk);
+                struct pt_regs *regs = task_pt_regs(tsk);
 
                 ret = -EINVAL;
                 if (valid_user_regs(&newregs)) {
@@ -604,7 +587,7 @@ static int ptrace_setregs(struct task_struct *tsk, void __user *uregs)
  */
 static int ptrace_getfpregs(struct task_struct *tsk, void __user *ufp)
 {
-        return copy_to_user(ufp, &tsk->thread_info->fpstate,
+        return copy_to_user(ufp, &task_thread_info(tsk)->fpstate,
                             sizeof(struct user_fp)) ? -EFAULT : 0;
 }
 
@@ -613,7 +596,7 @@ static int ptrace_getfpregs(struct task_struct *tsk, void __user *ufp)
  */
 static int ptrace_setfpregs(struct task_struct *tsk, void __user *ufp)
 {
-        struct thread_info *thread = tsk->thread_info;
+        struct thread_info *thread = task_thread_info(tsk);
         thread->used_cp[1] = thread->used_cp[2] = 1;
         return copy_from_user(&thread->fpstate, ufp,
                               sizeof(struct user_fp)) ? -EFAULT : 0;
@@ -626,7 +609,7 @@ static int ptrace_setfpregs(struct task_struct *tsk, void __user *ufp)
  */
 static int ptrace_getwmmxregs(struct task_struct *tsk, void __user *ufp)
 {
-        struct thread_info *thread = tsk->thread_info;
+        struct thread_info *thread = task_thread_info(tsk);
         void *ptr = &thread->fpstate;
 
         if (!test_ti_thread_flag(thread, TIF_USING_IWMMXT))
@@ -643,7 +626,7 @@ static int ptrace_getwmmxregs(struct task_struct *tsk, void __user *ufp)
  */
 static int ptrace_setwmmxregs(struct task_struct *tsk, void __user *ufp)
 {
-        struct thread_info *thread = tsk->thread_info;
+        struct thread_info *thread = task_thread_info(tsk);
         void *ptr = &thread->fpstate;
 
         if (!test_ti_thread_flag(thread, TIF_USING_IWMMXT))
@@ -779,7 +762,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 #endif
 
                 case PTRACE_GET_THREAD_AREA:
-                        ret = put_user(child->thread_info->tp_value,
+                        ret = put_user(task_thread_info(child)->tp_value,
                                        (unsigned long __user *) data);
                         break;
 
diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
index 373c0959bc2f..7338948bd7d3 100644
--- a/arch/arm/kernel/smp.c
+++ b/arch/arm/kernel/smp.c
@@ -114,7 +114,7 @@ int __cpuinit __cpu_up(unsigned int cpu)
          * We need to tell the secondary core where to find
          * its stack and the page tables.
          */
-        secondary_data.stack = (void *)idle->thread_info + THREAD_START_SP;
+        secondary_data.stack = task_stack_page(idle) + THREAD_START_SP;
         secondary_data.pgdir = virt_to_phys(pgd);
         wmb();
 
@@ -245,7 +245,7 @@ void __cpuexit cpu_die(void)
         __asm__("mov    sp, %0\n"
         "       b       secondary_start_kernel"
                 :
-                : "r" ((void *)current->thread_info + THREAD_SIZE - 8));
+                : "r" (task_stack_page(current) + THREAD_SIZE - 8));
 }
 #endif /* CONFIG_HOTPLUG_CPU */
 
diff --git a/arch/arm/kernel/traps.c b/arch/arm/kernel/traps.c
index c9fe6f5f7ee3..93cfd3ffcc72 100644
--- a/arch/arm/kernel/traps.c
+++ b/arch/arm/kernel/traps.c
@@ -164,7 +164,7 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
         } else if (verify_stack(fp)) {
                 printk("invalid frame pointer 0x%08x", fp);
                 ok = 0;
-        } else if (fp < (unsigned long)(tsk->thread_info + 1))
+        } else if (fp < (unsigned long)end_of_stack(tsk))
                 printk("frame pointer underflow");
         printk("\n");
 
@@ -210,7 +210,7 @@ static void __die(const char *str, int err, struct thread_info *thread, struct p
 
         if (!user_mode(regs) || in_interrupt()) {
                 dump_mem("Stack: ", regs->ARM_sp,
-                         THREAD_SIZE + (unsigned long)tsk->thread_info);
+                         THREAD_SIZE + (unsigned long)task_stack_page(tsk));
                 dump_backtrace(regs, tsk);
                 dump_instr(regs);
         }
diff --git a/arch/arm26/kernel/process.c b/arch/arm26/kernel/process.c
index 15833a0057dd..386305659171 100644
--- a/arch/arm26/kernel/process.c
+++ b/arch/arm26/kernel/process.c
@@ -277,10 +277,9 @@ int
 copy_thread(int nr, unsigned long clone_flags, unsigned long stack_start,
             unsigned long unused, struct task_struct *p, struct pt_regs *regs)
 {
-        struct thread_info *thread = p->thread_info;
-        struct pt_regs *childregs;
+        struct thread_info *thread = task_thread_info(p);
+        struct pt_regs *childregs = task_pt_regs(p);
 
-        childregs = __get_user_regs(thread);
         *childregs = *regs;
         childregs->ARM_r0 = 0;
         childregs->ARM_sp = stack_start;
diff --git a/arch/arm26/kernel/ptrace.c b/arch/arm26/kernel/ptrace.c
index 4e6b7356a722..3c3371d4683e 100644
--- a/arch/arm26/kernel/ptrace.c
+++ b/arch/arm26/kernel/ptrace.c
@@ -40,21 +40,6 @@
 #define BREAKINST_ARM   0xef9f0001
 
 /*
- * Get the address of the live pt_regs for the specified task.
- * These are saved onto the top kernel stack when the process
- * is not running.
- *
- * Note: if a user thread is execve'd from kernel space, the
- * kernel stack will not be empty on entry to the kernel, so
- * ptracing these tasks will fail.
- */
-static inline struct pt_regs *
-get_user_regs(struct task_struct *task)
-{
-        return __get_user_regs(task->thread_info);
-}
-
-/*
  * this routine will get a word off of the processes privileged stack.
  * the offset is how far from the base addr as stored in the THREAD.
  * this routine assumes that all the privileged stacks are in our
@@ -62,7 +47,7 @@ get_user_regs(struct task_struct *task)
  */
 static inline long get_user_reg(struct task_struct *task, int offset)
 {
-        return get_user_regs(task)->uregs[offset];
+        return task_pt_regs(task)->uregs[offset];
 }
 
 /*
@@ -74,7 +59,7 @@ static inline long get_user_reg(struct task_struct *task, int offset)
 static inline int
 put_user_reg(struct task_struct *task, int offset, long data)
 {
-        struct pt_regs newregs, *regs = get_user_regs(task);
+        struct pt_regs newregs, *regs = task_pt_regs(task);
         int ret = -EINVAL;
 
         newregs = *regs;
@@ -377,7 +362,7 @@ void ptrace_set_bpt(struct task_struct *child)
         u32 insn;
         int res;
 
-        regs = get_user_regs(child);
+        regs = task_pt_regs(child);
         pc = instruction_pointer(regs);
 
         res = read_instr(child, pc, &insn);
@@ -500,7 +485,7 @@ static int ptrace_write_user(struct task_struct *tsk, unsigned long off,
  */
 static int ptrace_getregs(struct task_struct *tsk, void *uregs)
 {
-        struct pt_regs *regs = get_user_regs(tsk);
+        struct pt_regs *regs = task_pt_regs(tsk);
 
         return copy_to_user(uregs, regs, sizeof(struct pt_regs)) ? -EFAULT : 0;
 }
@@ -515,7 +500,7 @@ static int ptrace_setregs(struct task_struct *tsk, void *uregs)
 
         ret = -EFAULT;
         if (copy_from_user(&newregs, uregs, sizeof(struct pt_regs)) == 0) {
-                struct pt_regs *regs = get_user_regs(tsk);
+                struct pt_regs *regs = task_pt_regs(tsk);
 
                 ret = -EINVAL;
                 if (valid_user_regs(&newregs)) {
@@ -532,7 +517,7 @@ static int ptrace_setregs(struct task_struct *tsk, void *uregs)
  */
 static int ptrace_getfpregs(struct task_struct *tsk, void *ufp)
 {
-        return copy_to_user(ufp, &tsk->thread_info->fpstate,
+        return copy_to_user(ufp, &task_thread_info(tsk)->fpstate,
                             sizeof(struct user_fp)) ? -EFAULT : 0;
 }
 
@@ -542,7 +527,7 @@ static int ptrace_getfpregs(struct task_struct *tsk, void *ufp)
 static int ptrace_setfpregs(struct task_struct *tsk, void *ufp)
 {
         set_stopped_child_used_math(tsk);
-        return copy_from_user(&tsk->thread_info->fpstate, ufp,
+        return copy_from_user(&task_threas_info(tsk)->fpstate, ufp,
                               sizeof(struct user_fp)) ? -EFAULT : 0;
 }
 
diff --git a/arch/arm26/kernel/traps.c b/arch/arm26/kernel/traps.c
index f64f59022392..5847ea5d7747 100644
--- a/arch/arm26/kernel/traps.c
+++ b/arch/arm26/kernel/traps.c
@@ -132,7 +132,7 @@ static void dump_instr(struct pt_regs *regs)
 
 /*static*/ void __dump_stack(struct task_struct *tsk, unsigned long sp)
 {
-        dump_mem("Stack: ", sp, 8192+(unsigned long)tsk->thread_info);
+        dump_mem("Stack: ", sp, 8192+(unsigned long)task_stack_page(tsk));
 }
 
 void dump_stack(void)
@@ -158,7 +158,7 @@ void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
         } else if (verify_stack(fp)) {
                 printk("invalid frame pointer 0x%08x", fp);
                 ok = 0;
-        } else if (fp < (unsigned long)(tsk->thread_info + 1))
+        } else if (fp < (unsigned long)end_of_stack(tsk))
                 printk("frame pointer underflow");
         printk("\n");
 
@@ -168,7 +168,7 @@ void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
 
 /* FIXME - this is probably wrong.. */
 void show_stack(struct task_struct *task, unsigned long *sp) {
-        dump_mem("Stack: ", (unsigned long)sp, 8192+(unsigned long)task->thread_info);
+        dump_mem("Stack: ", (unsigned long)sp, 8192+(unsigned long)task_stack_page(task));
 }
 
 DEFINE_SPINLOCK(die_lock);
@@ -187,7 +187,7 @@ NORET_TYPE void die(const char *str, struct pt_regs *regs, int err)
         printk("CPU: %d\n", smp_processor_id());
         show_regs(regs);
         printk("Process %s (pid: %d, stack limit = 0x%p)\n",
-                current->comm, current->pid, tsk->thread_info + 1);
+                current->comm, current->pid, end_of_stack(tsk));
 
         if (!user_mode(regs) || in_interrupt()) {
                 __dump_stack(tsk, (unsigned long)(regs + 1));
diff --git a/arch/cris/arch-v10/kernel/process.c b/arch/cris/arch-v10/kernel/process.c
index 69e28b4057e8..0a675ce9e099 100644
--- a/arch/cris/arch-v10/kernel/process.c
+++ b/arch/cris/arch-v10/kernel/process.c
@@ -79,7 +79,7 @@ void hard_reset_now (void)
  */
 unsigned long thread_saved_pc(struct task_struct *t)
 {
-        return (unsigned long)user_regs(t->thread_info)->irp;
+        return task_pt_regs(t)->irp;
 }
 
 static void kernel_thread_helper(void* dummy, int (*fn)(void *), void * arg)
@@ -128,7 +128,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
          * remember that the task_struct doubles as the kernel stack for the task
          */
 
-        childregs = user_regs(p->thread_info);        
+        childregs = task_pt_regs(p);
         
         *childregs = *regs;  /* struct copy of pt_regs */
         
diff --git a/arch/cris/arch-v10/kernel/ptrace.c b/arch/cris/arch-v10/kernel/ptrace.c
index 6cbd34a27b90..f214f74f264e 100644
--- a/arch/cris/arch-v10/kernel/ptrace.c
+++ b/arch/cris/arch-v10/kernel/ptrace.c
@@ -37,7 +37,7 @@ inline long get_reg(struct task_struct *task, unsigned int regno)
         if (regno == PT_USP)
                 return task->thread.usp;
         else if (regno < PT_MAX)
-                return ((unsigned long *)user_regs(task->thread_info))[regno];
+                return ((unsigned long *)task_pt_regs(task))[regno];
         else
                 return 0;
 }
@@ -51,7 +51,7 @@ inline int put_reg(struct task_struct *task, unsigned int regno,
         if (regno == PT_USP)
                 task->thread.usp = data;
         else if (regno < PT_MAX)
-                ((unsigned long *)user_regs(task->thread_info))[regno] = data;
+                ((unsigned long *)task_pt_regs(task))[regno] = data;
         else
                 return -1;
         return 0;
diff --git a/arch/cris/arch-v32/kernel/process.c b/arch/cris/arch-v32/kernel/process.c
index 882be42114f7..843513102d3c 100644
--- a/arch/cris/arch-v32/kernel/process.c
+++ b/arch/cris/arch-v32/kernel/process.c
@@ -96,7 +96,7 @@ hard_reset_now(void)
  */
 unsigned long thread_saved_pc(struct task_struct *t)
 {
-        return (unsigned long)user_regs(t->thread_info)->erp;
+        return task_pt_regs(t)->erp;
 }
 
 static void
@@ -148,7 +148,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
          * fix it up. Note: the task_struct doubles as the kernel stack for the
          * task.
          */
-        childregs = user_regs(p->thread_info);
+        childregs = task_pt_regs(p);
         *childregs = *regs;     /* Struct copy of pt_regs. */
         p->set_child_tid = p->clear_child_tid = NULL;
         childregs->r10 = 0;     /* Child returns 0 after a fork/clone. */
@@ -157,7 +157,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
          * The TLS is in $mof beacuse it is the 5th argument to sys_clone.
          */
         if (p->mm && (clone_flags & CLONE_SETTLS)) {
-                p->thread_info->tls = regs->mof;
+                task_thread_info(p)->tls = regs->mof;
         }
 
         /* Put the switch stack right below the pt_regs. */
diff --git a/arch/cris/arch-v32/kernel/ptrace.c b/arch/cris/arch-v32/kernel/ptrace.c
index 5528b83a622b..82cf2e3624a4 100644
--- a/arch/cris/arch-v32/kernel/ptrace.c
+++ b/arch/cris/arch-v32/kernel/ptrace.c
@@ -46,7 +46,7 @@ long get_reg(struct task_struct *task, unsigned int regno)
         unsigned long ret;
 
         if (regno <= PT_EDA)
-                ret = ((unsigned long *)user_regs(task->thread_info))[regno];
+                ret = ((unsigned long *)task_pt_regs(task))[regno];
         else if (regno == PT_USP)
                 ret = task->thread.usp;
         else if (regno == PT_PPC)
@@ -65,13 +65,13 @@ long get_reg(struct task_struct *task, unsigned int regno)
 int put_reg(struct task_struct *task, unsigned int regno, unsigned long data)
 {
         if (regno <= PT_EDA)
-                ((unsigned long *)user_regs(task->thread_info))[regno] = data;
+                ((unsigned long *)task_pt_regs(task))[regno] = data;
         else if (regno == PT_USP)
                 task->thread.usp = data;
         else if (regno == PT_PPC) {
                 /* Write pseudo-PC to ERP only if changed. */
                 if (data != get_pseudo_pc(task))
-                        ((unsigned long *)user_regs(task->thread_info))[PT_ERP] = data;
+                        task_pt_regs(task)->erp = data;
         } else if (regno <= PT_MAX)
                 return put_debugreg(task->pid, regno, data);
         else
diff --git a/arch/cris/arch-v32/kernel/smp.c b/arch/cris/arch-v32/kernel/smp.c
index 13867f4fad16..da40d19a151e 100644
--- a/arch/cris/arch-v32/kernel/smp.c
+++ b/arch/cris/arch-v32/kernel/smp.c
@@ -113,10 +113,10 @@ smp_boot_one_cpu(int cpuid)
         if (IS_ERR(idle))
                 panic("SMP: fork failed for CPU:%d", cpuid);
 
-        idle->thread_info->cpu = cpuid;
+        task_thread_info(idle)->cpu = cpuid;
 
         /* Information to the CPU that is about to boot */
-        smp_init_current_idle_thread = idle->thread_info;
+        smp_init_current_idle_thread = task_thread_info(idle);
         cpu_now_booting = cpuid;
 
         /* Wait for CPU to come online */
diff --git a/arch/cris/arch-v32/mm/tlb.c b/arch/cris/arch-v32/mm/tlb.c
index b08a28bb58ab..9d75d7692303 100644
--- a/arch/cris/arch-v32/mm/tlb.c
+++ b/arch/cris/arch-v32/mm/tlb.c
@@ -198,9 +198,9 @@ switch_mm(struct mm_struct *prev, struct mm_struct *next,
         per_cpu(current_pgd, cpu) = next->pgd;
 
         /* Switch context in the MMU. */
-        if (tsk && tsk->thread_info)
+        if (tsk && task_thread_info(tsk))
         {
-          SPEC_REG_WR(SPEC_REG_PID, next->context.page_id | tsk->thread_info->tls);
+          SPEC_REG_WR(SPEC_REG_PID, next->context.page_id | task_thread_info(tsk)->tls);
         }
         else
         {
diff --git a/arch/frv/kernel/process.c b/arch/frv/kernel/process.c
index c4488379ac3b..0fff8a61ef2a 100644
--- a/arch/frv/kernel/process.c
+++ b/arch/frv/kernel/process.c
@@ -204,7 +204,7 @@ int copy_thread(int nr, unsigned long clone_flags,
 
         regs0 = __kernel_frame0_ptr;
         childregs0 = (struct pt_regs *)
-                ((unsigned long) p->thread_info + THREAD_SIZE - USER_CONTEXT_SIZE);
+                (task_stack_page(p) + THREAD_SIZE - USER_CONTEXT_SIZE);
         childregs = childregs0;
 
         /* set up the userspace frame (the only place that the USP is stored) */
@@ -220,7 +220,7 @@ int copy_thread(int nr, unsigned long clone_flags,
                 *childregs = *regs;
                 childregs->sp = (unsigned long) childregs0;
                 childregs->next_frame = childregs0;
-                childregs->gr15 = (unsigned long) p->thread_info;
+                childregs->gr15 = (unsigned long) task_thread_info(p);
                 childregs->gr29 = (unsigned long) p;
         }
 
diff --git a/arch/h8300/kernel/process.c b/arch/h8300/kernel/process.c
index 585ed5efd0f7..ed79ae20e88d 100644
--- a/arch/h8300/kernel/process.c
+++ b/arch/h8300/kernel/process.c
@@ -195,7 +195,7 @@ int copy_thread(int nr, unsigned long clone_flags,
 {
         struct pt_regs * childregs;
 
-        childregs = ((struct pt_regs *) (THREAD_SIZE + (unsigned long) p->thread_info)) - 1;
+        childregs = (struct pt_regs *) (THREAD_SIZE + task_stack_page(p)) - 1;
 
         *childregs = *regs;
         childregs->retpc = (unsigned long) ret_from_fork;
diff --git a/arch/i386/kernel/process.c b/arch/i386/kernel/process.c
index 035928f3f6c1..2185377fdde1 100644
--- a/arch/i386/kernel/process.c
+++ b/arch/i386/kernel/process.c
@@ -424,18 +424,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long esp,
         struct task_struct *tsk;
         int err;
 
-        childregs = ((struct pt_regs *) (THREAD_SIZE + (unsigned long) p->thread_info)) - 1;
-        /*
-         * The below -8 is to reserve 8 bytes on top of the ring0 stack.
-         * This is necessary to guarantee that the entire "struct pt_regs"
-         * is accessable even if the CPU haven't stored the SS/ESP registers
-         * on the stack (interrupt gate does not save these registers
-         * when switching to the same priv ring).
-         * Therefore beware: accessing the xss/esp fields of the
-         * "struct pt_regs" is possible, but they may contain the
-         * completely wrong values.
-         */
-        childregs = (struct pt_regs *) ((unsigned long) childregs - 8);
+        childregs = task_pt_regs(p);
         *childregs = *regs;
         childregs->eax = 0;
         childregs->esp = esp;
@@ -540,12 +529,7 @@ EXPORT_SYMBOL(dump_thread);
  */
 int dump_task_regs(struct task_struct *tsk, elf_gregset_t *regs)
 {
-        struct pt_regs ptregs;
-        
-        ptregs = *(struct pt_regs *)
-                ((unsigned long)tsk->thread_info +
-                /* see comments in copy_thread() about -8 */
-                THREAD_SIZE - sizeof(ptregs) - 8);
+        struct pt_regs ptregs = *task_pt_regs(tsk);
         ptregs.xcs &= 0xffff;
         ptregs.xds &= 0xffff;
         ptregs.xes &= 0xffff;
@@ -601,8 +585,8 @@ static inline void disable_tsc(struct task_struct *prev_p,
          * gcc should eliminate the ->thread_info dereference if
          * has_secure_computing returns 0 at compile time (SECCOMP=n).
          */
-        prev = prev_p->thread_info;
-        next = next_p->thread_info;
+        prev = task_thread_info(prev_p);
+        next = task_thread_info(next_p);
 
         if (has_secure_computing(prev) || has_secure_computing(next)) {
                 /* slow path here */
@@ -787,7 +771,7 @@ unsigned long get_wchan(struct task_struct *p)
         int count = 0;
         if (!p || p == current || p->state == TASK_RUNNING)
                 return 0;
-        stack_page = (unsigned long)p->thread_info;
+        stack_page = (unsigned long)task_stack_page(p);
         esp = p->thread.esp;
         if (!stack_page || esp < stack_page || esp > top_esp+stack_page)
                 return 0;
diff --git a/arch/i386/kernel/smpboot.c b/arch/i386/kernel/smpboot.c
index b3c2e2c26743..255adb498268 100644
--- a/arch/i386/kernel/smpboot.c
+++ b/arch/i386/kernel/smpboot.c
@@ -875,8 +875,7 @@ static inline struct task_struct * alloc_idle_task(int cpu)
                 /* initialize thread_struct.  we really want to avoid destroy
                  * idle tread
                  */
-                idle->thread.esp = (unsigned long)(((struct pt_regs *)
-                        (THREAD_SIZE + (unsigned long) idle->thread_info)) - 1);
+                idle->thread.esp = (unsigned long)task_pt_regs(idle);
                 init_idle(idle, cpu);
                 return idle;
         }
@@ -1096,6 +1095,7 @@ static void smp_tune_scheduling (void)
                         cachesize = 16; /* Pentiums, 2x8kB cache */
                         bandwidth = 100;
                 }
+                max_cache_size = cachesize * 1024;
         }
 }
 
diff --git a/arch/i386/kernel/vm86.c b/arch/i386/kernel/vm86.c
index cbdb0afed76a..0c90ae54ddfa 100644
--- a/arch/i386/kernel/vm86.c
+++ b/arch/i386/kernel/vm86.c
@@ -311,7 +311,7 @@ static void do_sys_vm86(struct kernel_vm86_struct *info, struct task_struct *tsk
                 "movl %1,%%ebp\n\t"
                 "jmp resume_userspace"
                 : /* no outputs */
-                :"r" (&info->regs), "r" (tsk->thread_info) : "ax");
+                :"r" (&info->regs), "r" (task_thread_info(tsk)) : "ax");
         /* we never return here */
 }
 
diff --git a/arch/ia64/ia32/elfcore32.h b/arch/ia64/ia32/elfcore32.h
index b73b8b6b10c1..a47f63b204fb 100644
--- a/arch/ia64/ia32/elfcore32.h
+++ b/arch/ia64/ia32/elfcore32.h
@@ -95,8 +95,7 @@ static inline void elf_core_copy_regs(elf_gregset_t *elfregs,
 static inline int elf_core_copy_task_regs(struct task_struct *t,
                                           elf_gregset_t* elfregs)
 {
-        struct pt_regs *pp = ia64_task_regs(t);
-        ELF_CORE_COPY_REGS((*elfregs), pp);
+        ELF_CORE_COPY_REGS((*elfregs), task_pt_regs(t));
         return 1;
 }
 
diff --git a/arch/ia64/ia32/ia32_signal.c b/arch/ia64/ia32/ia32_signal.c
index aa891c9bc9b6..5856510210fa 100644
--- a/arch/ia64/ia32/ia32_signal.c
+++ b/arch/ia64/ia32/ia32_signal.c
@@ -255,7 +255,7 @@ save_ia32_fpstate_live (struct _fpstate_ia32 __user *save)
          */
         fp_tos = (fsr>>11)&0x7;
         fr8_st_map = (8-fp_tos)&0x7;
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         fpregp = (struct _fpreg_ia32 *)(((unsigned long)buf + 15) & ~15);
         ia64f2ia32f(fpregp, &ptp->f8);
         copy_to_user(&save->_st[(0+fr8_st_map)&0x7], fpregp, sizeof(struct _fpreg_ia32));
@@ -389,7 +389,7 @@ restore_ia32_fpstate_live (struct _fpstate_ia32 __user *save)
         fr8_st_map = (8-fp_tos)&0x7;
         fpregp = (struct _fpreg_ia32 *)(((unsigned long)buf + 15) & ~15);
 
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         copy_from_user(fpregp, &save->_st[(0+fr8_st_map)&0x7], sizeof(struct _fpreg_ia32));
         ia32f2ia64f(&ptp->f8, fpregp);
         copy_from_user(fpregp, &save->_st[(1+fr8_st_map)&0x7], sizeof(struct _fpreg_ia32));
diff --git a/arch/ia64/ia32/ia32_support.c b/arch/ia64/ia32/ia32_support.c
index 4f630043b3ae..c187743965a0 100644
--- a/arch/ia64/ia32/ia32_support.c
+++ b/arch/ia64/ia32/ia32_support.c
@@ -58,7 +58,7 @@ load_desc (u16 selector)
 void
 ia32_load_segment_descriptors (struct task_struct *task)
 {
-        struct pt_regs *regs = ia64_task_regs(task);
+        struct pt_regs *regs = task_pt_regs(task);
 
         /* Setup the segment descriptors */
         regs->r24 = load_desc(regs->r16 >> 16);         /* ESD */
@@ -113,7 +113,7 @@ void
 ia32_load_state (struct task_struct *t)
 {
         unsigned long eflag, fsr, fcr, fir, fdr, tssd;
-        struct pt_regs *regs = ia64_task_regs(t);
+        struct pt_regs *regs = task_pt_regs(t);
 
         eflag = t->thread.eflag;
         fsr = t->thread.fsr;
diff --git a/arch/ia64/ia32/sys_ia32.c b/arch/ia64/ia32/sys_ia32.c
index 0668b2b7714d..3945d378bd7e 100644
--- a/arch/ia64/ia32/sys_ia32.c
+++ b/arch/ia64/ia32/sys_ia32.c
@@ -1482,7 +1482,7 @@ getreg (struct task_struct *child, int regno)
 {
         struct pt_regs *child_regs;
 
-        child_regs = ia64_task_regs(child);
+        child_regs = task_pt_regs(child);
         switch (regno / sizeof(int)) {
               case PT_EBX: return child_regs->r11;
               case PT_ECX: return child_regs->r9;
@@ -1510,7 +1510,7 @@ putreg (struct task_struct *child, int regno, unsigned int value)
 {
         struct pt_regs *child_regs;
 
-        child_regs = ia64_task_regs(child);
+        child_regs = task_pt_regs(child);
         switch (regno / sizeof(int)) {
               case PT_EBX: child_regs->r11 = value; break;
               case PT_ECX: child_regs->r9 = value; break;
@@ -1626,7 +1626,7 @@ save_ia32_fpstate (struct task_struct *tsk, struct ia32_user_i387_struct __user
          *  Stack frames start with 16-bytes of temp space
          */
         swp = (struct switch_stack *)(tsk->thread.ksp + 16);
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         tos = (tsk->thread.fsr >> 11) & 7;
         for (i = 0; i < 8; i++)
                 put_fpreg(i, &save->st_space[i], ptp, swp, tos);
@@ -1659,7 +1659,7 @@ restore_ia32_fpstate (struct task_struct *tsk, struct ia32_user_i387_struct __us
          *  Stack frames start with 16-bytes of temp space
          */
         swp = (struct switch_stack *)(tsk->thread.ksp + 16);
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         tos = (tsk->thread.fsr >> 11) & 7;
         for (i = 0; i < 8; i++)
                 get_fpreg(i, &save->st_space[i], ptp, swp, tos);
@@ -1690,7 +1690,7 @@ save_ia32_fpxstate (struct task_struct *tsk, struct ia32_user_fxsr_struct __user
          *  Stack frames start with 16-bytes of temp space
          */
         swp = (struct switch_stack *)(tsk->thread.ksp + 16);
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         tos = (tsk->thread.fsr >> 11) & 7;
         for (i = 0; i < 8; i++)
                 put_fpreg(i, (struct _fpreg_ia32 __user *)&save->st_space[4*i], ptp, swp, tos);
@@ -1734,7 +1734,7 @@ restore_ia32_fpxstate (struct task_struct *tsk, struct ia32_user_fxsr_struct __u
          *  Stack frames start with 16-bytes of temp space
          */
         swp = (struct switch_stack *)(tsk->thread.ksp + 16);
-        ptp = ia64_task_regs(tsk);
+        ptp = task_pt_regs(tsk);
         tos = (tsk->thread.fsr >> 11) & 7;
         for (i = 0; i < 8; i++)
         get_fpreg(i, (struct _fpreg_ia32 __user *)&save->st_space[4*i], ptp, swp, tos);
diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c
index 355af15287c7..ee7eec9ee576 100644
--- a/arch/ia64/kernel/mca.c
+++ b/arch/ia64/kernel/mca.c
@@ -766,7 +766,7 @@ ia64_mca_modify_original_stack(struct pt_regs *regs,
                         l = strlen(previous_current->comm);
                 snprintf(comm, sizeof(comm), "%s %*s %d",
                         current->comm, l, previous_current->comm,
-                        previous_current->thread_info->cpu);
+                        task_thread_info(previous_current)->cpu);
         }
         memcpy(current->comm, comm, sizeof(current->comm));
 
@@ -1423,7 +1423,7 @@ format_mca_init_stack(void *mca_data, unsigned long offset,
         struct task_struct *p = (struct task_struct *)((char *)mca_data + offset);
         struct thread_info *ti;
         memset(p, 0, KERNEL_STACK_SIZE);
-        ti = (struct thread_info *)((char *)p + IA64_TASK_SIZE);
+        ti = task_thread_info(p);
         ti->flags = _TIF_MCA_INIT;
         ti->preempt_count = 1;
         ti->task = p;
diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
index c026ac1142a6..bd87cb6b7a81 100644
--- a/arch/ia64/kernel/perfmon.c
+++ b/arch/ia64/kernel/perfmon.c
@@ -1710,7 +1710,7 @@ static void
 pfm_syswide_force_stop(void *info)
 {
         pfm_context_t   *ctx = (pfm_context_t *)info;
-        struct pt_regs *regs = ia64_task_regs(current);
+        struct pt_regs *regs = task_pt_regs(current);
         struct task_struct *owner;
         unsigned long flags;
         int ret;
@@ -1815,7 +1815,7 @@ pfm_flush(struct file *filp)
         is_system = ctx->ctx_fl_system;
 
         task = PFM_CTX_TASK(ctx);
-        regs = ia64_task_regs(task);
+        regs = task_pt_regs(task);
 
         DPRINT(("ctx_state=%d is_current=%d\n",
                 state,
@@ -1945,7 +1945,7 @@ pfm_close(struct inode *inode, struct file *filp)
         is_system = ctx->ctx_fl_system;
 
         task = PFM_CTX_TASK(ctx);
-        regs = ia64_task_regs(task);
+        regs = task_pt_regs(task);
 
         DPRINT(("ctx_state=%d is_current=%d\n", 
                 state,
@@ -4052,7 +4052,7 @@ pfm_stop(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
                  */
                 ia64_psr(regs)->up = 0;
         } else {
-                tregs = ia64_task_regs(task);
+                tregs = task_pt_regs(task);
 
                 /*
                  * stop monitoring at the user level
@@ -4134,7 +4134,7 @@ pfm_start(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
                 ia64_psr(regs)->up = 1;
 
         } else {
-                tregs = ia64_task_regs(ctx->ctx_task);
+                tregs = task_pt_regs(ctx->ctx_task);
 
                 /*
                  * start monitoring at the kernel level the next
@@ -4404,7 +4404,7 @@ pfm_context_load(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
                 /*
                  * when not current, task MUST be stopped, so this is safe
                  */
-                regs = ia64_task_regs(task);
+                regs = task_pt_regs(task);
 
                 /* force a full reload */
                 ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
@@ -4530,7 +4530,7 @@ pfm_context_unload(pfm_context_t *ctx, void *arg, int count, struct pt_regs *reg
         /*
          * per-task mode
          */
-        tregs = task == current ? regs : ia64_task_regs(task);
+        tregs = task == current ? regs : task_pt_regs(task);
 
         if (task == current) {
                 /*
@@ -4593,7 +4593,7 @@ pfm_exit_thread(struct task_struct *task)
 {
         pfm_context_t *ctx;
         unsigned long flags;
-        struct pt_regs *regs = ia64_task_regs(task);
+        struct pt_regs *regs = task_pt_regs(task);
         int ret, state;
         int free_ok = 0;
 
@@ -4926,7 +4926,7 @@ restart_args:
         if (unlikely(ret)) goto abort_locked;
 
 skip_fd:
-        ret = (*func)(ctx, args_k, count, ia64_task_regs(current));
+        ret = (*func)(ctx, args_k, count, task_pt_regs(current));
 
         call_made = 1;
 
@@ -5050,7 +5050,7 @@ pfm_handle_work(void)
 
         pfm_clear_task_notify();
 
-        regs = ia64_task_regs(current);
+        regs = task_pt_regs(current);
 
         /*
          * extract reason for being here and clear
@@ -5794,7 +5794,7 @@ pfm_syst_wide_update_task(struct task_struct *task, unsigned long info, int is_c
          * on every CPU, so we can rely on the pid to identify the idle task.
          */
         if ((info & PFM_CPUINFO_EXCL_IDLE) == 0 || task->pid) {
-                regs = ia64_task_regs(task);
+                regs = task_pt_regs(task);
                 ia64_psr(regs)->pp = is_ctxswin ? dcr_pp : 0;
                 return;
         }
@@ -5877,7 +5877,7 @@ pfm_save_regs(struct task_struct *task)
         flags = pfm_protect_ctx_ctxsw(ctx);
 
         if (ctx->ctx_state == PFM_CTX_ZOMBIE) {
-                struct pt_regs *regs = ia64_task_regs(task);
+                struct pt_regs *regs = task_pt_regs(task);
 
                 pfm_clear_psr_up();
 
@@ -6077,7 +6077,7 @@ pfm_load_regs (struct task_struct *task)
         BUG_ON(psr & IA64_PSR_I);
 
         if (unlikely(ctx->ctx_state == PFM_CTX_ZOMBIE)) {
-                struct pt_regs *regs = ia64_task_regs(task);
+                struct pt_regs *regs = task_pt_regs(task);
 
                 BUG_ON(ctx->ctx_smpl_hdr);
 
@@ -6446,7 +6446,7 @@ pfm_alt_save_pmu_state(void *data)
 {
         struct pt_regs *regs;
 
-        regs = ia64_task_regs(current);
+        regs = task_pt_regs(current);
 
         DPRINT(("called\n"));
 
@@ -6472,7 +6472,7 @@ pfm_alt_restore_pmu_state(void *data)
 {
         struct pt_regs *regs;
 
-        regs = ia64_task_regs(current);
+        regs = task_pt_regs(current);
 
         DPRINT(("called\n"));
 
@@ -6754,7 +6754,7 @@ dump_pmu_state(const char *from)
         local_irq_save(flags);
 
         this_cpu = smp_processor_id();
-        regs     = ia64_task_regs(current);
+        regs     = task_pt_regs(current);
         info     = PFM_CPUINFO_GET();
         dcr      = ia64_getreg(_IA64_REG_CR_DCR);
 
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index e9904c74d2ba..309d59658e5f 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -328,7 +328,7 @@ ia64_save_extra (struct task_struct *task)
 #endif
 
 #ifdef CONFIG_IA32_SUPPORT
-        if (IS_IA32_PROCESS(ia64_task_regs(task)))
+        if (IS_IA32_PROCESS(task_pt_regs(task)))
                 ia32_save_state(task);
 #endif
 }
@@ -353,7 +353,7 @@ ia64_load_extra (struct task_struct *task)
 #endif
 
 #ifdef CONFIG_IA32_SUPPORT
-        if (IS_IA32_PROCESS(ia64_task_regs(task)))
+        if (IS_IA32_PROCESS(task_pt_regs(task)))
                 ia32_load_state(task);
 #endif
 }
@@ -488,7 +488,7 @@ copy_thread (int nr, unsigned long clone_flags,
          * If we're cloning an IA32 task then save the IA32 extra
          * state from the current task to the new task
          */
-        if (IS_IA32_PROCESS(ia64_task_regs(current))) {
+        if (IS_IA32_PROCESS(task_pt_regs(current))) {
                 ia32_save_state(p);
                 if (clone_flags & CLONE_SETTLS)
                         retval = ia32_clone_tls(p, child_ptregs);
@@ -701,7 +701,7 @@ int
 kernel_thread_helper (int (*fn)(void *), void *arg)
 {
 #ifdef CONFIG_IA32_SUPPORT
-        if (IS_IA32_PROCESS(ia64_task_regs(current))) {
+        if (IS_IA32_PROCESS(task_pt_regs(current))) {
                 /* A kernel thread is always a 64-bit process. */
                 current->thread.map_base  = DEFAULT_MAP_BASE;
                 current->thread.task_size = DEFAULT_TASK_SIZE;
@@ -722,7 +722,7 @@ flush_thread (void)
         current->thread.flags &= ~(IA64_THREAD_FPH_VALID | IA64_THREAD_DBG_VALID);
         ia64_drop_fpu(current);
 #ifdef CONFIG_IA32_SUPPORT
-        if (IS_IA32_PROCESS(ia64_task_regs(current))) {
+        if (IS_IA32_PROCESS(task_pt_regs(current))) {
                 ia32_drop_partial_page_list(current);
                 current->thread.task_size = IA32_PAGE_OFFSET;
                 set_fs(USER_DS);
@@ -755,7 +755,7 @@ exit_thread (void)
         if (current->thread.flags & IA64_THREAD_DBG_VALID)
                 pfm_release_debug_registers(current);
 #endif
-        if (IS_IA32_PROCESS(ia64_task_regs(current)))
+        if (IS_IA32_PROCESS(task_pt_regs(current)))
                 ia32_drop_partial_page_list(current);
 }
 
diff --git a/arch/ia64/kernel/ptrace.c b/arch/ia64/kernel/ptrace.c
index 8d88eeea02d1..eaed14aac6aa 100644
--- a/arch/ia64/kernel/ptrace.c
+++ b/arch/ia64/kernel/ptrace.c
@@ -254,7 +254,7 @@ get_rnat (struct task_struct *task, struct switch_stack *sw,
         long num_regs, nbits;
         struct pt_regs *pt;
 
-        pt = ia64_task_regs(task);
+        pt = task_pt_regs(task);
         kbsp = (unsigned long *) sw->ar_bspstore;
         ubspstore = (unsigned long *) pt->ar_bspstore;
 
@@ -314,7 +314,7 @@ put_rnat (struct task_struct *task, struct switch_stack *sw,
         struct pt_regs *pt;
         unsigned long cfm, *urbs_kargs;
 
-        pt = ia64_task_regs(task);
+        pt = task_pt_regs(task);
         kbsp = (unsigned long *) sw->ar_bspstore;
         ubspstore = (unsigned long *) pt->ar_bspstore;
 
@@ -407,7 +407,7 @@ ia64_peek (struct task_struct *child, struct switch_stack *child_stack,
 
         urbs_end = (long *) user_rbs_end;
         laddr = (unsigned long *) addr;
-        child_regs = ia64_task_regs(child);
+        child_regs = task_pt_regs(child);
         bspstore = (unsigned long *) child_regs->ar_bspstore;
         krbs = (unsigned long *) child + IA64_RBS_OFFSET/8;
         if (on_kernel_rbs(addr, (unsigned long) bspstore,
@@ -467,7 +467,7 @@ ia64_poke (struct task_struct *child, struct switch_stack *child_stack,
         struct pt_regs *child_regs;
 
         laddr = (unsigned long *) addr;
-        child_regs = ia64_task_regs(child);
+        child_regs = task_pt_regs(child);
         bspstore = (unsigned long *) child_regs->ar_bspstore;
         krbs = (unsigned long *) child + IA64_RBS_OFFSET/8;
         if (on_kernel_rbs(addr, (unsigned long) bspstore,
@@ -567,7 +567,7 @@ thread_matches (struct task_struct *thread, unsigned long addr)
                  */
                 return 0;
 
-        thread_regs = ia64_task_regs(thread);
+        thread_regs = task_pt_regs(thread);
         thread_rbs_end = ia64_get_user_rbs_end(thread, thread_regs, NULL);
         if (!on_kernel_rbs(addr, thread_regs->ar_bspstore, thread_rbs_end))
                 return 0;
@@ -627,7 +627,7 @@ find_thread_for_addr (struct task_struct *child, unsigned long addr)
 inline void
 ia64_flush_fph (struct task_struct *task)
 {
-        struct ia64_psr *psr = ia64_psr(ia64_task_regs(task));
+        struct ia64_psr *psr = ia64_psr(task_pt_regs(task));
 
         /*
          * Prevent migrating this task while
@@ -653,7 +653,7 @@ ia64_flush_fph (struct task_struct *task)
 void
 ia64_sync_fph (struct task_struct *task)
 {
-        struct ia64_psr *psr = ia64_psr(ia64_task_regs(task));
+        struct ia64_psr *psr = ia64_psr(task_pt_regs(task));
 
         ia64_flush_fph(task);
         if (!(task->thread.flags & IA64_THREAD_FPH_VALID)) {
@@ -794,7 +794,7 @@ access_uarea (struct task_struct *child, unsigned long addr,
                                           + offsetof(struct pt_regs, reg)))
 
 
-        pt = ia64_task_regs(child);
+        pt = task_pt_regs(child);
         sw = (struct switch_stack *) (child->thread.ksp + 16);
 
         if ((addr & 0x7) != 0) {
@@ -1120,7 +1120,7 @@ ptrace_getregs (struct task_struct *child, struct pt_all_user_regs __user *ppr)
         if (!access_ok(VERIFY_WRITE, ppr, sizeof(struct pt_all_user_regs)))
                 return -EIO;
 
-        pt = ia64_task_regs(child);
+        pt = task_pt_regs(child);
         sw = (struct switch_stack *) (child->thread.ksp + 16);
         unw_init_from_blocked_task(&info, child);
         if (unw_unwind_to_user(&info) < 0) {
@@ -1265,7 +1265,7 @@ ptrace_setregs (struct task_struct *child, struct pt_all_user_regs __user *ppr)
         if (!access_ok(VERIFY_READ, ppr, sizeof(struct pt_all_user_regs)))
                 return -EIO;
 
-        pt = ia64_task_regs(child);
+        pt = task_pt_regs(child);
         sw = (struct switch_stack *) (child->thread.ksp + 16);
         unw_init_from_blocked_task(&info, child);
         if (unw_unwind_to_user(&info) < 0) {
@@ -1403,7 +1403,7 @@ ptrace_setregs (struct task_struct *child, struct pt_all_user_regs __user *ppr)
 void
 ptrace_disable (struct task_struct *child)
 {
-        struct ia64_psr *child_psr = ia64_psr(ia64_task_regs(child));
+        struct ia64_psr *child_psr = ia64_psr(task_pt_regs(child));
 
         /* make sure the single step/taken-branch trap bits are not set: */
         child_psr->ss = 0;
@@ -1456,7 +1456,7 @@ sys_ptrace (long request, pid_t pid, unsigned long addr, unsigned long data)
         if (ret < 0)
                 goto out_tsk;
 
-        pt = ia64_task_regs(child);
+        pt = task_pt_regs(child);
         sw = (struct switch_stack *) (child->thread.ksp + 16);
 
         switch (request) {
diff --git a/arch/ia64/kernel/setup.c b/arch/ia64/kernel/setup.c
index c33305d8e5eb..c0766575a3a2 100644
--- a/arch/ia64/kernel/setup.c
+++ b/arch/ia64/kernel/setup.c
@@ -60,6 +60,7 @@
 #include <asm/smp.h>
 #include <asm/system.h>
 #include <asm/unistd.h>
+#include <asm/system.h>
 
 #if defined(CONFIG_SMP) && (IA64_CPU_SIZE > PAGE_SIZE)
 # error "struct cpuinfo_ia64 too big!"
@@ -695,6 +696,7 @@ static void
 get_max_cacheline_size (void)
 {
         unsigned long line_size, max = 1;
+        unsigned int cache_size = 0;
         u64 l, levels, unique_caches;
         pal_cache_config_info_t cci;
         s64 status;
@@ -724,6 +726,8 @@ get_max_cacheline_size (void)
                 line_size = 1 << cci.pcci_line_size;
                 if (line_size > max)
                         max = line_size;
+                if (cache_size < cci.pcci_cache_size)
+                        cache_size = cci.pcci_cache_size;
                 if (!cci.pcci_unified) {
                         status = ia64_pal_cache_config_info(l,
                                                     /* cache_type (instruction)= */ 1,
@@ -740,6 +744,9 @@ get_max_cacheline_size (void)
                         ia64_i_cache_stride_shift = cci.pcci_stride;
         }
   out:
+#ifdef CONFIG_SMP
+        max_cache_size = max(max_cache_size, cache_size);
+#endif
         if (max > ia64_max_cacheline_size)
                 ia64_max_cacheline_size = max;
 }
@@ -794,7 +801,7 @@ cpu_init (void)
 #endif
 
         /* Clear the stack memory reserved for pt_regs: */
-        memset(ia64_task_regs(current), 0, sizeof(struct pt_regs));
+        memset(task_pt_regs(current), 0, sizeof(struct pt_regs));
 
         ia64_set_kr(IA64_KR_FPU_OWNER, 0);
 
@@ -870,6 +877,15 @@ cpu_init (void)
         pm_idle = default_idle;
 }
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ */
+void sched_cacheflush(void)
+{
+        ia64_sal_cache_flush(3);
+}
+
 void
 check_bugs (void)
 {
diff --git a/arch/ia64/kernel/signal.c b/arch/ia64/kernel/signal.c
index 58ce07efc56e..463f6bb44d07 100644
--- a/arch/ia64/kernel/signal.c
+++ b/arch/ia64/kernel/signal.c
@@ -655,11 +655,11 @@ set_sigdelayed(pid_t pid, int signo, int code, void __user *addr)
 
                 if (!t)
                         return;
-                t->thread_info->sigdelayed.signo = signo;
-                t->thread_info->sigdelayed.code = code;
-                t->thread_info->sigdelayed.addr = addr;
-                t->thread_info->sigdelayed.start_time = start_time;
-                t->thread_info->sigdelayed.pid = pid;
+                task_thread_info(t)->sigdelayed.signo = signo;
+                task_thread_info(t)->sigdelayed.code = code;
+                task_thread_info(t)->sigdelayed.addr = addr;
+                task_thread_info(t)->sigdelayed.start_time = start_time;
+                task_thread_info(t)->sigdelayed.pid = pid;
                 wmb();
                 set_tsk_thread_flag(t, TIF_SIGDELAYED);
         }
diff --git a/arch/ia64/kernel/sys_ia64.c b/arch/ia64/kernel/sys_ia64.c
index f2dbcd1db0d4..c7b943f10199 100644
--- a/arch/ia64/kernel/sys_ia64.c
+++ b/arch/ia64/kernel/sys_ia64.c
@@ -151,7 +151,7 @@ out:
 asmlinkage long
 sys_pipe (void)
 {
-        struct pt_regs *regs = ia64_task_regs(current);
+        struct pt_regs *regs = task_pt_regs(current);
         int fd[2];
         int retval;
 
diff --git a/arch/m32r/kernel/process.c b/arch/m32r/kernel/process.c
index 2a1f250349b7..5dfc7ea45cf7 100644
--- a/arch/m32r/kernel/process.c
+++ b/arch/m32r/kernel/process.c
@@ -242,13 +242,10 @@ int dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpu)
 int copy_thread(int nr, unsigned long clone_flags, unsigned long spu,
         unsigned long unused, struct task_struct *tsk, struct pt_regs *regs)
 {
-        struct pt_regs *childregs;
-        unsigned long sp = (unsigned long)tsk->thread_info + THREAD_SIZE;
+        struct pt_regs *childregs = task_pt_regs(tsk);
         extern void ret_from_fork(void);
 
         /* Copy registers */
-        sp -= sizeof (struct pt_regs);
-        childregs = (struct pt_regs *)sp;
         *childregs = *regs;
 
         childregs->spu = spu;
diff --git a/arch/m32r/kernel/ptrace.c b/arch/m32r/kernel/ptrace.c
index 9b75caaf5cec..340a3bf59b88 100644
--- a/arch/m32r/kernel/ptrace.c
+++ b/arch/m32r/kernel/ptrace.c
@@ -35,23 +35,6 @@
 #include <asm/mmu_context.h>
 
 /*
- * Get the address of the live pt_regs for the specified task.
- * These are saved onto the top kernel stack when the process
- * is not running.
- *
- * Note: if a user thread is execve'd from kernel space, the
- * kernel stack will not be empty on entry to the kernel, so
- * ptracing these tasks will fail.
- */
-static inline struct pt_regs *
-get_user_regs(struct task_struct *task)
-{
-        return (struct pt_regs *)
-                ((unsigned long)task->thread_info + THREAD_SIZE
-                 - sizeof(struct pt_regs));
-}
-
-/*
  * This routine will get a word off of the process kernel stack.
  */
 static inline unsigned long int
@@ -59,7 +42,7 @@ get_stack_long(struct task_struct *task, int offset)
 {
         unsigned long *stack;
 
-        stack = (unsigned long *)get_user_regs(task);
+        stack = (unsigned long *)task_pt_regs(task);
 
         return stack[offset];
 }
@@ -72,7 +55,7 @@ put_stack_long(struct task_struct *task, int offset, unsigned long data)
 {
         unsigned long *stack;
 
-        stack = (unsigned long *)get_user_regs(task);
+        stack = (unsigned long *)task_pt_regs(task);
         stack[offset] = data;
 
         return 0;
@@ -208,7 +191,7 @@ static int ptrace_write_user(struct task_struct *tsk, unsigned long off,
  */
 static int ptrace_getregs(struct task_struct *tsk, void __user *uregs)
 {
-        struct pt_regs *regs = get_user_regs(tsk);
+        struct pt_regs *regs = task_pt_regs(tsk);
 
         return copy_to_user(uregs, regs, sizeof(struct pt_regs)) ? -EFAULT : 0;
 }
@@ -223,7 +206,7 @@ static int ptrace_setregs(struct task_struct *tsk, void __user *uregs)
 
         ret = -EFAULT;
         if (copy_from_user(&newregs, uregs, sizeof(struct pt_regs)) == 0) {
-                struct pt_regs *regs = get_user_regs(tsk);
+                struct pt_regs *regs = task_pt_regs(tsk);
                 *regs = newregs;
                 ret = 0;
         }
diff --git a/arch/m32r/kernel/smpboot.c b/arch/m32r/kernel/smpboot.c
index b90c54169fa5..d7ec16e7fb25 100644
--- a/arch/m32r/kernel/smpboot.c
+++ b/arch/m32r/kernel/smpboot.c
@@ -286,7 +286,7 @@ static void __init do_boot_cpu(int phys_id)
         /* So we see what's up   */
         printk("Booting processor %d/%d\n", phys_id, cpu_id);
         stack_start.spi = (void *)idle->thread.sp;
-        idle->thread_info->cpu = cpu_id;
+        task_thread_info(idle)->cpu = cpu_id;
 
         /*
          * Send Startup IPI
diff --git a/arch/m68k/amiga/amiints.c b/arch/m68k/amiga/amiints.c
index d9edf2d1a492..b0aa61bf8700 100644
--- a/arch/m68k/amiga/amiints.c
+++ b/arch/m68k/amiga/amiints.c
@@ -126,9 +126,9 @@ void __init amiga_init_IRQ(void)
                 gayle.inten = GAYLE_IRQ_IDE;
 
         /* turn off all interrupts and enable the master interrupt bit */
-        custom.intena = 0x7fff;
-        custom.intreq = 0x7fff;
-        custom.intena = IF_SETCLR | IF_INTEN;
+        amiga_custom.intena = 0x7fff;
+        amiga_custom.intreq = 0x7fff;
+        amiga_custom.intena = IF_SETCLR | IF_INTEN;
 
         cia_init_IRQ(&ciaa_base);
         cia_init_IRQ(&ciab_base);
@@ -245,7 +245,7 @@ int amiga_request_irq(unsigned int irq,
 
         /* enable the interrupt */
         if (irq < IRQ_AMIGA_PORTS && !ami_ablecount[irq])
-                custom.intena = IF_SETCLR | amiga_intena_vals[irq];
+                amiga_custom.intena = IF_SETCLR | amiga_intena_vals[irq];
 
         return error;
 }
@@ -274,7 +274,7 @@ void amiga_free_irq(unsigned int irq, void *dev_id)
                 amiga_delete_irq(&ami_irq_list[irq], dev_id);
                 /* if server list empty, disable the interrupt */
                 if (!ami_irq_list[irq] && irq < IRQ_AMIGA_PORTS)
-                        custom.intena = amiga_intena_vals[irq];
+                        amiga_custom.intena = amiga_intena_vals[irq];
         } else {
                 if (ami_irq_list[irq]->dev_id != dev_id)
                         printk("%s: removing probably wrong IRQ %d from %s\n",
@@ -283,7 +283,7 @@ void amiga_free_irq(unsigned int irq, void *dev_id)
                 ami_irq_list[irq]->flags   = 0;
                 ami_irq_list[irq]->dev_id  = NULL;
                 ami_irq_list[irq]->devname = NULL;
-                custom.intena = amiga_intena_vals[irq];
+                amiga_custom.intena = amiga_intena_vals[irq];
         }
 }
 
@@ -327,7 +327,7 @@ void amiga_enable_irq(unsigned int irq)
         }
 
         /* enable the interrupt */
-        custom.intena = IF_SETCLR | amiga_intena_vals[irq];
+        amiga_custom.intena = IF_SETCLR | amiga_intena_vals[irq];
 }
 
 void amiga_disable_irq(unsigned int irq)
@@ -358,7 +358,7 @@ void amiga_disable_irq(unsigned int irq)
         }
 
         /* disable the interrupt */
-        custom.intena = amiga_intena_vals[irq];
+        amiga_custom.intena = amiga_intena_vals[irq];
 }
 
 inline void amiga_do_irq(int irq, struct pt_regs *fp)
@@ -373,7 +373,7 @@ void amiga_do_irq_list(int irq, struct pt_regs *fp)
 
         kstat_cpu(0).irqs[SYS_IRQS + irq]++;
 
-        custom.intreq = amiga_intena_vals[irq];
+        amiga_custom.intreq = amiga_intena_vals[irq];
 
         for (node = ami_irq_list[irq]; node; node = node->next)
                 node->handler(irq, node->dev_id, fp);
@@ -385,23 +385,23 @@ void amiga_do_irq_list(int irq, struct pt_regs *fp)
 
 static irqreturn_t ami_int1(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if serial transmit buffer empty, interrupt */
         if (ints & IF_TBE) {
-                custom.intreq = IF_TBE;
+                amiga_custom.intreq = IF_TBE;
                 amiga_do_irq(IRQ_AMIGA_TBE, fp);
         }
 
         /* if floppy disk transfer complete, interrupt */
         if (ints & IF_DSKBLK) {
-                custom.intreq = IF_DSKBLK;
+                amiga_custom.intreq = IF_DSKBLK;
                 amiga_do_irq(IRQ_AMIGA_DSKBLK, fp);
         }
 
         /* if software interrupt set, interrupt */
         if (ints & IF_SOFT) {
-                custom.intreq = IF_SOFT;
+                amiga_custom.intreq = IF_SOFT;
                 amiga_do_irq(IRQ_AMIGA_SOFT, fp);
         }
         return IRQ_HANDLED;
@@ -409,17 +409,17 @@ static irqreturn_t ami_int1(int irq, void *dev_id, struct pt_regs *fp)
 
 static irqreturn_t ami_int3(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if a blitter interrupt */
         if (ints & IF_BLIT) {
-                custom.intreq = IF_BLIT;
+                amiga_custom.intreq = IF_BLIT;
                 amiga_do_irq(IRQ_AMIGA_BLIT, fp);
         }
 
         /* if a copper interrupt */
         if (ints & IF_COPER) {
-                custom.intreq = IF_COPER;
+                amiga_custom.intreq = IF_COPER;
                 amiga_do_irq(IRQ_AMIGA_COPPER, fp);
         }
 
@@ -431,29 +431,29 @@ static irqreturn_t ami_int3(int irq, void *dev_id, struct pt_regs *fp)
 
 static irqreturn_t ami_int4(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if audio 0 interrupt */
         if (ints & IF_AUD0) {
-                custom.intreq = IF_AUD0;
+                amiga_custom.intreq = IF_AUD0;
                 amiga_do_irq(IRQ_AMIGA_AUD0, fp);
         }
 
         /* if audio 1 interrupt */
         if (ints & IF_AUD1) {
-                custom.intreq = IF_AUD1;
+                amiga_custom.intreq = IF_AUD1;
                 amiga_do_irq(IRQ_AMIGA_AUD1, fp);
         }
 
         /* if audio 2 interrupt */
         if (ints & IF_AUD2) {
-                custom.intreq = IF_AUD2;
+                amiga_custom.intreq = IF_AUD2;
                 amiga_do_irq(IRQ_AMIGA_AUD2, fp);
         }
 
         /* if audio 3 interrupt */
         if (ints & IF_AUD3) {
-                custom.intreq = IF_AUD3;
+                amiga_custom.intreq = IF_AUD3;
                 amiga_do_irq(IRQ_AMIGA_AUD3, fp);
         }
         return IRQ_HANDLED;
@@ -461,7 +461,7 @@ static irqreturn_t ami_int4(int irq, void *dev_id, struct pt_regs *fp)
 
 static irqreturn_t ami_int5(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if serial receive buffer full interrupt */
         if (ints & IF_RBF) {
@@ -471,7 +471,7 @@ static irqreturn_t ami_int5(int irq, void *dev_id, struct pt_regs *fp)
 
         /* if a disk sync interrupt */
         if (ints & IF_DSKSYN) {
-                custom.intreq = IF_DSKSYN;
+                amiga_custom.intreq = IF_DSKSYN;
                 amiga_do_irq(IRQ_AMIGA_DSKSYN, fp);
         }
         return IRQ_HANDLED;
diff --git a/arch/m68k/amiga/amisound.c b/arch/m68k/amiga/amisound.c
index bd5d134e9f12..ae94db5d93b2 100644
--- a/arch/m68k/amiga/amisound.c
+++ b/arch/m68k/amiga/amisound.c
@@ -24,6 +24,8 @@ static const signed char sine_data[] = {
 };
 #define DATA_SIZE       (sizeof(sine_data)/sizeof(sine_data[0]))
 
+#define custom amiga_custom
+
     /*
      * The minimum period for audio may be modified by the frame buffer
      * device since it depends on htotal (for OCS/ECS/AGA)
diff --git a/arch/m68k/amiga/cia.c b/arch/m68k/amiga/cia.c
index 7d55682615e3..9476eb9440f5 100644
--- a/arch/m68k/amiga/cia.c
+++ b/arch/m68k/amiga/cia.c
@@ -60,7 +60,7 @@ unsigned char cia_set_irq(struct ciabase *base, unsigned char mask)
         else
                 base->icr_data &= ~mask;
         if (base->icr_data & base->icr_mask)
-                custom.intreq = IF_SETCLR | base->int_mask;
+                amiga_custom.intreq = IF_SETCLR | base->int_mask;
         return old & base->icr_mask;
 }
 
@@ -89,7 +89,7 @@ unsigned char cia_able_irq(struct ciabase *base, unsigned char mask)
                 }
         }
         if (base->icr_data & base->icr_mask)
-                custom.intreq = IF_SETCLR | base->int_mask;
+                amiga_custom.intreq = IF_SETCLR | base->int_mask;
         return old;
 }
 
@@ -133,7 +133,7 @@ static irqreturn_t cia_handler(int irq, void *dev_id, struct pt_regs *fp)
         mach_irq = base->cia_irq;
         irq = SYS_IRQS + mach_irq;
         ints = cia_set_irq(base, CIA_ICR_ALL);
-        custom.intreq = base->int_mask;
+        amiga_custom.intreq = base->int_mask;
         for (i = 0; i < CIA_IRQS; i++, irq++, mach_irq++) {
                 if (ints & 1) {
                         kstat_cpu(0).irqs[irq]++;
@@ -162,7 +162,7 @@ void __init cia_init_IRQ(struct ciabase *base)
         /* install CIA handler */
         request_irq(base->handler_irq, cia_handler, 0, base->name, base);
 
-        custom.intena = IF_SETCLR | base->int_mask;
+        amiga_custom.intena = IF_SETCLR | base->int_mask;
 }
 
 int cia_get_irq_list(struct ciabase *base, struct seq_file *p)
diff --git a/arch/m68k/amiga/config.c b/arch/m68k/amiga/config.c
index 4775e18a78f0..12e3706fe02c 100644
--- a/arch/m68k/amiga/config.c
+++ b/arch/m68k/amiga/config.c
@@ -105,9 +105,6 @@ static int a2000_hwclk (int, struct rtc_time *);
 static int amiga_set_clock_mmss (unsigned long);
 static unsigned int amiga_get_ss (void);
 extern void amiga_mksound( unsigned int count, unsigned int ticks );
-#ifdef CONFIG_AMIGA_FLOPPY
-extern void amiga_floppy_setup(char *, int *);
-#endif
 static void amiga_reset (void);
 extern void amiga_init_sound(void);
 static void amiga_savekmsg_init(void);
@@ -290,7 +287,7 @@ static void __init amiga_identify(void)
     case CS_OCS:
     case CS_ECS:
     case CS_AGA:
-      switch (custom.deniseid & 0xf) {
+      switch (amiga_custom.deniseid & 0xf) {
       case 0x0c:
         AMIGAHW_SET(DENISE_HR);
         break;
@@ -303,7 +300,7 @@ static void __init amiga_identify(void)
       AMIGAHW_SET(DENISE);
       break;
     }
-    switch ((custom.vposr>>8) & 0x7f) {
+    switch ((amiga_custom.vposr>>8) & 0x7f) {
     case 0x00:
       AMIGAHW_SET(AGNUS_PAL);
       break;
@@ -427,13 +424,7 @@ void __init config_amiga(void)
 
   mach_set_clock_mmss  = amiga_set_clock_mmss;
   mach_get_ss          = amiga_get_ss;
-#ifdef CONFIG_AMIGA_FLOPPY
-  mach_floppy_setup    = amiga_floppy_setup;
-#endif
   mach_reset           = amiga_reset;
-#ifdef CONFIG_DUMMY_CONSOLE
-  conswitchp           = &dummy_con;
-#endif
 #if defined(CONFIG_INPUT_M68K_BEEP) || defined(CONFIG_INPUT_M68K_BEEP_MODULE)
   mach_beep            = amiga_mksound;
 #endif
@@ -447,9 +438,9 @@ void __init config_amiga(void)
   amiga_colorclock = 5*amiga_eclock;    /* 3.5 MHz */
 
   /* clear all DMA bits */
-  custom.dmacon = DMAF_ALL;
+  amiga_custom.dmacon = DMAF_ALL;
   /* ensure that the DMA master bit is set */
-  custom.dmacon = DMAF_SETCLR | DMAF_MASTER;
+  amiga_custom.dmacon = DMAF_SETCLR | DMAF_MASTER;
 
   /* don't use Z2 RAM as system memory on Z3 capable machines */
   if (AMIGAHW_PRESENT(ZORRO3)) {
@@ -830,8 +821,8 @@ static void amiga_savekmsg_init(void)
 
 static void amiga_serial_putc(char c)
 {
-    custom.serdat = (unsigned char)c | 0x100;
-    while (!(custom.serdatr & 0x2000))
+    amiga_custom.serdat = (unsigned char)c | 0x100;
+    while (!(amiga_custom.serdatr & 0x2000))
         ;
 }
 
@@ -855,11 +846,11 @@ int amiga_serial_console_wait_key(struct console *co)
 {
     int ch;
 
-    while (!(custom.intreqr & IF_RBF))
+    while (!(amiga_custom.intreqr & IF_RBF))
         barrier();
-    ch = custom.serdatr & 0xff;
+    ch = amiga_custom.serdatr & 0xff;
     /* clear the interrupt, so that another character can be read */
-    custom.intreq = IF_RBF;
+    amiga_custom.intreq = IF_RBF;
     return ch;
 }
 
diff --git a/arch/m68k/apollo/config.c b/arch/m68k/apollo/config.c
index 264929471253..d401962d9b25 100644
--- a/arch/m68k/apollo/config.c
+++ b/arch/m68k/apollo/config.c
@@ -176,9 +176,6 @@ void config_apollo(void) {
         mach_set_clock_mmss  = dn_dummy_set_clock_mmss; /* */
         mach_process_int     = dn_process_int;
         mach_reset           = dn_dummy_reset;  /* */
-#ifdef CONFIG_DUMMY_CONSOLE
-        conswitchp           = &dummy_con;
-#endif
 #ifdef CONFIG_HEARTBEAT
         mach_heartbeat = dn_heartbeat;
 #endif
diff --git a/arch/m68k/atari/config.c b/arch/m68k/atari/config.c
index 9261d2deeaf5..1012b08e5522 100644
--- a/arch/m68k/atari/config.c
+++ b/arch/m68k/atari/config.c
@@ -52,9 +52,6 @@ int atari_rtc_year_offset;
 
 /* local function prototypes */
 static void atari_reset( void );
-#ifdef CONFIG_ATARI_FLOPPY
-extern void atari_floppy_setup(char *, int *);
-#endif
 static void atari_get_model(char *model);
 static int atari_get_hardware_list(char *buffer);
 
@@ -244,12 +241,6 @@ void __init config_atari(void)
     mach_get_irq_list    = show_atari_interrupts;
     mach_gettimeoffset   = atari_gettimeoffset;
     mach_reset           = atari_reset;
-#ifdef CONFIG_ATARI_FLOPPY
-    mach_floppy_setup    = atari_floppy_setup;
-#endif
-#ifdef CONFIG_DUMMY_CONSOLE
-    conswitchp           = &dummy_con;
-#endif
     mach_max_dma_address = 0xffffff;
 #if defined(CONFIG_INPUT_M68K_BEEP) || defined(CONFIG_INPUT_M68K_BEEP_MODULE)
     mach_beep          = atari_mksound;
diff --git a/arch/m68k/bvme6000/rtc.c b/arch/m68k/bvme6000/rtc.c
index f7573f2bcb9c..703cbc6dc9cc 100644
--- a/arch/m68k/bvme6000/rtc.c
+++ b/arch/m68k/bvme6000/rtc.c
@@ -47,6 +47,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
         unsigned char msr;
         unsigned long flags;
         struct rtc_time wtime;
+        void __user *argp = (void __user *)arg;
 
         switch (cmd) {
         case RTC_RD_TIME:       /* Read the time/date from RTC  */
@@ -69,7 +70,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
                 } while (wtime.tm_sec != BCD2BIN(rtc->bcd_sec));
                 rtc->msr = msr;
                 local_irq_restore(flags);
-                return copy_to_user((void *)arg, &wtime, sizeof wtime) ?
+                return copy_to_user(argp, &wtime, sizeof wtime) ?
                                                                 -EFAULT : 0;
         }
         case RTC_SET_TIME:      /* Set the RTC */
@@ -81,8 +82,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
                 if (!capable(CAP_SYS_ADMIN))
                         return -EACCES;
 
-                if (copy_from_user(&rtc_tm, (struct rtc_time*)arg,
-                                   sizeof(struct rtc_time)))
+                if (copy_from_user(&rtc_tm, argp, sizeof(struct rtc_time)))
                         return -EFAULT;
 
                 yrs = rtc_tm.tm_year;
diff --git a/arch/m68k/hp300/config.c b/arch/m68k/hp300/config.c
index a0b854f3f94a..6d129eef370f 100644
--- a/arch/m68k/hp300/config.c
+++ b/arch/m68k/hp300/config.c
@@ -261,9 +261,6 @@ void __init config_hp300(void)
 #ifdef CONFIG_HEARTBEAT
         mach_heartbeat       = hp300_pulse;
 #endif
-#ifdef CONFIG_DUMMY_CONSOLE
-        conswitchp           = &dummy_con;
-#endif
         mach_max_dma_address = 0xffffffff;
 
         if (hp300_model >= HP_330 && hp300_model <= HP_433S && hp300_model != HP_350) {
diff --git a/arch/m68k/kernel/asm-offsets.c b/arch/m68k/kernel/asm-offsets.c
index c787c5ba9513..246a8820c223 100644
--- a/arch/m68k/kernel/asm-offsets.c
+++ b/arch/m68k/kernel/asm-offsets.c
@@ -92,7 +92,7 @@ int main(void)
         DEFINE(TRAP_TRACE, TRAP_TRACE);
 
         /* offsets into the custom struct */
-        DEFINE(CUSTOMBASE, &custom);
+        DEFINE(CUSTOMBASE, &amiga_custom);
         DEFINE(C_INTENAR, offsetof(struct CUSTOM, intenar));
         DEFINE(C_INTREQR, offsetof(struct CUSTOM, intreqr));
         DEFINE(C_INTENA, offsetof(struct CUSTOM, intena));
diff --git a/arch/m68k/kernel/head.S b/arch/m68k/kernel/head.S
index d4336d846df1..70002c146eed 100644
--- a/arch/m68k/kernel/head.S
+++ b/arch/m68k/kernel/head.S
@@ -273,8 +273,10 @@
  * Macintosh console support
  */
 
+#ifdef CONFIG_FRAMEBUFFER_CONSOLE
 #define CONSOLE
 #define CONSOLE_PENGUIN
+#endif
 
 /*
  * Macintosh serial debug support; outputs boot info to the printer
diff --git a/arch/m68k/kernel/process.c b/arch/m68k/kernel/process.c
index 13d109328a42..3f9cb55d0356 100644
--- a/arch/m68k/kernel/process.c
+++ b/arch/m68k/kernel/process.c
@@ -238,10 +238,9 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 {
         struct pt_regs * childregs;
         struct switch_stack * childstack, *stack;
-        unsigned long stack_offset, *retp;
+        unsigned long *retp;
 
-        stack_offset = THREAD_SIZE - sizeof(struct pt_regs);
-        childregs = (struct pt_regs *) ((unsigned long) (p->thread_info) + stack_offset);
+        childregs = (struct pt_regs *) (task_stack_page(p) + THREAD_SIZE) - 1;
 
         *childregs = *regs;
         childregs->d0 = 0;
@@ -386,7 +385,7 @@ unsigned long get_wchan(struct task_struct *p)
         if (!p || p == current || p->state == TASK_RUNNING)
                 return 0;
 
-        stack_page = (unsigned long)(p->thread_info);
+        stack_page = (unsigned long)task_stack_page(p);
         fp = ((struct switch_stack *)p->thread.ksp)->a6;
         do {
                 if (fp < stack_page+sizeof(struct thread_info) ||
diff --git a/arch/m68k/kernel/setup.c b/arch/m68k/kernel/setup.c
index d6ca99242e5a..750d5b3c971f 100644
--- a/arch/m68k/kernel/setup.c
+++ b/arch/m68k/kernel/setup.c
@@ -84,9 +84,6 @@ void (*mach_reset)( void );
 void (*mach_halt)( void );
 void (*mach_power_off)( void );
 long mach_max_dma_address = 0x00ffffff; /* default set to the lower 16MB */
-#if defined(CONFIG_AMIGA_FLOPPY) || defined(CONFIG_ATARI_FLOPPY)
-void (*mach_floppy_setup) (char *, int *) __initdata = NULL;
-#endif
 #ifdef CONFIG_HEARTBEAT
 void (*mach_heartbeat) (int);
 EXPORT_SYMBOL(mach_heartbeat);
@@ -100,6 +97,8 @@ void (*mach_beep)(unsigned int, unsigned int);
 #if defined(CONFIG_ISA) && defined(MULTI_ISA)
 int isa_type;
 int isa_sex;
+EXPORT_SYMBOL(isa_type);
+EXPORT_SYMBOL(isa_sex);
 #endif
 
 extern int amiga_parse_bootinfo(const struct bi_record *);
@@ -280,6 +279,10 @@ void __init setup_arch(char **cmdline_p)
             }
         }
 
+#ifdef CONFIG_DUMMY_CONSOLE
+        conswitchp = &dummy_con;
+#endif
+
         switch (m68k_machtype) {
 #ifdef CONFIG_AMIGA
             case MACH_AMIGA:
@@ -521,16 +524,6 @@ int get_hardware_list(char *buffer)
     return(len);
 }
 
-
-#if defined(CONFIG_AMIGA_FLOPPY) || defined(CONFIG_ATARI_FLOPPY)
-void __init floppy_setup(char *str, int *ints)
-{
-        if (mach_floppy_setup)
-                mach_floppy_setup (str, ints);
-}
-
-#endif
-
 void check_bugs(void)
 {
 #ifndef CONFIG_M68KFPU_EMU
diff --git a/arch/m68k/kernel/signal.c b/arch/m68k/kernel/signal.c
index 9c636a4c238d..866917bfa028 100644
--- a/arch/m68k/kernel/signal.c
+++ b/arch/m68k/kernel/signal.c
@@ -96,7 +96,7 @@ asmlinkage int do_sigsuspend(struct pt_regs *regs)
 asmlinkage int
 do_rt_sigsuspend(struct pt_regs *regs)
 {
-        sigset_t *unewset = (sigset_t *)regs->d1;
+        sigset_t __user *unewset = (sigset_t __user *)regs->d1;
         size_t sigsetsize = (size_t)regs->d2;
         sigset_t saveset, newset;
 
@@ -122,8 +122,8 @@ do_rt_sigsuspend(struct pt_regs *regs)
 }
 
 asmlinkage int
-sys_sigaction(int sig, const struct old_sigaction *act,
-              struct old_sigaction *oact)
+sys_sigaction(int sig, const struct old_sigaction __user *act,
+              struct old_sigaction __user *oact)
 {
         struct k_sigaction new_ka, old_ka;
         int ret;
@@ -154,7 +154,7 @@ sys_sigaction(int sig, const struct old_sigaction *act,
 }
 
 asmlinkage int
-sys_sigaltstack(const stack_t *uss, stack_t *uoss)
+sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss)
 {
         return do_sigaltstack(uss, uoss, rdusp());
 }
@@ -169,10 +169,10 @@ sys_sigaltstack(const stack_t *uss, stack_t *uoss)
 
 struct sigframe
 {
-        char *pretcode;
+        char __user *pretcode;
         int sig;
         int code;
-        struct sigcontext *psc;
+        struct sigcontext __user *psc;
         char retcode[8];
         unsigned long extramask[_NSIG_WORDS-1];
         struct sigcontext sc;
@@ -180,10 +180,10 @@ struct sigframe
 
 struct rt_sigframe
 {
-        char *pretcode;
+        char __user *pretcode;
         int sig;
-        struct siginfo *pinfo;
-        void *puc;
+        struct siginfo __user *pinfo;
+        void __user *puc;
         char retcode[8];
         struct siginfo info;
         struct ucontext uc;
@@ -248,7 +248,7 @@ out:
 #define uc_formatvec    uc_filler[FPCONTEXT_SIZE/4]
 #define uc_extra        uc_filler[FPCONTEXT_SIZE/4+1]
 
-static inline int rt_restore_fpu_state(struct ucontext *uc)
+static inline int rt_restore_fpu_state(struct ucontext __user *uc)
 {
         unsigned char fpstate[FPCONTEXT_SIZE];
         int context_size = CPU_IS_060 ? 8 : 0;
@@ -267,7 +267,7 @@ static inline int rt_restore_fpu_state(struct ucontext *uc)
                 return 0;
         }
 
-        if (__get_user(*(long *)fpstate, (long *)&uc->uc_fpstate))
+        if (__get_user(*(long *)fpstate, (long __user *)&uc->uc_fpstate))
                 goto out;
         if (CPU_IS_060 ? fpstate[2] : fpstate[0]) {
                 if (!CPU_IS_060)
@@ -306,7 +306,7 @@ static inline int rt_restore_fpu_state(struct ucontext *uc)
                                     "m" (*fpregs.f_fpcntl));
         }
         if (context_size &&
-            __copy_from_user(fpstate + 4, (long *)&uc->uc_fpstate + 1,
+            __copy_from_user(fpstate + 4, (long __user *)&uc->uc_fpstate + 1,
                              context_size))
                 goto out;
         __asm__ volatile (".chip 68k/68881\n\t"
@@ -319,7 +319,7 @@ out:
 }
 
 static inline int
-restore_sigcontext(struct pt_regs *regs, struct sigcontext *usc, void *fp,
+restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *usc, void __user *fp,
                    int *pd0)
 {
         int fsize, formatvec;
@@ -404,10 +404,10 @@ badframe:
 
 static inline int
 rt_restore_ucontext(struct pt_regs *regs, struct switch_stack *sw,
-                    struct ucontext *uc, int *pd0)
+                    struct ucontext __user *uc, int *pd0)
 {
         int fsize, temp;
-        greg_t *gregs = uc->uc_mcontext.gregs;
+        greg_t __user *gregs = uc->uc_mcontext.gregs;
         unsigned long usp;
         int err;
 
@@ -506,7 +506,7 @@ asmlinkage int do_sigreturn(unsigned long __unused)
         struct switch_stack *sw = (struct switch_stack *) &__unused;
         struct pt_regs *regs = (struct pt_regs *) (sw + 1);
         unsigned long usp = rdusp();
-        struct sigframe *frame = (struct sigframe *)(usp - 4);
+        struct sigframe __user *frame = (struct sigframe __user *)(usp - 4);
         sigset_t set;
         int d0;
 
@@ -536,7 +536,7 @@ asmlinkage int do_rt_sigreturn(unsigned long __unused)
         struct switch_stack *sw = (struct switch_stack *) &__unused;
         struct pt_regs *regs = (struct pt_regs *) (sw + 1);
         unsigned long usp = rdusp();
-        struct rt_sigframe *frame = (struct rt_sigframe *)(usp - 4);
+        struct rt_sigframe __user *frame = (struct rt_sigframe __user *)(usp - 4);
         sigset_t set;
         int d0;
 
@@ -596,7 +596,7 @@ static inline void save_fpu_state(struct sigcontext *sc, struct pt_regs *regs)
         }
 }
 
-static inline int rt_save_fpu_state(struct ucontext *uc, struct pt_regs *regs)
+static inline int rt_save_fpu_state(struct ucontext __user *uc, struct pt_regs *regs)
 {
         unsigned char fpstate[FPCONTEXT_SIZE];
         int context_size = CPU_IS_060 ? 8 : 0;
@@ -617,7 +617,7 @@ static inline int rt_save_fpu_state(struct ucontext *uc, struct pt_regs *regs)
                           ".chip 68k"
                           : : "m" (*fpstate) : "memory");
 
-        err |= __put_user(*(long *)fpstate, (long *)&uc->uc_fpstate);
+        err |= __put_user(*(long *)fpstate, (long __user *)&uc->uc_fpstate);
         if (CPU_IS_060 ? fpstate[2] : fpstate[0]) {
                 fpregset_t fpregs;
                 if (!CPU_IS_060)
@@ -642,7 +642,7 @@ static inline int rt_save_fpu_state(struct ucontext *uc, struct pt_regs *regs)
                                     sizeof(fpregs));
         }
         if (context_size)
-                err |= copy_to_user((long *)&uc->uc_fpstate + 1, fpstate + 4,
+                err |= copy_to_user((long __user *)&uc->uc_fpstate + 1, fpstate + 4,
                                     context_size);
         return err;
 }
@@ -662,10 +662,10 @@ static void setup_sigcontext(struct sigcontext *sc, struct pt_regs *regs,
         save_fpu_state(sc, regs);
 }
 
-static inline int rt_setup_ucontext(struct ucontext *uc, struct pt_regs *regs)
+static inline int rt_setup_ucontext(struct ucontext __user *uc, struct pt_regs *regs)
 {
         struct switch_stack *sw = (struct switch_stack *)regs - 1;
-        greg_t *gregs = uc->uc_mcontext.gregs;
+        greg_t __user *gregs = uc->uc_mcontext.gregs;
         int err = 0;
 
         err |= __put_user(MCONTEXT_VERSION, &uc->uc_mcontext.version);
@@ -753,7 +753,7 @@ static inline void push_cache (unsigned long vaddr)
         }
 }
 
-static inline void *
+static inline void __user *
 get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size)
 {
         unsigned long usp;
@@ -766,13 +766,13 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size)
                 if (!on_sig_stack(usp))
                         usp = current->sas_ss_sp + current->sas_ss_size;
         }
-        return (void *)((usp - frame_size) & -8UL);
+        return (void __user *)((usp - frame_size) & -8UL);
 }
 
 static void setup_frame (int sig, struct k_sigaction *ka,
                          sigset_t *set, struct pt_regs *regs)
 {
-        struct sigframe *frame;
+        struct sigframe __user *frame;
         int fsize = frame_extra_sizes[regs->format];
         struct sigcontext context;
         int err = 0;
@@ -813,7 +813,7 @@ static void setup_frame (int sig, struct k_sigaction *ka,
         err |= __put_user(frame->retcode, &frame->pretcode);
         /* moveq #,d0; trap #0 */
         err |= __put_user(0x70004e40 + (__NR_sigreturn << 16),
-                          (long *)(frame->retcode));
+                          (long __user *)(frame->retcode));
 
         if (err)
                 goto give_sigsegv;
@@ -849,7 +849,7 @@ give_sigsegv:
 static void setup_rt_frame (int sig, struct k_sigaction *ka, siginfo_t *info,
                             sigset_t *set, struct pt_regs *regs)
 {
-        struct rt_sigframe *frame;
+        struct rt_sigframe __user *frame;
         int fsize = frame_extra_sizes[regs->format];
         int err = 0;
 
@@ -880,8 +880,8 @@ static void setup_rt_frame (int sig, struct k_sigaction *ka, siginfo_t *info,
 
         /* Create the ucontext.  */
         err |= __put_user(0, &frame->uc.uc_flags);
-        err |= __put_user(0, &frame->uc.uc_link);
-        err |= __put_user((void *)current->sas_ss_sp,
+        err |= __put_user(NULL, &frame->uc.uc_link);
+        err |= __put_user((void __user *)current->sas_ss_sp,
                           &frame->uc.uc_stack.ss_sp);
         err |= __put_user(sas_ss_flags(rdusp()),
                           &frame->uc.uc_stack.ss_flags);
@@ -893,8 +893,8 @@ static void setup_rt_frame (int sig, struct k_sigaction *ka, siginfo_t *info,
         err |= __put_user(frame->retcode, &frame->pretcode);
         /* moveq #,d0; notb d0; trap #0 */
         err |= __put_user(0x70004600 + ((__NR_rt_sigreturn ^ 0xff) << 16),
-                          (long *)(frame->retcode + 0));
-        err |= __put_user(0x4e40, (short *)(frame->retcode + 4));
+                          (long __user *)(frame->retcode + 0));
+        err |= __put_user(0x4e40, (short __user *)(frame->retcode + 4));
 
         if (err)
                 goto give_sigsegv;
diff --git a/arch/m68k/kernel/sys_m68k.c b/arch/m68k/kernel/sys_m68k.c
index 640895b2c51a..143c552d38f3 100644
--- a/arch/m68k/kernel/sys_m68k.c
+++ b/arch/m68k/kernel/sys_m68k.c
@@ -32,7 +32,7 @@
  * sys_pipe() is the normal C calling standard for creating
  * a pipe. It's not the way unix traditionally does this, though.
  */
-asmlinkage int sys_pipe(unsigned long * fildes)
+asmlinkage int sys_pipe(unsigned long __user * fildes)
 {
         int fd[2];
         int error;
@@ -94,7 +94,7 @@ struct mmap_arg_struct {
         unsigned long offset;
 };
 
-asmlinkage int old_mmap(struct mmap_arg_struct *arg)
+asmlinkage int old_mmap(struct mmap_arg_struct __user *arg)
 {
         struct mmap_arg_struct a;
         int error = -EFAULT;
@@ -160,11 +160,11 @@ out:
 
 struct sel_arg_struct {
         unsigned long n;
-        fd_set *inp, *outp, *exp;
-        struct timeval *tvp;
+        fd_set __user *inp, *outp, *exp;
+        struct timeval __user *tvp;
 };
 
-asmlinkage int old_select(struct sel_arg_struct *arg)
+asmlinkage int old_select(struct sel_arg_struct __user *arg)
 {
         struct sel_arg_struct a;
 
@@ -180,7 +180,7 @@ asmlinkage int old_select(struct sel_arg_struct *arg)
  * This is really horribly ugly.
  */
 asmlinkage int sys_ipc (uint call, int first, int second,
-                        int third, void *ptr, long fifth)
+                        int third, void __user *ptr, long fifth)
 {
         int version, ret;
 
@@ -190,14 +190,14 @@ asmlinkage int sys_ipc (uint call, int first, int second,
         if (call <= SEMCTL)
                 switch (call) {
                 case SEMOP:
-                        return sys_semop (first, (struct sembuf *)ptr, second);
+                        return sys_semop (first, ptr, second);
                 case SEMGET:
                         return sys_semget (first, second, third);
                 case SEMCTL: {
                         union semun fourth;
                         if (!ptr)
                                 return -EINVAL;
-                        if (get_user(fourth.__pad, (void **) ptr))
+                        if (get_user(fourth.__pad, (void __user *__user *) ptr))
                                 return -EFAULT;
                         return sys_semctl (first, second, third, fourth);
                         }
@@ -207,31 +207,26 @@ asmlinkage int sys_ipc (uint call, int first, int second,
         if (call <= MSGCTL)
                 switch (call) {
                 case MSGSND:
-                        return sys_msgsnd (first, (struct msgbuf *) ptr,
-                                          second, third);
+                        return sys_msgsnd (first, ptr, second, third);
                 case MSGRCV:
                         switch (version) {
                         case 0: {
                                 struct ipc_kludge tmp;
                                 if (!ptr)
                                         return -EINVAL;
-                                if (copy_from_user (&tmp,
-                                                    (struct ipc_kludge *)ptr,
-                                                    sizeof (tmp)))
+                                if (copy_from_user (&tmp, ptr, sizeof (tmp)))
                                         return -EFAULT;
                                 return sys_msgrcv (first, tmp.msgp, second,
                                                    tmp.msgtyp, third);
                                 }
                         default:
-                                return sys_msgrcv (first,
-                                                   (struct msgbuf *) ptr,
+                                return sys_msgrcv (first, ptr,
                                                    second, fifth, third);
                         }
                 case MSGGET:
                         return sys_msgget ((key_t) first, second);
                 case MSGCTL:
-                        return sys_msgctl (first, second,
-                                           (struct msqid_ds *) ptr);
+                        return sys_msgctl (first, second, ptr);
                 default:
                         return -ENOSYS;
                 }
@@ -241,20 +236,18 @@ asmlinkage int sys_ipc (uint call, int first, int second,
                         switch (version) {
                         default: {
                                 ulong raddr;
-                                ret = do_shmat (first, (char *) ptr,
-                                                 second, &raddr);
+                                ret = do_shmat (first, ptr, second, &raddr);
                                 if (ret)
                                         return ret;
-                                return put_user (raddr, (ulong *) third);
+                                return put_user (raddr, (ulong __user *) third);
                         }
                         }
                 case SHMDT:
-                        return sys_shmdt ((char *)ptr);
+                        return sys_shmdt (ptr);
                 case SHMGET:
                         return sys_shmget (first, second, third);
                 case SHMCTL:
-                        return sys_shmctl (first, second,
-                                           (struct shmid_ds *) ptr);
+                        return sys_shmctl (first, second, ptr);
                 default:
                         return -ENOSYS;
                 }
diff --git a/arch/m68k/kernel/traps.c b/arch/m68k/kernel/traps.c
index deb36e8b04a2..cdf58fbb3e73 100644
--- a/arch/m68k/kernel/traps.c
+++ b/arch/m68k/kernel/traps.c
@@ -169,25 +169,25 @@ void __init trap_init (void)
 
         if (CPU_IS_060 && !FPU_IS_EMU) {
                 /* set up IFPSP entry points */
-                asmlinkage void snan_vec(void) asm ("_060_fpsp_snan");
-                asmlinkage void operr_vec(void) asm ("_060_fpsp_operr");
-                asmlinkage void ovfl_vec(void) asm ("_060_fpsp_ovfl");
-                asmlinkage void unfl_vec(void) asm ("_060_fpsp_unfl");
-                asmlinkage void dz_vec(void) asm ("_060_fpsp_dz");
-                asmlinkage void inex_vec(void) asm ("_060_fpsp_inex");
-                asmlinkage void fline_vec(void) asm ("_060_fpsp_fline");
-                asmlinkage void unsupp_vec(void) asm ("_060_fpsp_unsupp");
-                asmlinkage void effadd_vec(void) asm ("_060_fpsp_effadd");
-
-                vectors[VEC_FPNAN] = snan_vec;
-                vectors[VEC_FPOE] = operr_vec;
-                vectors[VEC_FPOVER] = ovfl_vec;
-                vectors[VEC_FPUNDER] = unfl_vec;
-                vectors[VEC_FPDIVZ] = dz_vec;
-                vectors[VEC_FPIR] = inex_vec;
-                vectors[VEC_LINE11] = fline_vec;
-                vectors[VEC_FPUNSUP] = unsupp_vec;
-                vectors[VEC_UNIMPEA] = effadd_vec;
+                asmlinkage void snan_vec6(void) asm ("_060_fpsp_snan");
+                asmlinkage void operr_vec6(void) asm ("_060_fpsp_operr");
+                asmlinkage void ovfl_vec6(void) asm ("_060_fpsp_ovfl");
+                asmlinkage void unfl_vec6(void) asm ("_060_fpsp_unfl");
+                asmlinkage void dz_vec6(void) asm ("_060_fpsp_dz");
+                asmlinkage void inex_vec6(void) asm ("_060_fpsp_inex");
+                asmlinkage void fline_vec6(void) asm ("_060_fpsp_fline");
+                asmlinkage void unsupp_vec6(void) asm ("_060_fpsp_unsupp");
+                asmlinkage void effadd_vec6(void) asm ("_060_fpsp_effadd");
+
+                vectors[VEC_FPNAN] = snan_vec6;
+                vectors[VEC_FPOE] = operr_vec6;
+                vectors[VEC_FPOVER] = ovfl_vec6;
+                vectors[VEC_FPUNDER] = unfl_vec6;
+                vectors[VEC_FPDIVZ] = dz_vec6;
+                vectors[VEC_FPIR] = inex_vec6;
+                vectors[VEC_LINE11] = fline_vec6;
+                vectors[VEC_FPUNSUP] = unsupp_vec6;
+                vectors[VEC_UNIMPEA] = effadd_vec6;
         }
 
         /* if running on an amiga, make the NMI interrupt do nothing */
diff --git a/arch/m68k/kernel/vmlinux-std.lds b/arch/m68k/kernel/vmlinux-std.lds
index e58654f3f8dd..69d1d3d30c78 100644
--- a/arch/m68k/kernel/vmlinux-std.lds
+++ b/arch/m68k/kernel/vmlinux-std.lds
@@ -13,6 +13,7 @@ SECTIONS
   .text : {
         *(.text)
         SCHED_TEXT
+        LOCK_TEXT
         *(.fixup)
         *(.gnu.warning)
         } :text = 0x4e75
diff --git a/arch/m68k/kernel/vmlinux-sun3.lds b/arch/m68k/kernel/vmlinux-sun3.lds
index cc37e8d3c1e2..65cc39c24185 100644
--- a/arch/m68k/kernel/vmlinux-sun3.lds
+++ b/arch/m68k/kernel/vmlinux-sun3.lds
@@ -14,6 +14,7 @@ SECTIONS
         *(.head)
         *(.text)
         SCHED_TEXT
+        LOCK_TEXT
         *(.fixup)
         *(.gnu.warning)
         } :text = 0x4e75
@@ -66,7 +67,7 @@ __init_begin = .;
         __initramfs_end = .;
         . = ALIGN(8192);
         __init_end = .;
-        .init.task : { *(init_task) }
+        .data.init.task : { *(.data.init_task) }
 
 
   .bss : { *(.bss) }            /* BSS */
diff --git a/arch/m68k/lib/checksum.c b/arch/m68k/lib/checksum.c
index 4a5c5445c610..cb13c6e3ccae 100644
--- a/arch/m68k/lib/checksum.c
+++ b/arch/m68k/lib/checksum.c
@@ -134,7 +134,7 @@ EXPORT_SYMBOL(csum_partial);
  */
 
 unsigned int
-csum_partial_copy_from_user(const unsigned char *src, unsigned char *dst,
+csum_partial_copy_from_user(const unsigned char __user *src, unsigned char *dst,
                             int len, int sum, int *csum_err)
 {
         /*
diff --git a/arch/m68k/mac/config.c b/arch/m68k/mac/config.c
index cd19cbb213e8..14f8d3f4e195 100644
--- a/arch/m68k/mac/config.c
+++ b/arch/m68k/mac/config.c
@@ -212,9 +212,6 @@ void __init config_mac(void)
         mach_reset           = mac_reset;
         mach_halt            = mac_poweroff;
         mach_power_off       = mac_poweroff;
-#ifdef CONFIG_DUMMY_CONSOLE
-        conswitchp               = &dummy_con;
-#endif
         mach_max_dma_address = 0xffffffff;
 #if 0
         mach_debug_init  = mac_debug_init;
diff --git a/arch/m68k/mac/iop.c b/arch/m68k/mac/iop.c
index d889ba80ccdc..9179a3798407 100644
--- a/arch/m68k/mac/iop.c
+++ b/arch/m68k/mac/iop.c
@@ -293,8 +293,8 @@ void __init iop_init(void)
         }
 
         for (i = 0 ; i < NUM_IOP_CHAN ; i++) {
-                iop_send_queue[IOP_NUM_SCC][i] = 0;
-                iop_send_queue[IOP_NUM_ISM][i] = 0;
+                iop_send_queue[IOP_NUM_SCC][i] = NULL;
+                iop_send_queue[IOP_NUM_ISM][i] = NULL;
                 iop_listeners[IOP_NUM_SCC][i].devname = NULL;
                 iop_listeners[IOP_NUM_SCC][i].handler = NULL;
                 iop_listeners[IOP_NUM_ISM][i].devname = NULL;
diff --git a/arch/m68k/mac/misc.c b/arch/m68k/mac/misc.c
index 5b80d7cd954a..bbb0c3b95e9c 100644
--- a/arch/m68k/mac/misc.c
+++ b/arch/m68k/mac/misc.c
@@ -39,72 +39,163 @@
 extern struct mac_booter_data mac_bi_data;
 static void (*rom_reset)(void);
 
-#ifdef CONFIG_ADB
-/*
- * Return the current time as the number of seconds since January 1, 1904.
- */
-
-static long adb_read_time(void)
+#ifdef CONFIG_ADB_CUDA
+static long cuda_read_time(void)
 {
-        volatile struct adb_request req;
+        struct adb_request req;
         long time;
 
-        adb_request((struct adb_request *) &req, NULL,
-                        ADBREQ_RAW|ADBREQ_SYNC,
-                        2, CUDA_PACKET, CUDA_GET_TIME);
+        if (cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_GET_TIME) < 0)
+                return 0;
+        while (!req.complete)
+                cuda_poll();
 
         time = (req.reply[3] << 24) | (req.reply[4] << 16)
                 | (req.reply[5] << 8) | req.reply[6];
         return time - RTC_OFFSET;
 }
 
-/*
- * Set the current system time
- */
+static void cuda_write_time(long data)
+{
+        struct adb_request req;
+        data += RTC_OFFSET;
+        if (cuda_request(&req, NULL, 6, CUDA_PACKET, CUDA_SET_TIME,
+                        (data >> 24) & 0xFF, (data >> 16) & 0xFF,
+                        (data >> 8) & 0xFF, data & 0xFF) < 0)
+                return;
+        while (!req.complete)
+                cuda_poll();
+}
 
-static void adb_write_time(long data)
+static __u8 cuda_read_pram(int offset)
 {
-        volatile struct adb_request req;
+        struct adb_request req;
+        if (cuda_request(&req, NULL, 4, CUDA_PACKET, CUDA_GET_PRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF) < 0)
+                return 0;
+        while (!req.complete)
+                cuda_poll();
+        return req.reply[3];
+}
 
-        data += RTC_OFFSET;
+static void cuda_write_pram(int offset, __u8 data)
+{
+        struct adb_request req;
+        if (cuda_request(&req, NULL, 5, CUDA_PACKET, CUDA_SET_PRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF, data) < 0)
+                return;
+        while (!req.complete)
+                cuda_poll();
+}
+#else
+#define cuda_read_time() 0
+#define cuda_write_time(n)
+#define cuda_read_pram NULL
+#define cuda_write_pram NULL
+#endif
+
+#ifdef CONFIG_ADB_PMU68K
+static long pmu_read_time(void)
+{
+        struct adb_request req;
+        long time;
+
+        if (pmu_request(&req, NULL, 1, PMU_READ_RTC) < 0)
+                return 0;
+        while (!req.complete)
+                pmu_poll();
 
-        adb_request((struct adb_request *) &req, NULL,
-                        ADBREQ_RAW|ADBREQ_SYNC,
-                        6, CUDA_PACKET, CUDA_SET_TIME,
+        time = (req.reply[0] << 24) | (req.reply[1] << 16)
+                | (req.reply[2] << 8) | req.reply[3];
+        return time - RTC_OFFSET;
+}
+
+static void pmu_write_time(long data)
+{
+        struct adb_request req;
+        data += RTC_OFFSET;
+        if (pmu_request(&req, NULL, 5, PMU_SET_RTC,
                         (data >> 24) & 0xFF, (data >> 16) & 0xFF,
-                        (data >> 8) & 0xFF, data & 0xFF);
+                        (data >> 8) & 0xFF, data & 0xFF) < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
 }
 
-/*
- * Get a byte from the NVRAM
- */
+static __u8 pmu_read_pram(int offset)
+{
+        struct adb_request req;
+        if (pmu_request(&req, NULL, 3, PMU_READ_NVRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF) < 0)
+                return 0;
+        while (!req.complete)
+                pmu_poll();
+        return req.reply[3];
+}
 
-static __u8 adb_read_pram(int offset)
+static void pmu_write_pram(int offset, __u8 data)
 {
-        volatile struct adb_request req;
+        struct adb_request req;
+        if (pmu_request(&req, NULL, 4, PMU_WRITE_NVRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF, data) < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
+}
+#else
+#define pmu_read_time() 0
+#define pmu_write_time(n)
+#define pmu_read_pram NULL
+#define pmu_write_pram NULL
+#endif
 
-        adb_request((struct adb_request *) &req, NULL,
-                        ADBREQ_RAW|ADBREQ_SYNC,
-                        4, CUDA_PACKET, CUDA_GET_PRAM,
-                        (offset >> 8) & 0xFF, offset & 0xFF);
-        return req.reply[3];
+#ifdef CONFIG_ADB_MACIISI
+extern int maciisi_request(struct adb_request *req,
+                        void (*done)(struct adb_request *), int nbytes, ...);
+
+static long maciisi_read_time(void)
+{
+        struct adb_request req;
+        long time;
+
+        if (maciisi_request(&req, NULL, 2, CUDA_PACKET, CUDA_GET_TIME))
+                return 0;
+
+        time = (req.reply[3] << 24) | (req.reply[4] << 16)
+                | (req.reply[5] << 8) | req.reply[6];
+        return time - RTC_OFFSET;
 }
 
-/*
- * Write a byte to the NVRAM
- */
+static void maciisi_write_time(long data)
+{
+        struct adb_request req;
+        data += RTC_OFFSET;
+        maciisi_request(&req, NULL, 6, CUDA_PACKET, CUDA_SET_TIME,
+                        (data >> 24) & 0xFF, (data >> 16) & 0xFF,
+                        (data >> 8) & 0xFF, data & 0xFF);
+}
 
-static void adb_write_pram(int offset, __u8 data)
+static __u8 maciisi_read_pram(int offset)
 {
-        volatile struct adb_request req;
+        struct adb_request req;
+        if (maciisi_request(&req, NULL, 4, CUDA_PACKET, CUDA_GET_PRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF))
+                return 0;
+        return req.reply[3];
+}
 
-        adb_request((struct adb_request *) &req, NULL,
-                        ADBREQ_RAW|ADBREQ_SYNC,
-                        5, CUDA_PACKET, CUDA_SET_PRAM,
-                        (offset >> 8) & 0xFF, offset & 0xFF,
-                        data);
+static void maciisi_write_pram(int offset, __u8 data)
+{
+        struct adb_request req;
+        maciisi_request(&req, NULL, 5, CUDA_PACKET, CUDA_SET_PRAM,
+                        (offset >> 8) & 0xFF, offset & 0xFF, data);
 }
-#endif /* CONFIG_ADB */
+#else
+#define maciisi_read_time() 0
+#define maciisi_write_time(n)
+#define maciisi_read_pram NULL
+#define maciisi_write_pram NULL
+#endif
 
 /*
  * VIA PRAM/RTC access routines
@@ -305,42 +396,55 @@ static void oss_shutdown(void)
 
 static void cuda_restart(void)
 {
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        2, CUDA_PACKET, CUDA_RESET_SYSTEM);
+        struct adb_request req;
+        if (cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_RESET_SYSTEM) < 0)
+                return;
+        while (!req.complete)
+                cuda_poll();
 }
 
 static void cuda_shutdown(void)
 {
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        2, CUDA_PACKET, CUDA_POWERDOWN);
+        struct adb_request req;
+        if (cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_POWERDOWN) < 0)
+                return;
+        while (!req.complete)
+                cuda_poll();
 }
 
 #endif /* CONFIG_ADB_CUDA */
 
-#ifdef CONFIG_ADB_PMU
+#ifdef CONFIG_ADB_PMU68K
 
 void pmu_restart(void)
 {
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        3, PMU_PACKET, PMU_SET_INTR_MASK,
-                        PMU_INT_ADB|PMU_INT_TICK);
-
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        2, PMU_PACKET, PMU_RESET);
+        struct adb_request req;
+        if (pmu_request(&req, NULL,
+                        2, PMU_SET_INTR_MASK, PMU_INT_ADB|PMU_INT_TICK) < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
+        if (pmu_request(&req, NULL, 1, PMU_RESET) < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
 }
 
 void pmu_shutdown(void)
 {
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        3, PMU_PACKET, PMU_SET_INTR_MASK,
-                        PMU_INT_ADB|PMU_INT_TICK);
-
-        adb_request(NULL, NULL, ADBREQ_RAW|ADBREQ_SYNC,
-                        6, PMU_PACKET, PMU_SHUTDOWN,
-                        'M', 'A', 'T', 'T');
+        struct adb_request req;
+        if (pmu_request(&req, NULL,
+                        2, PMU_SET_INTR_MASK, PMU_INT_ADB|PMU_INT_TICK) < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
+        if (pmu_request(&req, NULL, 5, PMU_SHUTDOWN, 'M', 'A', 'T', 'T') < 0)
+                return;
+        while (!req.complete)
+                pmu_poll();
 }
 
-#endif /* CONFIG_ADB_PMU */
+#endif
 
 /*
  *-------------------------------------------------------------------
@@ -351,21 +455,22 @@ void pmu_shutdown(void)
 
 void mac_pram_read(int offset, __u8 *buffer, int len)
 {
-        __u8 (*func)(int) = NULL;
+        __u8 (*func)(int);
         int i;
 
-        if (macintosh_config->adb_type == MAC_ADB_IISI ||
-            macintosh_config->adb_type == MAC_ADB_PB1 ||
-            macintosh_config->adb_type == MAC_ADB_PB2 ||
-            macintosh_config->adb_type == MAC_ADB_CUDA) {
-#ifdef CONFIG_ADB
-                func = adb_read_pram;
-#else
-                return;
-#endif
-        } else {
+        switch(macintosh_config->adb_type) {
+        case MAC_ADB_IISI:
+                func = maciisi_read_pram; break;
+        case MAC_ADB_PB1:
+        case MAC_ADB_PB2:
+                func = pmu_read_pram; break;
+        case MAC_ADB_CUDA:
+                func = cuda_read_pram; break;
+        default:
                 func = via_read_pram;
         }
+        if (!func)
+                return;
         for (i = 0 ; i < len ; i++) {
                 buffer[i] = (*func)(offset++);
         }
@@ -373,21 +478,22 @@ void mac_pram_read(int offset, __u8 *buffer, int len)
 
 void mac_pram_write(int offset, __u8 *buffer, int len)
 {
-        void (*func)(int, __u8) = NULL;
+        void (*func)(int, __u8);
         int i;
 
-        if (macintosh_config->adb_type == MAC_ADB_IISI ||
-            macintosh_config->adb_type == MAC_ADB_PB1 ||
-            macintosh_config->adb_type == MAC_ADB_PB2 ||
-            macintosh_config->adb_type == MAC_ADB_CUDA) {
-#ifdef CONFIG_ADB
-                func = adb_write_pram;
-#else
-                return;
-#endif
-        } else {
+        switch(macintosh_config->adb_type) {
+        case MAC_ADB_IISI:
+                func = maciisi_write_pram; break;
+        case MAC_ADB_PB1:
+        case MAC_ADB_PB2:
+                func = pmu_write_pram; break;
+        case MAC_ADB_CUDA:
+                func = cuda_write_pram; break;
+        default:
                 func = via_write_pram;
         }
+        if (!func)
+                return;
         for (i = 0 ; i < len ; i++) {
                 (*func)(offset++, buffer[i]);
         }
@@ -408,7 +514,7 @@ void mac_poweroff(void)
         } else if (macintosh_config->adb_type == MAC_ADB_CUDA) {
                 cuda_shutdown();
 #endif
-#ifdef CONFIG_ADB_PMU
+#ifdef CONFIG_ADB_PMU68K
         } else if (macintosh_config->adb_type == MAC_ADB_PB1
                 || macintosh_config->adb_type == MAC_ADB_PB2) {
                 pmu_shutdown();
@@ -448,7 +554,7 @@ void mac_reset(void)
         } else if (macintosh_config->adb_type == MAC_ADB_CUDA) {
                 cuda_restart();
 #endif
-#ifdef CONFIG_ADB_PMU
+#ifdef CONFIG_ADB_PMU68K
         } else if (macintosh_config->adb_type == MAC_ADB_PB1
                 || macintosh_config->adb_type == MAC_ADB_PB2) {
                 pmu_restart();
@@ -466,12 +572,13 @@ void mac_reset(void)
                 /* make a 1-to-1 mapping, using the transparent tran. reg. */
                 unsigned long virt = (unsigned long) mac_reset;
                 unsigned long phys = virt_to_phys(mac_reset);
+                unsigned long addr = (phys&0xFF000000)|0x8777;
                 unsigned long offset = phys-virt;
                 local_irq_disable(); /* lets not screw this up, ok? */
                 __asm__ __volatile__(".chip 68030\n\t"
                                      "pmove %0,%/tt0\n\t"
                                      ".chip 68k"
-                                     : : "m" ((phys&0xFF000000)|0x8777));
+                                     : : "m" (addr));
                 /* Now jump to physical address so we can disable MMU */
                 __asm__ __volatile__(
                     ".chip 68030\n\t"
@@ -588,20 +695,22 @@ int mac_hwclk(int op, struct rtc_time *t)
         unsigned long now;
 
         if (!op) { /* read */
-                if (macintosh_config->adb_type == MAC_ADB_II) {
+                switch (macintosh_config->adb_type) {
+                case MAC_ADB_II:
+                case MAC_ADB_IOP:
                         now = via_read_time();
-                } else
-#ifdef CONFIG_ADB
-                if ((macintosh_config->adb_type == MAC_ADB_IISI) ||
-                           (macintosh_config->adb_type == MAC_ADB_PB1) ||
-                           (macintosh_config->adb_type == MAC_ADB_PB2) ||
-                           (macintosh_config->adb_type == MAC_ADB_CUDA)) {
-                        now = adb_read_time();
-                } else
-#endif
-                if (macintosh_config->adb_type == MAC_ADB_IOP) {
-                        now = via_read_time();
-                } else {
+                        break;
+                case MAC_ADB_IISI:
+                        now = maciisi_read_time();
+                        break;
+                case MAC_ADB_PB1:
+                case MAC_ADB_PB2:
+                        now = pmu_read_time();
+                        break;
+                case MAC_ADB_CUDA:
+                        now = cuda_read_time();
+                        break;
+                default:
                         now = 0;
                 }
 
@@ -619,15 +728,20 @@ int mac_hwclk(int op, struct rtc_time *t)
                 now = mktime(t->tm_year + 1900, t->tm_mon + 1, t->tm_mday,
                              t->tm_hour, t->tm_min, t->tm_sec);
 
-                if (macintosh_config->adb_type == MAC_ADB_II) {
-                        via_write_time(now);
-                } else if ((macintosh_config->adb_type == MAC_ADB_IISI) ||
-                           (macintosh_config->adb_type == MAC_ADB_PB1) ||
-                           (macintosh_config->adb_type == MAC_ADB_PB2) ||
-                           (macintosh_config->adb_type == MAC_ADB_CUDA)) {
-                        adb_write_time(now);
-                } else if (macintosh_config->adb_type == MAC_ADB_IOP) {
+                switch (macintosh_config->adb_type) {
+                case MAC_ADB_II:
+                case MAC_ADB_IOP:
                         via_write_time(now);
+                        break;
+                case MAC_ADB_CUDA:
+                        cuda_write_time(now);
+                        break;
+                case MAC_ADB_PB1:
+                case MAC_ADB_PB2:
+                        pmu_write_time(now);
+                        break;
+                case MAC_ADB_IISI:
+                        maciisi_write_time(now);
                 }
 #endif
         }
diff --git a/arch/m68k/math-emu/multi_arith.h b/arch/m68k/math-emu/multi_arith.h
index 02251e5afd89..4ad0ca918e2e 100644
--- a/arch/m68k/math-emu/multi_arith.h
+++ b/arch/m68k/math-emu/multi_arith.h
@@ -366,7 +366,7 @@ static inline void fp_submant(struct fp_ext *dest, struct fp_ext *src1,
 
 #define fp_mul64(desth, destl, src1, src2) ({                           \
         asm ("mulu.l %2,%1:%0" : "=d" (destl), "=d" (desth)             \
-                : "g" (src1), "0" (src2));                              \
+                : "dm" (src1), "0" (src2));                             \
 })
 #define fp_div64(quot, rem, srch, srcl, div)                            \
         asm ("divu.l %2,%1:%0" : "=d" (quot), "=d" (rem)                \
diff --git a/arch/m68k/mm/kmap.c b/arch/m68k/mm/kmap.c
index fe2383e36b06..85ad19a0ac79 100644
--- a/arch/m68k/mm/kmap.c
+++ b/arch/m68k/mm/kmap.c
@@ -102,7 +102,7 @@ static inline void free_io_area(void *addr)
  */
 /* Rewritten by Andreas Schwab to remove all races. */
 
-void *__ioremap(unsigned long physaddr, unsigned long size, int cacheflag)
+void __iomem *__ioremap(unsigned long physaddr, unsigned long size, int cacheflag)
 {
         struct vm_struct *area;
         unsigned long virtaddr, retaddr;
@@ -121,7 +121,7 @@ void *__ioremap(unsigned long physaddr, unsigned long size, int cacheflag)
         if (MACH_IS_AMIGA) {
                 if ((physaddr >= 0x40000000) && (physaddr + size < 0x60000000)
                     && (cacheflag == IOMAP_NOCACHE_SER))
-                        return (void *)physaddr;
+                        return (void __iomem *)physaddr;
         }
 #endif
 
@@ -218,21 +218,21 @@ void *__ioremap(unsigned long physaddr, unsigned long size, int cacheflag)
 #endif
         flush_tlb_all();
 
-        return (void *)retaddr;
+        return (void __iomem *)retaddr;
 }
 
 /*
  * Unmap a ioremap()ed region again
  */
-void iounmap(void *addr)
+void iounmap(void __iomem *addr)
 {
 #ifdef CONFIG_AMIGA
         if ((!MACH_IS_AMIGA) ||
             (((unsigned long)addr < 0x40000000) ||
              ((unsigned long)addr > 0x60000000)))
-                        free_io_area(addr);
+                        free_io_area((__force void *)addr);
 #else
-        free_io_area(addr);
+        free_io_area((__force void *)addr);
 #endif
 }
 
diff --git a/arch/m68k/mvme16x/rtc.c b/arch/m68k/mvme16x/rtc.c
index 30f5921ece9b..a69fe3048edc 100644
--- a/arch/m68k/mvme16x/rtc.c
+++ b/arch/m68k/mvme16x/rtc.c
@@ -45,6 +45,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
         volatile MK48T08ptr_t rtc = (MK48T08ptr_t)MVME_RTC_BASE;
         unsigned long flags;
         struct rtc_time wtime;
+        void __user *argp = (void __user *)arg;
 
         switch (cmd) {
         case RTC_RD_TIME:       /* Read the time/date from RTC  */
@@ -64,7 +65,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
                 wtime.tm_wday = BCD2BIN(rtc->bcd_dow)-1;
                 rtc->ctrl = 0;
                 local_irq_restore(flags);
-                return copy_to_user((void *)arg, &wtime, sizeof wtime) ?
+                return copy_to_user(argp, &wtime, sizeof wtime) ?
                                                                 -EFAULT : 0;
         }
         case RTC_SET_TIME:      /* Set the RTC */
@@ -76,8 +77,7 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
                 if (!capable(CAP_SYS_ADMIN))
                         return -EACCES;
 
-                if (copy_from_user(&rtc_tm, (struct rtc_time*)arg,
-                                   sizeof(struct rtc_time)))
+                if (copy_from_user(&rtc_tm, argp, sizeof(struct rtc_time)))
                         return -EFAULT;
 
                 yrs = rtc_tm.tm_year;
diff --git a/arch/m68k/q40/config.c b/arch/m68k/q40/config.c
index 02b626bae4ae..5e0f9b04d45e 100644
--- a/arch/m68k/q40/config.c
+++ b/arch/m68k/q40/config.c
@@ -36,8 +36,6 @@
 #include <asm/machdep.h>
 #include <asm/q40_master.h>
 
-extern void floppy_setup(char *str, int *ints);
-
 extern irqreturn_t q40_process_int (int level, struct pt_regs *regs);
 extern irqreturn_t (*q40_default_handler[]) (int, void *, struct pt_regs *);  /* added just for debugging */
 extern void q40_init_IRQ (void);
@@ -194,9 +192,6 @@ void __init config_q40(void)
     mach_heartbeat = q40_heartbeat;
 #endif
     mach_halt = q40_halt;
-#ifdef CONFIG_DUMMY_CONSOLE
-    conswitchp = &dummy_con;
-#endif
 
     /* disable a few things that SMSQ might have left enabled */
     q40_disable_irqs();
diff --git a/arch/m68k/sun3/config.c b/arch/m68k/sun3/config.c
index 77d05bcc3221..f1ca0dfbaa67 100644
--- a/arch/m68k/sun3/config.c
+++ b/arch/m68k/sun3/config.c
@@ -160,9 +160,6 @@ void __init config_sun3(void)
         mach_hwclk           =  sun3_hwclk;
         mach_halt            =  sun3_halt;
         mach_get_hardware_list = sun3_get_hardware_list;
-#if defined(CONFIG_DUMMY_CONSOLE)
-        conswitchp           = &dummy_con;
-#endif
 
         memory_start = ((((int)&_end) + 0x2000) & ~0x1fff);
 // PROM seems to want the last couple of physical pages. --m
diff --git a/arch/m68k/sun3x/config.c b/arch/m68k/sun3x/config.c
index 0ef547f5494d..0920f5d33606 100644
--- a/arch/m68k/sun3x/config.c
+++ b/arch/m68k/sun3x/config.c
@@ -71,10 +71,6 @@ void __init config_sun3x(void)
         mach_get_model       = sun3_get_model;
         mach_get_hardware_list = sun3x_get_hardware_list;
 
-#ifdef CONFIG_DUMMY_CONSOLE
-        conswitchp           = &dummy_con;
-#endif
-
         sun3_intreg = (unsigned char *)SUN3X_INTREG;
 
         /* only the serial console is known to work anyway... */
diff --git a/arch/m68knommu/kernel/process.c b/arch/m68knommu/kernel/process.c
index 8b3cf57ba706..99bf43824795 100644
--- a/arch/m68knommu/kernel/process.c
+++ b/arch/m68knommu/kernel/process.c
@@ -198,10 +198,9 @@ int copy_thread(int nr, unsigned long clone_flags,
 {
         struct pt_regs * childregs;
         struct switch_stack * childstack, *stack;
-        unsigned long stack_offset, *retp;
+        unsigned long *retp;
 
-        stack_offset = THREAD_SIZE - sizeof(struct pt_regs);
-        childregs = (struct pt_regs *) ((unsigned long) p->thread_info + stack_offset);
+        childregs = (struct pt_regs *) (task_stack_page(p) + THREAD_SIZE) - 1;
 
         *childregs = *regs;
         childregs->d0 = 0;
diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c
index 0476a4dce14e..fa98f10d0132 100644
--- a/arch/mips/kernel/process.c
+++ b/arch/mips/kernel/process.c
@@ -140,12 +140,12 @@ void flush_thread(void)
 int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
         unsigned long unused, struct task_struct *p, struct pt_regs *regs)
 {
-        struct thread_info *ti = p->thread_info;
+        struct thread_info *ti = task_thread_info(p);
         struct pt_regs *childregs;
         long childksp;
         p->set_child_tid = p->clear_child_tid = NULL;
 
-        childksp = (unsigned long)ti + THREAD_SIZE - 32;
+        childksp = (unsigned long)task_stack_page(p) + THREAD_SIZE - 32;
 
         preempt_disable();
 
@@ -229,9 +229,7 @@ void elf_dump_regs(elf_greg_t *gp, struct pt_regs *regs)
 
 int dump_task_regs (struct task_struct *tsk, elf_gregset_t *regs)
 {
-        struct thread_info *ti = tsk->thread_info;
-        long ksp = (unsigned long)ti + THREAD_SIZE - 32;
-        elf_dump_regs(&(*regs)[0], (struct pt_regs *) ksp - 1);
+        elf_dump_regs(*regs, task_pt_regs(tsk));
         return 1;
 }
 
@@ -409,7 +407,7 @@ unsigned long get_wchan(struct task_struct *p)
         if (!p || p == current || p->state == TASK_RUNNING)
                 return 0;
 
-        stack_page = (unsigned long)p->thread_info;
+        stack_page = (unsigned long)task_stack_page(p);
         if (!stack_page || !mips_frame_info_initialized)
                 return 0;
 
diff --git a/arch/mips/kernel/ptrace.c b/arch/mips/kernel/ptrace.c
index 8d2549335304..f838b36cc765 100644
--- a/arch/mips/kernel/ptrace.c
+++ b/arch/mips/kernel/ptrace.c
@@ -64,8 +64,7 @@ int ptrace_getregs (struct task_struct *child, __s64 __user *data)
         if (!access_ok(VERIFY_WRITE, data, 38 * 8))
                 return -EIO;
 
-        regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-               THREAD_SIZE - 32 - sizeof(struct pt_regs));
+        regs = task_pt_regs(child);
 
         for (i = 0; i < 32; i++)
                 __put_user (regs->regs[i], data + i);
@@ -92,8 +91,7 @@ int ptrace_setregs (struct task_struct *child, __s64 __user *data)
         if (!access_ok(VERIFY_READ, data, 38 * 8))
                 return -EIO;
 
-        regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-               THREAD_SIZE - 32 - sizeof(struct pt_regs));
+        regs = task_pt_regs(child);
 
         for (i = 0; i < 32; i++)
                 __get_user (regs->regs[i], data + i);
@@ -198,8 +196,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                 struct pt_regs *regs;
                 unsigned long tmp = 0;
 
-                regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-                       THREAD_SIZE - 32 - sizeof(struct pt_regs));
+                regs = task_pt_regs(child);
                 ret = 0;  /* Default return value. */
 
                 switch (addr) {
@@ -314,8 +311,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
         case PTRACE_POKEUSR: {
                 struct pt_regs *regs;
                 ret = 0;
-                regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-                       THREAD_SIZE - 32 - sizeof(struct pt_regs));
+                regs = task_pt_regs(child);
 
                 switch (addr) {
                 case 0 ... 31:
@@ -442,7 +438,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                 break;
 
         case PTRACE_GET_THREAD_AREA:
-                ret = put_user(child->thread_info->tp_value,
+                ret = put_user(task_thread_info(child)->tp_value,
                                 (unsigned long __user *) data);
                 break;
 
diff --git a/arch/mips/kernel/ptrace32.c b/arch/mips/kernel/ptrace32.c
index 1f998bfde165..0c82b25d8c6d 100644
--- a/arch/mips/kernel/ptrace32.c
+++ b/arch/mips/kernel/ptrace32.c
@@ -126,8 +126,7 @@ asmlinkage int sys32_ptrace(int request, int pid, int addr, int data)
                 struct pt_regs *regs;
                 unsigned int tmp;
 
-                regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-                       THREAD_SIZE - 32 - sizeof(struct pt_regs));
+                regs = task_pt_regs(child);
                 ret = 0;  /* Default return value. */
 
                 switch (addr) {
@@ -259,8 +258,7 @@ asmlinkage int sys32_ptrace(int request, int pid, int addr, int data)
         case PTRACE_POKEUSR: {
                 struct pt_regs *regs;
                 ret = 0;
-                regs = (struct pt_regs *) ((unsigned long) child->thread_info +
-                       THREAD_SIZE - 32 - sizeof(struct pt_regs));
+                regs = task_pt_regs(child);
 
                 switch (addr) {
                 case 0 ... 31:
@@ -377,7 +375,7 @@ asmlinkage int sys32_ptrace(int request, int pid, int addr, int data)
                 break;
 
         case PTRACE_GET_THREAD_AREA:
-                ret = put_user(child->thread_info->tp_value,
+                ret = put_user(task_thread_info(child)->tp_value,
                                 (unsigned int __user *) (unsigned long) data);
                 break;
 
@@ -391,7 +389,7 @@ asmlinkage int sys32_ptrace(int request, int pid, int addr, int data)
                 break;
 
         case PTRACE_GET_THREAD_AREA_3264:
-                ret = put_user(child->thread_info->tp_value,
+                ret = put_user(task_thread_info(child)->tp_value,
                                 (unsigned long __user *) (unsigned long) data);
                 break;
 
diff --git a/arch/mips/kernel/smp_mt.c b/arch/mips/kernel/smp_mt.c
index d429544ba4bc..794a1c3de2a4 100644
--- a/arch/mips/kernel/smp_mt.c
+++ b/arch/mips/kernel/smp_mt.c
@@ -287,6 +287,7 @@ void prom_prepare_cpus(unsigned int max_cpus)
  */
 void prom_boot_secondary(int cpu, struct task_struct *idle)
 {
+        struct thread_info *gp = task_thread_info(idle);
         dvpe();
         set_c0_mvpcontrol(MVPCONTROL_VPC);
 
@@ -307,11 +308,9 @@ void prom_boot_secondary(int cpu, struct task_struct *idle)
         write_tc_gpr_sp( __KSTK_TOS(idle));
 
         /* global pointer */
-        write_tc_gpr_gp((unsigned long)idle->thread_info);
+        write_tc_gpr_gp((unsigned long)gp);
 
-        flush_icache_range((unsigned long)idle->thread_info,
-                                           (unsigned long)idle->thread_info +
-                                           sizeof(struct thread_info));
+        flush_icache_range((unsigned long)gp, (unsigned long)(gp + 1));
 
         /* finally out of configuration and into chaos */
         clear_c0_mvpcontrol(MVPCONTROL_VPC);
diff --git a/arch/mips/kernel/syscall.c b/arch/mips/kernel/syscall.c
index 006881942aa2..332358430ff5 100644
--- a/arch/mips/kernel/syscall.c
+++ b/arch/mips/kernel/syscall.c
@@ -263,7 +263,7 @@ asmlinkage int sys_olduname(struct oldold_utsname * name)
 
 void sys_set_thread_area(unsigned long addr)
 {
-        struct thread_info *ti = current->thread_info;
+        struct thread_info *ti = task_thread_info(current);
 
         ti->tp_value = addr;
 
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index 7058893d5ad2..59a187956de0 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -519,7 +519,7 @@ static inline int simulate_llsc(struct pt_regs *regs)
  */
 static inline int simulate_rdhwr(struct pt_regs *regs)
 {
-        struct thread_info *ti = current->thread_info;
+        struct thread_info *ti = task_thread_info(current);
         unsigned int opcode;
 
         if (unlikely(get_insn_opcode(regs, &opcode)))
diff --git a/arch/mips/pmc-sierra/yosemite/smp.c b/arch/mips/pmc-sierra/yosemite/smp.c
index 0527170d6adb..f17f575f58f0 100644
--- a/arch/mips/pmc-sierra/yosemite/smp.c
+++ b/arch/mips/pmc-sierra/yosemite/smp.c
@@ -93,8 +93,8 @@ void __init prom_prepare_cpus(unsigned int max_cpus)
  */
 void prom_boot_secondary(int cpu, struct task_struct *idle)
 {
-        unsigned long gp = (unsigned long) idle->thread_info;
-        unsigned long sp = gp + THREAD_SIZE - 32;
+        unsigned long gp = (unsigned long) task_thread_info(idle);
+        unsigned long sp = __KSTK_TOP(idle);
 
         secondary_sp = sp;
         secondary_gp = gp;
diff --git a/arch/mips/sgi-ip27/ip27-smp.c b/arch/mips/sgi-ip27/ip27-smp.c
index 3a8291b7d26d..dbef3f6b5650 100644
--- a/arch/mips/sgi-ip27/ip27-smp.c
+++ b/arch/mips/sgi-ip27/ip27-smp.c
@@ -168,8 +168,8 @@ void __init prom_prepare_cpus(unsigned int max_cpus)
  */
 void __init prom_boot_secondary(int cpu, struct task_struct *idle)
 {
-        unsigned long gp = (unsigned long) idle->thread_info;
-        unsigned long sp = gp + THREAD_SIZE - 32;
+        unsigned long gp = (unsigned long)task_thread_info(idle);
+        unsigned long sp = __KSTK_TOS(idle);
 
         LAUNCH_SLAVE(cputonasid(cpu),cputoslice(cpu),
                 (launch_proc_t)MAPPED_KERN_RW_TO_K0(smp_bootstrap),
diff --git a/arch/mips/sibyte/cfe/smp.c b/arch/mips/sibyte/cfe/smp.c
index e8485124b8fc..4477af3d8074 100644
--- a/arch/mips/sibyte/cfe/smp.c
+++ b/arch/mips/sibyte/cfe/smp.c
@@ -60,7 +60,7 @@ void prom_boot_secondary(int cpu, struct task_struct *idle)
 
         retval = cfe_cpu_start(cpu_logical_map(cpu), &smp_bootstrap,
                                __KSTK_TOS(idle),
-                               (unsigned long)idle->thread_info, 0);
+                               (unsigned long)task_thread_info(idle), 0);
         if (retval != 0)
                 printk("cfe_start_cpu(%i) returned %i\n" , cpu, retval);
 }
diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c
index 4eb70a40ec7e..5da41677e70b 100644
--- a/arch/parisc/kernel/process.c
+++ b/arch/parisc/kernel/process.c
@@ -295,7 +295,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
             struct task_struct * p, struct pt_regs * pregs)
 {
         struct pt_regs * cregs = &(p->thread.regs);
-        struct thread_info *ti = p->thread_info;
+        void *stack = task_stack_page(p);
         
         /* We have to use void * instead of a function pointer, because
          * function pointers aren't a pointer to the function on 64-bit.
@@ -322,7 +322,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
          */
         if (usp == 1) {
                 /* kernel thread */
-                cregs->ksp = (((unsigned long)(ti)) + THREAD_SZ_ALGN);
+                cregs->ksp = (unsigned long)stack + THREAD_SZ_ALGN;
                 /* Must exit via ret_from_kernel_thread in order
                  * to call schedule_tail()
                  */
@@ -344,7 +344,7 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
                  */
 
                 /* Use same stack depth as parent */
-                cregs->ksp = ((unsigned long)(ti))
+                cregs->ksp = (unsigned long)stack
                         + (pregs->gr[21] & (THREAD_SIZE - 1));
                 cregs->gr[30] = usp;
                 if (p->personality == PER_HPUX) {
diff --git a/arch/parisc/kernel/smp.c b/arch/parisc/kernel/smp.c
index 17f23c26f1ca..25564b7ca6bb 100644
--- a/arch/parisc/kernel/smp.c
+++ b/arch/parisc/kernel/smp.c
@@ -517,7 +517,7 @@ int __init smp_boot_one_cpu(int cpuid)
         if (IS_ERR(idle))
                 panic("SMP: fork failed for CPU:%d", cpuid);
 
-        idle->thread_info->cpu = cpuid;
+        task_thread_info(idle)->cpu = cpuid;
 
         /* Let _start know what logical CPU we're booting
         ** (offset into init_tasks[],cpu_data[])
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 9101358cc6b3..57703994a063 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -426,7 +426,7 @@ void show_regs(struct pt_regs * regs)
         if (trap == 0x300 || trap == 0x600)
                 printk("DAR: "REG", DSISR: "REG"\n", regs->dar, regs->dsisr);
         printk("TASK = %p[%d] '%s' THREAD: %p",
-               current, current->pid, current->comm, current->thread_info);
+               current, current->pid, current->comm, task_thread_info(current));
 
 #ifdef CONFIG_SMP
         printk(" CPU: %d", smp_processor_id());
@@ -505,7 +505,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 {
         struct pt_regs *childregs, *kregs;
         extern void ret_from_fork(void);
-        unsigned long sp = (unsigned long)p->thread_info + THREAD_SIZE;
+        unsigned long sp = (unsigned long)task_stack_page(p) + THREAD_SIZE;
 
         CHECK_FULL_REGS(regs);
         /* Copy registers */
@@ -518,7 +518,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 #ifdef CONFIG_PPC32
                 childregs->gpr[2] = (unsigned long) p;
 #else
-                clear_ti_thread_flag(p->thread_info, TIF_32BIT);
+                clear_tsk_thread_flag(p, TIF_32BIT);
 #endif
                 p->thread.regs = NULL;  /* no user register state */
         } else {
@@ -590,10 +590,8 @@ void start_thread(struct pt_regs *regs, unsigned long start, unsigned long sp)
          * set.  Do it now.
          */
         if (!current->thread.regs) {
-                unsigned long childregs = (unsigned long)current->thread_info +
-                                                THREAD_SIZE;
-                childregs -= sizeof(struct pt_regs);
-                current->thread.regs = (struct pt_regs *)childregs;
+                struct pt_regs *regs = task_stack_page(current) + THREAD_SIZE;
+                current->thread.regs = regs - 1;
         }
 
         memset(regs->gpr, 0, sizeof(regs->gpr));
@@ -769,7 +767,7 @@ out:
 static int validate_sp(unsigned long sp, struct task_struct *p,
                        unsigned long nbytes)
 {
-        unsigned long stack_page = (unsigned long)p->thread_info;
+        unsigned long stack_page = (unsigned long)task_stack_page(p);
 
         if (sp >= stack_page + sizeof(struct thread_struct)
             && sp <= stack_page + THREAD_SIZE - nbytes)
diff --git a/arch/powerpc/kernel/ptrace-common.h b/arch/powerpc/kernel/ptrace-common.h
index b1babb729673..5ccbdbe0d5c9 100644
--- a/arch/powerpc/kernel/ptrace-common.h
+++ b/arch/powerpc/kernel/ptrace-common.h
@@ -62,7 +62,7 @@ static inline void set_single_step(struct task_struct *task)
         struct pt_regs *regs = task->thread.regs;
         if (regs != NULL)
                 regs->msr |= MSR_SE;
-        set_ti_thread_flag(task->thread_info, TIF_SINGLESTEP);
+        set_tsk_thread_flag(task, TIF_SINGLESTEP);
 }
 
 static inline void clear_single_step(struct task_struct *task)
@@ -70,7 +70,7 @@ static inline void clear_single_step(struct task_struct *task)
         struct pt_regs *regs = task->thread.regs;
         if (regs != NULL)
                 regs->msr &= ~MSR_SE;
-        clear_ti_thread_flag(task->thread_info, TIF_SINGLESTEP);
+        clear_tsk_thread_flag(task, TIF_SINGLESTEP);
 }
 
 #ifdef CONFIG_ALTIVEC
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
index d381ec90b759..c8458c531b25 100644
--- a/arch/powerpc/kernel/smp.c
+++ b/arch/powerpc/kernel/smp.c
@@ -338,8 +338,8 @@ static void __init smp_create_idle(unsigned int cpu)
 #ifdef CONFIG_PPC64
         paca[cpu].__current = p;
 #endif
-        current_set[cpu] = p->thread_info;
-        p->thread_info->cpu = cpu;
+        current_set[cpu] = task_thread_info(p);
+        task_thread_info(p)->cpu = cpu;
 }
 
 void __init smp_prepare_cpus(unsigned int max_cpus)
@@ -375,7 +375,7 @@ void __devinit smp_prepare_boot_cpu(void)
 #ifdef CONFIG_PPC64
         paca[boot_cpuid].__current = current;
 #endif
-        current_set[boot_cpuid] = current->thread_info;
+        current_set[boot_cpuid] = task_thread_info(current);
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
diff --git a/arch/powerpc/platforms/cell/smp.c b/arch/powerpc/platforms/cell/smp.c
index de96eadf419d..bdf6c5fe58c0 100644
--- a/arch/powerpc/platforms/cell/smp.c
+++ b/arch/powerpc/platforms/cell/smp.c
@@ -86,7 +86,7 @@ static inline int __devinit smp_startup_cpu(unsigned int lcpu)
         pcpu = get_hard_smp_processor_id(lcpu);
 
         /* Fixup atomic count: it exited inside IRQ handler. */
-        paca[lcpu].__current->thread_info->preempt_count        = 0;
+        task_thread_info(paca[lcpu].__current)->preempt_count   = 0;
 
         /*
          * If the RTAS start-cpu token does not exist then presume the
diff --git a/arch/powerpc/platforms/pseries/smp.c b/arch/powerpc/platforms/pseries/smp.c
index e4d017dd5ef3..8e6b1ed1396e 100644
--- a/arch/powerpc/platforms/pseries/smp.c
+++ b/arch/powerpc/platforms/pseries/smp.c
@@ -282,7 +282,7 @@ static inline int __devinit smp_startup_cpu(unsigned int lcpu)
         pcpu = get_hard_smp_processor_id(lcpu);
 
         /* Fixup atomic count: it exited inside IRQ handler. */
-        paca[lcpu].__current->thread_info->preempt_count        = 0;
+        task_thread_info(paca[lcpu].__current)->preempt_count   = 0;
 
         /* 
          * If the RTAS start-cpu token does not exist then presume the
diff --git a/arch/ppc/amiga/amiints.c b/arch/ppc/amiga/amiints.c
index 91195e2ce38d..5f35cf3986f7 100644
--- a/arch/ppc/amiga/amiints.c
+++ b/arch/ppc/amiga/amiints.c
@@ -96,8 +96,8 @@ void amiga_init_IRQ(void)
                 gayle.inten = GAYLE_IRQ_IDE;
 
         /* turn off all interrupts... */
-        custom.intena = 0x7fff;
-        custom.intreq = 0x7fff;
+        amiga_custom.intena = 0x7fff;
+        amiga_custom.intreq = 0x7fff;
 
 #ifdef CONFIG_APUS
         /* Clear any inter-CPU interrupt requests. Circumvents bug in
@@ -110,7 +110,7 @@ void amiga_init_IRQ(void)
         APUS_WRITE(APUS_IPL_EMU, IPLEMU_SETRESET | IPLEMU_IPLMASK);
 #endif
         /* ... and enable the master interrupt bit */
-        custom.intena = IF_SETCLR | IF_INTEN;
+        amiga_custom.intena = IF_SETCLR | IF_INTEN;
 
         cia_init_IRQ(&ciaa_base);
         cia_init_IRQ(&ciab_base);
@@ -151,7 +151,7 @@ void amiga_enable_irq(unsigned int irq)
         }
 
         /* enable the interrupt */
-        custom.intena = IF_SETCLR | ami_intena_vals[irq];
+        amiga_custom.intena = IF_SETCLR | ami_intena_vals[irq];
 }
 
 void amiga_disable_irq(unsigned int irq)
@@ -177,7 +177,7 @@ void amiga_disable_irq(unsigned int irq)
         }
 
         /* disable the interrupt */
-        custom.intena = ami_intena_vals[irq];
+        amiga_custom.intena = ami_intena_vals[irq];
 }
 
 inline void amiga_do_irq(int irq, struct pt_regs *fp)
@@ -196,7 +196,7 @@ void amiga_do_irq_list(int irq, struct pt_regs *fp)
 
         kstat_cpu(0).irqs[irq]++;
 
-        custom.intreq = ami_intena_vals[irq];
+        amiga_custom.intreq = ami_intena_vals[irq];
 
         for (action = desc->action; action; action = action->next)
                 action->handler(irq, action->dev_id, fp);
@@ -208,40 +208,40 @@ void amiga_do_irq_list(int irq, struct pt_regs *fp)
 
 static void ami_int1(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if serial transmit buffer empty, interrupt */
         if (ints & IF_TBE) {
-                custom.intreq = IF_TBE;
+                amiga_custom.intreq = IF_TBE;
                 amiga_do_irq(IRQ_AMIGA_TBE, fp);
         }
 
         /* if floppy disk transfer complete, interrupt */
         if (ints & IF_DSKBLK) {
-                custom.intreq = IF_DSKBLK;
+                amiga_custom.intreq = IF_DSKBLK;
                 amiga_do_irq(IRQ_AMIGA_DSKBLK, fp);
         }
 
         /* if software interrupt set, interrupt */
         if (ints & IF_SOFT) {
-                custom.intreq = IF_SOFT;
+                amiga_custom.intreq = IF_SOFT;
                 amiga_do_irq(IRQ_AMIGA_SOFT, fp);
         }
 }
 
 static void ami_int3(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if a blitter interrupt */
         if (ints & IF_BLIT) {
-                custom.intreq = IF_BLIT;
+                amiga_custom.intreq = IF_BLIT;
                 amiga_do_irq(IRQ_AMIGA_BLIT, fp);
         }
 
         /* if a copper interrupt */
         if (ints & IF_COPER) {
-                custom.intreq = IF_COPER;
+                amiga_custom.intreq = IF_COPER;
                 amiga_do_irq(IRQ_AMIGA_COPPER, fp);
         }
 
@@ -252,36 +252,36 @@ static void ami_int3(int irq, void *dev_id, struct pt_regs *fp)
 
 static void ami_int4(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if audio 0 interrupt */
         if (ints & IF_AUD0) {
-                custom.intreq = IF_AUD0;
+                amiga_custom.intreq = IF_AUD0;
                 amiga_do_irq(IRQ_AMIGA_AUD0, fp);
         }
 
         /* if audio 1 interrupt */
         if (ints & IF_AUD1) {
-                custom.intreq = IF_AUD1;
+                amiga_custom.intreq = IF_AUD1;
                 amiga_do_irq(IRQ_AMIGA_AUD1, fp);
         }
 
         /* if audio 2 interrupt */
         if (ints & IF_AUD2) {
-                custom.intreq = IF_AUD2;
+                amiga_custom.intreq = IF_AUD2;
                 amiga_do_irq(IRQ_AMIGA_AUD2, fp);
         }
 
         /* if audio 3 interrupt */
         if (ints & IF_AUD3) {
-                custom.intreq = IF_AUD3;
+                amiga_custom.intreq = IF_AUD3;
                 amiga_do_irq(IRQ_AMIGA_AUD3, fp);
         }
 }
 
 static void ami_int5(int irq, void *dev_id, struct pt_regs *fp)
 {
-        unsigned short ints = custom.intreqr & custom.intenar;
+        unsigned short ints = amiga_custom.intreqr & amiga_custom.intenar;
 
         /* if serial receive buffer full interrupt */
         if (ints & IF_RBF) {
@@ -291,7 +291,7 @@ static void ami_int5(int irq, void *dev_id, struct pt_regs *fp)
 
         /* if a disk sync interrupt */
         if (ints & IF_DSKSYN) {
-                custom.intreq = IF_DSKSYN;
+                amiga_custom.intreq = IF_DSKSYN;
                 amiga_do_irq(IRQ_AMIGA_DSKSYN, fp);
         }
 }
diff --git a/arch/ppc/amiga/cia.c b/arch/ppc/amiga/cia.c
index ad961465b6cb..4431c58f611a 100644
--- a/arch/ppc/amiga/cia.c
+++ b/arch/ppc/amiga/cia.c
@@ -66,7 +66,7 @@ static unsigned char cia_set_irq_private(struct ciabase *base,
         else
                 base->icr_data &= ~mask;
         if (base->icr_data & base->icr_mask)
-                custom.intreq = IF_SETCLR | base->int_mask;
+                amiga_custom.intreq = IF_SETCLR | base->int_mask;
         return old & base->icr_mask;
 }
 
@@ -114,7 +114,7 @@ static unsigned char cia_able_irq_private(struct ciabase *base,
         base->icr_mask &= CIA_ICR_ALL;
 
         if (base->icr_data & base->icr_mask)
-                custom.intreq = IF_SETCLR | base->int_mask;
+                amiga_custom.intreq = IF_SETCLR | base->int_mask;
         return old;
 }
 
@@ -145,7 +145,7 @@ static void cia_handler(int irq, void *dev_id, struct pt_regs *fp)
         irq = base->cia_irq;
         desc = irq_desc + irq;
         ints = cia_set_irq_private(base, CIA_ICR_ALL);
-        custom.intreq = base->int_mask;
+        amiga_custom.intreq = base->int_mask;
         for (i = 0; i < CIA_IRQS; i++, irq++) {
                 if (ints & 1) {
                         kstat_cpu(0).irqs[irq]++;
@@ -174,5 +174,5 @@ void __init cia_init_IRQ(struct ciabase *base)
         action->name = base->name;
         setup_irq(base->handler_irq, &amiga_sys_irqaction[base->handler_irq-IRQ_AMIGA_AUTO]);
 
-        custom.intena = IF_SETCLR | base->int_mask;
+        amiga_custom.intena = IF_SETCLR | base->int_mask;
 }
diff --git a/arch/ppc/amiga/config.c b/arch/ppc/amiga/config.c
index af881d7454dd..60e2da1c92c0 100644
--- a/arch/ppc/amiga/config.c
+++ b/arch/ppc/amiga/config.c
@@ -90,9 +90,6 @@ static void a3000_gettod (int *, int *, int *, int *, int *, int *);
 static void a2000_gettod (int *, int *, int *, int *, int *, int *);
 static int amiga_hwclk (int, struct hwclk_time *);
 static int amiga_set_clock_mmss (unsigned long);
-#ifdef CONFIG_AMIGA_FLOPPY
-extern void amiga_floppy_setup(char *, int *);
-#endif
 static void amiga_reset (void);
 extern void amiga_init_sound(void);
 static void amiga_savekmsg_init(void);
@@ -281,7 +278,7 @@ static void __init amiga_identify(void)
     case CS_OCS:
     case CS_ECS:
     case CS_AGA:
-      switch (custom.deniseid & 0xf) {
+      switch (amiga_custom.deniseid & 0xf) {
       case 0x0c:
         AMIGAHW_SET(DENISE_HR);
         break;
@@ -294,7 +291,7 @@ static void __init amiga_identify(void)
       AMIGAHW_SET(DENISE);
       break;
     }
-    switch ((custom.vposr>>8) & 0x7f) {
+    switch ((amiga_custom.vposr>>8) & 0x7f) {
     case 0x00:
       AMIGAHW_SET(AGNUS_PAL);
       break;
@@ -419,9 +416,6 @@ void __init config_amiga(void)
 
   mach_hwclk           = amiga_hwclk;
   mach_set_clock_mmss  = amiga_set_clock_mmss;
-#ifdef CONFIG_AMIGA_FLOPPY
-  mach_floppy_setup    = amiga_floppy_setup;
-#endif
   mach_reset           = amiga_reset;
 #ifdef CONFIG_HEARTBEAT
   mach_heartbeat = amiga_heartbeat;
@@ -432,9 +426,9 @@ void __init config_amiga(void)
   amiga_colorclock = 5*amiga_eclock;    /* 3.5 MHz */
 
   /* clear all DMA bits */
-  custom.dmacon = DMAF_ALL;
+  amiga_custom.dmacon = DMAF_ALL;
   /* ensure that the DMA master bit is set */
-  custom.dmacon = DMAF_SETCLR | DMAF_MASTER;
+  amiga_custom.dmacon = DMAF_SETCLR | DMAF_MASTER;
 
   /* request all RAM */
   for (i = 0; i < m68k_num_memory; i++) {
@@ -753,9 +747,9 @@ static void amiga_savekmsg_init(void)
 
 static void amiga_serial_putc(char c)
 {
-    custom.serdat = (unsigned char)c | 0x100;
+    amiga_custom.serdat = (unsigned char)c | 0x100;
     mb();
-    while (!(custom.serdatr & 0x2000))
+    while (!(amiga_custom.serdatr & 0x2000))
        ;
 }
 
@@ -785,11 +779,11 @@ int amiga_serial_console_wait_key(struct console *co)
 {
     int ch;
 
-    while (!(custom.intreqr & IF_RBF))
+    while (!(amiga_custom.intreqr & IF_RBF))
         barrier();
-    ch = custom.serdatr & 0xff;
+    ch = amiga_custom.serdatr & 0xff;
     /* clear the interrupt, so that another character can be read */
-    custom.intreq = IF_RBF;
+    amiga_custom.intreq = IF_RBF;
     return ch;
 }
 
diff --git a/arch/ppc/kernel/smp.c b/arch/ppc/kernel/smp.c
index becbfa397556..e55cdda6149a 100644
--- a/arch/ppc/kernel/smp.c
+++ b/arch/ppc/kernel/smp.c
@@ -318,7 +318,7 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
                 p = fork_idle(cpu);
                 if (IS_ERR(p))
                         panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
-                p->thread_info->cpu = cpu;
+                task_thread_info(p)->cpu = cpu;
                 idle_tasks[cpu] = p;
         }
 }
@@ -369,7 +369,7 @@ int __cpu_up(unsigned int cpu)
         char buf[32];
         int c;
 
-        secondary_ti = idle_tasks[cpu]->thread_info;
+        secondary_ti = task_thread_info(idle_tasks[cpu]);
         mb();
 
         /*
diff --git a/arch/ppc/platforms/apus_setup.c b/arch/ppc/platforms/apus_setup.c
index 2f74fde98ebc..c42c50073da5 100644
--- a/arch/ppc/platforms/apus_setup.c
+++ b/arch/ppc/platforms/apus_setup.c
@@ -55,9 +55,6 @@ int (*mach_hwclk) (int, struct hwclk_time*) = NULL;
 int (*mach_set_clock_mmss) (unsigned long) = NULL;
 void (*mach_reset)( void );
 long mach_max_dma_address = 0x00ffffff; /* default set to the lower 16MB */
-#if defined(CONFIG_AMIGA_FLOPPY)
-void (*mach_floppy_setup) (char *, int *) __initdata = NULL;
-#endif
 #ifdef CONFIG_HEARTBEAT
 void (*mach_heartbeat) (int) = NULL;
 extern void apus_heartbeat (void);
@@ -76,7 +73,6 @@ struct mem_info m68k_memory[NUM_MEMINFO];/* memory description */
 
 struct mem_info ramdisk;
 
-extern void amiga_floppy_setup(char *, int *);
 extern void config_amiga(void);
 
 static int __60nsram = 0;
@@ -305,16 +301,6 @@ void kbd_reset_setup(char *str, int *ints)
 {
 }
 
-/*********************************************************** FLOPPY */
-#if defined(CONFIG_AMIGA_FLOPPY)
-__init
-void floppy_setup(char *str, int *ints)
-{
-        if (mach_floppy_setup)
-                mach_floppy_setup (str, ints);
-}
-#endif
-
 /*********************************************************** MEMORY */
 #define KMAP_MAX 32
 unsigned long kmap_chunks[KMAP_MAX*3];
@@ -574,9 +560,9 @@ static __inline__ void ser_RTSon(void)
 
 int __debug_ser_out( unsigned char c )
 {
-        custom.serdat = c | 0x100;
+        amiga_custom.serdat = c | 0x100;
         mb();
-        while (!(custom.serdatr & 0x2000))
+        while (!(amiga_custom.serdatr & 0x2000))
                 barrier();
         return 1;
 }
@@ -586,11 +572,11 @@ unsigned char __debug_ser_in( void )
         unsigned char c;
 
         /* XXX: is that ok?? derived from amiga_ser.c... */
-        while( !(custom.intreqr & IF_RBF) )
+        while( !(amiga_custom.intreqr & IF_RBF) )
                 barrier();
-        c = custom.serdatr;
+        c = amiga_custom.serdatr;
         /* clear the interrupt, so that another character can be read */
-        custom.intreq = IF_RBF;
+        amiga_custom.intreq = IF_RBF;
         return c;
 }
 
@@ -601,10 +587,10 @@ int __debug_serinit( void )
         local_irq_save(flags);
 
         /* turn off Rx and Tx interrupts */
-        custom.intena = IF_RBF | IF_TBE;
+        amiga_custom.intena = IF_RBF | IF_TBE;
 
         /* clear any pending interrupt */
-        custom.intreq = IF_RBF | IF_TBE;
+        amiga_custom.intreq = IF_RBF | IF_TBE;
 
         local_irq_restore(flags);
 
@@ -617,7 +603,7 @@ int __debug_serinit( void )
 
 #ifdef CONFIG_KGDB
         /* turn Rx interrupts on for GDB */
-        custom.intena = IF_SETCLR | IF_RBF;
+        amiga_custom.intena = IF_SETCLR | IF_RBF;
         ser_RTSon();
 #endif
 
diff --git a/arch/ppc/xmon/xmon.c b/arch/ppc/xmon/xmon.c
index 2b483b4f1602..9075a7538e26 100644
--- a/arch/ppc/xmon/xmon.c
+++ b/arch/ppc/xmon/xmon.c
@@ -99,7 +99,7 @@ static void remove_bpts(void);
 static void insert_bpts(void);
 static struct bpt *at_breakpoint(unsigned pc);
 static void bpt_cmds(void);
-static void cacheflush(void);
+void cacheflush(void);
 #ifdef CONFIG_SMP
 static void cpu_cmd(void);
 #endif /* CONFIG_SMP */
diff --git a/arch/s390/kernel/binfmt_elf32.c b/arch/s390/kernel/binfmt_elf32.c
index 03ba5893f17b..1f451c2cb071 100644
--- a/arch/s390/kernel/binfmt_elf32.c
+++ b/arch/s390/kernel/binfmt_elf32.c
@@ -112,7 +112,7 @@ static inline int dump_regs32(struct pt_regs *ptregs, elf_gregset_t *regs)
 
 static inline int dump_task_regs32(struct task_struct *tsk, elf_gregset_t *regs)
 {
-        struct pt_regs *ptregs = __KSTK_PTREGS(tsk);
+        struct pt_regs *ptregs = task_pt_regs(tsk);
         int i;
 
         memcpy(&regs->psw.mask, &ptregs->psw.mask, 4);
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index 7dd58f8ac6b5..2ff90a1a1056 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -153,7 +153,7 @@ void show_regs(struct pt_regs *regs)
 {
         struct task_struct *tsk = current;
 
-        printk("CPU:    %d    %s\n", tsk->thread_info->cpu, print_tainted());
+        printk("CPU:    %d    %s\n", task_thread_info(tsk)->cpu, print_tainted());
         printk("Process %s (pid: %d, task: %p, ksp: %p)\n",
                current->comm, current->pid, (void *) tsk,
                (void *) tsk->thread.ksp);
@@ -217,8 +217,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long new_stackp,
             struct pt_regs childregs;
           } *frame;
 
-        frame = ((struct fake_frame *)
-                 (THREAD_SIZE + (unsigned long) p->thread_info)) - 1;
+        frame = container_of(task_pt_regs(p), struct fake_frame, childregs);
         p->thread.ksp = (unsigned long) frame;
         /* Store access registers to kernel stack of new process. */
         frame->childregs = *regs;
@@ -358,11 +357,10 @@ unsigned long get_wchan(struct task_struct *p)
         unsigned long return_address;
         int count;
 
-        if (!p || p == current || p->state == TASK_RUNNING || !p->thread_info)
+        if (!p || p == current || p->state == TASK_RUNNING || !task_stack_page(p))
                 return 0;
-        low = (struct stack_frame *) p->thread_info;
-        high = (struct stack_frame *)
-                ((unsigned long) p->thread_info + THREAD_SIZE) - 1;
+        low = task_stack_page(p);
+        high = (struct stack_frame *) task_pt_regs(p);
         sf = (struct stack_frame *) (p->thread.ksp & PSW_ADDR_INSN);
         if (sf <= low || sf > high)
                 return 0;
diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c
index cc02232aa96e..37dfe33dab73 100644
--- a/arch/s390/kernel/ptrace.c
+++ b/arch/s390/kernel/ptrace.c
@@ -52,7 +52,7 @@ FixPerRegisters(struct task_struct *task)
         struct pt_regs *regs;
         per_struct *per_info;
 
-        regs = __KSTK_PTREGS(task);
+        regs = task_pt_regs(task);
         per_info = (per_struct *) &task->thread.per_info;
         per_info->control_regs.bits.em_instruction_fetch =
                 per_info->single_step | per_info->instruction_fetch;
@@ -150,7 +150,7 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data)
                 /*
                  * psw and gprs are stored on the stack
                  */
-                tmp = *(addr_t *)((addr_t) &__KSTK_PTREGS(child)->psw + addr);
+                tmp = *(addr_t *)((addr_t) &task_pt_regs(child)->psw + addr);
                 if (addr == (addr_t) &dummy->regs.psw.mask)
                         /* Remove per bit from user psw. */
                         tmp &= ~PSW_MASK_PER;
@@ -176,7 +176,7 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data)
                 /*
                  * orig_gpr2 is stored on the kernel stack
                  */
-                tmp = (addr_t) __KSTK_PTREGS(child)->orig_gpr2;
+                tmp = (addr_t) task_pt_regs(child)->orig_gpr2;
 
         } else if (addr < (addr_t) (&dummy->regs.fp_regs + 1)) {
                 /* 
@@ -243,7 +243,7 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data)
                            high order bit but older gdb's rely on it */
                         data |= PSW_ADDR_AMODE;
 #endif
-                *(addr_t *)((addr_t) &__KSTK_PTREGS(child)->psw + addr) = data;
+                *(addr_t *)((addr_t) &task_pt_regs(child)->psw + addr) = data;
 
         } else if (addr < (addr_t) (&dummy->regs.orig_gpr2)) {
                 /*
@@ -267,7 +267,7 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data)
                 /*
                  * orig_gpr2 is stored on the kernel stack
                  */
-                __KSTK_PTREGS(child)->orig_gpr2 = data;
+                task_pt_regs(child)->orig_gpr2 = data;
 
         } else if (addr < (addr_t) (&dummy->regs.fp_regs + 1)) {
                 /*
@@ -393,15 +393,15 @@ peek_user_emu31(struct task_struct *child, addr_t addr, addr_t data)
                  */
                 if (addr == (addr_t) &dummy32->regs.psw.mask) {
                         /* Fake a 31 bit psw mask. */
-                        tmp = (__u32)(__KSTK_PTREGS(child)->psw.mask >> 32);
+                        tmp = (__u32)(task_pt_regs(child)->psw.mask >> 32);
                         tmp = PSW32_MASK_MERGE(PSW32_USER_BITS, tmp);
                 } else if (addr == (addr_t) &dummy32->regs.psw.addr) {
                         /* Fake a 31 bit psw address. */
-                        tmp = (__u32) __KSTK_PTREGS(child)->psw.addr |
+                        tmp = (__u32) task_pt_regs(child)->psw.addr |
                                 PSW32_ADDR_AMODE31;
                 } else {
                         /* gpr 0-15 */
-                        tmp = *(__u32 *)((addr_t) &__KSTK_PTREGS(child)->psw +
+                        tmp = *(__u32 *)((addr_t) &task_pt_regs(child)->psw +
                                          addr*2 + 4);
                 }
         } else if (addr < (addr_t) (&dummy32->regs.orig_gpr2)) {
@@ -415,7 +415,7 @@ peek_user_emu31(struct task_struct *child, addr_t addr, addr_t data)
                 /*
                  * orig_gpr2 is stored on the kernel stack
                  */
-                tmp = *(__u32*)((addr_t) &__KSTK_PTREGS(child)->orig_gpr2 + 4);
+                tmp = *(__u32*)((addr_t) &task_pt_regs(child)->orig_gpr2 + 4);
 
         } else if (addr < (addr_t) (&dummy32->regs.fp_regs + 1)) {
                 /*
@@ -472,15 +472,15 @@ poke_user_emu31(struct task_struct *child, addr_t addr, addr_t data)
                         if (tmp != PSW32_MASK_MERGE(PSW32_USER_BITS, tmp))
                                 /* Invalid psw mask. */
                                 return -EINVAL;
-                        __KSTK_PTREGS(child)->psw.mask =
+                        task_pt_regs(child)->psw.mask =
                                 PSW_MASK_MERGE(PSW_USER32_BITS, (__u64) tmp << 32);
                 } else if (addr == (addr_t) &dummy32->regs.psw.addr) {
                         /* Build a 64 bit psw address from 31 bit address. */
-                        __KSTK_PTREGS(child)->psw.addr = 
+                        task_pt_regs(child)->psw.addr =
                                 (__u64) tmp & PSW32_ADDR_INSN;
                 } else {
                         /* gpr 0-15 */
-                        *(__u32*)((addr_t) &__KSTK_PTREGS(child)->psw
+                        *(__u32*)((addr_t) &task_pt_regs(child)->psw
                                   + addr*2 + 4) = tmp;
                 }
         } else if (addr < (addr_t) (&dummy32->regs.orig_gpr2)) {
@@ -494,7 +494,7 @@ poke_user_emu31(struct task_struct *child, addr_t addr, addr_t data)
                 /*
                  * orig_gpr2 is stored on the kernel stack
                  */
-                *(__u32*)((addr_t) &__KSTK_PTREGS(child)->orig_gpr2 + 4) = tmp;
+                *(__u32*)((addr_t) &task_pt_regs(child)->orig_gpr2 + 4) = tmp;
 
         } else if (addr < (addr_t) (&dummy32->regs.fp_regs + 1)) {
                 /*
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index e10f4ca00499..cbfcfd02a43a 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -657,7 +657,7 @@ __cpu_up(unsigned int cpu)
         idle = current_set[cpu];
         cpu_lowcore = lowcore_ptr[cpu];
         cpu_lowcore->kernel_stack = (unsigned long)
-                idle->thread_info + (THREAD_SIZE);
+                task_stack_page(idle) + (THREAD_SIZE);
         sf = (struct stack_frame *) (cpu_lowcore->kernel_stack
                                      - sizeof(struct pt_regs)
                                      - sizeof(struct stack_frame));
diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c
index c36353e8c140..b0d8ca8e5eeb 100644
--- a/arch/s390/kernel/time.c
+++ b/arch/s390/kernel/time.c
@@ -282,7 +282,7 @@ static inline void start_hz_timer(void)
 {
         if (!cpu_isset(smp_processor_id(), nohz_cpu_mask))
                 return;
-        account_ticks(__KSTK_PTREGS(current));
+        account_ticks(task_pt_regs(current));
         cpu_clear(smp_processor_id(), nohz_cpu_mask);
 }
 
diff --git a/arch/s390/kernel/traps.c b/arch/s390/kernel/traps.c
index 95d109968619..5d21e9e6e7b4 100644
--- a/arch/s390/kernel/traps.c
+++ b/arch/s390/kernel/traps.c
@@ -136,8 +136,8 @@ void show_trace(struct task_struct *task, unsigned long * stack)
         sp = __show_trace(sp, S390_lowcore.async_stack - ASYNC_SIZE,
                           S390_lowcore.async_stack);
         if (task)
-                __show_trace(sp, (unsigned long) task->thread_info,
-                             (unsigned long) task->thread_info + THREAD_SIZE);
+                __show_trace(sp, (unsigned long) task_stack_page(task),
+                             (unsigned long) task_stack_page(task) + THREAD_SIZE);
         else
                 __show_trace(sp, S390_lowcore.thread_info,
                              S390_lowcore.thread_info + THREAD_SIZE);
@@ -240,7 +240,7 @@ char *task_show_regs(struct task_struct *task, char *buffer)
 {
         struct pt_regs *regs;
 
-        regs = __KSTK_PTREGS(task);
+        regs = task_pt_regs(task);
         buffer += sprintf(buffer, "task: %p, ksp: %p\n",
                        task, (void *)task->thread.ksp);
         buffer += sprintf(buffer, "User PSW : %p %p\n",
diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c
index 8a2bea34ddd2..aac15e42d03b 100644
--- a/arch/sh/kernel/process.c
+++ b/arch/sh/kernel/process.c
@@ -191,13 +191,8 @@ void flush_thread(void)
 {
 #if defined(CONFIG_SH_FPU)
         struct task_struct *tsk = current;
-        struct pt_regs *regs = (struct pt_regs *)
-                                ((unsigned long)tsk->thread_info
-                                 + THREAD_SIZE - sizeof(struct pt_regs)
-                                 - sizeof(unsigned long));
-
         /* Forget lazy FPU state */
-        clear_fpu(tsk, regs);
+        clear_fpu(tsk, task_pt_regs(tsk));
         clear_used_math();
 #endif
 }
@@ -232,13 +227,7 @@ int dump_task_regs(struct task_struct *tsk, elf_gregset_t *regs)
 {
         struct pt_regs ptregs;
         
-        ptregs = *(struct pt_regs *)
-                ((unsigned long)tsk->thread_info + THREAD_SIZE
-                 - sizeof(struct pt_regs)
-#ifdef CONFIG_SH_DSP
-                 - sizeof(struct pt_dspregs)
-#endif
-                 - sizeof(unsigned long));
+        ptregs = *task_pt_regs(tsk);
         elf_core_copy_regs(regs, &ptregs);
 
         return 1;
@@ -252,11 +241,7 @@ dump_task_fpu (struct task_struct *tsk, elf_fpregset_t *fpu)
 #if defined(CONFIG_SH_FPU)
         fpvalid = !!tsk_used_math(tsk);
         if (fpvalid) {
-                struct pt_regs *regs = (struct pt_regs *)
-                                        ((unsigned long)tsk->thread_info
-                                         + THREAD_SIZE - sizeof(struct pt_regs)
-                                         - sizeof(unsigned long));
-                unlazy_fpu(tsk, regs);
+                unlazy_fpu(tsk, task_pt_regs(tsk));
                 memcpy(fpu, &tsk->thread.fpu.hard, sizeof(*fpu));
         }
 #endif
@@ -279,18 +264,13 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
         copy_to_stopped_child_used_math(p);
 #endif
 
-        childregs = ((struct pt_regs *)
-                (THREAD_SIZE + (unsigned long) p->thread_info)
-#ifdef CONFIG_SH_DSP
-                - sizeof(struct pt_dspregs)
-#endif
-                - sizeof(unsigned long)) - 1;
+        childregs = task_pt_regs(p);
         *childregs = *regs;
 
         if (user_mode(regs)) {
                 childregs->regs[15] = usp;
         } else {
-                childregs->regs[15] = (unsigned long)p->thread_info + THREAD_SIZE;
+                childregs->regs[15] = (unsigned long)task_stack_page(p) + THREAD_SIZE;
         }
         if (clone_flags & CLONE_SETTLS) {
                 childregs->gbr = childregs->regs[0];
@@ -333,11 +313,7 @@ ubc_set_tracing(int asid, unsigned long pc)
 struct task_struct *__switch_to(struct task_struct *prev, struct task_struct *next)
 {
 #if defined(CONFIG_SH_FPU)
-        struct pt_regs *regs = (struct pt_regs *)
-                                ((unsigned long)prev->thread_info
-                                 + THREAD_SIZE - sizeof(struct pt_regs)
-                                 - sizeof(unsigned long));
-        unlazy_fpu(prev, regs);
+        unlazy_fpu(prev, task_pt_regs(prev));
 #endif
 
 #ifdef CONFIG_PREEMPT
@@ -346,13 +322,7 @@ struct task_struct *__switch_to(struct task_struct *prev, struct task_struct *ne
                 struct pt_regs *regs;
 
                 local_irq_save(flags);
-                regs = (struct pt_regs *)
-                        ((unsigned long)prev->thread_info
-                         + THREAD_SIZE - sizeof(struct pt_regs)
-#ifdef CONFIG_SH_DSP
-                         - sizeof(struct pt_dspregs)
-#endif
-                         - sizeof(unsigned long));
+                regs = task_pt_regs(prev);
                 if (user_mode(regs) && regs->regs[15] >= 0xc0000000) {
                         int offset = (int)regs->regs[15];
 
@@ -372,7 +342,7 @@ struct task_struct *__switch_to(struct task_struct *prev, struct task_struct *ne
          */
         asm volatile("ldc       %0, r7_bank"
                      : /* no output */
-                     : "r" (next->thread_info));
+                     : "r" (task_thread_info(next)));
 
 #ifdef CONFIG_MMU
         /* If no tasks are using the UBC, we're done */
diff --git a/arch/sh/kernel/ptrace.c b/arch/sh/kernel/ptrace.c
index 1a8be06519ec..3887b4f6feb2 100644
--- a/arch/sh/kernel/ptrace.c
+++ b/arch/sh/kernel/ptrace.c
@@ -41,12 +41,7 @@ static inline int get_stack_long(struct task_struct *task, int offset)
 {
         unsigned char *stack;
 
-        stack = (unsigned char *)
-                task->thread_info + THREAD_SIZE - sizeof(struct pt_regs)
-#ifdef CONFIG_SH_DSP
-                - sizeof(struct pt_dspregs)
-#endif
-                - sizeof(unsigned long);
+        stack = (unsigned char *)task_pt_regs(task);
         stack += offset;
         return (*((int *)stack));
 }
@@ -59,12 +54,7 @@ static inline int put_stack_long(struct task_struct *task, int offset,
 {
         unsigned char *stack;
 
-        stack = (unsigned char *)
-                task->thread_info + THREAD_SIZE - sizeof(struct pt_regs)
-#ifdef CONFIG_SH_DSP
-                - sizeof(struct pt_dspregs)
-#endif
-                - sizeof(unsigned long);
+        stack = (unsigned char *)task_pt_regs(task);
         stack += offset;
         *(unsigned long *) stack = data;
         return 0;
diff --git a/arch/sh/kernel/smp.c b/arch/sh/kernel/smp.c
index 59e49b18252c..62c7d1c0ad7b 100644
--- a/arch/sh/kernel/smp.c
+++ b/arch/sh/kernel/smp.c
@@ -103,7 +103,7 @@ int __cpu_up(unsigned int cpu)
         if (IS_ERR(tsk))
                 panic("Failed forking idle task for cpu %d\n", cpu);
         
-        tsk->thread_info->cpu = cpu;
+        task_thread_info(tsk)->cpu = cpu;
 
         cpu_set(cpu, cpu_online_map);
 
diff --git a/arch/sh64/kernel/process.c b/arch/sh64/kernel/process.c
index 419b5a710441..1da9c61d6823 100644
--- a/arch/sh64/kernel/process.c
+++ b/arch/sh64/kernel/process.c
@@ -744,7 +744,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
         }
 #endif
         /* Copy from sh version */
-        childregs = ((struct pt_regs *)(THREAD_SIZE + (unsigned long) p->thread_info )) - 1;
+        childregs = (struct pt_regs *)(THREAD_SIZE + task_stack_page(p)) - 1;
 
         *childregs = *regs;
 
@@ -752,7 +752,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
                 childregs->regs[15] = usp;
                 p->thread.uregs = childregs;
         } else {
-                childregs->regs[15] = (unsigned long)p->thread_info + THREAD_SIZE;
+                childregs->regs[15] = (unsigned long)task_stack_page(p) + THREAD_SIZE;
         }
 
         childregs->regs[9] = 0; /* Set return value for child */
diff --git a/arch/sh64/lib/dbg.c b/arch/sh64/lib/dbg.c
index 526fedae6db8..58087331b8a6 100644
--- a/arch/sh64/lib/dbg.c
+++ b/arch/sh64/lib/dbg.c
@@ -174,7 +174,7 @@ void evt_debug(int evt, int ret_addr, int event, int tra, struct pt_regs *regs)
         struct ring_node *rr;
 
         pid = current->pid;
-        stack_bottom = (unsigned long) current->thread_info;
+        stack_bottom = (unsigned long) task_stack_page(current);
         asm volatile("ori r15, 0, %0" : "=r" (sp));
         rr = event_ring + event_ptr;
         rr->evt = evt;
diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process.c
index ea8647411462..fbb05a452e51 100644
--- a/arch/sparc/kernel/process.c
+++ b/arch/sparc/kernel/process.c
@@ -302,7 +302,7 @@ void show_stack(struct task_struct *tsk, unsigned long *_ksp)
         int count = 0;
 
         if (tsk != NULL)
-                task_base = (unsigned long) tsk->thread_info;
+                task_base = (unsigned long) task_stack_page(tsk);
         else
                 task_base = (unsigned long) current_thread_info();
 
@@ -337,7 +337,7 @@ EXPORT_SYMBOL(dump_stack);
  */
 unsigned long thread_saved_pc(struct task_struct *tsk)
 {
-        return tsk->thread_info->kpc;
+        return task_thread_info(tsk)->kpc;
 }
 
 /*
@@ -392,7 +392,7 @@ void flush_thread(void)
                 /* We must fixup kregs as well. */
                 /* XXX This was not fixed for ti for a while, worked. Unused? */
                 current->thread.kregs = (struct pt_regs *)
-                    ((char *)current->thread_info + (THREAD_SIZE - TRACEREG_SZ));
+                    (task_stack_page(current) + (THREAD_SIZE - TRACEREG_SZ));
         }
 }
 
@@ -459,7 +459,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
                 unsigned long unused,
                 struct task_struct *p, struct pt_regs *regs)
 {
-        struct thread_info *ti = p->thread_info;
+        struct thread_info *ti = task_thread_info(p);
         struct pt_regs *childregs;
         char *new_stack;
 
@@ -482,7 +482,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
          *  V                      V (stk.fr.) V  (pt_regs)  { (stk.fr.) }
          *  +----- - - - - - ------+===========+============={+==========}+
          */
-        new_stack = (char*)ti + THREAD_SIZE;
+        new_stack = task_stack_page(p) + THREAD_SIZE;
         if (regs->psr & PSR_PS)
                 new_stack -= STACKFRAME_SZ;
         new_stack -= STACKFRAME_SZ + TRACEREG_SZ;
@@ -724,7 +724,7 @@ unsigned long get_wchan(struct task_struct *task)
             task->state == TASK_RUNNING)
                 goto out;
 
-        fp = task->thread_info->ksp + bias;
+        fp = task_thread_info(task)->ksp + bias;
         do {
                 /* Bogus frame pointer? */
                 if (fp < (task_base + sizeof(struct thread_info)) ||
diff --git a/arch/sparc/kernel/ptrace.c b/arch/sparc/kernel/ptrace.c
index fc470c0e9dc6..1baf13ed5c3a 100644
--- a/arch/sparc/kernel/ptrace.c
+++ b/arch/sparc/kernel/ptrace.c
@@ -75,7 +75,7 @@ static inline void read_sunos_user(struct pt_regs *regs, unsigned long offset,
                                    struct task_struct *tsk, long __user *addr)
 {
         struct pt_regs *cregs = tsk->thread.kregs;
-        struct thread_info *t = tsk->thread_info;
+        struct thread_info *t = task_thread_info(tsk);
         int v;
         
         if(offset >= 1024)
@@ -170,7 +170,7 @@ static inline void write_sunos_user(struct pt_regs *regs, unsigned long offset,
                                     struct task_struct *tsk)
 {
         struct pt_regs *cregs = tsk->thread.kregs;
-        struct thread_info *t = tsk->thread_info;
+        struct thread_info *t = task_thread_info(tsk);
         unsigned long value = regs->u_regs[UREG_I3];
 
         if(offset >= 1024)
diff --git a/arch/sparc/kernel/sun4d_smp.c b/arch/sparc/kernel/sun4d_smp.c
index cc1fc898495c..40d426cce824 100644
--- a/arch/sparc/kernel/sun4d_smp.c
+++ b/arch/sparc/kernel/sun4d_smp.c
@@ -200,7 +200,7 @@ void __init smp4d_boot_cpus(void)
                         /* Cook up an idler for this guy. */
                         p = fork_idle(i);
                         cpucount++;
-                        current_set[i] = p->thread_info;
+                        current_set[i] = task_thread_info(p);
                         for (no = 0; !cpu_find_by_instance(no, NULL, &mid)
                                      && mid != i; no++) ;
 
diff --git a/arch/sparc/kernel/sun4m_smp.c b/arch/sparc/kernel/sun4m_smp.c
index f113422a3727..a21f27d10e55 100644
--- a/arch/sparc/kernel/sun4m_smp.c
+++ b/arch/sparc/kernel/sun4m_smp.c
@@ -173,7 +173,7 @@ void __init smp4m_boot_cpus(void)
                         /* Cook up an idler for this guy. */
                         p = fork_idle(i);
                         cpucount++;
-                        current_set[i] = p->thread_info;
+                        current_set[i] = task_thread_info(p);
                         /* See trampoline.S for details... */
                         entry += ((i-1) * 3);
 
diff --git a/arch/sparc/kernel/traps.c b/arch/sparc/kernel/traps.c
index 3f451ae66482..41d45c298fb2 100644
--- a/arch/sparc/kernel/traps.c
+++ b/arch/sparc/kernel/traps.c
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 #ifndef CONFIG_SMP
         if(!fpt) {
 #else
-        if(!(fpt->thread_info->flags & _TIF_USEDFPU)) {
+        if(!(task_thread_info(fpt)->flags & _TIF_USEDFPU)) {
 #endif
                 fpsave(&fake_regs[0], &fake_fsr, &fake_queue[0], &fake_depth);
                 regs->psr &= ~PSR_EF;
@@ -334,7 +334,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
         /* nope, better SIGFPE the offending process... */
                
 #ifdef CONFIG_SMP
-        fpt->thread_info->flags &= ~_TIF_USEDFPU;
+        task_thread_info(fpt)->flags &= ~_TIF_USEDFPU;
 #endif
         if(psr & PSR_PS) {
                 /* The first fsr store/load we tried trapped,
diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index 02f9dec1d459..1dc3650c5cae 100644
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -390,7 +390,7 @@ void show_regs32(struct pt_regs32 *regs)
 
 unsigned long thread_saved_pc(struct task_struct *tsk)
 {
-        struct thread_info *ti = tsk->thread_info;
+        struct thread_info *ti = task_thread_info(tsk);
         unsigned long ret = 0xdeadbeefUL;
         
         if (ti && ti->ksp) {
@@ -616,11 +616,11 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
                 unsigned long unused,
                 struct task_struct *p, struct pt_regs *regs)
 {
-        struct thread_info *t = p->thread_info;
+        struct thread_info *t = task_thread_info(p);
         char *child_trap_frame;
 
         /* Calculate offset to stack_frame & pt_regs */
-        child_trap_frame = ((char *)t) + (THREAD_SIZE - (TRACEREG_SZ+STACKFRAME_SZ));
+        child_trap_frame = task_stack_page(p) + (THREAD_SIZE - (TRACEREG_SZ+STACKFRAME_SZ));
         memcpy(child_trap_frame, (((struct sparc_stackf *)regs)-1), (TRACEREG_SZ+STACKFRAME_SZ));
 
         t->flags = (t->flags & ~((0xffUL << TI_FLAG_CWP_SHIFT) | (0xffUL << TI_FLAG_CURRENT_DS_SHIFT))) |
@@ -845,9 +845,9 @@ unsigned long get_wchan(struct task_struct *task)
             task->state == TASK_RUNNING)
                 goto out;
 
-        thread_info_base = (unsigned long) task->thread_info;
+        thread_info_base = (unsigned long) task_stack_page(task);
         bias = STACK_BIAS;
-        fp = task->thread_info->ksp + bias;
+        fp = task_thread_info(task)->ksp + bias;
 
         do {
                 /* Bogus frame pointer? */
diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c
index 84d3df2264cb..3f9746f856d2 100644
--- a/arch/sparc64/kernel/ptrace.c
+++ b/arch/sparc64/kernel/ptrace.c
@@ -296,7 +296,7 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
         case PTRACE_GETREGS: {
                 struct pt_regs32 __user *pregs =
                         (struct pt_regs32 __user *) addr;
-                struct pt_regs *cregs = child->thread_info->kregs;
+                struct pt_regs *cregs = task_pt_regs(child);
                 int rval;
 
                 if (__put_user(tstate_to_psr(cregs->tstate), (&pregs->psr)) ||
@@ -320,11 +320,11 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
 
         case PTRACE_GETREGS64: {
                 struct pt_regs __user *pregs = (struct pt_regs __user *) addr;
-                struct pt_regs *cregs = child->thread_info->kregs;
+                struct pt_regs *cregs = task_pt_regs(child);
                 unsigned long tpc = cregs->tpc;
                 int rval;
 
-                if ((child->thread_info->flags & _TIF_32BIT) != 0)
+                if ((task_thread_info(child)->flags & _TIF_32BIT) != 0)
                         tpc &= 0xffffffff;
                 if (__put_user(cregs->tstate, (&pregs->tstate)) ||
                     __put_user(tpc, (&pregs->tpc)) ||
@@ -348,7 +348,7 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
         case PTRACE_SETREGS: {
                 struct pt_regs32 __user *pregs =
                         (struct pt_regs32 __user *) addr;
-                struct pt_regs *cregs = child->thread_info->kregs;
+                struct pt_regs *cregs = task_pt_regs(child);
                 unsigned int psr, pc, npc, y;
                 int i;
 
@@ -381,7 +381,7 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
 
         case PTRACE_SETREGS64: {
                 struct pt_regs __user *pregs = (struct pt_regs __user *) addr;
-                struct pt_regs *cregs = child->thread_info->kregs;
+                struct pt_regs *cregs = task_pt_regs(child);
                 unsigned long tstate, tpc, tnpc, y;
                 int i;
 
@@ -395,7 +395,7 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         pt_error_return(regs, EFAULT);
                         goto out_tsk;
                 }
-                if ((child->thread_info->flags & _TIF_32BIT) != 0) {
+                if ((task_thread_info(child)->flags & _TIF_32BIT) != 0) {
                         tpc &= 0xffffffff;
                         tnpc &= 0xffffffff;
                 }
@@ -430,11 +430,11 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         } fpq[16];
                 };
                 struct fps __user *fps = (struct fps __user *) addr;
-                unsigned long *fpregs = child->thread_info->fpregs;
+                unsigned long *fpregs = task_thread_info(child)->fpregs;
 
                 if (copy_to_user(&fps->regs[0], fpregs,
                                  (32 * sizeof(unsigned int))) ||
-                    __put_user(child->thread_info->xfsr[0], (&fps->fsr)) ||
+                    __put_user(task_thread_info(child)->xfsr[0], (&fps->fsr)) ||
                     __put_user(0, (&fps->fpqd)) ||
                     __put_user(0, (&fps->flags)) ||
                     __put_user(0, (&fps->extra)) ||
@@ -452,11 +452,11 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         unsigned long fsr;
                 };
                 struct fps __user *fps = (struct fps __user *) addr;
-                unsigned long *fpregs = child->thread_info->fpregs;
+                unsigned long *fpregs = task_thread_info(child)->fpregs;
 
                 if (copy_to_user(&fps->regs[0], fpregs,
                                  (64 * sizeof(unsigned int))) ||
-                    __put_user(child->thread_info->xfsr[0], (&fps->fsr))) {
+                    __put_user(task_thread_info(child)->xfsr[0], (&fps->fsr))) {
                         pt_error_return(regs, EFAULT);
                         goto out_tsk;
                 }
@@ -477,7 +477,7 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         } fpq[16];
                 };
                 struct fps __user *fps = (struct fps __user *) addr;
-                unsigned long *fpregs = child->thread_info->fpregs;
+                unsigned long *fpregs = task_thread_info(child)->fpregs;
                 unsigned fsr;
 
                 if (copy_from_user(fpregs, &fps->regs[0],
@@ -486,11 +486,11 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         pt_error_return(regs, EFAULT);
                         goto out_tsk;
                 }
-                child->thread_info->xfsr[0] &= 0xffffffff00000000UL;
-                child->thread_info->xfsr[0] |= fsr;
-                if (!(child->thread_info->fpsaved[0] & FPRS_FEF))
-                        child->thread_info->gsr[0] = 0;
-                child->thread_info->fpsaved[0] |= (FPRS_FEF | FPRS_DL);
+                task_thread_info(child)->xfsr[0] &= 0xffffffff00000000UL;
+                task_thread_info(child)->xfsr[0] |= fsr;
+                if (!(task_thread_info(child)->fpsaved[0] & FPRS_FEF))
+                        task_thread_info(child)->gsr[0] = 0;
+                task_thread_info(child)->fpsaved[0] |= (FPRS_FEF | FPRS_DL);
                 pt_succ_return(regs, 0);
                 goto out_tsk;
         }
@@ -501,17 +501,17 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
                         unsigned long fsr;
                 };
                 struct fps __user *fps = (struct fps __user *) addr;
-                unsigned long *fpregs = child->thread_info->fpregs;
+                unsigned long *fpregs = task_thread_info(child)->fpregs;
 
                 if (copy_from_user(fpregs, &fps->regs[0],
                                    (64 * sizeof(unsigned int))) ||
-                    __get_user(child->thread_info->xfsr[0], (&fps->fsr))) {
+                    __get_user(task_thread_info(child)->xfsr[0], (&fps->fsr))) {
                         pt_error_return(regs, EFAULT);
                         goto out_tsk;
                 }
-                if (!(child->thread_info->fpsaved[0] & FPRS_FEF))
-                        child->thread_info->gsr[0] = 0;
-                child->thread_info->fpsaved[0] |= (FPRS_FEF | FPRS_DL | FPRS_DU);
+                if (!(task_thread_info(child)->fpsaved[0] & FPRS_FEF))
+                        task_thread_info(child)->gsr[0] = 0;
+                task_thread_info(child)->fpsaved[0] |= (FPRS_FEF | FPRS_DL | FPRS_DU);
                 pt_succ_return(regs, 0);
                 goto out_tsk;
         }
@@ -562,8 +562,8 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
 #ifdef DEBUG_PTRACE
                 printk("CONT: %s [%d]: set exit_code = %x %lx %lx\n", child->comm,
                         child->pid, child->exit_code,
-                        child->thread_info->kregs->tpc,
-                        child->thread_info->kregs->tnpc);
+                        task_pt_regs(child)->tpc,
+                        task_pt_regs(child)->tnpc);
                        
 #endif
                 wake_up_process(child);
diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c
index 48180531562f..250745896aee 100644
--- a/arch/sparc64/kernel/setup.c
+++ b/arch/sparc64/kernel/setup.c
@@ -520,7 +520,7 @@ void __init setup_arch(char **cmdline_p)
         rd_doload = ((ram_flags & RAMDISK_LOAD_FLAG) != 0);     
 #endif
 
-        init_task.thread_info->kregs = &fake_swapper_regs;
+        task_thread_info(&init_task)->kregs = &fake_swapper_regs;
 
 #ifdef CONFIG_IP_PNP
         if (!ic_set_manually) {
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index 6efc03df51c3..1fb6323e65a4 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -335,7 +335,7 @@ static int __devinit smp_boot_one_cpu(unsigned int cpu)
 
         p = fork_idle(cpu);
         callin_flag = 0;
-        cpu_new_thread = p->thread_info;
+        cpu_new_thread = task_thread_info(p);
         cpu_set(cpu, cpu_callout_map);
 
         cpu_find_by_mid(cpu, &cpu_node);
diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c
index 5570e7bb22bb..8d44ae5a15e3 100644
--- a/arch/sparc64/kernel/traps.c
+++ b/arch/sparc64/kernel/traps.c
@@ -1808,7 +1808,7 @@ static void user_instruction_dump (unsigned int __user *pc)
 void show_stack(struct task_struct *tsk, unsigned long *_ksp)
 {
         unsigned long pc, fp, thread_base, ksp;
-        struct thread_info *tp = tsk->thread_info;
+        void *tp = task_stack_page(tsk);
         struct reg_window *rw;
         int count = 0;
 
@@ -1862,7 +1862,7 @@ static inline int is_kernel_stack(struct task_struct *task,
                         return 0;
         }
 
-        thread_base = (unsigned long) task->thread_info;
+        thread_base = (unsigned long) task_stack_page(task);
         thread_end = thread_base + sizeof(union thread_union);
         if (rw_addr >= thread_base &&
             rw_addr < thread_end &&
diff --git a/arch/um/kernel/process_kern.c b/arch/um/kernel/process_kern.c
index d2d3f256778c..7f13b85d2656 100644
--- a/arch/um/kernel/process_kern.c
+++ b/arch/um/kernel/process_kern.c
@@ -107,7 +107,7 @@ void set_current(void *t)
 {
         struct task_struct *task = t;
 
-        cpu_tasks[task->thread_info->cpu] = ((struct cpu_task) 
+        cpu_tasks[task_thread_info(task)->cpu] = ((struct cpu_task)
                 { external_pid(task), task });
 }
 
diff --git a/arch/um/kernel/skas/process_kern.c b/arch/um/kernel/skas/process_kern.c
index 09790ccb161c..dc41c6dc2f34 100644
--- a/arch/um/kernel/skas/process_kern.c
+++ b/arch/um/kernel/skas/process_kern.c
@@ -118,7 +118,7 @@ int copy_thread_skas(int nr, unsigned long clone_flags, unsigned long sp,
                 handler = new_thread_handler;
         }
 
-        new_thread(p->thread_info, &p->thread.mode.skas.switch_buf,
+        new_thread(task_stack_page(p), &p->thread.mode.skas.switch_buf,
                    &p->thread.mode.skas.fork_buf, handler);
         return(0);
 }
@@ -185,7 +185,7 @@ int start_uml_skas(void)
 
         init_task.thread.request.u.thread.proc = start_kernel_proc;
         init_task.thread.request.u.thread.arg = NULL;
-        return(start_idle_thread(init_task.thread_info,
+        return(start_idle_thread(task_stack_page(&init_task),
                                  &init_task.thread.mode.skas.switch_buf,
                                  &init_task.thread.mode.skas.fork_buf));
 }
diff --git a/arch/um/kernel/tt/exec_kern.c b/arch/um/kernel/tt/exec_kern.c
index 136e54c47d37..8f40e4838736 100644
--- a/arch/um/kernel/tt/exec_kern.c
+++ b/arch/um/kernel/tt/exec_kern.c
@@ -39,7 +39,7 @@ void flush_thread_tt(void)
                 do_exit(SIGKILL);
         }
                 
-        new_pid = start_fork_tramp(current->thread_info, stack, 0, exec_tramp);
+        new_pid = start_fork_tramp(task_stack_page(current), stack, 0, exec_tramp);
         if(new_pid < 0){
                 printk(KERN_ERR 
                        "flush_thread : new thread failed, errno = %d\n",
diff --git a/arch/um/kernel/tt/process_kern.c b/arch/um/kernel/tt/process_kern.c
index 14d4622a5fb8..62535303aa27 100644
--- a/arch/um/kernel/tt/process_kern.c
+++ b/arch/um/kernel/tt/process_kern.c
@@ -36,7 +36,7 @@ void switch_to_tt(void *prev, void *next)
         from = prev;
         to = next;
 
-        cpu = from->thread_info->cpu;
+        cpu = task_thread_info(from)->cpu;
         if(cpu == 0)
                 forward_interrupts(to->thread.mode.tt.extern_pid);
 #ifdef CONFIG_SMP
@@ -253,7 +253,7 @@ int copy_thread_tt(int nr, unsigned long clone_flags, unsigned long sp,
 
         clone_flags &= CLONE_VM;
         p->thread.temp_stack = stack;
-        new_pid = start_fork_tramp(p->thread_info, stack, clone_flags, tramp);
+        new_pid = start_fork_tramp(task_stack_page(p), stack, clone_flags, tramp);
         if(new_pid < 0){
                 printk(KERN_ERR "copy_thread : clone failed - errno = %d\n", 
                        -new_pid);
@@ -343,7 +343,7 @@ int do_proc_op(void *t, int proc_id)
                 pid = thread->request.u.exec.pid;
                 do_exec(thread->mode.tt.extern_pid, pid);
                 thread->mode.tt.extern_pid = pid;
-                cpu_tasks[task->thread_info->cpu].pid = pid;
+                cpu_tasks[task_thread_info(task)->cpu].pid = pid;
                 break;
         case OP_FORK:
                 attach_process(thread->request.u.fork.pid);
@@ -425,7 +425,7 @@ int start_uml_tt(void)
         int pages;
 
         pages = (1 << CONFIG_KERNEL_STACK_ORDER);
-        sp = (void *) ((unsigned long) init_task.thread_info) +
+        sp = task_stack_page(&init_task) +
                 pages * PAGE_SIZE - sizeof(unsigned long);
         return(tracer(start_kernel_proc, sp));
 }
diff --git a/arch/v850/kernel/process.c b/arch/v850/kernel/process.c
index 062ffa0a9998..eb909937958b 100644
--- a/arch/v850/kernel/process.c
+++ b/arch/v850/kernel/process.c
@@ -114,7 +114,7 @@ int copy_thread (int nr, unsigned long clone_flags,
                  struct task_struct *p, struct pt_regs *regs)
 {
         /* Start pushing stuff from the top of the child's kernel stack.  */
-        unsigned long orig_ksp = (unsigned long)p->thread_info + THREAD_SIZE;
+        unsigned long orig_ksp = task_tos(p);
         unsigned long ksp = orig_ksp;
         /* We push two `state save' stack fames (see entry.S) on the new
            kernel stack:
diff --git a/arch/v850/kernel/ptrace.c b/arch/v850/kernel/ptrace.c
index 18492d02aaf6..67e057509664 100644
--- a/arch/v850/kernel/ptrace.c
+++ b/arch/v850/kernel/ptrace.c
@@ -58,7 +58,7 @@ static v850_reg_t *reg_save_addr (unsigned reg_offs, struct task_struct *t)
                 regs = thread_saved_regs (t);
         else
                 /* Register saved during kernel entry (or not available).  */
-                regs = task_regs (t);
+                regs = task_pt_regs (t);
 
         return (v850_reg_t *)((char *)regs + reg_offs);
 }
diff --git a/arch/x86_64/ia32/ia32_binfmt.c b/arch/x86_64/ia32/ia32_binfmt.c
index 2b760d0d9ce2..029bddab0459 100644
--- a/arch/x86_64/ia32/ia32_binfmt.c
+++ b/arch/x86_64/ia32/ia32_binfmt.c
@@ -197,8 +197,7 @@ static inline void elf_core_copy_regs(elf_gregset_t *elfregs, struct pt_regs *re
 
 static inline int elf_core_copy_task_regs(struct task_struct *t, elf_gregset_t* elfregs)
 {       
-        struct pt_regs *pp = (struct pt_regs *)(t->thread.rsp0);
-        --pp;
+        struct pt_regs *pp = task_pt_regs(t);
         ELF_CORE_COPY_REGS((*elfregs), pp);
         /* fix wrong segments */ 
         (*elfregs)[7] = t->thread.ds; 
@@ -217,7 +216,7 @@ elf_core_copy_task_fpregs(struct task_struct *tsk, struct pt_regs *regs, elf_fpr
         if (!tsk_used_math(tsk))
                 return 0;
         if (!regs)
-                regs = ((struct pt_regs *)tsk->thread.rsp0) - 1;
+                regs = task_pt_regs(tsk);
         if (tsk == current)
                 unlazy_fpu(tsk);
         set_fs(KERNEL_DS); 
@@ -233,7 +232,7 @@ elf_core_copy_task_fpregs(struct task_struct *tsk, struct pt_regs *regs, elf_fpr
 static inline int 
 elf_core_copy_task_xfpregs(struct task_struct *t, elf_fpxregset_t *xfpu)
 {
-        struct pt_regs *regs = ((struct pt_regs *)(t->thread.rsp0))-1; 
+        struct pt_regs *regs = task_pt_regs(t);
         if (!tsk_used_math(t))
                 return 0;
         if (t == current)
diff --git a/arch/x86_64/ia32/ptrace32.c b/arch/x86_64/ia32/ptrace32.c
index ea4394e021d6..23a4515a73b4 100644
--- a/arch/x86_64/ia32/ptrace32.c
+++ b/arch/x86_64/ia32/ptrace32.c
@@ -41,7 +41,7 @@
 static int putreg32(struct task_struct *child, unsigned regno, u32 val)
 {
         int i;
-        __u64 *stack = (__u64 *)(child->thread.rsp0 - sizeof(struct pt_regs)); 
+        __u64 *stack = (__u64 *)task_pt_regs(child);
 
         switch (regno) {
         case offsetof(struct user32, regs.fs):
@@ -137,7 +137,7 @@ static int putreg32(struct task_struct *child, unsigned regno, u32 val)
 
 static int getreg32(struct task_struct *child, unsigned regno, u32 *val)
 {
-        __u64 *stack = (__u64 *)(child->thread.rsp0 - sizeof(struct pt_regs)); 
+        __u64 *stack = (__u64 *)task_pt_regs(child);
 
         switch (regno) {
         case offsetof(struct user32, regs.fs):
@@ -238,7 +238,7 @@ asmlinkage long sys32_ptrace(long request, u32 pid, u32 addr, u32 data)
         if (ret < 0)
                 goto out;
 
-        childregs = (struct pt_regs *)(child->thread.rsp0 - sizeof(struct pt_regs)); 
+        childregs = task_pt_regs(child);
 
         switch (request) {
         case PTRACE_PEEKDATA:
diff --git a/arch/x86_64/kernel/i387.c b/arch/x86_64/kernel/i387.c
index d9b22b633e39..a5d7e16b928e 100644
--- a/arch/x86_64/kernel/i387.c
+++ b/arch/x86_64/kernel/i387.c
@@ -95,7 +95,7 @@ int save_i387(struct _fpstate __user *buf)
         if (!used_math())
                 return 0;
         clear_used_math(); /* trigger finit */
-        if (tsk->thread_info->status & TS_USEDFPU) {
+        if (task_thread_info(tsk)->status & TS_USEDFPU) {
                 err = save_i387_checking((struct i387_fxsave_struct __user *)buf);
                 if (err) return err;
                 stts();
diff --git a/arch/x86_64/kernel/i8259.c b/arch/x86_64/kernel/i8259.c
index b61965f0fb34..5ecd34ab8c2b 100644
--- a/arch/x86_64/kernel/i8259.c
+++ b/arch/x86_64/kernel/i8259.c
@@ -133,7 +133,7 @@ static void end_8259A_irq (unsigned int irq)
 {
         if (irq > 256) { 
                 char var;
-                printk("return %p stack %p ti %p\n", __builtin_return_address(0), &var, current->thread_info); 
+                printk("return %p stack %p ti %p\n", __builtin_return_address(0), &var, task_thread_info(current));
 
                 BUG(); 
         }
diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c
index 669cf0ed3266..8ded407e4a94 100644
--- a/arch/x86_64/kernel/process.c
+++ b/arch/x86_64/kernel/process.c
@@ -451,7 +451,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long rsp,
         struct task_struct *me = current;
 
         childregs = ((struct pt_regs *)
-                        (THREAD_SIZE + (unsigned long) p->thread_info)) - 1;
+                        (THREAD_SIZE + task_stack_page(p))) - 1;
         *childregs = *regs;
 
         childregs->rax = 0;
@@ -463,7 +463,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long rsp,
         p->thread.rsp0 = (unsigned long) (childregs+1);
         p->thread.userrsp = me->thread.userrsp; 
 
-        set_ti_thread_flag(p->thread_info, TIF_FORK);
+        set_tsk_thread_flag(p, TIF_FORK);
 
         p->thread.fs = me->thread.fs;
         p->thread.gs = me->thread.gs;
@@ -590,7 +590,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
         write_pda(oldrsp, next->userrsp); 
         write_pda(pcurrent, next_p); 
         write_pda(kernelstack,
-            (unsigned long)next_p->thread_info + THREAD_SIZE - PDA_STACKOFFSET);
+                  task_stack_page(next_p) + THREAD_SIZE - PDA_STACKOFFSET);
 
         /*
          * Now maybe reload the debug registers
@@ -704,7 +704,7 @@ unsigned long get_wchan(struct task_struct *p)
 
         if (!p || p == current || p->state==TASK_RUNNING)
                 return 0; 
-        stack = (unsigned long)p->thread_info; 
+        stack = (unsigned long)task_stack_page(p);
         if (p->thread.rsp < stack || p->thread.rsp > stack+THREAD_SIZE)
                 return 0;
         fp = *(u64 *)(p->thread.rsp);
@@ -822,8 +822,7 @@ int dump_task_regs(struct task_struct *tsk, elf_gregset_t *regs)
 {
         struct pt_regs *pp, ptregs;
 
-        pp = (struct pt_regs *)(tsk->thread.rsp0);
-        --pp; 
+        pp = task_pt_regs(tsk);
 
         ptregs = *pp; 
         ptregs.cs &= 0xffff;
diff --git a/arch/x86_64/kernel/ptrace.c b/arch/x86_64/kernel/ptrace.c
index 86248bc9303e..53205622351c 100644
--- a/arch/x86_64/kernel/ptrace.c
+++ b/arch/x86_64/kernel/ptrace.c
@@ -67,12 +67,6 @@ static inline unsigned long get_stack_long(struct task_struct *task, int offset)
         return (*((unsigned long *)stack));
 }
 
-static inline struct pt_regs *get_child_regs(struct task_struct *task)
-{
-        struct pt_regs *regs = (void *)task->thread.rsp0;
-        return regs - 1;
-}
-
 /*
  * this routine will put a word on the processes privileged stack. 
  * the offset is how far from the base addr as stored in the TSS.  
@@ -170,7 +164,7 @@ static int is_at_popf(struct task_struct *child, struct pt_regs *regs)
 
 static void set_singlestep(struct task_struct *child)
 {
-        struct pt_regs *regs = get_child_regs(child);
+        struct pt_regs *regs = task_pt_regs(child);
 
         /*
          * Always set TIF_SINGLESTEP - this guarantees that
@@ -208,7 +202,7 @@ static void clear_singlestep(struct task_struct *child)
 
         /* But touch TF only if it was set by us.. */
         if (child->ptrace & PT_DTRACE) {
-                struct pt_regs *regs = get_child_regs(child);
+                struct pt_regs *regs = task_pt_regs(child);
                 regs->eflags &= ~TRAP_FLAG;
                 child->ptrace &= ~PT_DTRACE;
         }
diff --git a/arch/x86_64/kernel/smpboot.c b/arch/x86_64/kernel/smpboot.c
index c9e941ae5019..a28756ef7cef 100644
--- a/arch/x86_64/kernel/smpboot.c
+++ b/arch/x86_64/kernel/smpboot.c
@@ -776,7 +776,7 @@ static int __cpuinit do_boot_cpu(int cpu, int apicid)
 
         if (c_idle.idle) {
                 c_idle.idle->thread.rsp = (unsigned long) (((struct pt_regs *)
-                        (THREAD_SIZE + (unsigned long) c_idle.idle->thread_info)) - 1);
+                        (THREAD_SIZE +  task_stack_page(c_idle.idle))) - 1);
                 init_idle(c_idle.idle, cpu);
                 goto do_rest;
         }
@@ -814,7 +814,7 @@ do_rest:
         init_rsp = c_idle.idle->thread.rsp;
         per_cpu(init_tss,cpu).rsp0 = init_rsp;
         initial_code = start_secondary;
-        clear_ti_thread_flag(c_idle.idle->thread_info, TIF_FORK);
+        clear_tsk_thread_flag(c_idle.idle, TIF_FORK);
 
         printk(KERN_INFO "Booting processor %d/%d APIC 0x%x\n", cpu,
                 cpus_weight(cpu_present_map),
diff --git a/arch/x86_64/kernel/traps.c b/arch/x86_64/kernel/traps.c
index 2671fd46ea85..8bb0aeda78b9 100644
--- a/arch/x86_64/kernel/traps.c
+++ b/arch/x86_64/kernel/traps.c
@@ -308,7 +308,7 @@ void show_registers(struct pt_regs *regs)
         printk("CPU %d ", cpu);
         __show_regs(regs);
         printk("Process %s (pid: %d, threadinfo %p, task %p)\n",
-                cur->comm, cur->pid, cur->thread_info, cur);
+                cur->comm, cur->pid, task_thread_info(cur), cur);
 
         /*
          * When in-kernel, we also print out the stack and code at the
@@ -666,7 +666,7 @@ asmlinkage struct pt_regs *sync_regs(struct pt_regs *eregs)
                 ;
         /* Exception from user space */
         else if (user_mode(eregs))
-                regs = ((struct pt_regs *)current->thread.rsp0) - 1;
+                regs = task_pt_regs(current);
         /* Exception from kernel and interrupts are enabled. Move to
            kernel process stack. */
         else if (eregs->eflags & X86_EFLAGS_IF)
@@ -912,7 +912,7 @@ asmlinkage void math_state_restore(void)
         if (!used_math())
                 init_fpu(me);
         restore_fpu_checking(&me->thread.i387.fxsave);
-        me->thread_info->status |= TS_USEDFPU;
+        task_thread_info(me)->status |= TS_USEDFPU;
 }
 
 void __init trap_init(void)
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index 6a44b54ae817..f1f596644bfc 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -145,7 +145,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
         int user_mode = user_mode(regs);
 
         /* Set up new TSS. */
-        tos = (unsigned long)p->thread_info + THREAD_SIZE;
+        tos = (unsigned long)task_stack_page(p) + THREAD_SIZE;
         if (user_mode)
                 childregs = (struct pt_regs*)(tos - PT_USER_SIZE);
         else
@@ -217,7 +217,7 @@ int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
 unsigned long get_wchan(struct task_struct *p)
 {
         unsigned long sp, pc;
-        unsigned long stack_page = (unsigned long) p->thread_info;
+        unsigned long stack_page = (unsigned long) task_stack_page(p);
         int count = 0;
 
         if (!p || p == current || p->state == TASK_RUNNING)
diff --git a/arch/xtensa/kernel/ptrace.c b/arch/xtensa/kernel/ptrace.c
index ab5c4c65b5c4..4cc85285a70a 100644
--- a/arch/xtensa/kernel/ptrace.c
+++ b/arch/xtensa/kernel/ptrace.c
@@ -72,7 +72,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                 struct pt_regs *regs;
                 unsigned long tmp;
 
-                regs = xtensa_pt_regs(child);
+                regs = task_pt_regs(child);
                 tmp = 0;  /* Default return value. */
 
                 switch(addr) {
@@ -149,7 +149,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
         case PTRACE_POKEUSR:
                 {
                 struct pt_regs *regs;
-                regs = xtensa_pt_regs(child);
+                regs = task_pt_regs(child);
 
                 switch (addr) {
                 case REG_AR_BASE ... REG_AR_BASE + XCHAL_NUM_AREGS - 1:
@@ -240,7 +240,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                  * elf_gregset_t format. */
 
                 xtensa_gregset_t format;
-                struct pt_regs *regs = xtensa_pt_regs(child);
+                struct pt_regs *regs = task_pt_regs(child);
 
                 do_copy_regs (&format, regs, child);
 
@@ -257,7 +257,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                  * values in the elf_gregset_t format. */
 
                 xtensa_gregset_t format;
-                struct pt_regs *regs = xtensa_pt_regs(child);
+                struct pt_regs *regs = task_pt_regs(child);
 
                 if (copy_from_user(&format,(void *)data,sizeof(elf_gregset_t))){
                         ret = -EFAULT;
@@ -281,7 +281,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                  * elf_fpregset_t format. */
 
                 elf_fpregset_t fpregs;
-                struct pt_regs *regs = xtensa_pt_regs(child);
+                struct pt_regs *regs = task_pt_regs(child);
 
                 do_save_fpregs (&fpregs, regs, child);
 
@@ -299,7 +299,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data)
                  * values in the elf_fpregset_t format.
                  */
                 elf_fpregset_t fpregs;
-                struct pt_regs *regs = xtensa_pt_regs(child);
+                struct pt_regs *regs = task_pt_regs(child);
 
                 ret = 0;
                 if (copy_from_user(&fpregs, (void *)data, sizeof(elf_fpregset_t))) {
diff --git a/block/elevator.c b/block/elevator.c
index 99a4d7b2f8ad..1d0759178e4b 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -610,23 +610,23 @@ void elv_completed_request(request_queue_t *q, struct request *rq)
          * request is released from the driver, io must be done
          */
         if (blk_account_rq(rq)) {
-                struct request *first_rq = list_entry_rq(q->queue_head.next);
-
                 q->in_flight--;
+                if (blk_sorted_rq(rq) && e->ops->elevator_completed_req_fn)
+                        e->ops->elevator_completed_req_fn(q, rq);
+        }
 
-                /*
-                 * Check if the queue is waiting for fs requests to be
-                 * drained for flush sequence.
-                 */
-                if (q->ordseq && q->in_flight == 0 &&
+        /*
+         * Check if the queue is waiting for fs requests to be
+         * drained for flush sequence.
+         */
+        if (unlikely(q->ordseq)) {
+                struct request *first_rq = list_entry_rq(q->queue_head.next);
+                if (q->in_flight == 0 &&
                     blk_ordered_cur_seq(q) == QUEUE_ORDSEQ_DRAIN &&
                     blk_ordered_req_seq(first_rq) > QUEUE_ORDSEQ_DRAIN) {
                         blk_ordered_complete_seq(q, QUEUE_ORDSEQ_DRAIN, 0);
                         q->request_fn(q);
                 }
-
-                if (blk_sorted_rq(rq) && e->ops->elevator_completed_req_fn)
-                        e->ops->elevator_completed_req_fn(q, rq);
         }
 }
 
diff --git a/drivers/block/amiflop.c b/drivers/block/amiflop.c
index 3c679d30b698..b6e290956214 100644
--- a/drivers/block/amiflop.c
+++ b/drivers/block/amiflop.c
@@ -194,6 +194,8 @@ static DECLARE_WAIT_QUEUE_HEAD(ms_wait);
  */
 #define MAX_ERRORS 12
 
+#define custom amiga_custom
+
 /* Prevent "aliased" accesses. */
 static int fd_ref[4] = { 0,0,0,0 };
 static int fd_device[4] = { 0, 0, 0, 0 };
@@ -1439,6 +1441,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
 {
         int drive = iminor(inode) & 3;
         static struct floppy_struct getprm;
+        void __user *argp = (void __user *)param;
 
         switch(cmd){
         case FDFMTBEG:
@@ -1484,9 +1487,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
                 getprm.head=unit[drive].type->heads;
                 getprm.sect=unit[drive].dtype->sects * unit[drive].type->sect_mult;
                 getprm.size=unit[drive].blocks;
-                if (copy_to_user((void *)param,
-                                 (void *)&getprm,
-                                 sizeof(struct floppy_struct)))
+                if (copy_to_user(argp, &getprm, sizeof(struct floppy_struct)))
                         return -EFAULT;
                 break;
         case FDSETPRM:
@@ -1498,8 +1499,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
                 break;
 #ifdef RAW_IOCTL
         case IOCTL_RAW_TRACK:
-                if (copy_to_user((void *)param, raw_buf,
-                                 unit[drive].type->read_size))
+                if (copy_to_user(argp, raw_buf, unit[drive].type->read_size))
                         return -EFAULT;
                 else
                         return unit[drive].type->read_size;
@@ -1654,12 +1654,6 @@ static struct block_device_operations floppy_fops = {
         .media_changed  = amiga_floppy_change,
 };
 
-void __init amiga_floppy_setup (char *str, int *ints)
-{
-        printk (KERN_INFO "amiflop: Setting default df0 to %x\n", ints[1]);
-        fd_def_df0 = ints[1];
-}
-
 static int __init fd_probe_drives(void)
 {
         int drive,drives,nomem;
@@ -1845,4 +1839,18 @@ void cleanup_module(void)
         unregister_blkdev(FLOPPY_MAJOR, "fd");
 }
 #endif
+
+#else
+static int __init amiga_floppy_setup (char *str)
+{
+        int n;
+        if (!MACH_IS_AMIGA)
+                return 0;
+        if (!get_option(&str, &n))
+                return 0;
+        printk (KERN_INFO "amiflop: Setting default df0 to %x\n", n);
+        fd_def_df0 = n;
+}
+
+__setup("floppy=", amiga_floppy_setup);
 #endif
diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c
index 3aa68a5447d6..f8ce235ccfc3 100644
--- a/drivers/block/ataflop.c
+++ b/drivers/block/ataflop.c
@@ -1361,7 +1361,7 @@ static int floppy_revalidate(struct gendisk *disk)
                    formats, for 'permanent user-defined' parameter:
                    restore default_params[] here if flagged valid! */
                 if (default_params[drive].blocks == 0)
-                        UDT = 0;
+                        UDT = NULL;
                 else
                         UDT = &default_params[drive];
         }
@@ -1495,6 +1495,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
         struct floppy_struct getprm;
         int settype;
         struct floppy_struct setprm;
+        void __user *argp = (void __user *)param;
 
         switch (cmd) {
         case FDGETPRM:
@@ -1521,7 +1522,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
                 getprm.head = 2;
                 getprm.track = dtp->blocks/dtp->spt/2;
                 getprm.stretch = dtp->stretch;
-                if (copy_to_user((void *)param, &getprm, sizeof(getprm)))
+                if (copy_to_user(argp, &getprm, sizeof(getprm)))
                         return -EFAULT;
                 return 0;
         }
@@ -1540,7 +1541,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
                 /* get the parameters from user space */
                 if (floppy->ref != 1 && floppy->ref != -1)
                         return -EBUSY;
-                if (copy_from_user(&setprm, (void *) param, sizeof(setprm)))
+                if (copy_from_user(&setprm, argp, sizeof(setprm)))
                         return -EFAULT;
                 /* 
                  * first of all: check for floppy change and revalidate, 
@@ -1647,7 +1648,7 @@ static int fd_ioctl(struct inode *inode, struct file *filp,
         case FDFMTTRK:
                 if (floppy->ref != 1 && floppy->ref != -1)
                         return -EBUSY;
-                if (copy_from_user(&fmt_desc, (void *) param, sizeof(fmt_desc)))
+                if (copy_from_user(&fmt_desc, argp, sizeof(fmt_desc)))
                         return -EFAULT;
                 return do_format(drive, type, &fmt_desc);
         case FDCLRPRM:
@@ -1950,14 +1951,20 @@ Enomem:
         return -ENOMEM;
 }
 
-
-void __init atari_floppy_setup( char *str, int *ints )
+#ifndef MODULE
+static int __init atari_floppy_setup(char *str)
 {
+        int ints[3 + FD_MAX_UNITS];
         int i;
+
+        if (!MACH_IS_ATARI)
+                return 0;
+
+        str = get_options(str, 3 + FD_MAX_UNITS, ints);
         
         if (ints[0] < 1) {
                 printk(KERN_ERR "ataflop_setup: no arguments!\n" );
-                return;
+                return 0;
         }
         else if (ints[0] > 2+FD_MAX_UNITS) {
                 printk(KERN_ERR "ataflop_setup: too many arguments\n" );
@@ -1977,9 +1984,13 @@ void __init atari_floppy_setup( char *str, int *ints )
                 else
                         UserSteprate[i-3] = ints[i];
         }
+        return 1;
 }
 
-static void atari_floppy_exit(void)
+__setup("floppy=", atari_floppy_setup);
+#endif
+
+static void __exit atari_floppy_exit(void)
 {
         int i;
         blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256);
diff --git a/drivers/char/amiserial.c b/drivers/char/amiserial.c
index 869518e4035f..667a21c72edb 100644
--- a/drivers/char/amiserial.c
+++ b/drivers/char/amiserial.c
@@ -99,6 +99,7 @@ static char *serial_version = "4.30";
 #define _INLINE_ inline
 #endif
 
+#define custom amiga_custom
 static char *serial_name = "Amiga-builtin serial driver";
 
 static struct tty_driver *serial_driver;
@@ -1088,7 +1089,7 @@ static void rs_unthrottle(struct tty_struct * tty)
  */
 
 static int get_serial_info(struct async_struct * info,
-                           struct serial_struct * retinfo)
+                           struct serial_struct __user * retinfo)
 {
         struct serial_struct tmp;
         struct serial_state *state = info->state;
@@ -1112,7 +1113,7 @@ static int get_serial_info(struct async_struct * info,
 }
 
 static int set_serial_info(struct async_struct * info,
-                           struct serial_struct * new_info)
+                           struct serial_struct __user * new_info)
 {
         struct serial_struct new_serial;
         struct serial_state old_state, *state;
@@ -1193,7 +1194,7 @@ check_and_exit:
  *          transmit holding register is empty.  This functionality
  *          allows an RS485 driver to be written in user space. 
  */
-static int get_lsr_info(struct async_struct * info, unsigned int *value)
+static int get_lsr_info(struct async_struct * info, unsigned int __user *value)
 {
         unsigned char status;
         unsigned int result;
@@ -1284,6 +1285,7 @@ static int rs_ioctl(struct tty_struct *tty, struct file * file,
         struct async_struct * info = (struct async_struct *)tty->driver_data;
         struct async_icount cprev, cnow;        /* kernel counter temps */
         struct serial_icounter_struct icount;
+        void __user *argp = (void __user *)arg;
         unsigned long flags;
 
         if (serial_paranoia_check(info, tty->name, "rs_ioctl"))
@@ -1298,19 +1300,17 @@ static int rs_ioctl(struct tty_struct *tty, struct file * file,
 
         switch (cmd) {
                 case TIOCGSERIAL:
-                        return get_serial_info(info,
-                                               (struct serial_struct *) arg);
+                        return get_serial_info(info, argp);
                 case TIOCSSERIAL:
-                        return set_serial_info(info,
-                                               (struct serial_struct *) arg);
+                        return set_serial_info(info, argp);
                 case TIOCSERCONFIG:
                         return 0;
 
                 case TIOCSERGETLSR: /* Get line status register */
-                        return get_lsr_info(info, (unsigned int *) arg);
+                        return get_lsr_info(info, argp);
 
                 case TIOCSERGSTRUCT:
-                        if (copy_to_user((struct async_struct *) arg,
+                        if (copy_to_user(argp,
                                          info, sizeof(struct async_struct)))
                                 return -EFAULT;
                         return 0;
@@ -1369,7 +1369,7 @@ static int rs_ioctl(struct tty_struct *tty, struct file * file,
                         icount.brk = cnow.brk;
                         icount.buf_overrun = cnow.buf_overrun;
 
-                        if (copy_to_user((void *)arg, &icount, sizeof(icount)))
+                        if (copy_to_user(argp, &icount, sizeof(icount)))
                                 return -EFAULT;
                         return 0;
                 case TIOCSERGWILD:
diff --git a/drivers/char/dsp56k.c b/drivers/char/dsp56k.c
index 8693835cb2d5..e233cf280bc0 100644
--- a/drivers/char/dsp56k.c
+++ b/drivers/char/dsp56k.c
@@ -165,7 +165,7 @@ static int dsp56k_reset(void)
         return 0;
 }
 
-static int dsp56k_upload(u_char *bin, int len)
+static int dsp56k_upload(u_char __user *bin, int len)
 {
         int i;
         u_char *p;
@@ -199,7 +199,7 @@ static int dsp56k_upload(u_char *bin, int len)
         return 0;
 }
 
-static ssize_t dsp56k_read(struct file *file, char *buf, size_t count,
+static ssize_t dsp56k_read(struct file *file, char __user *buf, size_t count,
                            loff_t *ppos)
 {
         struct inode *inode = file->f_dentry->d_inode;
@@ -225,10 +225,10 @@ static ssize_t dsp56k_read(struct file *file, char *buf, size_t count,
                 }
                 case 2:  /* 16 bit */
                 {
-                        short *data;
+                        short __user *data;
 
                         count /= 2;
-                        data = (short*) buf;
+                        data = (short __user *) buf;
                         handshake(count, dsp56k.maxio, dsp56k.timeout, DSP56K_RECEIVE,
                                   put_user(dsp56k_host_interface.data.w[1], data+n++));
                         return 2*n;
@@ -244,10 +244,10 @@ static ssize_t dsp56k_read(struct file *file, char *buf, size_t count,
                 }
                 case 4:  /* 32 bit */
                 {
-                        long *data;
+                        long __user *data;
 
                         count /= 4;
-                        data = (long*) buf;
+                        data = (long __user *) buf;
                         handshake(count, dsp56k.maxio, dsp56k.timeout, DSP56K_RECEIVE,
                                   put_user(dsp56k_host_interface.data.l, data+n++));
                         return 4*n;
@@ -262,7 +262,7 @@ static ssize_t dsp56k_read(struct file *file, char *buf, size_t count,
         }
 }
 
-static ssize_t dsp56k_write(struct file *file, const char *buf, size_t count,
+static ssize_t dsp56k_write(struct file *file, const char __user *buf, size_t count,
                             loff_t *ppos)
 {
         struct inode *inode = file->f_dentry->d_inode;
@@ -287,10 +287,10 @@ static ssize_t dsp56k_write(struct file *file, const char *buf, size_t count,
                 }
                 case 2:  /* 16 bit */
                 {
-                        const short *data;
+                        const short __user *data;
 
                         count /= 2;
-                        data = (const short *)buf;
+                        data = (const short __user *)buf;
                         handshake(count, dsp56k.maxio, dsp56k.timeout, DSP56K_TRANSMIT,
                                   get_user(dsp56k_host_interface.data.w[1], data+n++));
                         return 2*n;
@@ -306,10 +306,10 @@ static ssize_t dsp56k_write(struct file *file, const char *buf, size_t count,
                 }
                 case 4:  /* 32 bit */
                 {
-                        const long *data;
+                        const long __user *data;
 
                         count /= 4;
-                        data = (const long *)buf;
+                        data = (const long __user *)buf;
                         handshake(count, dsp56k.maxio, dsp56k.timeout, DSP56K_TRANSMIT,
                                   get_user(dsp56k_host_interface.data.l, data+n++));
                         return 4*n;
@@ -328,6 +328,7 @@ static int dsp56k_ioctl(struct inode *inode, struct file *file,
                         unsigned int cmd, unsigned long arg)
 {
         int dev = iminor(inode) & 0x0f;
+        void __user *argp = (void __user *)arg;
 
         switch(dev)
         {
@@ -336,9 +337,9 @@ static int dsp56k_ioctl(struct inode *inode, struct file *file,
                 switch(cmd) {
                 case DSP56K_UPLOAD:
                 {
-                        char *bin;
+                        char __user *bin;
                         int r, len;
-                        struct dsp56k_upload *binary = (struct dsp56k_upload *) arg;
+                        struct dsp56k_upload __user *binary = argp;
     
                         if(get_user(len, &binary->len) < 0)
                                 return -EFAULT;
@@ -372,7 +373,7 @@ static int dsp56k_ioctl(struct inode *inode, struct file *file,
                 case DSP56K_HOST_FLAGS:
                 {
                         int dir, out, status;
-                        struct dsp56k_host_flags *hf = (struct dsp56k_host_flags*) arg;
+                        struct dsp56k_host_flags __user *hf = argp;
     
                         if(get_user(dir, &hf->dir) < 0)
                                 return -EFAULT;
diff --git a/drivers/char/scc.h b/drivers/char/scc.h
index 51810f72f1a9..93998f5baff5 100644
--- a/drivers/char/scc.h
+++ b/drivers/char/scc.h
@@ -399,7 +399,7 @@ struct scc_port {
                 __asm__ __volatile__ ( "tstb %0" : : "g" (*_scc_del) : "cc" );\
     } while (0)
 
-extern unsigned char scc_shadow[2][16];
+static unsigned char scc_shadow[2][16];
 
 /* The following functions should relax the somehow complicated
  * register access of the SCC. _SCCwrite() stores all written values
diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
index 362b33556b1a..745979f33dc2 100644
--- a/drivers/input/evdev.c
+++ b/drivers/input/evdev.c
@@ -159,7 +159,7 @@ struct input_event_compat {
 #ifdef CONFIG_X86_64
 #  define COMPAT_TEST is_compat_task()
 #elif defined(CONFIG_IA64)
-#  define COMPAT_TEST IS_IA32_PROCESS(ia64_task_regs(current))
+#  define COMPAT_TEST IS_IA32_PROCESS(task_pt_regs(current))
 #elif defined(CONFIG_S390)
 #  define COMPAT_TEST test_thread_flag(TIF_31BIT)
 #elif defined(CONFIG_MIPS)
diff --git a/drivers/input/joystick/amijoy.c b/drivers/input/joystick/amijoy.c
index 8558a99f6635..ec55a29fc861 100644
--- a/drivers/input/joystick/amijoy.c
+++ b/drivers/input/joystick/amijoy.c
@@ -64,8 +64,8 @@ static irqreturn_t amijoy_interrupt(int irq, void *dummy, struct pt_regs *fp)
                 if (amijoy[i]) {
 
                         switch (i) {
-                                case 0: data = ~custom.joy0dat; button = (~ciaa.pra >> 6) & 1; break;
-                                case 1: data = ~custom.joy1dat; button = (~ciaa.pra >> 7) & 1; break;
+                                case 0: data = ~amiga_custom.joy0dat; button = (~ciaa.pra >> 6) & 1; break;
+                                case 1: data = ~amiga_custom.joy1dat; button = (~ciaa.pra >> 7) & 1; break;
                         }
 
                         input_regs(amijoy_dev[i], fp);
diff --git a/drivers/input/mouse/amimouse.c b/drivers/input/mouse/amimouse.c
index d13d4c8fe3c5..c8b2cc9f184c 100644
--- a/drivers/input/mouse/amimouse.c
+++ b/drivers/input/mouse/amimouse.c
@@ -41,7 +41,7 @@ static irqreturn_t amimouse_interrupt(int irq, void *dummy, struct pt_regs *fp)
         unsigned short joy0dat, potgor;
         int nx, ny, dx, dy;
 
-        joy0dat = custom.joy0dat;
+        joy0dat = amiga_custom.joy0dat;
 
         nx = joy0dat & 0xff;
         ny = joy0dat >> 8;
@@ -57,7 +57,7 @@ static irqreturn_t amimouse_interrupt(int irq, void *dummy, struct pt_regs *fp)
         amimouse_lastx = nx;
         amimouse_lasty = ny;
 
-        potgor = custom.potgor;
+        potgor = amiga_custom.potgor;
 
         input_regs(amimouse_dev, fp);
 
@@ -77,7 +77,7 @@ static int amimouse_open(struct input_dev *dev)
 {
         unsigned short joy0dat;
 
-        joy0dat = custom.joy0dat;
+        joy0dat = amiga_custom.joy0dat;
 
         amimouse_lastx = joy0dat & 0xff;
         amimouse_lasty = joy0dat >> 8;
diff --git a/drivers/macintosh/adb-iop.c b/drivers/macintosh/adb-iop.c
index 71aeb912ec61..d56d400b6aaa 100644
--- a/drivers/macintosh/adb-iop.c
+++ b/drivers/macintosh/adb-iop.c
@@ -239,7 +239,7 @@ static int adb_iop_write(struct adb_request *req)
 
         local_irq_save(flags);
 
-        req->next = 0;
+        req->next = NULL;
         req->sent = 0;
         req->complete = 0;
         req->reply_len = 0;
diff --git a/drivers/macintosh/via-macii.c b/drivers/macintosh/via-macii.c
index e9a159ad3022..2a2ffe060169 100644
--- a/drivers/macintosh/via-macii.c
+++ b/drivers/macintosh/via-macii.c
@@ -260,7 +260,7 @@ static int macii_write(struct adb_request *req)
                 return -EINVAL;
         }
         
-        req->next = 0;
+        req->next = NULL;
         req->sent = 0;
         req->complete = 0;
         req->reply_len = 0;
@@ -295,7 +295,7 @@ static void macii_poll(void)
         unsigned long flags;
 
         local_irq_save(flags);
-        if (via[IFR] & SR_INT) macii_interrupt(0, 0, 0);
+        if (via[IFR] & SR_INT) macii_interrupt(0, NULL, NULL);
         local_irq_restore(flags);
 }
 
diff --git a/drivers/macintosh/via-maciisi.c b/drivers/macintosh/via-maciisi.c
index a1966975d58f..0129fcc3b183 100644
--- a/drivers/macintosh/via-maciisi.c
+++ b/drivers/macintosh/via-maciisi.c
@@ -294,6 +294,24 @@ static void maciisi_sync(struct adb_request *req)
                 printk(KERN_ERR "maciisi_send_request: poll timed out!\n");
 }
 
+int
+maciisi_request(struct adb_request *req, void (*done)(struct adb_request *),
+            int nbytes, ...)
+{
+        va_list list;
+        int i;
+
+        req->nbytes = nbytes;
+        req->done = done;
+        req->reply_expected = 0;
+        va_start(list, nbytes);
+        for (i = 0; i < nbytes; i++)
+                req->data[i++] = va_arg(list, int);
+        va_end(list);
+
+        return maciisi_send_request(req, 1);
+}
+
 /* Enqueue a request, and run the queue if possible */
 static int
 maciisi_write(struct adb_request* req)
@@ -308,7 +326,7 @@ maciisi_write(struct adb_request* req)
                 req->complete = 1;
                 return -EINVAL;
         }
-        req->next = 0;
+        req->next = NULL;
         req->sent = 0;
         req->complete = 0;
         req->reply_len = 0;
@@ -403,7 +421,7 @@ maciisi_poll(void)
 
         local_irq_save(flags);
         if (via[IFR] & SR_INT) {
-                maciisi_interrupt(0, 0, 0);
+                maciisi_interrupt(0, NULL, NULL);
         }
         else /* avoid calling this function too quickly in a loop */
                 udelay(ADB_DELAY);
diff --git a/drivers/macintosh/via-pmu68k.c b/drivers/macintosh/via-pmu68k.c
index 6f80d76ac17c..f08e52f2107b 100644
--- a/drivers/macintosh/via-pmu68k.c
+++ b/drivers/macintosh/via-pmu68k.c
@@ -493,7 +493,7 @@ pmu_queue_request(struct adb_request *req)
                 return -EINVAL;
         }
 
-        req->next = 0;
+        req->next = NULL;
         req->sent = 0;
         req->complete = 0;
         local_irq_save(flags);
@@ -717,7 +717,7 @@ pmu_handle_data(unsigned char *data, int len, struct pt_regs *regs)
                                 printk(KERN_ERR "PMU: extra ADB reply\n");
                                 return;
                         }
-                        req_awaiting_reply = 0;
+                        req_awaiting_reply = NULL;
                         if (len <= 2)
                                 req->reply_len = 0;
                         else {
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 0302723fa21f..1778104e106c 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1238,6 +1238,7 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
         mdk_rdev_t *same_pdev;
         char b[BDEVNAME_SIZE], b2[BDEVNAME_SIZE];
         struct kobject *ko;
+        char *s;
 
         if (rdev->mddev) {
                 MD_BUG();
@@ -1277,6 +1278,8 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
         bdevname(rdev->bdev,b);
         if (kobject_set_name(&rdev->kobj, "dev-%s", b) < 0)
                 return -ENOMEM;
+        while ( (s=strchr(rdev->kobj.k_name, '/')) != NULL)
+                *s = '!';
                         
         list_add(&rdev->same_set, &mddev->disks);
         rdev->mddev = mddev;
diff --git a/drivers/net/hplance.c b/drivers/net/hplance.c
index 08703d6f934c..d8410634bcaf 100644
--- a/drivers/net/hplance.c
+++ b/drivers/net/hplance.c
@@ -150,7 +150,7 @@ static void __init hplance_init(struct net_device *dev, struct dio_dev *d)
         lp->lance.name = (char*)d->name;                /* discards const, shut up gcc */
         lp->lance.base = va;
         lp->lance.init_block = (struct lance_init_block *)(va + HPLANCE_MEMOFF); /* CPU addr */
-        lp->lance.lance_init_block = 0;                 /* LANCE addr of same RAM */
+        lp->lance.lance_init_block = NULL;              /* LANCE addr of same RAM */
         lp->lance.busmaster_regval = LE_C3_BSWP;        /* we're bigendian */
         lp->lance.irq = d->ipl;
         lp->lance.writerap = hplance_writerap;
diff --git a/drivers/net/mac8390.c b/drivers/net/mac8390.c
index d8c99f038fa0..06cb460361a8 100644
--- a/drivers/net/mac8390.c
+++ b/drivers/net/mac8390.c
@@ -559,55 +559,52 @@ static void mac8390_no_reset(struct net_device *dev)
 /* directly from daynaport.c by Alan Cox */
 static void dayna_memcpy_fromcard(struct net_device *dev, void *to, int from, int count)
 {
-        volatile unsigned short *ptr;
-        unsigned short *target=to;
+        volatile unsigned char *ptr;
+        unsigned char *target=to;
         from<<=1;       /* word, skip overhead */
-        ptr=(unsigned short *)(dev->mem_start+from);
+        ptr=(unsigned char *)(dev->mem_start+from);
         /* Leading byte? */
         if (from&2) {
-                *((char *)target)++ = *(((char *)ptr++)-1);
+                *target++ = ptr[-1];
+                ptr += 2;
                 count--;
         }
         while(count>=2)
         {
-                *target++=*ptr++;       /* Copy and */
-                ptr++;                  /* skip cruft */
+                *(unsigned short *)target = *(unsigned short volatile *)ptr;
+                ptr += 4;                       /* skip cruft */
+                target += 2;
                 count-=2;
         }
         /* Trailing byte? */
         if(count)
-        {
-                /* Big endian */
-                unsigned short v=*ptr;
-                *((char *)target)=v>>8;
-        }
+                *target = *ptr;
 }
 
 static void dayna_memcpy_tocard(struct net_device *dev, int to, const void *from, int count)
 {
         volatile unsigned short *ptr;
-        const unsigned short *src=from;
+        const unsigned char *src=from;
         to<<=1; /* word, skip overhead */
         ptr=(unsigned short *)(dev->mem_start+to);
         /* Leading byte? */
         if (to&2) { /* avoid a byte write (stomps on other data) */
-                ptr[-1] = (ptr[-1]&0xFF00)|*((unsigned char *)src)++;
+                ptr[-1] = (ptr[-1]&0xFF00)|*src++;
                 ptr++;
                 count--;
         }
         while(count>=2)
         {
-                *ptr++=*src++;          /* Copy and */
+                *ptr++=*(unsigned short *)src;          /* Copy and */
                 ptr++;                  /* skip cruft */
+                src += 2;
                 count-=2;
         }
         /* Trailing byte? */
         if(count)
         {
-                /* Big endian */
-                unsigned short v=*src;
                 /* card doesn't like byte writes */
-                *ptr=(*ptr&0x00FF)|(v&0xFF00);
+                *ptr=(*ptr&0x00FF)|(*src << 8);
         }
 }
 
diff --git a/drivers/net/sun3lance.c b/drivers/net/sun3lance.c
index 5c8fcd40ef4d..01bdb2334058 100644
--- a/drivers/net/sun3lance.c
+++ b/drivers/net/sun3lance.c
@@ -389,7 +389,7 @@ static int __init lance_probe( struct net_device *dev)
         dev->stop = &lance_close;
         dev->get_stats = &lance_get_stats;
         dev->set_multicast_list = &set_multicast_list;
-        dev->set_mac_address = 0;
+        dev->set_mac_address = NULL;
 //      KLUDGE -- REMOVE ME
         set_bit(__LINK_STATE_PRESENT, &dev->state);
 
diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile
index f062ea0f813a..6e0c059df6a5 100644
--- a/drivers/scsi/Makefile
+++ b/drivers/scsi/Makefile
@@ -45,7 +45,7 @@ obj-$(CONFIG_CYBERSTORMII_SCSI)	+= NCR53C9x.o	cyberstormII.o
 obj-$(CONFIG_BLZ2060_SCSI)      += NCR53C9x.o   blz2060.o
 obj-$(CONFIG_BLZ1230_SCSI)      += NCR53C9x.o   blz1230.o
 obj-$(CONFIG_FASTLANE_SCSI)     += NCR53C9x.o   fastlane.o
-obj-$(CONFIG_OKTAGON_SCSI)      += NCR53C9x.o   oktagon_esp.o   oktagon_io.o
+obj-$(CONFIG_OKTAGON_SCSI)      += NCR53C9x.o   oktagon_esp_mod.o
 obj-$(CONFIG_ATARI_SCSI)        += atari_scsi.o
 obj-$(CONFIG_MAC_SCSI)          += mac_scsi.o
 obj-$(CONFIG_SCSI_MAC_ESP)      += mac_esp.o    NCR53C9x.o
@@ -164,6 +164,7 @@ CFLAGS_ncr53c8xx.o	:= $(ncr53c8xx-flags-y) $(ncr53c8xx-flags-m)
 zalon7xx-objs   := zalon.o ncr53c8xx.o
 NCR_Q720_mod-objs       := NCR_Q720.o ncr53c8xx.o
 libata-objs     := libata-core.o libata-scsi.o
+oktagon_esp_mod-objs    := oktagon_esp.o oktagon_io.o
 
 # Files generated that shall be removed upon make clean
 clean-files :=  53c7xx_d.h 53c700_d.h   \
diff --git a/drivers/scsi/NCR53C9x.c b/drivers/scsi/NCR53C9x.c
index 640590bd014a..c7dd0154d012 100644
--- a/drivers/scsi/NCR53C9x.c
+++ b/drivers/scsi/NCR53C9x.c
@@ -1799,6 +1799,7 @@ static int esp_do_data(struct NCR_ESP *esp, struct ESP_regs *eregs)
                  */
                 int oldphase, i = 0; /* or where we left off last time ?? esp->current_data ?? */
                 int fifocnt = 0;
+                unsigned char *p = phys_to_virt((unsigned long)SCptr->SCp.ptr);
 
                 oldphase = esp_read(eregs->esp_status) & ESP_STAT_PMASK;
 
@@ -1860,7 +1861,7 @@ static int esp_do_data(struct NCR_ESP *esp, struct ESP_regs *eregs)
 
                                 /* read fifo */
                                 for(j=0;j<fifocnt;j++)
-                                        SCptr->SCp.ptr[i++] = esp_read(eregs->esp_fdata);
+                                        p[i++] = esp_read(eregs->esp_fdata);
 
                                 ESPDATA(("(%d) ", i));
 
@@ -1882,7 +1883,7 @@ static int esp_do_data(struct NCR_ESP *esp, struct ESP_regs *eregs)
 
                                 /* fill fifo */
                                 for(j=0;j<this_count;j++)
-                                        esp_write(eregs->esp_fdata, SCptr->SCp.ptr[i++]);
+                                        esp_write(eregs->esp_fdata, p[i++]);
 
                                 /* how many left if this goes out ?? */
                                 hmuch -= this_count;
diff --git a/drivers/scsi/blz1230.c b/drivers/scsi/blz1230.c
index 763e409a1ff3..3867ac2de4c2 100644
--- a/drivers/scsi/blz1230.c
+++ b/drivers/scsi/blz1230.c
@@ -224,7 +224,7 @@ static int dma_can_transfer(struct NCR_ESP *esp, Scsi_Cmnd *sp)
 static void dma_dump_state(struct NCR_ESP *esp)
 {
         ESPLOG(("intreq:<%04x>, intena:<%04x>\n",
-                custom.intreqr, custom.intenar));
+                amiga_custom.intreqr, amiga_custom.intenar));
 }
 
 void dma_init_read(struct NCR_ESP *esp, __u32 addr, int length)
@@ -298,7 +298,7 @@ static int dma_irq_p(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/blz2060.c b/drivers/scsi/blz2060.c
index d72d05fffdfa..4ebe69e32756 100644
--- a/drivers/scsi/blz2060.c
+++ b/drivers/scsi/blz2060.c
@@ -190,7 +190,7 @@ static int dma_can_transfer(struct NCR_ESP *esp, Scsi_Cmnd *sp)
 static void dma_dump_state(struct NCR_ESP *esp)
 {
         ESPLOG(("intreq:<%04x>, intena:<%04x>\n",
-                custom.intreqr, custom.intenar));
+                amiga_custom.intreqr, amiga_custom.intenar));
 }
 
 static void dma_init_read(struct NCR_ESP *esp, __u32 addr, int length)
@@ -251,7 +251,7 @@ static void dma_led_on(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/cyberstorm.c b/drivers/scsi/cyberstorm.c
index f9b940e56430..a4a4fac5c0a1 100644
--- a/drivers/scsi/cyberstorm.c
+++ b/drivers/scsi/cyberstorm.c
@@ -223,7 +223,7 @@ static void dma_dump_state(struct NCR_ESP *esp)
                 esp->esp_id, ((struct cyber_dma_registers *)
                               (esp->dregs))->cond_reg));
         ESPLOG(("intreq:<%04x>, intena:<%04x>\n",
-                custom.intreqr, custom.intenar));
+                amiga_custom.intreqr, amiga_custom.intenar));
 }
 
 static void dma_init_read(struct NCR_ESP *esp, __u32 addr, int length)
@@ -322,7 +322,7 @@ static void dma_led_on(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/cyberstormII.c b/drivers/scsi/cyberstormII.c
index a3caabfd7557..3a803d73bc5f 100644
--- a/drivers/scsi/cyberstormII.c
+++ b/drivers/scsi/cyberstormII.c
@@ -200,7 +200,7 @@ static void dma_dump_state(struct NCR_ESP *esp)
                 esp->esp_id, ((struct cyberII_dma_registers *)
                               (esp->dregs))->cond_reg));
         ESPLOG(("intreq:<%04x>, intena:<%04x>\n",
-                custom.intreqr, custom.intenar));
+                amiga_custom.intreqr, amiga_custom.intenar));
 }
 
 static void dma_init_read(struct NCR_ESP *esp, __u32 addr, int length)
@@ -259,7 +259,7 @@ static void dma_led_on(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/fastlane.c b/drivers/scsi/fastlane.c
index ccee68b52f7e..8ae9c406a83b 100644
--- a/drivers/scsi/fastlane.c
+++ b/drivers/scsi/fastlane.c
@@ -268,7 +268,7 @@ static void dma_dump_state(struct NCR_ESP *esp)
                 esp->esp_id, ((struct fastlane_dma_registers *)
                               (esp->dregs))->cond_reg));
         ESPLOG(("intreq:<%04x>, intena:<%04x>\n",
-                custom.intreqr, custom.intenar));
+                amiga_custom.intreqr, amiga_custom.intenar));
 }
 
 static void dma_init_read(struct NCR_ESP *esp, __u32 addr, int length)
@@ -368,7 +368,7 @@ static void dma_led_on(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/oktagon_esp.c b/drivers/scsi/oktagon_esp.c
index 5d9c9ada814f..dee426f8c07b 100644
--- a/drivers/scsi/oktagon_esp.c
+++ b/drivers/scsi/oktagon_esp.c
@@ -490,7 +490,7 @@ static void dma_led_on(struct NCR_ESP *esp)
 
 static int dma_ports_p(struct NCR_ESP *esp)
 {
-        return ((custom.intenar) & IF_PORTS);
+        return ((amiga_custom.intenar) & IF_PORTS);
 }
 
 static void dma_setup(struct NCR_ESP *esp, __u32 addr, int count, int write)
diff --git a/drivers/scsi/wd33c93.c b/drivers/scsi/wd33c93.c
index fd63add6a577..fb53eeaee617 100644
--- a/drivers/scsi/wd33c93.c
+++ b/drivers/scsi/wd33c93.c
@@ -465,7 +465,7 @@ wd33c93_execute(struct Scsi_Host *instance)
          */
 
         cmd = (struct scsi_cmnd *) hostdata->input_Q;
-        prev = 0;
+        prev = NULL;
         while (cmd) {
                 if (!(hostdata->busy[cmd->device->id] & (1 << cmd->device->lun)))
                         break;
@@ -1569,7 +1569,7 @@ wd33c93_abort(struct scsi_cmnd * cmd)
  */
 
         tmp = (struct scsi_cmnd *) hostdata->input_Q;
-        prev = 0;
+        prev = NULL;
         while (tmp) {
                 if (tmp == cmd) {
                         if (prev)
diff --git a/drivers/video/amifb.c b/drivers/video/amifb.c
index d549e215f3c5..2c42a812655a 100644
--- a/drivers/video/amifb.c
+++ b/drivers/video/amifb.c
@@ -590,6 +590,8 @@ static u_short maxfmode, chipset;
 #define highw(x)        ((u_long)(x)>>16 & 0xffff)
 #define loww(x)         ((u_long)(x) & 0xffff)
 
+#define custom          amiga_custom
+
 #define VBlankOn()      custom.intena = IF_SETCLR|IF_COPER
 #define VBlankOff()     custom.intena = IF_COPER
 
@@ -1164,8 +1166,8 @@ static void ami_update_display(void);
 static void ami_init_display(void);
 static void ami_do_blank(void);
 static int ami_get_fix_cursorinfo(struct fb_fix_cursorinfo *fix);
-static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data);
-static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data);
+static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char __user *data);
+static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char __user *data);
 static int ami_get_cursorstate(struct fb_cursorstate *state);
 static int ami_set_cursorstate(struct fb_cursorstate *state);
 static void ami_set_sprite(void);
@@ -2179,6 +2181,7 @@ static int amifb_ioctl(struct inode *inode, struct file *file,
                 struct fb_var_cursorinfo var;
                 struct fb_cursorstate state;
         } crsr;
+        void __user *argp = (void __user *)arg;
         int i;
 
         switch (cmd) {
@@ -2186,33 +2189,32 @@ static int amifb_ioctl(struct inode *inode, struct file *file,
                         i = ami_get_fix_cursorinfo(&crsr.fix);
                         if (i)
                                 return i;
-                        return copy_to_user((void *)arg, &crsr.fix,
+                        return copy_to_user(argp, &crsr.fix,
                                             sizeof(crsr.fix)) ? -EFAULT : 0;
 
                 case FBIOGET_VCURSORINFO:
                         i = ami_get_var_cursorinfo(&crsr.var,
-                                ((struct fb_var_cursorinfo *)arg)->data);
+                                ((struct fb_var_cursorinfo __user *)arg)->data);
                         if (i)
                                 return i;
-                        return copy_to_user((void *)arg, &crsr.var,
+                        return copy_to_user(argp, &crsr.var,
                                             sizeof(crsr.var)) ? -EFAULT : 0;
 
                 case FBIOPUT_VCURSORINFO:
-                        if (copy_from_user(&crsr.var, (void *)arg,
-                                           sizeof(crsr.var)))
+                        if (copy_from_user(&crsr.var, argp, sizeof(crsr.var)))
                                 return -EFAULT;
                         return ami_set_var_cursorinfo(&crsr.var,
-                                ((struct fb_var_cursorinfo *)arg)->data);
+                                ((struct fb_var_cursorinfo __user *)arg)->data);
 
                 case FBIOGET_CURSORSTATE:
                         i = ami_get_cursorstate(&crsr.state);
                         if (i)
                                 return i;
-                        return copy_to_user((void *)arg, &crsr.state,
+                        return copy_to_user(argp, &crsr.state,
                                             sizeof(crsr.state)) ? -EFAULT : 0;
 
                 case FBIOPUT_CURSORSTATE:
-                        if (copy_from_user(&crsr.state, (void *)arg,
+                        if (copy_from_user(&crsr.state, argp,
                                            sizeof(crsr.state)))
                                 return -EFAULT;
                         return ami_set_cursorstate(&crsr.state);
@@ -3325,7 +3327,7 @@ static int ami_get_fix_cursorinfo(struct fb_fix_cursorinfo *fix)
         return 0;
 }
 
-static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
+static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char __user *data)
 {
         struct amifb_par *par = &currentpar;
         register u_short *lspr, *sspr;
@@ -3347,14 +3349,14 @@ static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
         var->yspot = par->crsr.spot_y;
         if (size > var->height*var->width)
                 return -ENAMETOOLONG;
-        if (!access_ok(VERIFY_WRITE, (void *)data, size))
+        if (!access_ok(VERIFY_WRITE, data, size))
                 return -EFAULT;
         delta = 1<<par->crsr.fmode;
         lspr = lofsprite + (delta<<1);
         if (par->bplcon0 & BPC0_LACE)
                 sspr = shfsprite + (delta<<1);
         else
-                sspr = 0;
+                sspr = NULL;
         for (height = (short)var->height-1; height >= 0; height--) {
                 bits = 0; words = delta; datawords = 0;
                 for (width = (short)var->width-1; width >= 0; width--) {
@@ -3400,7 +3402,7 @@ static int ami_get_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
         return 0;
 }
 
-static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
+static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char __user *data)
 {
         struct amifb_par *par = &currentpar;
         register u_short *lspr, *sspr;
@@ -3427,7 +3429,7 @@ static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
                 return -EINVAL;
         if (!var->height)
                 return -EINVAL;
-        if (!access_ok(VERIFY_READ, (void *)data, var->width*var->height))
+        if (!access_ok(VERIFY_READ, data, var->width*var->height))
                 return -EFAULT;
         delta = 1<<fmode;
         lofsprite = shfsprite = (u_short *)spritememory;
@@ -3442,13 +3444,13 @@ static int ami_set_var_cursorinfo(struct fb_var_cursorinfo *var, u_char *data)
                 if (((var->height+2)<<fmode<<2) > SPRITEMEMSIZE)
                         return -EINVAL;
                 memset(lspr, 0, (var->height+2)<<fmode<<2);
-                sspr = 0;
+                sspr = NULL;
         }
         for (height = (short)var->height-1; height >= 0; height--) {
                 bits = 16; words = delta; datawords = 0;
                 for (width = (short)var->width-1; width >= 0; width--) {
                         unsigned long tdata = 0;
-                        get_user(tdata, (char *)data);
+                        get_user(tdata, data);
                         data++;
 #ifdef __mc68000__
                         asm volatile (
diff --git a/drivers/video/aty/atyfb_base.c b/drivers/video/aty/atyfb_base.c
index e370125e4fbc..ed81005cbdba 100644
--- a/drivers/video/aty/atyfb_base.c
+++ b/drivers/video/aty/atyfb_base.c
@@ -3501,7 +3501,7 @@ err_release_mem:
 
 static int __devinit atyfb_atari_probe(void)
 {
-        struct aty_par *par;
+        struct atyfb_par *par;
         struct fb_info *info;
         int m64_num;
         u32 clock_r;
diff --git a/drivers/video/macfb.c b/drivers/video/macfb.c
index cfc748e94272..e6cbd9de944a 100644
--- a/drivers/video/macfb.c
+++ b/drivers/video/macfb.c
@@ -609,18 +609,19 @@ void __init macfb_setup(char *options)
         }
 }
 
-void __init macfb_init(void)
+static int __init macfb_init(void)
 {
         int video_cmap_len, video_is_nubus = 0;
         struct nubus_dev* ndev = NULL;
         char *option = NULL;
+        int err;
 
         if (fb_get_options("macfb", &option))
                 return -ENODEV;
         macfb_setup(option);
 
         if (!MACH_IS_MAC) 
-                return;
+                return -ENODEV;
 
         /* There can only be one internal video controller anyway so
            we're not too worried about this */
@@ -958,11 +959,11 @@ void __init macfb_init(void)
 
         fb_alloc_cmap(&fb_info.cmap, video_cmap_len, 0);
         
-        if (register_framebuffer(&fb_info) < 0)
-                return;
-
-        printk("fb%d: %s frame buffer device\n",
-               fb_info.node, fb_info.fix.id);
+        err = register_framebuffer(&fb_info);
+        if (!err)
+                printk("fb%d: %s frame buffer device\n",
+                       fb_info.node, fb_info.fix.id);
+        return err;
 }
 
 module_init(macfb_init);
diff --git a/drivers/zorro/proc.c b/drivers/zorro/proc.c
index 1a409c2c320c..7aa2d3de6d37 100644
--- a/drivers/zorro/proc.c
+++ b/drivers/zorro/proc.c
@@ -45,7 +45,7 @@ proc_bus_zorro_lseek(struct file *file, loff_t off, int whence)
 }
 
 static ssize_t
-proc_bus_zorro_read(struct file *file, char *buf, size_t nbytes, loff_t *ppos)
+proc_bus_zorro_read(struct file *file, char __user *buf, size_t nbytes, loff_t *ppos)
 {
         struct inode *ino = file->f_dentry->d_inode;
         struct proc_dir_entry *dp = PDE(ino);
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index f0b7256b2f87..5dd0207ffd46 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -122,6 +122,7 @@
 #include <linux/dvb/dmx.h>
 #include <linux/dvb/frontend.h>
 #include <linux/dvb/video.h>
+#include <linux/lp.h>
 
 /* Aiee. Someone does not find a difference between int and long */
 #define EXT2_IOC32_GETFLAGS               _IOR('f', 1, int)
@@ -2735,6 +2736,20 @@ static int do_ncp_setprivatedata(unsigned int fd, unsigned int cmd, unsigned lon
 }
 #endif
 
+static int
+lp_timeout_trans(unsigned int fd, unsigned int cmd, unsigned long arg)
+{
+        struct compat_timeval *tc = (struct compat_timeval *)arg;
+        struct timeval *tn = compat_alloc_user_space(sizeof(struct timeval));
+        struct timeval ts;
+        if (get_user(ts.tv_sec, &tc->tv_sec) ||
+            get_user(ts.tv_usec, &tc->tv_usec) ||
+            put_user(ts.tv_sec, &tn->tv_sec) ||
+            put_user(ts.tv_usec, &tn->tv_usec))
+                return -EFAULT;
+        return sys_ioctl(fd, cmd, (unsigned long)tn);
+}
+
 #define HANDLE_IOCTL(cmd,handler) \
         { (cmd), (ioctl_trans_handler_t)(handler) },
 
@@ -2962,6 +2977,20 @@ HANDLE_IOCTL(DMX_GET_EVENT, do_dmx_get_event)
 HANDLE_IOCTL(VIDEO_GET_EVENT, do_video_get_event)
 HANDLE_IOCTL(VIDEO_STILLPICTURE, do_video_stillpicture)
 HANDLE_IOCTL(VIDEO_SET_SPU_PALETTE, do_video_set_spu_palette)
+
+/* parport */
+COMPATIBLE_IOCTL(LPTIME)
+COMPATIBLE_IOCTL(LPCHAR)
+COMPATIBLE_IOCTL(LPABORTOPEN)
+COMPATIBLE_IOCTL(LPCAREFUL)
+COMPATIBLE_IOCTL(LPWAIT)
+COMPATIBLE_IOCTL(LPSETIRQ)
+COMPATIBLE_IOCTL(LPGETSTATUS)
+COMPATIBLE_IOCTL(LPGETSTATUS)
+COMPATIBLE_IOCTL(LPRESET)
+/*LPGETSTATS not implemented, but no kernels seem to compile it in anyways*/
+COMPATIBLE_IOCTL(LPGETFLAGS)
+HANDLE_IOCTL(LPSETTIMEOUT, lp_timeout_trans)
 };
 
 int ioctl_table_size = ARRAY_SIZE(ioctl_start);
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 94d3cdfbf9b8..d1db8c17a74e 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -40,11 +40,10 @@
 #include "xfs_rw.h"
 #include "xfs_iomap.h"
 #include <linux/mpage.h>
+#include <linux/pagevec.h>
 #include <linux/writeback.h>
 
 STATIC void xfs_count_page_state(struct page *, int *, int *, int *);
-STATIC void xfs_convert_page(struct inode *, struct page *, xfs_iomap_t *,
-                struct writeback_control *wbc, void *, int, int);
 
 #if defined(XFS_RW_TRACE)
 void
@@ -55,17 +54,15 @@ xfs_page_trace(
         int             mask)
 {
         xfs_inode_t     *ip;
-        bhv_desc_t      *bdp;
         vnode_t         *vp = LINVFS_GET_VP(inode);
         loff_t          isize = i_size_read(inode);
-        loff_t          offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
+        loff_t          offset = page_offset(page);
         int             delalloc = -1, unmapped = -1, unwritten = -1;
 
         if (page_has_buffers(page))
                 xfs_count_page_state(page, &delalloc, &unmapped, &unwritten);
 
-        bdp = vn_bhv_lookup(VN_BHV_HEAD(vp), &xfs_vnodeops);
-        ip = XFS_BHVTOI(bdp);
+        ip = xfs_vtoi(vp);
         if (!ip->i_rwtrace)
                 return;
 
@@ -103,15 +100,56 @@ xfs_finish_ioend(
                 queue_work(xfsdatad_workqueue, &ioend->io_work);
 }
 
+/*
+ * We're now finished for good with this ioend structure.
+ * Update the page state via the associated buffer_heads,
+ * release holds on the inode and bio, and finally free
+ * up memory.  Do not use the ioend after this.
+ */
 STATIC void
 xfs_destroy_ioend(
         xfs_ioend_t             *ioend)
 {
+        struct buffer_head      *bh, *next;
+
+        for (bh = ioend->io_buffer_head; bh; bh = next) {
+                next = bh->b_private;
+                bh->b_end_io(bh, ioend->io_uptodate);
+        }
+
         vn_iowake(ioend->io_vnode);
         mempool_free(ioend, xfs_ioend_pool);
 }
 
 /*
+ * Buffered IO write completion for delayed allocate extents.
+ * TODO: Update ondisk isize now that we know the file data
+ * has been flushed (i.e. the notorious "NULL file" problem).
+ */
+STATIC void
+xfs_end_bio_delalloc(
+        void                    *data)
+{
+        xfs_ioend_t             *ioend = data;
+
+        xfs_destroy_ioend(ioend);
+}
+
+/*
+ * Buffered IO write completion for regular, written extents.
+ */
+STATIC void
+xfs_end_bio_written(
+        void                    *data)
+{
+        xfs_ioend_t             *ioend = data;
+
+        xfs_destroy_ioend(ioend);
+}
+
+/*
+ * IO write completion for unwritten extents.
+ *
  * Issue transactions to convert a buffer range from unwritten
  * to written extents.
  */
@@ -123,21 +161,10 @@ xfs_end_bio_unwritten(
         vnode_t                 *vp = ioend->io_vnode;
         xfs_off_t               offset = ioend->io_offset;
         size_t                  size = ioend->io_size;
-        struct buffer_head      *bh, *next;
         int                     error;
 
         if (ioend->io_uptodate)
                 VOP_BMAP(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL, error);
-
-        /* ioend->io_buffer_head is only non-NULL for buffered I/O */
-        for (bh = ioend->io_buffer_head; bh; bh = next) {
-                next = bh->b_private;
-
-                bh->b_end_io = NULL;
-                clear_buffer_unwritten(bh);
-                end_buffer_async_write(bh, ioend->io_uptodate);
-        }
-
         xfs_destroy_ioend(ioend);
 }
 
@@ -149,7 +176,8 @@ xfs_end_bio_unwritten(
  */
 STATIC xfs_ioend_t *
 xfs_alloc_ioend(
-        struct inode            *inode)
+        struct inode            *inode,
+        unsigned int            type)
 {
         xfs_ioend_t             *ioend;
 
@@ -162,45 +190,25 @@ xfs_alloc_ioend(
          */
         atomic_set(&ioend->io_remaining, 1);
         ioend->io_uptodate = 1; /* cleared if any I/O fails */
+        ioend->io_list = NULL;
+        ioend->io_type = type;
         ioend->io_vnode = LINVFS_GET_VP(inode);
         ioend->io_buffer_head = NULL;
+        ioend->io_buffer_tail = NULL;
         atomic_inc(&ioend->io_vnode->v_iocount);
         ioend->io_offset = 0;
         ioend->io_size = 0;
 
-        INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
+        if (type == IOMAP_UNWRITTEN)
+                INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
+        else if (type == IOMAP_DELAY)
+                INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc, ioend);
+        else
+                INIT_WORK(&ioend->io_work, xfs_end_bio_written, ioend);
 
         return ioend;
 }
 
-void
-linvfs_unwritten_done(
-        struct buffer_head      *bh,
-        int                     uptodate)
-{
-        xfs_ioend_t             *ioend = bh->b_private;
-        static spinlock_t       unwritten_done_lock = SPIN_LOCK_UNLOCKED;
-        unsigned long           flags;
-
-        ASSERT(buffer_unwritten(bh));
-        bh->b_end_io = NULL;
-
-        if (!uptodate)
-                ioend->io_uptodate = 0;
-
-        /*
-         * Deep magic here.  We reuse b_private in the buffer_heads to build
-         * a chain for completing the I/O from user context after we've issued
-         * a transaction to convert the unwritten extent.
-         */
-        spin_lock_irqsave(&unwritten_done_lock, flags);
-        bh->b_private = ioend->io_buffer_head;
-        ioend->io_buffer_head = bh;
-        spin_unlock_irqrestore(&unwritten_done_lock, flags);
-
-        xfs_finish_ioend(ioend);
-}
-
 STATIC int
 xfs_map_blocks(
         struct inode            *inode,
@@ -218,138 +226,260 @@ xfs_map_blocks(
         return -error;
 }
 
+STATIC inline int
+xfs_iomap_valid(
+        xfs_iomap_t             *iomapp,
+        loff_t                  offset)
+{
+        return offset >= iomapp->iomap_offset &&
+                offset < iomapp->iomap_offset + iomapp->iomap_bsize;
+}
+
 /*
- * Finds the corresponding mapping in block @map array of the
- * given @offset within a @page.
+ * BIO completion handler for buffered IO.
  */
-STATIC xfs_iomap_t *
-xfs_offset_to_map(
+STATIC int
+xfs_end_bio(
+        struct bio              *bio,
+        unsigned int            bytes_done,
+        int                     error)
+{
+        xfs_ioend_t             *ioend = bio->bi_private;
+
+        if (bio->bi_size)
+                return 1;
+
+        ASSERT(ioend);
+        ASSERT(atomic_read(&bio->bi_cnt) >= 1);
+
+        /* Toss bio and pass work off to an xfsdatad thread */
+        if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
+                ioend->io_uptodate = 0;
+        bio->bi_private = NULL;
+        bio->bi_end_io = NULL;
+
+        bio_put(bio);
+        xfs_finish_ioend(ioend);
+        return 0;
+}
+
+STATIC void
+xfs_submit_ioend_bio(
+        xfs_ioend_t     *ioend,
+        struct bio      *bio)
+{
+        atomic_inc(&ioend->io_remaining);
+
+        bio->bi_private = ioend;
+        bio->bi_end_io = xfs_end_bio;
+
+        submit_bio(WRITE, bio);
+        ASSERT(!bio_flagged(bio, BIO_EOPNOTSUPP));
+        bio_put(bio);
+}
+
+STATIC struct bio *
+xfs_alloc_ioend_bio(
+        struct buffer_head      *bh)
+{
+        struct bio              *bio;
+        int                     nvecs = bio_get_nr_vecs(bh->b_bdev);
+
+        do {
+                bio = bio_alloc(GFP_NOIO, nvecs);
+                nvecs >>= 1;
+        } while (!bio);
+
+        ASSERT(bio->bi_private == NULL);
+        bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
+        bio->bi_bdev = bh->b_bdev;
+        bio_get(bio);
+        return bio;
+}
+
+STATIC void
+xfs_start_buffer_writeback(
+        struct buffer_head      *bh)
+{
+        ASSERT(buffer_mapped(bh));
+        ASSERT(buffer_locked(bh));
+        ASSERT(!buffer_delay(bh));
+        ASSERT(!buffer_unwritten(bh));
+
+        mark_buffer_async_write(bh);
+        set_buffer_uptodate(bh);
+        clear_buffer_dirty(bh);
+}
+
+STATIC void
+xfs_start_page_writeback(
         struct page             *page,
-        xfs_iomap_t             *iomapp,
-        unsigned long           offset)
+        struct writeback_control *wbc,
+        int                     clear_dirty,
+        int                     buffers)
+{
+        ASSERT(PageLocked(page));
+        ASSERT(!PageWriteback(page));
+        set_page_writeback(page);
+        if (clear_dirty)
+                clear_page_dirty(page);
+        unlock_page(page);
+        if (!buffers) {
+                end_page_writeback(page);
+                wbc->pages_skipped++;   /* We didn't write this page */
+        }
+}
+
+static inline int bio_add_buffer(struct bio *bio, struct buffer_head *bh)
+{
+        return bio_add_page(bio, bh->b_page, bh->b_size, bh_offset(bh));
+}
+
+/*
+ * Submit all of the bios for all of the ioends we have saved up,
+ * covering the initial writepage page and also any probed pages.
+ */
+STATIC void
+xfs_submit_ioend(
+        xfs_ioend_t             *ioend)
+{
+        xfs_ioend_t             *next;
+        struct buffer_head      *bh;
+        struct bio              *bio;
+        sector_t                lastblock = 0;
+
+        do {
+                next = ioend->io_list;
+                bio = NULL;
+
+                for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
+                        xfs_start_buffer_writeback(bh);
+
+                        if (!bio) {
+ retry:
+                                bio = xfs_alloc_ioend_bio(bh);
+                        } else if (bh->b_blocknr != lastblock + 1) {
+                                xfs_submit_ioend_bio(ioend, bio);
+                                goto retry;
+                        }
+
+                        if (bio_add_buffer(bio, bh) != bh->b_size) {
+                                xfs_submit_ioend_bio(ioend, bio);
+                                goto retry;
+                        }
+
+                        lastblock = bh->b_blocknr;
+                }
+                if (bio)
+                        xfs_submit_ioend_bio(ioend, bio);
+                xfs_finish_ioend(ioend);
+        } while ((ioend = next) != NULL);
+}
+
+/*
+ * Cancel submission of all buffer_heads so far in this endio.
+ * Toss the endio too.  Only ever called for the initial page
+ * in a writepage request, so only ever one page.
+ */
+STATIC void
+xfs_cancel_ioend(
+        xfs_ioend_t             *ioend)
+{
+        xfs_ioend_t             *next;
+        struct buffer_head      *bh, *next_bh;
+
+        do {
+                next = ioend->io_list;
+                bh = ioend->io_buffer_head;
+                do {
+                        next_bh = bh->b_private;
+                        clear_buffer_async_write(bh);
+                        unlock_buffer(bh);
+                } while ((bh = next_bh) != NULL);
+
+                vn_iowake(ioend->io_vnode);
+                mempool_free(ioend, xfs_ioend_pool);
+        } while ((ioend = next) != NULL);
+}
+
+/*
+ * Test to see if we've been building up a completion structure for
+ * earlier buffers -- if so, we try to append to this ioend if we
+ * can, otherwise we finish off any current ioend and start another.
+ * Return true if we've finished the given ioend.
+ */
+STATIC void
+xfs_add_to_ioend(
+        struct inode            *inode,
+        struct buffer_head      *bh,
+        xfs_off_t               offset,
+        unsigned int            type,
+        xfs_ioend_t             **result,
+        int                     need_ioend)
 {
-        loff_t                  full_offset;    /* offset from start of file */
+        xfs_ioend_t             *ioend = *result;
 
-        ASSERT(offset < PAGE_CACHE_SIZE);
+        if (!ioend || need_ioend || type != ioend->io_type) {
+                xfs_ioend_t     *previous = *result;
 
-        full_offset = page->index;              /* NB: using 64bit number */
-        full_offset <<= PAGE_CACHE_SHIFT;       /* offset from file start */
-        full_offset += offset;                  /* offset from page start */
+                ioend = xfs_alloc_ioend(inode, type);
+                ioend->io_offset = offset;
+                ioend->io_buffer_head = bh;
+                ioend->io_buffer_tail = bh;
+                if (previous)
+                        previous->io_list = ioend;
+                *result = ioend;
+        } else {
+                ioend->io_buffer_tail->b_private = bh;
+                ioend->io_buffer_tail = bh;
+        }
 
-        if (full_offset < iomapp->iomap_offset)
-                return NULL;
-        if (iomapp->iomap_offset + (iomapp->iomap_bsize -1) >= full_offset)
-                return iomapp;
-        return NULL;
+        bh->b_private = NULL;
+        ioend->io_size += bh->b_size;
 }
 
 STATIC void
 xfs_map_at_offset(
-        struct page             *page,
         struct buffer_head      *bh,
-        unsigned long           offset,
+        loff_t                  offset,
         int                     block_bits,
         xfs_iomap_t             *iomapp)
 {
         xfs_daddr_t             bn;
-        loff_t                  delta;
         int                     sector_shift;
 
         ASSERT(!(iomapp->iomap_flags & IOMAP_HOLE));
         ASSERT(!(iomapp->iomap_flags & IOMAP_DELAY));
         ASSERT(iomapp->iomap_bn != IOMAP_DADDR_NULL);
 
-        delta = page->index;
-        delta <<= PAGE_CACHE_SHIFT;
-        delta += offset;
-        delta -= iomapp->iomap_offset;
-        delta >>= block_bits;
-
         sector_shift = block_bits - BBSHIFT;
-        bn = iomapp->iomap_bn >> sector_shift;
-        bn += delta;
-        BUG_ON(!bn && !(iomapp->iomap_flags & IOMAP_REALTIME));
+        bn = (iomapp->iomap_bn >> sector_shift) +
+              ((offset - iomapp->iomap_offset) >> block_bits);
+
+        ASSERT(bn || (iomapp->iomap_flags & IOMAP_REALTIME));
         ASSERT((bn << sector_shift) >= iomapp->iomap_bn);
 
         lock_buffer(bh);
         bh->b_blocknr = bn;
-        bh->b_bdev = iomapp->iomap_target->pbr_bdev;
+        bh->b_bdev = iomapp->iomap_target->bt_bdev;
         set_buffer_mapped(bh);
         clear_buffer_delay(bh);
+        clear_buffer_unwritten(bh);
 }
 
 /*
- * Look for a page at index which is unlocked and contains our
- * unwritten extent flagged buffers at its head.  Returns page
- * locked and with an extra reference count, and length of the
- * unwritten extent component on this page that we can write,
- * in units of filesystem blocks.
- */
-STATIC struct page *
-xfs_probe_unwritten_page(
-        struct address_space    *mapping,
-        pgoff_t                 index,
-        xfs_iomap_t             *iomapp,
-        xfs_ioend_t             *ioend,
-        unsigned long           max_offset,
-        unsigned long           *fsbs,
-        unsigned int            bbits)
-{
-        struct page             *page;
-
-        page = find_trylock_page(mapping, index);
-        if (!page)
-                return NULL;
-        if (PageWriteback(page))
-                goto out;
-
-        if (page->mapping && page_has_buffers(page)) {
-                struct buffer_head      *bh, *head;
-                unsigned long           p_offset = 0;
-
-                *fsbs = 0;
-                bh = head = page_buffers(page);
-                do {
-                        if (!buffer_unwritten(bh) || !buffer_uptodate(bh))
-                                break;
-                        if (!xfs_offset_to_map(page, iomapp, p_offset))
-                                break;
-                        if (p_offset >= max_offset)
-                                break;
-                        xfs_map_at_offset(page, bh, p_offset, bbits, iomapp);
-                        set_buffer_unwritten_io(bh);
-                        bh->b_private = ioend;
-                        p_offset += bh->b_size;
-                        (*fsbs)++;
-                } while ((bh = bh->b_this_page) != head);
-
-                if (p_offset)
-                        return page;
-        }
-
-out:
-        unlock_page(page);
-        return NULL;
-}
-
-/*
- * Look for a page at index which is unlocked and not mapped
- * yet - clustering for mmap write case.
+ * Look for a page at index that is suitable for clustering.
  */
 STATIC unsigned int
-xfs_probe_unmapped_page(
-        struct address_space    *mapping,
-        pgoff_t                 index,
-        unsigned int            pg_offset)
+xfs_probe_page(
+        struct page             *page,
+        unsigned int            pg_offset,
+        int                     mapped)
 {
-        struct page             *page;
         int                     ret = 0;
 
-        page = find_trylock_page(mapping, index);
-        if (!page)
-                return 0;
         if (PageWriteback(page))
-                goto out;
+                return 0;
 
         if (page->mapping && PageDirty(page)) {
                 if (page_has_buffers(page)) {
@@ -357,79 +487,101 @@ xfs_probe_unmapped_page(
 
                         bh = head = page_buffers(page);
                         do {
-                                if (buffer_mapped(bh) || !buffer_uptodate(bh))
+                                if (!buffer_uptodate(bh))
+                                        break;
+                                if (mapped != buffer_mapped(bh))
                                         break;
                                 ret += bh->b_size;
                                 if (ret >= pg_offset)
                                         break;
                         } while ((bh = bh->b_this_page) != head);
                 } else
-                        ret = PAGE_CACHE_SIZE;
+                        ret = mapped ? 0 : PAGE_CACHE_SIZE;
         }
 
-out:
-        unlock_page(page);
         return ret;
 }
 
-STATIC unsigned int
-xfs_probe_unmapped_cluster(
+STATIC size_t
+xfs_probe_cluster(
         struct inode            *inode,
         struct page             *startpage,
         struct buffer_head      *bh,
-        struct buffer_head      *head)
+        struct buffer_head      *head,
+        int                     mapped)
 {
+        struct pagevec          pvec;
         pgoff_t                 tindex, tlast, tloff;
-        unsigned int            pg_offset, len, total = 0;
-        struct address_space    *mapping = inode->i_mapping;
+        size_t                  total = 0;
+        int                     done = 0, i;
 
         /* First sum forwards in this page */
         do {
-                if (buffer_mapped(bh))
-                        break;
+                if (mapped != buffer_mapped(bh))
+                        return total;
                 total += bh->b_size;
         } while ((bh = bh->b_this_page) != head);
 
-        /* If we reached the end of the page, sum forwards in
-         * following pages.
-         */
-        if (bh == head) {
-                tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
-                /* Prune this back to avoid pathological behavior */
-                tloff = min(tlast, startpage->index + 64);
-                for (tindex = startpage->index + 1; tindex < tloff; tindex++) {
-                        len = xfs_probe_unmapped_page(mapping, tindex,
-                                                        PAGE_CACHE_SIZE);
-                        if (!len)
-                                return total;
+        /* if we reached the end of the page, sum forwards in following pages */
+        tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
+        tindex = startpage->index + 1;
+
+        /* Prune this back to avoid pathological behavior */
+        tloff = min(tlast, startpage->index + 64);
+
+        pagevec_init(&pvec, 0);
+        while (!done && tindex <= tloff) {
+                unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
+
+                if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
+                        break;
+
+                for (i = 0; i < pagevec_count(&pvec); i++) {
+                        struct page *page = pvec.pages[i];
+                        size_t pg_offset, len = 0;
+
+                        if (tindex == tlast) {
+                                pg_offset =
+                                    i_size_read(inode) & (PAGE_CACHE_SIZE - 1);
+                                if (!pg_offset) {
+                                        done = 1;
+                                        break;
+                                }
+                        } else
+                                pg_offset = PAGE_CACHE_SIZE;
+
+                        if (page->index == tindex && !TestSetPageLocked(page)) {
+                                len = xfs_probe_page(page, pg_offset, mapped);
+                                unlock_page(page);
+                        }
+
+                        if (!len) {
+                                done = 1;
+                                break;
+                        }
+
                         total += len;
+                        tindex++;
                 }
-                if (tindex == tlast &&
-                    (pg_offset = i_size_read(inode) & (PAGE_CACHE_SIZE - 1))) {
-                        total += xfs_probe_unmapped_page(mapping,
-                                                        tindex, pg_offset);
-                }
+
+                pagevec_release(&pvec);
+                cond_resched();
         }
+
         return total;
 }
 
 /*
- * Probe for a given page (index) in the inode and test if it is delayed
- * and without unwritten buffers.  Returns page locked and with an extra
- * reference count.
+ * Test if a given page is suitable for writing as part of an unwritten
+ * or delayed allocate extent.
  */
-STATIC struct page *
-xfs_probe_delalloc_page(
-        struct inode            *inode,
-        pgoff_t                 index)
+STATIC int
+xfs_is_delayed_page(
+        struct page             *page,
+        unsigned int            type)
 {
-        struct page             *page;
-
-        page = find_trylock_page(inode->i_mapping, index);
-        if (!page)
-                return NULL;
         if (PageWriteback(page))
-                goto out;
+                return 0;
 
         if (page->mapping && page_has_buffers(page)) {
                 struct buffer_head      *bh, *head;
@@ -437,243 +589,156 @@ xfs_probe_delalloc_page(
 
                 bh = head = page_buffers(page);
                 do {
-                        if (buffer_unwritten(bh)) {
-                                acceptable = 0;
+                        if (buffer_unwritten(bh))
+                                acceptable = (type == IOMAP_UNWRITTEN);
+                        else if (buffer_delay(bh))
+                                acceptable = (type == IOMAP_DELAY);
+                        else if (buffer_mapped(bh))
+                                acceptable = (type == 0);
+                        else
                                 break;
-                        } else if (buffer_delay(bh)) {
-                                acceptable = 1;
-                        }
                 } while ((bh = bh->b_this_page) != head);
 
                 if (acceptable)
-                        return page;
-        }
-
-out:
-        unlock_page(page);
-        return NULL;
-}
-
-STATIC int
-xfs_map_unwritten(
-        struct inode            *inode,
-        struct page             *start_page,
-        struct buffer_head      *head,
-        struct buffer_head      *curr,
-        unsigned long           p_offset,
-        int                     block_bits,
-        xfs_iomap_t             *iomapp,
-        struct writeback_control *wbc,
-        int                     startio,
-        int                     all_bh)
-{
-        struct buffer_head      *bh = curr;
-        xfs_iomap_t             *tmp;
-        xfs_ioend_t             *ioend;
-        loff_t                  offset;
-        unsigned long           nblocks = 0;
-
-        offset = start_page->index;
-        offset <<= PAGE_CACHE_SHIFT;
-        offset += p_offset;
-
-        ioend = xfs_alloc_ioend(inode);
-
-        /* First map forwards in the page consecutive buffers
-         * covering this unwritten extent
-         */
-        do {
-                if (!buffer_unwritten(bh))
-                        break;
-                tmp = xfs_offset_to_map(start_page, iomapp, p_offset);
-                if (!tmp)
-                        break;
-                xfs_map_at_offset(start_page, bh, p_offset, block_bits, iomapp);
-                set_buffer_unwritten_io(bh);
-                bh->b_private = ioend;
-                p_offset += bh->b_size;
-                nblocks++;
-        } while ((bh = bh->b_this_page) != head);
-
-        atomic_add(nblocks, &ioend->io_remaining);
-
-        /* If we reached the end of the page, map forwards in any
-         * following pages which are also covered by this extent.
-         */
-        if (bh == head) {
-                struct address_space    *mapping = inode->i_mapping;
-                pgoff_t                 tindex, tloff, tlast;
-                unsigned long           bs;
-                unsigned int            pg_offset, bbits = inode->i_blkbits;
-                struct page             *page;
-
-                tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
-                tloff = (iomapp->iomap_offset + iomapp->iomap_bsize) >> PAGE_CACHE_SHIFT;
-                tloff = min(tlast, tloff);
-                for (tindex = start_page->index + 1; tindex < tloff; tindex++) {
-                        page = xfs_probe_unwritten_page(mapping,
-                                                tindex, iomapp, ioend,
-                                                PAGE_CACHE_SIZE, &bs, bbits);
-                        if (!page)
-                                break;
-                        nblocks += bs;
-                        atomic_add(bs, &ioend->io_remaining);
-                        xfs_convert_page(inode, page, iomapp, wbc, ioend,
-                                                        startio, all_bh);
-                        /* stop if converting the next page might add
-                         * enough blocks that the corresponding byte
-                         * count won't fit in our ulong page buf length */
-                        if (nblocks >= ((ULONG_MAX - PAGE_SIZE) >> block_bits))
-                                goto enough;
-                }
-
-                if (tindex == tlast &&
-                    (pg_offset = (i_size_read(inode) & (PAGE_CACHE_SIZE - 1)))) {
-                        page = xfs_probe_unwritten_page(mapping,
-                                                        tindex, iomapp, ioend,
-                                                        pg_offset, &bs, bbits);
-                        if (page) {
-                                nblocks += bs;
-                                atomic_add(bs, &ioend->io_remaining);
-                                xfs_convert_page(inode, page, iomapp, wbc, ioend,
-                                                        startio, all_bh);
-                                if (nblocks >= ((ULONG_MAX - PAGE_SIZE) >> block_bits))
-                                        goto enough;
-                        }
-                }
+                        return 1;
         }
 
-enough:
-        ioend->io_size = (xfs_off_t)nblocks << block_bits;
-        ioend->io_offset = offset;
-        xfs_finish_ioend(ioend);
         return 0;
 }
 
-STATIC void
-xfs_submit_page(
-        struct page             *page,
-        struct writeback_control *wbc,
-        struct buffer_head      *bh_arr[],
-        int                     bh_count,
-        int                     probed_page,
-        int                     clear_dirty)
-{
-        struct buffer_head      *bh;
-        int                     i;
-
-        BUG_ON(PageWriteback(page));
-        if (bh_count)
-                set_page_writeback(page);
-        if (clear_dirty)
-                clear_page_dirty(page);
-        unlock_page(page);
-
-        if (bh_count) {
-                for (i = 0; i < bh_count; i++) {
-                        bh = bh_arr[i];
-                        mark_buffer_async_write(bh);
-                        if (buffer_unwritten(bh))
-                                set_buffer_unwritten_io(bh);
-                        set_buffer_uptodate(bh);
-                        clear_buffer_dirty(bh);
-                }
-
-                for (i = 0; i < bh_count; i++)
-                        submit_bh(WRITE, bh_arr[i]);
-
-                if (probed_page && clear_dirty)
-                        wbc->nr_to_write--;     /* Wrote an "extra" page */
-        }
-}
-
 /*
  * Allocate & map buffers for page given the extent map. Write it out.
  * except for the original page of a writepage, this is called on
  * delalloc/unwritten pages only, for the original page it is possible
  * that the page has no mapping at all.
  */
-STATIC void
+STATIC int
 xfs_convert_page(
         struct inode            *inode,
         struct page             *page,
-        xfs_iomap_t             *iomapp,
+        loff_t                  tindex,
+        xfs_iomap_t             *mp,
+        xfs_ioend_t             **ioendp,
         struct writeback_control *wbc,
-        void                    *private,
         int                     startio,
         int                     all_bh)
 {
-        struct buffer_head      *bh_arr[MAX_BUF_PER_PAGE], *bh, *head;
-        xfs_iomap_t             *mp = iomapp, *tmp;
-        unsigned long           offset, end_offset;
-        int                     index = 0;
+        struct buffer_head      *bh, *head;
+        xfs_off_t               end_offset;
+        unsigned long           p_offset;
+        unsigned int            type;
         int                     bbits = inode->i_blkbits;
         int                     len, page_dirty;
+        int                     count = 0, done = 0, uptodate = 1;
+        xfs_off_t               offset = page_offset(page);
 
-        end_offset = (i_size_read(inode) & (PAGE_CACHE_SIZE - 1));
+        if (page->index != tindex)
+                goto fail;
+        if (TestSetPageLocked(page))
+                goto fail;
+        if (PageWriteback(page))
+                goto fail_unlock_page;
+        if (page->mapping != inode->i_mapping)
+                goto fail_unlock_page;
+        if (!xfs_is_delayed_page(page, (*ioendp)->io_type))
+                goto fail_unlock_page;
 
         /*
          * page_dirty is initially a count of buffers on the page before
          * EOF and is decrememted as we move each into a cleanable state.
+         *
+         * Derivation:
+         *
+         * End offset is the highest offset that this page should represent.
+         * If we are on the last page, (end_offset & (PAGE_CACHE_SIZE - 1))
+         * will evaluate non-zero and be less than PAGE_CACHE_SIZE and
+         * hence give us the correct page_dirty count. On any other page,
+         * it will be zero and in that case we need page_dirty to be the
+         * count of buffers on the page.
          */
+        end_offset = min_t(unsigned long long,
+                        (xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT,
+                        i_size_read(inode));
+
         len = 1 << inode->i_blkbits;
-        end_offset = max(end_offset, PAGE_CACHE_SIZE);
-        end_offset = roundup(end_offset, len);
-        page_dirty = end_offset / len;
+        p_offset = min_t(unsigned long, end_offset & (PAGE_CACHE_SIZE - 1),
+                                        PAGE_CACHE_SIZE);
+        p_offset = p_offset ? roundup(p_offset, len) : PAGE_CACHE_SIZE;
+        page_dirty = p_offset / len;
 
-        offset = 0;
         bh = head = page_buffers(page);
         do {
                 if (offset >= end_offset)
                         break;
-                if (!(PageUptodate(page) || buffer_uptodate(bh)))
+                if (!buffer_uptodate(bh))
+                        uptodate = 0;
+                if (!(PageUptodate(page) || buffer_uptodate(bh))) {
+                        done = 1;
                         continue;
-                if (buffer_mapped(bh) && all_bh &&
-                    !(buffer_unwritten(bh) || buffer_delay(bh))) {
+                }
+
+                if (buffer_unwritten(bh) || buffer_delay(bh)) {
+                        if (buffer_unwritten(bh))
+                                type = IOMAP_UNWRITTEN;
+                        else
+                                type = IOMAP_DELAY;
+
+                        if (!xfs_iomap_valid(mp, offset)) {
+                                done = 1;
+                                continue;
+                        }
+
+                        ASSERT(!(mp->iomap_flags & IOMAP_HOLE));
+                        ASSERT(!(mp->iomap_flags & IOMAP_DELAY));
+
+                        xfs_map_at_offset(bh, offset, bbits, mp);
                         if (startio) {
+                                xfs_add_to_ioend(inode, bh, offset,
+                                                type, ioendp, done);
+                        } else {
+                                set_buffer_dirty(bh);
+                                unlock_buffer(bh);
+                                mark_buffer_dirty(bh);
+                        }
+                        page_dirty--;
+                        count++;
+                } else {
+                        type = 0;
+                        if (buffer_mapped(bh) && all_bh && startio) {
                                 lock_buffer(bh);
-                                bh_arr[index++] = bh;
+                                xfs_add_to_ioend(inode, bh, offset,
+                                                type, ioendp, done);
+                                count++;
                                 page_dirty--;
+                        } else {
+                                done = 1;
                         }
-                        continue;
                 }
-                tmp = xfs_offset_to_map(page, mp, offset);
-                if (!tmp)
-                        continue;
-                ASSERT(!(tmp->iomap_flags & IOMAP_HOLE));
-                ASSERT(!(tmp->iomap_flags & IOMAP_DELAY));
+        } while (offset += len, (bh = bh->b_this_page) != head);
 
-                /* If this is a new unwritten extent buffer (i.e. one
-                 * that we haven't passed in private data for, we must
-                 * now map this buffer too.
-                 */
-                if (buffer_unwritten(bh) && !bh->b_end_io) {
-                        ASSERT(tmp->iomap_flags & IOMAP_UNWRITTEN);
-                        xfs_map_unwritten(inode, page, head, bh, offset,
-                                        bbits, tmp, wbc, startio, all_bh);
-                } else if (! (buffer_unwritten(bh) && buffer_locked(bh))) {
-                        xfs_map_at_offset(page, bh, offset, bbits, tmp);
-                        if (buffer_unwritten(bh)) {
-                                set_buffer_unwritten_io(bh);
-                                bh->b_private = private;
-                                ASSERT(private);
+        if (uptodate && bh == head)
+                SetPageUptodate(page);
+
+        if (startio) {
+                if (count) {
+                        struct backing_dev_info *bdi;
+
+                        bdi = inode->i_mapping->backing_dev_info;
+                        if (bdi_write_congested(bdi)) {
+                                wbc->encountered_congestion = 1;
+                                done = 1;
+                        } else if (--wbc->nr_to_write <= 0) {
+                                done = 1;
                         }
                 }
-                if (startio) {
-                        bh_arr[index++] = bh;
-                } else {
-                        set_buffer_dirty(bh);
-                        unlock_buffer(bh);
-                        mark_buffer_dirty(bh);
-                }
-                page_dirty--;
-        } while (offset += len, (bh = bh->b_this_page) != head);
-
-        if (startio && index) {
-                xfs_submit_page(page, wbc, bh_arr, index, 1, !page_dirty);
-        } else {
-                unlock_page(page);
+                xfs_start_page_writeback(page, wbc, !page_dirty, count);
         }
+
+        return done;
+ fail_unlock_page:
+        unlock_page(page);
+ fail:
+        return 1;
 }
 
 /*
@@ -685,19 +750,31 @@ xfs_cluster_write(
         struct inode            *inode,
         pgoff_t                 tindex,
         xfs_iomap_t             *iomapp,
+        xfs_ioend_t             **ioendp,
         struct writeback_control *wbc,
         int                     startio,
         int                     all_bh,
         pgoff_t                 tlast)
 {
-        struct page             *page;
+        struct pagevec          pvec;
+        int                     done = 0, i;
 
-        for (; tindex <= tlast; tindex++) {
-                page = xfs_probe_delalloc_page(inode, tindex);
-                if (!page)
+        pagevec_init(&pvec, 0);
+        while (!done && tindex <= tlast) {
+                unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
+
+                if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
                         break;
-                xfs_convert_page(inode, page, iomapp, wbc, NULL,
-                                startio, all_bh);
+
+                for (i = 0; i < pagevec_count(&pvec); i++) {
+                        done = xfs_convert_page(inode, pvec.pages[i], tindex++,
+                                        iomapp, ioendp, wbc, startio, all_bh);
+                        if (done)
+                                break;
+                }
+
+                pagevec_release(&pvec);
+                cond_resched();
         }
 }
 
@@ -728,18 +805,22 @@ xfs_page_state_convert(
         int             startio,
         int             unmapped) /* also implies page uptodate */
 {
-        struct buffer_head      *bh_arr[MAX_BUF_PER_PAGE], *bh, *head;
-        xfs_iomap_t             *iomp, iomap;
+        struct buffer_head      *bh, *head;
+        xfs_iomap_t             iomap;
+        xfs_ioend_t             *ioend = NULL, *iohead = NULL;
         loff_t                  offset;
         unsigned long           p_offset = 0;
+        unsigned int            type;
         __uint64_t              end_offset;
         pgoff_t                 end_index, last_index, tlast;
-        int                     len, err, i, cnt = 0, uptodate = 1;
-        int                     flags;
-        int                     page_dirty;
+        ssize_t                 size, len;
+        int                     flags, err, iomap_valid = 0, uptodate = 1;
+        int                     page_dirty, count = 0, trylock_flag = 0;
+        int                     all_bh = unmapped;
 
         /* wait for other IO threads? */
-        flags = (startio && wbc->sync_mode != WB_SYNC_NONE) ? 0 : BMAPI_TRYLOCK;
+        if (startio && (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking))
+                trylock_flag |= BMAPI_TRYLOCK;
 
         /* Is this page beyond the end of the file? */
         offset = i_size_read(inode);
@@ -754,161 +835,173 @@ xfs_page_state_convert(
                 }
         }
 
-        end_offset = min_t(unsigned long long,
-                        (loff_t)(page->index + 1) << PAGE_CACHE_SHIFT, offset);
-        offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
-
         /*
          * page_dirty is initially a count of buffers on the page before
          * EOF and is decrememted as we move each into a cleanable state.
-         */
+         *
+         * Derivation:
+         *
+         * End offset is the highest offset that this page should represent.
+         * If we are on the last page, (end_offset & (PAGE_CACHE_SIZE - 1))
+         * will evaluate non-zero and be less than PAGE_CACHE_SIZE and
+         * hence give us the correct page_dirty count. On any other page,
+         * it will be zero and in that case we need page_dirty to be the
+         * count of buffers on the page.
+         */
+        end_offset = min_t(unsigned long long,
+                        (xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT, offset);
         len = 1 << inode->i_blkbits;
-        p_offset = max(p_offset, PAGE_CACHE_SIZE);
-        p_offset = roundup(p_offset, len);
+        p_offset = min_t(unsigned long, end_offset & (PAGE_CACHE_SIZE - 1),
+                                        PAGE_CACHE_SIZE);
+        p_offset = p_offset ? roundup(p_offset, len) : PAGE_CACHE_SIZE;
         page_dirty = p_offset / len;
 
-        iomp = NULL;
-        p_offset = 0;
         bh = head = page_buffers(page);
+        offset = page_offset(page);
+        flags = -1;
+        type = 0;
+
+        /* TODO: cleanup count and page_dirty */
 
         do {
                 if (offset >= end_offset)
                         break;
                 if (!buffer_uptodate(bh))
                         uptodate = 0;
-                if (!(PageUptodate(page) || buffer_uptodate(bh)) && !startio)
+                if (!(PageUptodate(page) || buffer_uptodate(bh)) && !startio) {
+                        /*
+                         * the iomap is actually still valid, but the ioend
+                         * isn't.  shouldn't happen too often.
+                         */
+                        iomap_valid = 0;
                         continue;
-
-                if (iomp) {
-                        iomp = xfs_offset_to_map(page, &iomap, p_offset);
                 }
 
+                if (iomap_valid)
+                        iomap_valid = xfs_iomap_valid(&iomap, offset);
+
                 /*
                  * First case, map an unwritten extent and prepare for
                  * extent state conversion transaction on completion.
-                 */
-                if (buffer_unwritten(bh)) {
-                        if (!startio)
-                                continue;
-                        if (!iomp) {
-                                err = xfs_map_blocks(inode, offset, len, &iomap,
-                                                BMAPI_WRITE|BMAPI_IGNSTATE);
-                                if (err) {
-                                        goto error;
-                                }
-                                iomp = xfs_offset_to_map(page, &iomap,
-                                                                p_offset);
+                 *
+                 * Second case, allocate space for a delalloc buffer.
+                 * We can return EAGAIN here in the release page case.
+                 *
+                 * Third case, an unmapped buffer was found, and we are
+                 * in a path where we need to write the whole page out.
+                 */
+                if (buffer_unwritten(bh) || buffer_delay(bh) ||
+                    ((buffer_uptodate(bh) || PageUptodate(page)) &&
+                     !buffer_mapped(bh) && (unmapped || startio))) {
+                        /*
+                         * Make sure we don't use a read-only iomap
+                         */
+                        if (flags == BMAPI_READ)
+                                iomap_valid = 0;
+
+                        if (buffer_unwritten(bh)) {
+                                type = IOMAP_UNWRITTEN;
+                                flags = BMAPI_WRITE|BMAPI_IGNSTATE;
+                        } else if (buffer_delay(bh)) {
+                                type = IOMAP_DELAY;
+                                flags = BMAPI_ALLOCATE;
+                                if (!startio)
+                                        flags |= trylock_flag;
+                        } else {
+                                type = IOMAP_NEW;
+                                flags = BMAPI_WRITE|BMAPI_MMAP;
                         }
-                        if (iomp) {
-                                if (!bh->b_end_io) {
-                                        err = xfs_map_unwritten(inode, page,
-                                                        head, bh, p_offset,
-                                                        inode->i_blkbits, iomp,
-                                                        wbc, startio, unmapped);
-                                        if (err) {
-                                                goto error;
-                                        }
+
+                        if (!iomap_valid) {
+                                if (type == IOMAP_NEW) {
+                                        size = xfs_probe_cluster(inode,
+                                                        page, bh, head, 0);
                                 } else {
-                                        set_bit(BH_Lock, &bh->b_state);
+                                        size = len;
                                 }
-                                BUG_ON(!buffer_locked(bh));
-                                bh_arr[cnt++] = bh;
-                                page_dirty--;
-                        }
-                /*
-                 * Second case, allocate space for a delalloc buffer.
-                 * We can return EAGAIN here in the release page case.
-                 */
-                } else if (buffer_delay(bh)) {
-                        if (!iomp) {
-                                err = xfs_map_blocks(inode, offset, len, &iomap,
-                                                BMAPI_ALLOCATE | flags);
-                                if (err) {
+
+                                err = xfs_map_blocks(inode, offset, size,
+                                                &iomap, flags);
+                                if (err)
                                         goto error;
-                                }
-                                iomp = xfs_offset_to_map(page, &iomap,
-                                                                p_offset);
+                                iomap_valid = xfs_iomap_valid(&iomap, offset);
                         }
-                        if (iomp) {
-                                xfs_map_at_offset(page, bh, p_offset,
-                                                inode->i_blkbits, iomp);
+                        if (iomap_valid) {
+                                xfs_map_at_offset(bh, offset,
+                                                inode->i_blkbits, &iomap);
                                 if (startio) {
-                                        bh_arr[cnt++] = bh;
+                                        xfs_add_to_ioend(inode, bh, offset,
+                                                        type, &ioend,
+                                                        !iomap_valid);
                                 } else {
                                         set_buffer_dirty(bh);
                                         unlock_buffer(bh);
                                         mark_buffer_dirty(bh);
                                 }
                                 page_dirty--;
+                                count++;
+                        }
+                } else if (buffer_uptodate(bh) && startio) {
+                        /*
+                         * we got here because the buffer is already mapped.
+                         * That means it must already have extents allocated
+                         * underneath it. Map the extent by reading it.
+                         */
+                        if (!iomap_valid || type != 0) {
+                                flags = BMAPI_READ;
+                                size = xfs_probe_cluster(inode, page, bh,
+                                                                head, 1);
+                                err = xfs_map_blocks(inode, offset, size,
+                                                &iomap, flags);
+                                if (err)
+                                        goto error;
+                                iomap_valid = xfs_iomap_valid(&iomap, offset);
                         }
-                } else if ((buffer_uptodate(bh) || PageUptodate(page)) &&
-                           (unmapped || startio)) {
 
-                        if (!buffer_mapped(bh)) {
-                                int     size;
-
-                                /*
-                                 * Getting here implies an unmapped buffer
-                                 * was found, and we are in a path where we
-                                 * need to write the whole page out.
-                                 */
-                                if (!iomp) {
-                                        size = xfs_probe_unmapped_cluster(
-                                                        inode, page, bh, head);
-                                        err = xfs_map_blocks(inode, offset,
-                                                        size, &iomap,
-                                                        BMAPI_WRITE|BMAPI_MMAP);
-                                        if (err) {
-                                                goto error;
-                                        }
-                                        iomp = xfs_offset_to_map(page, &iomap,
-                                                                     p_offset);
-                                }
-                                if (iomp) {
-                                        xfs_map_at_offset(page,
-                                                        bh, p_offset,
-                                                        inode->i_blkbits, iomp);
-                                        if (startio) {
-                                                bh_arr[cnt++] = bh;
-                                        } else {
-                                                set_buffer_dirty(bh);
-                                                unlock_buffer(bh);
-                                                mark_buffer_dirty(bh);
-                                        }
-                                        page_dirty--;
-                                }
-                        } else if (startio) {
-                                if (buffer_uptodate(bh) &&
-                                    !test_and_set_bit(BH_Lock, &bh->b_state)) {
-                                        bh_arr[cnt++] = bh;
-                                        page_dirty--;
-                                }
+                        type = 0;
+                        if (!test_and_set_bit(BH_Lock, &bh->b_state)) {
+                                ASSERT(buffer_mapped(bh));
+                                if (iomap_valid)
+                                        all_bh = 1;
+                                xfs_add_to_ioend(inode, bh, offset, type,
+                                                &ioend, !iomap_valid);
+                                page_dirty--;
+                                count++;
+                        } else {
+                                iomap_valid = 0;
                         }
+                } else if ((buffer_uptodate(bh) || PageUptodate(page)) &&
+                           (unmapped || startio)) {
+                        iomap_valid = 0;
                 }
-        } while (offset += len, p_offset += len,
-                ((bh = bh->b_this_page) != head));
+
+                if (!iohead)
+                        iohead = ioend;
+
+        } while (offset += len, ((bh = bh->b_this_page) != head));
 
         if (uptodate && bh == head)
                 SetPageUptodate(page);
 
-        if (startio) {
-                xfs_submit_page(page, wbc, bh_arr, cnt, 0, !page_dirty);
-        }
+        if (startio)
+                xfs_start_page_writeback(page, wbc, 1, count);
 
-        if (iomp) {
-                offset = (iomp->iomap_offset + iomp->iomap_bsize - 1) >>
+        if (ioend && iomap_valid) {
+                offset = (iomap.iomap_offset + iomap.iomap_bsize - 1) >>
                                         PAGE_CACHE_SHIFT;
                 tlast = min_t(pgoff_t, offset, last_index);
-                xfs_cluster_write(inode, page->index + 1, iomp, wbc,
-                                        startio, unmapped, tlast);
+                xfs_cluster_write(inode, page->index + 1, &iomap, &ioend,
+                                        wbc, startio, all_bh, tlast);
         }
 
+        if (iohead)
+                xfs_submit_ioend(iohead);
+
         return page_dirty;
 
 error:
-        for (i = 0; i < cnt; i++) {
-                unlock_buffer(bh_arr[i]);
-        }
+        if (iohead)
+                xfs_cancel_ioend(iohead);
 
         /*
          * If it's delalloc and we have nowhere to put it,
@@ -916,9 +1009,8 @@ error:
          * us to try again.
          */
         if (err != -EAGAIN) {
-                if (!unmapped) {
+                if (!unmapped)
                         block_invalidatepage(page, 0);
-                }
                 ClearPageUptodate(page);
         }
         return err;
@@ -982,7 +1074,7 @@ __linvfs_get_block(
         }
 
         /* If this is a realtime file, data might be on a new device */
-        bh_result->b_bdev = iomap.iomap_target->pbr_bdev;
+        bh_result->b_bdev = iomap.iomap_target->bt_bdev;
 
         /* If we previously allocated a block out beyond eof and
          * we are now coming back to use it then we will need to
@@ -1094,10 +1186,10 @@ linvfs_direct_IO(
         if (error)
                 return -error;
 
-        iocb->private = xfs_alloc_ioend(inode);
+        iocb->private = xfs_alloc_ioend(inode, IOMAP_UNWRITTEN);
 
         ret = blockdev_direct_IO_own_locking(rw, iocb, inode,
-                iomap.iomap_target->pbr_bdev,
+                iomap.iomap_target->bt_bdev,
                 iov, offset, nr_segs,
                 linvfs_get_blocks_direct,
                 linvfs_end_io_direct);
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
index 4720758a9ade..55339dd5a30d 100644
--- a/fs/xfs/linux-2.6/xfs_aops.h
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -23,14 +23,24 @@ extern mempool_t *xfs_ioend_pool;
 
 typedef void (*xfs_ioend_func_t)(void *);
 
+/*
+ * xfs_ioend struct manages large extent writes for XFS.
+ * It can manage several multi-page bio's at once.
+ */
 typedef struct xfs_ioend {
+        struct xfs_ioend        *io_list;       /* next ioend in chain */
+        unsigned int            io_type;        /* delalloc / unwritten */
         unsigned int            io_uptodate;    /* I/O status register */
         atomic_t                io_remaining;   /* hold count */
         struct vnode            *io_vnode;      /* file being written to */
         struct buffer_head      *io_buffer_head;/* buffer linked list head */
+        struct buffer_head      *io_buffer_tail;/* buffer linked list tail */
         size_t                  io_size;        /* size of the extent */
         xfs_off_t               io_offset;      /* offset in the file */
         struct work_struct      io_work;        /* xfsdatad work queue */
 } xfs_ioend_t;
 
+extern struct address_space_operations linvfs_aops;
+extern int linvfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
+
 #endif /* __XFS_IOPS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 6fe21d2b8847..e44b7c1a3a36 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -31,76 +31,77 @@
 #include <linux/kthread.h>
 #include "xfs_linux.h"
 
-STATIC kmem_cache_t *pagebuf_zone;
-STATIC kmem_shaker_t pagebuf_shake;
+STATIC kmem_zone_t *xfs_buf_zone;
+STATIC kmem_shaker_t xfs_buf_shake;
+STATIC int xfsbufd(void *);
 STATIC int xfsbufd_wakeup(int, gfp_t);
-STATIC void pagebuf_delwri_queue(xfs_buf_t *, int);
+STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
 
 STATIC struct workqueue_struct *xfslogd_workqueue;
 struct workqueue_struct *xfsdatad_workqueue;
 
-#ifdef PAGEBUF_TRACE
+#ifdef XFS_BUF_TRACE
 void
-pagebuf_trace(
-        xfs_buf_t       *pb,
+xfs_buf_trace(
+        xfs_buf_t       *bp,
         char            *id,
         void            *data,
         void            *ra)
 {
-        ktrace_enter(pagebuf_trace_buf,
-                pb, id,
-                (void *)(unsigned long)pb->pb_flags,
-                (void *)(unsigned long)pb->pb_hold.counter,
-                (void *)(unsigned long)pb->pb_sema.count.counter,
+        ktrace_enter(xfs_buf_trace_buf,
+                bp, id,
+                (void *)(unsigned long)bp->b_flags,
+                (void *)(unsigned long)bp->b_hold.counter,
+                (void *)(unsigned long)bp->b_sema.count.counter,
                 (void *)current,
                 data, ra,
-                (void *)(unsigned long)((pb->pb_file_offset>>32) & 0xffffffff),
-                (void *)(unsigned long)(pb->pb_file_offset & 0xffffffff),
-                (void *)(unsigned long)pb->pb_buffer_length,
+                (void *)(unsigned long)((bp->b_file_offset>>32) & 0xffffffff),
+                (void *)(unsigned long)(bp->b_file_offset & 0xffffffff),
+                (void *)(unsigned long)bp->b_buffer_length,
                 NULL, NULL, NULL, NULL, NULL);
 }
-ktrace_t *pagebuf_trace_buf;
-#define PAGEBUF_TRACE_SIZE      4096
-#define PB_TRACE(pb, id, data)  \
-        pagebuf_trace(pb, id, (void *)data, (void *)__builtin_return_address(0))
+ktrace_t *xfs_buf_trace_buf;
+#define XFS_BUF_TRACE_SIZE      4096
+#define XB_TRACE(bp, id, data)  \
+        xfs_buf_trace(bp, id, (void *)data, (void *)__builtin_return_address(0))
 #else
-#define PB_TRACE(pb, id, data)  do { } while (0)
+#define XB_TRACE(bp, id, data)  do { } while (0)
 #endif
 
-#ifdef PAGEBUF_LOCK_TRACKING
-# define PB_SET_OWNER(pb)       ((pb)->pb_last_holder = current->pid)
-# define PB_CLEAR_OWNER(pb)     ((pb)->pb_last_holder = -1)
-# define PB_GET_OWNER(pb)       ((pb)->pb_last_holder)
+#ifdef XFS_BUF_LOCK_TRACKING
+# define XB_SET_OWNER(bp)       ((bp)->b_last_holder = current->pid)
+# define XB_CLEAR_OWNER(bp)     ((bp)->b_last_holder = -1)
+# define XB_GET_OWNER(bp)       ((bp)->b_last_holder)
 #else
-# define PB_SET_OWNER(pb)       do { } while (0)
-# define PB_CLEAR_OWNER(pb)     do { } while (0)
-# define PB_GET_OWNER(pb)       do { } while (0)
+# define XB_SET_OWNER(bp)       do { } while (0)
+# define XB_CLEAR_OWNER(bp)     do { } while (0)
+# define XB_GET_OWNER(bp)       do { } while (0)
 #endif
 
-#define pb_to_gfp(flags) \
-        ((((flags) & PBF_READ_AHEAD) ? __GFP_NORETRY : \
-          ((flags) & PBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
+#define xb_to_gfp(flags) \
+        ((((flags) & XBF_READ_AHEAD) ? __GFP_NORETRY : \
+          ((flags) & XBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
 
-#define pb_to_km(flags) \
-         (((flags) & PBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
+#define xb_to_km(flags) \
+         (((flags) & XBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
 
-#define pagebuf_allocate(flags) \
-        kmem_zone_alloc(pagebuf_zone, pb_to_km(flags))
-#define pagebuf_deallocate(pb) \
-        kmem_zone_free(pagebuf_zone, (pb));
+#define xfs_buf_allocate(flags) \
+        kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags))
+#define xfs_buf_deallocate(bp) \
+        kmem_zone_free(xfs_buf_zone, (bp));
 
 /*
- * Page Region interfaces.
+ *      Page Region interfaces.
  *
- * For pages in filesystems where the blocksize is smaller than the
- * pagesize, we use the page->private field (long) to hold a bitmap
- * of uptodate regions within the page.
+ *      For pages in filesystems where the blocksize is smaller than the
+ *      pagesize, we use the page->private field (long) to hold a bitmap
+ *      of uptodate regions within the page.
  *
- * Each such region is "bytes per page / bits per long" bytes long.
+ *      Each such region is "bytes per page / bits per long" bytes long.
  *
- * NBPPR == number-of-bytes-per-page-region
- * BTOPR == bytes-to-page-region (rounded up)
- * BTOPRT == bytes-to-page-region-truncated (rounded down)
+ *      NBPPR == number-of-bytes-per-page-region
+ *      BTOPR == bytes-to-page-region (rounded up)
+ *      BTOPRT == bytes-to-page-region-truncated (rounded down)
  */
 #if (BITS_PER_LONG == 32)
 #define PRSHIFT         (PAGE_CACHE_SHIFT - 5)  /* (32 == 1<<5) */
@@ -159,7 +160,7 @@ test_page_region(
 }
 
 /*
- * Mapping of multi-page buffers into contiguous virtual space
+ *      Mapping of multi-page buffers into contiguous virtual space
  */
 
 typedef struct a_list {
@@ -172,7 +173,7 @@ STATIC int		as_list_len;
 STATIC DEFINE_SPINLOCK(as_lock);
 
 /*
- * Try to batch vunmaps because they are costly.
+ *      Try to batch vunmaps because they are costly.
  */
 STATIC void
 free_address(
@@ -215,83 +216,83 @@ purge_addresses(void)
 }
 
 /*
- *      Internal pagebuf object manipulation
+ *      Internal xfs_buf_t object manipulation
  */
 
 STATIC void
-_pagebuf_initialize(
-        xfs_buf_t               *pb,
+_xfs_buf_initialize(
+        xfs_buf_t               *bp,
         xfs_buftarg_t           *target,
-        loff_t                  range_base,
+        xfs_off_t               range_base,
         size_t                  range_length,
-        page_buf_flags_t        flags)
+        xfs_buf_flags_t         flags)
 {
         /*
-         * We don't want certain flags to appear in pb->pb_flags.
+         * We don't want certain flags to appear in b_flags.
          */
-        flags &= ~(PBF_LOCK|PBF_MAPPED|PBF_DONT_BLOCK|PBF_READ_AHEAD);
-
-        memset(pb, 0, sizeof(xfs_buf_t));
-        atomic_set(&pb->pb_hold, 1);
-        init_MUTEX_LOCKED(&pb->pb_iodonesema);
-        INIT_LIST_HEAD(&pb->pb_list);
-        INIT_LIST_HEAD(&pb->pb_hash_list);
-        init_MUTEX_LOCKED(&pb->pb_sema); /* held, no waiters */
-        PB_SET_OWNER(pb);
-        pb->pb_target = target;
-        pb->pb_file_offset = range_base;
+        flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);
+
+        memset(bp, 0, sizeof(xfs_buf_t));
+        atomic_set(&bp->b_hold, 1);
+        init_MUTEX_LOCKED(&bp->b_iodonesema);
+        INIT_LIST_HEAD(&bp->b_list);
+        INIT_LIST_HEAD(&bp->b_hash_list);
+        init_MUTEX_LOCKED(&bp->b_sema); /* held, no waiters */
+        XB_SET_OWNER(bp);
+        bp->b_target = target;
+        bp->b_file_offset = range_base;
         /*
          * Set buffer_length and count_desired to the same value initially.
          * I/O routines should use count_desired, which will be the same in
          * most cases but may be reset (e.g. XFS recovery).
          */
-        pb->pb_buffer_length = pb->pb_count_desired = range_length;
-        pb->pb_flags = flags;
-        pb->pb_bn = XFS_BUF_DADDR_NULL;
-        atomic_set(&pb->pb_pin_count, 0);
-        init_waitqueue_head(&pb->pb_waiters);
-
-        XFS_STATS_INC(pb_create);
-        PB_TRACE(pb, "initialize", target);
+        bp->b_buffer_length = bp->b_count_desired = range_length;
+        bp->b_flags = flags;
+        bp->b_bn = XFS_BUF_DADDR_NULL;
+        atomic_set(&bp->b_pin_count, 0);
+        init_waitqueue_head(&bp->b_waiters);
+
+        XFS_STATS_INC(xb_create);
+        XB_TRACE(bp, "initialize", target);
 }
 
 /*
- * Allocate a page array capable of holding a specified number
- * of pages, and point the page buf at it.
+ *      Allocate a page array capable of holding a specified number
+ *      of pages, and point the page buf at it.
  */
 STATIC int
-_pagebuf_get_pages(
-        xfs_buf_t               *pb,
+_xfs_buf_get_pages(
+        xfs_buf_t               *bp,
         int                     page_count,
-        page_buf_flags_t        flags)
+        xfs_buf_flags_t         flags)
 {
         /* Make sure that we have a page list */
-        if (pb->pb_pages == NULL) {
-                pb->pb_offset = page_buf_poff(pb->pb_file_offset);
-                pb->pb_page_count = page_count;
-                if (page_count <= PB_PAGES) {
-                        pb->pb_pages = pb->pb_page_array;
+        if (bp->b_pages == NULL) {
+                bp->b_offset = xfs_buf_poff(bp->b_file_offset);
+                bp->b_page_count = page_count;
+                if (page_count <= XB_PAGES) {
+                        bp->b_pages = bp->b_page_array;
                 } else {
-                        pb->pb_pages = kmem_alloc(sizeof(struct page *) *
-                                        page_count, pb_to_km(flags));
-                        if (pb->pb_pages == NULL)
+                        bp->b_pages = kmem_alloc(sizeof(struct page *) *
+                                        page_count, xb_to_km(flags));
+                        if (bp->b_pages == NULL)
                                 return -ENOMEM;
                 }
-                memset(pb->pb_pages, 0, sizeof(struct page *) * page_count);
+                memset(bp->b_pages, 0, sizeof(struct page *) * page_count);
         }
         return 0;
 }
 
 /*
- *      Frees pb_pages if it was malloced.
+ *      Frees b_pages if it was allocated.
  */
 STATIC void
-_pagebuf_free_pages(
+_xfs_buf_free_pages(
         xfs_buf_t       *bp)
 {
-        if (bp->pb_pages != bp->pb_page_array) {
-                kmem_free(bp->pb_pages,
-                          bp->pb_page_count * sizeof(struct page *));
+        if (bp->b_pages != bp->b_page_array) {
+                kmem_free(bp->b_pages,
+                          bp->b_page_count * sizeof(struct page *));
         }
 }
 
@@ -299,79 +300,79 @@ _pagebuf_free_pages(
  *      Releases the specified buffer.
  *
  *      The modification state of any associated pages is left unchanged.
- *      The buffer most not be on any hash - use pagebuf_rele instead for
+ *      The buffer most not be on any hash - use xfs_buf_rele instead for
  *      hashed and refcounted buffers
  */
 void
-pagebuf_free(
+xfs_buf_free(
         xfs_buf_t               *bp)
 {
-        PB_TRACE(bp, "free", 0);
+        XB_TRACE(bp, "free", 0);
 
-        ASSERT(list_empty(&bp->pb_hash_list));
+        ASSERT(list_empty(&bp->b_hash_list));
 
-        if (bp->pb_flags & _PBF_PAGE_CACHE) {
+        if (bp->b_flags & _XBF_PAGE_CACHE) {
                 uint            i;
 
-                if ((bp->pb_flags & PBF_MAPPED) && (bp->pb_page_count > 1))
-                        free_address(bp->pb_addr - bp->pb_offset);
+                if ((bp->b_flags & XBF_MAPPED) && (bp->b_page_count > 1))
+                        free_address(bp->b_addr - bp->b_offset);
 
-                for (i = 0; i < bp->pb_page_count; i++)
-                        page_cache_release(bp->pb_pages[i]);
-                _pagebuf_free_pages(bp);
-        } else if (bp->pb_flags & _PBF_KMEM_ALLOC) {
+                for (i = 0; i < bp->b_page_count; i++)
+                        page_cache_release(bp->b_pages[i]);
+                _xfs_buf_free_pages(bp);
+        } else if (bp->b_flags & _XBF_KMEM_ALLOC) {
                  /*
-                  * XXX(hch): bp->pb_count_desired might be incorrect (see
-                  * pagebuf_associate_memory for details), but fortunately
+                  * XXX(hch): bp->b_count_desired might be incorrect (see
+                  * xfs_buf_associate_memory for details), but fortunately
                   * the Linux version of kmem_free ignores the len argument..
                   */
-                kmem_free(bp->pb_addr, bp->pb_count_desired);
-                _pagebuf_free_pages(bp);
+                kmem_free(bp->b_addr, bp->b_count_desired);
+                _xfs_buf_free_pages(bp);
         }
 
-        pagebuf_deallocate(bp);
+        xfs_buf_deallocate(bp);
 }
 
 /*
  *      Finds all pages for buffer in question and builds it's page list.
  */
 STATIC int
-_pagebuf_lookup_pages(
+_xfs_buf_lookup_pages(
         xfs_buf_t               *bp,
         uint                    flags)
 {
-        struct address_space    *mapping = bp->pb_target->pbr_mapping;
-        size_t                  blocksize = bp->pb_target->pbr_bsize;
-        size_t                  size = bp->pb_count_desired;
+        struct address_space    *mapping = bp->b_target->bt_mapping;
+        size_t                  blocksize = bp->b_target->bt_bsize;
+        size_t                  size = bp->b_count_desired;
         size_t                  nbytes, offset;
-        gfp_t                   gfp_mask = pb_to_gfp(flags);
+        gfp_t                   gfp_mask = xb_to_gfp(flags);
         unsigned short          page_count, i;
         pgoff_t                 first;
-        loff_t                  end;
+        xfs_off_t               end;
         int                     error;
 
-        end = bp->pb_file_offset + bp->pb_buffer_length;
-        page_count = page_buf_btoc(end) - page_buf_btoct(bp->pb_file_offset);
+        end = bp->b_file_offset + bp->b_buffer_length;
+        page_count = xfs_buf_btoc(end) - xfs_buf_btoct(bp->b_file_offset);
 
-        error = _pagebuf_get_pages(bp, page_count, flags);
+        error = _xfs_buf_get_pages(bp, page_count, flags);
         if (unlikely(error))
                 return error;
-        bp->pb_flags |= _PBF_PAGE_CACHE;
+        bp->b_flags |= _XBF_PAGE_CACHE;
 
-        offset = bp->pb_offset;
-        first = bp->pb_file_offset >> PAGE_CACHE_SHIFT;
+        offset = bp->b_offset;
+        first = bp->b_file_offset >> PAGE_CACHE_SHIFT;
 
-        for (i = 0; i < bp->pb_page_count; i++) {
+        for (i = 0; i < bp->b_page_count; i++) {
                 struct page     *page;
                 uint            retries = 0;
 
               retry:
                 page = find_or_create_page(mapping, first + i, gfp_mask);
                 if (unlikely(page == NULL)) {
-                        if (flags & PBF_READ_AHEAD) {
-                                bp->pb_page_count = i;
-                                for (i = 0; i < bp->pb_page_count; i++)
-                                        unlock_page(bp->pb_pages[i]);
+                        if (flags & XBF_READ_AHEAD) {
+                                bp->b_page_count = i;
+                                for (i = 0; i < bp->b_page_count; i++)
+                                        unlock_page(bp->b_pages[i]);
                                 return -ENOMEM;
                         }
 
@@ -387,13 +388,13 @@ _pagebuf_lookup_pages(
                                         "deadlock in %s (mode:0x%x)\n",
                                         __FUNCTION__, gfp_mask);
 
-                        XFS_STATS_INC(pb_page_retries);
+                        XFS_STATS_INC(xb_page_retries);
                         xfsbufd_wakeup(0, gfp_mask);
                         blk_congestion_wait(WRITE, HZ/50);
                         goto retry;
                 }
 
-                XFS_STATS_INC(pb_page_found);
+                XFS_STATS_INC(xb_page_found);
 
                 nbytes = min_t(size_t, size, PAGE_CACHE_SIZE - offset);
                 size -= nbytes;
@@ -401,27 +402,27 @@ _pagebuf_lookup_pages(
                 if (!PageUptodate(page)) {
                         page_count--;
                         if (blocksize >= PAGE_CACHE_SIZE) {
-                                if (flags & PBF_READ)
-                                        bp->pb_locked = 1;
+                                if (flags & XBF_READ)
+                                        bp->b_locked = 1;
                         } else if (!PagePrivate(page)) {
                                 if (test_page_region(page, offset, nbytes))
                                         page_count++;
                         }
                 }
 
-                bp->pb_pages[i] = page;
+                bp->b_pages[i] = page;
                 offset = 0;
         }
 
-        if (!bp->pb_locked) {
-                for (i = 0; i < bp->pb_page_count; i++)
-                        unlock_page(bp->pb_pages[i]);
+        if (!bp->b_locked) {
+                for (i = 0; i < bp->b_page_count; i++)
+                        unlock_page(bp->b_pages[i]);
         }
 
-        if (page_count == bp->pb_page_count)
-                bp->pb_flags |= PBF_DONE;
+        if (page_count == bp->b_page_count)
+                bp->b_flags |= XBF_DONE;
 
-        PB_TRACE(bp, "lookup_pages", (long)page_count);
+        XB_TRACE(bp, "lookup_pages", (long)page_count);
         return error;
 }
 
@@ -429,23 +430,23 @@ _pagebuf_lookup_pages(
  *      Map buffer into kernel address-space if nessecary.
  */
 STATIC int
-_pagebuf_map_pages(
+_xfs_buf_map_pages(
         xfs_buf_t               *bp,
         uint                    flags)
 {
         /* A single page buffer is always mappable */
-        if (bp->pb_page_count == 1) {
-                bp->pb_addr = page_address(bp->pb_pages[0]) + bp->pb_offset;
-                bp->pb_flags |= PBF_MAPPED;
-        } else if (flags & PBF_MAPPED) {
+        if (bp->b_page_count == 1) {
+                bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset;
+                bp->b_flags |= XBF_MAPPED;
+        } else if (flags & XBF_MAPPED) {
                 if (as_list_len > 64)
                         purge_addresses();
-                bp->pb_addr = vmap(bp->pb_pages, bp->pb_page_count,
-                                VM_MAP, PAGE_KERNEL);
-                if (unlikely(bp->pb_addr == NULL))
+                bp->b_addr = vmap(bp->b_pages, bp->b_page_count,
+                                        VM_MAP, PAGE_KERNEL);
+                if (unlikely(bp->b_addr == NULL))
                         return -ENOMEM;
-                bp->pb_addr += bp->pb_offset;
-                bp->pb_flags |= PBF_MAPPED;
+                bp->b_addr += bp->b_offset;
+                bp->b_flags |= XBF_MAPPED;
         }
 
         return 0;
@@ -456,9 +457,7 @@ _pagebuf_map_pages(
  */
 
 /*
- *      _pagebuf_find
- *
- *      Looks up, and creates if absent, a lockable buffer for
+ *      Look up, and creates if absent, a lockable buffer for
  *      a given range of an inode.  The buffer is returned
  *      locked.  If other overlapping buffers exist, they are
  *      released before the new buffer is created and locked,
@@ -466,55 +465,55 @@ _pagebuf_map_pages(
  *      are unlocked.  No I/O is implied by this call.
  */
 xfs_buf_t *
-_pagebuf_find(
+_xfs_buf_find(
         xfs_buftarg_t           *btp,   /* block device target          */
-        loff_t                  ioff,   /* starting offset of range     */
+        xfs_off_t               ioff,   /* starting offset of range     */
         size_t                  isize,  /* length of range              */
-        page_buf_flags_t        flags,  /* PBF_TRYLOCK                  */
-        xfs_buf_t               *new_pb)/* newly allocated buffer       */
+        xfs_buf_flags_t         flags,
+        xfs_buf_t               *new_bp)
 {
-        loff_t                  range_base;
+        xfs_off_t               range_base;
         size_t                  range_length;
         xfs_bufhash_t           *hash;
-        xfs_buf_t               *pb, *n;
+        xfs_buf_t               *bp, *n;
 
         range_base = (ioff << BBSHIFT);
         range_length = (isize << BBSHIFT);
 
         /* Check for IOs smaller than the sector size / not sector aligned */
-        ASSERT(!(range_length < (1 << btp->pbr_sshift)));
-        ASSERT(!(range_base & (loff_t)btp->pbr_smask));
+        ASSERT(!(range_length < (1 << btp->bt_sshift)));
+        ASSERT(!(range_base & (xfs_off_t)btp->bt_smask));
 
         hash = &btp->bt_hash[hash_long((unsigned long)ioff, btp->bt_hashshift)];
 
         spin_lock(&hash->bh_lock);
 
-        list_for_each_entry_safe(pb, n, &hash->bh_list, pb_hash_list) {
-                ASSERT(btp == pb->pb_target);
-                if (pb->pb_file_offset == range_base &&
-                    pb->pb_buffer_length == range_length) {
+        list_for_each_entry_safe(bp, n, &hash->bh_list, b_hash_list) {
+                ASSERT(btp == bp->b_target);
+                if (bp->b_file_offset == range_base &&
+                    bp->b_buffer_length == range_length) {
                         /*
-                         * If we look at something bring it to the
+                         * If we look at something, bring it to the
                          * front of the list for next time.
                          */
-                        atomic_inc(&pb->pb_hold);
-                        list_move(&pb->pb_hash_list, &hash->bh_list);
+                        atomic_inc(&bp->b_hold);
+                        list_move(&bp->b_hash_list, &hash->bh_list);
                         goto found;
                 }
         }
 
         /* No match found */
-        if (new_pb) {
-                _pagebuf_initialize(new_pb, btp, range_base,
+        if (new_bp) {
+                _xfs_buf_initialize(new_bp, btp, range_base,
                                 range_length, flags);
-                new_pb->pb_hash = hash;
-                list_add(&new_pb->pb_hash_list, &hash->bh_list);
+                new_bp->b_hash = hash;
+                list_add(&new_bp->b_hash_list, &hash->bh_list);
         } else {
-                XFS_STATS_INC(pb_miss_locked);
+                XFS_STATS_INC(xb_miss_locked);
         }
 
         spin_unlock(&hash->bh_lock);
-        return new_pb;
+        return new_bp;
 
 found:
         spin_unlock(&hash->bh_lock);
@@ -523,74 +522,72 @@ found:
          * if this does not work then we need to drop the
          * spinlock and do a hard attempt on the semaphore.
          */
-        if (down_trylock(&pb->pb_sema)) {
-                if (!(flags & PBF_TRYLOCK)) {
+        if (down_trylock(&bp->b_sema)) {
+                if (!(flags & XBF_TRYLOCK)) {
                         /* wait for buffer ownership */
-                        PB_TRACE(pb, "get_lock", 0);
-                        pagebuf_lock(pb);
-                        XFS_STATS_INC(pb_get_locked_waited);
+                        XB_TRACE(bp, "get_lock", 0);
+                        xfs_buf_lock(bp);
+                        XFS_STATS_INC(xb_get_locked_waited);
                 } else {
                         /* We asked for a trylock and failed, no need
                          * to look at file offset and length here, we
-                         * know that this pagebuf at least overlaps our
-                         * pagebuf and is locked, therefore our buffer
-                         * either does not exist, or is this buffer
+                         * know that this buffer at least overlaps our
+                         * buffer and is locked, therefore our buffer
+                         * either does not exist, or is this buffer.
                          */
-
-                        pagebuf_rele(pb);
-                        XFS_STATS_INC(pb_busy_locked);
-                        return (NULL);
+                        xfs_buf_rele(bp);
+                        XFS_STATS_INC(xb_busy_locked);
+                        return NULL;
                 }
         } else {
                 /* trylock worked */
-                PB_SET_OWNER(pb);
+                XB_SET_OWNER(bp);
         }
 
-        if (pb->pb_flags & PBF_STALE) {
-                ASSERT((pb->pb_flags & _PBF_DELWRI_Q) == 0);
-                pb->pb_flags &= PBF_MAPPED;
+        if (bp->b_flags & XBF_STALE) {
+                ASSERT((bp->b_flags & _XBF_DELWRI_Q) == 0);
+                bp->b_flags &= XBF_MAPPED;
         }
-        PB_TRACE(pb, "got_lock", 0);
-        XFS_STATS_INC(pb_get_locked);
-        return (pb);
+        XB_TRACE(bp, "got_lock", 0);
+        XFS_STATS_INC(xb_get_locked);
+        return bp;
 }
 
 /*
- *      xfs_buf_get_flags assembles a buffer covering the specified range.
- *
+ *      Assembles a buffer covering the specified range.
  *      Storage in memory for all portions of the buffer will be allocated,
  *      although backing storage may not be.
  */
 xfs_buf_t *
-xfs_buf_get_flags(                      /* allocate a buffer            */
+xfs_buf_get_flags(
         xfs_buftarg_t           *target,/* target for buffer            */
-        loff_t                  ioff,   /* starting offset of range     */
+        xfs_off_t               ioff,   /* starting offset of range     */
         size_t                  isize,  /* length of range              */
-        page_buf_flags_t        flags)  /* PBF_TRYLOCK                  */
+        xfs_buf_flags_t         flags)
 {
-        xfs_buf_t               *pb, *new_pb;
+        xfs_buf_t               *bp, *new_bp;
         int                     error = 0, i;
 
-        new_pb = pagebuf_allocate(flags);
-        if (unlikely(!new_pb))
+        new_bp = xfs_buf_allocate(flags);
+        if (unlikely(!new_bp))
                 return NULL;
 
-        pb = _pagebuf_find(target, ioff, isize, flags, new_pb);
-        if (pb == new_pb) {
-                error = _pagebuf_lookup_pages(pb, flags);
+        bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
+        if (bp == new_bp) {
+                error = _xfs_buf_lookup_pages(bp, flags);
                 if (error)
                         goto no_buffer;
         } else {
-                pagebuf_deallocate(new_pb);
-                if (unlikely(pb == NULL))
+                xfs_buf_deallocate(new_bp);
+                if (unlikely(bp == NULL))
                         return NULL;
         }
 
-        for (i = 0; i < pb->pb_page_count; i++)
-                mark_page_accessed(pb->pb_pages[i]);
+        for (i = 0; i < bp->b_page_count; i++)
+                mark_page_accessed(bp->b_pages[i]);
 
-        if (!(pb->pb_flags & PBF_MAPPED)) {
-                error = _pagebuf_map_pages(pb, flags);
+        if (!(bp->b_flags & XBF_MAPPED)) {
+                error = _xfs_buf_map_pages(bp, flags);
                 if (unlikely(error)) {
                         printk(KERN_WARNING "%s: failed to map pages\n",
                                         __FUNCTION__);
@@ -598,97 +595,97 @@ xfs_buf_get_flags(			/* allocate a buffer		*/
                 }
         }
 
-        XFS_STATS_INC(pb_get);
+        XFS_STATS_INC(xb_get);
 
         /*
          * Always fill in the block number now, the mapped cases can do
          * their own overlay of this later.
          */
-        pb->pb_bn = ioff;
-        pb->pb_count_desired = pb->pb_buffer_length;
+        bp->b_bn = ioff;
+        bp->b_count_desired = bp->b_buffer_length;
 
-        PB_TRACE(pb, "get", (unsigned long)flags);
-        return pb;
+        XB_TRACE(bp, "get", (unsigned long)flags);
+        return bp;
 
  no_buffer:
-        if (flags & (PBF_LOCK | PBF_TRYLOCK))
-                pagebuf_unlock(pb);
-        pagebuf_rele(pb);
+        if (flags & (XBF_LOCK | XBF_TRYLOCK))
+                xfs_buf_unlock(bp);
+        xfs_buf_rele(bp);
         return NULL;
 }
 
 xfs_buf_t *
 xfs_buf_read_flags(
         xfs_buftarg_t           *target,
-        loff_t                  ioff,
+        xfs_off_t               ioff,
         size_t                  isize,
-        page_buf_flags_t        flags)
+        xfs_buf_flags_t         flags)
 {
-        xfs_buf_t               *pb;
-
-        flags |= PBF_READ;
-
-        pb = xfs_buf_get_flags(target, ioff, isize, flags);
-        if (pb) {
-                if (!XFS_BUF_ISDONE(pb)) {
-                        PB_TRACE(pb, "read", (unsigned long)flags);
-                        XFS_STATS_INC(pb_get_read);
-                        pagebuf_iostart(pb, flags);
-                } else if (flags & PBF_ASYNC) {
-                        PB_TRACE(pb, "read_async", (unsigned long)flags);
+        xfs_buf_t               *bp;
+
+        flags |= XBF_READ;
+
+        bp = xfs_buf_get_flags(target, ioff, isize, flags);
+        if (bp) {
+                if (!XFS_BUF_ISDONE(bp)) {
+                        XB_TRACE(bp, "read", (unsigned long)flags);
+                        XFS_STATS_INC(xb_get_read);
+                        xfs_buf_iostart(bp, flags);
+                } else if (flags & XBF_ASYNC) {
+                        XB_TRACE(bp, "read_async", (unsigned long)flags);
                         /*
                          * Read ahead call which is already satisfied,
                          * drop the buffer
                          */
                         goto no_buffer;
                 } else {
-                        PB_TRACE(pb, "read_done", (unsigned long)flags);
+                        XB_TRACE(bp, "read_done", (unsigned long)flags);
                         /* We do not want read in the flags */
-                        pb->pb_flags &= ~PBF_READ;
+                        bp->b_flags &= ~XBF_READ;
                 }
         }
 
-        return pb;
+        return bp;
 
  no_buffer:
-        if (flags & (PBF_LOCK | PBF_TRYLOCK))
-                pagebuf_unlock(pb);
-        pagebuf_rele(pb);
+        if (flags & (XBF_LOCK | XBF_TRYLOCK))
+                xfs_buf_unlock(bp);
+        xfs_buf_rele(bp);
         return NULL;
 }
 
 /*
- * If we are not low on memory then do the readahead in a deadlock
- * safe manner.
+ *      If we are not low on memory then do the readahead in a deadlock
+ *      safe manner.
  */
 void
-pagebuf_readahead(
+xfs_buf_readahead(
         xfs_buftarg_t           *target,
-        loff_t                  ioff,
+        xfs_off_t               ioff,
         size_t                  isize,
-        page_buf_flags_t        flags)
+        xfs_buf_flags_t         flags)
 {
         struct backing_dev_info *bdi;
 
-        bdi = target->pbr_mapping->backing_dev_info;
+        bdi = target->bt_mapping->backing_dev_info;
         if (bdi_read_congested(bdi))
                 return;
 
-        flags |= (PBF_TRYLOCK|PBF_ASYNC|PBF_READ_AHEAD);
+        flags |= (XBF_TRYLOCK|XBF_ASYNC|XBF_READ_AHEAD);
         xfs_buf_read_flags(target, ioff, isize, flags);
 }
 
 xfs_buf_t *
-pagebuf_get_empty(
+xfs_buf_get_empty(
         size_t                  len,
         xfs_buftarg_t           *target)
 {
-        xfs_buf_t               *pb;
+        xfs_buf_t               *bp;
 
-        pb = pagebuf_allocate(0);
-        if (pb)
-                _pagebuf_initialize(pb, target, 0, len, 0);
-        return pb;
+        bp = xfs_buf_allocate(0);
+        if (bp)
+                _xfs_buf_initialize(bp, target, 0, len, 0);
+        return bp;
 }
 
 static inline struct page *
@@ -704,8 +701,8 @@ mem_to_page(
 }
 
 int
-pagebuf_associate_memory(
-        xfs_buf_t               *pb,
+xfs_buf_associate_memory(
+        xfs_buf_t               *bp,
         void                    *mem,
         size_t                  len)
 {
@@ -722,40 +719,40 @@ pagebuf_associate_memory(
                 page_count++;
 
         /* Free any previous set of page pointers */
-        if (pb->pb_pages)
-                _pagebuf_free_pages(pb);
+        if (bp->b_pages)
+                _xfs_buf_free_pages(bp);
 
-        pb->pb_pages = NULL;
-        pb->pb_addr = mem;
+        bp->b_pages = NULL;
+        bp->b_addr = mem;
 
-        rval = _pagebuf_get_pages(pb, page_count, 0);
+        rval = _xfs_buf_get_pages(bp, page_count, 0);
         if (rval)
                 return rval;
 
-        pb->pb_offset = offset;
+        bp->b_offset = offset;
         ptr = (size_t) mem & PAGE_CACHE_MASK;
         end = PAGE_CACHE_ALIGN((size_t) mem + len);
         end_cur = end;
         /* set up first page */
-        pb->pb_pages[0] = mem_to_page(mem);
+        bp->b_pages[0] = mem_to_page(mem);
 
         ptr += PAGE_CACHE_SIZE;
-        pb->pb_page_count = ++i;
+        bp->b_page_count = ++i;
         while (ptr < end) {
-                pb->pb_pages[i] = mem_to_page((void *)ptr);
-                pb->pb_page_count = ++i;
+                bp->b_pages[i] = mem_to_page((void *)ptr);
+                bp->b_page_count = ++i;
                 ptr += PAGE_CACHE_SIZE;
         }
-        pb->pb_locked = 0;
+        bp->b_locked = 0;
 
-        pb->pb_count_desired = pb->pb_buffer_length = len;
-        pb->pb_flags |= PBF_MAPPED;
+        bp->b_count_desired = bp->b_buffer_length = len;
+        bp->b_flags |= XBF_MAPPED;
 
         return 0;
 }
 
 xfs_buf_t *
-pagebuf_get_no_daddr(
+xfs_buf_get_noaddr(
         size_t                  len,
         xfs_buftarg_t           *target)
 {
@@ -764,10 +761,10 @@ pagebuf_get_no_daddr(
         void                    *data;
         int                     error;
 
-        bp = pagebuf_allocate(0);
+        bp = xfs_buf_allocate(0);
         if (unlikely(bp == NULL))
                 goto fail;
-        _pagebuf_initialize(bp, target, 0, len, 0);
+        _xfs_buf_initialize(bp, target, 0, len, 0);
 
  try_again:
         data = kmem_alloc(malloc_len, KM_SLEEP | KM_MAYFAIL);
@@ -776,78 +773,73 @@ pagebuf_get_no_daddr(
 
         /* check whether alignment matches.. */
         if ((__psunsigned_t)data !=
-            ((__psunsigned_t)data & ~target->pbr_smask)) {
+            ((__psunsigned_t)data & ~target->bt_smask)) {
                 /* .. else double the size and try again */
                 kmem_free(data, malloc_len);
                 malloc_len <<= 1;
                 goto try_again;
         }
 
-        error = pagebuf_associate_memory(bp, data, len);
+        error = xfs_buf_associate_memory(bp, data, len);
         if (error)
                 goto fail_free_mem;
-        bp->pb_flags |= _PBF_KMEM_ALLOC;
+        bp->b_flags |= _XBF_KMEM_ALLOC;
 
-        pagebuf_unlock(bp);
+        xfs_buf_unlock(bp);
 
-        PB_TRACE(bp, "no_daddr", data);
+        XB_TRACE(bp, "no_daddr", data);
         return bp;
  fail_free_mem:
         kmem_free(data, malloc_len);
  fail_free_buf:
-        pagebuf_free(bp);
+        xfs_buf_free(bp);
  fail:
         return NULL;
 }
 
 /*
- *      pagebuf_hold
- *
  *      Increment reference count on buffer, to hold the buffer concurrently
  *      with another thread which may release (free) the buffer asynchronously.
- *
  *      Must hold the buffer already to call this function.
  */
 void
-pagebuf_hold(
-        xfs_buf_t               *pb)
+xfs_buf_hold(
+        xfs_buf_t               *bp)
 {
-        atomic_inc(&pb->pb_hold);
-        PB_TRACE(pb, "hold", 0);
+        atomic_inc(&bp->b_hold);
+        XB_TRACE(bp, "hold", 0);
 }
 
 /*
- *      pagebuf_rele
- *
- *      pagebuf_rele releases a hold on the specified buffer.  If the
- *      the hold count is 1, pagebuf_rele calls pagebuf_free.
+ *      Releases a hold on the specified buffer.  If the
+ *      the hold count is 1, calls xfs_buf_free.
  */
 void
-pagebuf_rele(
-        xfs_buf_t               *pb)
+xfs_buf_rele(
+        xfs_buf_t               *bp)
 {
-        xfs_bufhash_t           *hash = pb->pb_hash;
+        xfs_bufhash_t           *hash = bp->b_hash;
 
-        PB_TRACE(pb, "rele", pb->pb_relse);
+        XB_TRACE(bp, "rele", bp->b_relse);
 
-        if (atomic_dec_and_lock(&pb->pb_hold, &hash->bh_lock)) {
-                if (pb->pb_relse) {
-                        atomic_inc(&pb->pb_hold);
+        if (atomic_dec_and_lock(&bp->b_hold, &hash->bh_lock)) {
+                if (bp->b_relse) {
+                        atomic_inc(&bp->b_hold);
                         spin_unlock(&hash->bh_lock);
-                        (*(pb->pb_relse)) (pb);
-                } else if (pb->pb_flags & PBF_FS_MANAGED) {
+                        (*(bp->b_relse)) (bp);
+                } else if (bp->b_flags & XBF_FS_MANAGED) {
                         spin_unlock(&hash->bh_lock);
                 } else {
-                        ASSERT(!(pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)));
-                        list_del_init(&pb->pb_hash_list);
+                        ASSERT(!(bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)));
+                        list_del_init(&bp->b_hash_list);
                         spin_unlock(&hash->bh_lock);
-                        pagebuf_free(pb);
+                        xfs_buf_free(bp);
                 }
         } else {
                 /*
                  * Catch reference count leaks
                  */
-                ASSERT(atomic_read(&pb->pb_hold) >= 0);
+                ASSERT(atomic_read(&bp->b_hold) >= 0);
         }
 }
 
@@ -863,168 +855,122 @@ pagebuf_rele(
  */
 
 /*
- *      pagebuf_cond_lock
- *
- *      pagebuf_cond_lock locks a buffer object, if it is not already locked.
- *      Note that this in no way
- *      locks the underlying pages, so it is only useful for synchronizing
- *      concurrent use of page buffer objects, not for synchronizing independent
- *      access to the underlying pages.
+ *      Locks a buffer object, if it is not already locked.
+ *      Note that this in no way locks the underlying pages, so it is only
+ *      useful for synchronizing concurrent use of buffer objects, not for
+ *      synchronizing independent access to the underlying pages.
  */
 int
-pagebuf_cond_lock(                      /* lock buffer, if not locked   */
-                                        /* returns -EBUSY if locked)    */
-        xfs_buf_t               *pb)
+xfs_buf_cond_lock(
+        xfs_buf_t               *bp)
 {
         int                     locked;
 
-        locked = down_trylock(&pb->pb_sema) == 0;
+        locked = down_trylock(&bp->b_sema) == 0;
         if (locked) {
-                PB_SET_OWNER(pb);
+                XB_SET_OWNER(bp);
         }
-        PB_TRACE(pb, "cond_lock", (long)locked);
-        return(locked ? 0 : -EBUSY);
+        XB_TRACE(bp, "cond_lock", (long)locked);
+        return locked ? 0 : -EBUSY;
 }
 
 #if defined(DEBUG) || defined(XFS_BLI_TRACE)
-/*
- *      pagebuf_lock_value
- *
- *      Return lock value for a pagebuf
- */
 int
-pagebuf_lock_value(
-        xfs_buf_t               *pb)
+xfs_buf_lock_value(
+        xfs_buf_t               *bp)
 {
-        return(atomic_read(&pb->pb_sema.count));
+        return atomic_read(&bp->b_sema.count);
 }
 #endif
 
 /*
- *      pagebuf_lock
- *
- *      pagebuf_lock locks a buffer object.  Note that this in no way
- *      locks the underlying pages, so it is only useful for synchronizing
- *      concurrent use of page buffer objects, not for synchronizing independent
- *      access to the underlying pages.
+ *      Locks a buffer object.
+ *      Note that this in no way locks the underlying pages, so it is only
+ *      useful for synchronizing concurrent use of buffer objects, not for
+ *      synchronizing independent access to the underlying pages.
  */
-int
-pagebuf_lock(
-        xfs_buf_t               *pb)
+void
+xfs_buf_lock(
+        xfs_buf_t               *bp)
 {
-        PB_TRACE(pb, "lock", 0);
-        if (atomic_read(&pb->pb_io_remaining))
-                blk_run_address_space(pb->pb_target->pbr_mapping);
-        down(&pb->pb_sema);
-        PB_SET_OWNER(pb);
-        PB_TRACE(pb, "locked", 0);
-        return 0;
+        XB_TRACE(bp, "lock", 0);
+        if (atomic_read(&bp->b_io_remaining))
+                blk_run_address_space(bp->b_target->bt_mapping);
+        down(&bp->b_sema);
+        XB_SET_OWNER(bp);
+        XB_TRACE(bp, "locked", 0);
 }
 
 /*
- *      pagebuf_unlock
- *
- *      pagebuf_unlock releases the lock on the buffer object created by
- *      pagebuf_lock or pagebuf_cond_lock (not any pinning of underlying pages
- *      created by pagebuf_pin).
- *
+ *      Releases the lock on the buffer object.
  *      If the buffer is marked delwri but is not queued, do so before we
- *      unlock the buffer as we need to set flags correctly. We also need to
+ *      unlock the buffer as we need to set flags correctly.  We also need to
  *      take a reference for the delwri queue because the unlocker is going to
  *      drop their's and they don't know we just queued it.
  */
 void
-pagebuf_unlock(                         /* unlock buffer                */
-        xfs_buf_t               *pb)    /* buffer to unlock             */
+xfs_buf_unlock(
+        xfs_buf_t               *bp)
 {
-        if ((pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q)) == PBF_DELWRI) {
-                atomic_inc(&pb->pb_hold);
-                pb->pb_flags |= PBF_ASYNC;
-                pagebuf_delwri_queue(pb, 0);
+        if ((bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)) == XBF_DELWRI) {
+                atomic_inc(&bp->b_hold);
+                bp->b_flags |= XBF_ASYNC;
+                xfs_buf_delwri_queue(bp, 0);
         }
 
-        PB_CLEAR_OWNER(pb);
-        up(&pb->pb_sema);
-        PB_TRACE(pb, "unlock", 0);
+        XB_CLEAR_OWNER(bp);
+        up(&bp->b_sema);
+        XB_TRACE(bp, "unlock", 0);
 }
 
 
 /*
  *      Pinning Buffer Storage in Memory
- */
-
-/*
- *      pagebuf_pin
- *
- *      pagebuf_pin locks all of the memory represented by a buffer in
- *      memory.  Multiple calls to pagebuf_pin and pagebuf_unpin, for
- *      the same or different buffers affecting a given page, will
- *      properly count the number of outstanding "pin" requests.  The
- *      buffer may be released after the pagebuf_pin and a different
- *      buffer used when calling pagebuf_unpin, if desired.
- *      pagebuf_pin should be used by the file system when it wants be
- *      assured that no attempt will be made to force the affected
- *      memory to disk.  It does not assure that a given logical page
- *      will not be moved to a different physical page.
+ *      Ensure that no attempt to force a buffer to disk will succeed.
  */
 void
-pagebuf_pin(
-        xfs_buf_t               *pb)
+xfs_buf_pin(
+        xfs_buf_t               *bp)
 {
-        atomic_inc(&pb->pb_pin_count);
-        PB_TRACE(pb, "pin", (long)pb->pb_pin_count.counter);
+        atomic_inc(&bp->b_pin_count);
+        XB_TRACE(bp, "pin", (long)bp->b_pin_count.counter);
 }
 
-/*
- *      pagebuf_unpin
- *
- *      pagebuf_unpin reverses the locking of memory performed by
- *      pagebuf_pin.  Note that both functions affected the logical
- *      pages associated with the buffer, not the buffer itself.
- */
 void
-pagebuf_unpin(
-        xfs_buf_t               *pb)
+xfs_buf_unpin(
+        xfs_buf_t               *bp)
 {
-        if (atomic_dec_and_test(&pb->pb_pin_count)) {
-                wake_up_all(&pb->pb_waiters);
-        }
-        PB_TRACE(pb, "unpin", (long)pb->pb_pin_count.counter);
+        if (atomic_dec_and_test(&bp->b_pin_count))
+                wake_up_all(&bp->b_waiters);
+        XB_TRACE(bp, "unpin", (long)bp->b_pin_count.counter);
 }
 
 int
-pagebuf_ispin(
-        xfs_buf_t               *pb)
+xfs_buf_ispin(
+        xfs_buf_t               *bp)
 {
-        return atomic_read(&pb->pb_pin_count);
+        return atomic_read(&bp->b_pin_count);
 }
 
-/*
- *      pagebuf_wait_unpin
- *
- *      pagebuf_wait_unpin waits until all of the memory associated
- *      with the buffer is not longer locked in memory.  It returns
- *      immediately if none of the affected pages are locked.
- */
-static inline void
-_pagebuf_wait_unpin(
-        xfs_buf_t               *pb)
+STATIC void
+xfs_buf_wait_unpin(
+        xfs_buf_t               *bp)
 {
         DECLARE_WAITQUEUE       (wait, current);
 
-        if (atomic_read(&pb->pb_pin_count) == 0)
+        if (atomic_read(&bp->b_pin_count) == 0)
                 return;
 
-        add_wait_queue(&pb->pb_waiters, &wait);
+        add_wait_queue(&bp->b_waiters, &wait);
         for (;;) {
                 set_current_state(TASK_UNINTERRUPTIBLE);
-                if (atomic_read(&pb->pb_pin_count) == 0)
+                if (atomic_read(&bp->b_pin_count) == 0)
                         break;
-                if (atomic_read(&pb->pb_io_remaining))
-                        blk_run_address_space(pb->pb_target->pbr_mapping);
+                if (atomic_read(&bp->b_io_remaining))
+                        blk_run_address_space(bp->b_target->bt_mapping);
                 schedule();
         }
-        remove_wait_queue(&pb->pb_waiters, &wait);
+        remove_wait_queue(&bp->b_waiters, &wait);
         set_current_state(TASK_RUNNING);
 }
 
@@ -1032,241 +978,216 @@ _pagebuf_wait_unpin(
  *      Buffer Utility Routines
  */
 
-/*
- *      pagebuf_iodone
- *
- *      pagebuf_iodone marks a buffer for which I/O is in progress
- *      done with respect to that I/O.  The pb_iodone routine, if
- *      present, will be called as a side-effect.
- */
 STATIC void
-pagebuf_iodone_work(
+xfs_buf_iodone_work(
         void                    *v)
 {
         xfs_buf_t               *bp = (xfs_buf_t *)v;
 
-        if (bp->pb_iodone)
-                (*(bp->pb_iodone))(bp);
-        else if (bp->pb_flags & PBF_ASYNC)
+        if (bp->b_iodone)
+                (*(bp->b_iodone))(bp);
+        else if (bp->b_flags & XBF_ASYNC)
                 xfs_buf_relse(bp);
 }
 
 void
-pagebuf_iodone(
-        xfs_buf_t               *pb,
+xfs_buf_ioend(
+        xfs_buf_t               *bp,
         int                     schedule)
 {
-        pb->pb_flags &= ~(PBF_READ | PBF_WRITE);
-        if (pb->pb_error == 0)
-                pb->pb_flags |= PBF_DONE;
+        bp->b_flags &= ~(XBF_READ | XBF_WRITE);
+        if (bp->b_error == 0)
+                bp->b_flags |= XBF_DONE;
 
-        PB_TRACE(pb, "iodone", pb->pb_iodone);
+        XB_TRACE(bp, "iodone", bp->b_iodone);
 
-        if ((pb->pb_iodone) || (pb->pb_flags & PBF_ASYNC)) {
+        if ((bp->b_iodone) || (bp->b_flags & XBF_ASYNC)) {
                 if (schedule) {
-                        INIT_WORK(&pb->pb_iodone_work, pagebuf_iodone_work, pb);
-                        queue_work(xfslogd_workqueue, &pb->pb_iodone_work);
+                        INIT_WORK(&bp->b_iodone_work, xfs_buf_iodone_work, bp);
+                        queue_work(xfslogd_workqueue, &bp->b_iodone_work);
                 } else {
-                        pagebuf_iodone_work(pb);
+                        xfs_buf_iodone_work(bp);
                 }
         } else {
-                up(&pb->pb_iodonesema);
+                up(&bp->b_iodonesema);
         }
 }
 
-/*
- *      pagebuf_ioerror
- *
- *      pagebuf_ioerror sets the error code for a buffer.
- */
 void
-pagebuf_ioerror(                        /* mark/clear buffer error flag */
-        xfs_buf_t               *pb,    /* buffer to mark               */
-        int                     error)  /* error to store (0 if none)   */
+xfs_buf_ioerror(
+        xfs_buf_t               *bp,
+        int                     error)
 {
         ASSERT(error >= 0 && error <= 0xffff);
-        pb->pb_error = (unsigned short)error;
-        PB_TRACE(pb, "ioerror", (unsigned long)error);
+        bp->b_error = (unsigned short)error;
+        XB_TRACE(bp, "ioerror", (unsigned long)error);
 }
 
 /*
- *      pagebuf_iostart
- *
- *      pagebuf_iostart initiates I/O on a buffer, based on the flags supplied.
- *      If necessary, it will arrange for any disk space allocation required,
- *      and it will break up the request if the block mappings require it.
- *      The pb_iodone routine in the buffer supplied will only be called
+ *      Initiate I/O on a buffer, based on the flags supplied.
+ *      The b_iodone routine in the buffer supplied will only be called
  *      when all of the subsidiary I/O requests, if any, have been completed.
- *      pagebuf_iostart calls the pagebuf_ioinitiate routine or
- *      pagebuf_iorequest, if the former routine is not defined, to start
- *      the I/O on a given low-level request.
  */
 int
-pagebuf_iostart(                        /* start I/O on a buffer          */
-        xfs_buf_t               *pb,    /* buffer to start                */
-        page_buf_flags_t        flags)  /* PBF_LOCK, PBF_ASYNC, PBF_READ, */
-                                        /* PBF_WRITE, PBF_DELWRI,         */
-                                        /* PBF_DONT_BLOCK                 */
+xfs_buf_iostart(
+        xfs_buf_t               *bp,
+        xfs_buf_flags_t         flags)
 {
         int                     status = 0;
 
-        PB_TRACE(pb, "iostart", (unsigned long)flags);
+        XB_TRACE(bp, "iostart", (unsigned long)flags);
 
-        if (flags & PBF_DELWRI) {
-                pb->pb_flags &= ~(PBF_READ | PBF_WRITE | PBF_ASYNC);
-                pb->pb_flags |= flags & (PBF_DELWRI | PBF_ASYNC);
-                pagebuf_delwri_queue(pb, 1);
+        if (flags & XBF_DELWRI) {
+                bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC);
+                bp->b_flags |= flags & (XBF_DELWRI | XBF_ASYNC);
+                xfs_buf_delwri_queue(bp, 1);
                 return status;
         }
 
-        pb->pb_flags &= ~(PBF_READ | PBF_WRITE | PBF_ASYNC | PBF_DELWRI | \
-                        PBF_READ_AHEAD | _PBF_RUN_QUEUES);
-        pb->pb_flags |= flags & (PBF_READ | PBF_WRITE | PBF_ASYNC | \
-                        PBF_READ_AHEAD | _PBF_RUN_QUEUES);
+        bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC | XBF_DELWRI | \
+                        XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+        bp->b_flags |= flags & (XBF_READ | XBF_WRITE | XBF_ASYNC | \
+                        XBF_READ_AHEAD | _XBF_RUN_QUEUES);
 
-        BUG_ON(pb->pb_bn == XFS_BUF_DADDR_NULL);
+        BUG_ON(bp->b_bn == XFS_BUF_DADDR_NULL);
 
         /* For writes allow an alternate strategy routine to precede
          * the actual I/O request (which may not be issued at all in
          * a shutdown situation, for example).
          */
-        status = (flags & PBF_WRITE) ?
-                pagebuf_iostrategy(pb) : pagebuf_iorequest(pb);
+        status = (flags & XBF_WRITE) ?
+                xfs_buf_iostrategy(bp) : xfs_buf_iorequest(bp);
 
         /* Wait for I/O if we are not an async request.
          * Note: async I/O request completion will release the buffer,
          * and that can already be done by this point.  So using the
          * buffer pointer from here on, after async I/O, is invalid.
          */
-        if (!status && !(flags & PBF_ASYNC))
-                status = pagebuf_iowait(pb);
+        if (!status && !(flags & XBF_ASYNC))
+                status = xfs_buf_iowait(bp);
 
         return status;
 }
 
-/*
- * Helper routine for pagebuf_iorequest
- */
-
 STATIC __inline__ int
-_pagebuf_iolocked(
-        xfs_buf_t               *pb)
+_xfs_buf_iolocked(
+        xfs_buf_t               *bp)
 {
-        ASSERT(pb->pb_flags & (PBF_READ|PBF_WRITE));
-        if (pb->pb_flags & PBF_READ)
-                return pb->pb_locked;
+        ASSERT(bp->b_flags & (XBF_READ | XBF_WRITE));
+        if (bp->b_flags & XBF_READ)
+                return bp->b_locked;
         return 0;
 }
 
 STATIC __inline__ void
-_pagebuf_iodone(
-        xfs_buf_t               *pb,
+_xfs_buf_ioend(
+        xfs_buf_t               *bp,
         int                     schedule)
 {
-        if (atomic_dec_and_test(&pb->pb_io_remaining) == 1) {
-                pb->pb_locked = 0;
-                pagebuf_iodone(pb, schedule);
+        if (atomic_dec_and_test(&bp->b_io_remaining) == 1) {
+                bp->b_locked = 0;
+                xfs_buf_ioend(bp, schedule);
         }
 }
 
 STATIC int
-bio_end_io_pagebuf(
+xfs_buf_bio_end_io(
         struct bio              *bio,
         unsigned int            bytes_done,
         int                     error)
 {
-        xfs_buf_t               *pb = (xfs_buf_t *)bio->bi_private;
-        unsigned int            blocksize = pb->pb_target->pbr_bsize;
+        xfs_buf_t               *bp = (xfs_buf_t *)bio->bi_private;
+        unsigned int            blocksize = bp->b_target->bt_bsize;
         struct bio_vec          *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
 
         if (bio->bi_size)
                 return 1;
 
         if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
-                pb->pb_error = EIO;
+                bp->b_error = EIO;
 
         do {
                 struct page     *page = bvec->bv_page;
 
-                if (unlikely(pb->pb_error)) {
-                        if (pb->pb_flags & PBF_READ)
+                if (unlikely(bp->b_error)) {
+                        if (bp->b_flags & XBF_READ)
                                 ClearPageUptodate(page);
                         SetPageError(page);
-                } else if (blocksize == PAGE_CACHE_SIZE) {
+                } else if (blocksize >= PAGE_CACHE_SIZE) {
                         SetPageUptodate(page);
                 } else if (!PagePrivate(page) &&
-                                (pb->pb_flags & _PBF_PAGE_CACHE)) {
+                                (bp->b_flags & _XBF_PAGE_CACHE)) {
                         set_page_region(page, bvec->bv_offset, bvec->bv_len);
                 }
 
                 if (--bvec >= bio->bi_io_vec)
                         prefetchw(&bvec->bv_page->flags);
 
-                if (_pagebuf_iolocked(pb)) {
+                if (_xfs_buf_iolocked(bp)) {
                         unlock_page(page);
                 }
         } while (bvec >= bio->bi_io_vec);
 
-        _pagebuf_iodone(pb, 1);
+        _xfs_buf_ioend(bp, 1);
         bio_put(bio);
         return 0;
 }
 
 STATIC void
-_pagebuf_ioapply(
-        xfs_buf_t               *pb)
+_xfs_buf_ioapply(
+        xfs_buf_t               *bp)
 {
         int                     i, rw, map_i, total_nr_pages, nr_pages;
         struct bio              *bio;
-        int                     offset = pb->pb_offset;
-        int                     size = pb->pb_count_desired;
-        sector_t                sector = pb->pb_bn;
-        unsigned int            blocksize = pb->pb_target->pbr_bsize;
-        int                     locking = _pagebuf_iolocked(pb);
+        int                     offset = bp->b_offset;
+        int                     size = bp->b_count_desired;
+        sector_t                sector = bp->b_bn;
+        unsigned int            blocksize = bp->b_target->bt_bsize;
+        int                     locking = _xfs_buf_iolocked(bp);
 
-        total_nr_pages = pb->pb_page_count;
+        total_nr_pages = bp->b_page_count;
         map_i = 0;
 
-        if (pb->pb_flags & _PBF_RUN_QUEUES) {
-                pb->pb_flags &= ~_PBF_RUN_QUEUES;
-                rw = (pb->pb_flags & PBF_READ) ? READ_SYNC : WRITE_SYNC;
+        if (bp->b_flags & _XBF_RUN_QUEUES) {
+                bp->b_flags &= ~_XBF_RUN_QUEUES;
+                rw = (bp->b_flags & XBF_READ) ? READ_SYNC : WRITE_SYNC;
         } else {
-                rw = (pb->pb_flags & PBF_READ) ? READ : WRITE;
+                rw = (bp->b_flags & XBF_READ) ? READ : WRITE;
         }
 
-        if (pb->pb_flags & PBF_ORDERED) {
-                ASSERT(!(pb->pb_flags & PBF_READ));
+        if (bp->b_flags & XBF_ORDERED) {
+                ASSERT(!(bp->b_flags & XBF_READ));
                 rw = WRITE_BARRIER;
         }
 
-        /* Special code path for reading a sub page size pagebuf in --
+        /* Special code path for reading a sub page size buffer in --
          * we populate up the whole page, and hence the other metadata
          * in the same page.  This optimization is only valid when the
-         * filesystem block size and the page size are equal.
+         * filesystem block size is not smaller than the page size.
          */
-        if ((pb->pb_buffer_length < PAGE_CACHE_SIZE) &&
-            (pb->pb_flags & PBF_READ) && locking &&
-            (blocksize == PAGE_CACHE_SIZE)) {
+        if ((bp->b_buffer_length < PAGE_CACHE_SIZE) &&
+            (bp->b_flags & XBF_READ) && locking &&
+            (blocksize >= PAGE_CACHE_SIZE)) {
                 bio = bio_alloc(GFP_NOIO, 1);
 
-                bio->bi_bdev = pb->pb_target->pbr_bdev;
+                bio->bi_bdev = bp->b_target->bt_bdev;
                 bio->bi_sector = sector - (offset >> BBSHIFT);
-                bio->bi_end_io = bio_end_io_pagebuf;
-                bio->bi_private = pb;
+                bio->bi_end_io = xfs_buf_bio_end_io;
+                bio->bi_private = bp;
 
-                bio_add_page(bio, pb->pb_pages[0], PAGE_CACHE_SIZE, 0);
+                bio_add_page(bio, bp->b_pages[0], PAGE_CACHE_SIZE, 0);
                 size = 0;
 
-                atomic_inc(&pb->pb_io_remaining);
+                atomic_inc(&bp->b_io_remaining);
 
                 goto submit_io;
         }
 
         /* Lock down the pages which we need to for the request */
-        if (locking && (pb->pb_flags & PBF_WRITE) && (pb->pb_locked == 0)) {
+        if (locking && (bp->b_flags & XBF_WRITE) && (bp->b_locked == 0)) {
                 for (i = 0; size; i++) {
                         int             nbytes = PAGE_CACHE_SIZE - offset;
-                        struct page     *page = pb->pb_pages[i];
+                        struct page     *page = bp->b_pages[i];
 
                         if (nbytes > size)
                                 nbytes = size;
@@ -1276,30 +1197,30 @@ _pagebuf_ioapply(
                         size -= nbytes;
                         offset = 0;
                 }
-                offset = pb->pb_offset;
-                size = pb->pb_count_desired;
+                offset = bp->b_offset;
+                size = bp->b_count_desired;
         }
 
 next_chunk:
-        atomic_inc(&pb->pb_io_remaining);
+        atomic_inc(&bp->b_io_remaining);
         nr_pages = BIO_MAX_SECTORS >> (PAGE_SHIFT - BBSHIFT);
         if (nr_pages > total_nr_pages)
                 nr_pages = total_nr_pages;
 
         bio = bio_alloc(GFP_NOIO, nr_pages);
-        bio->bi_bdev = pb->pb_target->pbr_bdev;
+        bio->bi_bdev = bp->b_target->bt_bdev;
         bio->bi_sector = sector;
-        bio->bi_end_io = bio_end_io_pagebuf;
-        bio->bi_private = pb;
+        bio->bi_end_io = xfs_buf_bio_end_io;
+        bio->bi_private = bp;
 
         for (; size && nr_pages; nr_pages--, map_i++) {
-                int     nbytes = PAGE_CACHE_SIZE - offset;
+                int     rbytes, nbytes = PAGE_CACHE_SIZE - offset;
 
                 if (nbytes > size)
                         nbytes = size;
 
-                if (bio_add_page(bio, pb->pb_pages[map_i],
-                                        nbytes, offset) < nbytes)
+                rbytes = bio_add_page(bio, bp->b_pages[map_i], nbytes, offset);
+                if (rbytes < nbytes)
                         break;
 
                 offset = 0;
@@ -1315,107 +1236,102 @@ submit_io:
                         goto next_chunk;
         } else {
                 bio_put(bio);
-                pagebuf_ioerror(pb, EIO);
+                xfs_buf_ioerror(bp, EIO);
         }
 }
 
-/*
- *      pagebuf_iorequest -- the core I/O request routine.
- */
 int
-pagebuf_iorequest(                      /* start real I/O               */
-        xfs_buf_t               *pb)    /* buffer to convey to device   */
+xfs_buf_iorequest(
+        xfs_buf_t               *bp)
 {
-        PB_TRACE(pb, "iorequest", 0);
+        XB_TRACE(bp, "iorequest", 0);
 
-        if (pb->pb_flags & PBF_DELWRI) {
-                pagebuf_delwri_queue(pb, 1);
+        if (bp->b_flags & XBF_DELWRI) {
+                xfs_buf_delwri_queue(bp, 1);
                 return 0;
         }
 
-        if (pb->pb_flags & PBF_WRITE) {
-                _pagebuf_wait_unpin(pb);
+        if (bp->b_flags & XBF_WRITE) {
+                xfs_buf_wait_unpin(bp);
         }
 
-        pagebuf_hold(pb);
+        xfs_buf_hold(bp);
 
         /* Set the count to 1 initially, this will stop an I/O
          * completion callout which happens before we have started
-         * all the I/O from calling pagebuf_iodone too early.
+         * all the I/O from calling xfs_buf_ioend too early.
          */
-        atomic_set(&pb->pb_io_remaining, 1);
-        _pagebuf_ioapply(pb);
-        _pagebuf_iodone(pb, 0);
+        atomic_set(&bp->b_io_remaining, 1);
+        _xfs_buf_ioapply(bp);
+        _xfs_buf_ioend(bp, 0);
 
-        pagebuf_rele(pb);
+        xfs_buf_rele(bp);
         return 0;
 }
 
 /*
- *      pagebuf_iowait
- *
- *      pagebuf_iowait waits for I/O to complete on the buffer supplied.
- *      It returns immediately if no I/O is pending.  In any case, it returns
- *      the error code, if any, or 0 if there is no error.
+ *      Waits for I/O to complete on the buffer supplied.
+ *      It returns immediately if no I/O is pending.
+ *      It returns the I/O error code, if any, or 0 if there was no error.
  */
 int
-pagebuf_iowait(
-        xfs_buf_t               *pb)
+xfs_buf_iowait(
+        xfs_buf_t               *bp)
 {
-        PB_TRACE(pb, "iowait", 0);
-        if (atomic_read(&pb->pb_io_remaining))
-                blk_run_address_space(pb->pb_target->pbr_mapping);
-        down(&pb->pb_iodonesema);
-        PB_TRACE(pb, "iowaited", (long)pb->pb_error);
-        return pb->pb_error;
+        XB_TRACE(bp, "iowait", 0);
+        if (atomic_read(&bp->b_io_remaining))
+                blk_run_address_space(bp->b_target->bt_mapping);
+        down(&bp->b_iodonesema);
+        XB_TRACE(bp, "iowaited", (long)bp->b_error);
+        return bp->b_error;
 }
 
-caddr_t
-pagebuf_offset(
-        xfs_buf_t               *pb,
+xfs_caddr_t
+xfs_buf_offset(
+        xfs_buf_t               *bp,
         size_t                  offset)
 {
         struct page             *page;
 
-        offset += pb->pb_offset;
+        if (bp->b_flags & XBF_MAPPED)
+                return XFS_BUF_PTR(bp) + offset;
 
-        page = pb->pb_pages[offset >> PAGE_CACHE_SHIFT];
-        return (caddr_t) page_address(page) + (offset & (PAGE_CACHE_SIZE - 1));
+        offset += bp->b_offset;
+        page = bp->b_pages[offset >> PAGE_CACHE_SHIFT];
+        return (xfs_caddr_t)page_address(page) + (offset & (PAGE_CACHE_SIZE-1));
 }
 
 /*
- *      pagebuf_iomove
- *
  *      Move data into or out of a buffer.
  */
 void
-pagebuf_iomove(
-        xfs_buf_t               *pb,    /* buffer to process            */
+xfs_buf_iomove(
+        xfs_buf_t               *bp,    /* buffer to process            */
         size_t                  boff,   /* starting buffer offset       */
         size_t                  bsize,  /* length to copy               */
         caddr_t                 data,   /* data address                 */
-        page_buf_rw_t           mode)   /* read/write flag              */
+        xfs_buf_rw_t            mode)   /* read/write/zero flag         */
 {
         size_t                  bend, cpoff, csize;
         struct page             *page;
 
         bend = boff + bsize;
         while (boff < bend) {
-                page = pb->pb_pages[page_buf_btoct(boff + pb->pb_offset)];
-                cpoff = page_buf_poff(boff + pb->pb_offset);
+                page = bp->b_pages[xfs_buf_btoct(boff + bp->b_offset)];
+                cpoff = xfs_buf_poff(boff + bp->b_offset);
                 csize = min_t(size_t,
-                              PAGE_CACHE_SIZE-cpoff, pb->pb_count_desired-boff);
+                              PAGE_CACHE_SIZE-cpoff, bp->b_count_desired-boff);
 
                 ASSERT(((csize + cpoff) <= PAGE_CACHE_SIZE));
 
                 switch (mode) {
-                case PBRW_ZERO:
+                case XBRW_ZERO:
                         memset(page_address(page) + cpoff, 0, csize);
                         break;
-                case PBRW_READ:
+                case XBRW_READ:
                         memcpy(data, page_address(page) + cpoff, csize);
                         break;
-                case PBRW_WRITE:
+                case XBRW_WRITE:
                         memcpy(page_address(page) + cpoff, data, csize);
                 }
 
@@ -1425,12 +1341,12 @@ pagebuf_iomove(
 }
 
 /*
- *      Handling of buftargs.
+ *      Handling of buffer targets (buftargs).
  */
 
 /*
- * Wait for any bufs with callbacks that have been submitted but
- * have not yet returned... walk the hash list for the target.
+ *      Wait for any bufs with callbacks that have been submitted but
+ *      have not yet returned... walk the hash list for the target.
  */
 void
 xfs_wait_buftarg(
@@ -1444,15 +1360,15 @@ xfs_wait_buftarg(
                 hash = &btp->bt_hash[i];
 again:
                 spin_lock(&hash->bh_lock);
-                list_for_each_entry_safe(bp, n, &hash->bh_list, pb_hash_list) {
-                        ASSERT(btp == bp->pb_target);
-                        if (!(bp->pb_flags & PBF_FS_MANAGED)) {
+                list_for_each_entry_safe(bp, n, &hash->bh_list, b_hash_list) {
+                        ASSERT(btp == bp->b_target);
+                        if (!(bp->b_flags & XBF_FS_MANAGED)) {
                                 spin_unlock(&hash->bh_lock);
                                 /*
                                  * Catch superblock reference count leaks
                                  * immediately
                                  */
-                                BUG_ON(bp->pb_bn == 0);
+                                BUG_ON(bp->b_bn == 0);
                                 delay(100);
                                 goto again;
                         }
@@ -1462,9 +1378,9 @@ again:
 }
 
 /*
- * Allocate buffer hash table for a given target.
- * For devices containing metadata (i.e. not the log/realtime devices)
- * we need to allocate a much larger hash table.
+ *      Allocate buffer hash table for a given target.
+ *      For devices containing metadata (i.e. not the log/realtime devices)
+ *      we need to allocate a much larger hash table.
  */
 STATIC void
 xfs_alloc_bufhash(
@@ -1487,11 +1403,34 @@ STATIC void
 xfs_free_bufhash(
         xfs_buftarg_t           *btp)
 {
-        kmem_free(btp->bt_hash,
-                        (1 << btp->bt_hashshift) * sizeof(xfs_bufhash_t));
+        kmem_free(btp->bt_hash, (1<<btp->bt_hashshift) * sizeof(xfs_bufhash_t));
         btp->bt_hash = NULL;
 }
 
+/*
+ *      buftarg list for delwrite queue processing
+ */
+STATIC LIST_HEAD(xfs_buftarg_list);
+STATIC DEFINE_SPINLOCK(xfs_buftarg_lock);
+
+STATIC void
+xfs_register_buftarg(
+        xfs_buftarg_t           *btp)
+{
+        spin_lock(&xfs_buftarg_lock);
+        list_add(&btp->bt_list, &xfs_buftarg_list);
+        spin_unlock(&xfs_buftarg_lock);
+}
+
+STATIC void
+xfs_unregister_buftarg(
+        xfs_buftarg_t           *btp)
+{
+        spin_lock(&xfs_buftarg_lock);
+        list_del(&btp->bt_list);
+        spin_unlock(&xfs_buftarg_lock);
+}
+
 void
 xfs_free_buftarg(
         xfs_buftarg_t           *btp,
@@ -1499,9 +1438,16 @@ xfs_free_buftarg(
 {
         xfs_flush_buftarg(btp, 1);
         if (external)
-                xfs_blkdev_put(btp->pbr_bdev);
+                xfs_blkdev_put(btp->bt_bdev);
         xfs_free_bufhash(btp);
-        iput(btp->pbr_mapping->host);
+        iput(btp->bt_mapping->host);
+
+        /* Unregister the buftarg first so that we don't get a
+         * wakeup finding a non-existent task
+         */
+        xfs_unregister_buftarg(btp);
+        kthread_stop(btp->bt_task);
+
         kmem_free(btp, sizeof(*btp));
 }
 
@@ -1512,11 +1458,11 @@ xfs_setsize_buftarg_flags(
         unsigned int            sectorsize,
         int                     verbose)
 {
-        btp->pbr_bsize = blocksize;
-        btp->pbr_sshift = ffs(sectorsize) - 1;
-        btp->pbr_smask = sectorsize - 1;
+        btp->bt_bsize = blocksize;
+        btp->bt_sshift = ffs(sectorsize) - 1;
+        btp->bt_smask = sectorsize - 1;
 
-        if (set_blocksize(btp->pbr_bdev, sectorsize)) {
+        if (set_blocksize(btp->bt_bdev, sectorsize)) {
                 printk(KERN_WARNING
                         "XFS: Cannot set_blocksize to %u on device %s\n",
                         sectorsize, XFS_BUFTARG_NAME(btp));
@@ -1536,10 +1482,10 @@ xfs_setsize_buftarg_flags(
 }
 
 /*
-* When allocating the initial buffer target we have not yet
-* read in the superblock, so don't know what sized sectors
-* are being used is at this early stage.  Play safe.
-*/
+ *      When allocating the initial buffer target we have not yet
+ *      read in the superblock, so don't know what sized sectors
+ *      are being used is at this early stage.  Play safe.
+ */
 STATIC int
 xfs_setsize_buftarg_early(
         xfs_buftarg_t           *btp,
@@ -1587,10 +1533,30 @@ xfs_mapping_buftarg(
         mapping->a_ops = &mapping_aops;
         mapping->backing_dev_info = bdi;
         mapping_set_gfp_mask(mapping, GFP_NOFS);
-        btp->pbr_mapping = mapping;
+        btp->bt_mapping = mapping;
         return 0;
 }
 
+STATIC int
+xfs_alloc_delwrite_queue(
+        xfs_buftarg_t           *btp)
+{
+        int     error = 0;
+
+        INIT_LIST_HEAD(&btp->bt_list);
+        INIT_LIST_HEAD(&btp->bt_delwrite_queue);
+        spinlock_init(&btp->bt_delwrite_lock, "delwri_lock");
+        btp->bt_flags = 0;
+        btp->bt_task = kthread_run(xfsbufd, btp, "xfsbufd");
+        if (IS_ERR(btp->bt_task)) {
+                error = PTR_ERR(btp->bt_task);
+                goto out_error;
+        }
+        xfs_register_buftarg(btp);
+out_error:
+        return error;
+}
+
 xfs_buftarg_t *
 xfs_alloc_buftarg(
         struct block_device     *bdev,
@@ -1600,12 +1566,14 @@ xfs_alloc_buftarg(
 
         btp = kmem_zalloc(sizeof(*btp), KM_SLEEP);
 
-        btp->pbr_dev =  bdev->bd_dev;
-        btp->pbr_bdev = bdev;
+        btp->bt_dev =  bdev->bd_dev;
+        btp->bt_bdev = bdev;
         if (xfs_setsize_buftarg_early(btp, bdev))
                 goto error;
         if (xfs_mapping_buftarg(btp, bdev))
                 goto error;
+        if (xfs_alloc_delwrite_queue(btp))
+                goto error;
         xfs_alloc_bufhash(btp, external);
         return btp;
 
@@ -1616,83 +1584,81 @@ error:
 
 
 /*
- * Pagebuf delayed write buffer handling
+ *      Delayed write buffer handling
  */
-
-STATIC LIST_HEAD(pbd_delwrite_queue);
-STATIC DEFINE_SPINLOCK(pbd_delwrite_lock);
-
 STATIC void
-pagebuf_delwri_queue(
-        xfs_buf_t               *pb,
+xfs_buf_delwri_queue(
+        xfs_buf_t               *bp,
         int                     unlock)
 {
-        PB_TRACE(pb, "delwri_q", (long)unlock);
-        ASSERT((pb->pb_flags & (PBF_DELWRI|PBF_ASYNC)) ==
-                                        (PBF_DELWRI|PBF_ASYNC));
+        struct list_head        *dwq = &bp->b_target->bt_delwrite_queue;
+        spinlock_t              *dwlk = &bp->b_target->bt_delwrite_lock;
+
+        XB_TRACE(bp, "delwri_q", (long)unlock);
+        ASSERT((bp->b_flags&(XBF_DELWRI|XBF_ASYNC)) == (XBF_DELWRI|XBF_ASYNC));
 
-        spin_lock(&pbd_delwrite_lock);
+        spin_lock(dwlk);
         /* If already in the queue, dequeue and place at tail */
-        if (!list_empty(&pb->pb_list)) {
-                ASSERT(pb->pb_flags & _PBF_DELWRI_Q);
-                if (unlock) {
-                        atomic_dec(&pb->pb_hold);
-                }
-                list_del(&pb->pb_list);
+        if (!list_empty(&bp->b_list)) {
+                ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+                if (unlock)
+                        atomic_dec(&bp->b_hold);
+                list_del(&bp->b_list);
         }
 
-        pb->pb_flags |= _PBF_DELWRI_Q;
-        list_add_tail(&pb->pb_list, &pbd_delwrite_queue);
-        pb->pb_queuetime = jiffies;
-        spin_unlock(&pbd_delwrite_lock);
+        bp->b_flags |= _XBF_DELWRI_Q;
+        list_add_tail(&bp->b_list, dwq);
+        bp->b_queuetime = jiffies;
+        spin_unlock(dwlk);
 
         if (unlock)
-                pagebuf_unlock(pb);
+                xfs_buf_unlock(bp);
 }
 
 void
-pagebuf_delwri_dequeue(
-        xfs_buf_t               *pb)
+xfs_buf_delwri_dequeue(
+        xfs_buf_t               *bp)
 {
+        spinlock_t              *dwlk = &bp->b_target->bt_delwrite_lock;
         int                     dequeued = 0;
 
-        spin_lock(&pbd_delwrite_lock);
-        if ((pb->pb_flags & PBF_DELWRI) && !list_empty(&pb->pb_list)) {
-                ASSERT(pb->pb_flags & _PBF_DELWRI_Q);
-                list_del_init(&pb->pb_list);
+        spin_lock(dwlk);
+        if ((bp->b_flags & XBF_DELWRI) && !list_empty(&bp->b_list)) {
+                ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+                list_del_init(&bp->b_list);
                 dequeued = 1;
         }
-        pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
-        spin_unlock(&pbd_delwrite_lock);
+        bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+        spin_unlock(dwlk);
 
         if (dequeued)
-                pagebuf_rele(pb);
+                xfs_buf_rele(bp);
 
-        PB_TRACE(pb, "delwri_dq", (long)dequeued);
+        XB_TRACE(bp, "delwri_dq", (long)dequeued);
 }
 
 STATIC void
-pagebuf_runall_queues(
+xfs_buf_runall_queues(
         struct workqueue_struct *queue)
 {
         flush_workqueue(queue);
 }
 
-/* Defines for pagebuf daemon */
-STATIC struct task_struct *xfsbufd_task;
-STATIC int xfsbufd_force_flush;
-STATIC int xfsbufd_force_sleep;
-
 STATIC int
 xfsbufd_wakeup(
         int                     priority,
         gfp_t                   mask)
 {
-        if (xfsbufd_force_sleep)
-                return 0;
-        xfsbufd_force_flush = 1;
-        barrier();
-        wake_up_process(xfsbufd_task);
+        xfs_buftarg_t           *btp;
+
+        spin_lock(&xfs_buftarg_lock);
+        list_for_each_entry(btp, &xfs_buftarg_list, bt_list) {
+                if (test_bit(XBT_FORCE_SLEEP, &btp->bt_flags))
+                        continue;
+                set_bit(XBT_FORCE_FLUSH, &btp->bt_flags);
+                wake_up_process(btp->bt_task);
+        }
+        spin_unlock(&xfs_buftarg_lock);
         return 0;
 }
 
@@ -1702,67 +1668,70 @@ xfsbufd(
 {
         struct list_head        tmp;
         unsigned long           age;
-        xfs_buftarg_t           *target;
-        xfs_buf_t               *pb, *n;
+        xfs_buftarg_t           *target = (xfs_buftarg_t *)data;
+        xfs_buf_t               *bp, *n;
+        struct list_head        *dwq = &target->bt_delwrite_queue;
+        spinlock_t              *dwlk = &target->bt_delwrite_lock;
 
         current->flags |= PF_MEMALLOC;
 
         INIT_LIST_HEAD(&tmp);
         do {
                 if (unlikely(freezing(current))) {
-                        xfsbufd_force_sleep = 1;
+                        set_bit(XBT_FORCE_SLEEP, &target->bt_flags);
                         refrigerator();
                 } else {
-                        xfsbufd_force_sleep = 0;
+                        clear_bit(XBT_FORCE_SLEEP, &target->bt_flags);
                 }
 
                 schedule_timeout_interruptible(
                         xfs_buf_timer_centisecs * msecs_to_jiffies(10));
 
                 age = xfs_buf_age_centisecs * msecs_to_jiffies(10);
-                spin_lock(&pbd_delwrite_lock);
-                list_for_each_entry_safe(pb, n, &pbd_delwrite_queue, pb_list) {
-                        PB_TRACE(pb, "walkq1", (long)pagebuf_ispin(pb));
-                        ASSERT(pb->pb_flags & PBF_DELWRI);
-
-                        if (!pagebuf_ispin(pb) && !pagebuf_cond_lock(pb)) {
-                                if (!xfsbufd_force_flush &&
+                spin_lock(dwlk);
+                list_for_each_entry_safe(bp, n, dwq, b_list) {
+                        XB_TRACE(bp, "walkq1", (long)xfs_buf_ispin(bp));
+                        ASSERT(bp->b_flags & XBF_DELWRI);
+
+                        if (!xfs_buf_ispin(bp) && !xfs_buf_cond_lock(bp)) {
+                                if (!test_bit(XBT_FORCE_FLUSH,
+                                                &target->bt_flags) &&
                                     time_before(jiffies,
-                                                pb->pb_queuetime + age)) {
-                                        pagebuf_unlock(pb);
+                                                bp->b_queuetime + age)) {
+                                        xfs_buf_unlock(bp);
                                         break;
                                 }
 
-                                pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
-                                pb->pb_flags |= PBF_WRITE;
-                                list_move(&pb->pb_list, &tmp);
+                                bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+                                bp->b_flags |= XBF_WRITE;
+                                list_move(&bp->b_list, &tmp);
                         }
                 }
-                spin_unlock(&pbd_delwrite_lock);
+                spin_unlock(dwlk);
 
                 while (!list_empty(&tmp)) {
-                        pb = list_entry(tmp.next, xfs_buf_t, pb_list);
-                        target = pb->pb_target;
+                        bp = list_entry(tmp.next, xfs_buf_t, b_list);
+                        ASSERT(target == bp->b_target);
 
-                        list_del_init(&pb->pb_list);
-                        pagebuf_iostrategy(pb);
+                        list_del_init(&bp->b_list);
+                        xfs_buf_iostrategy(bp);
 
-                        blk_run_address_space(target->pbr_mapping);
+                        blk_run_address_space(target->bt_mapping);
                 }
 
                 if (as_list_len > 0)
                         purge_addresses();
 
-                xfsbufd_force_flush = 0;
+                clear_bit(XBT_FORCE_FLUSH, &target->bt_flags);
         } while (!kthread_should_stop());
 
         return 0;
 }
 
 /*
- * Go through all incore buffers, and release buffers if they belong to
- * the given device. This is used in filesystem error handling to
- * preserve the consistency of its metadata.
+ *      Go through all incore buffers, and release buffers if they belong to
+ *      the given device. This is used in filesystem error handling to
+ *      preserve the consistency of its metadata.
  */
 int
 xfs_flush_buftarg(
@@ -1770,73 +1739,72 @@ xfs_flush_buftarg(
         int                     wait)
 {
         struct list_head        tmp;
-        xfs_buf_t               *pb, *n;
+        xfs_buf_t               *bp, *n;
         int                     pincount = 0;
+        struct list_head        *dwq = &target->bt_delwrite_queue;
+        spinlock_t              *dwlk = &target->bt_delwrite_lock;
 
-        pagebuf_runall_queues(xfsdatad_workqueue);
-        pagebuf_runall_queues(xfslogd_workqueue);
+        xfs_buf_runall_queues(xfsdatad_workqueue);
+        xfs_buf_runall_queues(xfslogd_workqueue);
 
         INIT_LIST_HEAD(&tmp);
-        spin_lock(&pbd_delwrite_lock);
-        list_for_each_entry_safe(pb, n, &pbd_delwrite_queue, pb_list) {
-
-                if (pb->pb_target != target)
-                        continue;
-
-                ASSERT(pb->pb_flags & (PBF_DELWRI|_PBF_DELWRI_Q));
-                PB_TRACE(pb, "walkq2", (long)pagebuf_ispin(pb));
-                if (pagebuf_ispin(pb)) {
+        spin_lock(dwlk);
+        list_for_each_entry_safe(bp, n, dwq, b_list) {
+                ASSERT(bp->b_target == target);
+                ASSERT(bp->b_flags & (XBF_DELWRI | _XBF_DELWRI_Q));
+                XB_TRACE(bp, "walkq2", (long)xfs_buf_ispin(bp));
+                if (xfs_buf_ispin(bp)) {
                         pincount++;
                         continue;
                 }
 
-                list_move(&pb->pb_list, &tmp);
+                list_move(&bp->b_list, &tmp);
         }
-        spin_unlock(&pbd_delwrite_lock);
+        spin_unlock(dwlk);
 
         /*
          * Dropped the delayed write list lock, now walk the temporary list
          */
-        list_for_each_entry_safe(pb, n, &tmp, pb_list) {
-                pagebuf_lock(pb);
-                pb->pb_flags &= ~(PBF_DELWRI|_PBF_DELWRI_Q);
-                pb->pb_flags |= PBF_WRITE;
+        list_for_each_entry_safe(bp, n, &tmp, b_list) {
+                xfs_buf_lock(bp);
+                bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+                bp->b_flags |= XBF_WRITE;
                 if (wait)
-                        pb->pb_flags &= ~PBF_ASYNC;
+                        bp->b_flags &= ~XBF_ASYNC;
                 else
-                        list_del_init(&pb->pb_list);
+                        list_del_init(&bp->b_list);
 
-                pagebuf_iostrategy(pb);
+                xfs_buf_iostrategy(bp);
         }
 
         /*
          * Remaining list items must be flushed before returning
          */
         while (!list_empty(&tmp)) {
-                pb = list_entry(tmp.next, xfs_buf_t, pb_list);
+                bp = list_entry(tmp.next, xfs_buf_t, b_list);
 
-                list_del_init(&pb->pb_list);
-                xfs_iowait(pb);
-                xfs_buf_relse(pb);
+                list_del_init(&bp->b_list);
+                xfs_iowait(bp);
+                xfs_buf_relse(bp);
         }
 
         if (wait)
-                blk_run_address_space(target->pbr_mapping);
+                blk_run_address_space(target->bt_mapping);
 
         return pincount;
 }
 
 int __init
-pagebuf_init(void)
+xfs_buf_init(void)
 {
         int             error = -ENOMEM;
 
-#ifdef PAGEBUF_TRACE
-        pagebuf_trace_buf = ktrace_alloc(PAGEBUF_TRACE_SIZE, KM_SLEEP);
+#ifdef XFS_BUF_TRACE
+        xfs_buf_trace_buf = ktrace_alloc(XFS_BUF_TRACE_SIZE, KM_SLEEP);
 #endif
 
-        pagebuf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
-        if (!pagebuf_zone)
+        xfs_buf_zone = kmem_zone_init(sizeof(xfs_buf_t), "xfs_buf");
+        if (!xfs_buf_zone)
                 goto out_free_trace_buf;
 
         xfslogd_workqueue = create_workqueue("xfslogd");
@@ -1847,42 +1815,33 @@ pagebuf_init(void)
         if (!xfsdatad_workqueue)
                 goto out_destroy_xfslogd_workqueue;
 
-        xfsbufd_task = kthread_run(xfsbufd, NULL, "xfsbufd");
-        if (IS_ERR(xfsbufd_task)) {
-                error = PTR_ERR(xfsbufd_task);
+        xfs_buf_shake = kmem_shake_register(xfsbufd_wakeup);
+        if (!xfs_buf_shake)
                 goto out_destroy_xfsdatad_workqueue;
-        }
-
-        pagebuf_shake = kmem_shake_register(xfsbufd_wakeup);
-        if (!pagebuf_shake)
-                goto out_stop_xfsbufd;
 
         return 0;
 
- out_stop_xfsbufd:
-        kthread_stop(xfsbufd_task);
  out_destroy_xfsdatad_workqueue:
         destroy_workqueue(xfsdatad_workqueue);
  out_destroy_xfslogd_workqueue:
         destroy_workqueue(xfslogd_workqueue);
  out_free_buf_zone:
-        kmem_zone_destroy(pagebuf_zone);
+        kmem_zone_destroy(xfs_buf_zone);
  out_free_trace_buf:
-#ifdef PAGEBUF_TRACE
-        ktrace_free(pagebuf_trace_buf);
+#ifdef XFS_BUF_TRACE
+        ktrace_free(xfs_buf_trace_buf);
 #endif
         return error;
 }
 
 void
-pagebuf_terminate(void)
+xfs_buf_terminate(void)
 {
-        kmem_shake_deregister(pagebuf_shake);
-        kthread_stop(xfsbufd_task);
+        kmem_shake_deregister(xfs_buf_shake);
         destroy_workqueue(xfsdatad_workqueue);
         destroy_workqueue(xfslogd_workqueue);
-        kmem_zone_destroy(pagebuf_zone);
-#ifdef PAGEBUF_TRACE
-        ktrace_free(pagebuf_trace_buf);
+        kmem_zone_destroy(xfs_buf_zone);
+#ifdef XFS_BUF_TRACE
+        ktrace_free(xfs_buf_trace_buf);
 #endif
 }
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index 237a35b915d1..4dd6592d5a4c 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -32,44 +32,47 @@
  *      Base types
  */
 
-#define XFS_BUF_DADDR_NULL ((xfs_daddr_t) (-1LL))
-
-#define page_buf_ctob(pp)       ((pp) * PAGE_CACHE_SIZE)
-#define page_buf_btoc(dd)       (((dd) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT)
-#define page_buf_btoct(dd)      ((dd) >> PAGE_CACHE_SHIFT)
-#define page_buf_poff(aa)       ((aa) & ~PAGE_CACHE_MASK)
-
-typedef enum page_buf_rw_e {
-        PBRW_READ = 1,                  /* transfer into target memory */
-        PBRW_WRITE = 2,                 /* transfer from target memory */
-        PBRW_ZERO = 3                   /* Zero target memory */
-} page_buf_rw_t;
-
-
-typedef enum page_buf_flags_e {         /* pb_flags values */
-        PBF_READ = (1 << 0),    /* buffer intended for reading from device */
-        PBF_WRITE = (1 << 1),   /* buffer intended for writing to device   */
-        PBF_MAPPED = (1 << 2),  /* buffer mapped (pb_addr valid)           */
-        PBF_ASYNC = (1 << 4),   /* initiator will not wait for completion  */
-        PBF_DONE = (1 << 5),    /* all pages in the buffer uptodate        */
-        PBF_DELWRI = (1 << 6),  /* buffer has dirty pages                  */
-        PBF_STALE = (1 << 7),   /* buffer has been staled, do not find it  */
-        PBF_FS_MANAGED = (1 << 8),  /* filesystem controls freeing memory  */
-        PBF_ORDERED = (1 << 11),    /* use ordered writes                  */
-        PBF_READ_AHEAD = (1 << 12), /* asynchronous read-ahead             */
+#define XFS_BUF_DADDR_NULL      ((xfs_daddr_t) (-1LL))
+
+#define xfs_buf_ctob(pp)        ((pp) * PAGE_CACHE_SIZE)
+#define xfs_buf_btoc(dd)        (((dd) + PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_btoct(dd)       ((dd) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_poff(aa)        ((aa) & ~PAGE_CACHE_MASK)
+
+typedef enum {
+        XBRW_READ = 1,                  /* transfer into target memory */
+        XBRW_WRITE = 2,                 /* transfer from target memory */
+        XBRW_ZERO = 3,                  /* Zero target memory */
+} xfs_buf_rw_t;
+
+typedef enum {
+        XBF_READ = (1 << 0),    /* buffer intended for reading from device */
+        XBF_WRITE = (1 << 1),   /* buffer intended for writing to device   */
+        XBF_MAPPED = (1 << 2),  /* buffer mapped (b_addr valid)            */
+        XBF_ASYNC = (1 << 4),   /* initiator will not wait for completion  */
+        XBF_DONE = (1 << 5),    /* all pages in the buffer uptodate        */
+        XBF_DELWRI = (1 << 6),  /* buffer has dirty pages                  */
+        XBF_STALE = (1 << 7),   /* buffer has been staled, do not find it  */
+        XBF_FS_MANAGED = (1 << 8),  /* filesystem controls freeing memory  */
+        XBF_ORDERED = (1 << 11),    /* use ordered writes                  */
+        XBF_READ_AHEAD = (1 << 12), /* asynchronous read-ahead             */
 
         /* flags used only as arguments to access routines */
-        PBF_LOCK = (1 << 14),       /* lock requested                      */
-        PBF_TRYLOCK = (1 << 15),    /* lock requested, but do not wait     */
-        PBF_DONT_BLOCK = (1 << 16), /* do not block in current thread      */
+        XBF_LOCK = (1 << 14),       /* lock requested                      */
+        XBF_TRYLOCK = (1 << 15),    /* lock requested, but do not wait     */
+        XBF_DONT_BLOCK = (1 << 16), /* do not block in current thread      */
 
         /* flags used only internally */
-        _PBF_PAGE_CACHE = (1 << 17),/* backed by pagecache                 */
-        _PBF_KMEM_ALLOC = (1 << 18),/* backed by kmem_alloc()              */
-        _PBF_RUN_QUEUES = (1 << 19),/* run block device task queue         */
-        _PBF_DELWRI_Q = (1 << 21),   /* buffer on delwri queue             */
-} page_buf_flags_t;
+        _XBF_PAGE_CACHE = (1 << 17),/* backed by pagecache                 */
+        _XBF_KMEM_ALLOC = (1 << 18),/* backed by kmem_alloc()              */
+        _XBF_RUN_QUEUES = (1 << 19),/* run block device task queue         */
+        _XBF_DELWRI_Q = (1 << 21),   /* buffer on delwri queue             */
+} xfs_buf_flags_t;
 
+typedef enum {
+        XBT_FORCE_SLEEP = (0 << 1),
+        XBT_FORCE_FLUSH = (1 << 1),
+} xfs_buftarg_flags_t;
 
 typedef struct xfs_bufhash {
         struct list_head        bh_list;
@@ -77,477 +80,350 @@ typedef struct xfs_bufhash {
 } xfs_bufhash_t;
 
 typedef struct xfs_buftarg {
-        dev_t                   pbr_dev;
-        struct block_device     *pbr_bdev;
-        struct address_space    *pbr_mapping;
-        unsigned int            pbr_bsize;
-        unsigned int            pbr_sshift;
-        size_t                  pbr_smask;
-
-        /* per-device buffer hash table */
+        dev_t                   bt_dev;
+        struct block_device     *bt_bdev;
+        struct address_space    *bt_mapping;
+        unsigned int            bt_bsize;
+        unsigned int            bt_sshift;
+        size_t                  bt_smask;
+
+        /* per device buffer hash table */
         uint                    bt_hashmask;
         uint                    bt_hashshift;
         xfs_bufhash_t           *bt_hash;
+
+        /* per device delwri queue */
+        struct task_struct      *bt_task;
+        struct list_head        bt_list;
+        struct list_head        bt_delwrite_queue;
+        spinlock_t              bt_delwrite_lock;
+        unsigned long           bt_flags;
 } xfs_buftarg_t;
 
 /*
- *      xfs_buf_t:  Buffer structure for page cache-based buffers
+ *      xfs_buf_t:  Buffer structure for pagecache-based buffers
+ *
+ * This buffer structure is used by the pagecache buffer management routines
+ * to refer to an assembly of pages forming a logical buffer.
  *
- * This buffer structure is used by the page cache buffer management routines
- * to refer to an assembly of pages forming a logical buffer.  The actual I/O
- * is performed with buffer_head structures, as required by drivers.
- * 
- * The buffer structure is used on temporary basis only, and discarded when
- * released.  The real data storage is recorded in the page cache.  Metadata is
+ * The buffer structure is used on a temporary basis only, and discarded when
+ * released.  The real data storage is recorded in the pagecache. Buffers are
  * hashed to the block device on which the file system resides.
  */
 
 struct xfs_buf;
+typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);
+typedef void (*xfs_buf_relse_t)(struct xfs_buf *);
+typedef int (*xfs_buf_bdstrat_t)(struct xfs_buf *);
 
-/* call-back function on I/O completion */
-typedef void (*page_buf_iodone_t)(struct xfs_buf *);
-/* call-back function on I/O completion */
-typedef void (*page_buf_relse_t)(struct xfs_buf *);
-/* pre-write function */
-typedef int (*page_buf_bdstrat_t)(struct xfs_buf *);
-
-#define PB_PAGES        2
+#define XB_PAGES        2
 
 typedef struct xfs_buf {
-        struct semaphore        pb_sema;        /* semaphore for lockables  */
-        unsigned long           pb_queuetime;   /* time buffer was queued   */
-        atomic_t                pb_pin_count;   /* pin count                */
-        wait_queue_head_t       pb_waiters;     /* unpin waiters            */
-        struct list_head        pb_list;
-        page_buf_flags_t        pb_flags;       /* status flags */
-        struct list_head        pb_hash_list;   /* hash table list */
-        xfs_bufhash_t           *pb_hash;       /* hash table list start */
-        xfs_buftarg_t           *pb_target;     /* buffer target (device) */
-        atomic_t                pb_hold;        /* reference count */
-        xfs_daddr_t             pb_bn;          /* block number for I/O */
-        loff_t                  pb_file_offset; /* offset in file */
-        size_t                  pb_buffer_length; /* size of buffer in bytes */
-        size_t                  pb_count_desired; /* desired transfer size */
-        void                    *pb_addr;       /* virtual address of buffer */
-        struct work_struct      pb_iodone_work;
-        atomic_t                pb_io_remaining;/* #outstanding I/O requests */
-        page_buf_iodone_t       pb_iodone;      /* I/O completion function */
-        page_buf_relse_t        pb_relse;       /* releasing function */
-        page_buf_bdstrat_t      pb_strat;       /* pre-write function */
-        struct semaphore        pb_iodonesema;  /* Semaphore for I/O waiters */
-        void                    *pb_fspriv;
-        void                    *pb_fspriv2;
-        void                    *pb_fspriv3;
-        unsigned short          pb_error;       /* error code on I/O */
-        unsigned short          pb_locked;      /* page array is locked */
-        unsigned int            pb_page_count;  /* size of page array */
-        unsigned int            pb_offset;      /* page offset in first page */
-        struct page             **pb_pages;     /* array of page pointers */
-        struct page             *pb_page_array[PB_PAGES]; /* inline pages */
-#ifdef PAGEBUF_LOCK_TRACKING
-        int                     pb_last_holder;
+        struct semaphore        b_sema;         /* semaphore for lockables */
+        unsigned long           b_queuetime;    /* time buffer was queued */
+        atomic_t                b_pin_count;    /* pin count */
+        wait_queue_head_t       b_waiters;      /* unpin waiters */
+        struct list_head        b_list;
+        xfs_buf_flags_t         b_flags;        /* status flags */
+        struct list_head        b_hash_list;    /* hash table list */
+        xfs_bufhash_t           *b_hash;        /* hash table list start */
+        xfs_buftarg_t           *b_target;      /* buffer target (device) */
+        atomic_t                b_hold;         /* reference count */
+        xfs_daddr_t             b_bn;           /* block number for I/O */
+        xfs_off_t               b_file_offset;  /* offset in file */
+        size_t                  b_buffer_length;/* size of buffer in bytes */
+        size_t                  b_count_desired;/* desired transfer size */
+        void                    *b_addr;        /* virtual address of buffer */
+        struct work_struct      b_iodone_work;
+        atomic_t                b_io_remaining; /* #outstanding I/O requests */
+        xfs_buf_iodone_t        b_iodone;       /* I/O completion function */
+        xfs_buf_relse_t         b_relse;        /* releasing function */
+        xfs_buf_bdstrat_t       b_strat;        /* pre-write function */
+        struct semaphore        b_iodonesema;   /* Semaphore for I/O waiters */
+        void                    *b_fspriv;
+        void                    *b_fspriv2;
+        void                    *b_fspriv3;
+        unsigned short          b_error;        /* error code on I/O */
+        unsigned short          b_locked;       /* page array is locked */
+        unsigned int            b_page_count;   /* size of page array */
+        unsigned int            b_offset;       /* page offset in first page */
+        struct page             **b_pages;      /* array of page pointers */
+        struct page             *b_page_array[XB_PAGES]; /* inline pages */
+#ifdef XFS_BUF_LOCK_TRACKING
+        int                     b_last_holder;
 #endif
 } xfs_buf_t;
 
 
 /* Finding and Reading Buffers */
-
-extern xfs_buf_t *_pagebuf_find(        /* find buffer for block if     */
-                                        /* the block is in memory       */
-                xfs_buftarg_t *,        /* inode for block              */
-                loff_t,                 /* starting offset of range     */
-                size_t,                 /* length of range              */
-                page_buf_flags_t,       /* PBF_LOCK                     */
-                xfs_buf_t *);           /* newly allocated buffer       */
-
+extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t,
+                                xfs_buf_flags_t, xfs_buf_t *);
 #define xfs_incore(buftarg,blkno,len,lockit) \
-        _pagebuf_find(buftarg, blkno ,len, lockit, NULL)
-
-extern xfs_buf_t *xfs_buf_get_flags(    /* allocate a buffer            */
-                xfs_buftarg_t *,        /* inode for buffer             */
-                loff_t,                 /* starting offset of range     */
-                size_t,                 /* length of range              */
-                page_buf_flags_t);      /* PBF_LOCK, PBF_READ,          */
-                                        /* PBF_ASYNC                    */
+        _xfs_buf_find(buftarg, blkno ,len, lockit, NULL)
 
+extern xfs_buf_t *xfs_buf_get_flags(xfs_buftarg_t *, xfs_off_t, size_t,
+                                xfs_buf_flags_t);
 #define xfs_buf_get(target, blkno, len, flags) \
-        xfs_buf_get_flags((target), (blkno), (len), PBF_LOCK | PBF_MAPPED)
-
-extern xfs_buf_t *xfs_buf_read_flags(   /* allocate and read a buffer   */
-                xfs_buftarg_t *,        /* inode for buffer             */
-                loff_t,                 /* starting offset of range     */
-                size_t,                 /* length of range              */
-                page_buf_flags_t);      /* PBF_LOCK, PBF_ASYNC          */
+        xfs_buf_get_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
 
+extern xfs_buf_t *xfs_buf_read_flags(xfs_buftarg_t *, xfs_off_t, size_t,
+                                xfs_buf_flags_t);
 #define xfs_buf_read(target, blkno, len, flags) \
-        xfs_buf_read_flags((target), (blkno), (len), PBF_LOCK | PBF_MAPPED)
-
-extern xfs_buf_t *pagebuf_get_empty(    /* allocate pagebuf struct with */
-                                        /*  no memory or disk address   */
-                size_t len,
-                xfs_buftarg_t *);       /* mount point "fake" inode     */
-
-extern xfs_buf_t *pagebuf_get_no_daddr(/* allocate pagebuf struct       */
-                                        /* without disk address         */
-                size_t len,
-                xfs_buftarg_t *);       /* mount point "fake" inode     */
-
-extern int pagebuf_associate_memory(
-                xfs_buf_t *,
-                void *,
-                size_t);
-
-extern void pagebuf_hold(               /* increment reference count    */
-                xfs_buf_t *);           /* buffer to hold               */
+        xfs_buf_read_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
 
-extern void pagebuf_readahead(          /* read ahead into cache        */
-                xfs_buftarg_t  *,       /* target for buffer (or NULL)  */
-                loff_t,                 /* starting offset of range     */
-                size_t,                 /* length of range              */
-                page_buf_flags_t);      /* additional read flags        */
+extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
+extern xfs_buf_t *xfs_buf_get_noaddr(size_t, xfs_buftarg_t *);
+extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
+extern void xfs_buf_hold(xfs_buf_t *);
+extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t,
+                                xfs_buf_flags_t);
 
 /* Releasing Buffers */
-
-extern void pagebuf_free(               /* deallocate a buffer          */
-                xfs_buf_t *);           /* buffer to deallocate         */
-
-extern void pagebuf_rele(               /* release hold on a buffer     */
-                xfs_buf_t *);           /* buffer to release            */
+extern void xfs_buf_free(xfs_buf_t *);
+extern void xfs_buf_rele(xfs_buf_t *);
 
 /* Locking and Unlocking Buffers */
-
-extern int pagebuf_cond_lock(           /* lock buffer, if not locked   */
-                                        /* (returns -EBUSY if locked)   */
-                xfs_buf_t *);           /* buffer to lock               */
-
-extern int pagebuf_lock_value(          /* return count on lock         */
-                xfs_buf_t *);          /* buffer to check              */
-
-extern int pagebuf_lock(                /* lock buffer                  */
-                xfs_buf_t *);          /* buffer to lock               */
-
-extern void pagebuf_unlock(             /* unlock buffer                */
-                xfs_buf_t *);           /* buffer to unlock             */
+extern int xfs_buf_cond_lock(xfs_buf_t *);
+extern int xfs_buf_lock_value(xfs_buf_t *);
+extern void xfs_buf_lock(xfs_buf_t *);
+extern void xfs_buf_unlock(xfs_buf_t *);
 
 /* Buffer Read and Write Routines */
-
-extern void pagebuf_iodone(             /* mark buffer I/O complete     */
-                xfs_buf_t *,            /* buffer to mark               */
-                int);                   /* run completion locally, or in
-                                         * a helper thread.             */
-
-extern void pagebuf_ioerror(            /* mark buffer in error (or not) */
-                xfs_buf_t *,            /* buffer to mark               */
-                int);                   /* error to store (0 if none)   */
-
-extern int pagebuf_iostart(             /* start I/O on a buffer        */
-                xfs_buf_t *,            /* buffer to start              */
-                page_buf_flags_t);      /* PBF_LOCK, PBF_ASYNC,         */
-                                        /* PBF_READ, PBF_WRITE,         */
-                                        /* PBF_DELWRI                   */
-
-extern int pagebuf_iorequest(           /* start real I/O               */
-                xfs_buf_t *);           /* buffer to convey to device   */
-
-extern int pagebuf_iowait(              /* wait for buffer I/O done     */
-                xfs_buf_t *);           /* buffer to wait on            */
-
-extern void pagebuf_iomove(             /* move data in/out of pagebuf  */
-                xfs_buf_t *,            /* buffer to manipulate         */
-                size_t,                 /* starting buffer offset       */
-                size_t,                 /* length in buffer             */
-                caddr_t,                /* data pointer                 */
-                page_buf_rw_t);         /* direction                    */
-
-static inline int pagebuf_iostrategy(xfs_buf_t *pb)
+extern void xfs_buf_ioend(xfs_buf_t *,  int);
+extern void xfs_buf_ioerror(xfs_buf_t *, int);
+extern int xfs_buf_iostart(xfs_buf_t *, xfs_buf_flags_t);
+extern int xfs_buf_iorequest(xfs_buf_t *);
+extern int xfs_buf_iowait(xfs_buf_t *);
+extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, xfs_caddr_t,
+                                xfs_buf_rw_t);
+
+static inline int xfs_buf_iostrategy(xfs_buf_t *bp)
 {
-        return pb->pb_strat ? pb->pb_strat(pb) : pagebuf_iorequest(pb);
+        return bp->b_strat ? bp->b_strat(bp) : xfs_buf_iorequest(bp);
 }
 
-static inline int pagebuf_geterror(xfs_buf_t *pb)
+static inline int xfs_buf_geterror(xfs_buf_t *bp)
 {
-        return pb ? pb->pb_error : ENOMEM;
+        return bp ? bp->b_error : ENOMEM;
 }
 
 /* Buffer Utility Routines */
-
-extern caddr_t pagebuf_offset(          /* pointer at offset in buffer  */
-                xfs_buf_t *,            /* buffer to offset into        */
-                size_t);                /* offset                       */
+extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t);
 
 /* Pinning Buffer Storage in Memory */
-
-extern void pagebuf_pin(                /* pin buffer in memory         */
-                xfs_buf_t *);           /* buffer to pin                */
-
-extern void pagebuf_unpin(              /* unpin buffered data          */
-                xfs_buf_t *);           /* buffer to unpin              */
-
-extern int pagebuf_ispin(               /* check if buffer is pinned    */
-                xfs_buf_t *);           /* buffer to check              */
+extern void xfs_buf_pin(xfs_buf_t *);
+extern void xfs_buf_unpin(xfs_buf_t *);
+extern int xfs_buf_ispin(xfs_buf_t *);
 
 /* Delayed Write Buffer Routines */
-
-extern void pagebuf_delwri_dequeue(xfs_buf_t *);
+extern void xfs_buf_delwri_dequeue(xfs_buf_t *);
 
 /* Buffer Daemon Setup Routines */
+extern int xfs_buf_init(void);
+extern void xfs_buf_terminate(void);
 
-extern int pagebuf_init(void);
-extern void pagebuf_terminate(void);
-
-
-#ifdef PAGEBUF_TRACE
-extern ktrace_t *pagebuf_trace_buf;
-extern void pagebuf_trace(
-                xfs_buf_t *,            /* buffer being traced          */
-                char *,                 /* description of operation     */
-                void *,                 /* arbitrary diagnostic value   */
-                void *);                /* return address               */
+#ifdef XFS_BUF_TRACE
+extern ktrace_t *xfs_buf_trace_buf;
+extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 #else
-# define pagebuf_trace(pb, id, ptr, ra) do { } while (0)
+#define xfs_buf_trace(bp,id,ptr,ra)     do { } while (0)
 #endif
 
-#define pagebuf_target_name(target)     \
-        ({ char __b[BDEVNAME_SIZE]; bdevname((target)->pbr_bdev, __b); __b; })
+#define xfs_buf_target_name(target)     \
+        ({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; })
 
 
+#define XFS_B_ASYNC             XBF_ASYNC
+#define XFS_B_DELWRI            XBF_DELWRI
+#define XFS_B_READ              XBF_READ
+#define XFS_B_WRITE             XBF_WRITE
+#define XFS_B_STALE             XBF_STALE
 
-/* These are just for xfs_syncsub... it sets an internal variable
- * then passes it to VOP_FLUSH_PAGES or adds the flags to a newly gotten buf_t
- */
-#define XFS_B_ASYNC             PBF_ASYNC
-#define XFS_B_DELWRI            PBF_DELWRI
-#define XFS_B_READ              PBF_READ
-#define XFS_B_WRITE             PBF_WRITE
-#define XFS_B_STALE             PBF_STALE
-
-#define XFS_BUF_TRYLOCK         PBF_TRYLOCK
-#define XFS_INCORE_TRYLOCK      PBF_TRYLOCK
-#define XFS_BUF_LOCK            PBF_LOCK
-#define XFS_BUF_MAPPED          PBF_MAPPED
-
-#define BUF_BUSY                PBF_DONT_BLOCK
-
-#define XFS_BUF_BFLAGS(x)       ((x)->pb_flags)
-#define XFS_BUF_ZEROFLAGS(x)    \
-        ((x)->pb_flags &= ~(PBF_READ|PBF_WRITE|PBF_ASYNC|PBF_DELWRI))
-
-#define XFS_BUF_STALE(x)        ((x)->pb_flags |= XFS_B_STALE)
-#define XFS_BUF_UNSTALE(x)      ((x)->pb_flags &= ~XFS_B_STALE)
-#define XFS_BUF_ISSTALE(x)      ((x)->pb_flags & XFS_B_STALE)
-#define XFS_BUF_SUPER_STALE(x)  do {                            \
-                                        XFS_BUF_STALE(x);       \
-                                        pagebuf_delwri_dequeue(x);      \
-                                        XFS_BUF_DONE(x);        \
-                                } while (0)
+#define XFS_BUF_TRYLOCK         XBF_TRYLOCK
+#define XFS_INCORE_TRYLOCK      XBF_TRYLOCK
+#define XFS_BUF_LOCK            XBF_LOCK
+#define XFS_BUF_MAPPED          XBF_MAPPED
 
-#define XFS_BUF_MANAGE          PBF_FS_MANAGED
-#define XFS_BUF_UNMANAGE(x)     ((x)->pb_flags &= ~PBF_FS_MANAGED)
-
-#define XFS_BUF_DELAYWRITE(x)    ((x)->pb_flags |= PBF_DELWRI)
-#define XFS_BUF_UNDELAYWRITE(x)  pagebuf_delwri_dequeue(x)
-#define XFS_BUF_ISDELAYWRITE(x)  ((x)->pb_flags & PBF_DELWRI)
-
-#define XFS_BUF_ERROR(x,no)      pagebuf_ioerror(x,no)
-#define XFS_BUF_GETERROR(x)      pagebuf_geterror(x)
-#define XFS_BUF_ISERROR(x)       (pagebuf_geterror(x)?1:0)
-
-#define XFS_BUF_DONE(x)          ((x)->pb_flags |= PBF_DONE)
-#define XFS_BUF_UNDONE(x)        ((x)->pb_flags &= ~PBF_DONE)
-#define XFS_BUF_ISDONE(x)        ((x)->pb_flags & PBF_DONE)
-
-#define XFS_BUF_BUSY(x)          do { } while (0)
-#define XFS_BUF_UNBUSY(x)        do { } while (0)
-#define XFS_BUF_ISBUSY(x)        (1)
-
-#define XFS_BUF_ASYNC(x)         ((x)->pb_flags |= PBF_ASYNC)
-#define XFS_BUF_UNASYNC(x)       ((x)->pb_flags &= ~PBF_ASYNC)
-#define XFS_BUF_ISASYNC(x)       ((x)->pb_flags & PBF_ASYNC)
-
-#define XFS_BUF_ORDERED(x)       ((x)->pb_flags |= PBF_ORDERED)
-#define XFS_BUF_UNORDERED(x)     ((x)->pb_flags &= ~PBF_ORDERED)
-#define XFS_BUF_ISORDERED(x)     ((x)->pb_flags & PBF_ORDERED)
-
-#define XFS_BUF_SHUT(x)          printk("XFS_BUF_SHUT not implemented yet\n")
-#define XFS_BUF_UNSHUT(x)        printk("XFS_BUF_UNSHUT not implemented yet\n")
-#define XFS_BUF_ISSHUT(x)        (0)
-
-#define XFS_BUF_HOLD(x)         pagebuf_hold(x)
-#define XFS_BUF_READ(x)         ((x)->pb_flags |= PBF_READ)
-#define XFS_BUF_UNREAD(x)       ((x)->pb_flags &= ~PBF_READ)
-#define XFS_BUF_ISREAD(x)       ((x)->pb_flags & PBF_READ)
-
-#define XFS_BUF_WRITE(x)        ((x)->pb_flags |= PBF_WRITE)
-#define XFS_BUF_UNWRITE(x)      ((x)->pb_flags &= ~PBF_WRITE)
-#define XFS_BUF_ISWRITE(x)      ((x)->pb_flags & PBF_WRITE)
-
-#define XFS_BUF_ISUNINITIAL(x)   (0)
-#define XFS_BUF_UNUNINITIAL(x)   (0)
-
-#define XFS_BUF_BP_ISMAPPED(bp)  1
-
-#define XFS_BUF_IODONE_FUNC(buf)        (buf)->pb_iodone
-#define XFS_BUF_SET_IODONE_FUNC(buf, func)      \
-                        (buf)->pb_iodone = (func)
-#define XFS_BUF_CLR_IODONE_FUNC(buf)            \
-                        (buf)->pb_iodone = NULL
-#define XFS_BUF_SET_BDSTRAT_FUNC(buf, func)     \
-                        (buf)->pb_strat = (func)
-#define XFS_BUF_CLR_BDSTRAT_FUNC(buf)           \
-                        (buf)->pb_strat = NULL
-
-#define XFS_BUF_FSPRIVATE(buf, type)            \
-                        ((type)(buf)->pb_fspriv)
-#define XFS_BUF_SET_FSPRIVATE(buf, value)       \
-                        (buf)->pb_fspriv = (void *)(value)
-#define XFS_BUF_FSPRIVATE2(buf, type)           \
-                        ((type)(buf)->pb_fspriv2)
-#define XFS_BUF_SET_FSPRIVATE2(buf, value)      \
-                        (buf)->pb_fspriv2 = (void *)(value)
-#define XFS_BUF_FSPRIVATE3(buf, type)           \
-                        ((type)(buf)->pb_fspriv3)
-#define XFS_BUF_SET_FSPRIVATE3(buf, value)      \
-                        (buf)->pb_fspriv3  = (void *)(value)
-#define XFS_BUF_SET_START(buf)
-
-#define XFS_BUF_SET_BRELSE_FUNC(buf, value) \
-                        (buf)->pb_relse = (value)
-
-#define XFS_BUF_PTR(bp)         (xfs_caddr_t)((bp)->pb_addr)
-
-static inline xfs_caddr_t xfs_buf_offset(xfs_buf_t *bp, size_t offset)
-{
-        if (bp->pb_flags & PBF_MAPPED)
-                return XFS_BUF_PTR(bp) + offset;
-        return (xfs_caddr_t) pagebuf_offset(bp, offset);
-}
+#define BUF_BUSY                XBF_DONT_BLOCK
+
+#define XFS_BUF_BFLAGS(bp)      ((bp)->b_flags)
+#define XFS_BUF_ZEROFLAGS(bp)   \
+        ((bp)->b_flags &= ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI))
+
+#define XFS_BUF_STALE(bp)       ((bp)->b_flags |= XFS_B_STALE)
+#define XFS_BUF_UNSTALE(bp)     ((bp)->b_flags &= ~XFS_B_STALE)
+#define XFS_BUF_ISSTALE(bp)     ((bp)->b_flags & XFS_B_STALE)
+#define XFS_BUF_SUPER_STALE(bp) do {                            \
+                                        XFS_BUF_STALE(bp);      \
+                                        xfs_buf_delwri_dequeue(bp);     \
+                                        XFS_BUF_DONE(bp);       \
+                                } while (0)
 
-#define XFS_BUF_SET_PTR(bp, val, count)         \
-                                pagebuf_associate_memory(bp, val, count)
-#define XFS_BUF_ADDR(bp)        ((bp)->pb_bn)
-#define XFS_BUF_SET_ADDR(bp, blk)               \
-                        ((bp)->pb_bn = (xfs_daddr_t)(blk))
-#define XFS_BUF_OFFSET(bp)      ((bp)->pb_file_offset)
-#define XFS_BUF_SET_OFFSET(bp, off)             \
-                        ((bp)->pb_file_offset = (off))
-#define XFS_BUF_COUNT(bp)       ((bp)->pb_count_desired)
-#define XFS_BUF_SET_COUNT(bp, cnt)              \
-                        ((bp)->pb_count_desired = (cnt))
-#define XFS_BUF_SIZE(bp)        ((bp)->pb_buffer_length)
-#define XFS_BUF_SET_SIZE(bp, cnt)               \
-                        ((bp)->pb_buffer_length = (cnt))
-#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)
-#define XFS_BUF_SET_VTYPE(bp, type)
-#define XFS_BUF_SET_REF(bp, ref)
-
-#define XFS_BUF_ISPINNED(bp)    pagebuf_ispin(bp)
-
-#define XFS_BUF_VALUSEMA(bp)    pagebuf_lock_value(bp)
-#define XFS_BUF_CPSEMA(bp)      (pagebuf_cond_lock(bp) == 0)
-#define XFS_BUF_VSEMA(bp)       pagebuf_unlock(bp)
-#define XFS_BUF_PSEMA(bp,x)     pagebuf_lock(bp)
-#define XFS_BUF_V_IODONESEMA(bp) up(&bp->pb_iodonesema);
-
-/* setup the buffer target from a buftarg structure */
-#define XFS_BUF_SET_TARGET(bp, target)  \
-                (bp)->pb_target = (target)
-#define XFS_BUF_TARGET(bp)      ((bp)->pb_target)
-#define XFS_BUFTARG_NAME(target)        \
-                pagebuf_target_name(target)
-
-#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)
-#define XFS_BUF_SET_VTYPE(bp, type)
-#define XFS_BUF_SET_REF(bp, ref)
-
-static inline int       xfs_bawrite(void *mp, xfs_buf_t *bp)
+#define XFS_BUF_MANAGE          XBF_FS_MANAGED
+#define XFS_BUF_UNMANAGE(bp)    ((bp)->b_flags &= ~XBF_FS_MANAGED)
+
+#define XFS_BUF_DELAYWRITE(bp)          ((bp)->b_flags |= XBF_DELWRI)
+#define XFS_BUF_UNDELAYWRITE(bp)        xfs_buf_delwri_dequeue(bp)
+#define XFS_BUF_ISDELAYWRITE(bp)        ((bp)->b_flags & XBF_DELWRI)
+
+#define XFS_BUF_ERROR(bp,no)    xfs_buf_ioerror(bp,no)
+#define XFS_BUF_GETERROR(bp)    xfs_buf_geterror(bp)
+#define XFS_BUF_ISERROR(bp)     (xfs_buf_geterror(bp) ? 1 : 0)
+
+#define XFS_BUF_DONE(bp)        ((bp)->b_flags |= XBF_DONE)
+#define XFS_BUF_UNDONE(bp)      ((bp)->b_flags &= ~XBF_DONE)
+#define XFS_BUF_ISDONE(bp)      ((bp)->b_flags & XBF_DONE)
+
+#define XFS_BUF_BUSY(bp)        do { } while (0)
+#define XFS_BUF_UNBUSY(bp)      do { } while (0)
+#define XFS_BUF_ISBUSY(bp)      (1)
+
+#define XFS_BUF_ASYNC(bp)       ((bp)->b_flags |= XBF_ASYNC)
+#define XFS_BUF_UNASYNC(bp)     ((bp)->b_flags &= ~XBF_ASYNC)
+#define XFS_BUF_ISASYNC(bp)     ((bp)->b_flags & XBF_ASYNC)
+
+#define XFS_BUF_ORDERED(bp)     ((bp)->b_flags |= XBF_ORDERED)
+#define XFS_BUF_UNORDERED(bp)   ((bp)->b_flags &= ~XBF_ORDERED)
+#define XFS_BUF_ISORDERED(bp)   ((bp)->b_flags & XBF_ORDERED)
+
+#define XFS_BUF_SHUT(bp)        do { } while (0)
+#define XFS_BUF_UNSHUT(bp)      do { } while (0)
+#define XFS_BUF_ISSHUT(bp)      (0)
+
+#define XFS_BUF_HOLD(bp)        xfs_buf_hold(bp)
+#define XFS_BUF_READ(bp)        ((bp)->b_flags |= XBF_READ)
+#define XFS_BUF_UNREAD(bp)      ((bp)->b_flags &= ~XBF_READ)
+#define XFS_BUF_ISREAD(bp)      ((bp)->b_flags & XBF_READ)
+
+#define XFS_BUF_WRITE(bp)       ((bp)->b_flags |= XBF_WRITE)
+#define XFS_BUF_UNWRITE(bp)     ((bp)->b_flags &= ~XBF_WRITE)
+#define XFS_BUF_ISWRITE(bp)     ((bp)->b_flags & XBF_WRITE)
+
+#define XFS_BUF_ISUNINITIAL(bp) (0)
+#define XFS_BUF_UNUNINITIAL(bp) (0)
+
+#define XFS_BUF_BP_ISMAPPED(bp) (1)
+
+#define XFS_BUF_IODONE_FUNC(bp)                 ((bp)->b_iodone)
+#define XFS_BUF_SET_IODONE_FUNC(bp, func)       ((bp)->b_iodone = (func))
+#define XFS_BUF_CLR_IODONE_FUNC(bp)             ((bp)->b_iodone = NULL)
+#define XFS_BUF_SET_BDSTRAT_FUNC(bp, func)      ((bp)->b_strat = (func))
+#define XFS_BUF_CLR_BDSTRAT_FUNC(bp)            ((bp)->b_strat = NULL)
+
+#define XFS_BUF_FSPRIVATE(bp, type)             ((type)(bp)->b_fspriv)
+#define XFS_BUF_SET_FSPRIVATE(bp, val)          ((bp)->b_fspriv = (void*)(val))
+#define XFS_BUF_FSPRIVATE2(bp, type)            ((type)(bp)->b_fspriv2)
+#define XFS_BUF_SET_FSPRIVATE2(bp, val)         ((bp)->b_fspriv2 = (void*)(val))
+#define XFS_BUF_FSPRIVATE3(bp, type)            ((type)(bp)->b_fspriv3)
+#define XFS_BUF_SET_FSPRIVATE3(bp, val)         ((bp)->b_fspriv3 = (void*)(val))
+#define XFS_BUF_SET_START(bp)                   do { } while (0)
+#define XFS_BUF_SET_BRELSE_FUNC(bp, func)       ((bp)->b_relse = (func))
+
+#define XFS_BUF_PTR(bp)                 (xfs_caddr_t)((bp)->b_addr)
+#define XFS_BUF_SET_PTR(bp, val, cnt)   xfs_buf_associate_memory(bp, val, cnt)
+#define XFS_BUF_ADDR(bp)                ((bp)->b_bn)
+#define XFS_BUF_SET_ADDR(bp, bno)       ((bp)->b_bn = (xfs_daddr_t)(bno))
+#define XFS_BUF_OFFSET(bp)              ((bp)->b_file_offset)
+#define XFS_BUF_SET_OFFSET(bp, off)     ((bp)->b_file_offset = (off))
+#define XFS_BUF_COUNT(bp)               ((bp)->b_count_desired)
+#define XFS_BUF_SET_COUNT(bp, cnt)      ((bp)->b_count_desired = (cnt))
+#define XFS_BUF_SIZE(bp)                ((bp)->b_buffer_length)
+#define XFS_BUF_SET_SIZE(bp, cnt)       ((bp)->b_buffer_length = (cnt))
+
+#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)    do { } while (0)
+#define XFS_BUF_SET_VTYPE(bp, type)             do { } while (0)
+#define XFS_BUF_SET_REF(bp, ref)                do { } while (0)
+
+#define XFS_BUF_ISPINNED(bp)    xfs_buf_ispin(bp)
+
+#define XFS_BUF_VALUSEMA(bp)    xfs_buf_lock_value(bp)
+#define XFS_BUF_CPSEMA(bp)      (xfs_buf_cond_lock(bp) == 0)
+#define XFS_BUF_VSEMA(bp)       xfs_buf_unlock(bp)
+#define XFS_BUF_PSEMA(bp,x)     xfs_buf_lock(bp)
+#define XFS_BUF_V_IODONESEMA(bp) up(&bp->b_iodonesema);
+
+#define XFS_BUF_SET_TARGET(bp, target)  ((bp)->b_target = (target))
+#define XFS_BUF_TARGET(bp)              ((bp)->b_target)
+#define XFS_BUFTARG_NAME(target)        xfs_buf_target_name(target)
+
+static inline int xfs_bawrite(void *mp, xfs_buf_t *bp)
 {
-        bp->pb_fspriv3 = mp;
-        bp->pb_strat = xfs_bdstrat_cb;
-        pagebuf_delwri_dequeue(bp);
-        return pagebuf_iostart(bp, PBF_WRITE | PBF_ASYNC | _PBF_RUN_QUEUES);
+        bp->b_fspriv3 = mp;
+        bp->b_strat = xfs_bdstrat_cb;
+        xfs_buf_delwri_dequeue(bp);
+        return xfs_buf_iostart(bp, XBF_WRITE | XBF_ASYNC | _XBF_RUN_QUEUES);
 }
 
-static inline void      xfs_buf_relse(xfs_buf_t *bp)
+static inline void xfs_buf_relse(xfs_buf_t *bp)
 {
-        if (!bp->pb_relse)
-                pagebuf_unlock(bp);
-        pagebuf_rele(bp);
+        if (!bp->b_relse)
+                xfs_buf_unlock(bp);
+        xfs_buf_rele(bp);
 }
 
-#define xfs_bpin(bp)            pagebuf_pin(bp)
-#define xfs_bunpin(bp)          pagebuf_unpin(bp)
+#define xfs_bpin(bp)            xfs_buf_pin(bp)
+#define xfs_bunpin(bp)          xfs_buf_unpin(bp)
 
 #define xfs_buftrace(id, bp)    \
-            pagebuf_trace(bp, id, NULL, (void *)__builtin_return_address(0))
+            xfs_buf_trace(bp, id, NULL, (void *)__builtin_return_address(0))
 
-#define xfs_biodone(pb)             \
-            pagebuf_iodone(pb, 0)
+#define xfs_biodone(bp)         xfs_buf_ioend(bp, 0)
 
-#define xfs_biomove(pb, off, len, data, rw) \
-            pagebuf_iomove((pb), (off), (len), (data), \
-                ((rw) == XFS_B_WRITE) ? PBRW_WRITE : PBRW_READ)
+#define xfs_biomove(bp, off, len, data, rw) \
+            xfs_buf_iomove((bp), (off), (len), (data), \
+                ((rw) == XFS_B_WRITE) ? XBRW_WRITE : XBRW_READ)
 
-#define xfs_biozero(pb, off, len) \
-            pagebuf_iomove((pb), (off), (len), NULL, PBRW_ZERO)
+#define xfs_biozero(bp, off, len) \
+            xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO)
 
 
-static inline int       XFS_bwrite(xfs_buf_t *pb)
+static inline int XFS_bwrite(xfs_buf_t *bp)
 {
-        int     iowait = (pb->pb_flags & PBF_ASYNC) == 0;
+        int     iowait = (bp->b_flags & XBF_ASYNC) == 0;
         int     error = 0;
 
         if (!iowait)
-                pb->pb_flags |= _PBF_RUN_QUEUES;
+                bp->b_flags |= _XBF_RUN_QUEUES;
 
-        pagebuf_delwri_dequeue(pb);
-        pagebuf_iostrategy(pb);
+        xfs_buf_delwri_dequeue(bp);
+        xfs_buf_iostrategy(bp);
         if (iowait) {
-                error = pagebuf_iowait(pb);
-                xfs_buf_relse(pb);
+                error = xfs_buf_iowait(bp);
+                xfs_buf_relse(bp);
         }
         return error;
 }
 
-#define XFS_bdwrite(pb)              \
-            pagebuf_iostart(pb, PBF_DELWRI | PBF_ASYNC)
+#define XFS_bdwrite(bp)         xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC)
 
 static inline int xfs_bdwrite(void *mp, xfs_buf_t *bp)
 {
-        bp->pb_strat = xfs_bdstrat_cb;
-        bp->pb_fspriv3 = mp;
-
-        return pagebuf_iostart(bp, PBF_DELWRI | PBF_ASYNC);
+        bp->b_strat = xfs_bdstrat_cb;
+        bp->b_fspriv3 = mp;
+        return xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC);
 }
 
-#define XFS_bdstrat(bp) pagebuf_iorequest(bp)
+#define XFS_bdstrat(bp) xfs_buf_iorequest(bp)
 
-#define xfs_iowait(pb)  pagebuf_iowait(pb)
+#define xfs_iowait(bp)  xfs_buf_iowait(bp)
 
 #define xfs_baread(target, rablkno, ralen)  \
-        pagebuf_readahead((target), (rablkno), (ralen), PBF_DONT_BLOCK)
-
-#define xfs_buf_get_empty(len, target)  pagebuf_get_empty((len), (target))
-#define xfs_buf_get_noaddr(len, target) pagebuf_get_no_daddr((len), (target))
-#define xfs_buf_free(bp)                pagebuf_free(bp)
+        xfs_buf_readahead((target), (rablkno), (ralen), XBF_DONT_BLOCK)
 
 
 /*
  *      Handling of buftargs.
  */
-
 extern xfs_buftarg_t *xfs_alloc_buftarg(struct block_device *, int);
 extern void xfs_free_buftarg(xfs_buftarg_t *, int);
 extern void xfs_wait_buftarg(xfs_buftarg_t *);
 extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int);
 extern int xfs_flush_buftarg(xfs_buftarg_t *, int);
 
-#define xfs_getsize_buftarg(buftarg) \
-        block_size((buftarg)->pbr_bdev)
-#define xfs_readonly_buftarg(buftarg) \
-        bdev_read_only((buftarg)->pbr_bdev)
-#define xfs_binval(buftarg) \
-        xfs_flush_buftarg(buftarg, 1)
-#define XFS_bflush(buftarg) \
-        xfs_flush_buftarg(buftarg, 1)
+#define xfs_getsize_buftarg(buftarg)    block_size((buftarg)->bt_bdev)
+#define xfs_readonly_buftarg(buftarg)   bdev_read_only((buftarg)->bt_bdev)
+
+#define xfs_binval(buftarg)             xfs_flush_buftarg(buftarg, 1)
+#define XFS_bflush(buftarg)             xfs_flush_buftarg(buftarg, 1)
 
 #endif  /* __XFS_BUF_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 06111d0bbae4..ced4404339c7 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -509,16 +509,14 @@ linvfs_open_exec(
         vnode_t         *vp = LINVFS_GET_VP(inode);
         xfs_mount_t     *mp = XFS_VFSTOM(vp->v_vfsp);
         int             error = 0;
-        bhv_desc_t      *bdp;
         xfs_inode_t     *ip;
 
         if (vp->v_vfsp->vfs_flag & VFS_DMI) {
-                bdp = vn_bhv_lookup(VN_BHV_HEAD(vp), &xfs_vnodeops);
-                if (!bdp) {
+                ip = xfs_vtoi(vp);
+                if (!ip) {
                         error = -EINVAL;
                         goto open_exec_out;
                 }
-                ip = XFS_BHVTOI(bdp);
                 if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ)) {
                         error = -XFS_SEND_DATA(mp, DM_EVENT_READ, vp,
                                                0, 0, 0, NULL);
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index 21667ba6dcd5..4db47790415c 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -146,13 +146,10 @@ xfs_find_handle(
 
         if (cmd != XFS_IOC_PATH_TO_FSHANDLE) {
                 xfs_inode_t     *ip;
-                bhv_desc_t      *bhv;
                 int             lock_mode;
 
                 /* need to get access to the xfs_inode to read the generation */
-                bhv = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
-                ASSERT(bhv);
-                ip = XFS_BHVTOI(bhv);
+                ip = xfs_vtoi(vp);
                 ASSERT(ip);
                 lock_mode = xfs_ilock_map_shared(ip);
 
@@ -751,9 +748,8 @@ xfs_ioctl(
                         (ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
                         mp->m_rtdev_targp : mp->m_ddev_targp;
 
-                da.d_mem = da.d_miniosz = 1 << target->pbr_sshift;
-                /* The size dio will do in one go */
-                da.d_maxiosz = 64 * PAGE_CACHE_SIZE;
+                da.d_mem = da.d_miniosz = 1 << target->bt_sshift;
+                da.d_maxiosz = INT_MAX & ~(da.d_miniosz - 1);
 
                 if (copy_to_user(arg, &da, sizeof(da)))
                         return -XFS_ERROR(EFAULT);
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index 9b8ee3470ecc..4bd3d03b23ed 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -54,11 +54,46 @@
 #include <linux/capability.h>
 #include <linux/xattr.h>
 #include <linux/namei.h>
+#include <linux/security.h>
 
 #define IS_NOATIME(inode) ((inode->i_sb->s_flags & MS_NOATIME) ||       \
         (S_ISDIR(inode->i_mode) && inode->i_sb->s_flags & MS_NODIRATIME))
 
 /*
+ * Get a XFS inode from a given vnode.
+ */
+xfs_inode_t *
+xfs_vtoi(
+        struct vnode    *vp)
+{
+        bhv_desc_t      *bdp;
+
+        bdp = bhv_lookup_range(VN_BHV_HEAD(vp),
+                        VNODE_POSITION_XFS, VNODE_POSITION_XFS);
+        if (unlikely(bdp == NULL))
+                return NULL;
+        return XFS_BHVTOI(bdp);
+}
+
+/*
+ * Bring the atime in the XFS inode uptodate.
+ * Used before logging the inode to disk or when the Linux inode goes away.
+ */
+void
+xfs_synchronize_atime(
+        xfs_inode_t     *ip)
+{
+        vnode_t         *vp;
+
+        vp = XFS_ITOV_NULL(ip);
+        if (vp) {
+                struct inode *inode = &vp->v_inode;
+                ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
+                ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
+        }
+}
+
+/*
  * Change the requested timestamp in the given inode.
  * We don't lock across timestamp updates, and we don't log them but
  * we do record the fact that there is dirty information in core.
@@ -77,23 +112,6 @@ xfs_ichgtime(
         struct inode    *inode = LINVFS_GET_IP(XFS_ITOV(ip));
         timespec_t      tv;
 
-        /*
-         * We're not supposed to change timestamps in readonly-mounted
-         * filesystems.  Throw it away if anyone asks us.
-         */
-        if (unlikely(IS_RDONLY(inode)))
-                return;
-
-        /*
-         * Don't update access timestamps on reads if mounted "noatime".
-         * Throw it away if anyone asks us.
-         */
-        if (unlikely(
-            (ip->i_mount->m_flags & XFS_MOUNT_NOATIME || IS_NOATIME(inode)) &&
-            (flags & (XFS_ICHGTIME_ACC|XFS_ICHGTIME_MOD|XFS_ICHGTIME_CHG)) ==
-                        XFS_ICHGTIME_ACC))
-                return;
-
         nanotime(&tv);
         if (flags & XFS_ICHGTIME_MOD) {
                 inode->i_mtime = tv;
@@ -130,8 +148,6 @@ xfs_ichgtime(
  * Variant on the above which avoids querying the system clock
  * in situations where we know the Linux inode timestamps have
  * just been updated (and so we can update our inode cheaply).
- * We also skip the readonly and noatime checks here, they are
- * also catered for already.
  */
 void
 xfs_ichgtime_fast(
@@ -142,20 +158,16 @@ xfs_ichgtime_fast(
         timespec_t      *tvp;
 
         /*
-         * We're not supposed to change timestamps in readonly-mounted
-         * filesystems.  Throw it away if anyone asks us.
+         * Atime updates for read() & friends are handled lazily now, and
+         * explicit updates must go through xfs_ichgtime()
          */
-        if (unlikely(IS_RDONLY(inode)))
-                return;
+        ASSERT((flags & XFS_ICHGTIME_ACC) == 0);
 
         /*
-         * Don't update access timestamps on reads if mounted "noatime".
-         * Throw it away if anyone asks us.
+         * We're not supposed to change timestamps in readonly-mounted
+         * filesystems.  Throw it away if anyone asks us.
          */
-        if (unlikely(
-            (ip->i_mount->m_flags & XFS_MOUNT_NOATIME || IS_NOATIME(inode)) &&
-            ((flags & (XFS_ICHGTIME_ACC|XFS_ICHGTIME_MOD|XFS_ICHGTIME_CHG)) ==
-                        XFS_ICHGTIME_ACC)))
+        if (unlikely(IS_RDONLY(inode)))
                 return;
 
         if (flags & XFS_ICHGTIME_MOD) {
@@ -163,11 +175,6 @@ xfs_ichgtime_fast(
                 ip->i_d.di_mtime.t_sec = (__int32_t)tvp->tv_sec;
                 ip->i_d.di_mtime.t_nsec = (__int32_t)tvp->tv_nsec;
         }
-        if (flags & XFS_ICHGTIME_ACC) {
-                tvp = &inode->i_atime;
-                ip->i_d.di_atime.t_sec = (__int32_t)tvp->tv_sec;
-                ip->i_d.di_atime.t_nsec = (__int32_t)tvp->tv_nsec;
-        }
         if (flags & XFS_ICHGTIME_CHG) {
                 tvp = &inode->i_ctime;
                 ip->i_d.di_ctime.t_sec = (__int32_t)tvp->tv_sec;
@@ -214,6 +221,39 @@ validate_fields(
 }
 
 /*
+ * Hook in SELinux.  This is not quite correct yet, what we really need
+ * here (as we do for default ACLs) is a mechanism by which creation of
+ * these attrs can be journalled at inode creation time (along with the
+ * inode, of course, such that log replay can't cause these to be lost).
+ */
+STATIC int
+linvfs_init_security(
+        struct vnode    *vp,
+        struct inode    *dir)
+{
+        struct inode    *ip = LINVFS_GET_IP(vp);
+        size_t          length;
+        void            *value;
+        char            *name;
+        int             error;
+
+        error = security_inode_init_security(ip, dir, &name, &value, &length);
+        if (error) {
+                if (error == -EOPNOTSUPP)
+                        return 0;
+                return -error;
+        }
+
+        VOP_ATTR_SET(vp, name, value, length, ATTR_SECURE, NULL, error);
+        if (!error)
+                VMODIFY(vp);
+
+        kfree(name);
+        kfree(value);
+        return error;
+}
+
+/*
  * Determine whether a process has a valid fs_struct (kernel daemons
  * like knfsd don't have an fs_struct).
  *
@@ -278,6 +318,9 @@ linvfs_mknod(
                 break;
         }
 
+        if (!error)
+                error = linvfs_init_security(vp, dir);
+
         if (default_acl) {
                 if (!error) {
                         error = _ACL_INHERIT(vp, &va, default_acl);
@@ -294,8 +337,6 @@ linvfs_mknod(
                                 teardown.d_inode = ip = LINVFS_GET_IP(vp);
                                 teardown.d_name = dentry->d_name;
 
-                                vn_mark_bad(vp);
-                                
                                 if (S_ISDIR(mode))
                                         VOP_RMDIR(dvp, &teardown, NULL, err2);
                                 else
@@ -506,7 +547,7 @@ linvfs_follow_link(
         ASSERT(dentry);
         ASSERT(nd);
 
-        link = (char *)kmalloc(MAXNAMELEN+1, GFP_KERNEL);
+        link = (char *)kmalloc(MAXPATHLEN+1, GFP_KERNEL);
         if (!link) {
                 nd_set_link(nd, ERR_PTR(-ENOMEM));
                 return NULL;
@@ -522,12 +563,12 @@ linvfs_follow_link(
         vp = LINVFS_GET_VP(dentry->d_inode);
 
         iov.iov_base = link;
-        iov.iov_len = MAXNAMELEN;
+        iov.iov_len = MAXPATHLEN;
 
         uio->uio_iov = &iov;
         uio->uio_offset = 0;
         uio->uio_segflg = UIO_SYSSPACE;
-        uio->uio_resid = MAXNAMELEN;
+        uio->uio_resid = MAXPATHLEN;
         uio->uio_iovcnt = 1;
 
         VOP_READLINK(vp, uio, 0, NULL, error);
@@ -535,7 +576,7 @@ linvfs_follow_link(
                 kfree(link);
                 link = ERR_PTR(-error);
         } else {
-                link[MAXNAMELEN - uio->uio_resid] = '\0';
+                link[MAXPATHLEN - uio->uio_resid] = '\0';
         }
         kfree(uio);
 
diff --git a/fs/xfs/linux-2.6/xfs_iops.h b/fs/xfs/linux-2.6/xfs_iops.h
index ee784b63acbf..6899a6b4a50a 100644
--- a/fs/xfs/linux-2.6/xfs_iops.h
+++ b/fs/xfs/linux-2.6/xfs_iops.h
@@ -26,11 +26,6 @@ extern struct file_operations linvfs_file_operations;
 extern struct file_operations linvfs_invis_file_operations;
 extern struct file_operations linvfs_dir_operations;
 
-extern struct address_space_operations linvfs_aops;
-
-extern int linvfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
-extern void linvfs_unwritten_done(struct buffer_head *, int);
-
 extern int xfs_ioctl(struct bhv_desc *, struct inode *, struct file *,
                         int, unsigned int, void __user *);
 
diff --git a/fs/xfs/linux-2.6/xfs_linux.h b/fs/xfs/linux-2.6/xfs_linux.h
index d8e21ba0cccc..67389b745526 100644
--- a/fs/xfs/linux-2.6/xfs_linux.h
+++ b/fs/xfs/linux-2.6/xfs_linux.h
@@ -110,10 +110,6 @@
  * delalloc and these ondisk-uninitialised buffers.
  */
 BUFFER_FNS(PrivateStart, unwritten);
-static inline void set_buffer_unwritten_io(struct buffer_head *bh)
-{
-        bh->b_end_io = linvfs_unwritten_done;
-}
 
 #define restricted_chown        xfs_params.restrict_chown.val
 #define irix_sgid_inherit       xfs_params.sgid_inherit.val
@@ -232,7 +228,7 @@ static inline void set_buffer_unwritten_io(struct buffer_head *bh)
 #define xfs_itruncate_data(ip, off)     \
         (-vmtruncate(LINVFS_GET_IP(XFS_ITOV(ip)), (off)))
 #define xfs_statvfs_fsid(statp, mp)     \
-        ({ u64 id = huge_encode_dev((mp)->m_dev);       \
+        ({ u64 id = huge_encode_dev((mp)->m_ddev_targp->bt_dev); \
            __kernel_fsid_t *fsid = &(statp)->f_fsid;    \
         (fsid->val[0] = (u32)id, fsid->val[1] = (u32)(id >> 32)); })
 
diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c
index 885dfafeabee..e0ab45fbfebd 100644
--- a/fs/xfs/linux-2.6/xfs_lrw.c
+++ b/fs/xfs/linux-2.6/xfs_lrw.c
@@ -233,8 +233,8 @@ xfs_read(
                 xfs_buftarg_t   *target =
                         (ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
                                 mp->m_rtdev_targp : mp->m_ddev_targp;
-                if ((*offset & target->pbr_smask) ||
-                    (size & target->pbr_smask)) {
+                if ((*offset & target->bt_smask) ||
+                    (size & target->bt_smask)) {
                         if (*offset == ip->i_d.di_size) {
                                 return (0);
                         }
@@ -281,9 +281,6 @@ xfs_read(
 
         xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
-        if (likely(!(ioflags & IO_INVIS)))
-                xfs_ichgtime_fast(ip, inode, XFS_ICHGTIME_ACC);
-
 unlock_isem:
         if (unlikely(ioflags & IO_ISDIRECT))
                 mutex_unlock(&inode->i_mutex);
@@ -346,9 +343,6 @@ xfs_sendfile(
         if (ret > 0)
                 XFS_STATS_ADD(xs_read_bytes, ret);
 
-        if (likely(!(ioflags & IO_INVIS)))
-                xfs_ichgtime_fast(ip, LINVFS_GET_IP(vp), XFS_ICHGTIME_ACC);
-
         return ret;
 }
 
@@ -362,7 +356,6 @@ STATIC int				/* error (positive) */
 xfs_zero_last_block(
         struct inode    *ip,
         xfs_iocore_t    *io,
-        xfs_off_t       offset,
         xfs_fsize_t     isize,
         xfs_fsize_t     end_size)
 {
@@ -371,19 +364,16 @@ xfs_zero_last_block(
         int             nimaps;
         int             zero_offset;
         int             zero_len;
-        int             isize_fsb_offset;
         int             error = 0;
         xfs_bmbt_irec_t imap;
         loff_t          loff;
-        size_t          lsize;
 
         ASSERT(ismrlocked(io->io_lock, MR_UPDATE) != 0);
-        ASSERT(offset > isize);
 
         mp = io->io_mount;
 
-        isize_fsb_offset = XFS_B_FSB_OFFSET(mp, isize);
-        if (isize_fsb_offset == 0) {
+        zero_offset = XFS_B_FSB_OFFSET(mp, isize);
+        if (zero_offset == 0) {
                 /*
                  * There are no extra bytes in the last block on disk to
                  * zero, so return.
@@ -413,10 +403,8 @@ xfs_zero_last_block(
          */
         XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL| XFS_EXTSIZE_RD);
         loff = XFS_FSB_TO_B(mp, last_fsb);
-        lsize = XFS_FSB_TO_B(mp, 1);
 
-        zero_offset = isize_fsb_offset;
-        zero_len = mp->m_sb.sb_blocksize - isize_fsb_offset;
+        zero_len = mp->m_sb.sb_blocksize - zero_offset;
 
         error = xfs_iozero(ip, loff + zero_offset, zero_len, end_size);
 
@@ -447,20 +435,17 @@ xfs_zero_eof(
         struct inode    *ip = LINVFS_GET_IP(vp);
         xfs_fileoff_t   start_zero_fsb;
         xfs_fileoff_t   end_zero_fsb;
-        xfs_fileoff_t   prev_zero_fsb;
         xfs_fileoff_t   zero_count_fsb;
         xfs_fileoff_t   last_fsb;
         xfs_extlen_t    buf_len_fsb;
-        xfs_extlen_t    prev_zero_count;
         xfs_mount_t     *mp;
         int             nimaps;
         int             error = 0;
         xfs_bmbt_irec_t imap;
-        loff_t          loff;
-        size_t          lsize;
 
         ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
         ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
+        ASSERT(offset > isize);
 
         mp = io->io_mount;
 
@@ -468,7 +453,7 @@ xfs_zero_eof(
          * First handle zeroing the block on which isize resides.
          * We only zero a part of that block so it is handled specially.
          */
-        error = xfs_zero_last_block(ip, io, offset, isize, end_size);
+        error = xfs_zero_last_block(ip, io, isize, end_size);
         if (error) {
                 ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
                 ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
@@ -496,8 +481,6 @@ xfs_zero_eof(
         }
 
         ASSERT(start_zero_fsb <= end_zero_fsb);
-        prev_zero_fsb = NULLFILEOFF;
-        prev_zero_count = 0;
         while (start_zero_fsb <= end_zero_fsb) {
                 nimaps = 1;
                 zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
@@ -519,10 +502,7 @@ xfs_zero_eof(
                          * that sits on a hole and sets the page as P_HOLE
                          * and calls remapf if it is a mapped file.
                          */
-                        prev_zero_fsb = NULLFILEOFF;
-                        prev_zero_count = 0;
-                        start_zero_fsb = imap.br_startoff +
-                                         imap.br_blockcount;
+                        start_zero_fsb = imap.br_startoff + imap.br_blockcount;
                         ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
                         continue;
                 }
@@ -543,17 +523,15 @@ xfs_zero_eof(
                  */
                 XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
 
-                loff = XFS_FSB_TO_B(mp, start_zero_fsb);
-                lsize = XFS_FSB_TO_B(mp, buf_len_fsb);
-
-                error = xfs_iozero(ip, loff, lsize, end_size);
+                error = xfs_iozero(ip,
+                                   XFS_FSB_TO_B(mp, start_zero_fsb),
+                                   XFS_FSB_TO_B(mp, buf_len_fsb),
+                                   end_size);
 
                 if (error) {
                         goto out_lock;
                 }
 
-                prev_zero_fsb = start_zero_fsb;
-                prev_zero_count = buf_len_fsb;
                 start_zero_fsb = imap.br_startoff + buf_len_fsb;
                 ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
 
@@ -640,7 +618,7 @@ xfs_write(
                         (xip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
                                 mp->m_rtdev_targp : mp->m_ddev_targp;
 
-                if ((pos & target->pbr_smask) || (count & target->pbr_smask))
+                if ((pos & target->bt_smask) || (count & target->bt_smask))
                         return XFS_ERROR(-EINVAL);
 
                 if (!VN_CACHED(vp) && pos < i_size_read(inode))
@@ -831,6 +809,10 @@ retry:
                 goto retry;
         }
 
+        isize = i_size_read(inode);
+        if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize))
+                *offset = isize;
+
         if (*offset > xip->i_d.di_size) {
                 xfs_ilock(xip, XFS_ILOCK_EXCL);
                 if (*offset > xip->i_d.di_size) {
@@ -956,7 +938,7 @@ xfs_bdstrat_cb(struct xfs_buf *bp)
 
         mp = XFS_BUF_FSPRIVATE3(bp, xfs_mount_t *);
         if (!XFS_FORCED_SHUTDOWN(mp)) {
-                pagebuf_iorequest(bp);
+                xfs_buf_iorequest(bp);
                 return 0;
         } else {
                 xfs_buftrace("XFS__BDSTRAT IOERROR", bp);
@@ -1009,7 +991,7 @@ xfsbdstrat(
                  * if (XFS_BUF_IS_GRIO(bp)) {
                  */
 
-                pagebuf_iorequest(bp);
+                xfs_buf_iorequest(bp);
                 return 0;
         }
 
diff --git a/fs/xfs/linux-2.6/xfs_stats.c b/fs/xfs/linux-2.6/xfs_stats.c
index 6c40a74be7c8..8955720a2c6b 100644
--- a/fs/xfs/linux-2.6/xfs_stats.c
+++ b/fs/xfs/linux-2.6/xfs_stats.c
@@ -34,7 +34,7 @@ xfs_read_xfsstats(
         __uint64_t      xs_write_bytes = 0;
         __uint64_t      xs_read_bytes = 0;
 
-        static struct xstats_entry {
+        static const struct xstats_entry {
                 char    *desc;
                 int     endpoint;
         } xstats[] = {
diff --git a/fs/xfs/linux-2.6/xfs_stats.h b/fs/xfs/linux-2.6/xfs_stats.h
index 50027c4a5618..8ba7a2fa6c1d 100644
--- a/fs/xfs/linux-2.6/xfs_stats.h
+++ b/fs/xfs/linux-2.6/xfs_stats.h
@@ -109,15 +109,15 @@ struct xfsstats {
         __uint32_t              vn_remove;      /* # times vn_remove called */
         __uint32_t              vn_free;        /* # times vn_free called */
 #define XFSSTAT_END_BUF                 (XFSSTAT_END_VNODE_OPS+9)
-        __uint32_t              pb_get;
-        __uint32_t              pb_create;
-        __uint32_t              pb_get_locked;
-        __uint32_t              pb_get_locked_waited;
-        __uint32_t              pb_busy_locked;
-        __uint32_t              pb_miss_locked;
-        __uint32_t              pb_page_retries;
-        __uint32_t              pb_page_found;
-        __uint32_t              pb_get_read;
+        __uint32_t              xb_get;
+        __uint32_t              xb_create;
+        __uint32_t              xb_get_locked;
+        __uint32_t              xb_get_locked_waited;
+        __uint32_t              xb_busy_locked;
+        __uint32_t              xb_miss_locked;
+        __uint32_t              xb_page_retries;
+        __uint32_t              xb_page_found;
+        __uint32_t              xb_get_read;
 /* Extra precision counters */
         __uint64_t              xs_xstrat_bytes;
         __uint64_t              xs_write_bytes;
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 6116b5bf433e..f22e426d9e42 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -306,13 +306,15 @@ xfs_mountfs_check_barriers(xfs_mount_t *mp)
                 xfs_fs_cmn_err(CE_NOTE, mp,
                   "Disabling barriers, not supported with external log device");
                 mp->m_flags &= ~XFS_MOUNT_BARRIER;
+                return;
         }
 
-        if (mp->m_ddev_targp->pbr_bdev->bd_disk->queue->ordered ==
+        if (mp->m_ddev_targp->bt_bdev->bd_disk->queue->ordered ==
                                         QUEUE_ORDERED_NONE) {
                 xfs_fs_cmn_err(CE_NOTE, mp,
                   "Disabling barriers, not supported by the underlying device");
                 mp->m_flags &= ~XFS_MOUNT_BARRIER;
+                return;
         }
 
         error = xfs_barrier_test(mp);
@@ -320,6 +322,7 @@ xfs_mountfs_check_barriers(xfs_mount_t *mp)
                 xfs_fs_cmn_err(CE_NOTE, mp,
                   "Disabling barriers, trial barrier write failed");
                 mp->m_flags &= ~XFS_MOUNT_BARRIER;
+                return;
         }
 }
 
@@ -327,7 +330,7 @@ void
 xfs_blkdev_issue_flush(
         xfs_buftarg_t           *buftarg)
 {
-        blkdev_issue_flush(buftarg->pbr_bdev, NULL);
+        blkdev_issue_flush(buftarg->bt_bdev, NULL);
 }
 
 STATIC struct inode *
@@ -576,7 +579,7 @@ xfssyncd(
                 timeleft = schedule_timeout_interruptible(timeleft);
                 /* swsusp */
                 try_to_freeze();
-                if (kthread_should_stop())
+                if (kthread_should_stop() && list_empty(&vfsp->vfs_sync_list))
                         break;
 
                 spin_lock(&vfsp->vfs_sync_lock);
@@ -966,9 +969,9 @@ init_xfs_fs( void )
         if (error < 0)
                 goto undo_zones;
 
-        error = pagebuf_init();
+        error = xfs_buf_init();
         if (error < 0)
-                goto undo_pagebuf;
+                goto undo_buffers;
 
         vn_init();
         xfs_init();
@@ -982,9 +985,9 @@ init_xfs_fs( void )
         return 0;
 
 undo_register:
-        pagebuf_terminate();
+        xfs_buf_terminate();
 
-undo_pagebuf:
+undo_buffers:
         linvfs_destroy_zones();
 
 undo_zones:
@@ -998,7 +1001,7 @@ exit_xfs_fs( void )
         XFS_DM_EXIT(&xfs_fs_type);
         unregister_filesystem(&xfs_fs_type);
         xfs_cleanup();
-        pagebuf_terminate();
+        xfs_buf_terminate();
         linvfs_destroy_zones();
         ktrace_uninit();
 }
diff --git a/fs/xfs/linux-2.6/xfs_vnode.c b/fs/xfs/linux-2.6/xfs_vnode.c
index e9bbcb4d6243..260dd8415dd7 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.c
+++ b/fs/xfs/linux-2.6/xfs_vnode.c
@@ -106,7 +106,6 @@ vn_revalidate_core(
         inode->i_blocks     = vap->va_nblocks;
         inode->i_mtime      = vap->va_mtime;
         inode->i_ctime      = vap->va_ctime;
-        inode->i_atime      = vap->va_atime;
         inode->i_blksize    = vap->va_blocksize;
         if (vap->va_xflags & XFS_XFLAG_IMMUTABLE)
                 inode->i_flags |= S_IMMUTABLE;
diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h
index f2bbb327c081..0fe2419461d6 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.h
+++ b/fs/xfs/linux-2.6/xfs_vnode.h
@@ -566,6 +566,25 @@ static inline int VN_BAD(struct vnode *vp)
 }
 
 /*
+ * Extracting atime values in various formats
+ */
+static inline void vn_atime_to_bstime(struct vnode *vp, xfs_bstime_t *bs_atime)
+{
+        bs_atime->tv_sec = vp->v_inode.i_atime.tv_sec;
+        bs_atime->tv_nsec = vp->v_inode.i_atime.tv_nsec;
+}
+
+static inline void vn_atime_to_timespec(struct vnode *vp, struct timespec *ts)
+{
+        *ts = vp->v_inode.i_atime;
+}
+
+static inline void vn_atime_to_time_t(struct vnode *vp, time_t *tt)
+{
+        *tt = vp->v_inode.i_atime.tv_sec;
+}
+
+/*
  * Some useful predicates.
  */
 #define VN_MAPPED(vp)   mapping_mapped(LINVFS_GET_IP(vp)->i_mapping)
diff --git a/fs/xfs/quota/xfs_dquot_item.c b/fs/xfs/quota/xfs_dquot_item.c
index 2f69822344e5..2ec6b441849c 100644
--- a/fs/xfs/quota/xfs_dquot_item.c
+++ b/fs/xfs/quota/xfs_dquot_item.c
@@ -239,7 +239,7 @@ xfs_qm_dquot_logitem_pushbuf(
          * trying to duplicate our effort.
          */
         ASSERT(qip->qli_pushbuf_flag != 0);
-        ASSERT(qip->qli_push_owner == get_thread_id());
+        ASSERT(qip->qli_push_owner == current_pid());
 
         /*
          * If flushlock isn't locked anymore, chances are that the
@@ -333,7 +333,7 @@ xfs_qm_dquot_logitem_trylock(
                         qip->qli_pushbuf_flag = 1;
                         ASSERT(qip->qli_format.qlf_blkno == dqp->q_blkno);
 #ifdef DEBUG
-                        qip->qli_push_owner = get_thread_id();
+                        qip->qli_push_owner = current_pid();
 #endif
                         /*
                          * The dquot is left locked.
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index bb6991a7a617..7dcdd0640c32 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -1392,11 +1392,12 @@ xfs_qm_qino_alloc(
 {
         xfs_trans_t     *tp;
         int             error;
-        unsigned long s;
+        unsigned long   s;
         cred_t          zerocr;
+        xfs_inode_t     zeroino;
         int             committed;
 
-        tp = xfs_trans_alloc(mp,XFS_TRANS_QM_QINOCREATE);
+        tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QINOCREATE);
         if ((error = xfs_trans_reserve(tp,
                                       XFS_QM_QINOCREATE_SPACE_RES(mp),
                                       XFS_CREATE_LOG_RES(mp), 0,
@@ -1406,8 +1407,9 @@ xfs_qm_qino_alloc(
                 return (error);
         }
         memset(&zerocr, 0, sizeof(zerocr));
+        memset(&zeroino, 0, sizeof(zeroino));
 
-        if ((error = xfs_dir_ialloc(&tp, mp->m_rootip, S_IFREG, 1, 0,
+        if ((error = xfs_dir_ialloc(&tp, &zeroino, S_IFREG, 1, 0,
                                    &zerocr, 0, 1, ip, &committed))) {
                 xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES |
                                  XFS_TRANS_ABORT);
@@ -1918,9 +1920,7 @@ xfs_qm_quotacheck(
          * at this point (because we intentionally didn't in dqget_noattach).
          */
         if (error) {
-                xfs_qm_dqpurge_all(mp,
-                                   XFS_QMOPT_UQUOTA|XFS_QMOPT_GQUOTA|
-                                   XFS_QMOPT_PQUOTA|XFS_QMOPT_QUOTAOFF);
+                xfs_qm_dqpurge_all(mp, XFS_QMOPT_QUOTALL | XFS_QMOPT_QUOTAOFF);
                 goto error_return;
         }
         /*
@@ -2743,6 +2743,7 @@ xfs_qm_vop_dqattach_and_dqmod_newinode(
                 xfs_dqunlock(udqp);
                 ASSERT(ip->i_udquot == NULL);
                 ip->i_udquot = udqp;
+                ASSERT(XFS_IS_UQUOTA_ON(tp->t_mountp));
                 ASSERT(ip->i_d.di_uid == be32_to_cpu(udqp->q_core.d_id));
                 xfs_trans_mod_dquot(tp, udqp, XFS_TRANS_DQ_ICOUNT, 1);
         }
@@ -2752,7 +2753,10 @@ xfs_qm_vop_dqattach_and_dqmod_newinode(
                 xfs_dqunlock(gdqp);
                 ASSERT(ip->i_gdquot == NULL);
                 ip->i_gdquot = gdqp;
-                ASSERT(ip->i_d.di_gid == be32_to_cpu(gdqp->q_core.d_id));
+                ASSERT(XFS_IS_OQUOTA_ON(tp->t_mountp));
+                ASSERT((XFS_IS_GQUOTA_ON(tp->t_mountp) ?
+                        ip->i_d.di_gid : ip->i_d.di_projid) ==
+                                be32_to_cpu(gdqp->q_core.d_id));
                 xfs_trans_mod_dquot(tp, gdqp, XFS_TRANS_DQ_ICOUNT, 1);
         }
 }
diff --git a/fs/xfs/support/debug.c b/fs/xfs/support/debug.c
index bb6dc91ea261..b08b3d9345b7 100644
--- a/fs/xfs/support/debug.c
+++ b/fs/xfs/support/debug.c
@@ -27,45 +27,12 @@ static DEFINE_SPINLOCK(xfs_err_lock);
 /* Translate from CE_FOO to KERN_FOO, err_level(CE_FOO) == KERN_FOO */
 #define XFS_MAX_ERR_LEVEL       7
 #define XFS_ERR_MASK            ((1 << 3) - 1)
-static char             *err_level[XFS_MAX_ERR_LEVEL+1] =
+static const char * const       err_level[XFS_MAX_ERR_LEVEL+1] =
                                         {KERN_EMERG, KERN_ALERT, KERN_CRIT,
                                          KERN_ERR, KERN_WARNING, KERN_NOTICE,
                                          KERN_INFO, KERN_DEBUG};
 
 void
-assfail(char *a, char *f, int l)
-{
-    printk("XFS assertion failed: %s, file: %s, line: %d\n", a, f, l);
-    BUG();
-}
-
-#if ((defined(DEBUG) || defined(INDUCE_IO_ERRROR)) && !defined(NO_WANT_RANDOM))
-
-unsigned long
-random(void)
-{
-        static unsigned long    RandomValue = 1;
-        /* cycles pseudo-randomly through all values between 1 and 2^31 - 2 */
-        register long   rv = RandomValue;
-        register long   lo;
-        register long   hi;
-
-        hi = rv / 127773;
-        lo = rv % 127773;
-        rv = 16807 * lo - 2836 * hi;
-        if( rv <= 0 ) rv += 2147483647;
-        return( RandomValue = rv );
-}
-
-int
-get_thread_id(void)
-{
-        return current->pid;
-}
-
-#endif /* DEBUG || INDUCE_IO_ERRROR || !NO_WANT_RANDOM */
-
-void
 cmn_err(register int level, char *fmt, ...)
 {
         char    *fp = fmt;
@@ -90,7 +57,6 @@ cmn_err(register int level, char *fmt, ...)
                 BUG();
 }
 
-
 void
 icmn_err(register int level, char *fmt, va_list ap)
 {
@@ -109,3 +75,27 @@ icmn_err(register int level, char *fmt, va_list ap)
         if (level == CE_PANIC)
                 BUG();
 }
+
+void
+assfail(char *expr, char *file, int line)
+{
+        printk("Assertion failed: %s, file: %s, line: %d\n", expr, file, line);
+        BUG();
+}
+
+#if ((defined(DEBUG) || defined(INDUCE_IO_ERRROR)) && !defined(NO_WANT_RANDOM))
+unsigned long random(void)
+{
+        static unsigned long    RandomValue = 1;
+        /* cycles pseudo-randomly through all values between 1 and 2^31 - 2 */
+        register long   rv = RandomValue;
+        register long   lo;
+        register long   hi;
+
+        hi = rv / 127773;
+        lo = rv % 127773;
+        rv = 16807 * lo - 2836 * hi;
+        if (rv <= 0) rv += 2147483647;
+        return RandomValue = rv;
+}
+#endif /* DEBUG || INDUCE_IO_ERRROR || !NO_WANT_RANDOM */
diff --git a/fs/xfs/support/debug.h b/fs/xfs/support/debug.h
index aff558664c32..e3bf58112e7e 100644
--- a/fs/xfs/support/debug.h
+++ b/fs/xfs/support/debug.h
@@ -31,24 +31,23 @@ extern void icmn_err(int, char *, va_list)
         __attribute__ ((format (printf, 2, 0)));
 extern void cmn_err(int, char *, ...)
         __attribute__ ((format (printf, 2, 3)));
+extern void assfail(char *expr, char *f, int l);
 
-#ifndef STATIC
-# define STATIC static
-#endif
+#define prdev(fmt,targ,args...) \
+        printk("Device %s - " fmt "\n", XFS_BUFTARG_NAME(targ), ## args)
 
-#ifdef DEBUG
-# define ASSERT(EX)     ((EX) ? ((void)0) : assfail(#EX, __FILE__, __LINE__))
-#else
-# define ASSERT(x)      ((void)0)
-#endif
+#define ASSERT_ALWAYS(expr)     \
+        (unlikely((expr) != 0) ? (void)0 : assfail(#expr, __FILE__, __LINE__))
 
-extern void assfail(char *, char *, int);
-#ifdef DEBUG
+#ifndef DEBUG
+# define ASSERT(expr)   ((void)0)
+#else
+# define ASSERT(expr)   ASSERT_ALWAYS(expr)
 extern unsigned long random(void);
-extern int get_thread_id(void);
 #endif
 
-#define ASSERT_ALWAYS(EX)  ((EX)?((void)0):assfail(#EX, __FILE__, __LINE__))
-#define debug_stop_all_cpus(param)      /* param is "cpumask_t *" */
+#ifndef STATIC
+# define STATIC static
+#endif
 
 #endif  /* __XFS_SUPPORT_DEBUG_H__ */
diff --git a/fs/xfs/support/uuid.c b/fs/xfs/support/uuid.c
index 69ec4f540c3a..a3d565a67734 100644
--- a/fs/xfs/support/uuid.c
+++ b/fs/xfs/support/uuid.c
@@ -27,6 +27,16 @@ uuid_init(void)
         mutex_init(&uuid_monitor);
 }
 
+
+/* IRIX interpretation of an uuid_t */
+typedef struct {
+        __be32  uu_timelow;
+        __be16  uu_timemid;
+        __be16  uu_timehi;
+        __be16  uu_clockseq;
+        __be16  uu_node[3];
+} xfs_uu_t;
+
 /*
  * uuid_getnodeuniq - obtain the node unique fields of a UUID.
  *
@@ -36,16 +46,11 @@ uuid_init(void)
 void
 uuid_getnodeuniq(uuid_t *uuid, int fsid [2])
 {
-        char    *uu = (char *)uuid;
-
-        /* on IRIX, this function assumes big-endian fields within
-         * the uuid, so we use INT_GET to get the same result on
-         * little-endian systems
-         */
+        xfs_uu_t *uup = (xfs_uu_t *)uuid;
 
-        fsid[0] = (INT_GET(*(u_int16_t*)(uu+8), ARCH_CONVERT) << 16) +
-                   INT_GET(*(u_int16_t*)(uu+4), ARCH_CONVERT);
-        fsid[1] =  INT_GET(*(u_int32_t*)(uu  ), ARCH_CONVERT);
+        fsid[0] = (be16_to_cpu(uup->uu_clockseq) << 16) |
+                   be16_to_cpu(uup->uu_timemid);
+        fsid[1] = be16_to_cpu(uup->uu_timelow);
 }
 
 void
diff --git a/fs/xfs/xfs_arch.h b/fs/xfs/xfs_arch.h
index 68e5051d8e24..c4836890b726 100644
--- a/fs/xfs/xfs_arch.h
+++ b/fs/xfs/xfs_arch.h
@@ -40,6 +40,22 @@
 #undef XFS_NATIVE_HOST
 #endif
 
+#ifdef XFS_NATIVE_HOST
+#define cpu_to_be16(val)        ((__be16)(val))
+#define cpu_to_be32(val)        ((__be32)(val))
+#define cpu_to_be64(val)        ((__be64)(val))
+#define be16_to_cpu(val)        ((__uint16_t)(val))
+#define be32_to_cpu(val)        ((__uint32_t)(val))
+#define be64_to_cpu(val)        ((__uint64_t)(val))
+#else
+#define cpu_to_be16(val)        (__swab16((__uint16_t)(val)))
+#define cpu_to_be32(val)        (__swab32((__uint32_t)(val)))
+#define cpu_to_be64(val)        (__swab64((__uint64_t)(val)))
+#define be16_to_cpu(val)        (__swab16((__be16)(val)))
+#define be32_to_cpu(val)        (__swab32((__be32)(val)))
+#define be64_to_cpu(val)        (__swab64((__be64)(val)))
+#endif
+
 #endif  /* __KERNEL__ */
 
 /* do we need conversion? */
@@ -186,7 +202,7 @@ static inline void be64_add(__be64 *a, __s64 b)
  */ 
 
 #define XFS_GET_DIR_INO4(di) \
-        (((u32)(di).i[0] << 24) | ((di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
+        (((__u32)(di).i[0] << 24) | ((di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
 
 #define XFS_PUT_DIR_INO4(from, di) \
 do { \
@@ -197,9 +213,9 @@ do { \
 } while (0)
 
 #define XFS_DI_HI(di) \
-        (((u32)(di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
+        (((__u32)(di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
 #define XFS_DI_LO(di) \
-        (((u32)(di).i[4] << 24) | ((di).i[5] << 16) | ((di).i[6] << 8) | ((di).i[7]))
+        (((__u32)(di).i[4] << 24) | ((di).i[5] << 16) | ((di).i[6] << 8) | ((di).i[7]))
 
 #define XFS_GET_DIR_INO8(di)        \
         (((xfs_ino_t)XFS_DI_LO(di) & 0xffffffffULL) | \
diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c
index 1c7421840c18..fe91eac4e2a7 100644
--- a/fs/xfs/xfs_attr_leaf.c
+++ b/fs/xfs/xfs_attr_leaf.c
@@ -128,7 +128,7 @@ xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes)
                 return (offset >= minforkoff) ? minforkoff : 0;
         }
 
-        if (unlikely(mp->m_flags & XFS_MOUNT_COMPAT_ATTR)) {
+        if (!(mp->m_flags & XFS_MOUNT_ATTR2)) {
                 if (bytes <= XFS_IFORK_ASIZE(dp))
                         return mp->m_attroffset >> 3;
                 return 0;
@@ -157,7 +157,7 @@ xfs_sbversion_add_attr2(xfs_mount_t *mp, xfs_trans_t *tp)
 {
         unsigned long s;
 
-        if (!(mp->m_flags & XFS_MOUNT_COMPAT_ATTR) &&
+        if ((mp->m_flags & XFS_MOUNT_ATTR2) &&
             !(XFS_SB_VERSION_HASATTR2(&mp->m_sb))) {
                 s = XFS_SB_LOCK(mp);
                 if (!XFS_SB_VERSION_HASATTR2(&mp->m_sb)) {
@@ -311,7 +311,7 @@ xfs_attr_shortform_remove(xfs_da_args_t *args)
          */
         totsize -= size;
         if (totsize == sizeof(xfs_attr_sf_hdr_t) && !args->addname &&
-            !(mp->m_flags & XFS_MOUNT_COMPAT_ATTR)) {
+            (mp->m_flags & XFS_MOUNT_ATTR2)) {
                 /*
                  * Last attribute now removed, revert to original
                  * inode format making all literal area available
@@ -330,7 +330,7 @@ xfs_attr_shortform_remove(xfs_da_args_t *args)
                 dp->i_d.di_forkoff = xfs_attr_shortform_bytesfit(dp, totsize);
                 ASSERT(dp->i_d.di_forkoff);
                 ASSERT(totsize > sizeof(xfs_attr_sf_hdr_t) || args->addname ||
-                        (mp->m_flags & XFS_MOUNT_COMPAT_ATTR));
+                        !(mp->m_flags & XFS_MOUNT_ATTR2));
                 dp->i_afp->if_ext_max =
                         XFS_IFORK_ASIZE(dp) / (uint)sizeof(xfs_bmbt_rec_t);
                 dp->i_df.if_ext_max =
@@ -739,7 +739,7 @@ xfs_attr_shortform_allfit(xfs_dabuf_t *bp, xfs_inode_t *dp)
                                 + name_loc->namelen
                                 + INT_GET(name_loc->valuelen, ARCH_CONVERT);
         }
-        if (!(dp->i_mount->m_flags & XFS_MOUNT_COMPAT_ATTR) &&
+        if ((dp->i_mount->m_flags & XFS_MOUNT_ATTR2) &&
             (bytes == sizeof(struct xfs_attr_sf_hdr)))
                 return(-1);
         return(xfs_attr_shortform_bytesfit(dp, bytes));
@@ -778,7 +778,7 @@ xfs_attr_leaf_to_shortform(xfs_dabuf_t *bp, xfs_da_args_t *args, int forkoff)
                 goto out;
 
         if (forkoff == -1) {
-                ASSERT(!(dp->i_mount->m_flags & XFS_MOUNT_COMPAT_ATTR));
+                ASSERT(dp->i_mount->m_flags & XFS_MOUNT_ATTR2);
 
                 /*
                  * Last attribute was removed, revert to original
diff --git a/fs/xfs/xfs_attr_leaf.h b/fs/xfs/xfs_attr_leaf.h
index f6143ff251a0..541e34109bb9 100644
--- a/fs/xfs/xfs_attr_leaf.h
+++ b/fs/xfs/xfs_attr_leaf.h
@@ -63,7 +63,7 @@ struct xfs_trans;
  * the leaf_entry.  The namespaces are independent only because we also look
  * at the namespace bit when we are looking for a matching attribute name.
  *
- * We also store a "incomplete" bit in the leaf_entry.  It shows that an
+ * We also store an "incomplete" bit in the leaf_entry.  It shows that an
  * attribute is in the middle of being created and should not be shown to
  * the user if we crash during the time that the bit is set.  We clear the
  * bit when we have finished setting up the attribute.  We do this because
@@ -72,42 +72,48 @@ struct xfs_trans;
  */
 #define XFS_ATTR_LEAF_MAPSIZE   3       /* how many freespace slots */
 
+typedef struct xfs_attr_leaf_map {      /* RLE map of free bytes */
+        __uint16_t      base;           /* base of free region */
+        __uint16_t      size;           /* length of free region */
+} xfs_attr_leaf_map_t;
+
+typedef struct xfs_attr_leaf_hdr {      /* constant-structure header block */
+        xfs_da_blkinfo_t info;          /* block type, links, etc. */
+        __uint16_t      count;          /* count of active leaf_entry's */
+        __uint16_t      usedbytes;      /* num bytes of names/values stored */
+        __uint16_t      firstused;      /* first used byte in name area */
+        __uint8_t       holes;          /* != 0 if blk needs compaction */
+        __uint8_t       pad1;
+        xfs_attr_leaf_map_t freemap[XFS_ATTR_LEAF_MAPSIZE];
+                                        /* N largest free regions */
+} xfs_attr_leaf_hdr_t;
+
+typedef struct xfs_attr_leaf_entry {    /* sorted on key, not name */
+        xfs_dahash_t    hashval;        /* hash value of name */
+        __uint16_t      nameidx;        /* index into buffer of name/value */
+        __uint8_t       flags;          /* LOCAL/ROOT/SECURE/INCOMPLETE flag */
+        __uint8_t       pad2;           /* unused pad byte */
+} xfs_attr_leaf_entry_t;
+
+typedef struct xfs_attr_leaf_name_local {
+        __uint16_t      valuelen;       /* number of bytes in value */
+        __uint8_t       namelen;        /* length of name bytes */
+        __uint8_t       nameval[1];     /* name/value bytes */
+} xfs_attr_leaf_name_local_t;
+
+typedef struct xfs_attr_leaf_name_remote {
+        xfs_dablk_t     valueblk;       /* block number of value bytes */
+        __uint32_t      valuelen;       /* number of bytes in value */
+        __uint8_t       namelen;        /* length of name bytes */
+        __uint8_t       name[1];        /* name bytes */
+} xfs_attr_leaf_name_remote_t;
+
 typedef struct xfs_attr_leafblock {
-        struct xfs_attr_leaf_hdr {      /* constant-structure header block */
-                xfs_da_blkinfo_t info;  /* block type, links, etc. */
-                __uint16_t count;       /* count of active leaf_entry's */
-                __uint16_t usedbytes;   /* num bytes of names/values stored */
-                __uint16_t firstused;   /* first used byte in name area */
-                __uint8_t  holes;       /* != 0 if blk needs compaction */
-                __uint8_t  pad1;
-                struct xfs_attr_leaf_map {        /* RLE map of free bytes */
-                        __uint16_t base;          /* base of free region */
-                        __uint16_t size;          /* length of free region */
-                } freemap[XFS_ATTR_LEAF_MAPSIZE]; /* N largest free regions */
-        } hdr;
-        struct xfs_attr_leaf_entry {    /* sorted on key, not name */
-                xfs_dahash_t hashval;   /* hash value of name */
-                __uint16_t nameidx;     /* index into buffer of name/value */
-                __uint8_t flags;        /* LOCAL/ROOT/SECURE/INCOMPLETE flag */
-                __uint8_t pad2;         /* unused pad byte */
-        } entries[1];                   /* variable sized array */
-        struct xfs_attr_leaf_name_local {
-                __uint16_t valuelen;    /* number of bytes in value */
-                __uint8_t namelen;      /* length of name bytes */
-                __uint8_t nameval[1];   /* name/value bytes */
-        } namelist;                     /* grows from bottom of buf */
-        struct xfs_attr_leaf_name_remote {
-                xfs_dablk_t valueblk;   /* block number of value bytes */
-                __uint32_t valuelen;    /* number of bytes in value */
-                __uint8_t namelen;      /* length of name bytes */
-                __uint8_t name[1];      /* name bytes */
-        } valuelist;                    /* grows from bottom of buf */
+        xfs_attr_leaf_hdr_t     hdr;    /* constant-structure header block */
+        xfs_attr_leaf_entry_t   entries[1];     /* sorted on key, not name */
+        xfs_attr_leaf_name_local_t namelist;    /* grows from bottom of buf */
+        xfs_attr_leaf_name_remote_t valuelist;  /* grows from bottom of buf */
 } xfs_attr_leafblock_t;
-typedef struct xfs_attr_leaf_hdr xfs_attr_leaf_hdr_t;
-typedef struct xfs_attr_leaf_map xfs_attr_leaf_map_t;
-typedef struct xfs_attr_leaf_entry xfs_attr_leaf_entry_t;
-typedef struct xfs_attr_leaf_name_local xfs_attr_leaf_name_local_t;
-typedef struct xfs_attr_leaf_name_remote xfs_attr_leaf_name_remote_t;
 
 /*
  * Flags used in the leaf_entry[i].flags field.
@@ -150,7 +156,8 @@ xfs_attr_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
                 (leafp))[INT_GET((leafp)->entries[idx].nameidx, ARCH_CONVERT)];
 }
 
-#define XFS_ATTR_LEAF_NAME(leafp,idx)           xfs_attr_leaf_name(leafp,idx)
+#define XFS_ATTR_LEAF_NAME(leafp,idx)           \
+        xfs_attr_leaf_name(leafp,idx)
 static inline char *xfs_attr_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
 {
         return (&((char *)
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index e415a4698e9c..70625e577c70 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -2146,13 +2146,176 @@ xfs_bmap_add_extent_hole_real(
         return 0; /* keep gcc quite */
 }
 
+/*
+ * Adjust the size of the new extent based on di_extsize and rt extsize.
+ */
+STATIC int
+xfs_bmap_extsize_align(
+        xfs_mount_t     *mp,
+        xfs_bmbt_irec_t *gotp,          /* next extent pointer */
+        xfs_bmbt_irec_t *prevp,         /* previous extent pointer */
+        xfs_extlen_t    extsz,          /* align to this extent size */
+        int             rt,             /* is this a realtime inode? */
+        int             eof,            /* is extent at end-of-file? */
+        int             delay,          /* creating delalloc extent? */
+        int             convert,        /* overwriting unwritten extent? */
+        xfs_fileoff_t   *offp,          /* in/out: aligned offset */
+        xfs_extlen_t    *lenp)          /* in/out: aligned length */
+{
+        xfs_fileoff_t   orig_off;       /* original offset */
+        xfs_extlen_t    orig_alen;      /* original length */
+        xfs_fileoff_t   orig_end;       /* original off+len */
+        xfs_fileoff_t   nexto;          /* next file offset */
+        xfs_fileoff_t   prevo;          /* previous file offset */
+        xfs_fileoff_t   align_off;      /* temp for offset */
+        xfs_extlen_t    align_alen;     /* temp for length */
+        xfs_extlen_t    temp;           /* temp for calculations */
+
+        if (convert)
+                return 0;
+
+        orig_off = align_off = *offp;
+        orig_alen = align_alen = *lenp;
+        orig_end = orig_off + orig_alen;
+
+        /*
+         * If this request overlaps an existing extent, then don't
+         * attempt to perform any additional alignment.
+         */
+        if (!delay && !eof &&
+            (orig_off >= gotp->br_startoff) &&
+            (orig_end <= gotp->br_startoff + gotp->br_blockcount)) {
+                return 0;
+        }
+
+        /*
+         * If the file offset is unaligned vs. the extent size
+         * we need to align it.  This will be possible unless
+         * the file was previously written with a kernel that didn't
+         * perform this alignment, or if a truncate shot us in the
+         * foot.
+         */
+        temp = do_mod(orig_off, extsz);
+        if (temp) {
+                align_alen += temp;
+                align_off -= temp;
+        }
+        /*
+         * Same adjustment for the end of the requested area.
+         */
+        if ((temp = (align_alen % extsz))) {
+                align_alen += extsz - temp;
+        }
+        /*
+         * If the previous block overlaps with this proposed allocation
+         * then move the start forward without adjusting the length.
+         */
+        if (prevp->br_startoff != NULLFILEOFF) {
+                if (prevp->br_startblock == HOLESTARTBLOCK)
+                        prevo = prevp->br_startoff;
+                else
+                        prevo = prevp->br_startoff + prevp->br_blockcount;
+        } else
+                prevo = 0;
+        if (align_off != orig_off && align_off < prevo)
+                align_off = prevo;
+        /*
+         * If the next block overlaps with this proposed allocation
+         * then move the start back without adjusting the length,
+         * but not before offset 0.
+         * This may of course make the start overlap previous block,
+         * and if we hit the offset 0 limit then the next block
+         * can still overlap too.
+         */
+        if (!eof && gotp->br_startoff != NULLFILEOFF) {
+                if ((delay && gotp->br_startblock == HOLESTARTBLOCK) ||
+                    (!delay && gotp->br_startblock == DELAYSTARTBLOCK))
+                        nexto = gotp->br_startoff + gotp->br_blockcount;
+                else
+                        nexto = gotp->br_startoff;
+        } else
+                nexto = NULLFILEOFF;
+        if (!eof &&
+            align_off + align_alen != orig_end &&
+            align_off + align_alen > nexto)
+                align_off = nexto > align_alen ? nexto - align_alen : 0;
+        /*
+         * If we're now overlapping the next or previous extent that
+         * means we can't fit an extsz piece in this hole.  Just move
+         * the start forward to the first valid spot and set
+         * the length so we hit the end.
+         */
+        if (align_off != orig_off && align_off < prevo)
+                align_off = prevo;
+        if (align_off + align_alen != orig_end &&
+            align_off + align_alen > nexto &&
+            nexto != NULLFILEOFF) {
+                ASSERT(nexto > prevo);
+                align_alen = nexto - align_off;
+        }
+
+        /*
+         * If realtime, and the result isn't a multiple of the realtime
+         * extent size we need to remove blocks until it is.
+         */
+        if (rt && (temp = (align_alen % mp->m_sb.sb_rextsize))) {
+                /*
+                 * We're not covering the original request, or
+                 * we won't be able to once we fix the length.
+                 */
+                if (orig_off < align_off ||
+                    orig_end > align_off + align_alen ||
+                    align_alen - temp < orig_alen)
+                        return XFS_ERROR(EINVAL);
+                /*
+                 * Try to fix it by moving the start up.
+                 */
+                if (align_off + temp <= orig_off) {
+                        align_alen -= temp;
+                        align_off += temp;
+                }
+                /*
+                 * Try to fix it by moving the end in.
+                 */
+                else if (align_off + align_alen - temp >= orig_end)
+                        align_alen -= temp;
+                /*
+                 * Set the start to the minimum then trim the length.
+                 */
+                else {
+                        align_alen -= orig_off - align_off;
+                        align_off = orig_off;
+                        align_alen -= align_alen % mp->m_sb.sb_rextsize;
+                }
+                /*
+                 * Result doesn't cover the request, fail it.
+                 */
+                if (orig_off < align_off || orig_end > align_off + align_alen)
+                        return XFS_ERROR(EINVAL);
+        } else {
+                ASSERT(orig_off >= align_off);
+                ASSERT(orig_end <= align_off + align_alen);
+        }
+
+#ifdef DEBUG
+        if (!eof && gotp->br_startoff != NULLFILEOFF)
+                ASSERT(align_off + align_alen <= gotp->br_startoff);
+        if (prevp->br_startoff != NULLFILEOFF)
+                ASSERT(align_off >= prevp->br_startoff + prevp->br_blockcount);
+#endif
+
+        *lenp = align_alen;
+        *offp = align_off;
+        return 0;
+}
+
 #define XFS_ALLOC_GAP_UNITS     4
 
 /*
  * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file.
  * It figures out where to ask the underlying allocator to put the new extent.
  */
-STATIC int                              /* error */
+STATIC int
 xfs_bmap_alloc(
         xfs_bmalloca_t  *ap)            /* bmap alloc argument struct */
 {
@@ -2163,10 +2326,10 @@ xfs_bmap_alloc(
         xfs_mount_t     *mp;            /* mount point structure */
         int             nullfb;         /* true if ap->firstblock isn't set */
         int             rt;             /* true if inode is realtime */
-#ifdef __KERNEL__
-        xfs_extlen_t    prod=0;         /* product factor for allocators */
-        xfs_extlen_t    ralen=0;        /* realtime allocation length */
-#endif
+        xfs_extlen_t    prod = 0;       /* product factor for allocators */
+        xfs_extlen_t    ralen = 0;      /* realtime allocation length */
+        xfs_extlen_t    align;          /* minimum allocation alignment */
+        xfs_rtblock_t   rtx;
 
 #define ISVALID(x,y)    \
         (rt ? \
@@ -2182,125 +2345,25 @@ xfs_bmap_alloc(
         nullfb = ap->firstblock == NULLFSBLOCK;
         rt = XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata;
         fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock);
-#ifdef __KERNEL__
         if (rt) {
-                xfs_extlen_t    extsz;          /* file extent size for rt */
-                xfs_fileoff_t   nexto;          /* next file offset */
-                xfs_extlen_t    orig_alen;      /* original ap->alen */
-                xfs_fileoff_t   orig_end;       /* original off+len */
-                xfs_fileoff_t   orig_off;       /* original ap->off */
-                xfs_extlen_t    mod_off;        /* modulus calculations */
-                xfs_fileoff_t   prevo;          /* previous file offset */
-                xfs_rtblock_t   rtx;            /* realtime extent number */
-                xfs_extlen_t    temp;           /* temp for rt calculations */
-
-                /*
-                 * Set prod to match the realtime extent size.
-                 */
-                if (!(extsz = ap->ip->i_d.di_extsize))
-                        extsz = mp->m_sb.sb_rextsize;
-                prod = extsz / mp->m_sb.sb_rextsize;
-                orig_off = ap->off;
-                orig_alen = ap->alen;
-                orig_end = orig_off + orig_alen;
-                /*
-                 * If the file offset is unaligned vs. the extent size
-                 * we need to align it.  This will be possible unless
-                 * the file was previously written with a kernel that didn't
-                 * perform this alignment.
-                 */
-                mod_off = do_mod(orig_off, extsz);
-                if (mod_off) {
-                        ap->alen += mod_off;
-                        ap->off -= mod_off;
-                }
-                /*
-                 * Same adjustment for the end of the requested area.
-                 */
-                if ((temp = (ap->alen % extsz)))
-                        ap->alen += extsz - temp;
-                /*
-                 * If the previous block overlaps with this proposed allocation
-                 * then move the start forward without adjusting the length.
-                 */
-                prevo =
-                        ap->prevp->br_startoff == NULLFILEOFF ?
-                                0 :
-                                (ap->prevp->br_startoff +
-                                 ap->prevp->br_blockcount);
-                if (ap->off != orig_off && ap->off < prevo)
-                        ap->off = prevo;
-                /*
-                 * If the next block overlaps with this proposed allocation
-                 * then move the start back without adjusting the length,
-                 * but not before offset 0.
-                 * This may of course make the start overlap previous block,
-                 * and if we hit the offset 0 limit then the next block
-                 * can still overlap too.
-                 */
-                nexto = (ap->eof || ap->gotp->br_startoff == NULLFILEOFF) ?
-                        NULLFILEOFF : ap->gotp->br_startoff;
-                if (!ap->eof &&
-                    ap->off + ap->alen != orig_end &&
-                    ap->off + ap->alen > nexto)
-                        ap->off = nexto > ap->alen ? nexto - ap->alen : 0;
-                /*
-                 * If we're now overlapping the next or previous extent that
-                 * means we can't fit an extsz piece in this hole.  Just move
-                 * the start forward to the first valid spot and set
-                 * the length so we hit the end.
-                 */
-                if ((ap->off != orig_off && ap->off < prevo) ||
-                    (ap->off + ap->alen != orig_end &&
-                     ap->off + ap->alen > nexto)) {
-                        ap->off = prevo;
-                        ap->alen = nexto - prevo;
-                }
-                /*
-                 * If the result isn't a multiple of rtextents we need to
-                 * remove blocks until it is.
-                 */
-                if ((temp = (ap->alen % mp->m_sb.sb_rextsize))) {
-                        /*
-                         * We're not covering the original request, or
-                         * we won't be able to once we fix the length.
-                         */
-                        if (orig_off < ap->off ||
-                            orig_end > ap->off + ap->alen ||
-                            ap->alen - temp < orig_alen)
-                                return XFS_ERROR(EINVAL);
-                        /*
-                         * Try to fix it by moving the start up.
-                         */
-                        if (ap->off + temp <= orig_off) {
-                                ap->alen -= temp;
-                                ap->off += temp;
-                        }
-                        /*
-                         * Try to fix it by moving the end in.
-                         */
-                        else if (ap->off + ap->alen - temp >= orig_end)
-                                ap->alen -= temp;
-                        /*
-                         * Set the start to the minimum then trim the length.
-                         */
-                        else {
-                                ap->alen -= orig_off - ap->off;
-                                ap->off = orig_off;
-                                ap->alen -= ap->alen % mp->m_sb.sb_rextsize;
-                        }
-                        /*
-                         * Result doesn't cover the request, fail it.
-                         */
-                        if (orig_off < ap->off || orig_end > ap->off + ap->alen)
-                                return XFS_ERROR(EINVAL);
-                }
+                align = ap->ip->i_d.di_extsize ?
+                        ap->ip->i_d.di_extsize : mp->m_sb.sb_rextsize;
+                /* Set prod to match the extent size */
+                prod = align / mp->m_sb.sb_rextsize;
+
+                error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+                                                align, rt, ap->eof, 0,
+                                                ap->conv, &ap->off, &ap->alen);
+                if (error)
+                        return error;
+                ASSERT(ap->alen);
                 ASSERT(ap->alen % mp->m_sb.sb_rextsize == 0);
+
                 /*
                  * If the offset & length are not perfectly aligned
                  * then kill prod, it will just get us in trouble.
                  */
-                if (do_mod(ap->off, extsz) || ap->alen % extsz)
+                if (do_mod(ap->off, align) || ap->alen % align)
                         prod = 1;
                 /*
                  * Set ralen to be the actual requested length in rtextents.
@@ -2326,15 +2389,24 @@ xfs_bmap_alloc(
                         ap->rval = rtx * mp->m_sb.sb_rextsize;
                 } else
                         ap->rval = 0;
+        } else {
+                align = (ap->userdata && ap->ip->i_d.di_extsize &&
+                        (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)) ?
+                        ap->ip->i_d.di_extsize : 0;
+                if (unlikely(align)) {
+                        error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+                                                        align, rt,
+                                                        ap->eof, 0, ap->conv,
+                                                        &ap->off, &ap->alen);
+                        ASSERT(!error);
+                        ASSERT(ap->alen);
+                }
+                if (nullfb)
+                        ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
+                else
+                        ap->rval = ap->firstblock;
         }
-#else
-        if (rt)
-                ap->rval = 0;
-#endif  /* __KERNEL__ */
-        else if (nullfb)
-                ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
-        else
-                ap->rval = ap->firstblock;
+
         /*
          * If allocating at eof, and there's a previous real block,
          * try to use it's last block as our starting point.
@@ -2598,11 +2670,12 @@ xfs_bmap_alloc(
                         args.total = ap->total;
                         args.minlen = ap->minlen;
                 }
-                if (ap->ip->i_d.di_extsize) {
+                if (unlikely(ap->userdata && ap->ip->i_d.di_extsize &&
+                            (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE))) {
                         args.prod = ap->ip->i_d.di_extsize;
                         if ((args.mod = (xfs_extlen_t)do_mod(ap->off, args.prod)))
                                 args.mod = (xfs_extlen_t)(args.prod - args.mod);
-                } else if (mp->m_sb.sb_blocksize >= NBPP) {
+                } else if (unlikely(mp->m_sb.sb_blocksize >= NBPP)) {
                         args.prod = 1;
                         args.mod = 0;
                 } else {
@@ -3580,14 +3653,16 @@ xfs_bmap_search_extents(
 
         ep = xfs_bmap_do_search_extents(base, lastx, nextents, bno, eofp,
                                           lastxp, gotp, prevp);
-        rt = ip->i_d.di_flags & XFS_DIFLAG_REALTIME;
-        if(!rt && !gotp->br_startblock && (*lastxp != NULLEXTNUM)) {
+        rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+        if (unlikely(!rt && !gotp->br_startblock && (*lastxp != NULLEXTNUM))) {
                 cmn_err(CE_PANIC,"Access to block zero: fs: <%s> inode: %lld "
                         "start_block : %llx start_off : %llx blkcnt : %llx "
                         "extent-state : %x \n",
-                        (ip->i_mount)->m_fsname,(long long)ip->i_ino,
-                        gotp->br_startblock, gotp->br_startoff,
-                        gotp->br_blockcount,gotp->br_state);
+                        (ip->i_mount)->m_fsname, (long long)ip->i_ino,
+                        (unsigned long long)gotp->br_startblock,
+                        (unsigned long long)gotp->br_startoff,
+                        (unsigned long long)gotp->br_blockcount,
+                        gotp->br_state);
         }
         return ep;
 }
@@ -3875,7 +3950,7 @@ xfs_bmap_add_attrfork(
                 ip->i_d.di_forkoff = xfs_attr_shortform_bytesfit(ip, size);
                 if (!ip->i_d.di_forkoff)
                         ip->i_d.di_forkoff = mp->m_attroffset >> 3;
-                else if (!(mp->m_flags & XFS_MOUNT_COMPAT_ATTR))
+                else if (mp->m_flags & XFS_MOUNT_ATTR2)
                         version = 2;
                 break;
         default:
@@ -4023,13 +4098,13 @@ xfs_bmap_compute_maxlevels(
          */
         if (whichfork == XFS_DATA_FORK) {
                 maxleafents = MAXEXTNUM;
-                sz = (mp->m_flags & XFS_MOUNT_COMPAT_ATTR) ?
-                        mp->m_attroffset : XFS_BMDR_SPACE_CALC(MINDBTPTRS);
+                sz = (mp->m_flags & XFS_MOUNT_ATTR2) ?
+                        XFS_BMDR_SPACE_CALC(MINDBTPTRS) : mp->m_attroffset;
         } else {
                 maxleafents = MAXAEXTNUM;
-                sz = (mp->m_flags & XFS_MOUNT_COMPAT_ATTR) ?
-                        mp->m_sb.sb_inodesize - mp->m_attroffset :
-                        XFS_BMDR_SPACE_CALC(MINABTPTRS);
+                sz = (mp->m_flags & XFS_MOUNT_ATTR2) ?
+                        XFS_BMDR_SPACE_CALC(MINABTPTRS) :
+                        mp->m_sb.sb_inodesize - mp->m_attroffset;
         }
         maxrootrecs = (int)XFS_BTREE_BLOCK_MAXRECS(sz, xfs_bmdr, 0);
         minleafrecs = mp->m_bmap_dmnr[0];
@@ -4418,8 +4493,8 @@ xfs_bmap_read_extents(
                 num_recs = be16_to_cpu(block->bb_numrecs);
                 if (unlikely(i + num_recs > room)) {
                         ASSERT(i + num_recs <= room);
-                        xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                                "corrupt dinode %Lu, (btree extents).  Unmount and run xfs_repair.",
+                        xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                                "corrupt dinode %Lu, (btree extents).",
                                 (unsigned long long) ip->i_ino);
                         XFS_ERROR_REPORT("xfs_bmap_read_extents(1)",
                                          XFS_ERRLEVEL_LOW,
@@ -4590,6 +4665,7 @@ xfs_bmapi(
         char            contig;         /* allocation must be one extent */
         char            delay;          /* this request is for delayed alloc */
         char            exact;          /* don't do all of wasdelayed extent */
+        char            convert;        /* unwritten extent I/O completion */
         xfs_bmbt_rec_t  *ep;            /* extent list entry pointer */
         int             error;          /* error return */
         xfs_bmbt_irec_t got;            /* current extent list record */
@@ -4643,7 +4719,7 @@ xfs_bmapi(
         }
         if (XFS_FORCED_SHUTDOWN(mp))
                 return XFS_ERROR(EIO);
-        rt = XFS_IS_REALTIME_INODE(ip);
+        rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
         ifp = XFS_IFORK_PTR(ip, whichfork);
         ASSERT(ifp->if_ext_max ==
                XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
@@ -4654,6 +4730,7 @@ xfs_bmapi(
         delay = (flags & XFS_BMAPI_DELAY) != 0;
         trim = (flags & XFS_BMAPI_ENTIRE) == 0;
         userdata = (flags & XFS_BMAPI_METADATA) == 0;
+        convert = (flags & XFS_BMAPI_CONVERT) != 0;
         exact = (flags & XFS_BMAPI_EXACT) != 0;
         rsvd = (flags & XFS_BMAPI_RSVBLOCKS) != 0;
         contig = (flags & XFS_BMAPI_CONTIG) != 0;
@@ -4748,15 +4825,25 @@ xfs_bmapi(
                         }
                         minlen = contig ? alen : 1;
                         if (delay) {
-                                xfs_extlen_t    extsz = 0;
+                                xfs_extlen_t    extsz;
 
                                 /* Figure out the extent size, adjust alen */
                                 if (rt) {
                                         if (!(extsz = ip->i_d.di_extsize))
                                                 extsz = mp->m_sb.sb_rextsize;
-                                        alen = roundup(alen, extsz);
-                                        extsz = alen / mp->m_sb.sb_rextsize;
+                                } else {
+                                        extsz = ip->i_d.di_extsize;
                                 }
+                                if (extsz) {
+                                        error = xfs_bmap_extsize_align(mp,
+                                                        &got, &prev, extsz,
+                                                        rt, eof, delay, convert,
+                                                        &aoff, &alen);
+                                        ASSERT(!error);
+                                }
+
+                                if (rt)
+                                        extsz = alen / mp->m_sb.sb_rextsize;
 
                                 /*
                                  * Make a transaction-less quota reservation for
@@ -4785,32 +4872,33 @@ xfs_bmapi(
                                         xfs_bmap_worst_indlen(ip, alen);
                                 ASSERT(indlen > 0);
 
-                                if (rt)
+                                if (rt) {
                                         error = xfs_mod_incore_sb(mp,
                                                         XFS_SBS_FREXTENTS,
                                                         -(extsz), rsvd);
-                                else
+                                } else {
                                         error = xfs_mod_incore_sb(mp,
                                                         XFS_SBS_FDBLOCKS,
                                                         -(alen), rsvd);
+                                }
                                 if (!error) {
                                         error = xfs_mod_incore_sb(mp,
                                                         XFS_SBS_FDBLOCKS,
                                                         -(indlen), rsvd);
-                                        if (error && rt) {
-                                                xfs_mod_incore_sb(ip->i_mount,
+                                        if (error && rt)
+                                                xfs_mod_incore_sb(mp,
                                                         XFS_SBS_FREXTENTS,
                                                         extsz, rsvd);
-                                        } else if (error) {
-                                                xfs_mod_incore_sb(ip->i_mount,
+                                        else if (error)
+                                                xfs_mod_incore_sb(mp,
                                                         XFS_SBS_FDBLOCKS,
                                                         alen, rsvd);
-                                        }
                                 }
 
                                 if (error) {
-                                        if (XFS_IS_QUOTA_ON(ip->i_mount))
+                                        if (XFS_IS_QUOTA_ON(mp))
                                                 /* unreserve the blocks now */
+                                                (void)
                                                 XFS_TRANS_UNRESERVE_QUOTA_NBLKS(
                                                         mp, NULL, ip,
                                                         (long)alen, 0, rt ?
@@ -4849,6 +4937,7 @@ xfs_bmapi(
                                 bma.firstblock = *firstblock;
                                 bma.alen = alen;
                                 bma.off = aoff;
+                                bma.conv = convert;
                                 bma.wasdel = wasdelay;
                                 bma.minlen = minlen;
                                 bma.low = flist->xbf_low;
@@ -5270,8 +5359,7 @@ xfs_bunmapi(
                 return 0;
         }
         XFS_STATS_INC(xs_blk_unmap);
-        isrt = (whichfork == XFS_DATA_FORK) &&
-               (ip->i_d.di_flags & XFS_DIFLAG_REALTIME);
+        isrt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
         start = bno;
         bno = start + len - 1;
         ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
@@ -5443,7 +5531,7 @@ xfs_bunmapi(
                 }
                 if (wasdel) {
                         ASSERT(STARTBLOCKVAL(del.br_startblock) > 0);
-                        /* Update realtim/data freespace, unreserve quota */
+                        /* Update realtime/data freespace, unreserve quota */
                         if (isrt) {
                                 xfs_filblks_t rtexts;
 
@@ -5451,14 +5539,14 @@ xfs_bunmapi(
                                 do_div(rtexts, mp->m_sb.sb_rextsize);
                                 xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS,
                                                 (int)rtexts, rsvd);
-                                XFS_TRANS_RESERVE_QUOTA_NBLKS(mp, NULL, ip,
-                                        -((long)del.br_blockcount), 0,
+                                (void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
+                                        NULL, ip, -((long)del.br_blockcount), 0,
                                         XFS_QMOPT_RES_RTBLKS);
                         } else {
                                 xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS,
                                                 (int)del.br_blockcount, rsvd);
-                                XFS_TRANS_RESERVE_QUOTA_NBLKS(mp, NULL, ip,
-                                        -((long)del.br_blockcount), 0,
+                                (void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
+                                        NULL, ip, -((long)del.br_blockcount), 0,
                                         XFS_QMOPT_RES_REGBLKS);
                         }
                         ip->i_delayed_blks -= del.br_blockcount;
@@ -5652,7 +5740,9 @@ xfs_getbmap(
                    ip->i_d.di_format != XFS_DINODE_FMT_LOCAL)
                 return XFS_ERROR(EINVAL);
         if (whichfork == XFS_DATA_FORK) {
-                if (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC) {
+                if ((ip->i_d.di_extsize && (ip->i_d.di_flags &
+                                (XFS_DIFLAG_REALTIME|XFS_DIFLAG_EXTSIZE))) ||
+                    ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC|XFS_DIFLAG_APPEND)){
                         prealloced = 1;
                         fixlen = XFS_MAXIOFFSET(mp);
                 } else {
diff --git a/fs/xfs/xfs_bmap.h b/fs/xfs/xfs_bmap.h
index 2e0717a01309..12cc63dfc2c4 100644
--- a/fs/xfs/xfs_bmap.h
+++ b/fs/xfs/xfs_bmap.h
@@ -62,6 +62,10 @@ typedef	struct xfs_bmap_free
 #define XFS_BMAPI_IGSTATE       0x200   /* Ignore state - */
                                         /* combine contig. space */
 #define XFS_BMAPI_CONTIG        0x400   /* must allocate only one extent */
+/*      XFS_BMAPI_DIRECT_IO     0x800   */
+#define XFS_BMAPI_CONVERT       0x1000  /* unwritten extent conversion - */
+                                        /* need write cache flushing and no */
+                                        /* additional allocation alignments */
 
 #define XFS_BMAPI_AFLAG(w)      xfs_bmapi_aflag(w)
 static inline int xfs_bmapi_aflag(int w)
@@ -101,7 +105,8 @@ typedef struct xfs_bmalloca {
         char                    wasdel; /* replacing a delayed allocation */
         char                    userdata;/* set if is user data */
         char                    low;    /* low on space, using seq'l ags */
-        char                    aeof;   /* allocated space at eof */
+        char                    aeof;   /* allocated space at eof */
+        char                    conv;   /* overwriting unwritten extents */
 } xfs_bmalloca_t;
 
 #ifdef __KERNEL__
diff --git a/fs/xfs/xfs_clnt.h b/fs/xfs/xfs_clnt.h
index 328a528b926d..f57cc9ac875e 100644
--- a/fs/xfs/xfs_clnt.h
+++ b/fs/xfs/xfs_clnt.h
@@ -57,7 +57,7 @@ struct xfs_mount_args {
 /*
  * XFS mount option flags -- args->flags1
  */
-#define XFSMNT_COMPAT_ATTR      0x00000001      /* do not use ATTR2 format */
+#define XFSMNT_ATTR2            0x00000001      /* allow ATTR2 EA format */
 #define XFSMNT_WSYNC            0x00000002      /* safe mode nfs mount
                                                  * compatible */
 #define XFSMNT_INO64            0x00000004      /* move inode numbers up
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 070259a4254c..c6191d00ad27 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -60,8 +60,6 @@ xfs_swapext(
         xfs_bstat_t     *sbp;
         struct file     *fp = NULL, *tfp = NULL;
         vnode_t         *vp, *tvp;
-        bhv_desc_t      *bdp, *tbdp;
-        vn_bhv_head_t   *bhp, *tbhp;
         static uint     lock_flags = XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL;
         int             ilf_fields, tilf_fields;
         int             error = 0;
@@ -90,13 +88,10 @@ xfs_swapext(
                 goto error0;
         }
 
-        bhp = VN_BHV_HEAD(vp);
-        bdp = vn_bhv_lookup(bhp, &xfs_vnodeops);
-        if (bdp == NULL) {
+        ip = xfs_vtoi(vp);
+        if (ip == NULL) {
                 error = XFS_ERROR(EBADF);
                 goto error0;
-        } else {
-                ip = XFS_BHVTOI(bdp);
         }
 
         if (((tfp = fget((int)sxp->sx_fdtmp)) == NULL) ||
@@ -105,13 +100,10 @@ xfs_swapext(
                 goto error0;
         }
 
-        tbhp = VN_BHV_HEAD(tvp);
-        tbdp = vn_bhv_lookup(tbhp, &xfs_vnodeops);
-        if (tbdp == NULL) {
+        tip = xfs_vtoi(tvp);
+        if (tip == NULL) {
                 error = XFS_ERROR(EBADF);
                 goto error0;
-        } else {
-                tip = XFS_BHVTOI(tbdp);
         }
 
         if (ip->i_mount != tip->i_mount) {
diff --git a/fs/xfs/xfs_dinode.h b/fs/xfs/xfs_dinode.h
index c5a0e537ff1a..79d0d9e1fbab 100644
--- a/fs/xfs/xfs_dinode.h
+++ b/fs/xfs/xfs_dinode.h
@@ -199,10 +199,16 @@ typedef enum xfs_dinode_fmt
 
 #define XFS_DFORK_DSIZE(dip,mp) \
         XFS_CFORK_DSIZE_DISK(&(dip)->di_core, mp)
+#define XFS_DFORK_DSIZE_HOST(dip,mp) \
+        XFS_CFORK_DSIZE(&(dip)->di_core, mp)
 #define XFS_DFORK_ASIZE(dip,mp) \
         XFS_CFORK_ASIZE_DISK(&(dip)->di_core, mp)
+#define XFS_DFORK_ASIZE_HOST(dip,mp) \
+        XFS_CFORK_ASIZE(&(dip)->di_core, mp)
 #define XFS_DFORK_SIZE(dip,mp,w) \
         XFS_CFORK_SIZE_DISK(&(dip)->di_core, mp, w)
+#define XFS_DFORK_SIZE_HOST(dip,mp,w) \
+        XFS_CFORK_SIZE(&(dip)->di_core, mp, w)
 
 #define XFS_DFORK_Q(dip)                    XFS_CFORK_Q_DISK(&(dip)->di_core)
 #define XFS_DFORK_BOFF(dip)                 XFS_CFORK_BOFF_DISK(&(dip)->di_core)
@@ -216,6 +222,7 @@ typedef enum xfs_dinode_fmt
 #define XFS_CFORK_FMT_SET(dcp,w,n) \
         ((w) == XFS_DATA_FORK ? \
                 ((dcp)->di_format = (n)) : ((dcp)->di_aformat = (n)))
+#define XFS_DFORK_FORMAT(dip,w) XFS_CFORK_FORMAT(&(dip)->di_core, w)
 
 #define XFS_CFORK_NEXTENTS_DISK(dcp,w) \
         ((w) == XFS_DATA_FORK ? \
@@ -223,13 +230,13 @@ typedef enum xfs_dinode_fmt
                 INT_GET((dcp)->di_anextents, ARCH_CONVERT))
 #define XFS_CFORK_NEXTENTS(dcp,w) \
         ((w) == XFS_DATA_FORK ? (dcp)->di_nextents : (dcp)->di_anextents)
+#define XFS_DFORK_NEXTENTS(dip,w) XFS_CFORK_NEXTENTS_DISK(&(dip)->di_core, w)
+#define XFS_DFORK_NEXTENTS_HOST(dip,w) XFS_CFORK_NEXTENTS(&(dip)->di_core, w)
 
 #define XFS_CFORK_NEXT_SET(dcp,w,n) \
         ((w) == XFS_DATA_FORK ? \
                 ((dcp)->di_nextents = (n)) : ((dcp)->di_anextents = (n)))
 
-#define XFS_DFORK_NEXTENTS(dip,w) XFS_CFORK_NEXTENTS_DISK(&(dip)->di_core, w)
-
 #define XFS_BUF_TO_DINODE(bp)   ((xfs_dinode_t *)XFS_BUF_PTR(bp))
 
 /*
@@ -246,8 +253,10 @@ typedef enum xfs_dinode_fmt
 #define XFS_DIFLAG_NOATIME_BIT   6      /* do not update atime */
 #define XFS_DIFLAG_NODUMP_BIT    7      /* do not dump */
 #define XFS_DIFLAG_RTINHERIT_BIT 8      /* create with realtime bit set */
-#define XFS_DIFLAG_PROJINHERIT_BIT  9   /* create with parents projid */
-#define XFS_DIFLAG_NOSYMLINKS_BIT  10   /* disallow symlink creation */
+#define XFS_DIFLAG_PROJINHERIT_BIT   9  /* create with parents projid */
+#define XFS_DIFLAG_NOSYMLINKS_BIT   10  /* disallow symlink creation */
+#define XFS_DIFLAG_EXTSIZE_BIT      11  /* inode extent size allocator hint */
+#define XFS_DIFLAG_EXTSZINHERIT_BIT 12  /* inherit inode extent size */
 #define XFS_DIFLAG_REALTIME      (1 << XFS_DIFLAG_REALTIME_BIT)
 #define XFS_DIFLAG_PREALLOC      (1 << XFS_DIFLAG_PREALLOC_BIT)
 #define XFS_DIFLAG_NEWRTBM       (1 << XFS_DIFLAG_NEWRTBM_BIT)
@@ -259,11 +268,14 @@ typedef enum xfs_dinode_fmt
 #define XFS_DIFLAG_RTINHERIT     (1 << XFS_DIFLAG_RTINHERIT_BIT)
 #define XFS_DIFLAG_PROJINHERIT   (1 << XFS_DIFLAG_PROJINHERIT_BIT)
 #define XFS_DIFLAG_NOSYMLINKS    (1 << XFS_DIFLAG_NOSYMLINKS_BIT)
+#define XFS_DIFLAG_EXTSIZE       (1 << XFS_DIFLAG_EXTSIZE_BIT)
+#define XFS_DIFLAG_EXTSZINHERIT  (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
 
 #define XFS_DIFLAG_ANY \
         (XFS_DIFLAG_REALTIME | XFS_DIFLAG_PREALLOC | XFS_DIFLAG_NEWRTBM | \
          XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
          XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
-         XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS)
+         XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
+         XFS_DIFLAG_EXTSZINHERIT)
 
 #endif  /* __XFS_DINODE_H__ */
diff --git a/fs/xfs/xfs_dir.c b/fs/xfs/xfs_dir.c
index 3dd30391f551..bb87d2a700a9 100644
--- a/fs/xfs/xfs_dir.c
+++ b/fs/xfs/xfs_dir.c
@@ -176,7 +176,7 @@ xfs_dir_mount(xfs_mount_t *mp)
         uint shortcount, leafcount, count;
 
         mp->m_dirversion = 1;
-        if (mp->m_flags & XFS_MOUNT_COMPAT_ATTR) {
+        if (!(mp->m_flags & XFS_MOUNT_ATTR2)) {
                 shortcount = (mp->m_attroffset -
                                 (uint)sizeof(xfs_dir_sf_hdr_t)) /
                                  (uint)sizeof(xfs_dir_sf_entry_t);
diff --git a/fs/xfs/xfs_dir.h b/fs/xfs/xfs_dir.h
index 488defe86ba6..8cc8afb9f6c0 100644
--- a/fs/xfs/xfs_dir.h
+++ b/fs/xfs/xfs_dir.h
@@ -135,6 +135,8 @@ void	xfs_dir_startup(void);	/* called exactly once */
         ((mp)->m_dirops.xd_shortform_to_single(args))
 
 #define XFS_DIR_IS_V1(mp)       ((mp)->m_dirversion == 1)
+#define XFS_DIR_IS_V2(mp)       ((mp)->m_dirversion == 2)
 extern xfs_dirops_t xfsv1_dirops;
+extern xfs_dirops_t xfsv2_dirops;
 
 #endif  /* __XFS_DIR_H__ */
diff --git a/fs/xfs/xfs_dir2.h b/fs/xfs/xfs_dir2.h
index 7e24ffeda9e1..3158f5dc431f 100644
--- a/fs/xfs/xfs_dir2.h
+++ b/fs/xfs/xfs_dir2.h
@@ -72,9 +72,6 @@ typedef struct xfs_dir2_put_args {
         struct uio      *uio;           /* uio control structure */
 } xfs_dir2_put_args_t;
 
-#define XFS_DIR_IS_V2(mp)       ((mp)->m_dirversion == 2)
-extern xfs_dirops_t     xfsv2_dirops;
-
 /*
  * Other interfaces used by the rest of the dir v2 code.
  */
diff --git a/fs/xfs/xfs_dir_leaf.h b/fs/xfs/xfs_dir_leaf.h
index ab6b09eef9ab..eb8cd9a4667f 100644
--- a/fs/xfs/xfs_dir_leaf.h
+++ b/fs/xfs/xfs_dir_leaf.h
@@ -67,34 +67,38 @@ struct xfs_trans;
  */
 #define XFS_DIR_LEAF_MAPSIZE    3       /* how many freespace slots */
 
+typedef struct xfs_dir_leaf_map {       /* RLE map of free bytes */
+        __uint16_t      base;           /* base of free region */
+        __uint16_t      size;           /* run length of free region */
+} xfs_dir_leaf_map_t;
+
+typedef struct xfs_dir_leaf_hdr {       /* constant-structure header block */
+        xfs_da_blkinfo_t info;          /* block type, links, etc. */
+        __uint16_t      count;          /* count of active leaf_entry's */
+        __uint16_t      namebytes;      /* num bytes of name strings stored */
+        __uint16_t      firstused;      /* first used byte in name area */
+        __uint8_t       holes;          /* != 0 if blk needs compaction */
+        __uint8_t       pad1;
+        xfs_dir_leaf_map_t freemap[XFS_DIR_LEAF_MAPSIZE];
+} xfs_dir_leaf_hdr_t;
+
+typedef struct xfs_dir_leaf_entry {     /* sorted on key, not name */
+        xfs_dahash_t    hashval;        /* hash value of name */
+        __uint16_t      nameidx;        /* index into buffer of name */
+        __uint8_t       namelen;        /* length of name string */
+        __uint8_t       pad2;
+} xfs_dir_leaf_entry_t;
+
+typedef struct xfs_dir_leaf_name {
+        xfs_dir_ino_t   inumber;        /* inode number for this key */
+        __uint8_t       name[1];        /* name string itself */
+} xfs_dir_leaf_name_t;
+
 typedef struct xfs_dir_leafblock {
-        struct xfs_dir_leaf_hdr {       /* constant-structure header block */
-                xfs_da_blkinfo_t info;  /* block type, links, etc. */
-                __uint16_t count;       /* count of active leaf_entry's */
-                __uint16_t namebytes;   /* num bytes of name strings stored */
-                __uint16_t firstused;   /* first used byte in name area */
-                __uint8_t  holes;       /* != 0 if blk needs compaction */
-                __uint8_t  pad1;
-                struct xfs_dir_leaf_map {/* RLE map of free bytes */
-                        __uint16_t base; /* base of free region */
-                        __uint16_t size; /* run length of free region */
-                } freemap[XFS_DIR_LEAF_MAPSIZE]; /* N largest free regions */
-        } hdr;
-        struct xfs_dir_leaf_entry {     /* sorted on key, not name */
-                xfs_dahash_t hashval;   /* hash value of name */
-                __uint16_t nameidx;     /* index into buffer of name */
-                __uint8_t namelen;      /* length of name string */
-                __uint8_t pad2;
-        } entries[1];                   /* var sized array */
-        struct xfs_dir_leaf_name {
-                xfs_dir_ino_t inumber;  /* inode number for this key */
-                __uint8_t name[1];      /* name string itself */
-        } namelist[1];                  /* grows from bottom of buf */
+        xfs_dir_leaf_hdr_t      hdr;    /* constant-structure header block */
+        xfs_dir_leaf_entry_t    entries[1];     /* var sized array */
+        xfs_dir_leaf_name_t     namelist[1];    /* grows from bottom of buf */
 } xfs_dir_leafblock_t;
-typedef struct xfs_dir_leaf_hdr xfs_dir_leaf_hdr_t;
-typedef struct xfs_dir_leaf_map xfs_dir_leaf_map_t;
-typedef struct xfs_dir_leaf_entry xfs_dir_leaf_entry_t;
-typedef struct xfs_dir_leaf_name xfs_dir_leaf_name_t;
 
 /*
  * Length of name for which a 512-byte block filesystem
@@ -126,11 +130,10 @@ typedef union {
 #define XFS_PUT_COOKIE(c,mp,bno,entry,hash)     \
         ((c).s.be = XFS_DA_MAKE_BNOENTRY(mp, bno, entry), (c).s.h = (hash))
 
-typedef struct xfs_dir_put_args
-{
+typedef struct xfs_dir_put_args {
         xfs_dircook_t   cook;           /* cookie of (next) entry */
         xfs_intino_t    ino;            /* inode number */
-        struct xfs_dirent       *dbp;           /* buffer pointer */
+        struct xfs_dirent *dbp;         /* buffer pointer */
         char            *name;          /* directory entry name */
         int             namelen;        /* length of name */
         int             done;           /* output: set if value was stored */
@@ -138,7 +141,8 @@ typedef struct xfs_dir_put_args
         struct uio      *uio;           /* uio control structure */
 } xfs_dir_put_args_t;
 
-#define XFS_DIR_LEAF_ENTSIZE_BYNAME(len)        xfs_dir_leaf_entsize_byname(len)
+#define XFS_DIR_LEAF_ENTSIZE_BYNAME(len)        \
+        xfs_dir_leaf_entsize_byname(len)
 static inline int xfs_dir_leaf_entsize_byname(int len)
 {
         return (uint)sizeof(xfs_dir_leaf_name_t)-1 + len;
diff --git a/fs/xfs/xfs_error.c b/fs/xfs/xfs_error.c
index d7b6b5d16704..2a21c5024017 100644
--- a/fs/xfs/xfs_error.c
+++ b/fs/xfs/xfs_error.c
@@ -54,7 +54,6 @@ xfs_error_trap(int e)
                 if (e != xfs_etrap[i])
                         continue;
                 cmn_err(CE_NOTE, "xfs_error_trap: error %d", e);
-                debug_stop_all_cpus((void *)-1LL);
                 BUG();
                 break;
         }
diff --git a/fs/xfs/xfs_error.h b/fs/xfs/xfs_error.h
index 06d8a8426c16..26b8e709a569 100644
--- a/fs/xfs/xfs_error.h
+++ b/fs/xfs/xfs_error.h
@@ -18,9 +18,6 @@
 #ifndef __XFS_ERROR_H__
 #define __XFS_ERROR_H__
 
-#define prdev(fmt,targ,args...) \
-        printk("XFS: device %s - " fmt "\n", XFS_BUFTARG_NAME(targ), ## args)
-
 #define XFS_ERECOVER    1       /* Failure to recover log */
 #define XFS_ELOGSTAT    2       /* Failure to stat log in user space */
 #define XFS_ENOLOGSPACE 3       /* Reservation too large */
@@ -182,8 +179,11 @@ extern int xfs_errortag_clearall_umount(int64_t fsid, char *fsname, int loud);
 struct xfs_mount;
 /* PRINTFLIKE4 */
 extern void xfs_cmn_err(int panic_tag, int level, struct xfs_mount *mp,
-                            char *fmt, ...);
+                        char *fmt, ...);
 /* PRINTFLIKE3 */
 extern void xfs_fs_cmn_err(int level, struct xfs_mount *mp, char *fmt, ...);
 
+#define xfs_fs_repair_cmn_err(level, mp, fmt, args...) \
+        xfs_fs_cmn_err(level, mp, fmt "  Unmount and run xfs_repair.", ## args)
+
 #endif  /* __XFS_ERROR_H__ */
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
index ba096f80f48d..14010f1fa82f 100644
--- a/fs/xfs/xfs_fs.h
+++ b/fs/xfs/xfs_fs.h
@@ -3,15 +3,15 @@
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
+ * modify it under the terms of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
  *
  * This program is distributed in the hope that it would be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
+ * GNU Lesser General Public License for more details.
  *
- * You should have received a copy of the GNU General Public License
+ * You should have received a copy of the GNU Lesser General Public License
  * along with this program; if not, write the Free Software Foundation,
  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  */
@@ -65,6 +65,8 @@ struct fsxattr {
 #define XFS_XFLAG_RTINHERIT     0x00000100      /* create with rt bit set */
 #define XFS_XFLAG_PROJINHERIT   0x00000200      /* create with parents projid */
 #define XFS_XFLAG_NOSYMLINKS    0x00000400      /* disallow symlink creation */
+#define XFS_XFLAG_EXTSIZE       0x00000800      /* extent size allocator hint */
+#define XFS_XFLAG_EXTSZINHERIT  0x00001000      /* inherit inode extent size */
 #define XFS_XFLAG_HASATTR       0x80000000      /* no DIFLAG for this   */
 
 /*
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
index d1236d6f4045..163031c1e394 100644
--- a/fs/xfs/xfs_fsops.c
+++ b/fs/xfs/xfs_fsops.c
@@ -540,6 +540,32 @@ xfs_reserve_blocks(
         return(0);
 }
 
+void
+xfs_fs_log_dummy(xfs_mount_t *mp)
+{
+        xfs_trans_t *tp;
+        xfs_inode_t *ip;
+
+
+        tp = _xfs_trans_alloc(mp, XFS_TRANS_DUMMY1);
+        atomic_inc(&mp->m_active_trans);
+        if (xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0)) {
+                xfs_trans_cancel(tp, 0);
+                return;
+        }
+
+        ip = mp->m_rootip;
+        xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+        xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
+        xfs_trans_ihold(tp, ip);
+        xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+        xfs_trans_set_sync(tp);
+        xfs_trans_commit(tp, 0, NULL);
+
+        xfs_iunlock(ip, XFS_ILOCK_EXCL);
+}
+
 int
 xfs_fs_goingdown(
         xfs_mount_t     *mp,
diff --git a/fs/xfs/xfs_fsops.h b/fs/xfs/xfs_fsops.h
index f32713f14f9a..300d0c9d61ad 100644
--- a/fs/xfs/xfs_fsops.h
+++ b/fs/xfs/xfs_fsops.h
@@ -25,5 +25,6 @@ extern int xfs_fs_counts(xfs_mount_t *mp, xfs_fsop_counts_t *cnt);
 extern int xfs_reserve_blocks(xfs_mount_t *mp, __uint64_t *inval,
                                 xfs_fsop_resblks_t *outval);
 extern int xfs_fs_goingdown(xfs_mount_t *mp, __uint32_t inflags);
+extern void xfs_fs_log_dummy(xfs_mount_t *mp);
 
 #endif  /* __XFS_FSOPS_H__ */
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index fc19eedbd11b..8e380a1fb79b 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -493,7 +493,6 @@ xfs_iget(
 
 retry:
         if ((inode = iget_locked(XFS_MTOVFS(mp)->vfs_super, ino))) {
-                bhv_desc_t      *bdp;
                 xfs_inode_t     *ip;
 
                 vp = LINVFS_GET_VP(inode);
@@ -517,14 +516,12 @@ retry:
                          * to wait for the inode to go away.
                          */
                         if (is_bad_inode(inode) ||
-                            ((bdp = vn_bhv_lookup(VN_BHV_HEAD(vp),
-                                                  &xfs_vnodeops)) == NULL)) {
+                            ((ip = xfs_vtoi(vp)) == NULL)) {
                                 iput(inode);
                                 delay(1);
                                 goto retry;
                         }
 
-                        ip = XFS_BHVTOI(bdp);
                         if (lock_flags != 0)
                                 xfs_ilock(ip, lock_flags);
                         XFS_STATS_INC(xs_ig_found);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index df0d4572d70a..1d7f5a7e063e 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -404,9 +404,8 @@ xfs_iformat(
             INT_GET(dip->di_core.di_nextents, ARCH_CONVERT) +
                 INT_GET(dip->di_core.di_anextents, ARCH_CONVERT) >
             INT_GET(dip->di_core.di_nblocks, ARCH_CONVERT))) {
-                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                        "corrupt dinode %Lu, extent total = %d, nblocks = %Lu."
-                        "  Unmount and run xfs_repair.",
+                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                        "corrupt dinode %Lu, extent total = %d, nblocks = %Lu.",
                         (unsigned long long)ip->i_ino,
                         (int)(INT_GET(dip->di_core.di_nextents, ARCH_CONVERT)
                             + INT_GET(dip->di_core.di_anextents, ARCH_CONVERT)),
@@ -418,9 +417,8 @@ xfs_iformat(
         }
 
         if (unlikely(INT_GET(dip->di_core.di_forkoff, ARCH_CONVERT) > ip->i_mount->m_sb.sb_inodesize)) {
-                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                        "corrupt dinode %Lu, forkoff = 0x%x."
-                        "  Unmount and run xfs_repair.",
+                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                        "corrupt dinode %Lu, forkoff = 0x%x.",
                         (unsigned long long)ip->i_ino,
                         (int)(INT_GET(dip->di_core.di_forkoff, ARCH_CONVERT)));
                 XFS_CORRUPTION_ERROR("xfs_iformat(2)", XFS_ERRLEVEL_LOW,
@@ -451,8 +449,9 @@ xfs_iformat(
                          * no local regular files yet
                          */
                         if (unlikely((INT_GET(dip->di_core.di_mode, ARCH_CONVERT) & S_IFMT) == S_IFREG)) {
-                                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                                        "corrupt inode (local format for regular file) %Lu.  Unmount and run xfs_repair.",
+                                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                                        "corrupt inode %Lu "
+                                        "(local format for regular file).",
                                         (unsigned long long) ip->i_ino);
                                 XFS_CORRUPTION_ERROR("xfs_iformat(4)",
                                                      XFS_ERRLEVEL_LOW,
@@ -462,8 +461,9 @@ xfs_iformat(
 
                         di_size = INT_GET(dip->di_core.di_size, ARCH_CONVERT);
                         if (unlikely(di_size > XFS_DFORK_DSIZE(dip, ip->i_mount))) {
-                                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                                        "corrupt inode %Lu (bad size %Ld for local inode).  Unmount and run xfs_repair.",
+                                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                                        "corrupt inode %Lu "
+                                        "(bad size %Ld for local inode).",
                                         (unsigned long long) ip->i_ino,
                                         (long long) di_size);
                                 XFS_CORRUPTION_ERROR("xfs_iformat(5)",
@@ -551,8 +551,9 @@ xfs_iformat_local(
          * kmem_alloc() or memcpy() below.
          */
         if (unlikely(size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
-                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                        "corrupt inode %Lu (bad size %d for local fork, size = %d).  Unmount and run xfs_repair.",
+                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                        "corrupt inode %Lu "
+                        "(bad size %d for local fork, size = %d).",
                         (unsigned long long) ip->i_ino, size,
                         XFS_DFORK_SIZE(dip, ip->i_mount, whichfork));
                 XFS_CORRUPTION_ERROR("xfs_iformat_local", XFS_ERRLEVEL_LOW,
@@ -610,8 +611,8 @@ xfs_iformat_extents(
          * kmem_alloc() or memcpy() below.
          */
         if (unlikely(size < 0 || size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
-                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                        "corrupt inode %Lu ((a)extents = %d).  Unmount and run xfs_repair.",
+                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                        "corrupt inode %Lu ((a)extents = %d).",
                         (unsigned long long) ip->i_ino, nex);
                 XFS_CORRUPTION_ERROR("xfs_iformat_extents(1)", XFS_ERRLEVEL_LOW,
                                      ip->i_mount, dip);
@@ -692,8 +693,8 @@ xfs_iformat_btree(
             || XFS_BMDR_SPACE_CALC(nrecs) >
                         XFS_DFORK_SIZE(dip, ip->i_mount, whichfork)
             || XFS_IFORK_NEXTENTS(ip, whichfork) > ip->i_d.di_nblocks)) {
-                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
-                        "corrupt inode %Lu (btree).  Unmount and run xfs_repair.",
+                xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
+                        "corrupt inode %Lu (btree).",
                         (unsigned long long) ip->i_ino);
                 XFS_ERROR_REPORT("xfs_iformat_btree", XFS_ERRLEVEL_LOW,
                                  ip->i_mount);
@@ -809,6 +810,10 @@ _xfs_dic2xflags(
                         flags |= XFS_XFLAG_PROJINHERIT;
                 if (di_flags & XFS_DIFLAG_NOSYMLINKS)
                         flags |= XFS_XFLAG_NOSYMLINKS;
+                if (di_flags & XFS_DIFLAG_EXTSIZE)
+                        flags |= XFS_XFLAG_EXTSIZE;
+                if (di_flags & XFS_DIFLAG_EXTSZINHERIT)
+                        flags |= XFS_XFLAG_EXTSZINHERIT;
         }
 
         return flags;
@@ -1192,11 +1197,19 @@ xfs_ialloc(
                         if ((mode & S_IFMT) == S_IFDIR) {
                                 if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT)
                                         di_flags |= XFS_DIFLAG_RTINHERIT;
-                        } else {
+                                if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+                                        di_flags |= XFS_DIFLAG_EXTSZINHERIT;
+                                        ip->i_d.di_extsize = pip->i_d.di_extsize;
+                                }
+                        } else if ((mode & S_IFMT) == S_IFREG) {
                                 if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT) {
                                         di_flags |= XFS_DIFLAG_REALTIME;
                                         ip->i_iocore.io_flags |= XFS_IOCORE_RT;
                                 }
+                                if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+                                        di_flags |= XFS_DIFLAG_EXTSIZE;
+                                        ip->i_d.di_extsize = pip->i_d.di_extsize;
+                                }
                         }
                         if ((pip->i_d.di_flags & XFS_DIFLAG_NOATIME) &&
                             xfs_inherit_noatime)
@@ -1262,7 +1275,7 @@ xfs_isize_check(
         if ((ip->i_d.di_mode & S_IFMT) != S_IFREG)
                 return;
 
-        if ( ip->i_d.di_flags & XFS_DIFLAG_REALTIME )
+        if (ip->i_d.di_flags & (XFS_DIFLAG_REALTIME | XFS_DIFLAG_EXTSIZE))
                 return;
 
         nimaps = 2;
@@ -1765,22 +1778,19 @@ xfs_igrow_start(
         xfs_fsize_t     new_size,
         cred_t          *credp)
 {
-        xfs_fsize_t     isize;
         int             error;
 
         ASSERT(ismrlocked(&(ip->i_lock), MR_UPDATE) != 0);
         ASSERT(ismrlocked(&(ip->i_iolock), MR_UPDATE) != 0);
         ASSERT(new_size > ip->i_d.di_size);
 
-        error = 0;
-        isize = ip->i_d.di_size;
         /*
          * Zero any pages that may have been created by
          * xfs_write_file() beyond the end of the file
          * and any blocks between the old and new file sizes.
          */
-        error = xfs_zero_eof(XFS_ITOV(ip), &ip->i_iocore, new_size, isize,
-                                new_size);
+        error = xfs_zero_eof(XFS_ITOV(ip), &ip->i_iocore, new_size,
+                             ip->i_d.di_size, new_size);
         return error;
 }
 
@@ -3355,6 +3365,11 @@ xfs_iflush_int(
         ip->i_update_core = 0;
         SYNCHRONIZE();
 
+        /*
+         * Make sure to get the latest atime from the Linux inode.
+         */
+        xfs_synchronize_atime(ip);
+
         if (XFS_TEST_ERROR(INT_GET(dip->di_core.di_magic,ARCH_CONVERT) != XFS_DINODE_MAGIC,
                                mp, XFS_ERRTAG_IFLUSH_1, XFS_RANDOM_IFLUSH_1)) {
                 xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 124d30e6143b..1cfbcf18ce86 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -436,6 +436,10 @@ void		xfs_ichgtime(xfs_inode_t *, int);
 xfs_fsize_t     xfs_file_last_byte(xfs_inode_t *);
 void            xfs_lock_inodes(xfs_inode_t **, int, int, uint);
 
+xfs_inode_t     *xfs_vtoi(struct vnode *vp);
+
+void            xfs_synchronize_atime(xfs_inode_t *);
+
 #define xfs_ipincount(ip)       ((unsigned int) atomic_read(&ip->i_pincount))
 
 #ifdef DEBUG
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 7f3363c621e1..36aa1fcb90a5 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -271,6 +271,11 @@ xfs_inode_item_format(
         if (ip->i_update_size)
                 ip->i_update_size = 0;
 
+        /*
+         * Make sure to get the latest atime from the Linux inode.
+         */
+        xfs_synchronize_atime(ip);
+
         vecp->i_addr = (xfs_caddr_t)&ip->i_d;
         vecp->i_len  = sizeof(xfs_dinode_core_t);
         XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_ICORE);
@@ -603,7 +608,7 @@ xfs_inode_item_trylock(
                 if (iip->ili_pushbuf_flag == 0) {
                         iip->ili_pushbuf_flag = 1;
 #ifdef DEBUG
-                        iip->ili_push_owner = get_thread_id();
+                        iip->ili_push_owner = current_pid();
 #endif
                         /*
                          * Inode is left locked in shared mode.
@@ -782,7 +787,7 @@ xfs_inode_item_pushbuf(
          * trying to duplicate our effort.
          */
         ASSERT(iip->ili_pushbuf_flag != 0);
-        ASSERT(iip->ili_push_owner == get_thread_id());
+        ASSERT(iip->ili_push_owner == current_pid());
 
         /*
          * If flushlock isn't locked anymore, chances are that the
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index ca7afc83a893..788917f355c4 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -262,7 +262,7 @@ phase2:
         case BMAPI_WRITE:
                 /* If we found an extent, return it */
                 if (nimaps &&
-                    (imap.br_startblock != HOLESTARTBLOCK) && 
+                    (imap.br_startblock != HOLESTARTBLOCK) &&
                     (imap.br_startblock != DELAYSTARTBLOCK)) {
                         xfs_iomap_map_trace(XFS_IOMAP_WRITE_MAP, io,
                                         offset, count, iomapp, &imap, flags);
@@ -317,6 +317,58 @@ out:
 }
 
 STATIC int
+xfs_iomap_eof_align_last_fsb(
+        xfs_mount_t     *mp,
+        xfs_iocore_t    *io,
+        xfs_fsize_t     isize,
+        xfs_extlen_t    extsize,
+        xfs_fileoff_t   *last_fsb)
+{
+        xfs_fileoff_t   new_last_fsb = 0;
+        xfs_extlen_t    align;
+        int             eof, error;
+
+        if (io->io_flags & XFS_IOCORE_RT)
+                ;
+        /*
+         * If mounted with the "-o swalloc" option, roundup the allocation
+         * request to a stripe width boundary if the file size is >=
+         * stripe width and we are allocating past the allocation eof.
+         */
+        else if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC) &&
+                (isize >= XFS_FSB_TO_B(mp, mp->m_swidth)))
+                new_last_fsb = roundup_64(*last_fsb, mp->m_swidth);
+        /*
+         * Roundup the allocation request to a stripe unit (m_dalign) boundary
+         * if the file size is >= stripe unit size, and we are allocating past
+         * the allocation eof.
+         */
+        else if (mp->m_dalign && (isize >= XFS_FSB_TO_B(mp, mp->m_dalign)))
+                new_last_fsb = roundup_64(*last_fsb, mp->m_dalign);
+
+        /*
+         * Always round up the allocation request to an extent boundary
+         * (when file on a real-time subvolume or has di_extsize hint).
+         */
+        if (extsize) {
+                if (new_last_fsb)
+                        align = roundup_64(new_last_fsb, extsize);
+                else
+                        align = extsize;
+                new_last_fsb = roundup_64(*last_fsb, align);
+        }
+
+        if (new_last_fsb) {
+                error = XFS_BMAP_EOF(mp, io, new_last_fsb, XFS_DATA_FORK, &eof);
+                if (error)
+                        return error;
+                if (eof)
+                        *last_fsb = new_last_fsb;
+        }
+        return 0;
+}
+
+STATIC int
 xfs_flush_space(
         xfs_inode_t     *ip,
         int             *fsynced,
@@ -362,19 +414,20 @@ xfs_iomap_write_direct(
         xfs_iocore_t    *io = &ip->i_iocore;
         xfs_fileoff_t   offset_fsb;
         xfs_fileoff_t   last_fsb;
-        xfs_filblks_t   count_fsb;
+        xfs_filblks_t   count_fsb, resaligned;
         xfs_fsblock_t   firstfsb;
+        xfs_extlen_t    extsz, temp;
+        xfs_fsize_t     isize;
         int             nimaps;
-        int             error;
         int             bmapi_flag;
         int             quota_flag;
         int             rt;
         xfs_trans_t     *tp;
         xfs_bmbt_irec_t imap;
         xfs_bmap_free_t free_list;
-        xfs_filblks_t   qblocks, resblks;
+        uint            qblocks, resblks, resrtextents;
         int             committed;
-        int             resrtextents;
+        int             error;
 
         /*
          * Make sure that the dquots are there. This doesn't hold
@@ -384,37 +437,52 @@ xfs_iomap_write_direct(
         if (error)
                 return XFS_ERROR(error);
 
-        offset_fsb = XFS_B_TO_FSBT(mp, offset);
-        last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
-        count_fsb = last_fsb - offset_fsb;
-        if (found && (ret_imap->br_startblock == HOLESTARTBLOCK)) {
-                xfs_fileoff_t   map_last_fsb;
-
-                map_last_fsb = ret_imap->br_blockcount + ret_imap->br_startoff;
-                if (map_last_fsb < last_fsb) {
-                        last_fsb = map_last_fsb;
-                        count_fsb = last_fsb - offset_fsb;
-                }
-                ASSERT(count_fsb > 0);
+        rt = XFS_IS_REALTIME_INODE(ip);
+        if (unlikely(rt)) {
+                if (!(extsz = ip->i_d.di_extsize))
+                        extsz = mp->m_sb.sb_rextsize;
+        } else {
+                extsz = ip->i_d.di_extsize;
         }
 
-        /*
-         * Determine if reserving space on the data or realtime partition.
-         */
-        if ((rt = XFS_IS_REALTIME_INODE(ip))) {
-                xfs_extlen_t    extsz;
+        isize = ip->i_d.di_size;
+        if (io->io_new_size > isize)
+                isize = io->io_new_size;
 
-                if (!(extsz = ip->i_d.di_extsize))
-                        extsz = mp->m_sb.sb_rextsize;
-                resrtextents = qblocks = (count_fsb + extsz - 1);
-                do_div(resrtextents, mp->m_sb.sb_rextsize);
-                resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
-                quota_flag = XFS_QMOPT_RES_RTBLKS;
+        offset_fsb = XFS_B_TO_FSBT(mp, offset);
+        last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
+        if ((offset + count) > isize) {
+                error = xfs_iomap_eof_align_last_fsb(mp, io, isize, extsz,
+                                                        &last_fsb);
+                if (error)
+                        goto error_out;
         } else {
-                resrtextents = 0;
-                resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, count_fsb);
-                quota_flag = XFS_QMOPT_RES_REGBLKS;
+                if (found && (ret_imap->br_startblock == HOLESTARTBLOCK))
+                        last_fsb = MIN(last_fsb, (xfs_fileoff_t)
+                                        ret_imap->br_blockcount +
+                                        ret_imap->br_startoff);
         }
+        count_fsb = last_fsb - offset_fsb;
+        ASSERT(count_fsb > 0);
+
+        resaligned = count_fsb;
+        if (unlikely(extsz)) {
+                if ((temp = do_mod(offset_fsb, extsz)))
+                        resaligned += temp;
+                if ((temp = do_mod(resaligned, extsz)))
+                        resaligned += extsz - temp;
+        }
+
+        if (unlikely(rt)) {
+                resrtextents = qblocks = resaligned;
+                resrtextents /= mp->m_sb.sb_rextsize;
+                resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+                quota_flag = XFS_QMOPT_RES_RTBLKS;
+        } else {
+                resrtextents = 0;
+                resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
+                quota_flag = XFS_QMOPT_RES_REGBLKS;
+        }
 
         /*
          * Allocate and setup the transaction
@@ -425,7 +493,6 @@ xfs_iomap_write_direct(
                         XFS_WRITE_LOG_RES(mp), resrtextents,
                         XFS_TRANS_PERM_LOG_RES,
                         XFS_WRITE_LOG_COUNT);
-
         /*
          * Check for running out of space, note: need lock to return
          */
@@ -435,20 +502,20 @@ xfs_iomap_write_direct(
         if (error)
                 goto error_out;
 
-        if (XFS_TRANS_RESERVE_QUOTA_NBLKS(mp, tp, ip, qblocks, 0, quota_flag)) {
-                error = (EDQUOT);
+        error = XFS_TRANS_RESERVE_QUOTA_NBLKS(mp, tp, ip,
+                                              qblocks, 0, quota_flag);
+        if (error)
                 goto error1;
-        }
 
-        bmapi_flag = XFS_BMAPI_WRITE;
         xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
         xfs_trans_ihold(tp, ip);
 
-        if (!(flags & BMAPI_MMAP) && (offset < ip->i_d.di_size || rt))
+        bmapi_flag = XFS_BMAPI_WRITE;
+        if ((flags & BMAPI_DIRECT) && (offset < ip->i_d.di_size || extsz))
                 bmapi_flag |= XFS_BMAPI_PREALLOC;
 
         /*
-         * Issue the bmapi() call to allocate the blocks
+         * Issue the xfs_bmapi() call to allocate the blocks
          */
         XFS_BMAP_INIT(&free_list, &firstfsb);
         nimaps = 1;
@@ -483,8 +550,10 @@ xfs_iomap_write_direct(
                         "extent-state : %x \n",
                         (ip->i_mount)->m_fsname,
                         (long long)ip->i_ino,
-                        ret_imap->br_startblock, ret_imap->br_startoff,
-                        ret_imap->br_blockcount,ret_imap->br_state);
+                        (unsigned long long)ret_imap->br_startblock,
+                        (unsigned long long)ret_imap->br_startoff,
+                        (unsigned long long)ret_imap->br_blockcount,
+                        ret_imap->br_state);
         }
         return 0;
 
@@ -500,6 +569,63 @@ error_out:
         return XFS_ERROR(error);
 }
 
+/*
+ * If the caller is doing a write at the end of the file,
+ * then extend the allocation out to the file system's write
+ * iosize.  We clean up any extra space left over when the
+ * file is closed in xfs_inactive().
+ *
+ * For sync writes, we are flushing delayed allocate space to
+ * try to make additional space available for allocation near
+ * the filesystem full boundary - preallocation hurts in that
+ * situation, of course.
+ */
+STATIC int
+xfs_iomap_eof_want_preallocate(
+        xfs_mount_t     *mp,
+        xfs_iocore_t    *io,
+        xfs_fsize_t     isize,
+        xfs_off_t       offset,
+        size_t          count,
+        int             ioflag,
+        xfs_bmbt_irec_t *imap,
+        int             nimaps,
+        int             *prealloc)
+{
+        xfs_fileoff_t   start_fsb;
+        xfs_filblks_t   count_fsb;
+        xfs_fsblock_t   firstblock;
+        int             n, error, imaps;
+
+        *prealloc = 0;
+        if ((ioflag & BMAPI_SYNC) || (offset + count) <= isize)
+                return 0;
+
+        /*
+         * If there are any real blocks past eof, then don't
+         * do any speculative allocation.
+         */
+        start_fsb = XFS_B_TO_FSBT(mp, ((xfs_ufsize_t)(offset + count - 1)));
+        count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+        while (count_fsb > 0) {
+                imaps = nimaps;
+                firstblock = NULLFSBLOCK;
+                error = XFS_BMAPI(mp, NULL, io, start_fsb, count_fsb,
+                                  0, &firstblock, 0, imap, &imaps, NULL);
+                if (error)
+                        return error;
+                for (n = 0; n < imaps; n++) {
+                        if ((imap[n].br_startblock != HOLESTARTBLOCK) &&
+                            (imap[n].br_startblock != DELAYSTARTBLOCK))
+                                return 0;
+                        start_fsb += imap[n].br_blockcount;
+                        count_fsb -= imap[n].br_blockcount;
+                }
+        }
+        *prealloc = 1;
+        return 0;
+}
+
 int
 xfs_iomap_write_delay(
         xfs_inode_t     *ip,
@@ -513,13 +639,15 @@ xfs_iomap_write_delay(
         xfs_iocore_t    *io = &ip->i_iocore;
         xfs_fileoff_t   offset_fsb;
         xfs_fileoff_t   last_fsb;
-        xfs_fsize_t     isize;
+        xfs_off_t       aligned_offset;
+        xfs_fileoff_t   ioalign;
         xfs_fsblock_t   firstblock;
+        xfs_extlen_t    extsz;
+        xfs_fsize_t     isize;
         int             nimaps;
-        int             error;
         xfs_bmbt_irec_t imap[XFS_WRITE_IMAPS];
-        int             aeof;
-        int             fsynced = 0;
+        int             prealloc, fsynced = 0;
+        int             error;
 
         ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE) != 0);
 
@@ -527,152 +655,57 @@ xfs_iomap_write_delay(
          * Make sure that the dquots are there. This doesn't hold
          * the ilock across a disk read.
          */
-
         error = XFS_QM_DQATTACH(mp, ip, XFS_QMOPT_ILOCKED);
         if (error)
                 return XFS_ERROR(error);
 
+        if (XFS_IS_REALTIME_INODE(ip)) {
+                if (!(extsz = ip->i_d.di_extsize))
+                        extsz = mp->m_sb.sb_rextsize;
+        } else {
+                extsz = ip->i_d.di_extsize;
+        }
+
+        offset_fsb = XFS_B_TO_FSBT(mp, offset);
+
 retry:
         isize = ip->i_d.di_size;
-        if (io->io_new_size > isize) {
+        if (io->io_new_size > isize)
                 isize = io->io_new_size;
-        }
 
-        aeof = 0;
-        offset_fsb = XFS_B_TO_FSBT(mp, offset);
-        last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
-        /*
-         * If the caller is doing a write at the end of the file,
-         * then extend the allocation (and the buffer used for the write)
-         * out to the file system's write iosize.  We clean up any extra
-         * space left over when the file is closed in xfs_inactive().
-         *
-         * For sync writes, we are flushing delayed allocate space to
-         * try to make additional space available for allocation near
-         * the filesystem full boundary - preallocation hurts in that
-         * situation, of course.
-         */
-        if (!(ioflag & BMAPI_SYNC) && ((offset + count) > ip->i_d.di_size)) {
-                xfs_off_t       aligned_offset;
-                xfs_filblks_t   count_fsb;
-                unsigned int    iosize;
-                xfs_fileoff_t   ioalign;
-                int             n;
-                xfs_fileoff_t   start_fsb;
+        error = xfs_iomap_eof_want_preallocate(mp, io, isize, offset, count,
+                                ioflag, imap, XFS_WRITE_IMAPS, &prealloc);
+        if (error)
+                return error;
 
-                /*
-                 * If there are any real blocks past eof, then don't
-                 * do any speculative allocation.
-                 */
-                start_fsb = XFS_B_TO_FSBT(mp,
-                                        ((xfs_ufsize_t)(offset + count - 1)));
-                count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
-                while (count_fsb > 0) {
-                        nimaps = XFS_WRITE_IMAPS;
-                        error = XFS_BMAPI(mp, NULL, io, start_fsb, count_fsb,
-                                        0, &firstblock, 0, imap, &nimaps, NULL);
-                        if (error) {
-                                return error;
-                        }
-                        for (n = 0; n < nimaps; n++) {
-                                if ( !(io->io_flags & XFS_IOCORE_RT)  && 
-                                        !imap[n].br_startblock) {
-                                        cmn_err(CE_PANIC,"Access to block "
-                                                "zero:  fs <%s> inode: %lld "
-                                                "start_block : %llx start_off "
-                                                ": %llx blkcnt : %llx "
-                                                "extent-state : %x \n",
-                                                (ip->i_mount)->m_fsname,
-                                                (long long)ip->i_ino,
-                                                imap[n].br_startblock,
-                                                imap[n].br_startoff,
-                                                imap[n].br_blockcount,
-                                                imap[n].br_state);
-                                }
-                                if ((imap[n].br_startblock != HOLESTARTBLOCK) &&
-                                    (imap[n].br_startblock != DELAYSTARTBLOCK)) {
-                                        goto write_map;
-                                }
-                                start_fsb += imap[n].br_blockcount;
-                                count_fsb -= imap[n].br_blockcount;
-                        }
-                }
-                iosize = mp->m_writeio_blocks;
+        if (prealloc) {
                 aligned_offset = XFS_WRITEIO_ALIGN(mp, (offset + count - 1));
                 ioalign = XFS_B_TO_FSBT(mp, aligned_offset);
-                last_fsb = ioalign + iosize;
-                aeof = 1;
+                last_fsb = ioalign + mp->m_writeio_blocks;
+        } else {
+                last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
         }
-write_map:
-        nimaps = XFS_WRITE_IMAPS;
-        firstblock = NULLFSBLOCK;
 
-        /*
-         * If mounted with the "-o swalloc" option, roundup the allocation
-         * request to a stripe width boundary if the file size is >=
-         * stripe width and we are allocating past the allocation eof.
-         */
-        if (!(io->io_flags & XFS_IOCORE_RT) && mp->m_swidth 
-            && (mp->m_flags & XFS_MOUNT_SWALLOC)
-            && (isize >= XFS_FSB_TO_B(mp, mp->m_swidth)) && aeof) {
-                int eof;
-                xfs_fileoff_t new_last_fsb;
-
-                new_last_fsb = roundup_64(last_fsb, mp->m_swidth);
-                error = xfs_bmap_eof(ip, new_last_fsb, XFS_DATA_FORK, &eof);
-                if (error) {
-                        return error;
-                }
-                if (eof) {
-                        last_fsb = new_last_fsb;
-                }
-        /*
-         * Roundup the allocation request to a stripe unit (m_dalign) boundary
-         * if the file size is >= stripe unit size, and we are allocating past
-         * the allocation eof.
-         */
-        } else if (!(io->io_flags & XFS_IOCORE_RT) && mp->m_dalign &&
-                   (isize >= XFS_FSB_TO_B(mp, mp->m_dalign)) && aeof) {
-                int eof;
-                xfs_fileoff_t new_last_fsb;
-                new_last_fsb = roundup_64(last_fsb, mp->m_dalign);
-                error = xfs_bmap_eof(ip, new_last_fsb, XFS_DATA_FORK, &eof);
-                if (error) {
-                        return error;
-                }
-                if (eof) {
-                        last_fsb = new_last_fsb;
-                }
-        /*
-         * Round up the allocation request to a real-time extent boundary
-         * if the file is on the real-time subvolume.
-         */
-        } else if (io->io_flags & XFS_IOCORE_RT && aeof) {
-                int eof;
-                xfs_fileoff_t new_last_fsb;
-
-                new_last_fsb = roundup_64(last_fsb, mp->m_sb.sb_rextsize);
-                error = XFS_BMAP_EOF(mp, io, new_last_fsb, XFS_DATA_FORK, &eof);
-                if (error) {
+        if (prealloc || extsz) {
+                error = xfs_iomap_eof_align_last_fsb(mp, io, isize, extsz,
+                                                        &last_fsb);
+                if (error)
                         return error;
-                }
-                if (eof)
-                        last_fsb = new_last_fsb;
         }
+
+        nimaps = XFS_WRITE_IMAPS;
+        firstblock = NULLFSBLOCK;
         error = xfs_bmapi(NULL, ip, offset_fsb,
                           (xfs_filblks_t)(last_fsb - offset_fsb),
                           XFS_BMAPI_DELAY | XFS_BMAPI_WRITE |
                           XFS_BMAPI_ENTIRE, &firstblock, 1, imap,
                           &nimaps, NULL);
-        /*
-         * This can be EDQUOT, if nimaps == 0
-         */
-        if (error && (error != ENOSPC)) {
+        if (error && (error != ENOSPC))
                 return XFS_ERROR(error);
-        }
+
         /*
          * If bmapi returned us nothing, and if we didn't get back EDQUOT,
-         * then we must have run out of space.
+         * then we must have run out of space - flush delalloc, and retry..
          */
         if (nimaps == 0) {
                 xfs_iomap_enter_trace(XFS_IOMAP_WRITE_NOSPACE,
@@ -684,17 +717,21 @@ write_map:
                 goto retry;
         }
 
-        *ret_imap = imap[0];
-        *nmaps = 1;
-        if ( !(io->io_flags & XFS_IOCORE_RT)  && !ret_imap->br_startblock) {
+        if (!(io->io_flags & XFS_IOCORE_RT)  && !ret_imap->br_startblock) {
                 cmn_err(CE_PANIC,"Access to block zero:  fs <%s> inode: %lld "
                         "start_block : %llx start_off : %llx blkcnt : %llx "
                         "extent-state : %x \n",
                         (ip->i_mount)->m_fsname,
                         (long long)ip->i_ino,
-                        ret_imap->br_startblock, ret_imap->br_startoff,
-                        ret_imap->br_blockcount,ret_imap->br_state);
+                        (unsigned long long)ret_imap->br_startblock,
+                        (unsigned long long)ret_imap->br_startoff,
+                        (unsigned long long)ret_imap->br_blockcount,
+                        ret_imap->br_state);
         }
+
+        *ret_imap = imap[0];
+        *nmaps = 1;
+
         return 0;
 }
 
@@ -820,17 +857,21 @@ xfs_iomap_write_allocate(
                  */
 
                 for (i = 0; i < nimaps; i++) {
-                        if ( !(io->io_flags & XFS_IOCORE_RT)  && 
-                                !imap[i].br_startblock) {
+                        if (!(io->io_flags & XFS_IOCORE_RT)  &&
+                            !imap[i].br_startblock) {
                                 cmn_err(CE_PANIC,"Access to block zero:  "
                                         "fs <%s> inode: %lld "
-                                        "start_block : %llx start_off : %llx " 
+                                        "start_block : %llx start_off : %llx "
                                         "blkcnt : %llx extent-state : %x \n",
                                         (ip->i_mount)->m_fsname,
                                         (long long)ip->i_ino,
-                                        imap[i].br_startblock,
-                                        imap[i].br_startoff,
-                                        imap[i].br_blockcount,imap[i].br_state);
+                                        (unsigned long long)
+                                                imap[i].br_startblock,
+                                        (unsigned long long)
+                                                imap[i].br_startoff,
+                                        (unsigned long long)
+                                                imap[i].br_blockcount,
+                                        imap[i].br_state);
                         }
                         if ((offset_fsb >= imap[i].br_startoff) &&
                             (offset_fsb < (imap[i].br_startoff +
@@ -867,17 +908,17 @@ xfs_iomap_write_unwritten(
 {
         xfs_mount_t     *mp = ip->i_mount;
         xfs_iocore_t    *io = &ip->i_iocore;
-        xfs_trans_t     *tp;
         xfs_fileoff_t   offset_fsb;
         xfs_filblks_t   count_fsb;
         xfs_filblks_t   numblks_fsb;
-        xfs_bmbt_irec_t imap;
+        xfs_fsblock_t   firstfsb;
+        int             nimaps;
+        xfs_trans_t     *tp;
+        xfs_bmbt_irec_t imap;
+        xfs_bmap_free_t free_list;
+        uint            resblks;
         int             committed;
         int             error;
-        int             nres;
-        int             nimaps;
-        xfs_fsblock_t   firstfsb;
-        xfs_bmap_free_t free_list;
 
         xfs_iomap_enter_trace(XFS_IOMAP_UNWRITTEN,
                                 &ip->i_iocore, offset, count);
@@ -886,9 +927,9 @@ xfs_iomap_write_unwritten(
         count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
         count_fsb = (xfs_filblks_t)(count_fsb - offset_fsb);
 
-        do {
-                nres = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+        resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0) << 1;
 
+        do {
                 /*
                  * set up a transaction to convert the range of extents
                  * from unwritten to real. Do allocations in a loop until
@@ -896,7 +937,7 @@ xfs_iomap_write_unwritten(
                  */
 
                 tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
-                error = xfs_trans_reserve(tp, nres,
+                error = xfs_trans_reserve(tp, resblks,
                                 XFS_WRITE_LOG_RES(mp), 0,
                                 XFS_TRANS_PERM_LOG_RES,
                                 XFS_WRITE_LOG_COUNT);
@@ -915,7 +956,7 @@ xfs_iomap_write_unwritten(
                 XFS_BMAP_INIT(&free_list, &firstfsb);
                 nimaps = 1;
                 error = xfs_bmapi(tp, ip, offset_fsb, count_fsb,
-                                  XFS_BMAPI_WRITE, &firstfsb,
+                                  XFS_BMAPI_WRITE|XFS_BMAPI_CONVERT, &firstfsb,
                                   1, &imap, &nimaps, &free_list);
                 if (error)
                         goto error_on_bmapi_transaction;
@@ -929,15 +970,17 @@ xfs_iomap_write_unwritten(
                 xfs_iunlock(ip, XFS_ILOCK_EXCL);
                 if (error)
                         goto error0;
-                
+
                 if ( !(io->io_flags & XFS_IOCORE_RT)  && !imap.br_startblock) {
                         cmn_err(CE_PANIC,"Access to block zero:  fs <%s> "
                                 "inode: %lld start_block : %llx start_off : "
                                 "%llx blkcnt : %llx extent-state : %x \n",
                                 (ip->i_mount)->m_fsname,
                                 (long long)ip->i_ino,
-                                imap.br_startblock,imap.br_startoff,
-                                imap.br_blockcount,imap.br_state);
+                                (unsigned long long)imap.br_startblock,
+                                (unsigned long long)imap.br_startoff,
+                                (unsigned long long)imap.br_blockcount,
+                                imap.br_state);
                 }
 
                 if ((numblks_fsb = imap.br_blockcount) == 0) {
diff --git a/fs/xfs/xfs_itable.c b/fs/xfs/xfs_itable.c
index f63646ead816..c59450e1be40 100644
--- a/fs/xfs/xfs_itable.c
+++ b/fs/xfs/xfs_itable.c
@@ -56,6 +56,7 @@ xfs_bulkstat_one_iget(
 {
         xfs_dinode_core_t *dic;         /* dinode core info pointer */
         xfs_inode_t     *ip;            /* incore inode pointer */
+        vnode_t         *vp;
         int             error;
 
         error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_SHARED, &ip, bno);
@@ -72,6 +73,7 @@ xfs_bulkstat_one_iget(
                 goto out_iput;
         }
 
+        vp = XFS_ITOV(ip);
         dic = &ip->i_d;
 
         /* xfs_iget returns the following without needing
@@ -84,8 +86,7 @@ xfs_bulkstat_one_iget(
         buf->bs_uid = dic->di_uid;
         buf->bs_gid = dic->di_gid;
         buf->bs_size = dic->di_size;
-        buf->bs_atime.tv_sec = dic->di_atime.t_sec;
-        buf->bs_atime.tv_nsec = dic->di_atime.t_nsec;
+        vn_atime_to_bstime(vp, &buf->bs_atime);
         buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
         buf->bs_mtime.tv_nsec = dic->di_mtime.t_nsec;
         buf->bs_ctime.tv_sec = dic->di_ctime.t_sec;
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 29af51275ca9..3d9a36e77363 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -178,6 +178,83 @@ xlog_trace_iclog(xlog_in_core_t *iclog, uint state)
 #define xlog_trace_iclog(iclog,state)
 #endif /* XFS_LOG_TRACE */
 
+
+static void
+xlog_ins_ticketq(struct xlog_ticket **qp, struct xlog_ticket *tic)
+{
+        if (*qp) {
+                tic->t_next         = (*qp);
+                tic->t_prev         = (*qp)->t_prev;
+                (*qp)->t_prev->t_next = tic;
+                (*qp)->t_prev       = tic;
+        } else {
+                tic->t_prev = tic->t_next = tic;
+                *qp = tic;
+        }
+
+        tic->t_flags |= XLOG_TIC_IN_Q;
+}
+
+static void
+xlog_del_ticketq(struct xlog_ticket **qp, struct xlog_ticket *tic)
+{
+        if (tic == tic->t_next) {
+                *qp = NULL;
+        } else {
+                *qp = tic->t_next;
+                tic->t_next->t_prev = tic->t_prev;
+                tic->t_prev->t_next = tic->t_next;
+        }
+
+        tic->t_next = tic->t_prev = NULL;
+        tic->t_flags &= ~XLOG_TIC_IN_Q;
+}
+
+static void
+xlog_grant_sub_space(struct log *log, int bytes)
+{
+        log->l_grant_write_bytes -= bytes;
+        if (log->l_grant_write_bytes < 0) {
+                log->l_grant_write_bytes += log->l_logsize;
+                log->l_grant_write_cycle--;
+        }
+
+        log->l_grant_reserve_bytes -= bytes;
+        if ((log)->l_grant_reserve_bytes < 0) {
+                log->l_grant_reserve_bytes += log->l_logsize;
+                log->l_grant_reserve_cycle--;
+        }
+
+}
+
+static void
+xlog_grant_add_space_write(struct log *log, int bytes)
+{
+        log->l_grant_write_bytes += bytes;
+        if (log->l_grant_write_bytes > log->l_logsize) {
+                log->l_grant_write_bytes -= log->l_logsize;
+                log->l_grant_write_cycle++;
+        }
+}
+
+static void
+xlog_grant_add_space_reserve(struct log *log, int bytes)
+{
+        log->l_grant_reserve_bytes += bytes;
+        if (log->l_grant_reserve_bytes > log->l_logsize) {
+                log->l_grant_reserve_bytes -= log->l_logsize;
+                log->l_grant_reserve_cycle++;
+        }
+}
+
+static inline void
+xlog_grant_add_space(struct log *log, int bytes)
+{
+        xlog_grant_add_space_write(log, bytes);
+        xlog_grant_add_space_reserve(log, bytes);
+}
+
+
 /*
  * NOTES:
  *
@@ -428,7 +505,7 @@ xfs_log_mount(xfs_mount_t	*mp,
                 if (readonly)
                         vfsp->vfs_flag &= ~VFS_RDONLY;
 
-                error = xlog_recover(mp->m_log, readonly);
+                error = xlog_recover(mp->m_log);
 
                 if (readonly)
                         vfsp->vfs_flag |= VFS_RDONLY;
@@ -1320,8 +1397,7 @@ xlog_sync(xlog_t		*log,
 
         /* move grant heads by roundoff in sync */
         s = GRANT_LOCK(log);
-        XLOG_GRANT_ADD_SPACE(log, roundoff, 'w');
-        XLOG_GRANT_ADD_SPACE(log, roundoff, 'r');
+        xlog_grant_add_space(log, roundoff);
         GRANT_UNLOCK(log, s);
 
         /* put cycle number in every block */
@@ -1515,7 +1591,6 @@ xlog_state_finish_copy(xlog_t		*log,
  * print out info relating to regions written which consume
  * the reservation
  */
-#if defined(XFS_LOG_RES_DEBUG)
 STATIC void
 xlog_print_tic_res(xfs_mount_t *mp, xlog_ticket_t *ticket)
 {
@@ -1605,11 +1680,11 @@ xlog_print_tic_res(xfs_mount_t *mp, xlog_ticket_t *ticket)
                         ticket->t_res_arr_sum, ticket->t_res_o_flow,
                         ticket->t_res_num_ophdrs, ophdr_spc,
                         ticket->t_res_arr_sum + 
-                          ticket->t_res_o_flow + ophdr_spc,
+                        ticket->t_res_o_flow + ophdr_spc,
                         ticket->t_res_num);
 
         for (i = 0; i < ticket->t_res_num; i++) {
-                uint r_type = ticket->t_res_arr[i].r_type; 
+                uint r_type = ticket->t_res_arr[i].r_type; 
                 cmn_err(CE_WARN,
                             "region[%u]: %s - %u bytes\n",
                             i, 
@@ -1618,9 +1693,6 @@ xlog_print_tic_res(xfs_mount_t *mp, xlog_ticket_t *ticket)
                             ticket->t_res_arr[i].r_len);
         }
 }
-#else
-#define xlog_print_tic_res(mp, ticket)
-#endif
 
 /*
  * Write some region out to in-core log
@@ -2389,7 +2461,7 @@ xlog_grant_log_space(xlog_t	   *log,
 
         /* something is already sleeping; insert new transaction at end */
         if (log->l_reserve_headq) {
-                XLOG_INS_TICKETQ(log->l_reserve_headq, tic);
+                xlog_ins_ticketq(&log->l_reserve_headq, tic);
                 xlog_trace_loggrant(log, tic,
                                     "xlog_grant_log_space: sleep 1");
                 /*
@@ -2422,7 +2494,7 @@ redo:
                                      log->l_grant_reserve_bytes);
         if (free_bytes < need_bytes) {
                 if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-                        XLOG_INS_TICKETQ(log->l_reserve_headq, tic);
+                        xlog_ins_ticketq(&log->l_reserve_headq, tic);
                 xlog_trace_loggrant(log, tic,
                                     "xlog_grant_log_space: sleep 2");
                 XFS_STATS_INC(xs_sleep_logspace);
@@ -2439,11 +2511,10 @@ redo:
                 s = GRANT_LOCK(log);
                 goto redo;
         } else if (tic->t_flags & XLOG_TIC_IN_Q)
-                XLOG_DEL_TICKETQ(log->l_reserve_headq, tic);
+                xlog_del_ticketq(&log->l_reserve_headq, tic);
 
         /* we've got enough space */
-        XLOG_GRANT_ADD_SPACE(log, need_bytes, 'w');
-        XLOG_GRANT_ADD_SPACE(log, need_bytes, 'r');
+        xlog_grant_add_space(log, need_bytes);
 #ifdef DEBUG
         tail_lsn = log->l_tail_lsn;
         /*
@@ -2464,7 +2535,7 @@ redo:
 
  error_return:
         if (tic->t_flags & XLOG_TIC_IN_Q)
-                XLOG_DEL_TICKETQ(log->l_reserve_headq, tic);
+                xlog_del_ticketq(&log->l_reserve_headq, tic);
         xlog_trace_loggrant(log, tic, "xlog_grant_log_space: err_ret");
         /*
          * If we are failing, make sure the ticket doesn't have any
@@ -2533,7 +2604,7 @@ xlog_regrant_write_log_space(xlog_t	   *log,
 
                 if (ntic != log->l_write_headq) {
                         if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-                                XLOG_INS_TICKETQ(log->l_write_headq, tic);
+                                xlog_ins_ticketq(&log->l_write_headq, tic);
 
                         xlog_trace_loggrant(log, tic,
                                     "xlog_regrant_write_log_space: sleep 1");
@@ -2565,7 +2636,7 @@ redo:
                                      log->l_grant_write_bytes);
         if (free_bytes < need_bytes) {
                 if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-                        XLOG_INS_TICKETQ(log->l_write_headq, tic);
+                        xlog_ins_ticketq(&log->l_write_headq, tic);
                 XFS_STATS_INC(xs_sleep_logspace);
                 sv_wait(&tic->t_sema, PINOD|PLTWAIT, &log->l_grant_lock, s);
 
@@ -2581,9 +2652,10 @@ redo:
                 s = GRANT_LOCK(log);
                 goto redo;
         } else if (tic->t_flags & XLOG_TIC_IN_Q)
-                XLOG_DEL_TICKETQ(log->l_write_headq, tic);
+                xlog_del_ticketq(&log->l_write_headq, tic);
 
-        XLOG_GRANT_ADD_SPACE(log, need_bytes, 'w'); /* we've got enough space */
+        /* we've got enough space */
+        xlog_grant_add_space_write(log, need_bytes);
 #ifdef DEBUG
         tail_lsn = log->l_tail_lsn;
         if (CYCLE_LSN(tail_lsn) != log->l_grant_write_cycle) {
@@ -2600,7 +2672,7 @@ redo:
 
  error_return:
         if (tic->t_flags & XLOG_TIC_IN_Q)
-                XLOG_DEL_TICKETQ(log->l_reserve_headq, tic);
+                xlog_del_ticketq(&log->l_reserve_headq, tic);
         xlog_trace_loggrant(log, tic, "xlog_regrant_write_log_space: err_ret");
         /*
          * If we are failing, make sure the ticket doesn't have any
@@ -2633,8 +2705,7 @@ xlog_regrant_reserve_log_space(xlog_t	     *log,
                 ticket->t_cnt--;
 
         s = GRANT_LOCK(log);
-        XLOG_GRANT_SUB_SPACE(log, ticket->t_curr_res, 'w');
-        XLOG_GRANT_SUB_SPACE(log, ticket->t_curr_res, 'r');
+        xlog_grant_sub_space(log, ticket->t_curr_res);
         ticket->t_curr_res = ticket->t_unit_res;
         XLOG_TIC_RESET_RES(ticket);
         xlog_trace_loggrant(log, ticket,
@@ -2647,7 +2718,7 @@ xlog_regrant_reserve_log_space(xlog_t	     *log,
                 return;
         }
 
-        XLOG_GRANT_ADD_SPACE(log, ticket->t_unit_res, 'r');
+        xlog_grant_add_space_reserve(log, ticket->t_unit_res);
         xlog_trace_loggrant(log, ticket,
                             "xlog_regrant_reserve_log_space: exit");
         xlog_verify_grant_head(log, 0);
@@ -2683,8 +2754,7 @@ xlog_ungrant_log_space(xlog_t	     *log,
         s = GRANT_LOCK(log);
         xlog_trace_loggrant(log, ticket, "xlog_ungrant_log_space: enter");
 
-        XLOG_GRANT_SUB_SPACE(log, ticket->t_curr_res, 'w');
-        XLOG_GRANT_SUB_SPACE(log, ticket->t_curr_res, 'r');
+        xlog_grant_sub_space(log, ticket->t_curr_res);
 
         xlog_trace_loggrant(log, ticket, "xlog_ungrant_log_space: sub current");
 
@@ -2693,8 +2763,7 @@ xlog_ungrant_log_space(xlog_t	     *log,
          */
         if (ticket->t_cnt > 0) {
                 ASSERT(ticket->t_flags & XLOG_TIC_PERM_RESERV);
-                XLOG_GRANT_SUB_SPACE(log, ticket->t_unit_res*ticket->t_cnt,'w');
-                XLOG_GRANT_SUB_SPACE(log, ticket->t_unit_res*ticket->t_cnt,'r');
+                xlog_grant_sub_space(log, ticket->t_unit_res*ticket->t_cnt);
         }
 
         xlog_trace_loggrant(log, ticket, "xlog_ungrant_log_space: exit");
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index f40d4391fcfc..4b2ac88dbb83 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -96,7 +96,6 @@ static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
 
 
 /* Region types for iovec's i_type */
-#if defined(XFS_LOG_RES_DEBUG)
 #define XLOG_REG_TYPE_BFORMAT           1
 #define XLOG_REG_TYPE_BCHUNK            2
 #define XLOG_REG_TYPE_EFI_FORMAT        3
@@ -117,21 +116,13 @@ static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
 #define XLOG_REG_TYPE_COMMIT            18
 #define XLOG_REG_TYPE_TRANSHDR          19
 #define XLOG_REG_TYPE_MAX               19
-#endif
 
-#if defined(XFS_LOG_RES_DEBUG)
 #define XLOG_VEC_SET_TYPE(vecp, t) ((vecp)->i_type = (t))
-#else
-#define XLOG_VEC_SET_TYPE(vecp, t)
-#endif
-
 
 typedef struct xfs_log_iovec {
         xfs_caddr_t             i_addr;         /* beginning address of region */
         int             i_len;          /* length in bytes of region */
-#if defined(XFS_LOG_RES_DEBUG)
-        uint            i_type;         /* type of region */
-#endif
+        uint            i_type;         /* type of region */
 } xfs_log_iovec_t;
 
 typedef void* xfs_log_ticket_t;
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 4518b188ade6..34bcbf50789c 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -253,7 +253,6 @@ typedef __uint32_t xlog_tid_t;
 
 
 /* Ticket reservation region accounting */ 
-#if defined(XFS_LOG_RES_DEBUG)
 #define XLOG_TIC_LEN_MAX        15
 #define XLOG_TIC_RESET_RES(t) ((t)->t_res_num = \
                                 (t)->t_res_arr_sum = (t)->t_res_num_ophdrs = 0)
@@ -278,15 +277,9 @@ typedef __uint32_t xlog_tid_t;
  * we don't care about.
  */
 typedef struct xlog_res {
-        uint    r_len;
-        uint    r_type;
+        uint    r_len;  /* region length                :4 */
+        uint    r_type; /* region's transaction type    :4 */
 } xlog_res_t;
-#else
-#define XLOG_TIC_RESET_RES(t)
-#define XLOG_TIC_ADD_OPHDR(t)
-#define XLOG_TIC_ADD_REGION(t, len, type)
-#endif
-
 
 typedef struct xlog_ticket {
         sv_t               t_sema;       /* sleep on this semaphore      : 20 */
@@ -301,14 +294,12 @@ typedef struct xlog_ticket {
         char               t_flags;      /* properties of reservation    : 1  */
         uint               t_trans_type; /* transaction type             : 4  */
 
-#if defined (XFS_LOG_RES_DEBUG)
         /* reservation array fields */
         uint               t_res_num;                    /* num in array : 4 */
-        xlog_res_t         t_res_arr[XLOG_TIC_LEN_MAX];  /* array of res : X */ 
         uint               t_res_num_ophdrs;             /* num op hdrs  : 4 */
         uint               t_res_arr_sum;                /* array sum    : 4 */
         uint               t_res_o_flow;                 /* sum overflow : 4 */
-#endif
+        xlog_res_t         t_res_arr[XLOG_TIC_LEN_MAX];  /* array of res : 8 * 15 */ 
 } xlog_ticket_t;
 
 #endif
@@ -494,71 +485,13 @@ typedef struct log {
 
 #define XLOG_FORCED_SHUTDOWN(log)       ((log)->l_flags & XLOG_IO_ERROR)
 
-#define XLOG_GRANT_SUB_SPACE(log,bytes,type)                            \
-    {                                                                   \
-        if (type == 'w') {                                              \
-                (log)->l_grant_write_bytes -= (bytes);                  \
-                if ((log)->l_grant_write_bytes < 0) {                   \
-                        (log)->l_grant_write_bytes += (log)->l_logsize; \
-                        (log)->l_grant_write_cycle--;                   \
-                }                                                       \
-        } else {                                                        \
-                (log)->l_grant_reserve_bytes -= (bytes);                \
-                if ((log)->l_grant_reserve_bytes < 0) {                 \
-                        (log)->l_grant_reserve_bytes += (log)->l_logsize;\
-                        (log)->l_grant_reserve_cycle--;                 \
-                }                                                       \
-         }                                                              \
-    }
-#define XLOG_GRANT_ADD_SPACE(log,bytes,type)                            \
-    {                                                                   \
-        if (type == 'w') {                                              \
-                (log)->l_grant_write_bytes += (bytes);                  \
-                if ((log)->l_grant_write_bytes > (log)->l_logsize) {    \
-                        (log)->l_grant_write_bytes -= (log)->l_logsize; \
-                        (log)->l_grant_write_cycle++;                   \
-                }                                                       \
-        } else {                                                        \
-                (log)->l_grant_reserve_bytes += (bytes);                \
-                if ((log)->l_grant_reserve_bytes > (log)->l_logsize) {  \
-                        (log)->l_grant_reserve_bytes -= (log)->l_logsize;\
-                        (log)->l_grant_reserve_cycle++;                 \
-                }                                                       \
-         }                                                              \
-    }
-#define XLOG_INS_TICKETQ(q, tic)                        \
-    {                                                   \
-        if (q) {                                        \
-                (tic)->t_next       = (q);              \
-                (tic)->t_prev       = (q)->t_prev;      \
-                (q)->t_prev->t_next = (tic);            \
-                (q)->t_prev         = (tic);            \
-        } else {                                        \
-                (tic)->t_prev = (tic)->t_next = (tic);  \
-                (q) = (tic);                            \
-        }                                               \
-        (tic)->t_flags |= XLOG_TIC_IN_Q;                \
-    }
-#define XLOG_DEL_TICKETQ(q, tic)                        \
-    {                                                   \
-        if ((tic) == (tic)->t_next) {                   \
-                (q) = NULL;                             \
-        } else {                                        \
-                (q) = (tic)->t_next;                    \
-                (tic)->t_next->t_prev = (tic)->t_prev;  \
-                (tic)->t_prev->t_next = (tic)->t_next;  \
-        }                                               \
-        (tic)->t_next = (tic)->t_prev = NULL;           \
-        (tic)->t_flags &= ~XLOG_TIC_IN_Q;               \
-    }
 
 /* common routines */
 extern xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
 extern int       xlog_find_tail(xlog_t  *log,
                                 xfs_daddr_t *head_blk,
-                                xfs_daddr_t *tail_blk,
-                                int readonly);
-extern int       xlog_recover(xlog_t *log, int readonly);
+                                xfs_daddr_t *tail_blk);
+extern int       xlog_recover(xlog_t *log);
 extern int       xlog_recover_finish(xlog_t *log, int mfsi_flags);
 extern void      xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
 extern void      xlog_recover_process_iunlinks(xlog_t *log);
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 8ab7df768063..7d46cbd6a07a 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -783,8 +783,7 @@ int
 xlog_find_tail(
         xlog_t                  *log,
         xfs_daddr_t             *head_blk,
-        xfs_daddr_t             *tail_blk,
-        int                     readonly)
+        xfs_daddr_t             *tail_blk)
 {
         xlog_rec_header_t       *rhead;
         xlog_op_header_t        *op_head;
@@ -2563,10 +2562,12 @@ xlog_recover_do_quotaoff_trans(
 
         /*
          * The logitem format's flag tells us if this was user quotaoff,
-         * group quotaoff or both.
+         * group/project quotaoff or both.
          */
         if (qoff_f->qf_flags & XFS_UQUOTA_ACCT)
                 log->l_quotaoffs_flag |= XFS_DQ_USER;
+        if (qoff_f->qf_flags & XFS_PQUOTA_ACCT)
+                log->l_quotaoffs_flag |= XFS_DQ_PROJ;
         if (qoff_f->qf_flags & XFS_GQUOTA_ACCT)
                 log->l_quotaoffs_flag |= XFS_DQ_GROUP;
 
@@ -3890,14 +3891,13 @@ xlog_do_recover(
  */
 int
 xlog_recover(
-        xlog_t          *log,
-        int             readonly)
+        xlog_t          *log)
 {
         xfs_daddr_t     head_blk, tail_blk;
         int             error;
 
         /* find the tail of the log */
-        if ((error = xlog_find_tail(log, &head_blk, &tail_blk, readonly)))
+        if ((error = xlog_find_tail(log, &head_blk, &tail_blk)))
                 return error;
 
         if (tail_blk != head_blk) {
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 303af86739bf..6088e14f84e3 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -51,7 +51,7 @@ STATIC int	xfs_uuid_mount(xfs_mount_t *);
 STATIC void     xfs_uuid_unmount(xfs_mount_t *mp);
 STATIC void     xfs_unmountfs_wait(xfs_mount_t *);
 
-static struct {
+static const struct {
     short offset;
     short type;     /* 0 = integer
                 * 1 = binary / string (no translation)
@@ -1077,8 +1077,7 @@ xfs_unmountfs(xfs_mount_t *mp, struct cred *cr)
 
         xfs_iflush_all(mp);
 
-        XFS_QM_DQPURGEALL(mp,
-                XFS_QMOPT_UQUOTA | XFS_QMOPT_GQUOTA | XFS_QMOPT_UMOUNTING);
+        XFS_QM_DQPURGEALL(mp, XFS_QMOPT_QUOTALL | XFS_QMOPT_UMOUNTING);
 
         /*
          * Flush out the log synchronously so that we know for sure
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index 3432fd5a3986..cd3cf9613a00 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -308,7 +308,6 @@ typedef struct xfs_mount {
         xfs_buftarg_t           *m_ddev_targp;  /* saves taking the address */
         xfs_buftarg_t           *m_logdev_targp;/* ptr to log device */
         xfs_buftarg_t           *m_rtdev_targp; /* ptr to rt device */
-#define m_dev           m_ddev_targp->pbr_dev
         __uint8_t               m_dircook_elog; /* log d-cookie entry bits */
         __uint8_t               m_blkbit_log;   /* blocklog + NBBY */
         __uint8_t               m_blkbb_log;    /* blocklog - BBSHIFT */
@@ -393,7 +392,7 @@ typedef struct xfs_mount {
                                                    user */
 #define XFS_MOUNT_NOALIGN       (1ULL << 7)     /* turn off stripe alignment
                                                    allocations */
-#define XFS_MOUNT_COMPAT_ATTR   (1ULL << 8)     /* do not use attr2 format */
+#define XFS_MOUNT_ATTR2         (1ULL << 8)     /* allow use of attr2 format */
                              /* (1ULL << 9)     -- currently unused */
 #define XFS_MOUNT_NORECOVERY    (1ULL << 10)    /* no recovery - dirty fs */
 #define XFS_MOUNT_SHARED        (1ULL << 11)    /* shared mount */
diff --git a/fs/xfs/xfs_rename.c b/fs/xfs/xfs_rename.c
index 4d4e8f4e768e..81a05cfd77d2 100644
--- a/fs/xfs/xfs_rename.c
+++ b/fs/xfs/xfs_rename.c
@@ -243,7 +243,6 @@ xfs_rename(
         xfs_inode_t     *inodes[4];
         int             target_ip_dropped = 0;  /* dropped target_ip link? */
         vnode_t         *src_dir_vp;
-        bhv_desc_t      *target_dir_bdp;
         int             spaceres;
         int             target_link_zero = 0;
         int             num_inodes;
@@ -260,14 +259,12 @@ xfs_rename(
          * Find the XFS behavior descriptor for the target directory
          * vnode since it was not handed to us.
          */
-        target_dir_bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(target_dir_vp),
-                                                &xfs_vnodeops);
-        if (target_dir_bdp == NULL) {
+        target_dp = xfs_vtoi(target_dir_vp);
+        if (target_dp == NULL) {
                 return XFS_ERROR(EXDEV);
         }
 
         src_dp = XFS_BHVTOI(src_dir_bdp);
-        target_dp = XFS_BHVTOI(target_dir_bdp);
         mp = src_dp->i_mount;
 
         if (DM_EVENT_ENABLED(src_dir_vp->v_vfsp, src_dp, DM_EVENT_RENAME) ||
diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
index c4b20872f07d..a59c102cf214 100644
--- a/fs/xfs/xfs_rw.c
+++ b/fs/xfs/xfs_rw.c
@@ -238,6 +238,7 @@ xfs_bioerror_relse(
         }
         return (EIO);
 }
+
 /*
  * Prints out an ALERT message about I/O error.
  */
@@ -252,11 +253,9 @@ xfs_ioerror_alert(
  "I/O error in filesystem (\"%s\") meta-data dev %s block 0x%llx"
  "       (\"%s\") error %d buf count %zd",
                 (!mp || !mp->m_fsname) ? "(fs name not set)" : mp->m_fsname,
-                XFS_BUFTARG_NAME(bp->pb_target),
-                (__uint64_t)blkno,
-                func,
-                XFS_BUF_GETERROR(bp),
-                XFS_BUF_COUNT(bp));
+                XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
+                (__uint64_t)blkno, func,
+                XFS_BUF_GETERROR(bp), XFS_BUF_COUNT(bp));
 }
 
 /*
diff --git a/fs/xfs/xfs_sb.h b/fs/xfs/xfs_sb.h
index 4a17d335f897..bf168a91ddb8 100644
--- a/fs/xfs/xfs_sb.h
+++ b/fs/xfs/xfs_sb.h
@@ -68,18 +68,6 @@ struct xfs_mount;
         (XFS_SB_VERSION_NUMBITS | \
          XFS_SB_VERSION_OKREALFBITS | \
          XFS_SB_VERSION_OKSASHFBITS)
-#define XFS_SB_VERSION_MKFS(ia,dia,extflag,dirv2,na,sflag,morebits)     \
-        (((ia) || (dia) || (extflag) || (dirv2) || (na) || (sflag) || \
-          (morebits)) ? \
-                (XFS_SB_VERSION_4 | \
-                 ((ia) ? XFS_SB_VERSION_ALIGNBIT : 0) | \
-                 ((dia) ? XFS_SB_VERSION_DALIGNBIT : 0) | \
-                 ((extflag) ? XFS_SB_VERSION_EXTFLGBIT : 0) | \
-                 ((dirv2) ? XFS_SB_VERSION_DIRV2BIT : 0) | \
-                 ((na) ? XFS_SB_VERSION_LOGV2BIT : 0) | \
-                 ((sflag) ? XFS_SB_VERSION_SECTORBIT : 0) | \
-                 ((morebits) ? XFS_SB_VERSION_MOREBITSBIT : 0)) : \
-                XFS_SB_VERSION_1)
 
 /*
  * There are two words to hold XFS "feature" bits: the original
@@ -105,11 +93,6 @@ struct xfs_mount;
         (XFS_SB_VERSION2_OKREALFBITS |  \
          XFS_SB_VERSION2_OKSASHFBITS )
 
-/*
- * mkfs macro to set up sb_features2 word
- */
-#define XFS_SB_VERSION2_MKFS(resvd1, sbcntr)    0
-
 typedef struct xfs_sb
 {
         __uint32_t      sb_magicnum;    /* magic number == XFS_SB_MAGIC */
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 279e043d7323..d3d714e6b32a 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -1014,6 +1014,7 @@ xfs_trans_cancel(
         xfs_log_item_t          *lip;
         int                     i;
 #endif
+        xfs_mount_t             *mp = tp->t_mountp;
 
         /*
          * See if the caller is being too lazy to figure out if
@@ -1026,9 +1027,10 @@ xfs_trans_cancel(
          * filesystem.  This happens in paths where we detect
          * corruption and decide to give up.
          */
-        if ((tp->t_flags & XFS_TRANS_DIRTY) &&
-            !XFS_FORCED_SHUTDOWN(tp->t_mountp))
-                xfs_force_shutdown(tp->t_mountp, XFS_CORRUPT_INCORE);
+        if ((tp->t_flags & XFS_TRANS_DIRTY) && !XFS_FORCED_SHUTDOWN(mp)) {
+                XFS_ERROR_REPORT("xfs_trans_cancel", XFS_ERRLEVEL_LOW, mp);
+                xfs_force_shutdown(mp, XFS_CORRUPT_INCORE);
+        }
 #ifdef DEBUG
         if (!(flags & XFS_TRANS_ABORT)) {
                 licp = &(tp->t_items);
@@ -1040,7 +1042,7 @@ xfs_trans_cancel(
                                 }
 
                                 lip = lidp->lid_item;
-                                if (!XFS_FORCED_SHUTDOWN(tp->t_mountp))
+                                if (!XFS_FORCED_SHUTDOWN(mp))
                                         ASSERT(!(lip->li_type == XFS_LI_EFD));
                         }
                         licp = licp->lic_next;
@@ -1048,7 +1050,7 @@ xfs_trans_cancel(
         }
 #endif
         xfs_trans_unreserve_and_mod_sb(tp);
-        XFS_TRANS_UNRESERVE_AND_MOD_DQUOTS(tp->t_mountp, tp);
+        XFS_TRANS_UNRESERVE_AND_MOD_DQUOTS(mp, tp);
 
         if (tp->t_ticket) {
                 if (flags & XFS_TRANS_RELEASE_LOG_RES) {
@@ -1057,7 +1059,7 @@ xfs_trans_cancel(
                 } else {
                         log_flags = 0;
                 }
-                xfs_log_done(tp->t_mountp, tp->t_ticket, NULL, log_flags);
+                xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
         }
 
         /* mark this thread as no longer being in a transaction */
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index a889963fdd14..d77901c07f63 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -973,7 +973,6 @@ void		xfs_trans_bhold(xfs_trans_t *, struct xfs_buf *);
 void            xfs_trans_bhold_release(xfs_trans_t *, struct xfs_buf *);
 void            xfs_trans_binval(xfs_trans_t *, struct xfs_buf *);
 void            xfs_trans_inode_buf(xfs_trans_t *, struct xfs_buf *);
-void            xfs_trans_inode_buf(xfs_trans_t *, struct xfs_buf *);
 void            xfs_trans_stale_inode_buf(xfs_trans_t *, struct xfs_buf *);
 void            xfs_trans_dquot_buf(xfs_trans_t *, struct xfs_buf *, uint);
 void            xfs_trans_inode_alloc_buf(xfs_trans_t *, struct xfs_buf *);
diff --git a/fs/xfs/xfs_utils.c b/fs/xfs/xfs_utils.c
index fefe1d60377f..34654ec6ae10 100644
--- a/fs/xfs/xfs_utils.c
+++ b/fs/xfs/xfs_utils.c
@@ -55,16 +55,13 @@ xfs_get_dir_entry(
         xfs_inode_t     **ipp)
 {
         vnode_t         *vp;
-        bhv_desc_t      *bdp;
 
         vp = VNAME_TO_VNODE(dentry);
-        bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(vp), &xfs_vnodeops);
-        if (!bdp) {
-                *ipp = NULL;
+
+        *ipp = xfs_vtoi(vp);
+        if (!*ipp)
                 return XFS_ERROR(ENOENT);
-        }
         VN_HOLD(vp);
-        *ipp = XFS_BHVTOI(bdp);
         return 0;
 }
 
diff --git a/fs/xfs/xfs_vfsops.c b/fs/xfs/xfs_vfsops.c
index 7bdbd991ab1c..b6ad370fab3d 100644
--- a/fs/xfs/xfs_vfsops.c
+++ b/fs/xfs/xfs_vfsops.c
@@ -53,6 +53,7 @@
 #include "xfs_acl.h"
 #include "xfs_attr.h"
 #include "xfs_clnt.h"
+#include "xfs_fsops.h"
 
 STATIC int xfs_sync(bhv_desc_t *, int, cred_t *);
 
@@ -290,8 +291,8 @@ xfs_start_flags(
                 mp->m_flags |= XFS_MOUNT_IDELETE;
         if (ap->flags & XFSMNT_DIRSYNC)
                 mp->m_flags |= XFS_MOUNT_DIRSYNC;
-        if (ap->flags & XFSMNT_COMPAT_ATTR)
-                mp->m_flags |= XFS_MOUNT_COMPAT_ATTR;
+        if (ap->flags & XFSMNT_ATTR2)
+                mp->m_flags |= XFS_MOUNT_ATTR2;
 
         if (ap->flags2 & XFSMNT2_COMPAT_IOSIZE)
                 mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
@@ -312,6 +313,8 @@ xfs_start_flags(
                 mp->m_flags |= XFS_MOUNT_NOUUID;
         if (ap->flags & XFSMNT_BARRIER)
                 mp->m_flags |= XFS_MOUNT_BARRIER;
+        else
+                mp->m_flags &= ~XFS_MOUNT_BARRIER;
 
         return 0;
 }
@@ -330,10 +333,11 @@ xfs_finish_flags(
 
         /* Fail a mount where the logbuf is smaller then the log stripe */
         if (XFS_SB_VERSION_HASLOGV2(&mp->m_sb)) {
-                if ((ap->logbufsize == -1) &&
+                if ((ap->logbufsize <= 0) &&
                     (mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE)) {
                         mp->m_logbsize = mp->m_sb.sb_logsunit;
-                } else if (ap->logbufsize < mp->m_sb.sb_logsunit) {
+                } else if (ap->logbufsize > 0 &&
+                           ap->logbufsize < mp->m_sb.sb_logsunit) {
                         cmn_err(CE_WARN,
         "XFS: logbuf size must be greater than or equal to log stripe size");
                         return XFS_ERROR(EINVAL);
@@ -347,6 +351,10 @@ xfs_finish_flags(
                 }
         }
 
+        if (XFS_SB_VERSION_HASATTR2(&mp->m_sb)) {
+                mp->m_flags |= XFS_MOUNT_ATTR2;
+        }
+
         /*
          * prohibit r/w mounts of read-only filesystems
          */
@@ -382,10 +390,6 @@ xfs_finish_flags(
                         return XFS_ERROR(EINVAL);
         }
 
-        if (XFS_SB_VERSION_HASATTR2(&mp->m_sb)) {
-                mp->m_flags &= ~XFS_MOUNT_COMPAT_ATTR;
-        }
-
         return 0;
 }
 
@@ -504,13 +508,13 @@ xfs_mount(
         if (error)
                 goto error2;
 
+        if ((mp->m_flags & XFS_MOUNT_BARRIER) && !(vfsp->vfs_flag & VFS_RDONLY))
+                xfs_mountfs_check_barriers(mp);
+
         error = XFS_IOINIT(vfsp, args, flags);
         if (error)
                 goto error2;
 
-        if ((args->flags & XFSMNT_BARRIER) &&
-            !(XFS_MTOVFS(mp)->vfs_flag & VFS_RDONLY))
-                xfs_mountfs_check_barriers(mp);
         return 0;
 
 error2:
@@ -655,6 +659,11 @@ xfs_mntupdate(
         else
                 mp->m_flags &= ~XFS_MOUNT_NOATIME;
 
+        if (args->flags & XFSMNT_BARRIER)
+                mp->m_flags |= XFS_MOUNT_BARRIER;
+        else
+                mp->m_flags &= ~XFS_MOUNT_BARRIER;
+
         if ((vfsp->vfs_flag & VFS_RDONLY) &&
             !(*flags & MS_RDONLY)) {
                 vfsp->vfs_flag &= ~VFS_RDONLY;
@@ -1634,6 +1643,7 @@ xfs_vget(
 #define MNTOPT_NORECOVERY   "norecovery"   /* don't run XFS recovery */
 #define MNTOPT_BARRIER  "barrier"       /* use writer barriers for log write and
                                          * unwritten extent conversion */
+#define MNTOPT_NOBARRIER "nobarrier"    /* .. disable */
 #define MNTOPT_OSYNCISOSYNC "osyncisosync" /* o_sync is REALLY o_sync */
 #define MNTOPT_64BITINODE   "inode64"   /* inodes can be allocated anywhere */
 #define MNTOPT_IKEEP    "ikeep"         /* do not free empty inode clusters */
@@ -1680,7 +1690,6 @@ xfs_parseargs(
         int                     iosize;
 
         args->flags2 |= XFSMNT2_COMPAT_IOSIZE;
-        args->flags |= XFSMNT_COMPAT_ATTR;
 
 #if 0   /* XXX: off by default, until some remaining issues ironed out */
         args->flags |= XFSMNT_IDELETE; /* default to on */
@@ -1806,6 +1815,8 @@ xfs_parseargs(
                         args->flags |= XFSMNT_NOUUID;
                 } else if (!strcmp(this_char, MNTOPT_BARRIER)) {
                         args->flags |= XFSMNT_BARRIER;
+                } else if (!strcmp(this_char, MNTOPT_NOBARRIER)) {
+                        args->flags &= ~XFSMNT_BARRIER;
                 } else if (!strcmp(this_char, MNTOPT_IKEEP)) {
                         args->flags &= ~XFSMNT_IDELETE;
                 } else if (!strcmp(this_char, MNTOPT_NOIKEEP)) {
@@ -1815,9 +1826,9 @@ xfs_parseargs(
                 } else if (!strcmp(this_char, MNTOPT_NOLARGEIO)) {
                         args->flags2 |= XFSMNT2_COMPAT_IOSIZE;
                 } else if (!strcmp(this_char, MNTOPT_ATTR2)) {
-                        args->flags &= ~XFSMNT_COMPAT_ATTR;
+                        args->flags |= XFSMNT_ATTR2;
                 } else if (!strcmp(this_char, MNTOPT_NOATTR2)) {
-                        args->flags |= XFSMNT_COMPAT_ATTR;
+                        args->flags &= ~XFSMNT_ATTR2;
                 } else if (!strcmp(this_char, "osyncisdsync")) {
                         /* no-op, this is now the default */
 printk("XFS: osyncisdsync is now the default, option is deprecated.\n");
@@ -1892,7 +1903,6 @@ xfs_showargs(
                 { XFS_MOUNT_NOUUID,             "," MNTOPT_NOUUID },
                 { XFS_MOUNT_NORECOVERY,         "," MNTOPT_NORECOVERY },
                 { XFS_MOUNT_OSYNCISOSYNC,       "," MNTOPT_OSYNCISOSYNC },
-                { XFS_MOUNT_BARRIER,            "," MNTOPT_BARRIER },
                 { XFS_MOUNT_IDELETE,            "," MNTOPT_NOIKEEP },
                 { 0, NULL }
         };
@@ -1914,33 +1924,28 @@ xfs_showargs(
 
         if (mp->m_logbufs > 0)
                 seq_printf(m, "," MNTOPT_LOGBUFS "=%d", mp->m_logbufs);
-
         if (mp->m_logbsize > 0)
                 seq_printf(m, "," MNTOPT_LOGBSIZE "=%dk", mp->m_logbsize >> 10);
 
         if (mp->m_logname)
                 seq_printf(m, "," MNTOPT_LOGDEV "=%s", mp->m_logname);
-
         if (mp->m_rtname)
                 seq_printf(m, "," MNTOPT_RTDEV "=%s", mp->m_rtname);
 
         if (mp->m_dalign > 0)
                 seq_printf(m, "," MNTOPT_SUNIT "=%d",
                                 (int)XFS_FSB_TO_BB(mp, mp->m_dalign));
-
         if (mp->m_swidth > 0)
                 seq_printf(m, "," MNTOPT_SWIDTH "=%d",
                                 (int)XFS_FSB_TO_BB(mp, mp->m_swidth));
 
-        if (!(mp->m_flags & XFS_MOUNT_COMPAT_ATTR))
-                seq_printf(m, "," MNTOPT_ATTR2);
-
         if (!(mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE))
                 seq_printf(m, "," MNTOPT_LARGEIO);
+        if (mp->m_flags & XFS_MOUNT_BARRIER)
+                seq_printf(m, "," MNTOPT_BARRIER);
 
         if (!(vfsp->vfs_flag & VFS_32BITINODES))
                 seq_printf(m, "," MNTOPT_64BITINODE);
-
         if (vfsp->vfs_flag & VFS_GRPID)
                 seq_printf(m, "," MNTOPT_GRPID);
 
@@ -1959,6 +1964,7 @@ xfs_freeze(
         /* Push the superblock and write an unmount record */
         xfs_log_unmount_write(mp);
         xfs_unmountfs_writesb(mp);
+        xfs_fs_log_dummy(mp);
 }
 
 
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index e92cacde02f5..8076cc981e11 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -185,8 +185,7 @@ xfs_getattr(
                 break;
         }
 
-        vap->va_atime.tv_sec = ip->i_d.di_atime.t_sec;
-        vap->va_atime.tv_nsec = ip->i_d.di_atime.t_nsec;
+        vn_atime_to_timespec(vp, &vap->va_atime);
         vap->va_mtime.tv_sec = ip->i_d.di_mtime.t_sec;
         vap->va_mtime.tv_nsec = ip->i_d.di_mtime.t_nsec;
         vap->va_ctime.tv_sec = ip->i_d.di_ctime.t_sec;
@@ -544,24 +543,6 @@ xfs_setattr(
                 }
 
                 /*
-                 * Can't set extent size unless the file is marked, or
-                 * about to be marked as a realtime file.
-                 *
-                 * This check will be removed when fixed size extents
-                 * with buffered data writes is implemented.
-                 *
-                 */
-                if ((mask & XFS_AT_EXTSIZE)                     &&
-                    ((ip->i_d.di_extsize << mp->m_sb.sb_blocklog) !=
-                     vap->va_extsize) &&
-                    (!((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ||
-                       ((mask & XFS_AT_XFLAGS) &&
-                        (vap->va_xflags & XFS_XFLAG_REALTIME))))) {
-                        code = XFS_ERROR(EINVAL);
-                        goto error_return;
-                }
-
-                /*
                  * Can't change realtime flag if any extents are allocated.
                  */
                 if ((ip->i_d.di_nextents || ip->i_delayed_blks) &&
@@ -823,13 +804,17 @@ xfs_setattr(
                                         di_flags |= XFS_DIFLAG_RTINHERIT;
                                 if (vap->va_xflags & XFS_XFLAG_NOSYMLINKS)
                                         di_flags |= XFS_DIFLAG_NOSYMLINKS;
-                        } else {
+                                if (vap->va_xflags & XFS_XFLAG_EXTSZINHERIT)
+                                        di_flags |= XFS_DIFLAG_EXTSZINHERIT;
+                        } else if ((ip->i_d.di_mode & S_IFMT) == S_IFREG) {
                                 if (vap->va_xflags & XFS_XFLAG_REALTIME) {
                                         di_flags |= XFS_DIFLAG_REALTIME;
                                         ip->i_iocore.io_flags |= XFS_IOCORE_RT;
                                 } else {
                                         ip->i_iocore.io_flags &= ~XFS_IOCORE_RT;
                                 }
+                                if (vap->va_xflags & XFS_XFLAG_EXTSIZE)
+                                        di_flags |= XFS_DIFLAG_EXTSIZE;
                         }
                         ip->i_d.di_flags = di_flags;
                 }
@@ -999,10 +984,6 @@ xfs_readlink(
                 goto error_return;
         }
 
-        if (!(ioflags & IO_INVIS)) {
-                xfs_ichgtime(ip, XFS_ICHGTIME_ACC);
-        }
-
         /*
          * See if the symlink is stored inline.
          */
@@ -1234,7 +1215,8 @@ xfs_inactive_free_eofblocks(
         xfs_iunlock(ip, XFS_ILOCK_SHARED);
 
         if (!error && (nimaps != 0) &&
-            (imap.br_startblock != HOLESTARTBLOCK)) {
+            (imap.br_startblock != HOLESTARTBLOCK ||
+             ip->i_delayed_blks)) {
                 /*
                  * Attach the dquots to the inode up front.
                  */
@@ -1569,9 +1551,11 @@ xfs_release(
 
         if (ip->i_d.di_nlink != 0) {
                 if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
-                     ((ip->i_d.di_size > 0) || (VN_CACHED(vp) > 0)) &&
+                     ((ip->i_d.di_size > 0) || (VN_CACHED(vp) > 0 ||
+                       ip->i_delayed_blks > 0)) &&
                      (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
-                    (!(ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC|XFS_DIFLAG_APPEND)))) {
+                    (!(ip->i_d.di_flags &
+                                (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
                         if ((error = xfs_inactive_free_eofblocks(mp, ip)))
                                 return (error);
                         /* Update linux inode block count after free above */
@@ -1628,7 +1612,8 @@ xfs_inactive(
          * only one with a reference to the inode.
          */
         truncate = ((ip->i_d.di_nlink == 0) &&
-            ((ip->i_d.di_size != 0) || (ip->i_d.di_nextents > 0)) &&
+            ((ip->i_d.di_size != 0) || (ip->i_d.di_nextents > 0) ||
+             (ip->i_delayed_blks > 0)) &&
             ((ip->i_d.di_mode & S_IFMT) == S_IFREG));
 
         mp = ip->i_mount;
@@ -1646,10 +1631,12 @@ xfs_inactive(
 
         if (ip->i_d.di_nlink != 0) {
                 if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
-                     ((ip->i_d.di_size > 0) || (VN_CACHED(vp) > 0)) &&
-                     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
-                    (!(ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC|XFS_DIFLAG_APPEND)) ||
-                     (ip->i_delayed_blks != 0))) {
+                     ((ip->i_d.di_size > 0) || (VN_CACHED(vp) > 0 ||
+                       ip->i_delayed_blks > 0)) &&
+                      (ip->i_df.if_flags & XFS_IFEXTENTS) &&
+                     (!(ip->i_d.di_flags &
+                                (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)) ||
+                      (ip->i_delayed_blks != 0)))) {
                         if ((error = xfs_inactive_free_eofblocks(mp, ip)))
                                 return (VN_INACTIVE_CACHE);
                         /* Update linux inode block count after free above */
@@ -2593,7 +2580,6 @@ xfs_link(
         int                     cancel_flags;
         int                     committed;
         vnode_t                 *target_dir_vp;
-        bhv_desc_t              *src_bdp;
         int                     resblks;
         char                    *target_name = VNAME(dentry);
         int                     target_namelen;
@@ -2606,8 +2592,7 @@ xfs_link(
         if (VN_ISDIR(src_vp))
                 return XFS_ERROR(EPERM);
 
-        src_bdp = vn_bhv_lookup_unlocked(VN_BHV_HEAD(src_vp), &xfs_vnodeops);
-        sip = XFS_BHVTOI(src_bdp);
+        sip = xfs_vtoi(src_vp);
         tdp = XFS_BHVTOI(target_dir_bdp);
         mp = tdp->i_mount;
         if (XFS_FORCED_SHUTDOWN(mp))
@@ -3240,7 +3225,6 @@ xfs_readdir(
         xfs_trans_t     *tp = NULL;
         int             error = 0;
         uint            lock_mode;
-        xfs_off_t       start_offset;
 
         vn_trace_entry(BHV_TO_VNODE(dir_bdp), __FUNCTION__,
                                                (inst_t *)__return_address);
@@ -3251,11 +3235,7 @@ xfs_readdir(
         }
 
         lock_mode = xfs_ilock_map_shared(dp);
-        start_offset = uiop->uio_offset;
         error = XFS_DIR_GETDENTS(dp->i_mount, tp, dp, uiop, eofp);
-        if (start_offset != uiop->uio_offset) {
-                xfs_ichgtime(dp, XFS_ICHGTIME_ACC);
-        }
         xfs_iunlock_map_shared(dp, lock_mode);
         return error;
 }
@@ -3832,7 +3812,12 @@ xfs_reclaim(
         vn_iowait(vp);
 
         ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);
-        ASSERT(VN_CACHED(vp) == 0);
+
+        /*
+         * Make sure the atime in the XFS inode is correct before freeing the
+         * Linux inode.
+         */
+        xfs_synchronize_atime(ip);
 
         /* If we have nothing to flush with this inode then complete the
          * teardown now, otherwise break the link between the xfs inode
@@ -4002,42 +3987,36 @@ xfs_alloc_file_space(
         int                     alloc_type,
         int                     attr_flags)
 {
+        xfs_mount_t             *mp = ip->i_mount;
+        xfs_off_t               count;
         xfs_filblks_t           allocated_fsb;
         xfs_filblks_t           allocatesize_fsb;
-        int                     committed;
-        xfs_off_t               count;
-        xfs_filblks_t           datablocks;
-        int                     error;
+        xfs_extlen_t            extsz, temp;
+        xfs_fileoff_t           startoffset_fsb;
         xfs_fsblock_t           firstfsb;
-        xfs_bmap_free_t         free_list;
-        xfs_bmbt_irec_t         *imapp;
-        xfs_bmbt_irec_t         imaps[1];
-        xfs_mount_t             *mp;
-        int                     numrtextents;
-        int                     reccount;
-        uint                    resblks;
+        int                     nimaps;
+        int                     bmapi_flag;
+        int                     quota_flag;
         int                     rt;
-        int                     rtextsize;
-        xfs_fileoff_t           startoffset_fsb;
         xfs_trans_t             *tp;
-        int                     xfs_bmapi_flags;
+        xfs_bmbt_irec_t         imaps[1], *imapp;
+        xfs_bmap_free_t         free_list;
+        uint                    qblocks, resblks, resrtextents;
+        int                     committed;
+        int                     error;
 
         vn_trace_entry(XFS_ITOV(ip), __FUNCTION__, (inst_t *)__return_address);
-        mp = ip->i_mount;
 
         if (XFS_FORCED_SHUTDOWN(mp))
                 return XFS_ERROR(EIO);
 
-        /*
-         * determine if this is a realtime file
-         */
-        if ((rt = XFS_IS_REALTIME_INODE(ip)) != 0) {
-                if (ip->i_d.di_extsize)
-                        rtextsize = ip->i_d.di_extsize;
-                else
-                        rtextsize = mp->m_sb.sb_rextsize;
-        } else
-                rtextsize = 0;
+        rt = XFS_IS_REALTIME_INODE(ip);
+        if (unlikely(rt)) {
+                if (!(extsz = ip->i_d.di_extsize))
+                        extsz = mp->m_sb.sb_rextsize;
+        } else {
+                extsz = ip->i_d.di_extsize;
+        }
 
         if ((error = XFS_QM_DQATTACH(mp, ip, 0)))
                 return error;
@@ -4048,8 +4027,8 @@ xfs_alloc_file_space(
         count = len;
         error = 0;
         imapp = &imaps[0];
-        reccount = 1;
-        xfs_bmapi_flags = XFS_BMAPI_WRITE | (alloc_type ? XFS_BMAPI_PREALLOC : 0);
+        nimaps = 1;
+        bmapi_flag = XFS_BMAPI_WRITE | (alloc_type ? XFS_BMAPI_PREALLOC : 0);
         startoffset_fsb = XFS_B_TO_FSBT(mp, offset);
         allocatesize_fsb = XFS_B_TO_FSB(mp, count);
 
@@ -4070,43 +4049,51 @@ xfs_alloc_file_space(
         }
 
         /*
-         * allocate file space until done or until there is an error
+         * Allocate file space until done or until there is an error
          */
 retry:
         while (allocatesize_fsb && !error) {
+                xfs_fileoff_t   s, e;
+
                 /*
-                 * determine if reserving space on
-                 * the data or realtime partition.
+                 * Determine space reservations for data/realtime.
                  */
-                if (rt) {
-                        xfs_fileoff_t s, e;
-
+                if (unlikely(extsz)) {
                         s = startoffset_fsb;
-                        do_div(s, rtextsize);
-                        s *= rtextsize;
-                        e = roundup_64(startoffset_fsb + allocatesize_fsb,
-                                rtextsize);
-                        numrtextents = (int)(e - s) / mp->m_sb.sb_rextsize;
-                        datablocks = 0;
+                        do_div(s, extsz);
+                        s *= extsz;
+                        e = startoffset_fsb + allocatesize_fsb;
+                        if ((temp = do_mod(startoffset_fsb, extsz)))
+                                e += temp;
+                        if ((temp = do_mod(e, extsz)))
+                                e += extsz - temp;
+                } else {
+                        s = 0;
+                        e = allocatesize_fsb;
+                }
+
+                if (unlikely(rt)) {
+                        resrtextents = qblocks = (uint)(e - s);
+                        resrtextents /= mp->m_sb.sb_rextsize;
+                        resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+                        quota_flag = XFS_QMOPT_RES_RTBLKS;
                 } else {
-                        datablocks = allocatesize_fsb;
-                        numrtextents = 0;
+                        resrtextents = 0;
+                        resblks = qblocks = \
+                                XFS_DIOSTRAT_SPACE_RES(mp, (uint)(e - s));
+                        quota_flag = XFS_QMOPT_RES_REGBLKS;
                 }
 
                 /*
-                 * allocate and setup the transaction
+                 * Allocate and setup the transaction.
                  */
                 tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
-                resblks = XFS_DIOSTRAT_SPACE_RES(mp, datablocks);
-                error = xfs_trans_reserve(tp,
-                                          resblks,
-                                          XFS_WRITE_LOG_RES(mp),
-                                          numrtextents,
+                error = xfs_trans_reserve(tp, resblks,
+                                          XFS_WRITE_LOG_RES(mp), resrtextents,
                                           XFS_TRANS_PERM_LOG_RES,
                                           XFS_WRITE_LOG_COUNT);
-
                 /*
-                 * check for running out of space
+                 * Check for running out of space
                  */
                 if (error) {
                         /*
@@ -4117,8 +4104,8 @@ retry:
                         break;
                 }
                 xfs_ilock(ip, XFS_ILOCK_EXCL);
-                error = XFS_TRANS_RESERVE_QUOTA(mp, tp,
-                                ip->i_udquot, ip->i_gdquot, resblks, 0, 0);
+                error = XFS_TRANS_RESERVE_QUOTA_NBLKS(mp, tp, ip,
+                                                      qblocks, 0, quota_flag);
                 if (error)
                         goto error1;
 
@@ -4126,19 +4113,19 @@ retry:
                 xfs_trans_ihold(tp, ip);
 
                 /*
-                 * issue the bmapi() call to allocate the blocks
+                 * Issue the xfs_bmapi() call to allocate the blocks
                  */
                 XFS_BMAP_INIT(&free_list, &firstfsb);
                 error = xfs_bmapi(tp, ip, startoffset_fsb,
-                                  allocatesize_fsb, xfs_bmapi_flags,
-                                  &firstfsb, 0, imapp, &reccount,
+                                  allocatesize_fsb, bmapi_flag,
+                                  &firstfsb, 0, imapp, &nimaps,
                                   &free_list);
                 if (error) {
                         goto error0;
                 }
 
                 /*
-                 * complete the transaction
+                 * Complete the transaction
                  */
                 error = xfs_bmap_finish(&tp, &free_list, firstfsb, &committed);
                 if (error) {
@@ -4153,7 +4140,7 @@ retry:
 
                 allocated_fsb = imapp->br_blockcount;
 
-                if (reccount == 0) {
+                if (nimaps == 0) {
                         error = XFS_ERROR(ENOSPC);
                         break;
                 }
@@ -4176,9 +4163,11 @@ dmapi_enospc_check:
 
         return error;
 
- error0:
+error0: /* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
         xfs_bmap_cancel(&free_list);
- error1:
+        XFS_TRANS_UNRESERVE_QUOTA_NBLKS(mp, tp, ip, qblocks, 0, quota_flag);
+
+error1: /* Just cancel transaction */
         xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
         xfs_iunlock(ip, XFS_ILOCK_EXCL);
         goto dmapi_enospc_check;
@@ -4423,8 +4412,8 @@ xfs_free_file_space(
                 }
                 xfs_ilock(ip, XFS_ILOCK_EXCL);
                 error = XFS_TRANS_RESERVE_QUOTA(mp, tp,
-                                ip->i_udquot, ip->i_gdquot, resblks, 0, rt ?
-                                XFS_QMOPT_RES_RTBLKS : XFS_QMOPT_RES_REGBLKS);
+                                ip->i_udquot, ip->i_gdquot, resblks, 0,
+                                XFS_QMOPT_RES_REGBLKS);
                 if (error)
                         goto error1;
 
diff --git a/include/asm-alpha/mmu_context.h b/include/asm-alpha/mmu_context.h
index a714d0cdc204..6f92482cc96c 100644
--- a/include/asm-alpha/mmu_context.h
+++ b/include/asm-alpha/mmu_context.h
@@ -156,7 +156,7 @@ ev5_switch_mm(struct mm_struct *prev_mm, struct mm_struct *next_mm,
         /* Always update the PCB ASN.  Another thread may have allocated
            a new mm->context (via flush_tlb_mm) without the ASN serial
            number wrapping.  We have no way to detect when this is needed.  */
-        next->thread_info->pcb.asn = mmc & HARDWARE_ASN_MASK;
+        task_thread_info(next)->pcb.asn = mmc & HARDWARE_ASN_MASK;
 }
 
 __EXTERN_INLINE void
@@ -235,7 +235,7 @@ init_new_context(struct task_struct *tsk, struct mm_struct *mm)
                 if (cpu_online(i))
                         mm->context[i] = 0;
         if (tsk != current)
-                tsk->thread_info->pcb.ptbr
+                task_thread_info(tsk)->pcb.ptbr
                   = ((unsigned long)mm->pgd - IDENT_ADDR) >> PAGE_SHIFT;
         return 0;
 }
@@ -249,7 +249,7 @@ destroy_context(struct mm_struct *mm)
 static inline void
 enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk)
 {
-        tsk->thread_info->pcb.ptbr
+        task_thread_info(tsk)->pcb.ptbr
           = ((unsigned long)mm->pgd - IDENT_ADDR) >> PAGE_SHIFT;
 }
 
diff --git a/include/asm-alpha/processor.h b/include/asm-alpha/processor.h
index bb1a7a3abb8b..425b7b6d28cb 100644
--- a/include/asm-alpha/processor.h
+++ b/include/asm-alpha/processor.h
@@ -52,19 +52,10 @@ extern long kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
 
 unsigned long get_wchan(struct task_struct *p);
 
-/* See arch/alpha/kernel/ptrace.c for details.  */
-#define PT_REG(reg) \
-  (PAGE_SIZE*2 - sizeof(struct pt_regs) + offsetof(struct pt_regs, reg))
-
-#define SW_REG(reg) \
- (PAGE_SIZE*2 - sizeof(struct pt_regs) - sizeof(struct switch_stack) \
-  + offsetof(struct switch_stack, reg))
-
-#define KSTK_EIP(tsk) \
-  (*(unsigned long *)(PT_REG(pc) + (unsigned long) ((tsk)->thread_info)))
+#define KSTK_EIP(tsk) (task_pt_regs(tsk)->pc)
 
 #define KSTK_ESP(tsk) \
-  ((tsk) == current ? rdusp() : (tsk)->thread_info->pcb.usp)
+  ((tsk) == current ? rdusp() : task_thread_info(tsk)->pcb.usp)
 
 #define cpu_relax()     barrier()
 
diff --git a/include/asm-alpha/ptrace.h b/include/asm-alpha/ptrace.h
index 072375c135b4..9933b8b3612e 100644
--- a/include/asm-alpha/ptrace.h
+++ b/include/asm-alpha/ptrace.h
@@ -75,10 +75,10 @@ struct switch_stack {
 #define profile_pc(regs) instruction_pointer(regs)
 extern void show_regs(struct pt_regs *);
 
-#define alpha_task_regs(task) \
-  ((struct pt_regs *) ((long) (task)->thread_info + 2*PAGE_SIZE) - 1)
+#define task_pt_regs(task) \
+  ((struct pt_regs *) (task_stack_page(task) + 2*PAGE_SIZE) - 1)
 
-#define force_successful_syscall_return() (alpha_task_regs(current)->r0 = 0)
+#define force_successful_syscall_return() (task_pt_regs(current)->r0 = 0)
 
 #endif
 
diff --git a/include/asm-alpha/system.h b/include/asm-alpha/system.h
index 050e86d12891..cc9c7e8cced5 100644
--- a/include/asm-alpha/system.h
+++ b/include/asm-alpha/system.h
@@ -131,15 +131,25 @@ struct el_common_EV6_mcheck {
 extern void halt(void) __attribute__((noreturn));
 #define __halt() __asm__ __volatile__ ("call_pal %0 #halt" : : "i" (PAL_halt))
 
-#define switch_to(P,N,L)                                                \
-  do {                                                                  \
-    (L) = alpha_switch_to(virt_to_phys(&(N)->thread_info->pcb), (P));   \
-    check_mmu_context();                                                \
+#define switch_to(P,N,L)                                                 \
+  do {                                                                   \
+    (L) = alpha_switch_to(virt_to_phys(&task_thread_info(N)->pcb), (P)); \
+    check_mmu_context();                                                 \
   } while (0)
 
 struct task_struct;
 extern struct task_struct *alpha_switch_to(unsigned long, struct task_struct*);
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 #define imb() \
 __asm__ __volatile__ ("call_pal %0 #imb" : : "i" (PAL_imb) : "memory")
 
diff --git a/include/asm-alpha/thread_info.h b/include/asm-alpha/thread_info.h
index d51491ed00b8..69ffd93f8e22 100644
--- a/include/asm-alpha/thread_info.h
+++ b/include/asm-alpha/thread_info.h
@@ -54,8 +54,6 @@ register struct thread_info *__current_thread_info __asm__("$8");
 #define alloc_thread_info(tsk) \
   ((struct thread_info *) __get_free_pages(GFP_KERNEL,1))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #endif /* __ASSEMBLY__ */
 
diff --git a/include/asm-arm/processor.h b/include/asm-arm/processor.h
index 7d4118e09054..31290694648b 100644
--- a/include/asm-arm/processor.h
+++ b/include/asm-arm/processor.h
@@ -85,9 +85,11 @@ unsigned long get_wchan(struct task_struct *p);
  */
 extern int kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
 
-#define KSTK_REGS(tsk)  (((struct pt_regs *)(THREAD_START_SP + (unsigned long)(tsk)->thread_info)) - 1)
-#define KSTK_EIP(tsk)   KSTK_REGS(tsk)->ARM_pc
-#define KSTK_ESP(tsk)   KSTK_REGS(tsk)->ARM_sp
+#define task_pt_regs(p) \
+        ((struct pt_regs *)(THREAD_START_SP + task_stack_page(p)) - 1)
+
+#define KSTK_EIP(tsk)   task_pt_regs(tsk)->ARM_pc
+#define KSTK_ESP(tsk)   task_pt_regs(tsk)->ARM_sp
 
 /*
  * Prefetching support - only ARMv5.
diff --git a/include/asm-arm/system.h b/include/asm-arm/system.h
index 5621d61ebc07..eb2de8c10515 100644
--- a/include/asm-arm/system.h
+++ b/include/asm-arm/system.h
@@ -168,10 +168,20 @@ extern struct task_struct *__switch_to(struct task_struct *, struct thread_info
 
 #define switch_to(prev,next,last)                                       \
 do {                                                                    \
-        last = __switch_to(prev,prev->thread_info,next->thread_info);   \
+        last = __switch_to(prev,task_thread_info(prev), task_thread_info(next));        \
 } while (0)
 
 /*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
+/*
  * CPU interrupt mask handling.
  */
 #if __LINUX_ARM_ARCH__ >= 6
diff --git a/include/asm-arm/thread_info.h b/include/asm-arm/thread_info.h
index 7c98557b717f..33a33cbb6329 100644
--- a/include/asm-arm/thread_info.h
+++ b/include/asm-arm/thread_info.h
@@ -96,13 +96,10 @@ static inline struct thread_info *current_thread_info(void)
 extern struct thread_info *alloc_thread_info(struct task_struct *task);
 extern void free_thread_info(struct thread_info *);
 
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
-
 #define thread_saved_pc(tsk)    \
-        ((unsigned long)(pc_pointer((tsk)->thread_info->cpu_context.pc)))
+        ((unsigned long)(pc_pointer(task_thread_info(tsk)->cpu_context.pc)))
 #define thread_saved_fp(tsk)    \
-        ((unsigned long)((tsk)->thread_info->cpu_context.fp))
+        ((unsigned long)(task_thread_info(tsk)->cpu_context.fp))
 
 extern void iwmmxt_task_disable(struct thread_info *);
 extern void iwmmxt_task_copy(struct thread_info *, void *);
diff --git a/include/asm-arm26/system.h b/include/asm-arm26/system.h
index f23fac1938f3..ca4ccfc4b578 100644
--- a/include/asm-arm26/system.h
+++ b/include/asm-arm26/system.h
@@ -111,10 +111,20 @@ extern struct task_struct *__switch_to(struct task_struct *, struct thread_info
 
 #define switch_to(prev,next,last)                                       \
 do {                                                                    \
-        last = __switch_to(prev,prev->thread_info,next->thread_info);   \
+        last = __switch_to(prev,task_thread_info(prev),task_thread_info(next)); \
 } while (0)
 
 /*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
+/*
  * Save the current interrupt enable state & disable IRQs
  */
 #define local_irq_save(x)                               \
diff --git a/include/asm-arm26/thread_info.h b/include/asm-arm26/thread_info.h
index aff3e5699c64..a65e58a0a767 100644
--- a/include/asm-arm26/thread_info.h
+++ b/include/asm-arm26/thread_info.h
@@ -82,18 +82,15 @@ static inline struct thread_info *current_thread_info(void)
 
 /* FIXME - PAGE_SIZE < 32K */
 #define THREAD_SIZE             (8*32768) // FIXME - this needs attention (see kernel/fork.c which gets a nice div by zero if this is lower than 8*32768
-#define __get_user_regs(x) (((struct pt_regs *)((unsigned long)(x) + THREAD_SIZE - 8)) - 1)
+#define task_pt_regs(task) ((struct pt_regs *)(task_stack_page(task) + THREAD_SIZE - 8) - 1)
 
 extern struct thread_info *alloc_thread_info(struct task_struct *task);
 extern void free_thread_info(struct thread_info *);
 
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
-
 #define thread_saved_pc(tsk)    \
-        ((unsigned long)(pc_pointer((tsk)->thread_info->cpu_context.pc)))
+        ((unsigned long)(pc_pointer(task_thread_info(tsk)->cpu_context.pc)))
 #define thread_saved_fp(tsk)    \
-        ((unsigned long)((tsk)->thread_info->cpu_context.fp))
+        ((unsigned long)(task_thread_info(tsk)->cpu_context.fp))
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/asm-cris/arch-v10/processor.h b/include/asm-cris/arch-v10/processor.h
index e23df8dc96e8..cc692c7a0660 100644
--- a/include/asm-cris/arch-v10/processor.h
+++ b/include/asm-cris/arch-v10/processor.h
@@ -40,7 +40,7 @@ struct thread_struct {
 #define KSTK_EIP(tsk)   \
 ({                      \
         unsigned long eip = 0;   \
-        unsigned long regs = (unsigned long)user_regs(tsk); \
+        unsigned long regs = (unsigned long)task_pt_regs(tsk); \
         if (regs > PAGE_SIZE && \
                 virt_addr_valid(regs)) \
         eip = ((struct pt_regs *)regs)->irp; \
diff --git a/include/asm-cris/arch-v32/processor.h b/include/asm-cris/arch-v32/processor.h
index 8c939bf27987..32bf2e538ced 100644
--- a/include/asm-cris/arch-v32/processor.h
+++ b/include/asm-cris/arch-v32/processor.h
@@ -36,7 +36,7 @@ struct thread_struct {
 #define KSTK_EIP(tsk)           \
 ({                              \
         unsigned long eip = 0;  \
-        unsigned long regs = (unsigned long)user_regs(tsk); \
+        unsigned long regs = (unsigned long)task_pt_regs(tsk); \
         if (regs > PAGE_SIZE && virt_addr_valid(regs))      \
                 eip = ((struct pt_regs *)regs)->erp;        \
         eip; \
diff --git a/include/asm-cris/processor.h b/include/asm-cris/processor.h
index dce41009eeb0..961e2bceadbc 100644
--- a/include/asm-cris/processor.h
+++ b/include/asm-cris/processor.h
@@ -45,7 +45,8 @@ struct task_struct;
  * Dito but for the currently running task
  */
 
-#define current_regs() user_regs(current->thread_info)
+#define task_pt_regs(task) user_regs(task_thread_info(task))
+#define current_regs() task_pt_regs(current)
 
 static inline void prepare_to_copy(struct task_struct *tsk)
 {
diff --git a/include/asm-cris/thread_info.h b/include/asm-cris/thread_info.h
index cef0140fc104..7ad853c3f74e 100644
--- a/include/asm-cris/thread_info.h
+++ b/include/asm-cris/thread_info.h
@@ -69,8 +69,6 @@ struct thread_info {
 /* thread information allocation */
 #define alloc_thread_info(tsk) ((struct thread_info *) __get_free_pages(GFP_KERNEL,1))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #endif /* !__ASSEMBLY__ */
 
diff --git a/include/asm-frv/thread_info.h b/include/asm-frv/thread_info.h
index 60f6b2aee76d..a5576e02dd1d 100644
--- a/include/asm-frv/thread_info.h
+++ b/include/asm-frv/thread_info.h
@@ -110,8 +110,6 @@ register struct thread_info *__current_thread_info asm("gr15");
 #endif
 
 #define free_thread_info(info)  kfree(info)
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/asm-h8300/thread_info.h b/include/asm-h8300/thread_info.h
index bfcc755c3bb1..45f09dc9caff 100644
--- a/include/asm-h8300/thread_info.h
+++ b/include/asm-h8300/thread_info.h
@@ -69,8 +69,6 @@ static inline struct thread_info *current_thread_info(void)
 #define alloc_thread_info(tsk) ((struct thread_info *) \
                                 __get_free_pages(GFP_KERNEL, 1))
 #define free_thread_info(ti)    free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
 #endif /* __ASSEMBLY__ */
 
 /*
diff --git a/include/asm-i386/i387.h b/include/asm-i386/i387.h
index 6747006743f9..152d0baa576a 100644
--- a/include/asm-i386/i387.h
+++ b/include/asm-i386/i387.h
@@ -49,19 +49,19 @@ static inline void __save_init_fpu( struct task_struct *tsk )
                 X86_FEATURE_FXSR,
                 "m" (tsk->thread.i387.fxsave)
                 :"memory");
-        tsk->thread_info->status &= ~TS_USEDFPU;
+        task_thread_info(tsk)->status &= ~TS_USEDFPU;
 }
 
 #define __unlazy_fpu( tsk ) do { \
-        if ((tsk)->thread_info->status & TS_USEDFPU) \
+        if (task_thread_info(tsk)->status & TS_USEDFPU) \
                 save_init_fpu( tsk ); \
 } while (0)
 
 #define __clear_fpu( tsk )                                      \
 do {                                                            \
-        if ((tsk)->thread_info->status & TS_USEDFPU) {          \
+        if (task_thread_info(tsk)->status & TS_USEDFPU) {               \
                 asm volatile("fnclex ; fwait");                         \
-                (tsk)->thread_info->status &= ~TS_USEDFPU;      \
+                task_thread_info(tsk)->status &= ~TS_USEDFPU;   \
                 stts();                                         \
         }                                                       \
 } while (0)
diff --git a/include/asm-i386/processor.h b/include/asm-i386/processor.h
index 13ecf66b098c..feca5d961e2b 100644
--- a/include/asm-i386/processor.h
+++ b/include/asm-i386/processor.h
@@ -561,10 +561,20 @@ unsigned long get_wchan(struct task_struct *p);
        (unsigned long)(&__ptr[THREAD_SIZE_LONGS]);                     \
 })
 
+/*
+ * The below -8 is to reserve 8 bytes on top of the ring0 stack.
+ * This is necessary to guarantee that the entire "struct pt_regs"
+ * is accessable even if the CPU haven't stored the SS/ESP registers
+ * on the stack (interrupt gate does not save these registers
+ * when switching to the same priv ring).
+ * Therefore beware: accessing the xss/esp fields of the
+ * "struct pt_regs" is possible, but they may contain the
+ * completely wrong values.
+ */
 #define task_pt_regs(task)                                             \
 ({                                                                     \
        struct pt_regs *__regs__;                                       \
-       __regs__ = (struct pt_regs *)KSTK_TOP((task)->thread_info);     \
+       __regs__ = (struct pt_regs *)(KSTK_TOP(task_stack_page(task))-8); \
        __regs__ - 1;                                                   \
 })
 
diff --git a/include/asm-i386/system.h b/include/asm-i386/system.h
index 9c0593b7a94e..36a92ed6a9d0 100644
--- a/include/asm-i386/system.h
+++ b/include/asm-i386/system.h
@@ -548,6 +548,15 @@ void enable_hlt(void);
 extern int es7000_plat;
 void cpu_idle_wait(void);
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible:
+ */
+static inline void sched_cacheflush(void)
+{
+        wbinvd();
+}
+
 extern unsigned long arch_align_stack(unsigned long sp);
 
 #endif
diff --git a/include/asm-i386/thread_info.h b/include/asm-i386/thread_info.h
index 8fbf791651bf..2493e77e8c30 100644
--- a/include/asm-i386/thread_info.h
+++ b/include/asm-i386/thread_info.h
@@ -111,8 +111,6 @@ register unsigned long current_stack_pointer asm("esp") __attribute_used__;
 #endif
 
 #define free_thread_info(info)  kfree(info)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/asm-i386/topology.h b/include/asm-i386/topology.h
index 0ec27c9e8e45..d7e19eb344b7 100644
--- a/include/asm-i386/topology.h
+++ b/include/asm-i386/topology.h
@@ -72,7 +72,6 @@ static inline int node_to_first_cpu(int node)
         .max_interval           = 32,                   \
         .busy_factor            = 32,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000000),         \
         .cache_nice_tries       = 1,                    \
         .busy_idx               = 3,                    \
         .idle_idx               = 1,                    \
diff --git a/include/asm-ia64/compat.h b/include/asm-ia64/compat.h
index aaf11f4e9169..c0b19106665c 100644
--- a/include/asm-ia64/compat.h
+++ b/include/asm-ia64/compat.h
@@ -192,7 +192,7 @@ compat_ptr (compat_uptr_t uptr)
 static __inline__ void __user *
 compat_alloc_user_space (long len)
 {
-        struct pt_regs *regs = ia64_task_regs(current);
+        struct pt_regs *regs = task_pt_regs(current);
         return (void __user *) (((regs->r12 & 0xffffffff) & -16) - len);
 }
 
diff --git a/include/asm-ia64/processor.h b/include/asm-ia64/processor.h
index 94e07e727395..8c648bf72bbd 100644
--- a/include/asm-ia64/processor.h
+++ b/include/asm-ia64/processor.h
@@ -352,7 +352,7 @@ extern unsigned long get_wchan (struct task_struct *p);
 /* Return instruction pointer of blocked task TSK.  */
 #define KSTK_EIP(tsk)                                   \
   ({                                                    \
-        struct pt_regs *_regs = ia64_task_regs(tsk);    \
+        struct pt_regs *_regs = task_pt_regs(tsk);      \
         _regs->cr_iip + ia64_psr(_regs)->ri;            \
   })
 
diff --git a/include/asm-ia64/ptrace.h b/include/asm-ia64/ptrace.h
index 2c703d6e0c86..9471cdc3f4c0 100644
--- a/include/asm-ia64/ptrace.h
+++ b/include/asm-ia64/ptrace.h
@@ -248,7 +248,7 @@ struct switch_stack {
 })
 
   /* given a pointer to a task_struct, return the user's pt_regs */
-# define ia64_task_regs(t)              (((struct pt_regs *) ((char *) (t) + IA64_STK_OFFSET)) - 1)
+# define task_pt_regs(t)                (((struct pt_regs *) ((char *) (t) + IA64_STK_OFFSET)) - 1)
 # define ia64_psr(regs)                 ((struct ia64_psr *) &(regs)->cr_ipsr)
 # define user_mode(regs)                (((struct ia64_psr *) &(regs)->cr_ipsr)->cpl != 0)
 # define user_stack(task,regs)  ((long) regs - (long) task == IA64_STK_OFFSET - sizeof(*regs))
@@ -271,7 +271,7 @@ struct switch_stack {
    *
    * On ia64, we can clear the user's pt_regs->r8 to force a successful syscall.
    */
-# define force_successful_syscall_return()      (ia64_task_regs(current)->r8 = 0)
+# define force_successful_syscall_return()      (task_pt_regs(current)->r8 = 0)
 
   struct task_struct;                   /* forward decl */
   struct unw_frame_info;                /* forward decl */
diff --git a/include/asm-ia64/system.h b/include/asm-ia64/system.h
index 635235fa1e32..80c5a234e259 100644
--- a/include/asm-ia64/system.h
+++ b/include/asm-ia64/system.h
@@ -219,14 +219,14 @@ extern void ia64_load_extra (struct task_struct *task);
 
 #define IA64_HAS_EXTRA_STATE(t)                                                 \
         ((t)->thread.flags & (IA64_THREAD_DBG_VALID|IA64_THREAD_PM_VALID)       \
-         || IS_IA32_PROCESS(ia64_task_regs(t)) || PERFMON_IS_SYSWIDE())
+         || IS_IA32_PROCESS(task_pt_regs(t)) || PERFMON_IS_SYSWIDE())
 
 #define __switch_to(prev,next,last) do {                                                         \
         if (IA64_HAS_EXTRA_STATE(prev))                                                          \
                 ia64_save_extra(prev);                                                           \
         if (IA64_HAS_EXTRA_STATE(next))                                                          \
                 ia64_load_extra(next);                                                           \
-        ia64_psr(ia64_task_regs(next))->dfh = !ia64_is_local_fpu_owner(next);                    \
+        ia64_psr(task_pt_regs(next))->dfh = !ia64_is_local_fpu_owner(next);                      \
         (last) = ia64_switch_to((next));                                                         \
 } while (0)
 
@@ -238,8 +238,8 @@ extern void ia64_load_extra (struct task_struct *task);
  * the latest fph state from another CPU.  In other words: eager save, lazy restore.
  */
 # define switch_to(prev,next,last) do {                                         \
-        if (ia64_psr(ia64_task_regs(prev))->mfh && ia64_is_local_fpu_owner(prev)) {                             \
-                ia64_psr(ia64_task_regs(prev))->mfh = 0;                        \
+        if (ia64_psr(task_pt_regs(prev))->mfh && ia64_is_local_fpu_owner(prev)) {                               \
+                ia64_psr(task_pt_regs(prev))->mfh = 0;                  \
                 (prev)->thread.flags |= IA64_THREAD_FPH_VALID;                  \
                 __ia64_save_fpu((prev)->thread.fph);                            \
         }                                                                       \
@@ -279,6 +279,7 @@ extern void ia64_load_extra (struct task_struct *task);
 #define ia64_platform_is(x) (strcmp(x, platform_name) == 0)
 
 void cpu_idle_wait(void);
+void sched_cacheflush(void);
 
 #define arch_align_stack(x) (x)
 
diff --git a/include/asm-ia64/thread_info.h b/include/asm-ia64/thread_info.h
index 171b2207bde4..653bb7f9a753 100644
--- a/include/asm-ia64/thread_info.h
+++ b/include/asm-ia64/thread_info.h
@@ -57,11 +57,20 @@ struct thread_info {
 /* how to get the thread information struct from C */
 #define current_thread_info()   ((struct thread_info *) ((char *) current + IA64_TASK_SIZE))
 #define alloc_thread_info(tsk)  ((struct thread_info *) ((char *) (tsk) + IA64_TASK_SIZE))
+#define task_thread_info(tsk)   ((struct thread_info *) ((char *) (tsk) + IA64_TASK_SIZE))
 #else
 #define current_thread_info()   ((struct thread_info *) 0)
 #define alloc_thread_info(tsk)  ((struct thread_info *) 0)
+#define task_thread_info(tsk)   ((struct thread_info *) 0)
 #endif
 #define free_thread_info(ti)    /* nothing */
+#define task_stack_page(tsk)    ((void *)(tsk))
+
+#define __HAVE_THREAD_FUNCTIONS
+#define setup_thread_stack(p, org) \
+        *task_thread_info(p) = *task_thread_info(org); \
+        task_thread_info(p)->task = (p);
+#define end_of_stack(p) (unsigned long *)((void *)(p) + IA64_RBS_OFFSET)
 
 #define __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
 #define alloc_task_struct()     ((task_t *)__get_free_pages(GFP_KERNEL, KERNEL_STACK_SIZE_ORDER))
diff --git a/include/asm-ia64/topology.h b/include/asm-ia64/topology.h
index f7c330467e7e..d8aae4da3978 100644
--- a/include/asm-ia64/topology.h
+++ b/include/asm-ia64/topology.h
@@ -55,7 +55,6 @@ void build_cpu_to_node_map(void);
         .max_interval           = 4,                    \
         .busy_factor            = 64,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000000),         \
         .per_cpu_gain           = 100,                  \
         .cache_nice_tries       = 2,                    \
         .busy_idx               = 2,                    \
@@ -81,7 +80,6 @@ void build_cpu_to_node_map(void);
         .max_interval           = 8*(min(num_online_cpus(), 32)), \
         .busy_factor            = 64,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000000),         \
         .cache_nice_tries       = 2,                    \
         .busy_idx               = 3,                    \
         .idle_idx               = 2,                    \
diff --git a/include/asm-m32r/ptrace.h b/include/asm-m32r/ptrace.h
index 55cd7ecfde43..0d058b2d844e 100644
--- a/include/asm-m32r/ptrace.h
+++ b/include/asm-m32r/ptrace.h
@@ -163,6 +163,9 @@ extern void show_regs(struct pt_regs *);
 
 extern void withdraw_debug_trap(struct pt_regs *regs);
 
+#define task_pt_regs(task) \
+        ((struct pt_regs *)(task_stack_page(task) + THREAD_SIZE) - 1)
+
 #endif /* __KERNEL */
 
 #endif /* _ASM_M32R_PTRACE_H */
diff --git a/include/asm-m32r/system.h b/include/asm-m32r/system.h
index dcf619a0a0b0..06c12a037cba 100644
--- a/include/asm-m32r/system.h
+++ b/include/asm-m32r/system.h
@@ -68,6 +68,16 @@
         last = __last; \
 } while(0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 /* Interrupt Control */
 #if !defined(CONFIG_CHIP_M32102) && !defined(CONFIG_CHIP_M32104)
 #define local_irq_enable() \
diff --git a/include/asm-m32r/thread_info.h b/include/asm-m32r/thread_info.h
index 0f589363f619..22aff3222d22 100644
--- a/include/asm-m32r/thread_info.h
+++ b/include/asm-m32r/thread_info.h
@@ -110,8 +110,6 @@ static inline struct thread_info *current_thread_info(void)
 #endif
 
 #define free_thread_info(info) kfree(info)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #define TI_FLAG_FAULT_CODE_SHIFT        28
 
diff --git a/include/asm-m68k/amigahw.h b/include/asm-m68k/amigahw.h
index 3ae5d8d55ba9..a16fe4e5a28a 100644
--- a/include/asm-m68k/amigahw.h
+++ b/include/asm-m68k/amigahw.h
@@ -274,7 +274,7 @@ struct CIA {
 #define ZTWO_VADDR(x) (((unsigned long)(x))+zTwoBase)
 
 #define CUSTOM_PHYSADDR     (0xdff000)
-#define custom ((*(volatile struct CUSTOM *)(zTwoBase+CUSTOM_PHYSADDR)))
+#define amiga_custom ((*(volatile struct CUSTOM *)(zTwoBase+CUSTOM_PHYSADDR)))
 
 #define CIAA_PHYSADDR     (0xbfe001)
 #define CIAB_PHYSADDR     (0xbfd000)
@@ -294,12 +294,12 @@ static inline void amifb_video_off(void)
 {
         if (amiga_chipset == CS_ECS || amiga_chipset == CS_AGA) {
                 /* program Denise/Lisa for a higher maximum play rate */
-                custom.htotal = 113;        /* 31 kHz */
-                custom.vtotal = 223;        /* 70 Hz */
-                custom.beamcon0 = 0x4390;   /* HARDDIS, VAR{BEAM,VSY,HSY,CSY}EN */
+                amiga_custom.htotal = 113;        /* 31 kHz */
+                amiga_custom.vtotal = 223;        /* 70 Hz */
+                amiga_custom.beamcon0 = 0x4390;   /* HARDDIS, VAR{BEAM,VSY,HSY,CSY}EN */
                 /* suspend the monitor */
-                custom.hsstrt = custom.hsstop = 116;
-                custom.vsstrt = custom.vsstop = 226;
+                amiga_custom.hsstrt = amiga_custom.hsstop = 116;
+                amiga_custom.vsstrt = amiga_custom.vsstop = 226;
                 amiga_audio_min_period = 57;
         }
 }
diff --git a/include/asm-m68k/amigaints.h b/include/asm-m68k/amigaints.h
index 2aff4cfbf7b3..aa968d014bb6 100644
--- a/include/asm-m68k/amigaints.h
+++ b/include/asm-m68k/amigaints.h
@@ -109,8 +109,6 @@
 extern void amiga_do_irq(int irq, struct pt_regs *fp);
 extern void amiga_do_irq_list(int irq, struct pt_regs *fp);
 
-extern unsigned short amiga_intena_vals[];
-
 /* CIA interrupt control register bits */
 
 #define CIA_ICR_TA      0x01
diff --git a/include/asm-m68k/checksum.h b/include/asm-m68k/checksum.h
index 78860c20db01..17280ef719f5 100644
--- a/include/asm-m68k/checksum.h
+++ b/include/asm-m68k/checksum.h
@@ -25,7 +25,7 @@ unsigned int csum_partial(const unsigned char * buff, int len, unsigned int sum)
  * better 64-bit) boundary
  */
 
-extern unsigned int csum_partial_copy_from_user(const unsigned char *src,
+extern unsigned int csum_partial_copy_from_user(const unsigned char __user *src,
                                                 unsigned char *dst,
                                                 int len, int sum,
                                                 int *csum_err);
diff --git a/include/asm-m68k/dsp56k.h b/include/asm-m68k/dsp56k.h
index ab3dd33e23a1..2d8c0c9f794b 100644
--- a/include/asm-m68k/dsp56k.h
+++ b/include/asm-m68k/dsp56k.h
@@ -13,7 +13,7 @@
 /* Used for uploading DSP binary code */
 struct dsp56k_upload {
         int len;
-        char *bin;
+        char __user *bin;
 };
 
 /* For the DSP host flags */
diff --git a/include/asm-m68k/floppy.h b/include/asm-m68k/floppy.h
index c6e708dd9f62..63a05ed95c17 100644
--- a/include/asm-m68k/floppy.h
+++ b/include/asm-m68k/floppy.h
@@ -46,7 +46,7 @@ asmlinkage irqreturn_t floppy_hardint(int irq, void *dev_id,
 
 static int virtual_dma_count=0;
 static int virtual_dma_residue=0;
-static char *virtual_dma_addr=0;
+static char *virtual_dma_addr=NULL;
 static int virtual_dma_mode=0;
 static int doing_pdma=0;
 
diff --git a/include/asm-m68k/hardirq.h b/include/asm-m68k/hardirq.h
index 728318bf7f0e..5e1c5826c83d 100644
--- a/include/asm-m68k/hardirq.h
+++ b/include/asm-m68k/hardirq.h
@@ -14,13 +14,4 @@ typedef struct {
 
 #define HARDIRQ_BITS    8
 
-/*
- * The hardirq mask has to be large enough to have
- * space for potentially all IRQ sources in the system
- * nesting on a single CPU:
- */
-#if (1 << HARDIRQ_BITS) < NR_IRQS
-# error HARDIRQ_BITS is too low!
-#endif
-
 #endif
diff --git a/include/asm-m68k/io.h b/include/asm-m68k/io.h
index 6bb8b0d8f99d..dcfaa352d34c 100644
--- a/include/asm-m68k/io.h
+++ b/include/asm-m68k/io.h
@@ -24,6 +24,7 @@
 #ifdef __KERNEL__
 
 #include <linux/config.h>
+#include <linux/compiler.h>
 #include <asm/raw_io.h>
 #include <asm/virtconvert.h>
 
@@ -120,68 +121,68 @@ extern int isa_sex;
  * be compiled in so the case statement will be optimised away
  */
 
-static inline u8 *isa_itb(unsigned long addr)
+static inline u8 __iomem *isa_itb(unsigned long addr)
 {
   switch(ISA_TYPE)
     {
 #ifdef CONFIG_Q40
-    case Q40_ISA: return (u8 *)Q40_ISA_IO_B(addr);
+    case Q40_ISA: return (u8 __iomem *)Q40_ISA_IO_B(addr);
 #endif
 #ifdef CONFIG_GG2
-    case GG2_ISA: return (u8 *)GG2_ISA_IO_B(addr);
+    case GG2_ISA: return (u8 __iomem *)GG2_ISA_IO_B(addr);
 #endif
 #ifdef CONFIG_AMIGA_PCMCIA
-    case AG_ISA: return (u8 *)AG_ISA_IO_B(addr);
+    case AG_ISA: return (u8 __iomem *)AG_ISA_IO_B(addr);
 #endif
-    default: return 0; /* avoid warnings, just in case */
+    default: return NULL; /* avoid warnings, just in case */
     }
 }
-static inline u16 *isa_itw(unsigned long addr)
+static inline u16 __iomem *isa_itw(unsigned long addr)
 {
   switch(ISA_TYPE)
     {
 #ifdef CONFIG_Q40
-    case Q40_ISA: return (u16 *)Q40_ISA_IO_W(addr);
+    case Q40_ISA: return (u16 __iomem *)Q40_ISA_IO_W(addr);
 #endif
 #ifdef CONFIG_GG2
-    case GG2_ISA: return (u16 *)GG2_ISA_IO_W(addr);
+    case GG2_ISA: return (u16 __iomem *)GG2_ISA_IO_W(addr);
 #endif
 #ifdef CONFIG_AMIGA_PCMCIA
-    case AG_ISA: return (u16 *)AG_ISA_IO_W(addr);
+    case AG_ISA: return (u16 __iomem *)AG_ISA_IO_W(addr);
 #endif
-    default: return 0; /* avoid warnings, just in case */
+    default: return NULL; /* avoid warnings, just in case */
     }
 }
-static inline u8 *isa_mtb(unsigned long addr)
+static inline u8 __iomem *isa_mtb(unsigned long addr)
 {
   switch(ISA_TYPE)
     {
 #ifdef CONFIG_Q40
-    case Q40_ISA: return (u8 *)Q40_ISA_MEM_B(addr);
+    case Q40_ISA: return (u8 __iomem *)Q40_ISA_MEM_B(addr);
 #endif
 #ifdef CONFIG_GG2
-    case GG2_ISA: return (u8 *)GG2_ISA_MEM_B(addr);
+    case GG2_ISA: return (u8 __iomem *)GG2_ISA_MEM_B(addr);
 #endif
 #ifdef CONFIG_AMIGA_PCMCIA
-    case AG_ISA: return (u8 *)addr;
+    case AG_ISA: return (u8 __iomem *)addr;
 #endif
-    default: return 0; /* avoid warnings, just in case */
+    default: return NULL; /* avoid warnings, just in case */
     }
 }
-static inline u16 *isa_mtw(unsigned long addr)
+static inline u16 __iomem *isa_mtw(unsigned long addr)
 {
   switch(ISA_TYPE)
     {
 #ifdef CONFIG_Q40
-    case Q40_ISA: return (u16 *)Q40_ISA_MEM_W(addr);
+    case Q40_ISA: return (u16 __iomem *)Q40_ISA_MEM_W(addr);
 #endif
 #ifdef CONFIG_GG2
-    case GG2_ISA: return (u16 *)GG2_ISA_MEM_W(addr);
+    case GG2_ISA: return (u16 __iomem *)GG2_ISA_MEM_W(addr);
 #endif
 #ifdef CONFIG_AMIGA_PCMCIA
-    case AG_ISA: return (u16 *)addr;
+    case AG_ISA: return (u16 __iomem *)addr;
 #endif
-    default: return 0; /* avoid warnings, just in case */
+    default: return NULL; /* avoid warnings, just in case */
     }
 }
 
@@ -326,20 +327,20 @@ static inline void isa_delay(void)
 
 #define mmiowb()
 
-static inline void *ioremap(unsigned long physaddr, unsigned long size)
+static inline void __iomem *ioremap(unsigned long physaddr, unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_NOCACHE_SER);
 }
-static inline void *ioremap_nocache(unsigned long physaddr, unsigned long size)
+static inline void __iomem *ioremap_nocache(unsigned long physaddr, unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_NOCACHE_SER);
 }
-static inline void *ioremap_writethrough(unsigned long physaddr,
+static inline void __iomem *ioremap_writethrough(unsigned long physaddr,
                                          unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_WRITETHROUGH);
 }
-static inline void *ioremap_fullcache(unsigned long physaddr,
+static inline void __iomem *ioremap_fullcache(unsigned long physaddr,
                                       unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_FULL_CACHING);
diff --git a/include/asm-m68k/irq.h b/include/asm-m68k/irq.h
index 127ad190cf2d..325c86f8512d 100644
--- a/include/asm-m68k/irq.h
+++ b/include/asm-m68k/irq.h
@@ -23,6 +23,15 @@
 #endif
 
 /*
+ * The hardirq mask has to be large enough to have
+ * space for potentially all IRQ sources in the system
+ * nesting on a single CPU:
+ */
+#if (1 << HARDIRQ_BITS) < NR_IRQS
+# error HARDIRQ_BITS is too low!
+#endif
+
+/*
  * Interrupt source definitions
  * General interrupt sources are the level 1-7.
  * Adding an interrupt service routine for one of these sources
diff --git a/include/asm-m68k/machdep.h b/include/asm-m68k/machdep.h
index a0dd5c47002c..7d3fee342369 100644
--- a/include/asm-m68k/machdep.h
+++ b/include/asm-m68k/machdep.h
@@ -34,7 +34,6 @@ extern void (*mach_power_off)( void );
 extern unsigned long (*mach_hd_init) (unsigned long, unsigned long);
 extern void (*mach_hd_setup)(char *, int *);
 extern long mach_max_dma_address;
-extern void (*mach_floppy_setup)(char *, int *);
 extern void (*mach_heartbeat) (int);
 extern void (*mach_l2_flush) (int);
 extern void (*mach_beep) (unsigned int, unsigned int);
diff --git a/include/asm-m68k/raw_io.h b/include/asm-m68k/raw_io.h
index 041f0a87b25d..5439bcaa57c6 100644
--- a/include/asm-m68k/raw_io.h
+++ b/include/asm-m68k/raw_io.h
@@ -19,9 +19,9 @@
 #define IOMAP_NOCACHE_NONSER            2
 #define IOMAP_WRITETHROUGH              3
 
-extern void iounmap(void *addr);
+extern void iounmap(void __iomem *addr);
 
-extern void *__ioremap(unsigned long physaddr, unsigned long size,
+extern void __iomem *__ioremap(unsigned long physaddr, unsigned long size,
                        int cacheflag);
 extern void __iounmap(void *addr, unsigned long size);
 
@@ -30,21 +30,21 @@ extern void __iounmap(void *addr, unsigned long size);
  * two accesses to memory, which may be undesirable for some devices.
  */
 #define in_8(addr) \
-    ({ u8 __v = (*(volatile u8 *) (addr)); __v; })
+    ({ u8 __v = (*(__force volatile u8 *) (addr)); __v; })
 #define in_be16(addr) \
-    ({ u16 __v = (*(volatile u16 *) (addr)); __v; })
+    ({ u16 __v = (*(__force volatile u16 *) (addr)); __v; })
 #define in_be32(addr) \
-    ({ u32 __v = (*(volatile u32 *) (addr)); __v; })
+    ({ u32 __v = (*(__force volatile u32 *) (addr)); __v; })
 #define in_le16(addr) \
-    ({ u16 __v = le16_to_cpu(*(volatile u16 *) (addr)); __v; })
+    ({ u16 __v = le16_to_cpu(*(__force volatile u16 *) (addr)); __v; })
 #define in_le32(addr) \
-    ({ u32 __v = le32_to_cpu(*(volatile u32 *) (addr)); __v; })
+    ({ u32 __v = le32_to_cpu(*(__force volatile u32 *) (addr)); __v; })
 
-#define out_8(addr,b) (void)((*(volatile u8 *) (addr)) = (b))
-#define out_be16(addr,w) (void)((*(volatile u16 *) (addr)) = (w))
-#define out_be32(addr,l) (void)((*(volatile u32 *) (addr)) = (l))
-#define out_le16(addr,w) (void)((*(volatile u16 *) (addr)) = cpu_to_le16(w))
-#define out_le32(addr,l) (void)((*(volatile u32 *) (addr)) = cpu_to_le32(l))
+#define out_8(addr,b) (void)((*(__force volatile u8 *) (addr)) = (b))
+#define out_be16(addr,w) (void)((*(__force volatile u16 *) (addr)) = (w))
+#define out_be32(addr,l) (void)((*(__force volatile u32 *) (addr)) = (l))
+#define out_le16(addr,w) (void)((*(__force volatile u16 *) (addr)) = cpu_to_le16(w))
+#define out_le32(addr,l) (void)((*(__force volatile u32 *) (addr)) = cpu_to_le32(l))
 
 #define raw_inb in_8
 #define raw_inw in_be16
@@ -54,7 +54,7 @@ extern void __iounmap(void *addr, unsigned long size);
 #define raw_outw(val,port) out_be16((port),(val))
 #define raw_outl(val,port) out_be32((port),(val))
 
-static inline void raw_insb(volatile u8 *port, u8 *buf, unsigned int len)
+static inline void raw_insb(volatile u8 __iomem *port, u8 *buf, unsigned int len)
 {
         unsigned int i;
 
@@ -62,7 +62,7 @@ static inline void raw_insb(volatile u8 *port, u8 *buf, unsigned int len)
                 *buf++ = in_8(port);
 }
 
-static inline void raw_outsb(volatile u8 *port, const u8 *buf,
+static inline void raw_outsb(volatile u8 __iomem *port, const u8 *buf,
                              unsigned int len)
 {
         unsigned int i;
@@ -71,7 +71,7 @@ static inline void raw_outsb(volatile u8 *port, const u8 *buf,
                 out_8(port, *buf++);
 }
 
-static inline void raw_insw(volatile u16 *port, u16 *buf, unsigned int nr)
+static inline void raw_insw(volatile u16 __iomem *port, u16 *buf, unsigned int nr)
 {
         unsigned int tmp;
 
@@ -110,7 +110,7 @@ static inline void raw_insw(volatile u16 *port, u16 *buf, unsigned int nr)
         }
 }
 
-static inline void raw_outsw(volatile u16 *port, const u16 *buf,
+static inline void raw_outsw(volatile u16 __iomem *port, const u16 *buf,
                              unsigned int nr)
 {
         unsigned int tmp;
@@ -150,7 +150,7 @@ static inline void raw_outsw(volatile u16 *port, const u16 *buf,
         }
 }
 
-static inline void raw_insl(volatile u32 *port, u32 *buf, unsigned int nr)
+static inline void raw_insl(volatile u32 __iomem *port, u32 *buf, unsigned int nr)
 {
         unsigned int tmp;
 
@@ -189,7 +189,7 @@ static inline void raw_insl(volatile u32 *port, u32 *buf, unsigned int nr)
         }
 }
 
-static inline void raw_outsl(volatile u32 *port, const u32 *buf,
+static inline void raw_outsl(volatile u32 __iomem *port, const u32 *buf,
                              unsigned int nr)
 {
         unsigned int tmp;
@@ -230,7 +230,7 @@ static inline void raw_outsl(volatile u32 *port, const u32 *buf,
 }
 
 
-static inline void raw_insw_swapw(volatile u16 *port, u16 *buf,
+static inline void raw_insw_swapw(volatile u16 __iomem *port, u16 *buf,
                                   unsigned int nr)
 {
     if ((nr) % 8)
@@ -283,7 +283,7 @@ static inline void raw_insw_swapw(volatile u16 *port, u16 *buf,
                 : "d0", "a0", "a1", "d6");
 }
 
-static inline void raw_outsw_swapw(volatile u16 *port, const u16 *buf,
+static inline void raw_outsw_swapw(volatile u16 __iomem *port, const u16 *buf,
                                    unsigned int nr)
 {
     if ((nr) % 8)
diff --git a/include/asm-m68k/signal.h b/include/asm-m68k/signal.h
index a0cdf9082372..b7b7ea20caab 100644
--- a/include/asm-m68k/signal.h
+++ b/include/asm-m68k/signal.h
@@ -144,7 +144,7 @@ struct sigaction {
 #endif /* __KERNEL__ */
 
 typedef struct sigaltstack {
-        void *ss_sp;
+        void __user *ss_sp;
         int ss_flags;
         size_t ss_size;
 } stack_t;
diff --git a/include/asm-m68k/sun3_pgtable.h b/include/asm-m68k/sun3_pgtable.h
index e974bb072047..5156a28a18d8 100644
--- a/include/asm-m68k/sun3_pgtable.h
+++ b/include/asm-m68k/sun3_pgtable.h
@@ -211,7 +211,7 @@ static inline unsigned long pte_to_pgoff(pte_t pte)
         return pte.pte & SUN3_PAGE_PGNUM_MASK;
 }
 
-static inline pte_t pgoff_to_pte(inline unsigned off)
+static inline pte_t pgoff_to_pte(unsigned off)
 {
         pte_t pte = { off + SUN3_PAGE_ACCESSED };
         return pte;
diff --git a/include/asm-m68k/sun3ints.h b/include/asm-m68k/sun3ints.h
index fd838eb14213..bd038fccb64b 100644
--- a/include/asm-m68k/sun3ints.h
+++ b/include/asm-m68k/sun3ints.h
@@ -31,7 +31,6 @@ int sun3_request_irq(unsigned int irq,
                     );
 extern void sun3_init_IRQ (void);
 extern irqreturn_t (*sun3_default_handler[]) (int, void *, struct pt_regs *);
-extern irqreturn_t (*sun3_inthandler[]) (int, void *, struct pt_regs *);
 extern void sun3_free_irq (unsigned int irq, void *dev_id);
 extern void sun3_enable_interrupts (void);
 extern void sun3_disable_interrupts (void);
diff --git a/include/asm-m68k/sun3xflop.h b/include/asm-m68k/sun3xflop.h
index fda1eccf10aa..98a9f79dab29 100644
--- a/include/asm-m68k/sun3xflop.h
+++ b/include/asm-m68k/sun3xflop.h
@@ -208,7 +208,7 @@ static int sun3xflop_request_irq(void)
 
         if(!once) {
                 once = 1;
-                error = request_irq(FLOPPY_IRQ, sun3xflop_hardint, SA_INTERRUPT, "floppy", 0);
+                error = request_irq(FLOPPY_IRQ, sun3xflop_hardint, SA_INTERRUPT, "floppy", NULL);
                 return ((error == 0) ? 0 : -1);
         } else return 0;
 }
@@ -238,7 +238,7 @@ static int sun3xflop_init(void)
         *sun3x_fdc.fcr_r = 0;
 
         /* Success... */
-        floppy_set_flags(0, 1, FD_BROKEN_DCL); // I don't know how to detect this.
+        floppy_set_flags(NULL, 1, FD_BROKEN_DCL); // I don't know how to detect this.
         allowed_drive_mask = 0x01;
         return (int) SUN3X_FDC;
 }
diff --git a/include/asm-m68k/thread_info.h b/include/asm-m68k/thread_info.h
index 9532ca3c45cb..c4d622a57dfb 100644
--- a/include/asm-m68k/thread_info.h
+++ b/include/asm-m68k/thread_info.h
@@ -37,6 +37,7 @@ struct thread_info {
 #define init_stack              (init_thread_union.stack)
 
 #define task_thread_info(tsk)   (&(tsk)->thread.info)
+#define task_stack_page(tsk)    ((void *)(tsk)->thread_info)
 #define current_thread_info()   task_thread_info(current)
 
 #define __HAVE_THREAD_FUNCTIONS
diff --git a/include/asm-m68k/uaccess.h b/include/asm-m68k/uaccess.h
index f5cedf19cf68..2ffd87b0a769 100644
--- a/include/asm-m68k/uaccess.h
+++ b/include/asm-m68k/uaccess.h
@@ -42,6 +42,7 @@ struct exception_table_entry
 ({                                                      \
     int __pu_err;                                       \
     typeof(*(ptr)) __pu_val = (x);                      \
+    __chk_user_ptr(ptr);                                \
     switch (sizeof (*(ptr))) {                          \
     case 1:                                             \
         __put_user_asm(__pu_err, __pu_val, ptr, b);     \
@@ -91,6 +92,7 @@ __asm__ __volatile__					\
 ({                                                              \
     int __gu_err;                                               \
     typeof(*(ptr)) __gu_val;                                    \
+    __chk_user_ptr(ptr);                                        \
     switch (sizeof(*(ptr))) {                                   \
     case 1:                                                     \
         __get_user_asm(__gu_err, __gu_val, ptr, b, "=d");       \
@@ -105,7 +107,7 @@ __asm__ __volatile__					\
         __gu_err = __constant_copy_from_user(&__gu_val, ptr, 8);  \
         break;                                                  \
     default:                                                    \
-        __gu_val = 0;                                           \
+        __gu_val = (typeof(*(ptr)))0;                           \
         __gu_err = __get_user_bad();                            \
         break;                                                  \
     }                                                           \
@@ -134,7 +136,7 @@ __asm__ __volatile__				\
      : "m"(*(ptr)), "i" (-EFAULT), "0"(0))
 
 static inline unsigned long
-__generic_copy_from_user(void *to, const void *from, unsigned long n)
+__generic_copy_from_user(void *to, const void __user *from, unsigned long n)
 {
     unsigned long tmp;
     __asm__ __volatile__
@@ -189,7 +191,7 @@ __generic_copy_from_user(void *to, const void *from, unsigned long n)
 }
 
 static inline unsigned long
-__generic_copy_to_user(void *to, const void *from, unsigned long n)
+__generic_copy_to_user(void __user *to, const void *from, unsigned long n)
 {
     unsigned long tmp;
     __asm__ __volatile__
@@ -264,7 +266,7 @@ __generic_copy_to_user(void *to, const void *from, unsigned long n)
          : "d0", "memory")
 
 static inline unsigned long
-__constant_copy_from_user(void *to, const void *from, unsigned long n)
+__constant_copy_from_user(void *to, const void __user *from, unsigned long n)
 {
     switch (n) {
     case 0:
@@ -520,7 +522,7 @@ __constant_copy_from_user(void *to, const void *from, unsigned long n)
 #define __copy_from_user_inatomic __copy_from_user
 
 static inline unsigned long
-__constant_copy_to_user(void *to, const void *from, unsigned long n)
+__constant_copy_to_user(void __user *to, const void *from, unsigned long n)
 {
     switch (n) {
     case 0:
@@ -766,7 +768,7 @@ __constant_copy_to_user(void *to, const void *from, unsigned long n)
  */
 
 static inline long
-strncpy_from_user(char *dst, const char *src, long count)
+strncpy_from_user(char *dst, const char __user *src, long count)
 {
     long res;
     if (count == 0) return count;
@@ -799,11 +801,11 @@ strncpy_from_user(char *dst, const char *src, long count)
  *
  * Return 0 on exception, a value greater than N if too long
  */
-static inline long strnlen_user(const char *src, long n)
+static inline long strnlen_user(const char __user *src, long n)
 {
         long res;
 
-        res = -(long)src;
+        res = -(unsigned long)src;
         __asm__ __volatile__
                 ("1:\n"
                  "   tstl %2\n"
@@ -842,7 +844,7 @@ static inline long strnlen_user(const char *src, long n)
  */
 
 static inline unsigned long
-clear_user(void *to, unsigned long n)
+clear_user(void __user *to, unsigned long n)
 {
     __asm__ __volatile__
         ("   tstl %1\n"
diff --git a/include/asm-m68k/zorro.h b/include/asm-m68k/zorro.h
index cf816588bedb..5ce97c22b582 100644
--- a/include/asm-m68k/zorro.h
+++ b/include/asm-m68k/zorro.h
@@ -15,24 +15,24 @@
 #define z_memcpy_fromio(a,b,c)  memcpy((a),(void *)(b),(c))
 #define z_memcpy_toio(a,b,c)    memcpy((void *)(a),(b),(c))
 
-static inline void *z_remap_nocache_ser(unsigned long physaddr,
+static inline void __iomem *z_remap_nocache_ser(unsigned long physaddr,
                                         unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_NOCACHE_SER);
 }
 
-static inline void *z_remap_nocache_nonser(unsigned long physaddr,
+static inline void __iomem *z_remap_nocache_nonser(unsigned long physaddr,
                                            unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_NOCACHE_NONSER);
 }
 
-static inline void *z_remap_writethrough(unsigned long physaddr,
+static inline void __iomem *z_remap_writethrough(unsigned long physaddr,
                                          unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_WRITETHROUGH);
 }
-static inline void *z_remap_fullcache(unsigned long physaddr,
+static inline void __iomem *z_remap_fullcache(unsigned long physaddr,
                                       unsigned long size)
 {
         return __ioremap(physaddr, size, IOMAP_FULL_CACHING);
diff --git a/include/asm-m68knommu/machdep.h b/include/asm-m68knommu/machdep.h
index 5a9f9c297f79..27c90afd3339 100644
--- a/include/asm-m68knommu/machdep.h
+++ b/include/asm-m68knommu/machdep.h
@@ -38,7 +38,6 @@ extern void (*mach_power_off)( void );
 extern unsigned long (*mach_hd_init) (unsigned long, unsigned long);
 extern void (*mach_hd_setup)(char *, int *);
 extern long mach_max_dma_address;
-extern void (*mach_floppy_setup)(char *, int *);
 extern void (*mach_floppy_eject)(void);
 extern void (*mach_heartbeat) (int);
 extern void (*mach_l2_flush) (int);
diff --git a/include/asm-m68knommu/thread_info.h b/include/asm-m68knommu/thread_info.h
index 7b9a3fa3af5d..b8f009edf2b2 100644
--- a/include/asm-m68knommu/thread_info.h
+++ b/include/asm-m68knommu/thread_info.h
@@ -75,8 +75,6 @@ static inline struct thread_info *current_thread_info(void)
 #define alloc_thread_info(tsk) ((struct thread_info *) \
                                 __get_free_pages(GFP_KERNEL, THREAD_SIZE_ORDER))
 #define free_thread_info(ti)    free_pages((unsigned long) (ti), THREAD_SIZE_ORDER)
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
 #endif /* __ASSEMBLY__ */
 
 #define PREEMPT_ACTIVE  0x4000000
diff --git a/include/asm-mips/mach-ip27/topology.h b/include/asm-mips/mach-ip27/topology.h
index 82141c711c33..59d26b52ba32 100644
--- a/include/asm-mips/mach-ip27/topology.h
+++ b/include/asm-mips/mach-ip27/topology.h
@@ -27,7 +27,6 @@ extern unsigned char __node_distances[MAX_COMPACT_NODES][MAX_COMPACT_NODES];
         .max_interval           = 32,                   \
         .busy_factor            = 32,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000),            \
         .cache_nice_tries       = 1,                    \
         .per_cpu_gain           = 100,                  \
         .flags                  = SD_LOAD_BALANCE       \
diff --git a/include/asm-mips/processor.h b/include/asm-mips/processor.h
index de53055a62ae..39d2bd50fece 100644
--- a/include/asm-mips/processor.h
+++ b/include/asm-mips/processor.h
@@ -200,11 +200,11 @@ extern void start_thread(struct pt_regs * regs, unsigned long pc, unsigned long
 
 unsigned long get_wchan(struct task_struct *p);
 
-#define __PT_REG(reg) ((long)&((struct pt_regs *)0)->reg - sizeof(struct pt_regs))
-#define __KSTK_TOS(tsk) ((unsigned long)(tsk->thread_info) + THREAD_SIZE - 32)
-#define KSTK_EIP(tsk) (*(unsigned long *)(__KSTK_TOS(tsk) + __PT_REG(cp0_epc)))
-#define KSTK_ESP(tsk) (*(unsigned long *)(__KSTK_TOS(tsk) + __PT_REG(regs[29])))
-#define KSTK_STATUS(tsk) (*(unsigned long *)(__KSTK_TOS(tsk) + __PT_REG(cp0_status)))
+#define __KSTK_TOS(tsk) ((unsigned long)task_stack_page(tsk) + THREAD_SIZE - 32)
+#define task_pt_regs(tsk) ((struct pt_regs *)__KSTK_TOS(tsk) - 1)
+#define KSTK_EIP(tsk) (task_pt_regs(tsk)->cp0_epc)
+#define KSTK_ESP(tsk) (task_pt_regs(tsk)->regs[29])
+#define KSTK_STATUS(tsk) (task_pt_regs(tsk)->cp0_status)
 
 #define cpu_relax()     barrier()
 
diff --git a/include/asm-mips/system.h b/include/asm-mips/system.h
index 330c4e497af3..e8e5d4143377 100644
--- a/include/asm-mips/system.h
+++ b/include/asm-mips/system.h
@@ -159,11 +159,21 @@ struct task_struct;
 do {                                                                    \
         if (cpu_has_dsp)                                                \
                 __save_dsp(prev);                                       \
-        (last) = resume(prev, next, next->thread_info);                 \
+        (last) = resume(prev, next, task_thread_info(next));            \
         if (cpu_has_dsp)                                                \
                 __restore_dsp(current);                                 \
 } while(0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 static inline unsigned long __xchg_u32(volatile int * m, unsigned int val)
 {
         __u32 retval;
diff --git a/include/asm-mips/thread_info.h b/include/asm-mips/thread_info.h
index e6c24472e03f..1612b3fe1080 100644
--- a/include/asm-mips/thread_info.h
+++ b/include/asm-mips/thread_info.h
@@ -97,8 +97,6 @@ register struct thread_info *__current_thread_info __asm__("$28");
 #endif
 
 #define free_thread_info(info) kfree(info)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #endif /* !__ASSEMBLY__ */
 
diff --git a/include/asm-parisc/system.h b/include/asm-parisc/system.h
index f3928d3a80cb..a5a973c0c07f 100644
--- a/include/asm-parisc/system.h
+++ b/include/asm-parisc/system.h
@@ -49,6 +49,15 @@ extern struct task_struct *_switch_to(struct task_struct *, struct task_struct *
         (last) = _switch_to(prev, next);                        \
 } while(0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
 
 
 /* interrupt control */
diff --git a/include/asm-parisc/thread_info.h b/include/asm-parisc/thread_info.h
index 57bbb76cb6c1..ac32f140b83a 100644
--- a/include/asm-parisc/thread_info.h
+++ b/include/asm-parisc/thread_info.h
@@ -43,9 +43,6 @@ struct thread_info {
 #define alloc_thread_info(tsk) ((struct thread_info *) \
                         __get_free_pages(GFP_KERNEL, THREAD_ORDER))
 #define free_thread_info(ti)    free_pages((unsigned long) (ti), THREAD_ORDER)
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
-
 
 /* how to get the thread information struct from C */
 #define current_thread_info()   ((struct thread_info *)mfctl(30))
diff --git a/include/asm-powerpc/system.h b/include/asm-powerpc/system.h
index 4c888303e85b..9b822afa7d0e 100644
--- a/include/asm-powerpc/system.h
+++ b/include/asm-powerpc/system.h
@@ -183,6 +183,16 @@ struct thread_struct;
 extern struct task_struct *_switch(struct thread_struct *prev,
                                    struct thread_struct *next);
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 extern unsigned int rtas_data;
 extern int mem_init_done;       /* set on boot once kmalloc can be called */
 extern unsigned long memory_limit;
diff --git a/include/asm-powerpc/thread_info.h b/include/asm-powerpc/thread_info.h
index ac1e80e6033e..7e09d7cda933 100644
--- a/include/asm-powerpc/thread_info.h
+++ b/include/asm-powerpc/thread_info.h
@@ -89,9 +89,6 @@ struct thread_info {
 
 #endif /* THREAD_SHIFT < PAGE_SHIFT */
 
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
-
 /* how to get the thread information struct from C */
 static inline struct thread_info *current_thread_info(void)
 {
diff --git a/include/asm-powerpc/topology.h b/include/asm-powerpc/topology.h
index 9f3d4da261c4..1e19cd00af25 100644
--- a/include/asm-powerpc/topology.h
+++ b/include/asm-powerpc/topology.h
@@ -39,7 +39,6 @@ static inline int node_to_first_cpu(int node)
         .max_interval           = 32,                   \
         .busy_factor            = 32,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000000),         \
         .cache_nice_tries       = 1,                    \
         .per_cpu_gain           = 100,                  \
         .busy_idx               = 3,                    \
diff --git a/include/asm-ppc/system.h b/include/asm-ppc/system.h
index b97037348277..fb49c0c49ea1 100644
--- a/include/asm-ppc/system.h
+++ b/include/asm-ppc/system.h
@@ -131,6 +131,16 @@ extern struct task_struct *__switch_to(struct task_struct *,
         struct task_struct *);
 #define switch_to(prev, next, last)     ((last) = __switch_to((prev), (next)))
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 struct thread_struct;
 extern struct task_struct *_switch(struct thread_struct *prev,
                                    struct thread_struct *next);
diff --git a/include/asm-s390/elf.h b/include/asm-s390/elf.h
index 372d51cccd53..710646e64f7d 100644
--- a/include/asm-s390/elf.h
+++ b/include/asm-s390/elf.h
@@ -163,7 +163,7 @@ static inline int dump_regs(struct pt_regs *ptregs, elf_gregset_t *regs)
 
 static inline int dump_task_regs(struct task_struct *tsk, elf_gregset_t *regs)
 {
-        struct pt_regs *ptregs = __KSTK_PTREGS(tsk);
+        struct pt_regs *ptregs = task_pt_regs(tsk);
         memcpy(&regs->psw, &ptregs->psw, sizeof(regs->psw)+sizeof(regs->gprs));
         memcpy(regs->acrs, tsk->thread.acrs, sizeof(regs->acrs));
         regs->orig_gpr2 = ptregs->orig_gpr2;
diff --git a/include/asm-s390/processor.h b/include/asm-s390/processor.h
index 4ec652ebb3b1..c5cbc4bd8414 100644
--- a/include/asm-s390/processor.h
+++ b/include/asm-s390/processor.h
@@ -191,10 +191,10 @@ extern void show_registers(struct pt_regs *regs);
 extern void show_trace(struct task_struct *task, unsigned long *sp);
 
 unsigned long get_wchan(struct task_struct *p);
-#define __KSTK_PTREGS(tsk) ((struct pt_regs *) \
-        ((unsigned long) tsk->thread_info + THREAD_SIZE - sizeof(struct pt_regs)))
-#define KSTK_EIP(tsk)   (__KSTK_PTREGS(tsk)->psw.addr)
-#define KSTK_ESP(tsk)   (__KSTK_PTREGS(tsk)->gprs[15])
+#define task_pt_regs(tsk) ((struct pt_regs *) \
+        (task_stack_page(tsk) + THREAD_SIZE) - 1)
+#define KSTK_EIP(tsk)   (task_pt_regs(tsk)->psw.addr)
+#define KSTK_ESP(tsk)   (task_pt_regs(tsk)->gprs[15])
 
 /*
  * Give up the time slice of the virtual PU.
diff --git a/include/asm-s390/system.h b/include/asm-s390/system.h
index 864cae7e1fd6..c7c3a9ad593f 100644
--- a/include/asm-s390/system.h
+++ b/include/asm-s390/system.h
@@ -104,6 +104,16 @@ static inline void restore_access_regs(unsigned int *acrs)
         prev = __switch_to(prev,next);                                       \
 } while (0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 #ifdef CONFIG_VIRT_CPU_ACCOUNTING
 extern void account_user_vtime(struct task_struct *);
 extern void account_system_vtime(struct task_struct *);
diff --git a/include/asm-s390/thread_info.h b/include/asm-s390/thread_info.h
index 6c18a3f24316..f3797a52c4ea 100644
--- a/include/asm-s390/thread_info.h
+++ b/include/asm-s390/thread_info.h
@@ -81,8 +81,6 @@ static inline struct thread_info *current_thread_info(void)
 #define alloc_thread_info(tsk) ((struct thread_info *) \
         __get_free_pages(GFP_KERNEL,THREAD_ORDER))
 #define free_thread_info(ti) free_pages((unsigned long) (ti),THREAD_ORDER)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #endif
 
diff --git a/include/asm-sh/ptrace.h b/include/asm-sh/ptrace.h
index 0f75e16a7415..792fc35bd624 100644
--- a/include/asm-sh/ptrace.h
+++ b/include/asm-sh/ptrace.h
@@ -91,6 +91,16 @@ struct pt_dspregs {
 #define instruction_pointer(regs) ((regs)->pc)
 extern void show_regs(struct pt_regs *);
 
+#ifdef CONFIG_SH_DSP
+#define task_pt_regs(task) \
+        ((struct pt_regs *) (task_stack_page(task) + THREAD_SIZE \
+                 - sizeof(struct pt_dspregs) - sizeof(unsigned long)) - 1)
+#else
+#define task_pt_regs(task) \
+        ((struct pt_regs *) (task_stack_page(task) + THREAD_SIZE \
+                 - sizeof(unsigned long)) - 1)
+#endif
+
 static inline unsigned long profile_pc(struct pt_regs *regs)
 {
         unsigned long pc = instruction_pointer(regs);
diff --git a/include/asm-sh/system.h b/include/asm-sh/system.h
index 28a3c2d8bcd7..bb0330499bdf 100644
--- a/include/asm-sh/system.h
+++ b/include/asm-sh/system.h
@@ -57,6 +57,16 @@
         last = __last;                                                  \
 } while (0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 #define nop() __asm__ __volatile__ ("nop")
 
 
diff --git a/include/asm-sh/thread_info.h b/include/asm-sh/thread_info.h
index 46080cefaff8..85f0c11b4319 100644
--- a/include/asm-sh/thread_info.h
+++ b/include/asm-sh/thread_info.h
@@ -60,8 +60,6 @@ static inline struct thread_info *current_thread_info(void)
 #define THREAD_SIZE (2*PAGE_SIZE)
 #define alloc_thread_info(ti) ((struct thread_info *) __get_free_pages(GFP_KERNEL,1))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/asm-sh64/thread_info.h b/include/asm-sh64/thread_info.h
index 10f024c6a2e3..1f825cb163c3 100644
--- a/include/asm-sh64/thread_info.h
+++ b/include/asm-sh64/thread_info.h
@@ -66,8 +66,6 @@ static inline struct thread_info *current_thread_info(void)
 
 #define alloc_thread_info(ti) ((struct thread_info *) __get_free_pages(GFP_KERNEL,1))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #endif /* __ASSEMBLY__ */
 
diff --git a/include/asm-sparc/system.h b/include/asm-sparc/system.h
index 1f6b71f9e1b6..58dd162927bb 100644
--- a/include/asm-sparc/system.h
+++ b/include/asm-sparc/system.h
@@ -155,7 +155,7 @@ extern void fpsave(unsigned long *fpregs, unsigned long *fsr,
         "here:\n"                                                                       \
         : "=&r" (last)                                                                  \
         : "r" (&(current_set[hard_smp_processor_id()])),        \
-          "r" ((next)->thread_info),                            \
+          "r" (task_thread_info(next)),                         \
           "i" (TI_KPSR),                                        \
           "i" (TI_KSP),                                         \
           "i" (TI_TASK)                                         \
@@ -166,6 +166,16 @@ extern void fpsave(unsigned long *fpregs, unsigned long *fsr,
         } while(0)
 
 /*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
+/*
  * Changing the IRQ level on the Sparc.
  */
 extern void local_irq_restore(unsigned long);
diff --git a/include/asm-sparc/thread_info.h b/include/asm-sparc/thread_info.h
index ff6ccb3d24c6..65f060b040ab 100644
--- a/include/asm-sparc/thread_info.h
+++ b/include/asm-sparc/thread_info.h
@@ -92,9 +92,6 @@ BTFIXUPDEF_CALL(struct thread_info *, alloc_thread_info, void)
 BTFIXUPDEF_CALL(void, free_thread_info, struct thread_info *)
 #define free_thread_info(ti) BTFIXUP_CALL(free_thread_info)(ti)
 
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
-
 #endif /* __ASSEMBLY__ */
 
 /*
diff --git a/include/asm-sparc64/elf.h b/include/asm-sparc64/elf.h
index 91458118277e..69539a8ab833 100644
--- a/include/asm-sparc64/elf.h
+++ b/include/asm-sparc64/elf.h
@@ -119,7 +119,7 @@ typedef struct {
 #endif
 
 #define ELF_CORE_COPY_TASK_REGS(__tsk, __elf_regs)      \
-        ({ ELF_CORE_COPY_REGS((*(__elf_regs)), (__tsk)->thread_info->kregs); 1; })
+        ({ ELF_CORE_COPY_REGS((*(__elf_regs)), task_pt_regs(__tsk)); 1; })
 
 /*
  * This is used to ensure we don't load something for the wrong architecture.
diff --git a/include/asm-sparc64/mmu_context.h b/include/asm-sparc64/mmu_context.h
index 08ba72d7722c..57ee7b306189 100644
--- a/include/asm-sparc64/mmu_context.h
+++ b/include/asm-sparc64/mmu_context.h
@@ -60,7 +60,7 @@ do { \
         register unsigned long pgd_cache asm("o4"); \
         paddr = __pa((__mm)->pgd); \
         pgd_cache = 0UL; \
-        if ((__tsk)->thread_info->flags & _TIF_32BIT) \
+        if (task_thread_info(__tsk)->flags & _TIF_32BIT) \
                 pgd_cache = get_pgd_cache((__mm)->pgd); \
         __asm__ __volatile__("wrpr      %%g0, 0x494, %%pstate\n\t" \
                              "mov       %3, %%g4\n\t" \
diff --git a/include/asm-sparc64/processor.h b/include/asm-sparc64/processor.h
index 3169f3e2237e..cd8d9b4c8658 100644
--- a/include/asm-sparc64/processor.h
+++ b/include/asm-sparc64/processor.h
@@ -186,8 +186,9 @@ extern pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
 
 extern unsigned long get_wchan(struct task_struct *task);
 
-#define KSTK_EIP(tsk)  ((tsk)->thread_info->kregs->tpc)
-#define KSTK_ESP(tsk)  ((tsk)->thread_info->kregs->u_regs[UREG_FP])
+#define task_pt_regs(tsk) (task_thread_info(tsk)->kregs)
+#define KSTK_EIP(tsk)  (task_pt_regs(tsk)->tpc)
+#define KSTK_ESP(tsk)  (task_pt_regs(tsk)->u_regs[UREG_FP])
 
 #define cpu_relax()     barrier()
 
diff --git a/include/asm-sparc64/system.h b/include/asm-sparc64/system.h
index 309f1466b6fa..af254e581834 100644
--- a/include/asm-sparc64/system.h
+++ b/include/asm-sparc64/system.h
@@ -208,7 +208,7 @@ do {	if (test_thread_flag(TIF_PERFCTR)) {				\
         /* If you are tempted to conditionalize the following */        \
         /* so that ASI is only written if it changes, think again. */   \
         __asm__ __volatile__("wr %%g0, %0, %%asi"                       \
-        : : "r" (__thread_flag_byte_ptr(next->thread_info)[TI_FLAG_BYTE_CURRENT_DS]));\
+        : : "r" (__thread_flag_byte_ptr(task_thread_info(next))[TI_FLAG_BYTE_CURRENT_DS]));\
         __asm__ __volatile__(                                           \
         "mov    %%g4, %%g7\n\t"                                         \
         "wrpr   %%g0, 0x95, %%pstate\n\t"                               \
@@ -238,7 +238,7 @@ do {	if (test_thread_flag(TIF_PERFCTR)) {				\
         "b,a ret_from_syscall\n\t"                                      \
         "1:\n\t"                                                        \
         : "=&r" (last)                                                  \
-        : "0" (next->thread_info),                                      \
+        : "0" (task_thread_info(next)),                                 \
           "i" (TI_WSTATE), "i" (TI_KSP), "i" (TI_NEW_CHILD),            \
           "i" (TI_CWP), "i" (TI_TASK)                                   \
         : "cc",                                                         \
@@ -253,6 +253,16 @@ do {	if (test_thread_flag(TIF_PERFCTR)) {				\
         }                                                               \
 } while(0)
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ *
+ * TODO: fill this in!
+ */
+static inline void sched_cacheflush(void)
+{
+}
+
 static inline unsigned long xchg32(__volatile__ unsigned int *m, unsigned int val)
 {
         unsigned long tmp1, tmp2;
diff --git a/include/asm-um/thread_info.h b/include/asm-um/thread_info.h
index 97267f059ef5..705c71972c32 100644
--- a/include/asm-um/thread_info.h
+++ b/include/asm-um/thread_info.h
@@ -56,9 +56,6 @@ static inline struct thread_info *current_thread_info(void)
         ((struct thread_info *) kmalloc(THREAD_SIZE, GFP_KERNEL))
 #define free_thread_info(ti) kfree(ti)
 
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
-
 #endif
 
 #define PREEMPT_ACTIVE          0x10000000
diff --git a/include/asm-v850/processor.h b/include/asm-v850/processor.h
index 98f929427d3d..2d31308935a0 100644
--- a/include/asm-v850/processor.h
+++ b/include/asm-v850/processor.h
@@ -98,10 +98,10 @@ unsigned long get_wchan (struct task_struct *p);
 
 
 /* Return some info about the user process TASK.  */
-#define task_tos(task)  ((unsigned long)(task)->thread_info + THREAD_SIZE)
-#define task_regs(task) ((struct pt_regs *)task_tos (task) - 1)
-#define task_sp(task)   (task_regs (task)->gpr[GPR_SP])
-#define task_pc(task)   (task_regs (task)->pc)
+#define task_tos(task)  ((unsigned long)task_stack_page(task) + THREAD_SIZE)
+#define task_pt_regs(task) ((struct pt_regs *)task_tos (task) - 1)
+#define task_sp(task)   (task_pt_regs (task)->gpr[GPR_SP])
+#define task_pc(task)   (task_pt_regs (task)->pc)
 /* Grotty old names for some.  */
 #define KSTK_EIP(task)  task_pc (task)
 #define KSTK_ESP(task)  task_sp (task)
diff --git a/include/asm-v850/thread_info.h b/include/asm-v850/thread_info.h
index e4cfad94a553..82b8f2846207 100644
--- a/include/asm-v850/thread_info.h
+++ b/include/asm-v850/thread_info.h
@@ -58,8 +58,6 @@ struct thread_info {
 #define alloc_thread_info(tsk) ((struct thread_info *) \
                                 __get_free_pages(GFP_KERNEL, 1))
 #define free_thread_info(ti)    free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti)     get_task_struct((ti)->task)
-#define put_thread_info(ti)     put_task_struct((ti)->task)
 
 #endif /* __ASSEMBLY__ */
 
diff --git a/include/asm-x86_64/compat.h b/include/asm-x86_64/compat.h
index 3863a7da372b..b37ab8218ef0 100644
--- a/include/asm-x86_64/compat.h
+++ b/include/asm-x86_64/compat.h
@@ -198,7 +198,7 @@ static inline compat_uptr_t ptr_to_compat(void __user *uptr)
 
 static __inline__ void __user *compat_alloc_user_space(long len)
 {
-        struct pt_regs *regs = (void *)current->thread.rsp0 - sizeof(struct pt_regs); 
+        struct pt_regs *regs = task_pt_regs(current);
         return (void __user *)regs->rsp - len; 
 }
 
diff --git a/include/asm-x86_64/i387.h b/include/asm-x86_64/i387.h
index 57f7e1433849..876eb9a2fe78 100644
--- a/include/asm-x86_64/i387.h
+++ b/include/asm-x86_64/i387.h
@@ -30,7 +30,7 @@ extern int save_i387(struct _fpstate __user *buf);
  */
 
 #define unlazy_fpu(tsk) do { \
-        if ((tsk)->thread_info->status & TS_USEDFPU) \
+        if (task_thread_info(tsk)->status & TS_USEDFPU) \
                 save_init_fpu(tsk); \
 } while (0)
 
@@ -46,9 +46,9 @@ static inline void tolerant_fwait(void)
 }
 
 #define clear_fpu(tsk) do { \
-        if ((tsk)->thread_info->status & TS_USEDFPU) {          \
+        if (task_thread_info(tsk)->status & TS_USEDFPU) {       \
                 tolerant_fwait();                               \
-                (tsk)->thread_info->status &= ~TS_USEDFPU;      \
+                task_thread_info(tsk)->status &= ~TS_USEDFPU;   \
                 stts();                                         \
         }                                                       \
 } while (0)
@@ -170,10 +170,10 @@ static inline void kernel_fpu_end(void)
         preempt_enable();
 }
 
-static inline void save_init_fpu( struct task_struct *tsk )
+static inline void save_init_fpu(struct task_struct *tsk)
 {
         __fxsave_clear(tsk);
-        tsk->thread_info->status &= ~TS_USEDFPU;
+        task_thread_info(tsk)->status &= ~TS_USEDFPU;
         stts();
 }
 
diff --git a/include/asm-x86_64/processor.h b/include/asm-x86_64/processor.h
index 394dd729752d..87a282b1043a 100644
--- a/include/asm-x86_64/processor.h
+++ b/include/asm-x86_64/processor.h
@@ -321,8 +321,8 @@ extern long kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
 #define thread_saved_pc(t) (*(unsigned long *)((t)->thread.rsp - 8))
 
 extern unsigned long get_wchan(struct task_struct *p);
-#define KSTK_EIP(tsk) \
-        (((struct pt_regs *)(tsk->thread.rsp0 - sizeof(struct pt_regs)))->rip)
+#define task_pt_regs(tsk) ((struct pt_regs *)(tsk)->thread.rsp0 - 1)
+#define KSTK_EIP(tsk) (task_pt_regs(tsk)->rip)
 #define KSTK_ESP(tsk) -1 /* sorry. doesn't work for syscall. */
 
 
diff --git a/include/asm-x86_64/system.h b/include/asm-x86_64/system.h
index 38c1e8a69c9c..0eacbefb7dd0 100644
--- a/include/asm-x86_64/system.h
+++ b/include/asm-x86_64/system.h
@@ -193,6 +193,15 @@ static inline void write_cr4(unsigned long val)
 #define wbinvd() \
         __asm__ __volatile__ ("wbinvd": : :"memory");
 
+/*
+ * On SMP systems, when the scheduler does migration-cost autodetection,
+ * it needs a way to flush as much of the CPU's caches as possible.
+ */
+static inline void sched_cacheflush(void)
+{
+        wbinvd();
+}
+
 #endif  /* __KERNEL__ */
 
 #define nop() __asm__ __volatile__ ("nop")
diff --git a/include/asm-x86_64/thread_info.h b/include/asm-x86_64/thread_info.h
index eb7c5fda1870..4ac0e0a36934 100644
--- a/include/asm-x86_64/thread_info.h
+++ b/include/asm-x86_64/thread_info.h
@@ -76,8 +76,6 @@ static inline struct thread_info *stack_thread_info(void)
 #define alloc_thread_info(tsk) \
         ((struct thread_info *) __get_free_pages(GFP_KERNEL,THREAD_ORDER))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), THREAD_ORDER)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/asm-x86_64/topology.h b/include/asm-x86_64/topology.h
index 7d82bc56b9fa..2fa7f27381b4 100644
--- a/include/asm-x86_64/topology.h
+++ b/include/asm-x86_64/topology.h
@@ -39,7 +39,6 @@ extern int __node_distance(int, int);
         .max_interval           = 32,                   \
         .busy_factor            = 32,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (10*1000000),         \
         .cache_nice_tries       = 2,                    \
         .busy_idx               = 3,                    \
         .idle_idx               = 2,                    \
diff --git a/include/asm-xtensa/processor.h b/include/asm-xtensa/processor.h
index 9cab5e4298b9..d1d72ad36f08 100644
--- a/include/asm-xtensa/processor.h
+++ b/include/asm-xtensa/processor.h
@@ -184,12 +184,12 @@ extern int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
 #define release_segments(mm)    do { } while(0)
 #define forget_segments()       do { } while (0)
 
-#define thread_saved_pc(tsk)    (xtensa_pt_regs(tsk)->pc)
+#define thread_saved_pc(tsk)    (task_pt_regs(tsk)->pc)
 
 extern unsigned long get_wchan(struct task_struct *p);
 
-#define KSTK_EIP(tsk)           (xtensa_pt_regs(tsk)->pc)
-#define KSTK_ESP(tsk)           (xtensa_pt_regs(tsk)->areg[1])
+#define KSTK_EIP(tsk)           (task_pt_regs(tsk)->pc)
+#define KSTK_ESP(tsk)           (task_pt_regs(tsk)->areg[1])
 
 #define cpu_relax()  do { } while (0)
 
diff --git a/include/asm-xtensa/ptrace.h b/include/asm-xtensa/ptrace.h
index aa4fd7fb3ce7..a5ac71a5205c 100644
--- a/include/asm-xtensa/ptrace.h
+++ b/include/asm-xtensa/ptrace.h
@@ -113,8 +113,8 @@ struct pt_regs {
 };
 
 #ifdef __KERNEL__
-# define xtensa_pt_regs(tsk) ((struct pt_regs*) \
-  (((long)(tsk)->thread_info + KERNEL_STACK_SIZE - (XCHAL_NUM_AREGS-16)*4)) - 1)
+# define task_pt_regs(tsk) ((struct pt_regs*) \
+  (task_stack_page(tsk) + KERNEL_STACK_SIZE - (XCHAL_NUM_AREGS-16)*4) - 1)
 # define user_mode(regs) (((regs)->ps & 0x00000020)!=0)
 # define instruction_pointer(regs) ((regs)->pc)
 extern void show_regs(struct pt_regs *);
diff --git a/include/asm-xtensa/thread_info.h b/include/asm-xtensa/thread_info.h
index af208d41fd82..5ae34ab71597 100644
--- a/include/asm-xtensa/thread_info.h
+++ b/include/asm-xtensa/thread_info.h
@@ -93,8 +93,6 @@ static inline struct thread_info *current_thread_info(void)
 /* thread information allocation */
 #define alloc_thread_info(tsk) ((struct thread_info *) __get_free_pages(GFP_KERNEL,1))
 #define free_thread_info(ti) free_pages((unsigned long) (ti), 1)
-#define get_thread_info(ti) get_task_struct((ti)->task)
-#define put_thread_info(ti) put_task_struct((ti)->task)
 
 #else /* !__ASSEMBLY__ */
 
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 3b74c4bf2934..a72e17135421 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -631,7 +631,14 @@ struct sched_domain {
 
 extern void partition_sched_domains(cpumask_t *partition1,
                                     cpumask_t *partition2);
-#endif /* CONFIG_SMP */
+
+/*
+ * Maximum cache size the migration-costs auto-tuning code will
+ * search from:
+ */
+extern unsigned int max_cache_size;
+
+#endif  /* CONFIG_SMP */
 
 
 struct io_context;                      /* See blkdev.h */
@@ -689,9 +696,12 @@ struct task_struct {
 
         int lock_depth;         /* BKL lock depth */
 
-#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
+#if defined(CONFIG_SMP)
+        int last_waker_cpu;     /* CPU that last woke this task up */
+#if defined(__ARCH_WANT_UNLOCKED_CTXSW)
         int oncpu;
 #endif
+#endif
         int prio, static_prio;
         struct list_head run_list;
         prio_array_t *array;
@@ -1230,6 +1240,7 @@ static inline void task_unlock(struct task_struct *p)
 #ifndef __HAVE_THREAD_FUNCTIONS
 
 #define task_thread_info(task) (task)->thread_info
+#define task_stack_page(task) ((void*)((task)->thread_info))
 
 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
 {
diff --git a/include/linux/topology.h b/include/linux/topology.h
index 3df1d474e5c5..315a5163d6a0 100644
--- a/include/linux/topology.h
+++ b/include/linux/topology.h
@@ -86,7 +86,6 @@
         .max_interval           = 2,                    \
         .busy_factor            = 8,                    \
         .imbalance_pct          = 110,                  \
-        .cache_hot_time         = 0,                    \
         .cache_nice_tries       = 0,                    \
         .per_cpu_gain           = 25,                   \
         .busy_idx               = 0,                    \
@@ -117,7 +116,6 @@
         .max_interval           = 4,                    \
         .busy_factor            = 64,                   \
         .imbalance_pct          = 125,                  \
-        .cache_hot_time         = (5*1000000/2),        \
         .cache_nice_tries       = 1,                    \
         .per_cpu_gain           = 100,                  \
         .busy_idx               = 2,                    \
diff --git a/kernel/sched.c b/kernel/sched.c
index c0c60c926d5e..c9dec2aa1976 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -34,6 +34,7 @@
 #include <linux/notifier.h>
 #include <linux/profile.h>
 #include <linux/suspend.h>
+#include <linux/vmalloc.h>
 #include <linux/blkdev.h>
 #include <linux/delay.h>
 #include <linux/smp.h>
@@ -1289,6 +1290,9 @@ static int try_to_wake_up(task_t *p, unsigned int state, int sync)
                 }
         }
 
+        if (p->last_waker_cpu != this_cpu)
+                goto out_set_cpu;
+
         if (unlikely(!cpu_isset(this_cpu, p->cpus_allowed)))
                 goto out_set_cpu;
 
@@ -1359,6 +1363,8 @@ out_set_cpu:
                 cpu = task_cpu(p);
         }
 
+        p->last_waker_cpu = this_cpu;
+
 out_activate:
 #endif /* CONFIG_SMP */
         if (old_state == TASK_UNINTERRUPTIBLE) {
@@ -1440,9 +1446,12 @@ void fastcall sched_fork(task_t *p, int clone_flags)
 #ifdef CONFIG_SCHEDSTATS
         memset(&p->sched_info, 0, sizeof(p->sched_info));
 #endif
-#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
+#if defined(CONFIG_SMP)
+        p->last_waker_cpu = cpu;
+#if defined(__ARCH_WANT_UNLOCKED_CTXSW)
         p->oncpu = 0;
 #endif
+#endif
 #ifdef CONFIG_PREEMPT
         /* Want to start with kernel preemption disabled. */
         task_thread_info(p)->preempt_count = 1;
@@ -5082,7 +5091,470 @@ static void init_sched_build_groups(struct sched_group groups[], cpumask_t span,
 
 #define SD_NODES_PER_DOMAIN 16
 
+/*
+ * Self-tuning task migration cost measurement between source and target CPUs.
+ *
+ * This is done by measuring the cost of manipulating buffers of varying
+ * sizes. For a given buffer-size here are the steps that are taken:
+ *
+ * 1) the source CPU reads+dirties a shared buffer
+ * 2) the target CPU reads+dirties the same shared buffer
+ *
+ * We measure how long they take, in the following 4 scenarios:
+ *
+ *  - source: CPU1, target: CPU2 | cost1
+ *  - source: CPU2, target: CPU1 | cost2
+ *  - source: CPU1, target: CPU1 | cost3
+ *  - source: CPU2, target: CPU2 | cost4
+ *
+ * We then calculate the cost3+cost4-cost1-cost2 difference - this is
+ * the cost of migration.
+ *
+ * We then start off from a small buffer-size and iterate up to larger
+ * buffer sizes, in 5% steps - measuring each buffer-size separately, and
+ * doing a maximum search for the cost. (The maximum cost for a migration
+ * normally occurs when the working set size is around the effective cache
+ * size.)
+ */
+#define SEARCH_SCOPE            2
+#define MIN_CACHE_SIZE          (64*1024U)
+#define DEFAULT_CACHE_SIZE      (5*1024*1024U)
+#define ITERATIONS              2
+#define SIZE_THRESH             130
+#define COST_THRESH             130
+
+/*
+ * The migration cost is a function of 'domain distance'. Domain
+ * distance is the number of steps a CPU has to iterate down its
+ * domain tree to share a domain with the other CPU. The farther
+ * two CPUs are from each other, the larger the distance gets.
+ *
+ * Note that we use the distance only to cache measurement results,
+ * the distance value is not used numerically otherwise. When two
+ * CPUs have the same distance it is assumed that the migration
+ * cost is the same. (this is a simplification but quite practical)
+ */
+#define MAX_DOMAIN_DISTANCE 32
+
+static unsigned long long migration_cost[MAX_DOMAIN_DISTANCE] =
+                { [ 0 ... MAX_DOMAIN_DISTANCE-1 ] = -1LL };
+
+/*
+ * Allow override of migration cost - in units of microseconds.
+ * E.g. migration_cost=1000,2000,3000 will set up a level-1 cost
+ * of 1 msec, level-2 cost of 2 msecs and level3 cost of 3 msecs:
+ */
+static int __init migration_cost_setup(char *str)
+{
+        int ints[MAX_DOMAIN_DISTANCE+1], i;
+
+        str = get_options(str, ARRAY_SIZE(ints), ints);
+
+        printk("#ints: %d\n", ints[0]);
+        for (i = 1; i <= ints[0]; i++) {
+                migration_cost[i-1] = (unsigned long long)ints[i]*1000;
+                printk("migration_cost[%d]: %Ld\n", i-1, migration_cost[i-1]);
+        }
+        return 1;
+}
+
+__setup ("migration_cost=", migration_cost_setup);
+
+/*
+ * Global multiplier (divisor) for migration-cutoff values,
+ * in percentiles. E.g. use a value of 150 to get 1.5 times
+ * longer cache-hot cutoff times.
+ *
+ * (We scale it from 100 to 128 to long long handling easier.)
+ */
+
+#define MIGRATION_FACTOR_SCALE 128
+
+static unsigned int migration_factor = MIGRATION_FACTOR_SCALE;
+
+static int __init setup_migration_factor(char *str)
+{
+        get_option(&str, &migration_factor);
+        migration_factor = migration_factor * MIGRATION_FACTOR_SCALE / 100;
+        return 1;
+}
+
+__setup("migration_factor=", setup_migration_factor);
+
+/*
+ * Estimated distance of two CPUs, measured via the number of domains
+ * we have to pass for the two CPUs to be in the same span:
+ */
+static unsigned long domain_distance(int cpu1, int cpu2)
+{
+        unsigned long distance = 0;
+        struct sched_domain *sd;
+
+        for_each_domain(cpu1, sd) {
+                WARN_ON(!cpu_isset(cpu1, sd->span));
+                if (cpu_isset(cpu2, sd->span))
+                        return distance;
+                distance++;
+        }
+        if (distance >= MAX_DOMAIN_DISTANCE) {
+                WARN_ON(1);
+                distance = MAX_DOMAIN_DISTANCE-1;
+        }
+
+        return distance;
+}
+
+static unsigned int migration_debug;
+
+static int __init setup_migration_debug(char *str)
+{
+        get_option(&str, &migration_debug);
+        return 1;
+}
+
+__setup("migration_debug=", setup_migration_debug);
+
+/*
+ * Maximum cache-size that the scheduler should try to measure.
+ * Architectures with larger caches should tune this up during
+ * bootup. Gets used in the domain-setup code (i.e. during SMP
+ * bootup).
+ */
+unsigned int max_cache_size;
+
+static int __init setup_max_cache_size(char *str)
+{
+        get_option(&str, &max_cache_size);
+        return 1;
+}
+
+__setup("max_cache_size=", setup_max_cache_size);
+
+/*
+ * Dirty a big buffer in a hard-to-predict (for the L2 cache) way. This
+ * is the operation that is timed, so we try to generate unpredictable
+ * cachemisses that still end up filling the L2 cache:
+ */
+static void touch_cache(void *__cache, unsigned long __size)
+{
+        unsigned long size = __size/sizeof(long), chunk1 = size/3,
+                        chunk2 = 2*size/3;
+        unsigned long *cache = __cache;
+        int i;
+
+        for (i = 0; i < size/6; i += 8) {
+                switch (i % 6) {
+                        case 0: cache[i]++;
+                        case 1: cache[size-1-i]++;
+                        case 2: cache[chunk1-i]++;
+                        case 3: cache[chunk1+i]++;
+                        case 4: cache[chunk2-i]++;
+                        case 5: cache[chunk2+i]++;
+                }
+        }
+}
+
+/*
+ * Measure the cache-cost of one task migration. Returns in units of nsec.
+ */
+static unsigned long long measure_one(void *cache, unsigned long size,
+                                      int source, int target)
+{
+        cpumask_t mask, saved_mask;
+        unsigned long long t0, t1, t2, t3, cost;
+
+        saved_mask = current->cpus_allowed;
+
+        /*
+         * Flush source caches to RAM and invalidate them:
+         */
+        sched_cacheflush();
+
+        /*
+         * Migrate to the source CPU:
+         */
+        mask = cpumask_of_cpu(source);
+        set_cpus_allowed(current, mask);
+        WARN_ON(smp_processor_id() != source);
+
+        /*
+         * Dirty the working set:
+         */
+        t0 = sched_clock();
+        touch_cache(cache, size);
+        t1 = sched_clock();
+
+        /*
+         * Migrate to the target CPU, dirty the L2 cache and access
+         * the shared buffer. (which represents the working set
+         * of a migrated task.)
+         */
+        mask = cpumask_of_cpu(target);
+        set_cpus_allowed(current, mask);
+        WARN_ON(smp_processor_id() != target);
+
+        t2 = sched_clock();
+        touch_cache(cache, size);
+        t3 = sched_clock();
+
+        cost = t1-t0 + t3-t2;
+
+        if (migration_debug >= 2)
+                printk("[%d->%d]: %8Ld %8Ld %8Ld => %10Ld.\n",
+                        source, target, t1-t0, t1-t0, t3-t2, cost);
+        /*
+         * Flush target caches to RAM and invalidate them:
+         */
+        sched_cacheflush();
+
+        set_cpus_allowed(current, saved_mask);
+
+        return cost;
+}
+
+/*
+ * Measure a series of task migrations and return the average
+ * result. Since this code runs early during bootup the system
+ * is 'undisturbed' and the average latency makes sense.
+ *
+ * The algorithm in essence auto-detects the relevant cache-size,
+ * so it will properly detect different cachesizes for different
+ * cache-hierarchies, depending on how the CPUs are connected.
+ *
+ * Architectures can prime the upper limit of the search range via
+ * max_cache_size, otherwise the search range defaults to 20MB...64K.
+ */
+static unsigned long long
+measure_cost(int cpu1, int cpu2, void *cache, unsigned int size)
+{
+        unsigned long long cost1, cost2;
+        int i;
+
+        /*
+         * Measure the migration cost of 'size' bytes, over an
+         * average of 10 runs:
+         *
+         * (We perturb the cache size by a small (0..4k)
+         *  value to compensate size/alignment related artifacts.
+         *  We also subtract the cost of the operation done on
+         *  the same CPU.)
+         */
+        cost1 = 0;
+
+        /*
+         * dry run, to make sure we start off cache-cold on cpu1,
+         * and to get any vmalloc pagefaults in advance:
+         */
+        measure_one(cache, size, cpu1, cpu2);
+        for (i = 0; i < ITERATIONS; i++)
+                cost1 += measure_one(cache, size - i*1024, cpu1, cpu2);
+
+        measure_one(cache, size, cpu2, cpu1);
+        for (i = 0; i < ITERATIONS; i++)
+                cost1 += measure_one(cache, size - i*1024, cpu2, cpu1);
+
+        /*
+         * (We measure the non-migrating [cached] cost on both
+         *  cpu1 and cpu2, to handle CPUs with different speeds)
+         */
+        cost2 = 0;
+
+        measure_one(cache, size, cpu1, cpu1);
+        for (i = 0; i < ITERATIONS; i++)
+                cost2 += measure_one(cache, size - i*1024, cpu1, cpu1);
+
+        measure_one(cache, size, cpu2, cpu2);
+        for (i = 0; i < ITERATIONS; i++)
+                cost2 += measure_one(cache, size - i*1024, cpu2, cpu2);
+
+        /*
+         * Get the per-iteration migration cost:
+         */
+        do_div(cost1, 2*ITERATIONS);
+        do_div(cost2, 2*ITERATIONS);
+
+        return cost1 - cost2;
+}
+
+static unsigned long long measure_migration_cost(int cpu1, int cpu2)
+{
+        unsigned long long max_cost = 0, fluct = 0, avg_fluct = 0;
+        unsigned int max_size, size, size_found = 0;
+        long long cost = 0, prev_cost;
+        void *cache;
+
+        /*
+         * Search from max_cache_size*5 down to 64K - the real relevant
+         * cachesize has to lie somewhere inbetween.
+         */
+        if (max_cache_size) {
+                max_size = max(max_cache_size * SEARCH_SCOPE, MIN_CACHE_SIZE);
+                size = max(max_cache_size / SEARCH_SCOPE, MIN_CACHE_SIZE);
+        } else {
+                /*
+                 * Since we have no estimation about the relevant
+                 * search range
+                 */
+                max_size = DEFAULT_CACHE_SIZE * SEARCH_SCOPE;
+                size = MIN_CACHE_SIZE;
+        }
+
+        if (!cpu_online(cpu1) || !cpu_online(cpu2)) {
+                printk("cpu %d and %d not both online!\n", cpu1, cpu2);
+                return 0;
+        }
+
+        /*
+         * Allocate the working set:
+         */
+        cache = vmalloc(max_size);
+        if (!cache) {
+                printk("could not vmalloc %d bytes for cache!\n", 2*max_size);
+                return 1000000; // return 1 msec on very small boxen
+        }
+
+        while (size <= max_size) {
+                prev_cost = cost;
+                cost = measure_cost(cpu1, cpu2, cache, size);
+
+                /*
+                 * Update the max:
+                 */
+                if (cost > 0) {
+                        if (max_cost < cost) {
+                                max_cost = cost;
+                                size_found = size;
+                        }
+                }
+                /*
+                 * Calculate average fluctuation, we use this to prevent
+                 * noise from triggering an early break out of the loop:
+                 */
+                fluct = abs(cost - prev_cost);
+                avg_fluct = (avg_fluct + fluct)/2;
+
+                if (migration_debug)
+                        printk("-> [%d][%d][%7d] %3ld.%ld [%3ld.%ld] (%ld): (%8Ld %8Ld)\n",
+                                cpu1, cpu2, size,
+                                (long)cost / 1000000,
+                                ((long)cost / 100000) % 10,
+                                (long)max_cost / 1000000,
+                                ((long)max_cost / 100000) % 10,
+                                domain_distance(cpu1, cpu2),
+                                cost, avg_fluct);
+
+                /*
+                 * If we iterated at least 20% past the previous maximum,
+                 * and the cost has dropped by more than 20% already,
+                 * (taking fluctuations into account) then we assume to
+                 * have found the maximum and break out of the loop early:
+                 */
+                if (size_found && (size*100 > size_found*SIZE_THRESH))
+                        if (cost+avg_fluct <= 0 ||
+                                max_cost*100 > (cost+avg_fluct)*COST_THRESH) {
+
+                                if (migration_debug)
+                                        printk("-> found max.\n");
+                                break;
+                        }
+                /*
+                 * Increase the cachesize in 5% steps:
+                 */
+                size = size * 20 / 19;
+        }
+
+        if (migration_debug)
+                printk("[%d][%d] working set size found: %d, cost: %Ld\n",
+                        cpu1, cpu2, size_found, max_cost);
+
+        vfree(cache);
+
+        /*
+         * A task is considered 'cache cold' if at least 2 times
+         * the worst-case cost of migration has passed.
+         *
+         * (this limit is only listened to if the load-balancing
+         * situation is 'nice' - if there is a large imbalance we
+         * ignore it for the sake of CPU utilization and
+         * processing fairness.)
+         */
+        return 2 * max_cost * migration_factor / MIGRATION_FACTOR_SCALE;
+}
+
+static void calibrate_migration_costs(const cpumask_t *cpu_map)
+{
+        int cpu1 = -1, cpu2 = -1, cpu, orig_cpu = raw_smp_processor_id();
+        unsigned long j0, j1, distance, max_distance = 0;
+        struct sched_domain *sd;
+
+        j0 = jiffies;
+
+        /*
+         * First pass - calculate the cacheflush times:
+         */
+        for_each_cpu_mask(cpu1, *cpu_map) {
+                for_each_cpu_mask(cpu2, *cpu_map) {
+                        if (cpu1 == cpu2)
+                                continue;
+                        distance = domain_distance(cpu1, cpu2);
+                        max_distance = max(max_distance, distance);
+                        /*
+                         * No result cached yet?
+                         */
+                        if (migration_cost[distance] == -1LL)
+                                migration_cost[distance] =
+                                        measure_migration_cost(cpu1, cpu2);
+                }
+        }
+        /*
+         * Second pass - update the sched domain hierarchy with
+         * the new cache-hot-time estimations:
+         */
+        for_each_cpu_mask(cpu, *cpu_map) {
+                distance = 0;
+                for_each_domain(cpu, sd) {
+                        sd->cache_hot_time = migration_cost[distance];
+                        distance++;
+                }
+        }
+        /*
+         * Print the matrix:
+         */
+        if (migration_debug)
+                printk("migration: max_cache_size: %d, cpu: %d MHz:\n",
+                        max_cache_size,
+#ifdef CONFIG_X86
+                        cpu_khz/1000
+#else
+                        -1
+#endif
+                );
+        printk("migration_cost=");
+        for (distance = 0; distance <= max_distance; distance++) {
+                if (distance)
+                        printk(",");
+                printk("%ld", (long)migration_cost[distance] / 1000);
+        }
+        printk("\n");
+        j1 = jiffies;
+        if (migration_debug)
+                printk("migration: %ld seconds\n", (j1-j0)/HZ);
+
+        /*
+         * Move back to the original CPU. NUMA-Q gets confused
+         * if we migrate to another quad during bootup.
+         */
+        if (raw_smp_processor_id() != orig_cpu) {
+                cpumask_t mask = cpumask_of_cpu(orig_cpu),
+                        saved_mask = current->cpus_allowed;
+
+                set_cpus_allowed(current, mask);
+                set_cpus_allowed(current, saved_mask);
+        }
+}
+
 #ifdef CONFIG_NUMA
+
 /**
  * find_next_best_node - find the next node to include in a sched_domain
  * @node: node whose sched_domain we're building
@@ -5448,6 +5920,10 @@ next_sg:
 #endif
                 cpu_attach_domain(sd, i);
         }
+        /*
+         * Tune cache-hot values:
+         */
+        calibrate_migration_costs(cpu_map);
 }
 /*
  * Set up scheduler domains and groups.  Callers must hold the hotplug lock.
diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 1850d0aef4ac..b62cab575a84 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -208,6 +208,8 @@ static int check_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
                 page = vm_normal_page(vma, addr, *pte);
                 if (!page)
                         continue;
+                if (PageReserved(page))
+                        continue;
                 nid = page_to_nid(page);
                 if (node_isset(nid, *nodes) == !!(flags & MPOL_MF_INVERT))
                         continue;
@@ -290,7 +292,7 @@ static inline int check_pgd_range(struct vm_area_struct *vma,
 static inline int vma_migratable(struct vm_area_struct *vma)
 {
         if (vma->vm_flags & (
-                VM_LOCKED|VM_IO|VM_HUGETLB|VM_PFNMAP))
+                VM_LOCKED|VM_IO|VM_HUGETLB|VM_PFNMAP|VM_RESERVED))
                 return 0;
         return 1;
 }
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index d41a0662d4da..8c960b469593 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1742,7 +1742,7 @@ void __devinit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
         unsigned long end_pfn = start_pfn + size;
         unsigned long pfn;
 
-        for (pfn = start_pfn; pfn < end_pfn; pfn++, page++) {
+        for (pfn = start_pfn; pfn < end_pfn; pfn++) {
                 if (!early_pfn_valid(pfn))
                         continue;
                 page = pfn_to_page(pfn);
diff --git a/mm/swap.c b/mm/swap.c
index ee6d71ccfa56..cbb48e721ab9 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -384,6 +384,8 @@ unsigned pagevec_lookup(struct pagevec *pvec, struct address_space *mapping,
         return pagevec_count(pvec);
 }
 
+EXPORT_SYMBOL(pagevec_lookup);
+
 unsigned pagevec_lookup_tag(struct pagevec *pvec, struct address_space *mapping,
                 pgoff_t *index, int tag, unsigned nr_pages)
 {
diff --git a/mm/tiny-shmem.c b/mm/tiny-shmem.c
index cdc6d431972b..f9d6a9cc91c4 100644
--- a/mm/tiny-shmem.c
+++ b/mm/tiny-shmem.c
@@ -90,7 +90,7 @@ struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags)
         file->f_mode = FMODE_WRITE | FMODE_READ;
 
         /* notify everyone as to the change of file size */
-        error = do_truncate(dentry, size, file);
+        error = do_truncate(dentry, size, 0, file);
         if (error < 0)
                 goto close_file;
 
diff --git a/sound/oss/dmasound/dmasound.h b/sound/oss/dmasound/dmasound.h
index 222014cafc1a..a1b0b92af4b5 100644
--- a/sound/oss/dmasound/dmasound.h
+++ b/sound/oss/dmasound/dmasound.h
@@ -270,7 +270,6 @@ extern int dmasound_catchRadius;
 #define SW_INPUT_VOLUME_SCALE   4
 #define SW_INPUT_VOLUME_DEFAULT (128 / SW_INPUT_VOLUME_SCALE)
 
-extern int expand_bal;  /* Balance factor for expanding (not volume!) */
 extern int expand_read_bal;     /* Balance factor for reading */
 extern uint software_input_volume; /* software implemented recording volume! */
 
diff --git a/sound/oss/dmasound/dmasound_atari.c b/sound/oss/dmasound/dmasound_atari.c
index 59eb53f89318..dc31373069a5 100644
--- a/sound/oss/dmasound/dmasound_atari.c
+++ b/sound/oss/dmasound/dmasound_atari.c
@@ -67,46 +67,46 @@ static int expand_data;	/* Data for expanding */
  * ++geert: split in even more functions (one per format)
  */
 
-static ssize_t ata_ct_law(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_law(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft);
-static ssize_t ata_ct_s8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s8(const u_char __user *userPtr, size_t userCount,
                          u_char frame[], ssize_t *frameUsed,
                          ssize_t frameLeft);
-static ssize_t ata_ct_u8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u8(const u_char __user *userPtr, size_t userCount,
                          u_char frame[], ssize_t *frameUsed,
                          ssize_t frameLeft);
-static ssize_t ata_ct_s16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s16be(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft);
-static ssize_t ata_ct_u16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u16be(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft);
-static ssize_t ata_ct_s16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s16le(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft);
-static ssize_t ata_ct_u16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u16le(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft);
-static ssize_t ata_ctx_law(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_law(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft);
-static ssize_t ata_ctx_s8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft);
-static ssize_t ata_ctx_u8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft);
-static ssize_t ata_ctx_s16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s16be(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft);
-static ssize_t ata_ctx_u16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u16be(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft);
-static ssize_t ata_ctx_s16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s16le(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft);
-static ssize_t ata_ctx_u16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u16le(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft);
 
@@ -151,7 +151,7 @@ static int FalconStateInfo(char *buffer, size_t space);
 /*** Translations ************************************************************/
 
 
-static ssize_t ata_ct_law(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_law(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft)
 {
@@ -176,7 +176,7 @@ static ssize_t ata_ct_law(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_s8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s8(const u_char __user *userPtr, size_t userCount,
                          u_char frame[], ssize_t *frameUsed,
                          ssize_t frameLeft)
 {
@@ -194,7 +194,7 @@ static ssize_t ata_ct_s8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_u8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u8(const u_char __user *userPtr, size_t userCount,
                          u_char frame[], ssize_t *frameUsed,
                          ssize_t frameLeft)
 {
@@ -217,8 +217,9 @@ static ssize_t ata_ct_u8(const u_char *userPtr, size_t userCount,
                 used = count*2;
                 while (count > 0) {
                         u_short data;
-                        if (get_user(data, ((u_short *)userPtr)++))
+                        if (get_user(data, (u_short __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 2;
                         *p++ = data ^ 0x8080;
                         count--;
                 }
@@ -228,7 +229,7 @@ static ssize_t ata_ct_u8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_s16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s16be(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -240,8 +241,9 @@ static ssize_t ata_ct_s16be(const u_char *userPtr, size_t userCount,
                 used = count*2;
                 while (count > 0) {
                         u_short data;
-                        if (get_user(data, ((u_short *)userPtr)++))
+                        if (get_user(data, (u_short __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 2;
                         *p++ = data;
                         *p++ = data;
                         count--;
@@ -259,7 +261,7 @@ static ssize_t ata_ct_s16be(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_u16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u16be(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -271,8 +273,9 @@ static ssize_t ata_ct_u16be(const u_char *userPtr, size_t userCount,
                 used = count*2;
                 while (count > 0) {
                         u_short data;
-                        if (get_user(data, ((u_short *)userPtr)++))
+                        if (get_user(data, (u_short __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 2;
                         data ^= 0x8000;
                         *p++ = data;
                         *p++ = data;
@@ -284,9 +287,10 @@ static ssize_t ata_ct_u16be(const u_char *userPtr, size_t userCount,
                 count = min_t(unsigned long, userCount, frameLeft)>>2;
                 used = count*4;
                 while (count > 0) {
-                        u_long data;
-                        if (get_user(data, ((u_int *)userPtr)++))
+                        u_int data;
+                        if (get_user(data, (u_int __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 4;
                         *p++ = data ^ 0x80008000;
                         count--;
                 }
@@ -296,7 +300,7 @@ static ssize_t ata_ct_u16be(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_s16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_s16le(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -309,8 +313,9 @@ static ssize_t ata_ct_s16le(const u_char *userPtr, size_t userCount,
                 used = count*2;
                 while (count > 0) {
                         u_short data;
-                        if (get_user(data, ((u_short *)userPtr)++))
+                        if (get_user(data, (u_short __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 2;
                         data = le2be16(data);
                         *p++ = data;
                         *p++ = data;
@@ -323,8 +328,9 @@ static ssize_t ata_ct_s16le(const u_char *userPtr, size_t userCount,
                 used = count*4;
                 while (count > 0) {
                         u_long data;
-                        if (get_user(data, ((u_int *)userPtr)++))
+                        if (get_user(data, (u_int __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 4;
                         data = le2be16dbl(data);
                         *p++ = data;
                         count--;
@@ -335,7 +341,7 @@ static ssize_t ata_ct_s16le(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ct_u16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ct_u16le(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -348,8 +354,9 @@ static ssize_t ata_ct_u16le(const u_char *userPtr, size_t userCount,
                 used = count*2;
                 while (count > 0) {
                         u_short data;
-                        if (get_user(data, ((u_short *)userPtr)++))
+                        if (get_user(data, (u_short __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 2;
                         data = le2be16(data) ^ 0x8000;
                         *p++ = data;
                         *p++ = data;
@@ -361,8 +368,9 @@ static ssize_t ata_ct_u16le(const u_char *userPtr, size_t userCount,
                 used = count;
                 while (count > 0) {
                         u_long data;
-                        if (get_user(data, ((u_int *)userPtr)++))
+                        if (get_user(data, (u_int __user *)userPtr))
                                 return -EFAULT;
+                        userPtr += 4;
                         data = le2be16dbl(data) ^ 0x80008000;
                         *p++ = data;
                         count--;
@@ -373,7 +381,7 @@ static ssize_t ata_ct_u16le(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_law(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_law(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
@@ -435,7 +443,7 @@ static ssize_t ata_ctx_law(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_s8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft)
 {
@@ -470,8 +478,9 @@ static ssize_t ata_ctx_s8(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 userCount -= 2;
                                 bal += hSpeed;
                         }
@@ -488,7 +497,7 @@ static ssize_t ata_ctx_s8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_u8(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft)
 {
@@ -524,8 +533,9 @@ static ssize_t ata_ctx_u8(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 data ^= 0x8080;
                                 userCount -= 2;
                                 bal += hSpeed;
@@ -543,7 +553,7 @@ static ssize_t ata_ctx_u8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_s16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s16be(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft)
 {
@@ -561,8 +571,9 @@ static ssize_t ata_ctx_s16be(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 userCount -= 2;
                                 bal += hSpeed;
                         }
@@ -579,8 +590,9 @@ static ssize_t ata_ctx_s16be(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 4)
                                         break;
-                                if (get_user(data, ((u_int *)userPtr)++))
+                                if (get_user(data, (u_int __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 4;
                                 userCount -= 4;
                                 bal += hSpeed;
                         }
@@ -597,7 +609,7 @@ static ssize_t ata_ctx_s16be(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_u16be(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u16be(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft)
 {
@@ -615,8 +627,9 @@ static ssize_t ata_ctx_u16be(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 data ^= 0x8000;
                                 userCount -= 2;
                                 bal += hSpeed;
@@ -634,8 +647,9 @@ static ssize_t ata_ctx_u16be(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 4)
                                         break;
-                                if (get_user(data, ((u_int *)userPtr)++))
+                                if (get_user(data, (u_int __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 4;
                                 data ^= 0x80008000;
                                 userCount -= 4;
                                 bal += hSpeed;
@@ -653,7 +667,7 @@ static ssize_t ata_ctx_u16be(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_s16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_s16le(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft)
 {
@@ -671,8 +685,9 @@ static ssize_t ata_ctx_s16le(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 data = le2be16(data);
                                 userCount -= 2;
                                 bal += hSpeed;
@@ -690,8 +705,9 @@ static ssize_t ata_ctx_s16le(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 4)
                                         break;
-                                if (get_user(data, ((u_int *)userPtr)++))
+                                if (get_user(data, (u_int __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 4;
                                 data = le2be16dbl(data);
                                 userCount -= 4;
                                 bal += hSpeed;
@@ -709,7 +725,7 @@ static ssize_t ata_ctx_s16le(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t ata_ctx_u16le(const u_char *userPtr, size_t userCount,
+static ssize_t ata_ctx_u16le(const u_char __user *userPtr, size_t userCount,
                              u_char frame[], ssize_t *frameUsed,
                              ssize_t frameLeft)
 {
@@ -727,8 +743,9 @@ static ssize_t ata_ctx_u16le(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 2)
                                         break;
-                                if (get_user(data, ((u_short *)userPtr)++))
+                                if (get_user(data, (u_short __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 2;
                                 data = le2be16(data) ^ 0x8000;
                                 userCount -= 2;
                                 bal += hSpeed;
@@ -746,8 +763,9 @@ static ssize_t ata_ctx_u16le(const u_char *userPtr, size_t userCount,
                         if (bal < 0) {
                                 if (userCount < 4)
                                         break;
-                                if (get_user(data, ((u_int *)userPtr)++))
+                                if (get_user(data, (u_int __user *)userPtr))
                                         return -EFAULT;
+                                userPtr += 4;
                                 data = le2be16dbl(data) ^ 0x80008000;
                                 userCount -= 4;
                                 bal += hSpeed;
diff --git a/sound/oss/dmasound/dmasound_paula.c b/sound/oss/dmasound/dmasound_paula.c
index d59f60b26410..494070a3f870 100644
--- a/sound/oss/dmasound/dmasound_paula.c
+++ b/sound/oss/dmasound/dmasound_paula.c
@@ -34,6 +34,7 @@
 #define DMASOUND_PAULA_REVISION 0
 #define DMASOUND_PAULA_EDITION 4
 
+#define custom amiga_custom
    /*
     *   The minimum period for audio depends on htotal (for OCS/ECS/AGA)
     *   (Imported from arch/m68k/amiga/amisound.c)
@@ -156,7 +157,7 @@ static int AmiStateInfo(char *buffer, size_t space);
      *  Native format
      */
 
-static ssize_t ami_ct_s8(const u_char *userPtr, size_t userCount,
+static ssize_t ami_ct_s8(const u_char __user *userPtr, size_t userCount,
                          u_char frame[], ssize_t *frameUsed, ssize_t frameLeft)
 {
         ssize_t count, used;
@@ -189,7 +190,7 @@ static ssize_t ami_ct_s8(const u_char *userPtr, size_t userCount,
      */
 
 #define GENERATE_AMI_CT8(funcname, convsample)                          \
-static ssize_t funcname(const u_char *userPtr, size_t userCount,        \
+static ssize_t funcname(const u_char __user *userPtr, size_t userCount, \
                         u_char frame[], ssize_t *frameUsed,             \
                         ssize_t frameLeft)                              \
 {                                                                       \
@@ -240,10 +241,11 @@ GENERATE_AMI_CT8(ami_ct_u8, AMI_CT_U8)
      */
 
 #define GENERATE_AMI_CT_16(funcname, convsample)                        \
-static ssize_t funcname(const u_char *userPtr, size_t userCount,        \
+static ssize_t funcname(const u_char __user *userPtr, size_t userCount, \
                         u_char frame[], ssize_t *frameUsed,             \
                         ssize_t frameLeft)                              \
 {                                                                       \
+        const u_short __user *ptr = (const u_short __user *)userPtr;    \
         ssize_t count, used;                                            \
         u_short data;                                                   \
                                                                         \
@@ -253,7 +255,7 @@ static ssize_t funcname(const u_char *userPtr, size_t userCount,	\
                 count = min_t(size_t, userCount, frameLeft)>>1 & ~1;    \
                 used = count*2;                                         \
                 while (count > 0) {                                     \
-                        if (get_user(data, ((u_short *)userPtr)++))     \
+                        if (get_user(data, ptr++))                      \
                                 return -EFAULT;                         \
                         data = convsample(data);                        \
                         *high++ = data>>8;                              \
@@ -268,12 +270,12 @@ static ssize_t funcname(const u_char *userPtr, size_t userCount,	\
                 count = min_t(size_t, userCount, frameLeft)>>2 & ~1;    \
                 used = count*4;                                         \
                 while (count > 0) {                                     \
-                        if (get_user(data, ((u_short *)userPtr)++))     \
+                        if (get_user(data, ptr++))                      \
                                 return -EFAULT;                         \
                         data = convsample(data);                        \
                         *lefth++ = data>>8;                             \
                         *leftl++ = (data>>2) & 0x3f;                    \
-                        if (get_user(data, ((u_short *)userPtr)++))     \
+                        if (get_user(data, ptr++))                      \
                                 return -EFAULT;                         \
                         data = convsample(data);                        \
                         *righth++ = data>>8;                            \
diff --git a/sound/oss/dmasound/dmasound_q40.c b/sound/oss/dmasound/dmasound_q40.c
index 1ddaa6284b08..e2081f32b0c4 100644
--- a/sound/oss/dmasound/dmasound_q40.c
+++ b/sound/oss/dmasound/dmasound_q40.c
@@ -58,7 +58,7 @@ static void Q40Interrupt(void);
 
 
 /* userCount, frameUsed, frameLeft == byte counts */
-static ssize_t q40_ct_law(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ct_law(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
@@ -79,7 +79,7 @@ static ssize_t q40_ct_law(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t q40_ct_s8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ct_s8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft)
 {
@@ -98,7 +98,7 @@ static ssize_t q40_ct_s8(const u_char *userPtr, size_t userCount,
         return used;
 }
 
-static ssize_t q40_ct_u8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ct_u8(const u_char __user *userPtr, size_t userCount,
                           u_char frame[], ssize_t *frameUsed,
                           ssize_t frameLeft)
 {
@@ -114,7 +114,7 @@ static ssize_t q40_ct_u8(const u_char *userPtr, size_t userCount,
 
 
 /* a bit too complicated to optimise right now ..*/
-static ssize_t q40_ctx_law(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctx_law(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -152,7 +152,7 @@ static ssize_t q40_ctx_law(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t q40_ctx_s8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctx_s8(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
@@ -189,7 +189,7 @@ static ssize_t q40_ctx_s8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t q40_ctx_u8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctx_u8(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
@@ -224,7 +224,7 @@ static ssize_t q40_ctx_u8(const u_char *userPtr, size_t userCount,
 }
 
 /* compressing versions */
-static ssize_t q40_ctc_law(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctc_law(const u_char __user *userPtr, size_t userCount,
                             u_char frame[], ssize_t *frameUsed,
                             ssize_t frameLeft)
 {
@@ -265,7 +265,7 @@ static ssize_t q40_ctc_law(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t q40_ctc_s8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctc_s8(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
@@ -304,7 +304,7 @@ static ssize_t q40_ctc_s8(const u_char *userPtr, size_t userCount,
 }
 
 
-static ssize_t q40_ctc_u8(const u_char *userPtr, size_t userCount,
+static ssize_t q40_ctc_u8(const u_char __user *userPtr, size_t userCount,
                            u_char frame[], ssize_t *frameUsed,
                            ssize_t frameLeft)
 {
diff --git a/sound/oss/dmasound/trans_16.c b/sound/oss/dmasound/trans_16.c
index 23562e947806..ca973ac2a30a 100644
--- a/sound/oss/dmasound/trans_16.c
+++ b/sound/oss/dmasound/trans_16.c
@@ -17,6 +17,7 @@
 #include <asm/uaccess.h>
 #include "dmasound.h"
 
+extern int expand_bal;  /* Balance factor for expanding (not volume!) */
 static short dmasound_alaw2dma16[] ;
 static short dmasound_ulaw2dma16[] ;