Merge git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] wire up pselect, ppoll
  [IA64] Add TIF_RESTORE_SIGMASK
  [IA64] unwind did not work for processes born with CLONE_STOPPED
  [IA64] Optional method to purge the TLB on SN systems
  [IA64] SPIN_LOCK_UNLOCKED macro cleanup in arch/ia64
  [IA64-SN2][KJ] mmtimer.c-kzalloc
  [IA64] fix stack alignment for ia32 signal handlers
  [IA64] - Altix: hotplug after intr redirect can crash system
  [IA64] save and restore cpus_allowed in cpu_idle_wait
  [IA64] Removal of percpu TR cleanup in kexec code
  [IA64] Fix some section mismatch errors

23 files changed, 276 insertions, 221 deletions

diff --git a/arch/ia64/ia32/ia32_entry.S b/arch/ia64/ia32/ia32_entry.S
index 687e5fdc9683..99b665e2b1d5 100644
--- a/arch/ia64/ia32/ia32_entry.S
+++ b/arch/ia64/ia32/ia32_entry.S
@@ -52,43 +52,6 @@ ENTRY(ia32_clone)
         br.ret.sptk.many rp
 END(ia32_clone)
 
-ENTRY(sys32_rt_sigsuspend)
-        .prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
-        alloc loc1=ar.pfs,8,2,3,0               // preserve all eight input regs
-        mov loc0=rp
-        mov out0=in0                            // mask
-        mov out1=in1                            // sigsetsize
-        mov out2=sp                             // out2 = &sigscratch
-        .fframe 16
-        adds sp=-16,sp                          // allocate dummy "sigscratch"
-        ;;
-        .body
-        br.call.sptk.many rp=ia32_rt_sigsuspend
-1:      .restore sp
-        adds sp=16,sp
-        mov rp=loc0
-        mov ar.pfs=loc1
-        br.ret.sptk.many rp
-END(sys32_rt_sigsuspend)
-
-ENTRY(sys32_sigsuspend)
-        .prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
-        alloc loc1=ar.pfs,8,2,3,0               // preserve all eight input regs
-        mov loc0=rp
-        mov out0=in2                            // mask (first two args are ignored)
-        ;;
-        mov out1=sp                             // out1 = &sigscratch
-        .fframe 16
-        adds sp=-16,sp                          // allocate dummy "sigscratch"
-        .body
-        br.call.sptk.many rp=ia32_sigsuspend
-1:      .restore sp
-        adds sp=16,sp
-        mov rp=loc0
-        mov ar.pfs=loc1
-        br.ret.sptk.many rp
-END(sys32_sigsuspend)
-
 GLOBAL_ENTRY(ia32_ret_from_clone)
         PT_REGS_UNWIND_INFO(0)
 {       /*
@@ -389,7 +352,7 @@ ia32_syscall_table:
         data8 sys_rt_sigpending
         data8 compat_sys_rt_sigtimedwait
         data8 sys32_rt_sigqueueinfo
-        data8 sys32_rt_sigsuspend
+        data8 compat_sys_rt_sigsuspend
         data8 sys32_pread         /* 180 */
         data8 sys32_pwrite
         data8 sys_chown /* 16-bit version */
diff --git a/arch/ia64/ia32/ia32_signal.c b/arch/ia64/ia32/ia32_signal.c
index 10510e585204..85e82f32e480 100644
--- a/arch/ia64/ia32/ia32_signal.c
+++ b/arch/ia64/ia32/ia32_signal.c
@@ -451,59 +451,20 @@ sigact_set_handler (struct k_sigaction *sa, unsigned int handler, unsigned int r
                 sa->sa.sa_handler = (__sighandler_t) (((unsigned long) restorer << 32) | handler);
 }
 
-long
-__ia32_rt_sigsuspend (compat_sigset_t *sset, unsigned int sigsetsize, struct sigscratch *scr)
+asmlinkage long
+sys32_sigsuspend (int history0, int history1, old_sigset_t mask)
 {
-        extern long ia64_do_signal (sigset_t *oldset, struct sigscratch *scr, long in_syscall);
-        sigset_t oldset, set;
-
-        scr->scratch_unat = 0;  /* avoid leaking kernel bits to user level */
-        memset(&set, 0, sizeof(set));
-
-        memcpy(&set.sig, &sset->sig, sigsetsize);
-
-        sigdelsetmask(&set, ~_BLOCKABLE);
-
+        mask &= _BLOCKABLE;
         spin_lock_irq(&current->sighand->siglock);
-        {
-                oldset = current->blocked;
-                current->blocked = set;
-                recalc_sigpending();
-        }
+        current->saved_sigmask = current->blocked;
+        siginitset(&current->blocked, mask);
+        recalc_sigpending();
         spin_unlock_irq(&current->sighand->siglock);
 
-        /*
-         * The return below usually returns to the signal handler.  We need to pre-set the
-         * correct error code here to ensure that the right values get saved in sigcontext
-         * by ia64_do_signal.
-         */
-        scr->pt.r8 = -EINTR;
-        while (1) {
-                current->state = TASK_INTERRUPTIBLE;
-                schedule();
-                if (ia64_do_signal(&oldset, scr, 1))
-                        return -EINTR;
-        }
-}
-
-asmlinkage long
-ia32_rt_sigsuspend (compat_sigset_t __user *uset, unsigned int sigsetsize, struct sigscratch *scr)
-{
-        compat_sigset_t set;
-
-        if (sigsetsize > sizeof(compat_sigset_t))
-                return -EINVAL;
-
-        if (copy_from_user(&set.sig, &uset->sig, sigsetsize))
-                return -EFAULT;
-
-        return __ia32_rt_sigsuspend(&set, sigsetsize, scr);
-}
-
-asmlinkage long
-ia32_sigsuspend (unsigned int mask, struct sigscratch *scr)
-{
-        return __ia32_rt_sigsuspend((compat_sigset_t *) &mask, sizeof(mask), scr);
+        current->state = TASK_INTERRUPTIBLE;
+        schedule();
+        set_thread_flag(TIF_RESTORE_SIGMASK);
+        return -ERESTARTNOHAND;
 }
 
 asmlinkage long
@@ -810,7 +771,11 @@ get_sigframe (struct k_sigaction *ka, struct pt_regs * regs, size_t frame_size)
         }
         /* Legacy stack switching not supported */
 
-        return (void __user *)((esp - frame_size) & -8ul);
+        esp -= frame_size;
+        /* Align the stack pointer according to the i386 ABI,
+         * i.e. so that on function entry ((sp + 4) & 15) == 0. */
+        esp = ((esp + 4) & -16ul) - 4;
+        return (void __user *) esp;
 }
 
 static int
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index 55fd2d5471e1..b50bf208678e 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -1199,32 +1199,6 @@ ENTRY(notify_resume_user)
         br.ret.sptk.many rp
 END(notify_resume_user)
 
-GLOBAL_ENTRY(sys_rt_sigsuspend)
-        .prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
-        alloc loc1=ar.pfs,8,2,3,0 // preserve all eight input regs in case of syscall restart!
-        mov r9=ar.unat
-        mov loc0=rp                             // save return address
-        mov out0=in0                            // mask
-        mov out1=in1                            // sigsetsize
-        adds out2=8,sp                          // out2=&sigscratch->ar_pfs
-        ;;
-        .fframe 16
-        .spillsp ar.unat, 16
-        st8 [sp]=r9,-16                         // allocate space for ar.unat and save it
-        st8 [out2]=loc1,-8                      // save ar.pfs, out2=&sigscratch
-        .body
-        br.call.sptk.many rp=ia64_rt_sigsuspend
-.ret17: .restore sp
-        adds sp=16,sp                           // pop scratch stack space
-        ;;
-        ld8 r9=[sp]                             // load new unat from sw->caller_unat
-        mov rp=loc0
-        ;;
-        mov ar.unat=r9
-        mov ar.pfs=loc1
-        br.ret.sptk.many rp
-END(sys_rt_sigsuspend)
-
 ENTRY(sys_rt_sigreturn)
         PT_REGS_UNWIND_INFO(0)
         /*
@@ -1598,8 +1572,8 @@ sys_call_table:
         data8 sys_readlinkat
         data8 sys_fchmodat
         data8 sys_faccessat
-        data8 sys_ni_syscall                    // reserved for pselect
-        data8 sys_ni_syscall                    // 1295 reserved for ppoll
+        data8 sys_pselect6
+        data8 sys_ppoll
         data8 sys_unshare
         data8 sys_splice
         data8 sys_set_robust_list
diff --git a/arch/ia64/kernel/iosapic.c b/arch/ia64/kernel/iosapic.c
index 93d9ab14ba24..37f46527d233 100644
--- a/arch/ia64/kernel/iosapic.c
+++ b/arch/ia64/kernel/iosapic.c
@@ -1012,7 +1012,7 @@ iosapic_register_platform_intr (u32 int_type, unsigned int gsi,
 /*
  * ACPI calls this when it finds an entry for a legacy ISA IRQ override.
  */
-void __init
+void __devinit
 iosapic_override_isa_irq (unsigned int isa_irq, unsigned int gsi,
                           unsigned long polarity,
                           unsigned long trigger)
diff --git a/arch/ia64/kernel/irq_ia64.c b/arch/ia64/kernel/irq_ia64.c
index 1c5044a80958..dce5341303de 100644
--- a/arch/ia64/kernel/irq_ia64.c
+++ b/arch/ia64/kernel/irq_ia64.c
@@ -38,6 +38,7 @@
 #include <asm/machvec.h>
 #include <asm/pgtable.h>
 #include <asm/system.h>
+#include <asm/tlbflush.h>
 
 #ifdef CONFIG_PERFMON
 # include <asm/perfmon.h>
@@ -126,8 +127,10 @@ void destroy_irq(unsigned int irq)
 
 #ifdef CONFIG_SMP
 #       define IS_RESCHEDULE(vec)       (vec == IA64_IPI_RESCHEDULE)
+#       define IS_LOCAL_TLB_FLUSH(vec)  (vec == IA64_IPI_LOCAL_TLB_FLUSH)
 #else
 #       define IS_RESCHEDULE(vec)       (0)
+#       define IS_LOCAL_TLB_FLUSH(vec)  (0)
 #endif
 /*
  * That's where the IVT branches when we get an external
@@ -179,8 +182,11 @@ ia64_handle_irq (ia64_vector vector, struct pt_regs *regs)
         saved_tpr = ia64_getreg(_IA64_REG_CR_TPR);
         ia64_srlz_d();
         while (vector != IA64_SPURIOUS_INT_VECTOR) {
-                if (unlikely(IS_RESCHEDULE(vector)))
-                         kstat_this_cpu.irqs[vector]++;
+                if (unlikely(IS_LOCAL_TLB_FLUSH(vector))) {
+                        smp_local_flush_tlb();
+                        kstat_this_cpu.irqs[vector]++;
+                } else if (unlikely(IS_RESCHEDULE(vector)))
+                        kstat_this_cpu.irqs[vector]++;
                 else {
                         ia64_setreg(_IA64_REG_CR_TPR, vector);
                         ia64_srlz_d();
@@ -226,8 +232,11 @@ void ia64_process_pending_intr(void)
           * Perform normal interrupt style processing
           */
         while (vector != IA64_SPURIOUS_INT_VECTOR) {
-                if (unlikely(IS_RESCHEDULE(vector)))
-                         kstat_this_cpu.irqs[vector]++;
+                if (unlikely(IS_LOCAL_TLB_FLUSH(vector))) {
+                        smp_local_flush_tlb();
+                        kstat_this_cpu.irqs[vector]++;
+                } else if (unlikely(IS_RESCHEDULE(vector)))
+                        kstat_this_cpu.irqs[vector]++;
                 else {
                         struct pt_regs *old_regs = set_irq_regs(NULL);
 
@@ -259,12 +268,12 @@ void ia64_process_pending_intr(void)
 
 
 #ifdef CONFIG_SMP
-extern irqreturn_t handle_IPI (int irq, void *dev_id);
 
 static irqreturn_t dummy_handler (int irq, void *dev_id)
 {
         BUG();
 }
+extern irqreturn_t handle_IPI (int irq, void *dev_id);
 
 static struct irqaction ipi_irqaction = {
         .handler =      handle_IPI,
@@ -277,6 +286,13 @@ static struct irqaction resched_irqaction = {
         .flags =        IRQF_DISABLED,
         .name =         "resched"
 };
+
+static struct irqaction tlb_irqaction = {
+        .handler =      dummy_handler,
+        .flags =        SA_INTERRUPT,
+        .name =         "tlb_flush"
+};
+
 #endif
 
 void
@@ -302,6 +318,7 @@ init_IRQ (void)
 #ifdef CONFIG_SMP
         register_percpu_irq(IA64_IPI_VECTOR, &ipi_irqaction);
         register_percpu_irq(IA64_IPI_RESCHEDULE, &resched_irqaction);
+        register_percpu_irq(IA64_IPI_LOCAL_TLB_FLUSH, &tlb_irqaction);
 #endif
 #ifdef CONFIG_PERFMON
         pfm_init_percpu();
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index 8bb571a8a738..d1c3ed9943e5 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -155,7 +155,7 @@ show_regs (struct pt_regs *regs)
 }
 
 void
-do_notify_resume_user (sigset_t *oldset, struct sigscratch *scr, long in_syscall)
+do_notify_resume_user (sigset_t *unused, struct sigscratch *scr, long in_syscall)
 {
         if (fsys_mode(current, &scr->pt)) {
                 /* defer signal-handling etc. until we return to privilege-level 0.  */
@@ -170,8 +170,8 @@ do_notify_resume_user (sigset_t *oldset, struct sigscratch *scr, long in_syscall
 #endif
 
         /* deal with pending signal delivery */
-        if (test_thread_flag(TIF_SIGPENDING))
-                ia64_do_signal(oldset, scr, in_syscall);
+        if (test_thread_flag(TIF_SIGPENDING)||test_thread_flag(TIF_RESTORE_SIGMASK))
+                ia64_do_signal(scr, in_syscall);
 }
 
 static int pal_halt        = 1;
@@ -236,6 +236,7 @@ void cpu_idle_wait(void)
 {
         unsigned int cpu, this_cpu = get_cpu();
         cpumask_t map;
+        cpumask_t tmp = current->cpus_allowed;
 
         set_cpus_allowed(current, cpumask_of_cpu(this_cpu));
         put_cpu();
@@ -257,6 +258,7 @@ void cpu_idle_wait(void)
                 }
                 cpus_and(map, map, cpu_online_map);
         } while (!cpus_empty(map));
+        set_cpus_allowed(current, tmp);
 }
 EXPORT_SYMBOL_GPL(cpu_idle_wait);
 
diff --git a/arch/ia64/kernel/relocate_kernel.S b/arch/ia64/kernel/relocate_kernel.S
index ae473e3f2a0d..903babd22d62 100644
--- a/arch/ia64/kernel/relocate_kernel.S
+++ b/arch/ia64/kernel/relocate_kernel.S
@@ -94,7 +94,7 @@ GLOBAL_ENTRY(relocate_new_kernel)
 4:
         srlz.i
         ;;
-        //purge TR entry for kernel text and data
+        // purge TR entry for kernel text and data
         movl r16=KERNEL_START
         mov r18=KERNEL_TR_PAGE_SHIFT<<2
         ;;
@@ -104,15 +104,6 @@ GLOBAL_ENTRY(relocate_new_kernel)
         srlz.i
         ;;
 
-        // purge TR entry for percpu data
-        movl r16=PERCPU_ADDR
-        mov r18=PERCPU_PAGE_SHIFT<<2
-        ;;
-        ptr.d r16,r18
-        ;;
-        srlz.d
-        ;;
-
         // purge TR entry for pal code
         mov r16=in3
         mov r18=IA64_GRANULE_SHIFT<<2
diff --git a/arch/ia64/kernel/setup.c b/arch/ia64/kernel/setup.c
index 6e19da122ae3..9df1efe7487d 100644
--- a/arch/ia64/kernel/setup.c
+++ b/arch/ia64/kernel/setup.c
@@ -786,7 +786,7 @@ identify_cpu (struct cpuinfo_ia64 *c)
         c->unimpl_pa_mask = ~((1L<<63) | ((1L << phys_addr_size) - 1));
 }
 
-void
+void __init
 setup_per_cpu_areas (void)
 {
         /* start_kernel() requires this... */
diff --git a/arch/ia64/kernel/sigframe.h b/arch/ia64/kernel/sigframe.h
index 37b986cb86e0..9fd9a1933b3d 100644
--- a/arch/ia64/kernel/sigframe.h
+++ b/arch/ia64/kernel/sigframe.h
@@ -22,4 +22,4 @@ struct sigframe {
         struct sigcontext sc;
 };
 
-extern long ia64_do_signal (sigset_t *, struct sigscratch *, long);
+extern void ia64_do_signal (struct sigscratch *, long);
diff --git a/arch/ia64/kernel/signal.c b/arch/ia64/kernel/signal.c
index 0dcd56da6001..aeec8184e862 100644
--- a/arch/ia64/kernel/signal.c
+++ b/arch/ia64/kernel/signal.c
@@ -40,47 +40,6 @@
 # define GET_SIGSET(k,u)        __get_user((k)->sig[0], &(u)->sig[0])
 #endif
 
-long
-ia64_rt_sigsuspend (sigset_t __user *uset, size_t sigsetsize, struct sigscratch *scr)
-{
-        sigset_t oldset, set;
-
-        /* XXX: Don't preclude handling different sized sigset_t's.  */
-        if (sigsetsize != sizeof(sigset_t))
-                return -EINVAL;
-
-        if (!access_ok(VERIFY_READ, uset, sigsetsize))
-                return -EFAULT;
-
-        if (GET_SIGSET(&set, uset))
-                return -EFAULT;
-
-        sigdelsetmask(&set, ~_BLOCKABLE);
-
-        spin_lock_irq(&current->sighand->siglock);
-        {
-                oldset = current->blocked;
-                current->blocked = set;
-                recalc_sigpending();
-        }
-        spin_unlock_irq(&current->sighand->siglock);
-
-        /*
-         * The return below usually returns to the signal handler.  We need to
-         * pre-set the correct error code here to ensure that the right values
-         * get saved in sigcontext by ia64_do_signal.
-         */
-        scr->pt.r8 = EINTR;
-        scr->pt.r10 = -1;
-
-        while (1) {
-                current->state = TASK_INTERRUPTIBLE;
-                schedule();
-                if (ia64_do_signal(&oldset, scr, 1))
-                        return -EINTR;
-        }
-}
-
 asmlinkage long
 sys_sigaltstack (const stack_t __user *uss, stack_t __user *uoss, long arg2,
                  long arg3, long arg4, long arg5, long arg6, long arg7,
@@ -477,10 +436,11 @@ handle_signal (unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigse
  * Note that `init' is a special process: it doesn't get signals it doesn't want to
  * handle.  Thus you cannot kill init even with a SIGKILL even by mistake.
  */
-long
-ia64_do_signal (sigset_t *oldset, struct sigscratch *scr, long in_syscall)
+void
+ia64_do_signal (struct sigscratch *scr, long in_syscall)
 {
         struct k_sigaction ka;
+        sigset_t *oldset;
         siginfo_t info;
         long restart = in_syscall;
         long errno = scr->pt.r8;
@@ -492,9 +452,11 @@ ia64_do_signal (sigset_t *oldset, struct sigscratch *scr, long in_syscall)
          * doing anything if so.
          */
         if (!user_mode(&scr->pt))
-                return 0;
+                return;
 
-        if (!oldset)
+        if (test_thread_flag(TIF_RESTORE_SIGMASK))
+                oldset = &current->saved_sigmask;
+        else
                 oldset = &current->blocked;
 
         /*
@@ -557,8 +519,15 @@ ia64_do_signal (sigset_t *oldset, struct sigscratch *scr, long in_syscall)
                  * Whee!  Actually deliver the signal.  If the delivery failed, we need to
                  * continue to iterate in this loop so we can deliver the SIGSEGV...
                  */
-                if (handle_signal(signr, &ka, &info, oldset, scr))
-                        return 1;
+                if (handle_signal(signr, &ka, &info, oldset, scr)) {
+                        /* a signal was successfully delivered; the saved
+                         * sigmask will have been stored in the signal frame,
+                         * and will be restored by sigreturn, so we can simply
+                         * clear the TIF_RESTORE_SIGMASK flag */
+                        if (test_thread_flag(TIF_RESTORE_SIGMASK))
+                                clear_thread_flag(TIF_RESTORE_SIGMASK);
+                        return;
+                }
         }
 
         /* Did we come from a system call? */
@@ -584,5 +553,11 @@ ia64_do_signal (sigset_t *oldset, struct sigscratch *scr, long in_syscall)
                         }
                 }
         }
-        return 0;
+
+        /* if there's no signal to deliver, we just put the saved sigmask
+         * back */
+        if (test_thread_flag(TIF_RESTORE_SIGMASK)) {
+                clear_thread_flag(TIF_RESTORE_SIGMASK);
+                sigprocmask(SIG_SETMASK, &current->saved_sigmask, NULL);
+        }
 }
diff --git a/arch/ia64/kernel/smp.c b/arch/ia64/kernel/smp.c
index 55ddd809b02d..221de3804560 100644
--- a/arch/ia64/kernel/smp.c
+++ b/arch/ia64/kernel/smp.c
@@ -50,6 +50,18 @@
 #include <asm/mca.h>
 
 /*
+ * Note: alignment of 4 entries/cacheline was empirically determined
+ * to be a good tradeoff between hot cachelines & spreading the array
+ * across too many cacheline.
+ */
+static struct local_tlb_flush_counts {
+        unsigned int count;
+} __attribute__((__aligned__(32))) local_tlb_flush_counts[NR_CPUS];
+
+static DEFINE_PER_CPU(unsigned int, shadow_flush_counts[NR_CPUS]) ____cacheline_aligned;
+
+
+/*
  * Structure and data for smp_call_function(). This is designed to minimise static memory
  * requirements. It also looks cleaner.
  */
@@ -248,6 +260,62 @@ smp_send_reschedule (int cpu)
         platform_send_ipi(cpu, IA64_IPI_RESCHEDULE, IA64_IPI_DM_INT, 0);
 }
 
+/*
+ * Called with preeemption disabled.
+ */
+static void
+smp_send_local_flush_tlb (int cpu)
+{
+        platform_send_ipi(cpu, IA64_IPI_LOCAL_TLB_FLUSH, IA64_IPI_DM_INT, 0);
+}
+
+void
+smp_local_flush_tlb(void)
+{
+        /*
+         * Use atomic ops. Otherwise, the load/increment/store sequence from
+         * a "++" operation can have the line stolen between the load & store.
+         * The overhead of the atomic op in negligible in this case & offers
+         * significant benefit for the brief periods where lots of cpus
+         * are simultaneously flushing TLBs.
+         */
+        ia64_fetchadd(1, &local_tlb_flush_counts[smp_processor_id()].count, acq);
+        local_flush_tlb_all();
+}
+
+#define FLUSH_DELAY     5 /* Usec backoff to eliminate excessive cacheline bouncing */
+
+void
+smp_flush_tlb_cpumask(cpumask_t xcpumask)
+{
+        unsigned int *counts = __ia64_per_cpu_var(shadow_flush_counts);
+        cpumask_t cpumask = xcpumask;
+        int mycpu, cpu, flush_mycpu = 0;
+
+        preempt_disable();
+        mycpu = smp_processor_id();
+
+        for_each_cpu_mask(cpu, cpumask)
+                counts[cpu] = local_tlb_flush_counts[cpu].count;
+
+        mb();
+        for_each_cpu_mask(cpu, cpumask) {
+                if (cpu == mycpu)
+                        flush_mycpu = 1;
+                else
+                        smp_send_local_flush_tlb(cpu);
+        }
+
+        if (flush_mycpu)
+                smp_local_flush_tlb();
+
+        for_each_cpu_mask(cpu, cpumask)
+                while(counts[cpu] == local_tlb_flush_counts[cpu].count)
+                        udelay(FLUSH_DELAY);
+
+        preempt_enable();
+}
+
 void
 smp_flush_tlb_all (void)
 {
diff --git a/arch/ia64/kernel/traps.c b/arch/ia64/kernel/traps.c
index 5bfb8be02b70..b8e0d70bf989 100644
--- a/arch/ia64/kernel/traps.c
+++ b/arch/ia64/kernel/traps.c
@@ -43,9 +43,9 @@ die (const char *str, struct pt_regs *regs, long err)
                 u32 lock_owner;
                 int lock_owner_depth;
         } die = {
-                .lock =                 SPIN_LOCK_UNLOCKED,
-                .lock_owner =           -1,
-                .lock_owner_depth =     0
+                .lock = __SPIN_LOCK_UNLOCKED(die.lock),
+                .lock_owner = -1,
+                .lock_owner_depth = 0
         };
         static int die_counter;
         int cpu = get_cpu();
diff --git a/arch/ia64/kernel/unwind.c b/arch/ia64/kernel/unwind.c
index 93d5a3b41f69..fe1426266b9b 100644
--- a/arch/ia64/kernel/unwind.c
+++ b/arch/ia64/kernel/unwind.c
@@ -60,6 +60,7 @@
 #  define UNW_DEBUG_ON(n)       unw_debug_level >= n
    /* Do not code a printk level, not all debug lines end in newline */
 #  define UNW_DPRINT(n, ...)  if (UNW_DEBUG_ON(n)) printk(__VA_ARGS__)
+#  undef inline
 #  define inline
 #else /* !UNW_DEBUG */
 #  define UNW_DEBUG_ON(n)  0
@@ -145,7 +146,7 @@ static struct {
 # endif
 } unw = {
         .tables = &unw.kernel_table,
-        .lock = SPIN_LOCK_UNLOCKED,
+        .lock = __SPIN_LOCK_UNLOCKED(unw.lock),
         .save_order = {
                 UNW_REG_RP, UNW_REG_PFS, UNW_REG_PSP, UNW_REG_PR,
                 UNW_REG_UNAT, UNW_REG_LC, UNW_REG_FPSR, UNW_REG_PRI_UNAT_GR
@@ -1943,9 +1944,9 @@ EXPORT_SYMBOL(unw_unwind);
 int
 unw_unwind_to_user (struct unw_frame_info *info)
 {
-        unsigned long ip, sp, pr = 0;
+        unsigned long ip, sp, pr = info->pr;
 
-        while (unw_unwind(info) >= 0) {
+        do {
                 unw_get_sp(info, &sp);
                 if ((long)((unsigned long)info->task + IA64_STK_OFFSET - sp)
                     < IA64_PT_REGS_SIZE) {
@@ -1963,7 +1964,7 @@ unw_unwind_to_user (struct unw_frame_info *info)
                                 __FUNCTION__, ip);
                         return -1;
                 }
-        }
+        } while (unw_unwind(info) >= 0);
         unw_get_ip(info, &ip);
         UNW_DPRINT(0, "unwind.%s: failed to unwind to user-level (ip=0x%lx)\n",
                    __FUNCTION__, ip);
diff --git a/arch/ia64/mm/tlb.c b/arch/ia64/mm/tlb.c
index ffad7624436c..fa4e6d4810f3 100644
--- a/arch/ia64/mm/tlb.c
+++ b/arch/ia64/mm/tlb.c
@@ -32,9 +32,9 @@ static struct {
 } purge;
 
 struct ia64_ctx ia64_ctx = {
-        .lock =         SPIN_LOCK_UNLOCKED,
-        .next =         1,
-        .max_ctx =      ~0U
+        .lock = __SPIN_LOCK_UNLOCKED(ia64_ctx.lock),
+        .next = 1,
+        .max_ctx = ~0U
 };
 
 DEFINE_PER_CPU(u8, ia64_need_tlb_flush);
diff --git a/arch/ia64/sn/kernel/irq.c b/arch/ia64/sn/kernel/irq.c
index 8d2a1bfbfe7c..7f6d2360a262 100644
--- a/arch/ia64/sn/kernel/irq.c
+++ b/arch/ia64/sn/kernel/irq.c
@@ -59,6 +59,22 @@ void sn_intr_free(nasid_t local_nasid, int local_widget,
                         (u64) sn_irq_info->irq_cookie, 0, 0);
 }
 
+u64 sn_intr_redirect(nasid_t local_nasid, int local_widget,
+                      struct sn_irq_info *sn_irq_info,
+                      nasid_t req_nasid, int req_slice)
+{
+        struct ia64_sal_retval ret_stuff;
+        ret_stuff.status = 0;
+        ret_stuff.v0 = 0;
+
+        SAL_CALL_NOLOCK(ret_stuff, (u64) SN_SAL_IOIF_INTERRUPT,
+                        (u64) SAL_INTR_REDIRECT, (u64) local_nasid,
+                        (u64) local_widget, __pa(sn_irq_info),
+                        (u64) req_nasid, (u64) req_slice, 0);
+
+        return ret_stuff.status;
+}
+
 static unsigned int sn_startup_irq(unsigned int irq)
 {
         return 0;
@@ -127,15 +143,8 @@ struct sn_irq_info *sn_retarget_vector(struct sn_irq_info *sn_irq_info,
         struct sn_irq_info *new_irq_info;
         struct sn_pcibus_provider *pci_provider;
 
-        new_irq_info = kmalloc(sizeof(struct sn_irq_info), GFP_ATOMIC);
-        if (new_irq_info == NULL)
-                return NULL;
-
-        memcpy(new_irq_info, sn_irq_info, sizeof(struct sn_irq_info));
-
-        bridge = (u64) new_irq_info->irq_bridge;
+        bridge = (u64) sn_irq_info->irq_bridge;
         if (!bridge) {
-                kfree(new_irq_info);
                 return NULL; /* irq is not a device interrupt */
         }
 
@@ -145,8 +154,25 @@ struct sn_irq_info *sn_retarget_vector(struct sn_irq_info *sn_irq_info,
                 local_widget = TIO_SWIN_WIDGETNUM(bridge);
         else
                 local_widget = SWIN_WIDGETNUM(bridge);
-
         vector = sn_irq_info->irq_irq;
+
+        /* Make use of SAL_INTR_REDIRECT if PROM supports it */
+        status = sn_intr_redirect(local_nasid, local_widget, sn_irq_info, nasid, slice);
+        if (!status) {
+                new_irq_info = sn_irq_info;
+                goto finish_up;
+        }
+
+        /*
+         * PROM does not support SAL_INTR_REDIRECT, or it failed.
+         * Revert to old method.
+         */
+        new_irq_info = kmalloc(sizeof(struct sn_irq_info), GFP_ATOMIC);
+        if (new_irq_info == NULL)
+                return NULL;
+
+        memcpy(new_irq_info, sn_irq_info, sizeof(struct sn_irq_info));
+
         /* Free the old PROM new_irq_info structure */
         sn_intr_free(local_nasid, local_widget, new_irq_info);
         unregister_intr_pda(new_irq_info);
@@ -162,11 +188,18 @@ struct sn_irq_info *sn_retarget_vector(struct sn_irq_info *sn_irq_info,
                 return NULL;
         }
 
+        register_intr_pda(new_irq_info);
+        spin_lock(&sn_irq_info_lock);
+        list_replace_rcu(&sn_irq_info->list, &new_irq_info->list);
+        spin_unlock(&sn_irq_info_lock);
+        call_rcu(&sn_irq_info->rcu, sn_irq_info_free);
+
+
+finish_up:
         /* Update kernels new_irq_info with new target info */
         cpuid = nasid_slice_to_cpuid(new_irq_info->irq_nasid,
                                      new_irq_info->irq_slice);
         new_irq_info->irq_cpuid = cpuid;
-        register_intr_pda(new_irq_info);
 
         pci_provider = sn_pci_provider[new_irq_info->irq_bridge_type];
 
@@ -178,11 +211,6 @@ struct sn_irq_info *sn_retarget_vector(struct sn_irq_info *sn_irq_info,
             pci_provider && pci_provider->target_interrupt)
                 (pci_provider->target_interrupt)(new_irq_info);
 
-        spin_lock(&sn_irq_info_lock);
-        list_replace_rcu(&sn_irq_info->list, &new_irq_info->list);
-        spin_unlock(&sn_irq_info_lock);
-        call_rcu(&sn_irq_info->rcu, sn_irq_info_free);
-
 #ifdef CONFIG_SMP
         cpuphys = cpu_physical_id(cpuid);
         set_irq_affinity_info((vector & 0xff), cpuphys, 0);
diff --git a/arch/ia64/sn/kernel/sn2/sn2_smp.c b/arch/ia64/sn/kernel/sn2/sn2_smp.c
index 601747b1e22a..5d318b579fb1 100644
--- a/arch/ia64/sn/kernel/sn2/sn2_smp.c
+++ b/arch/ia64/sn/kernel/sn2/sn2_smp.c
@@ -46,6 +46,9 @@ DECLARE_PER_CPU(struct ptc_stats, ptcstats);
 
 static  __cacheline_aligned DEFINE_SPINLOCK(sn2_global_ptc_lock);
 
+/* 0 = old algorithm (no IPI flushes), 1 = ipi deadlock flush, 2 = ipi instead of SHUB ptc, >2 = always ipi */
+static int sn2_flush_opt = 0;
+
 extern unsigned long
 sn2_ptc_deadlock_recovery_core(volatile unsigned long *, unsigned long,
                                volatile unsigned long *, unsigned long,
@@ -76,6 +79,8 @@ struct ptc_stats {
         unsigned long shub_itc_clocks;
         unsigned long shub_itc_clocks_max;
         unsigned long shub_ptc_flushes_not_my_mm;
+        unsigned long shub_ipi_flushes;
+        unsigned long shub_ipi_flushes_itc_clocks;
 };
 
 #define sn2_ptctest     0
@@ -121,6 +126,18 @@ void sn_tlb_migrate_finish(struct mm_struct *mm)
                 flush_tlb_mm(mm);
 }
 
+static void
+sn2_ipi_flush_all_tlb(struct mm_struct *mm)
+{
+        unsigned long itc;
+
+        itc = ia64_get_itc();
+        smp_flush_tlb_cpumask(mm->cpu_vm_mask);
+        itc = ia64_get_itc() - itc;
+        __get_cpu_var(ptcstats).shub_ipi_flushes_itc_clocks += itc;
+        __get_cpu_var(ptcstats).shub_ipi_flushes++;
+}
+
 /**
  * sn2_global_tlb_purge - globally purge translation cache of virtual address range
  * @mm: mm_struct containing virtual address range
@@ -154,7 +171,12 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start,
         unsigned long itc, itc2, flags, data0 = 0, data1 = 0, rr_value, old_rr = 0;
         short nasids[MAX_NUMNODES], nix;
         nodemask_t nodes_flushed;
-        int active, max_active, deadlock;
+        int active, max_active, deadlock, flush_opt = sn2_flush_opt;
+
+        if (flush_opt > 2) {
+                sn2_ipi_flush_all_tlb(mm);
+                return;
+        }
 
         nodes_clear(nodes_flushed);
         i = 0;
@@ -189,6 +211,12 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start,
                 return;
         }
 
+        if (flush_opt == 2) {
+                sn2_ipi_flush_all_tlb(mm);
+                preempt_enable();
+                return;
+        }
+
         itc = ia64_get_itc();
         nix = 0;
         for_each_node_mask(cnode, nodes_flushed)
@@ -256,6 +284,8 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start,
                         }
                         if (active >= max_active || i == (nix - 1)) {
                                 if ((deadlock = wait_piowc())) {
+                                        if (flush_opt == 1)
+                                                goto done;
                                         sn2_ptc_deadlock_recovery(nasids, ibegin, i, mynasid, ptc0, data0, ptc1, data1);
                                         if (reset_max_active_on_deadlock())
                                                 max_active = 1;
@@ -267,6 +297,7 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start,
                 start += (1UL << nbits);
         } while (start < end);
 
+done:
         itc2 = ia64_get_itc() - itc2;
         __get_cpu_var(ptcstats).shub_itc_clocks += itc2;
         if (itc2 > __get_cpu_var(ptcstats).shub_itc_clocks_max)
@@ -279,6 +310,11 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start,
 
         spin_unlock_irqrestore(PTC_LOCK(shub1), flags);
 
+        if (flush_opt == 1 && deadlock) {
+                __get_cpu_var(ptcstats).deadlocks++;
+                sn2_ipi_flush_all_tlb(mm);
+        }
+
         preempt_enable();
 }
 
@@ -425,24 +461,42 @@ static int sn2_ptc_seq_show(struct seq_file *file, void *data)
 
         if (!cpu) {
                 seq_printf(file,
-                           "# cpu ptc_l newrid ptc_flushes nodes_flushed deadlocks lock_nsec shub_nsec shub_nsec_max not_my_mm deadlock2\n");
-                seq_printf(file, "# ptctest %d\n", sn2_ptctest);
+                           "# cpu ptc_l newrid ptc_flushes nodes_flushed deadlocks lock_nsec shub_nsec shub_nsec_max not_my_mm deadlock2 ipi_fluches ipi_nsec\n");
+                seq_printf(file, "# ptctest %d, flushopt %d\n", sn2_ptctest, sn2_flush_opt);
         }
 
         if (cpu < NR_CPUS && cpu_online(cpu)) {
                 stat = &per_cpu(ptcstats, cpu);
-                seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n", cpu, stat->ptc_l,
+                seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n", cpu, stat->ptc_l,
                                 stat->change_rid, stat->shub_ptc_flushes, stat->nodes_flushed,
                                 stat->deadlocks,
                                 1000 * stat->lock_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
                                 1000 * stat->shub_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec,
                                 1000 * stat->shub_itc_clocks_max / per_cpu(cpu_info, cpu).cyc_per_usec,
                                 stat->shub_ptc_flushes_not_my_mm,
-                                stat->deadlocks2);
+                                stat->deadlocks2,
+                                stat->shub_ipi_flushes,
+                                1000 * stat->shub_ipi_flushes_itc_clocks / per_cpu(cpu_info, cpu).cyc_per_usec);
         }
         return 0;
 }
 
+static ssize_t sn2_ptc_proc_write(struct file *file, const char __user *user, size_t count, loff_t *data)
+{
+        int cpu;
+        char optstr[64];
+
+        if (copy_from_user(optstr, user, count))
+                return -EFAULT;
+        optstr[count - 1] = '\0';
+        sn2_flush_opt = simple_strtoul(optstr, NULL, 0);
+
+        for_each_online_cpu(cpu)
+                memset(&per_cpu(ptcstats, cpu), 0, sizeof(struct ptc_stats));
+
+        return count;
+}
+
 static struct seq_operations sn2_ptc_seq_ops = {
         .start = sn2_ptc_seq_start,
         .next = sn2_ptc_seq_next,
@@ -458,6 +512,7 @@ static int sn2_ptc_proc_open(struct inode *inode, struct file *file)
 static const struct file_operations proc_sn2_ptc_operations = {
         .open = sn2_ptc_proc_open,
         .read = seq_read,
+        .write = sn2_ptc_proc_write,
         .llseek = seq_lseek,
         .release = seq_release,
 };
diff --git a/drivers/char/mmtimer.c b/drivers/char/mmtimer.c
index c09160383a53..6e55cfb9c65a 100644
--- a/drivers/char/mmtimer.c
+++ b/drivers/char/mmtimer.c
@@ -705,15 +705,13 @@ static int __init mmtimer_init(void)
         maxn++;
 
         /* Allocate list of node ptrs to mmtimer_t's */
-        timers = kmalloc(sizeof(mmtimer_t *)*maxn, GFP_KERNEL);
+        timers = kzalloc(sizeof(mmtimer_t *)*maxn, GFP_KERNEL);
         if (timers == NULL) {
                 printk(KERN_ERR "%s: failed to allocate memory for device\n",
                                 MMTIMER_NAME);
                 goto out3;
         }
 
-        memset(timers,0,(sizeof(mmtimer_t *)*maxn));
-
         /* Allocate mmtimer_t's for each online node */
         for_each_online_node(node) {
                 timers[node] = kmalloc_node(sizeof(mmtimer_t)*NUM_COMPARATORS, GFP_KERNEL, node);
diff --git a/include/asm-ia64/hw_irq.h b/include/asm-ia64/hw_irq.h
index 27f9df6b9145..c054d7a9aaa7 100644
--- a/include/asm-ia64/hw_irq.h
+++ b/include/asm-ia64/hw_irq.h
@@ -66,6 +66,7 @@ extern int ia64_last_device_vector;
 #define IA64_PERFMON_VECTOR             0xee    /* performanc monitor interrupt vector */
 #define IA64_TIMER_VECTOR               0xef    /* use highest-prio group 15 interrupt for timer */
 #define IA64_MCA_WAKEUP_VECTOR          0xf0    /* MCA wakeup (must be >MCA_RENDEZ_VECTOR) */
+#define IA64_IPI_LOCAL_TLB_FLUSH        0xfc    /* SMP flush local TLB */
 #define IA64_IPI_RESCHEDULE             0xfd    /* SMP reschedule */
 #define IA64_IPI_VECTOR                 0xfe    /* inter-processor interrupt vector */
 
diff --git a/include/asm-ia64/iosapic.h b/include/asm-ia64/iosapic.h
index 20f98f1751a1..421cb6b62a7c 100644
--- a/include/asm-ia64/iosapic.h
+++ b/include/asm-ia64/iosapic.h
@@ -83,7 +83,7 @@ extern int gsi_to_irq (unsigned int gsi);
 extern int iosapic_register_intr (unsigned int gsi, unsigned long polarity,
                                   unsigned long trigger);
 extern void iosapic_unregister_intr (unsigned int irq);
-extern void __init iosapic_override_isa_irq (unsigned int isa_irq, unsigned int gsi,
+extern void __devinit iosapic_override_isa_irq (unsigned int isa_irq, unsigned int gsi,
                                       unsigned long polarity,
                                       unsigned long trigger);
 extern int __init iosapic_register_platform_intr (u32 int_type,
diff --git a/include/asm-ia64/sn/sn_sal.h b/include/asm-ia64/sn/sn_sal.h
index 2c4004eb5a68..291e8ceed6e6 100644
--- a/include/asm-ia64/sn/sn_sal.h
+++ b/include/asm-ia64/sn/sn_sal.h
@@ -106,6 +106,7 @@
 /* interrupt handling */
 #define SAL_INTR_ALLOC          1
 #define SAL_INTR_FREE           2
+#define SAL_INTR_REDIRECT       3
 
 /*
  * operations available on the generic SN_SAL_SYSCTL_OP
diff --git a/include/asm-ia64/thread_info.h b/include/asm-ia64/thread_info.h
index d28147506585..7d0241db622b 100644
--- a/include/asm-ia64/thread_info.h
+++ b/include/asm-ia64/thread_info.h
@@ -85,6 +85,7 @@ struct thread_info {
 #define TIF_SYSCALL_TRACE       3       /* syscall trace active */
 #define TIF_SYSCALL_AUDIT       4       /* syscall auditing active */
 #define TIF_SINGLESTEP          5       /* restore singlestep on return to user mode */
+#define TIF_RESTORE_SIGMASK     6       /* restore signal mask in do_signal() */
 #define TIF_POLLING_NRFLAG      16      /* true if poll_idle() is polling TIF_NEED_RESCHED */
 #define TIF_MEMDIE              17
 #define TIF_MCA_INIT            18      /* this task is processing MCA or INIT */
@@ -96,6 +97,7 @@ struct thread_info {
 #define _TIF_SINGLESTEP         (1 << TIF_SINGLESTEP)
 #define _TIF_SYSCALL_TRACEAUDIT (_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT|_TIF_SINGLESTEP)
 #define _TIF_NOTIFY_RESUME      (1 << TIF_NOTIFY_RESUME)
+#define _TIF_RESTORE_SIGMASK    (1 << TIF_RESTORE_SIGMASK)
 #define _TIF_SIGPENDING         (1 << TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED       (1 << TIF_NEED_RESCHED)
 #define _TIF_POLLING_NRFLAG     (1 << TIF_POLLING_NRFLAG)
@@ -104,7 +106,7 @@ struct thread_info {
 #define _TIF_FREEZE             (1 << TIF_FREEZE)
 
 /* "work to do on user-return" bits */
-#define TIF_ALLWORK_MASK        (_TIF_NOTIFY_RESUME|_TIF_SIGPENDING|_TIF_NEED_RESCHED|_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT)
+#define TIF_ALLWORK_MASK        (_TIF_NOTIFY_RESUME|_TIF_SIGPENDING|_TIF_NEED_RESCHED|_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT|_TIF_RESTORE_SIGMASK)
 /* like TIF_ALLWORK_BITS but sans TIF_SYSCALL_TRACE or TIF_SYSCALL_AUDIT */
 #define TIF_WORK_MASK           (TIF_ALLWORK_MASK&~(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT))
 
diff --git a/include/asm-ia64/tlbflush.h b/include/asm-ia64/tlbflush.h
index cf9acb9bb1fb..e37f9fbf33af 100644
--- a/include/asm-ia64/tlbflush.h
+++ b/include/asm-ia64/tlbflush.h
@@ -27,9 +27,11 @@ extern void local_flush_tlb_all (void);
 #ifdef CONFIG_SMP
   extern void smp_flush_tlb_all (void);
   extern void smp_flush_tlb_mm (struct mm_struct *mm);
+  extern void smp_flush_tlb_cpumask (cpumask_t xcpumask);
 # define flush_tlb_all()        smp_flush_tlb_all()
 #else
 # define flush_tlb_all()        local_flush_tlb_all()
+# define smp_flush_tlb_cpumask(m) local_flush_tlb_all()
 #endif
 
 static inline void
@@ -94,6 +96,15 @@ flush_tlb_pgtables (struct mm_struct *mm, unsigned long start, unsigned long end
          */
 }
 
+/*
+ * Flush the local TLB. Invoked from another cpu using an IPI.
+ */
+#ifdef CONFIG_SMP
+void smp_local_flush_tlb(void);
+#else
+#define smp_local_flush_tlb()
+#endif
+
 #define flush_tlb_kernel_range(start, end)      flush_tlb_all() /* XXX fix me */
 
 #endif /* _ASM_IA64_TLBFLUSH_H */
diff --git a/include/asm-ia64/unistd.h b/include/asm-ia64/unistd.h
index a9e1fa4cac4d..861c8ec87b09 100644
--- a/include/asm-ia64/unistd.h
+++ b/include/asm-ia64/unistd.h
@@ -283,7 +283,8 @@
 #define __NR_readlinkat                 1291
 #define __NR_fchmodat                   1292
 #define __NR_faccessat                  1293
-/* 1294, 1295 reserved for pselect/ppoll */
+#define __NR_pselect6                   1294
+#define __NR_ppoll                      1295
 #define __NR_unshare                    1296
 #define __NR_splice                     1297
 #define __NR_set_robust_list            1298
@@ -300,6 +301,7 @@
 #define NR_syscalls                     281 /* length of syscall table */
 
 #define __ARCH_WANT_SYS_RT_SIGACTION
+#define __ARCH_WANT_SYS_RT_SIGSUSPEND
 
 #ifdef CONFIG_IA32_SUPPORT
 # define __ARCH_WANT_SYS_FADVISE64
@@ -310,6 +312,7 @@
 # define __ARCH_WANT_SYS_OLDUMOUNT
 # define __ARCH_WANT_SYS_SIGPENDING
 # define __ARCH_WANT_SYS_SIGPROCMASK
+# define __ARCH_WANT_COMPAT_SYS_RT_SIGSUSPEND
 # define __ARCH_WANT_COMPAT_SYS_TIME
 #endif