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

11 files changed, 132 insertions, 104 deletions

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);