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

2 files changed, 103 insertions, 20 deletions

diff --git a/arch/ia64/sn/kernel/irq.c b/arch/ia64/sn/kernel/irq.c
index 8d2a1bf..7f6d236 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 601747b..5d318b5 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,
 };