Merge commit '8700c95adb03' into timers/nohz

The full dynticks tree needs the latest RCU and sched
upstream updates in order to fix some dependencies.

Merge a common upstream merge point that has these
updates.

Conflicts:
	include/linux/perf_event.h
	kernel/rcutree.h
	kernel/rcutree_plugin.h

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>

9 files changed, 103 insertions, 204 deletions

diff --git a/arch/ia64/kernel/fsys.S b/arch/ia64/kernel/fsys.S
index c4cd45d97749..abc6dee3799c 100644
--- a/arch/ia64/kernel/fsys.S
+++ b/arch/ia64/kernel/fsys.S
@@ -90,53 +90,6 @@ ENTRY(fsys_getpid)
 	FSYS_RETURN
 END(fsys_getpid)
 
-ENTRY(fsys_getppid)
-	.prologue
-	.altrp b6
-	.body
-	add r17=IA64_TASK_GROUP_LEADER_OFFSET,r16
-	;;
-	ld8 r17=[r17]				// r17 = current->group_leader
-	add r9=TI_FLAGS+IA64_TASK_SIZE,r16
-	;;
-
-	ld4 r9=[r9]
-	add r17=IA64_TASK_REAL_PARENT_OFFSET,r17 // r17 = &current->group_leader->real_parent
-	;;
-	and r9=TIF_ALLWORK_MASK,r9
-
-1:	ld8 r18=[r17]				// r18 = current->group_leader->real_parent
-	;;
-	cmp.ne p8,p0=0,r9
-	add r8=IA64_TASK_TGID_OFFSET,r18	// r8 = &current->group_leader->real_parent->tgid
-	;;
-
-	/*
-	 * The .acq is needed to ensure that the read of tgid has returned its data before
-	 * we re-check "real_parent".
-	 */
-	ld4.acq r8=[r8]				// r8 = current->group_leader->real_parent->tgid
-#ifdef CONFIG_SMP
-	/*
-	 * Re-read current->group_leader->real_parent.
-	 */
-	ld8 r19=[r17]				// r19 = current->group_leader->real_parent
-(p8)	br.spnt.many fsys_fallback_syscall
-	;;
-	cmp.ne p6,p0=r18,r19			// did real_parent change?
-	mov r19=0			// i must not leak kernel bits...
-(p6)	br.cond.spnt.few 1b			// yes -> redo the read of tgid and the check
-	;;
-	mov r17=0			// i must not leak kernel bits...
-	mov r18=0			// i must not leak kernel bits...
-#else
-	mov r17=0			// i must not leak kernel bits...
-	mov r18=0			// i must not leak kernel bits...
-	mov r19=0			// i must not leak kernel bits...
-#endif
-	FSYS_RETURN
-END(fsys_getppid)
-
 ENTRY(fsys_set_tid_address)
 	.prologue
 	.altrp b6
@@ -614,7 +567,7 @@ paravirt_fsyscall_table:
 	data8 0				// chown
 	data8 0				// lseek		// 1040
 	data8 fsys_getpid		// getpid
-	data8 fsys_getppid		// getppid
+	data8 0				// getppid
 	data8 0				// mount
 	data8 0				// umount
 	data8 0				// setuid		// 1045
diff --git a/arch/ia64/kernel/iosapic.c b/arch/ia64/kernel/iosapic.c
index ee33c3aaa2fc..19f107be734e 100644
--- a/arch/ia64/kernel/iosapic.c
+++ b/arch/ia64/kernel/iosapic.c
@@ -76,7 +76,7 @@
  *	PCI pin -> global system interrupt (GSI) -> IA-64 vector <-> IRQ
  *
  * Note: The term "IRQ" is loosely used everywhere in Linux kernel to
- * describeinterrupts.  Now we use "IRQ" only for Linux IRQ's.  ISA IRQ
+ * describe interrupts.  Now we use "IRQ" only for Linux IRQ's.  ISA IRQ
  * (isa_irq) is the only exception in this source code.
  */
 
@@ -1010,6 +1010,26 @@ iosapic_check_gsi_range (unsigned int gsi_base, unsigned int ver)
 	return 0;
 }
 
+static int
+iosapic_delete_rte(unsigned int irq, unsigned int gsi)
+{
+	struct iosapic_rte_info *rte, *temp;
+
+	list_for_each_entry_safe(rte, temp, &iosapic_intr_info[irq].rtes,
+								rte_list) {
+		if (rte->iosapic->gsi_base + rte->rte_index == gsi) {
+			if (rte->refcnt)
+				return -EBUSY;
+
+			list_del(&rte->rte_list);
+			kfree(rte);
+			return 0;
+		}
+	}
+
+	return -EINVAL;
+}
+
 int iosapic_init(unsigned long phys_addr, unsigned int gsi_base)
 {
 	int num_rte, err, index;
@@ -1069,7 +1089,7 @@ int iosapic_init(unsigned long phys_addr, unsigned int gsi_base)
 
 int iosapic_remove(unsigned int gsi_base)
 {
-	int index, err = 0;
+	int i, irq, index, err = 0;
 	unsigned long flags;
 
 	spin_lock_irqsave(&iosapic_lock, flags);
@@ -1087,6 +1107,16 @@ int iosapic_remove(unsigned int gsi_base)
 		goto out;
 	}
 
+	for (i = gsi_base; i < gsi_base + iosapic_lists[index].num_rte; i++) {
+		irq = __gsi_to_irq(i);
+		if (irq < 0)
+			continue;
+
+		err = iosapic_delete_rte(irq, i);
+		if (err)
+			goto out;
+	}
+
 	iounmap(iosapic_lists[index].addr);
 	iosapic_free(index);
  out:
diff --git a/arch/ia64/kernel/irq.c b/arch/ia64/kernel/irq.c
index ad69606613eb..f2c418281130 100644
--- a/arch/ia64/kernel/irq.c
+++ b/arch/ia64/kernel/irq.c
@@ -23,6 +23,8 @@
 #include <linux/interrupt.h>
 #include <linux/kernel_stat.h>
 
+#include <asm/mca.h>
+
 /*
  * 'what should we do if we get a hw irq event on an illegal vector'.
  * each architecture has to answer this themselves.
@@ -83,6 +85,12 @@ bool is_affinity_mask_valid(const struct cpumask *cpumask)
 
 #endif /* CONFIG_SMP */
 
+int __init arch_early_irq_init(void)
+{
+	ia64_mca_irq_init();
+	return 0;
+}
+
 #ifdef CONFIG_HOTPLUG_CPU
 unsigned int vectors_in_migration[NR_IRQS];
 
diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c
index 65bf9cd39044..d7396dbb07bb 100644
--- a/arch/ia64/kernel/mca.c
+++ b/arch/ia64/kernel/mca.c
@@ -2074,22 +2074,16 @@ ia64_mca_init(void)
 	printk(KERN_INFO "MCA related initialization done\n");
 }
 
+
 /*
- * ia64_mca_late_init
- *
- *	Opportunity to setup things that require initialization later
- *	than ia64_mca_init.  Setup a timer to poll for CPEs if the
- *	platform doesn't support an interrupt driven mechanism.
- *
- *  Inputs  :   None
- *  Outputs :   Status
+ * These pieces cannot be done in ia64_mca_init() because it is called before
+ * early_irq_init() which would wipe out our percpu irq registrations. But we
+ * cannot leave them until ia64_mca_late_init() because by then all the other
+ * processors have been brought online and have set their own CMC vectors to
+ * point at a non-existant action. Called from arch_early_irq_init().
  */
-static int __init
-ia64_mca_late_init(void)
+void __init ia64_mca_irq_init(void)
 {
-	if (!mca_init)
-		return 0;
-
 	/*
 	 *  Configure the CMCI/P vector and handler. Interrupts for CMC are
 	 *  per-processor, so AP CMC interrupts are setup in smp_callin() (smpboot.c).
@@ -2108,6 +2102,23 @@ ia64_mca_late_init(void)
 	/* Setup the CPEI/P handler */
 	register_percpu_irq(IA64_CPEP_VECTOR, &mca_cpep_irqaction);
 #endif
+}
+
+/*
+ * ia64_mca_late_init
+ *
+ *	Opportunity to setup things that require initialization later
+ *	than ia64_mca_init.  Setup a timer to poll for CPEs if the
+ *	platform doesn't support an interrupt driven mechanism.
+ *
+ *  Inputs  :   None
+ *  Outputs :   Status
+ */
+static int __init
+ia64_mca_late_init(void)
+{
+	if (!mca_init)
+		return 0;
 
 	register_hotcpu_notifier(&mca_cpu_notifier);
 
diff --git a/arch/ia64/kernel/mca_drv.c b/arch/ia64/kernel/mca_drv.c
index 9392e021c93b..94f8bf777afa 100644
--- a/arch/ia64/kernel/mca_drv.c
+++ b/arch/ia64/kernel/mca_drv.c
@@ -349,7 +349,7 @@ init_record_index_pools(void)
 
 	/* - 3 - */
 	slidx_pool.max_idx = (rec_max_size/sect_min_size) * 2 + 1;
-	slidx_pool.buffer = (slidx_list_t *)
+	slidx_pool.buffer =
 		kmalloc(slidx_pool.max_idx * sizeof(slidx_list_t), GFP_KERNEL);
 
 	return slidx_pool.buffer ? 0 : -ENOMEM;
diff --git a/arch/ia64/kernel/palinfo.c b/arch/ia64/kernel/palinfo.c
index 77597e5ea60a..79521d5499f9 100644
--- a/arch/ia64/kernel/palinfo.c
+++ b/arch/ia64/kernel/palinfo.c
@@ -849,17 +849,6 @@ static palinfo_entry_t palinfo_entries[]={
 
 #define NR_PALINFO_ENTRIES	(int) ARRAY_SIZE(palinfo_entries)
 
-/*
- * this array is used to keep track of the proc entries we create. This is
- * required in the module mode when we need to remove all entries. The procfs code
- * does not do recursion of deletion
- *
- * Notes:
- *	- +1 accounts for the cpuN directory entry in /proc/pal
- */
-#define NR_PALINFO_PROC_ENTRIES	(NR_CPUS*(NR_PALINFO_ENTRIES+1))
-
-static struct proc_dir_entry *palinfo_proc_entries[NR_PALINFO_PROC_ENTRIES];
 static struct proc_dir_entry *palinfo_dir;
 
 /*
@@ -971,60 +960,32 @@ palinfo_read_entry(char *page, char **start, off_t off, int count, int *eof, voi
 static void __cpuinit
 create_palinfo_proc_entries(unsigned int cpu)
 {
-#	define CPUSTR	"cpu%d"
-
 	pal_func_cpu_u_t f;
-	struct proc_dir_entry **pdir;
 	struct proc_dir_entry *cpu_dir;
 	int j;
-	char cpustr[sizeof(CPUSTR)];
-
-
-	/*
-	 * we keep track of created entries in a depth-first order for
-	 * cleanup purposes. Each entry is stored into palinfo_proc_entries
-	 */
-	sprintf(cpustr,CPUSTR, cpu);
+	char cpustr[3+4+1];	/* cpu numbers are up to 4095 on itanic */
+	sprintf(cpustr, "cpu%d", cpu);
 
 	cpu_dir = proc_mkdir(cpustr, palinfo_dir);
+	if (!cpu_dir)
+		return;
 
 	f.req_cpu = cpu;
 
-	/*
-	 * Compute the location to store per cpu entries
-	 * We dont store the top level entry in this list, but
-	 * remove it finally after removing all cpu entries.
-	 */
-	pdir = &palinfo_proc_entries[cpu*(NR_PALINFO_ENTRIES+1)];
-	*pdir++ = cpu_dir;
 	for (j=0; j < NR_PALINFO_ENTRIES; j++) {
 		f.func_id = j;
-		*pdir = create_proc_read_entry(
-				palinfo_entries[j].name, 0, cpu_dir,
-				palinfo_read_entry, (void *)f.value);
-		pdir++;
+		create_proc_read_entry(
+			palinfo_entries[j].name, 0, cpu_dir,
+			palinfo_read_entry, (void *)f.value);
 	}
 }
 
 static void
 remove_palinfo_proc_entries(unsigned int hcpu)
 {
-	int j;
-	struct proc_dir_entry *cpu_dir, **pdir;
-
-	pdir = &palinfo_proc_entries[hcpu*(NR_PALINFO_ENTRIES+1)];
-	cpu_dir = *pdir;
-	*pdir++=NULL;
-	for (j=0; j < (NR_PALINFO_ENTRIES); j++) {
-		if ((*pdir)) {
-			remove_proc_entry ((*pdir)->name, cpu_dir);
-			*pdir ++= NULL;
-		}
-	}
-
-	if (cpu_dir) {
-		remove_proc_entry(cpu_dir->name, palinfo_dir);
-	}
+	char cpustr[3+4+1];	/* cpu numbers are up to 4095 on itanic */
+	sprintf(cpustr, "cpu%d", hcpu);
+	remove_proc_subtree(cpustr, palinfo_dir);
 }
 
 static int __cpuinit palinfo_cpu_callback(struct notifier_block *nfb,
@@ -1058,6 +1019,8 @@ palinfo_init(void)
 
 	printk(KERN_INFO "PAL Information Facility v%s\n", PALINFO_VERSION);
 	palinfo_dir = proc_mkdir("pal", NULL);
+	if (!palinfo_dir)
+		return -ENOMEM;
 
 	/* Create palinfo dirs in /proc for all online cpus */
 	for_each_online_cpu(i) {
@@ -1073,22 +1036,8 @@ palinfo_init(void)
 static void __exit
 palinfo_exit(void)
 {
-	int i = 0;
-
-	/* remove all nodes: depth first pass. Could optimize this  */
-	for_each_online_cpu(i) {
-		remove_palinfo_proc_entries(i);
-	}
-
-	/*
-	 * Remove the top level entry finally
-	 */
-	remove_proc_entry(palinfo_dir->name, NULL);
-
-	/*
-	 * Unregister from cpu notifier callbacks
-	 */
 	unregister_hotcpu_notifier(&palinfo_cpu_notifier);
+	remove_proc_subtree("pal", NULL);
 }
 
 module_init(palinfo_init);
diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
index 433f5e8a2cd1..9ea25fce06d5 100644
--- a/arch/ia64/kernel/perfmon.c
+++ b/arch/ia64/kernel/perfmon.c
@@ -42,6 +42,7 @@
 #include <linux/completion.h>
 #include <linux/tracehook.h>
 #include <linux/slab.h>
+#include <linux/cpu.h>
 
 #include <asm/errno.h>
 #include <asm/intrinsics.h>
@@ -619,6 +620,7 @@ static struct file_system_type pfm_fs_type = {
 	.mount    = pfmfs_mount,
 	.kill_sb  = kill_anon_super,
 };
+MODULE_ALIAS_FS("pfmfs");
 
 DEFINE_PER_CPU(unsigned long, pfm_syst_info);
 DEFINE_PER_CPU(struct task_struct *, pmu_owner);
@@ -1321,8 +1323,6 @@ out:
 }
 EXPORT_SYMBOL(pfm_unregister_buffer_fmt);
 
-extern void update_pal_halt_status(int);
-
 static int
 pfm_reserve_session(struct task_struct *task, int is_syswide, unsigned int cpu)
 {
@@ -1370,9 +1370,9 @@ pfm_reserve_session(struct task_struct *task, int is_syswide, unsigned int cpu)
 		cpu));
 
 	/*
-	 * disable default_idle() to go to PAL_HALT
+	 * Force idle() into poll mode
 	 */
-	update_pal_halt_status(0);
+	cpu_idle_poll_ctrl(true);
 
 	UNLOCK_PFS(flags);
 
@@ -1429,11 +1429,8 @@ pfm_unreserve_session(pfm_context_t *ctx, int is_syswide, unsigned int cpu)
 		is_syswide,
 		cpu));
 
-	/*
-	 * if possible, enable default_idle() to go into PAL_HALT
-	 */
-	if (pfm_sessions.pfs_task_sessions == 0 && pfm_sessions.pfs_sys_sessions == 0)
-		update_pal_halt_status(1);
+	/* Undo forced polling. Last session reenables pal_halt */
+	cpu_idle_poll_ctrl(false);
 
 	UNLOCK_PFS(flags);
 
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index e34f565f595a..a26fc640e4ce 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -209,41 +209,13 @@ do_notify_resume_user(sigset_t *unused, struct sigscratch *scr, long in_syscall)
 	local_irq_disable();	/* force interrupt disable */
 }
 
-static int pal_halt        = 1;
-static int can_do_pal_halt = 1;
-
 static int __init nohalt_setup(char * str)
 {
-	pal_halt = can_do_pal_halt = 0;
+	cpu_idle_poll_ctrl(true);
 	return 1;
 }
 __setup("nohalt", nohalt_setup);
 
-void
-update_pal_halt_status(int status)
-{
-	can_do_pal_halt = pal_halt && status;
-}
-
-/*
- * We use this if we don't have any better idle routine..
- */
-void
-default_idle (void)
-{
-	local_irq_enable();
-	while (!need_resched()) {
-		if (can_do_pal_halt) {
-			local_irq_disable();
-			if (!need_resched()) {
-				safe_halt();
-			}
-			local_irq_enable();
-		} else
-			cpu_relax();
-	}
-}
-
 #ifdef CONFIG_HOTPLUG_CPU
 /* We don't actually take CPU down, just spin without interrupts. */
 static inline void play_dead(void)
@@ -270,50 +242,29 @@ static inline void play_dead(void)
 }
 #endif /* CONFIG_HOTPLUG_CPU */
 
-void __attribute__((noreturn))
-cpu_idle (void)
+void arch_cpu_idle_dead(void)
+{
+	play_dead();
+}
+
+void arch_cpu_idle(void)
 {
 	void (*mark_idle)(int) = ia64_mark_idle;
-  	int cpu = smp_processor_id();
-
-	/* endless idle loop with no priority at all */
-	while (1) {
-		rcu_idle_enter();
-		if (can_do_pal_halt) {
-			current_thread_info()->status &= ~TS_POLLING;
-			/*
-			 * TS_POLLING-cleared state must be visible before we
-			 * test NEED_RESCHED:
-			 */
-			smp_mb();
-		} else {
-			current_thread_info()->status |= TS_POLLING;
-		}
 
-		if (!need_resched()) {
-			void (*idle)(void);
 #ifdef CONFIG_SMP
-			min_xtp();
+	min_xtp();
 #endif
-			rmb();
-			if (mark_idle)
-				(*mark_idle)(1);
-
-			if (!idle)
-				idle = default_idle;
-			(*idle)();
-			if (mark_idle)
-				(*mark_idle)(0);
+	rmb();
+	if (mark_idle)
+		(*mark_idle)(1);
+
+	safe_halt();
+
+	if (mark_idle)
+		(*mark_idle)(0);
 #ifdef CONFIG_SMP
-			normal_xtp();
+	normal_xtp();
 #endif
-		}
-		rcu_idle_exit();
-		schedule_preempt_disabled();
-		check_pgt_cache();
-		if (cpu_is_offline(cpu))
-			play_dead();
-	}
 }
 
 void
diff --git a/arch/ia64/kernel/smpboot.c b/arch/ia64/kernel/smpboot.c
index 500f1e4d9f9d..8d87168d218d 100644
--- a/arch/ia64/kernel/smpboot.c
+++ b/arch/ia64/kernel/smpboot.c
@@ -455,7 +455,7 @@ start_secondary (void *unused)
 	preempt_disable();
 	smp_callin();
 
-	cpu_idle();
+	cpu_startup_entry(CPUHP_ONLINE);
 	return 0;
 }