diff options
Diffstat (limited to 'kernel')
| -rw-r--r-- | kernel/Makefile | 2 | ||||
| -rw-r--r-- | kernel/exit.c | 4 | ||||
| -rw-r--r-- | kernel/hw_breakpoint.c | 67 | ||||
| -rw-r--r-- | kernel/jump_label.c | 429 | ||||
| -rw-r--r-- | kernel/kprobes.c | 26 | ||||
| -rw-r--r-- | kernel/module.c | 6 | ||||
| -rw-r--r-- | kernel/perf_event.c | 2355 | ||||
| -rw-r--r-- | kernel/sched.c | 2 | ||||
| -rw-r--r-- | kernel/trace/ftrace.c | 123 | ||||
| -rw-r--r-- | kernel/trace/ring_buffer.c | 21 | ||||
| -rw-r--r-- | kernel/trace/trace_event_perf.c | 28 | ||||
| -rw-r--r-- | kernel/trace/trace_events.c | 55 | ||||
| -rw-r--r-- | kernel/trace/trace_functions_graph.c | 126 | ||||
| -rw-r--r-- | kernel/trace/trace_workqueue.c | 10 | ||||
| -rw-r--r-- | kernel/tracepoint.c | 14 | ||||
| -rw-r--r-- | kernel/watchdog.c | 41 |
16 files changed, 2206 insertions, 1103 deletions
diff --git a/kernel/Makefile b/kernel/Makefile index 0b72d1a74be..d52b473c99a 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -10,7 +10,7 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o \ kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o cred.o \ - async.o range.o + async.o range.o jump_label.o obj-$(CONFIG_HAVE_EARLY_RES) += early_res.o obj-y += groups.o diff --git a/kernel/exit.c b/kernel/exit.c index 03120229db2..e2bdf37f9fd 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -149,9 +149,7 @@ static void delayed_put_task_struct(struct rcu_head *rhp) { struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); -#ifdef CONFIG_PERF_EVENTS - WARN_ON_ONCE(tsk->perf_event_ctxp); -#endif + perf_event_delayed_put(tsk); trace_sched_process_free(tsk); put_task_struct(tsk); } diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index c7c2aed9e2d..3b714e839c1 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -433,8 +433,7 @@ register_user_hw_breakpoint(struct perf_event_attr *attr, perf_overflow_handler_t triggered, struct task_struct *tsk) { - return perf_event_create_kernel_counter(attr, -1, task_pid_vnr(tsk), - triggered); + return perf_event_create_kernel_counter(attr, -1, tsk, triggered); } EXPORT_SYMBOL_GPL(register_user_hw_breakpoint); @@ -516,7 +515,7 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr, get_online_cpus(); for_each_online_cpu(cpu) { pevent = per_cpu_ptr(cpu_events, cpu); - bp = perf_event_create_kernel_counter(attr, cpu, -1, triggered); + bp = perf_event_create_kernel_counter(attr, cpu, NULL, triggered); *pevent = bp; @@ -566,6 +565,61 @@ static struct notifier_block hw_breakpoint_exceptions_nb = { .priority = 0x7fffffff }; +static void bp_perf_event_destroy(struct perf_event *event) +{ + release_bp_slot(event); +} + +static int hw_breakpoint_event_init(struct perf_event *bp) +{ + int err; + + if (bp->attr.type != PERF_TYPE_BREAKPOINT) + return -ENOENT; + + err = register_perf_hw_breakpoint(bp); + if (err) + return err; + + bp->destroy = bp_perf_event_destroy; + + return 0; +} + +static int hw_breakpoint_add(struct perf_event *bp, int flags) +{ + if (!(flags & PERF_EF_START)) + bp->hw.state = PERF_HES_STOPPED; + + return arch_install_hw_breakpoint(bp); +} + +static void hw_breakpoint_del(struct perf_event *bp, int flags) +{ + arch_uninstall_hw_breakpoint(bp); +} + +static void hw_breakpoint_start(struct perf_event *bp, int flags) +{ + bp->hw.state = 0; +} + +static void hw_breakpoint_stop(struct perf_event *bp, int flags) +{ + bp->hw.state = PERF_HES_STOPPED; +} + +static struct pmu perf_breakpoint = { + .task_ctx_nr = perf_sw_context, /* could eventually get its own */ + + .event_init = hw_breakpoint_event_init, + .add = hw_breakpoint_add, + .del = hw_breakpoint_del, + .start = hw_breakpoint_start, + .stop = hw_breakpoint_stop, + .read = hw_breakpoint_pmu_read, +}; + static int __init init_hw_breakpoint(void) { unsigned int **task_bp_pinned; @@ -587,6 +641,8 @@ static int __init init_hw_breakpoint(void) constraints_initialized = 1; + perf_pmu_register(&perf_breakpoint); + return register_die_notifier(&hw_breakpoint_exceptions_nb); err_alloc: @@ -602,8 +658,3 @@ static int __init init_hw_breakpoint(void) core_initcall(init_hw_breakpoint); -struct pmu perf_ops_bp = { - .enable = arch_install_hw_breakpoint, - .disable = arch_uninstall_hw_breakpoint, - .read = hw_breakpoint_pmu_read, -}; diff --git a/kernel/jump_label.c b/kernel/jump_label.c new file mode 100644 index 00000000000..7be868bf25c --- /dev/null +++ b/kernel/jump_label.c @@ -0,0 +1,429 @@ +/* + * jump label support + * + * Copyright (C) 2009 Jason Baron <jbaron@redhat.com> + * + */ +#include <linux/jump_label.h> +#include <linux/memory.h> +#include <linux/uaccess.h> +#include <linux/module.h> +#include <linux/list.h> +#include <linux/jhash.h> +#include <linux/slab.h> +#include <linux/sort.h> +#include <linux/err.h> + +#ifdef HAVE_JUMP_LABEL + +#define JUMP_LABEL_HASH_BITS 6 +#define JUMP_LABEL_TABLE_SIZE (1 << JUMP_LABEL_HASH_BITS) +static struct hlist_head jump_label_table[JUMP_LABEL_TABLE_SIZE]; + +/* mutex to protect coming/going of the the jump_label table */ +static DEFINE_MUTEX(jump_label_mutex); + +struct jump_label_entry { + struct hlist_node hlist; + struct jump_entry *table; + int nr_entries; + /* hang modules off here */ + struct hlist_head modules; + unsigned long key; +}; + +struct jump_label_module_entry { + struct hlist_node hlist; + struct jump_entry *table; + int nr_entries; + struct module *mod; +}; + +static int jump_label_cmp(const void *a, const void *b) +{ + const struct jump_entry *jea = a; + const struct jump_entry *jeb = b; + + if (jea->key < jeb->key) + return -1; + + if (jea->key > jeb->key) + return 1; + + return 0; +} + +static void +sort_jump_label_entries(struct jump_entry *start, struct jump_entry *stop) +{ + unsigned long size; + + size = (((unsigned long)stop - (unsigned long)start) + / sizeof(struct jump_entry)); + sort(start, size, sizeof(struct jump_entry), jump_label_cmp, NULL); +} + +static struct jump_label_entry *get_jump_label_entry(jump_label_t key) +{ + struct hlist_head *head; + struct hlist_node *node; + struct jump_label_entry *e; + u32 hash = jhash((void *)&key, sizeof(jump_label_t), 0); + + head = &jump_label_table[hash & (JUMP_LABEL_TABLE_SIZE - 1)]; + hlist_for_each_entry(e, node, head, hlist) { + if (key == e->key) + return e; + } + return NULL; +} + +static struct jump_label_entry * +add_jump_label_entry(jump_label_t key, int nr_entries, struct jump_entry *table) +{ + struct hlist_head *head; + struct jump_label_entry *e; + u32 hash; + + e = get_jump_label_entry(key); + if (e) + return ERR_PTR(-EEXIST); + + e = kmalloc(sizeof(struct jump_label_entry), GFP_KERNEL); + if (!e) + return ERR_PTR(-ENOMEM); + + hash = jhash((void *)&key, sizeof(jump_label_t), 0); + head = &jump_label_table[hash & (JUMP_LABEL_TABLE_SIZE - 1)]; + e->key = key; + e->table = table; + e->nr_entries = nr_entries; + INIT_HLIST_HEAD(&(e->modules)); + hlist_add_head(&e->hlist, head); + return e; +} + +static int +build_jump_label_hashtable(struct jump_entry *start, struct jump_entry *stop) +{ + struct jump_entry *iter, *iter_begin; + struct jump_label_entry *entry; + int count; + + sort_jump_label_entries(start, stop); + iter = start; + while (iter < stop) { + entry = get_jump_label_entry(iter->key); + if (!entry) { + iter_begin = iter; + count = 0; + while ((iter < stop) && + (iter->key == iter_begin->key)) { + iter++; + count++; + } + entry = add_jump_label_entry(iter_begin->key, + count, iter_begin); + if (IS_ERR(entry)) + return PTR_ERR(entry); + } else { + WARN_ONCE(1, KERN_ERR "build_jump_hashtable: unexpected entry!\n"); + return -1; + } + } + return 0; +} + +/*** + * jump_label_update - update jump label text + * @key - key value associated with a a jump label + * @type - enum set to JUMP_LABEL_ENABLE or JUMP_LABEL_DISABLE + * + * Will enable/disable the jump for jump label @key, depending on the + * value of @type. + * + */ + +void jump_label_update(unsigned long key, enum jump_label_type type) +{ + struct jump_entry *iter; + struct jump_label_entry *entry; + struct hlist_node *module_node; + struct jump_label_module_entry *e_module; + int count; + + mutex_lock(&jump_label_mutex); + entry = get_jump_label_entry((jump_label_t)key); + if (entry) { + count = entry->nr_entries; + iter = entry->table; + while (count--) { + if (kernel_text_address(iter->code)) + arch_jump_label_transform(iter, type); + iter++; + } + /* eanble/disable jump labels in modules */ + hlist_for_each_entry(e_module, module_node, &(entry->modules), + hlist) { + count = e_module->nr_entries; + iter = e_module->table; + while (count--) { + if (kernel_text_address(iter->code)) + arch_jump_label_transform(iter, type); + iter++; + } + } + } + mutex_unlock(&jump_label_mutex); +} + +static int addr_conflict(struct jump_entry *entry, void *start, void *end) +{ + if (entry->code <= (unsigned long)end && + entry->code + JUMP_LABEL_NOP_SIZE > (unsigned long)start) + return 1; + + return 0; +} + +#ifdef CONFIG_MODULES + +static int module_conflict(void *start, void *end) +{ + struct hlist_head *head; + struct hlist_node *node, *node_next, *module_node, *module_node_next; + struct jump_label_entry *e; + struct jump_label_module_entry *e_module; + struct jump_entry *iter; + int i, count; + int conflict = 0; + + for (i = 0; i < JUMP_LABEL_TABLE_SIZE; i++) { + head = &jump_label_table[i]; + hlist_for_each_entry_safe(e, node, node_next, head, hlist) { + hlist_for_each_entry_safe(e_module, module_node, + module_node_next, + &(e->modules), hlist) { + count = e_module->nr_entries; + iter = e_module->table; + while (count--) { + if (addr_conflict(iter, start, end)) { + conflict = 1; + goto out; + } + iter++; + } + } + } + } +out: + return conflict; +} + +#endif + +/*** + * jump_label_text_reserved - check if addr range is reserved + * @start: start text addr + * @end: end text addr + * + * checks if the text addr located between @start and @end + * overlaps with any of the jump label patch addresses. Code + * that wants to modify kernel text should first verify that + * it does not overlap with any of the jump label addresses. + * + * returns 1 if there is an overlap, 0 otherwise + */ +int jump_label_text_reserved(void *start, void *end) +{ + struct jump_entry *iter; + struct jump_entry *iter_start = __start___jump_table; + struct jump_entry *iter_stop = __start___jump_table; + int conflict = 0; + + mutex_lock(&jump_label_mutex); + iter = iter_start; + while (iter < iter_stop) { + if (addr_conflict(iter, start, end)) { + conflict = 1; + goto out; + } + iter++; + } + + /* now check modules */ +#ifdef CONFIG_MODULES + conflict = module_conflict(start, end); +#endif +out: + mutex_unlock(&jump_label_mutex); + return conflict; +} + +static __init int init_jump_label(void) +{ + int ret; + struct jump_entry *iter_start = __start___jump_table; + struct jump_entry *iter_stop = __stop___jump_table; + struct jump_entry *iter; + + mutex_lock(&jump_label_mutex); + ret = build_jump_label_hashtable(__start___jump_table, + __stop___jump_table); + iter = iter_start; + while (iter < iter_stop) { + arch_jump_label_text_poke_early(iter->code); + iter++; + } + mutex_unlock(&jump_label_mutex); + return ret; +} +early_initcall(init_jump_label); + +#ifdef CONFIG_MODULES + +static struct jump_label_module_entry * +add_jump_label_module_entry(struct jump_label_entry *entry, + struct jump_entry *iter_begin, + int count, struct module *mod) +{ + struct jump_label_module_entry *e; + + e = kmalloc(sizeof(struct jump_label_module_entry), GFP_KERNEL); + if (!e) + return ERR_PTR(-ENOMEM); + e->mod = mod; + e->nr_entries = count; + e->table = iter_begin; + hlist_add_head(&e->hlist, &entry->modules); + return e; +} + +static int add_jump_label_module(struct module *mod) +{ + struct jump_entry *iter, *iter_begin; + struct jump_label_entry *entry; + struct jump_label_module_entry *module_entry; + int count; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return 0; + + sort_jump_label_entries(mod->jump_entries, + mod->jump_entries + mod->num_jump_entries); + iter = mod->jump_entries; + while (iter < mod->jump_entries + mod->num_jump_entries) { + entry = get_jump_label_entry(iter->key); + iter_begin = iter; + count = 0; + while ((iter < mod->jump_entries + mod->num_jump_entries) && + (iter->key == iter_begin->key)) { + iter++; + count++; + } + if (!entry) { + entry = add_jump_label_entry(iter_begin->key, 0, NULL); + if (IS_ERR(entry)) + return PTR_ERR(entry); + } + module_entry = add_jump_label_module_entry(entry, iter_begin, + count, mod); + if (IS_ERR(module_entry)) + return PTR_ERR(module_entry); + } + return 0; +} + +static void remove_jump_label_module(struct module *mod) +{ + struct hlist_head *head; + struct hlist_node *node, *node_next, *module_node, *module_node_next; + struct jump_label_entry *e; + struct jump_label_module_entry *e_module; + int i; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return; + + for (i = 0; i < JUMP_LABEL_TABLE_SIZE; i++) { + head = &jump_label_table[i]; + hlist_for_each_entry_safe(e, node, node_next, head, hlist) { + hlist_for_each_entry_safe(e_module, module_node, + module_node_next, + &(e->modules), hlist) { + if (e_module->mod == mod) { + hlist_del(&e_module->hlist); + kfree(e_module); + } + } + if (hlist_empty(&e->modules) && (e->nr_entries == 0)) { + hlist_del(&e->hlist); + kfree(e); + } + } + } +} + +static int +jump_label_module_notify(struct notifier_block *self, unsigned long val, + void *data) +{ + struct module *mod = data; + int ret = 0; + + switch (val) { + case MODULE_STATE_COMING: + mutex_lock(&jump_label_mutex); + ret = add_jump_label_module(mod); + if (ret) + remove_jump_label_module(mod); + mutex_unlock(&jump_label_mutex); + break; + case MODULE_STATE_GOING: + mutex_lock(&jump_label_mutex); + remove_jump_label_module(mod); + mutex_unlock(&jump_label_mutex); + break; + } + return ret; +} + +/*** + * apply_jump_label_nops - patch module jump labels with arch_get_jump_label_nop() + * @mod: module to patch + * + * Allow for run-time selection of the optimal nops. Before the module + * loads patch these with arch_get_jump_label_nop(), which is specified by + * the arch specific jump label code. + */ +void jump_label_apply_nops(struct module *mod) +{ + struct jump_entry *iter; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return; + + iter = mod->jump_entries; + while (iter < mod->jump_entries + mod->num_jump_entries) { + arch_jump_label_text_poke_early(iter->code); + iter++; + } +} + +struct notifier_block jump_label_module_nb = { + .notifier_call = jump_label_module_notify, + .priority = 0, +}; + +static __init int init_jump_label_module(void) +{ + return register_module_notifier(&jump_label_module_nb); +} +early_initcall(init_jump_label_module); + +#endif /* CONFIG_MODULES */ + +#endif diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 282035f3ae9..ec4210c6501 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -47,6 +47,7 @@ #include <linux/memory.h> #include <linux/ftrace.h> #include <linux/cpu.h> +#include <linux/jump_label.h> #include <asm-generic/sections.h> #include <asm/cacheflush.h> @@ -399,7 +400,7 @@ static inline int kprobe_optready(struct kprobe *p) * Return an optimized kprobe whose optimizing code replaces * instructions including addr (exclude breakpoint). */ -struct kprobe *__kprobes get_optimized_kprobe(unsigned long addr) +static struct kprobe *__kprobes get_optimized_kprobe(unsigned long addr) { int i; struct kprobe *p = NULL; @@ -831,6 +832,7 @@ void __kprobes recycle_rp_inst(struct kretprobe_instance *ri, void __kprobes kretprobe_hash_lock(struct task_struct *tsk, struct hlist_head **head, unsigned long *flags) +__acquires(hlist_lock) { unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); spinlock_t *hlist_lock; @@ -842,6 +844,7 @@ void __kprobes kretprobe_hash_lock(struct task_struct *tsk, static void __kprobes kretprobe_table_lock(unsigned long hash, unsigned long *flags) +__acquires(hlist_lock) { spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); spin_lock_irqsave(hlist_lock, *flags); @@ -849,6 +852,7 @@ static void __kprobes kretprobe_table_lock(unsigned long hash, void __kprobes kretprobe_hash_unlock(struct task_struct *tsk, unsigned long *flags) +__releases(hlist_lock) { unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); spinlock_t *hlist_lock; @@ -857,7 +861,9 @@ void __kprobes kretprobe_hash_unlock(struct task_struct *tsk, spin_unlock_irqrestore(hlist_lock, *flags); } -void __kprobes kretprobe_table_unlock(unsigned long hash, unsigned long *flags) +static void __kprobes kretprobe_table_unlock(unsigned long hash, + unsigned long *flags) +__releases(hlist_lock) { spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); spin_unlock_irqrestore(hlist_lock, *flags); @@ -1141,7 +1147,8 @@ int __kprobes register_kprobe(struct kprobe *p) preempt_disable(); if (!kernel_text_address((unsigned long) p->addr) || in_kprobes_functions((unsigned long) p->addr) || - ftrace_text_reserved(p->addr, p->addr)) { + ftrace_text_reserved(p->addr, p->addr) || + jump_label_text_reserved(p->addr, p->addr)) { preempt_enable(); return -EINVAL; } @@ -1339,18 +1346,19 @@ int __kprobes register_jprobes(struct jprobe **jps, int num) if (num <= 0) return -EINVAL; for (i = 0; i < num; i++) { - unsigned long addr; + unsigned long addr, offset; jp = jps[i]; addr = arch_deref_entry_point(jp->entry); - if (!kernel_text_address(addr)) - ret = -EINVAL; - else { - /* Todo: Verify probepoint is a function entry point */ + /* Verify probepoint is a function entry point */ + if (kallsyms_lookup_size_offset(addr, NULL, &offset) && + offset == 0) { jp->kp.pre_handler = setjmp_pre_handler; jp->kp.break_handler = longjmp_break_handler; ret = register_kprobe(&jp->kp); - } + } else + ret = -EINVAL; + if (ret < 0) { if (i > 0) unregister_jprobes(jps, i); diff --git a/kernel/module.c b/kernel/module.c index ccd64199184..2df46301a7a 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -55,6 +55,7 @@ #include <linux/async.h> #include <linux/percpu.h> #include <linux/kmemleak.h> +#include <linux/jump_label.h> #define CREATE_TRACE_POINTS #include <trace/events/module.h> @@ -2309,6 +2310,11 @@ static void find_module_sections(struct module *mod, struct load_info *info) sizeof(*mod->tracepoints), &mod->num_tracepoints); #endif +#ifdef HAVE_JUMP_LABEL + mod->jump_entries = section_objs(info, "__jump_table", + sizeof(*mod->jump_entries), + &mod->num_jump_entries); +#endif #ifdef CONFIG_EVENT_TRACING mod->trace_events = section_objs(info, "_ftrace_events", sizeof(*mod->trace_events), diff --git a/kernel/perf_event.c b/kernel/perf_event.c index db5b5606468..64507eaa2d9 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -31,24 +31,18 @@ #include <linux/kernel_stat.h> #include <linux/perf_event.h> #include <linux/ftrace_event.h> -#include <linux/hw_breakpoint.h> #include <asm/irq_regs.h> -/* - * Each CPU has a list of per CPU events: - */ -static DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context); - -int perf_max_events __read_mostly = 1; -static int perf_reserved_percpu __read_mostly; -static int perf_overcommit __read_mostly = 1; - static atomic_t nr_events __read_mostly; static atomic_t nr_mmap_events __read_mostly; static atomic_t nr_comm_events __read_mostly; static atomic_t nr_task_events __read_mostly; +static LIST_HEAD(pmus); +static DEFINE_MUTEX(pmus_lock); +static struct srcu_struct pmus_srcu; + /* * perf event paranoia level: * -1 - not paranoid at all @@ -67,36 +61,38 @@ int sysctl_perf_event_sample_rate __read_mostly = 100000; static atomic64_t perf_event_id; -/* - * Lock for (sysadmin-configurable) event reservations: - */ -static DEFINE_SPINLOCK(perf_resource_lock); +void __weak perf_event_print_debug(void) { } -/* - * Architecture provided APIs - weak aliases: - */ -extern __weak const struct pmu *hw_perf_event_init(struct perf_event *event) +void perf_pmu_disable(struct pmu *pmu) { - return NULL; + int *count = this_cpu_ptr(pmu->pmu_disable_count); + if (!(*count)++) + pmu->pmu_disable(pmu); } -void __weak hw_perf_disable(void) { barrier(); } -void __weak hw_perf_enable(void) { barrier(); } - -void __weak perf_event_print_debug(void) { } - -static DEFINE_PER_CPU(int, perf_disable_count); - -void perf_disable(void) +void perf_pmu_enable(struct pmu *pmu) { - if (!__get_cpu_var(perf_disable_count)++) - hw_perf_disable(); + int *count = this_cpu_ptr(pmu->pmu_disable_count); + if (!--(*count)) + pmu->pmu_enable(pmu); } -void perf_enable(void) +static DEFINE_PER_CPU(struct list_head, rotation_list); + +/* + * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized + * because they're strictly cpu affine and rotate_start is called with IRQs + * disabled, while rotate_context is called from IRQ context. + */ +static void perf_pmu_rotate_start(struct pmu *pmu) { - if (!--__get_cpu_var(perf_disable_count)) - hw_perf_enable(); + struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + struct list_head *head = &__get_cpu_var(rotation_list); + + WARN_ON(!irqs_disabled()); + + if (list_empty(&cpuctx->rotation_list)) + list_add(&cpuctx->rotation_list, head); } static void get_ctx(struct perf_event_context *ctx) @@ -151,13 +147,13 @@ static u64 primary_event_id(struct perf_event *event) * the context could get moved to another task. */ static struct perf_event_context * -perf_lock_task_context(struct task_struct *task, unsigned long *flags) +perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags) { struct perf_event_context *ctx; rcu_read_lock(); - retry: - ctx = rcu_dereference(task->perf_event_ctxp); +retry: + ctx = rcu_dereference(task->perf_event_ctxp[ctxn]); if (ctx) { /* * If this context is a clone of another, it might @@ -170,7 +166,7 @@ perf_lock_task_context(struct task_struct *task, unsigned long *flags) * can't get swapped on us any more. */ raw_spin_lock_irqsave(&ctx->lock, *flags); - if (ctx != rcu_dereference(task->perf_event_ctxp)) { + if (ctx != rcu_dereference(task->perf_event_ctxp[ctxn])) { raw_spin_unlock_irqrestore(&ctx->lock, *flags); goto retry; } @@ -189,12 +185,13 @@ perf_lock_task_context(struct task_struct *task, unsigned long *flags) * can't get swapped to another task. This also increments its * reference count so that the context can't get freed. */ -static struct perf_event_context *perf_pin_task_context(struct task_struct *task) +static struct perf_event_context * +perf_pin_task_context(struct task_struct *task, int ctxn) { struct perf_event_context *ctx; unsigned long flags; - ctx = perf_lock_task_context(task, &flags); + ctx = perf_lock_task_context(task, ctxn, &flags); if (ctx) { ++ctx->pin_count; raw_spin_unlock_irqrestore(&ctx->lock, flags); @@ -302,6 +299,8 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) } list_add_rcu(&event->event_entry, &ctx->event_list); + if (!ctx->nr_events) + perf_pmu_rotate_start(ctx->pmu); ctx->nr_events++; if (event->attr.inherit_stat) ctx->nr_stat++; @@ -436,7 +435,7 @@ event_sched_out(struct perf_event *event, event->state = PERF_EVENT_STATE_OFF; } event->tstamp_stopped = ctx->time; - event->pmu->disable(event); + event->pmu->del(event, 0); event->oncpu = -1; if (!is_software_event(event)) @@ -466,6 +465,12 @@ group_sched_out(struct perf_event *group_event, cpuctx->exclusive = 0; } +static inline struct perf_cpu_context * +__get_cpu_context(struct perf_event_context *ctx) +{ + return this_cpu_ptr(ctx->pmu->pmu_cpu_context); +} + /* * Cross CPU call to remove a performance event * @@ -474,9 +479,9 @@ group_sched_out(struct perf_event *group_event, */ static void __perf_event_remove_from_context(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a task context, we need to check whether it is @@ -487,27 +492,11 @@ static void __perf_event_remove_from_context(void *info) return; raw_spin_lock(&ctx->lock); - /* - * Protect the list operation against NMI by disabling the - * events on a global level. - */ - perf_disable(); event_sched_out(event, cpuctx, ctx); list_del_event(event, ctx); - if (!ctx->task) { - /* - * Allow more per task events with respect to the - * reservation: - */ - cpuctx->max_pertask = - min(perf_max_events - ctx->nr_events, - perf_max_events - perf_reserved_percpu); - } - - perf_enable(); raw_spin_unlock(&ctx->lock); } @@ -572,8 +561,8 @@ retry: static void __perf_event_disable(void *info) { struct perf_event *event = info; - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a per-task event, need to check whether this @@ -628,7 +617,7 @@ void perf_event_disable(struct perf_event *event) return; } - retry: +retry: task_oncpu_function_call(task, __perf_event_disable, event); raw_spin_lock_irq(&ctx->lock); @@ -667,7 +656,7 @@ event_sched_in(struct perf_event *event, */ smp_wmb(); - if (event->pmu->enable(event)) { + if (event->pmu->add(event, PERF_EF_START)) { event->state = PERF_EVENT_STATE_INACTIVE; event->oncpu = -1; return -EAGAIN; @@ -691,22 +680,15 @@ group_sched_in(struct perf_event *group_event, struct perf_event_context *ctx) { struct perf_event *event, *partial_group = NULL; - const struct pmu *pmu = group_event->pmu; - bool txn = false; + struct pmu *pmu = group_event->pmu; if (group_event->state == PERF_EVENT_STATE_OFF) return 0; - /* Check if group transaction availabe */ - if (pmu->start_txn) - txn = true; - - if (txn) - pmu->start_txn(pmu); + pmu->start_txn(pmu); if (event_sched_in(group_event, cpuctx, ctx)) { - if (txn) - pmu->cancel_txn(pmu); + pmu->cancel_txn(pmu); return -EAGAIN; } @@ -720,7 +702,7 @@ group_sched_in(struct perf_event *group_event, } } - if (!txn || !pmu->commit_txn(pmu)) + if (!pmu->commit_txn(pmu)) return 0; group_error: @@ -735,8 +717,7 @@ group_error: } event_sched_out(group_event, cpuctx, ctx); - if (txn) - pmu->cancel_txn(pmu); + pmu->cancel_txn(pmu); return -EAGAIN; } @@ -789,10 +770,10 @@ static void add_event_to_ctx(struct perf_event *event, */ static void __perf_install_in_context(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; struct perf_event *leader = event->group_leader; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); int err; /* @@ -812,12 +793,6 @@ static void __perf_install_in_context(void *info) ctx->is_active = 1; update_context_time(ctx); - /* - * Protect the list operation against NMI by disabling the - * events on a global level. NOP for non NMI based events. - */ - perf_disable(); - add_event_to_ctx(event, ctx); if (event->cpu != -1 && event->cpu != smp_processor_id()) @@ -855,12 +830,7 @@ static void __perf_install_in_context(void *info) } } - if (!err && !ctx->task && cpuctx->max_pertask) - cpuctx->max_pertask--; - - unlock: - perf_enable(); - +unlock: raw_spin_unlock(&ctx->lock); } @@ -883,6 +853,8 @@ perf_install_in_context(struct perf_event_context *ctx, { struct task_struct *task = ctx->task; + event->ctx = ctx; + if (!task) { /* * Per cpu events are installed via an smp call and @@ -931,10 +903,12 @@ static void __perf_event_mark_enabled(struct perf_event *event, event->state = PERF_EVENT_STATE_INACTIVE; event->tstamp_enabled = ctx->time - event->total_time_enabled; - list_for_each_entry(sub, &event->sibling_list, group_entry) - if (sub->state >= PERF_EVENT_STATE_INACTIVE) + list_for_each_entry(sub, &event->sibling_list, group_entry) { + if (sub->state >= PERF_EVENT_STATE_INACTIVE) { sub->tstamp_enabled = ctx->time - sub->total_time_enabled; + } + } } /* @@ -943,9 +917,9 @@ static void __perf_event_mark_enabled(struct perf_event *event, static void __perf_event_enable(void *info) { struct perf_event *event = info; - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = event->ctx; struct perf_event *leader = event->group_leader; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); int err; /* @@ -979,12 +953,10 @@ static void __perf_event_enable(void *info) if (!group_can_go_on(event, cpuctx, 1)) { err = -EEXIST; } else { - perf_disable(); if (event == leader) err = group_sched_in(event, cpuctx, ctx); else err = event_sched_in(event, cpuctx, ctx); - perf_enable(); } if (err) { @@ -1000,7 +972,7 @@ static void __perf_event_enable(void *info) } } - unlock: +unlock: raw_spin_unlock(&ctx->lock); } @@ -1041,7 +1013,7 @@ void perf_event_enable(struct perf_event *event) if (event->state == PERF_EVENT_STATE_ERROR) event->state = PERF_EVENT_STATE_OFF; - retry: +retry: raw_spin_unlock_irq(&ctx->lock); task_oncpu_function_call(task, __perf_event_enable, event); @@ -1061,7 +1033,7 @@ void perf_event_enable(struct perf_event *event) if (event->state == PERF_EVENT_STATE_OFF) __perf_event_mark_enabled(event, ctx); - out: +out: raw_spin_unlock_irq(&ctx->lock); } @@ -1092,26 +1064,26 @@ static void ctx_sched_out(struct perf_event_context *ctx, struct perf_event *event; raw_spin_lock(&ctx->lock); + perf_pmu_disable(ctx->pmu); ctx->is_active = 0; if (likely(!ctx->nr_events)) goto out; update_context_time(ctx); - perf_disable(); if (!ctx->nr_active) - goto out_enable; + goto out; - if (event_type & EVENT_PINNED) + if (event_type & EVENT_PINNED) { list_for_each_entry(event, &ctx->pinned_groups, group_entry) group_sched_out(event, cpuctx, ctx); + } - if (event_type & EVENT_FLEXIBLE) + if (event_type & EVENT_FLEXIBLE) { list_for_each_entry(event, &ctx->flexible_groups, group_entry) group_sched_out(event, cpuctx, ctx); - - out_enable: - perf_enable(); - out: + } +out: + perf_pmu_enable(ctx->pmu); raw_spin_unlock(&ctx->lock); } @@ -1209,34 +1181,25 @@ static void perf_event_sync_stat(struct perf_event_context *ctx, } } -/* - * Called from scheduler to remove the events of the current task, - * with interrupts disabled. - * - * We stop each event and update the event value in event->count. - * - * This does not protect us against NMI, but disable() - * sets the disabled bit in the control field of event _before_ - * accessing the event control register. If a NMI hits, then it will - * not restart the event. - */ -void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next) +void perf_event_context_sched_out(struct task_struct *task, int ctxn, + struct task_struct *next) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_event_context *ctx = task->perf_event_ctxp[ctxn]; struct perf_event_context *next_ctx; struct perf_event_context *parent; + struct perf_cpu_context *cpuctx; int do_switch = 1; - perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); + if (likely(!ctx)) + return; - if (likely(!ctx || !cpuctx->task_ctx)) + cpuctx = __get_cpu_context(ctx); + if (!cpuctx->task_ctx) return; rcu_read_lock(); parent = rcu_dereference(ctx->parent_ctx); - next_ctx = next->perf_event_ctxp; + next_ctx = next->perf_event_ctxp[ctxn]; if (parent && next_ctx && rcu_dereference(next_ctx->parent_ctx) == parent) { /* @@ -1255,8 +1218,8 @@ void perf_event_task_sched_out(struct task_struct *task, * XXX do we need a memory barrier of sorts * wrt to rcu_dereference() of perf_event_ctxp */ - task->perf_event_ctxp = next_ctx; - next->perf_event_ctxp = ctx; + task->perf_event_ctxp[ctxn] = next_ctx; + next->perf_event_ctxp[ctxn] = ctx; ctx->task = next; next_ctx->task = task; do_switch = 0; @@ -1274,10 +1237,35 @@ void perf_event_task_sched_out(struct task_struct *task, } } +#define for_each_task_context_nr(ctxn) \ + for ((ctxn) = 0; (ctxn) < perf_nr_task_contexts; (ctxn)++) + +/* + * Called from scheduler to remove the events of the current task, + * with interrupts disabled. + * + * We stop each event and update the event value in event->count. + * + * This does not protect us against NMI, but disable() + * sets the disabled bit in the control field of event _before_ + * accessing the event control register. If a NMI hits, then it will + * not restart the event. + */ +void perf_event_task_sched_out(struct task_struct *task, + struct task_struct *next) +{ + int ctxn; + + perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); + + for_each_task_context_nr(ctxn) + perf_event_context_sched_out(task, ctxn, next); +} + static void task_ctx_sched_out(struct perf_event_context *ctx, enum event_type_t event_type) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); if (!cpuctx->task_ctx) return; @@ -1350,9 +1338,10 @@ ctx_flexible_sched_in(struct perf_event_context *ctx, if (event->cpu != -1 && event->cpu != smp_processor_id()) continue; - if (group_can_go_on(event, cpuctx, can_add_hw)) + if (group_can_go_on(event, cpuctx, can_add_hw)) { if (group_sched_in(event, cpuctx, ctx)) can_add_hw = 0; + } } } @@ -1368,8 +1357,6 @@ ctx_sched_in(struct perf_event_context *ctx, ctx->timestamp = perf_clock(); - perf_disable(); - /* * First go through the list and put on any pinned groups * in order to give them the best chance of going on. @@ -1381,8 +1368,7 @@ ctx_sched_in(struct perf_event_context *ctx, if (event_type & EVENT_FLEXIBLE) ctx_flexible_sched_in(ctx, cpuctx); - perf_enable(); - out: +out: raw_spin_unlock(&ctx->lock); } @@ -1394,43 +1380,28 @@ static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx, ctx_sched_in(ctx, cpuctx, event_type); } -static void task_ctx_sched_in(struct task_struct *task, +static void task_ctx_sched_in(struct perf_event_context *ctx, enum event_type_t event_type) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_cpu_context *cpuctx; - if (likely(!ctx)) - return; + cpuctx = __get_cpu_context(ctx); if (cpuctx->task_ctx == ctx) return; + ctx_sched_in(ctx, cpuctx, event_type); cpuctx->task_ctx = ctx; } -/* - * Called from scheduler to add the events of the current task - * with interrupts disabled. - * - * We restore the event value and then enable it. - * - * This does not protect us against NMI, but enable() - * sets the enabled bit in the control field of event _before_ - * accessing the event control register. If a NMI hits, then it will - * keep the event running. - */ -void perf_event_task_sched_in(struct task_struct *task) -{ - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - struct perf_event_context *ctx = task->perf_event_ctxp; - if (likely(!ctx)) - return; +void perf_event_context_sched_in(struct perf_event_context *ctx) +{ + struct perf_cpu_context *cpuctx; + cpuctx = __get_cpu_context(ctx); if (cpuctx->task_ctx == ctx) return; - perf_disable(); - + perf_pmu_disable(ctx->pmu); /* * We want to keep the following priority order: * cpu pinned (that don't need to move), task pinned, @@ -1444,7 +1415,37 @@ void perf_event_task_sched_in(struct task_struct *task) cpuctx->task_ctx = ctx; - perf_enable(); + /* + * Since these rotations are per-cpu, we need to ensure the + * cpu-context we got scheduled on is actually rotating. + */ + perf_pmu_rotate_start(ctx->pmu); + perf_pmu_enable(ctx->pmu); +} + +/* + * Called from scheduler to add the events of the current task + * with interrupts disabled. + * + * We restore the event value and then enable it. + * + * This does not protect us against NMI, but enable() + * sets the enabled bit in the control field of event _before_ + * accessing the event control register. If a NMI hits, then it will + * keep the event running. + */ +void perf_event_task_sched_in(struct task_struct *task) +{ + struct perf_event_context *ctx; + int ctxn; + + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (likely(!ctx)) + continue; + + perf_event_context_sched_in(ctx); + } } #define MAX_INTERRUPTS (~0ULL) @@ -1524,22 +1525,6 @@ do { \ return div64_u64(dividend, divisor); } -static void perf_event_stop(struct perf_event *event) -{ - if (!event->pmu->stop) - return event->pmu->disable(event); - - return event->pmu->stop(event); -} - -static int perf_event_start(struct perf_event *event) -{ - if (!event->pmu->start) - return event->pmu->enable(event); - - return event->pmu->start(event); -} - static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) { struct hw_perf_event *hwc = &event->hw; @@ -1559,15 +1544,13 @@ static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) hwc->sample_period = sample_period; if (local64_read(&hwc->period_left) > 8*sample_period) { - perf_disable(); - perf_event_stop(event); + event->pmu->stop(event, PERF_EF_UPDATE); local64_set(&hwc->period_left, 0); - perf_event_start(event); - perf_enable(); + event->pmu->start(event, PERF_EF_RELOAD); } } -static void perf_ctx_adjust_freq(struct perf_event_context *ctx) +static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period) { struct perf_event *event; struct hw_perf_event *hwc; @@ -1592,23 +1575,19 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) */ if (interrupts == MAX_INTERRUPTS) { perf_log_throttle(event, 1); - perf_disable(); - event->pmu->unthrottle(event); - perf_enable(); + event->pmu->start(event, 0); } if (!event->attr.freq || !event->attr.sample_freq) continue; - perf_disable(); event->pmu->read(event); now = local64_read(&event->count); delta = now - hwc->freq_count_stamp; hwc->freq_count_stamp = now; if (delta > 0) - perf_adjust_period(event, TICK_NSEC, delta); - perf_enable(); + perf_adjust_period(event, period, delta); } raw_spin_unlock(&ctx->lock); } @@ -1626,32 +1605,38 @@ static void rotate_ctx(struct perf_event_context *ctx) raw_spin_unlock(&ctx->lock); } -void perf_event_task_tick(struct task_struct *curr) +/* + * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized + * because they're strictly cpu affine and rotate_start is called with IRQs + * disabled, while rotate_context is called from IRQ context. + */ +static void perf_rotate_context(struct perf_cpu_context *cpuctx) { - struct perf_cpu_context *cpuctx; - struct perf_event_context *ctx; - int rotate = 0; + u64 interval = (u64)cpuctx->jiffies_interval * TICK_NSEC; + struct perf_event_context *ctx = NULL; + int rotate = 0, remove = 1; - if (!atomic_read(&nr_events)) - return; - - cpuctx = &__get_cpu_var(perf_cpu_context); - if (cpuctx->ctx.nr_events && - cpuctx->ctx.nr_events != cpuctx->ctx.nr_active) - rotate = 1; + if (cpuctx->ctx.nr_events) { + remove = 0; + if (cpuctx->ctx.nr_events != cpuctx->ctx.nr_active) + rotate = 1; + } - ctx = curr->perf_event_ctxp; - if (ctx && ctx->nr_events && ctx->nr_events != ctx->nr_active) - rotate = 1; + ctx = cpuctx->task_ctx; + if (ctx && ctx->nr_events) { + remove = 0; + if (ctx->nr_events != ctx->nr_active) + rotate = 1; + } - perf_ctx_adjust_freq(&cpuctx->ctx); + perf_pmu_disable(cpuctx->ctx.pmu); + perf_ctx_adjust_freq(&cpuctx->ctx, interval); if (ctx) - perf_ctx_adjust_freq(ctx); + perf_ctx_adjust_freq(ctx, interval); if (!rotate) - return; + goto done; - perf_disable(); cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE); if (ctx) task_ctx_sched_out(ctx, EVENT_FLEXIBLE); @@ -1662,8 +1647,27 @@ void perf_event_task_tick(struct task_struct *curr) cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE); if (ctx) - task_ctx_sched_in(curr, EVENT_FLEXIBLE); - perf_enable(); + task_ctx_sched_in(ctx, EVENT_FLEXIBLE); + +done: + if (remove) + list_del_init(&cpuctx->rotation_list); + + perf_pmu_enable(cpuctx->ctx.pmu); +} + +void perf_event_task_tick(void) +{ + struct list_head *head = &__get_cpu_var(rotation_list); + struct perf_cpu_context *cpuctx, *tmp; + + WARN_ON(!irqs_disabled()); + + list_for_each_entry_safe(cpuctx, tmp, head, rotation_list) { + if (cpuctx->jiffies_interval == 1 || + !(jiffies % cpuctx->jiffies_interval)) + perf_rotate_context(cpuctx); + } } static int event_enable_on_exec(struct perf_event *event, @@ -1685,20 +1689,18 @@ static int event_enable_on_exec(struct perf_event *event, * Enable all of a task's events that have been marked enable-on-exec. * This expects task == current. */ -static void perf_event_enable_on_exec(struct task_struct *task) +static void perf_event_enable_on_exec(struct perf_event_context *ctx) { - struct perf_event_context *ctx; struct perf_event *event; unsigned long flags; int enabled = 0; int ret; local_irq_save(flags); - ctx = task->perf_event_ctxp; if (!ctx || !ctx->nr_events) goto out; - __perf_event_task_sched_out(ctx); + task_ctx_sched_out(ctx, EVENT_ALL); raw_spin_lock(&ctx->lock); @@ -1722,8 +1724,8 @@ static void perf_event_enable_on_exec(struct task_struct *task) raw_spin_unlock(&ctx->lock); - perf_event_task_sched_in(task); - out: + perf_event_context_sched_in(ctx); +out: local_irq_restore(flags); } @@ -1732,9 +1734,9 @@ static void perf_event_enable_on_exec(struct task_struct *task) */ static void __perf_event_read(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a task context, we need to check whether it is @@ -1782,11 +1784,219 @@ static u64 perf_event_read(struct perf_event *event) } /* - * Initialize the perf_event context in a task_struct: + * Callchain support */ + +struct callchain_cpus_entries { + struct rcu_head rcu_head; + struct perf_callchain_entry *cpu_entries[0]; +}; + +static DEFINE_PER_CPU(int, callchain_recursion[PERF_NR_CONTEXTS]); +static atomic_t nr_callchain_events; +static DEFINE_MUTEX(callchain_mutex); +struct callchain_cpus_entries *callchain_cpus_entries; + + +__weak void perf_callchain_kernel(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +__weak void perf_callchain_user(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +static void release_callchain_buffers_rcu(struct rcu_head *head) +{ + struct callchain_cpus_entries *entries; + int cpu; + + entries = container_of(head, struct callchain_cpus_entries, rcu_head); + + for_each_possible_cpu(cpu) + kfree(entries->cpu_entries[cpu]); + + kfree(entries); +} + +static void release_callchain_buffers(void) +{ + struct callchain_cpus_entries *entries; + + entries = callchain_cpus_entries; + rcu_assign_pointer(callchain_cpus_entries, NULL); + call_rcu(&entries->rcu_head, release_callchain_buffers_rcu); +} + +static int alloc_callchain_buffers(void) +{ + int cpu; + int size; + struct callchain_cpus_entries *entries; + + /* + * We can't use the percpu allocation API for data that can be + * accessed from NMI. Use a temporary manual per cpu allocation + * until that gets sorted out. + */ + size = sizeof(*entries) + sizeof(struct perf_callchain_entry *) * + num_possible_cpus(); + + entries = kzalloc(size, GFP_KERNEL); + if (!entries) + return -ENOMEM; + + size = sizeof(struct perf_callchain_entry) * PERF_NR_CONTEXTS; + + for_each_possible_cpu(cpu) { + entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL, + cpu_to_node(cpu)); + if (!entries->cpu_entries[cpu]) + goto fail; + } + + rcu_assign_pointer(callchain_cpus_entries, entries); + + return 0; + +fail: + for_each_possible_cpu(cpu) + kfree(entries->cpu_entries[cpu]); + kfree(entries); + + return -ENOMEM; +} + +static int get_callchain_buffers(void) +{ + int err = 0; + int count; + + mutex_lock(&callchain_mutex); + + count = atomic_inc_return(&nr_callchain_events); + if (WARN_ON_ONCE(count < 1)) { + err = -EINVAL; + goto exit; + } + + if (count > 1) { + /* If the allocation failed, give up */ + if (!callchain_cpus_entries) + err = -ENOMEM; + goto exit; + } + + err = alloc_callchain_buffers(); + if (err) + release_callchain_buffers(); +exit: + mutex_unlock(&callchain_mutex); + + return err; +} + +static void put_callchain_buffers(void) +{ + if (atomic_dec_and_mutex_lock(&nr_callchain_events, &callchain_mutex)) { + release_callchain_buffers(); + mutex_unlock(&callchain_mutex); + } +} + +static int get_recursion_context(int *recursion) +{ + int rctx; + + if (in_nmi()) + rctx = 3; + else if (in_irq()) + rctx = 2; + else if (in_softirq()) + rctx = 1; + else + rctx = 0; + + if (recursion[rctx]) + return -1; + + recursion[rctx]++; + barrier(); + + return rctx; +} + +static inline void put_recursion_context(int *recursion, int rctx) +{ + barrier(); + recursion[rctx]--; +} + +static struct perf_callchain_entry *get_callchain_entry(int *rctx) +{ + int cpu; + struct callchain_cpus_entries *entries; + + *rctx = get_recursion_context(__get_cpu_var(callchain_recursion)); + if (*rctx == -1) + return NULL; + + entries = rcu_dereference(callchain_cpus_entries); + if (!entries) + return NULL; + + cpu = smp_processor_id(); + + return &entries->cpu_entries[cpu][*rctx]; +} + static void -__perf_event_init_context(struct perf_event_context *ctx, - struct task_struct *task) +put_callchain_entry(int rctx) +{ + put_recursion_context(__get_cpu_var(callchain_recursion), rctx); +} + +static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) +{ + int rctx; + struct perf_callchain_entry *entry; + + + entry = get_callchain_entry(&rctx); + if (rctx == -1) + return NULL; + + if (!entry) + goto exit_put; + + entry->nr = 0; + + if (!user_mode(regs)) { + perf_callchain_store(entry, PERF_CONTEXT_KERNEL); + perf_callchain_kernel(entry, regs); + if (current->mm) + regs = task_pt_regs(current); + else + regs = NULL; + } + + if (regs) { + perf_callchain_store(entry, PERF_CONTEXT_USER); + perf_callchain_user(entry, regs); + } + +exit_put: + put_callchain_entry(rctx); + + return entry; +} + +/* + * Initialize the perf_event context in a task_struct: + */ +static void __perf_event_init_context(struct perf_event_context *ctx) { raw_spin_lock_init(&ctx->lock); mutex_init(&ctx->mutex); @@ -1794,45 +2004,38 @@ __perf_event_init_context(struct perf_event_context *ctx, INIT_LIST_HEAD(&ctx->flexible_groups); INIT_LIST_HEAD(&ctx->event_list); atomic_set(&ctx->refcount, 1); - ctx->task = task; } -static struct perf_event_context *find_get_context(pid_t pid, int cpu) +static struct perf_event_context * +alloc_perf_context(struct pmu *pmu, struct task_struct *task) { struct perf_event_context *ctx; - struct perf_cpu_context *cpuctx; - struct task_struct *task; - unsigned long flags; - int err; - - if (pid == -1 && cpu != -1) { - /* Must be root to operate on a CPU event: */ - if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN)) - return ERR_PTR(-EACCES); - if (cpu < 0 || cpu >= nr_cpumask_bits) - return ERR_PTR(-EINVAL); + ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL); + if (!ctx) + return NULL; - /* - * We could be clever and allow to attach a event to an - * offline CPU and activate it when the CPU comes up, but - * that's for later. - */ - if (!cpu_online(cpu)) - return ERR_PTR(-ENODEV); + __perf_event_init_context(ctx); + if (task) { + ctx->task = task; + get_task_struct(task); + } + ctx->pmu = pmu; - cpuctx = &per_cpu(perf_cpu_context, cpu); - ctx = &cpuctx->ctx; - get_ctx(ctx); + return ctx; +} - return ctx; - } +static struct task_struct * +find_lively_task_by_vpid(pid_t vpid) +{ + struct task_struct *task; + int err; rcu_read_lock(); - if (!pid) + if (!vpid) task = current; else - task = find_task_by_vpid(pid); + task = find_task_by_vpid(vpid); if (task) get_task_struct(task); rcu_read_unlock(); @@ -1852,35 +2055,79 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu) if (!ptrace_may_access(task, PTRACE_MODE_READ)) goto errout; - retry: - ctx = perf_lock_task_context(task, &flags); + return task; +errout: + put_task_struct(task); + return ERR_PTR(err); + +} + +static struct perf_event_context * +find_get_context(struct pmu *pmu, struct task_struct *task, int cpu) +{ + struct perf_event_context *ctx; + struct perf_cpu_context *cpuctx; + unsigned long flags; + int ctxn, err; + + if (!task && cpu != -1) { + /* Must be root to operate on a CPU event: */ + if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN)) + return ERR_PTR(-EACCES); + + if (cpu < 0 || cpu >= nr_cpumask_bits) + return ERR_PTR(-EINVAL); + + /* + * We could be clever and allow to attach a event to an + * offline CPU and activate it when the CPU comes up, but + * that's for later. + */ + if (!cpu_online(cpu)) + return ERR_PTR(-ENODEV); + + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); + ctx = &cpuctx->ctx; + get_ctx(ctx); + + return ctx; + } + + err = -EINVAL; + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + goto errout; + +retry: + ctx = perf_lock_task_context(task, ctxn, &flags); if (ctx) { unclone_ctx(ctx); raw_spin_unlock_irqrestore(&ctx->lock, flags); } if (!ctx) { - ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL); + ctx = alloc_perf_context(pmu, task); err = -ENOMEM; if (!ctx) goto errout; - __perf_event_init_context(ctx, task); + get_ctx(ctx); - if (cmpxchg(&task->perf_event_ctxp, NULL, ctx)) { + + if (cmpxchg(&task->perf_event_ctxp[ctxn], NULL, ctx)) { /* * We raced with some other task; use * the context they set. */ + put_task_struct(task); kfree(ctx); goto retry; } - get_task_struct(task); } put_task_struct(task); return ctx; - errout: +errout: put_task_struct(task); return ERR_PTR(err); } @@ -1913,6 +2160,8 @@ static void free_event(struct perf_event *event) atomic_dec(&nr_comm_events); if (event->attr.task) atomic_dec(&nr_task_events); + if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) + put_callchain_buffers(); } if (event->buffer) { @@ -1923,7 +2172,9 @@ static void free_event(struct perf_event *event) if (event->destroy) event->destroy(event); - put_ctx(event->ctx); + if (event->ctx) + put_ctx(event->ctx); + call_rcu(&event->rcu_head, free_event_rcu); } @@ -2344,6 +2595,9 @@ int perf_event_task_disable(void) static int perf_event_index(struct perf_event *event) { + if (event->hw.state & PERF_HES_STOPPED) + return 0; + if (event->state != PERF_EVENT_STATE_ACTIVE) return 0; @@ -2956,16 +3210,6 @@ void perf_event_do_pending(void) } /* - * Callchain support -- arch specific - */ - -__weak struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) -{ - return NULL; -} - - -/* * We assume there is only KVM supporting the callbacks. * Later on, we might change it to a list if there is * another virtualization implementation supporting the callbacks. @@ -3071,7 +3315,7 @@ again: if (handle->wakeup != local_read(&buffer->wakeup)) perf_output_wakeup(handle); - out: +out: preempt_enable(); } @@ -3459,14 +3703,20 @@ static void perf_event_output(struct perf_event *event, int nmi, struct perf_output_handle handle; struct perf_event_header header; + /* protect the callchain buffers */ + rcu_read_lock(); + perf_prepare_sample(&header, data, event, regs); if (perf_output_begin(&handle, event, header.size, nmi, 1)) - return; + goto exit; perf_output_sample(&handle, &header, data, event); perf_output_end(&handle); + +exit: + rcu_read_unlock(); } /* @@ -3580,16 +3830,27 @@ static void perf_event_task_ctx(struct perf_event_context *ctx, static void perf_event_task_event(struct perf_task_event *task_event) { struct perf_cpu_context *cpuctx; - struct perf_event_context *ctx = task_event->task_ctx; + struct perf_event_context *ctx; + struct pmu *pmu; + int ctxn; rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_task_ctx(&cpuctx->ctx, task_event); - if (!ctx) - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_task_ctx(ctx, task_event); - put_cpu_var(perf_cpu_context); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + perf_event_task_ctx(&cpuctx->ctx, task_event); + + ctx = task_event->task_ctx; + if (!ctx) { + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + goto next; + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + } + if (ctx) + perf_event_task_ctx(ctx, task_event); +next: + put_cpu_ptr(pmu->pmu_cpu_context); + } rcu_read_unlock(); } @@ -3694,8 +3955,10 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) { struct perf_cpu_context *cpuctx; struct perf_event_context *ctx; - unsigned int size; char comm[TASK_COMM_LEN]; + unsigned int size; + struct pmu *pmu; + int ctxn; memset(comm, 0, sizeof(comm)); strlcpy(comm, comm_event->task->comm, sizeof(comm)); @@ -3707,21 +3970,36 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) comm_event->event_id.header.size = sizeof(comm_event->event_id) + size; rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_comm_ctx(&cpuctx->ctx, comm_event); - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_comm_ctx(ctx, comm_event); - put_cpu_var(perf_cpu_context); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + perf_event_comm_ctx(&cpuctx->ctx, comm_event); + + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + goto next; + + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + if (ctx) + perf_event_comm_ctx(ctx, comm_event); +next: + put_cpu_ptr(pmu->pmu_cpu_context); + } rcu_read_unlock(); } void perf_event_comm(struct task_struct *task) { struct perf_comm_event comm_event; + struct perf_event_context *ctx; + int ctxn; - if (task->perf_event_ctxp) - perf_event_enable_on_exec(task); + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (!ctx) + continue; + + perf_event_enable_on_exec(ctx); + } if (!atomic_read(&nr_comm_events)) return; @@ -3823,6 +4101,8 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event) char tmp[16]; char *buf = NULL; const char *name; + struct pmu *pmu; + int ctxn; memset(tmp, 0, sizeof(tmp)); @@ -3875,12 +4155,23 @@ got_name: mmap_event->event_id.header.size = sizeof(mmap_event->event_id) + size; rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC); - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_mmap_ctx(ctx, mmap_event, vma->vm_flags & VM_EXEC); - put_cpu_var(perf_cpu_context); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, + vma->vm_flags & VM_EXEC); + + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + goto next; + + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + if (ctx) { + perf_event_mmap_ctx(ctx, mmap_event, + vma->vm_flags & VM_EXEC); + } +next: + put_cpu_ptr(pmu->pmu_cpu_context); + } rcu_read_unlock(); kfree(buf); @@ -3962,8 +4253,6 @@ static int __perf_event_overflow(struct perf_event *event, int nmi, struct hw_perf_event *hwc = &event->hw; int ret = 0; - throttle = (throttle && event->pmu->unthrottle != NULL); - if (!throttle) { hwc->interrupts++; } else { @@ -4031,6 +4320,17 @@ int perf_event_overflow(struct perf_event *event, int nmi, * Generic software event infrastructure */ +struct swevent_htable { + struct swevent_hlist *swevent_hlist; + struct mutex hlist_mutex; + int hlist_refcount; + + /* Recursion avoidance in each contexts */ + int recursion[PERF_NR_CONTEXTS]; +}; + +static DEFINE_PER_CPU(struct swevent_htable, swevent_htable); + /* * We directly increment event->count and keep a second value in * event->hw.period_left to count intervals. This period event @@ -4088,7 +4388,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow, } } -static void perf_swevent_add(struct perf_event *event, u64 nr, +static void perf_swevent_event(struct perf_event *event, u64 nr, int nmi, struct perf_sample_data *data, struct pt_regs *regs) { @@ -4114,6 +4414,9 @@ static void perf_swevent_add(struct perf_event *event, u64 nr, static int perf_exclude_event(struct perf_event *event, struct pt_regs *regs) { + if (event->hw.state & PERF_HES_STOPPED) + return 0; + if (regs) { if (event->attr.exclude_user && user_mode(regs)) return 1; @@ -4160,11 +4463,11 @@ __find_swevent_head(struct swevent_hlist *hlist, u64 type, u32 event_id) /* For the read side: events when they trigger */ static inline struct hlist_head * -find_swevent_head_rcu(struct perf_cpu_context *ctx, u64 type, u32 event_id) +find_swevent_head_rcu(struct swevent_htable *swhash, u64 type, u32 event_id) { struct swevent_hlist *hlist; - hlist = rcu_dereference(ctx->swevent_hlist); + hlist = rcu_dereference(swhash->swevent_hlist); if (!hlist) return NULL; @@ -4173,7 +4476,7 @@ find_swevent_head_rcu(struct perf_cpu_context *ctx, u64 type, u32 event_id) /* For the event head insertion and removal in the hlist */ static inline struct hlist_head * -find_swevent_head(struct perf_cpu_context *ctx, struct perf_event *event) +find_swevent_head(struct swevent_htable *swhash, struct perf_event *event) { struct swevent_hlist *hlist; u32 event_id = event->attr.config; @@ -4184,7 +4487,7 @@ find_swevent_head(struct perf_cpu_context *ctx, struct perf_event *event) * and release. Which makes the protected version suitable here. * The context lock guarantees that. */ - hlist = rcu_dereference_protected(ctx->swevent_hlist, + hlist = rcu_dereference_protected(swhash->swevent_hlist, lockdep_is_held(&event->ctx->lock)); if (!hlist) return NULL; @@ -4197,23 +4500,19 @@ static void do_perf_sw_event(enum perf_type_id type, u32 event_id, struct perf_sample_data *data, struct pt_regs *regs) { - struct perf_cpu_context *cpuctx; + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); struct perf_event *event; struct hlist_node *node; struct hlist_head *head; - cpuctx = &__get_cpu_var(perf_cpu_context); - rcu_read_lock(); - - head = find_swevent_head_rcu(cpuctx, type, event_id); - + head = find_swevent_head_rcu(swhash, type, event_id); if (!head) goto end; hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_swevent_match(event, type, event_id, data, regs)) - perf_swevent_add(event, nr, nmi, data, regs); + perf_swevent_event(event, nr, nmi, data, regs); } end: rcu_read_unlock(); @@ -4221,33 +4520,17 @@ end: int perf_swevent_get_recursion_context(void) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - int rctx; - - if (in_nmi()) - rctx = 3; - else if (in_irq()) - rctx = 2; - else if (in_softirq()) - rctx = 1; - else - rctx = 0; + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); - if (cpuctx->recursion[rctx]) - return -1; - - cpuctx->recursion[rctx]++; - barrier(); - - return rctx; + return get_recursion_context(swhash->recursion); } EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context); void inline perf_swevent_put_recursion_context(int rctx) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - barrier(); - cpuctx->recursion[rctx]--; + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); + + put_recursion_context(swhash->recursion, rctx); } void __perf_sw_event(u32 event_id, u64 nr, int nmi, @@ -4273,20 +4556,20 @@ static void perf_swevent_read(struct perf_event *event) { } -static int perf_swevent_enable(struct perf_event *event) +static int perf_swevent_add(struct perf_event *event, int flags) { + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); struct hw_perf_event *hwc = &event->hw; - struct perf_cpu_context *cpuctx; struct hlist_head *head; - cpuctx = &__get_cpu_var(perf_cpu_context); - if (hwc->sample_period) { hwc->last_period = hwc->sample_period; perf_swevent_set_period(event); } - head = find_swevent_head(cpuctx, event); + hwc->state = !(flags & PERF_EF_START); + + head = find_swevent_head(swhash, event); if (WARN_ON_ONCE(!head)) return -EINVAL; @@ -4295,202 +4578,27 @@ static int perf_swevent_enable(struct perf_event *event) return 0; } -static void perf_swevent_disable(struct perf_event *event) +static void perf_swevent_del(struct perf_event *event, int flags) { hlist_del_rcu(&event->hlist_entry); } -static void perf_swevent_void(struct perf_event *event) -{ -} - -static int perf_swevent_int(struct perf_event *event) -{ - return 0; -} - -static const struct pmu perf_ops_generic = { - .enable = perf_swevent_enable, - .disable = perf_swevent_disable, - .start = perf_swevent_int, - .stop = perf_swevent_void, - .read = perf_swevent_read, - .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */ -}; - -/* - * hrtimer based swevent callback - */ - -static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) -{ - enum hrtimer_restart ret = HRTIMER_RESTART; - struct perf_sample_data data; - struct pt_regs *regs; - struct perf_event *event; - u64 period; - - event = container_of(hrtimer, struct perf_event, hw.hrtimer); - event->pmu->read(event); - - perf_sample_data_init(&data, 0); - data.period = event->hw.last_period; - regs = get_irq_regs(); - - if (regs && !perf_exclude_event(event, regs)) { - if (!(event->attr.exclude_idle && current->pid == 0)) - if (perf_event_overflow(event, 0, &data, regs)) - ret = HRTIMER_NORESTART; - } - - period = max_t(u64, 10000, event->hw.sample_period); - hrtimer_forward_now(hrtimer, ns_to_ktime(period)); - - return ret; -} - -static void perf_swevent_start_hrtimer(struct perf_event *event) +static void perf_swevent_start(struct perf_event *event, int flags) { - struct hw_perf_event *hwc = &event->hw; - - hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); - hwc->hrtimer.function = perf_swevent_hrtimer; - if (hwc->sample_period) { - u64 period; - - if (hwc->remaining) { - if (hwc->remaining < 0) - period = 10000; - else - period = hwc->remaining; - hwc->remaining = 0; - } else { - period = max_t(u64, 10000, hwc->sample_period); - } - __hrtimer_start_range_ns(&hwc->hrtimer, - ns_to_ktime(period), 0, - HRTIMER_MODE_REL, 0); - } + event->hw.state = 0; } -static void perf_swevent_cancel_hrtimer(struct perf_event *event) +static void perf_swevent_stop(struct perf_event *event, int flags) { - struct hw_perf_event *hwc = &event->hw; - - if (hwc->sample_period) { - ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer); - hwc->remaining = ktime_to_ns(remaining); - - hrtimer_cancel(&hwc->hrtimer); - } + event->hw.state = PERF_HES_STOPPED; } -/* - * Software event: cpu wall time clock - */ - -static void cpu_clock_perf_event_update(struct perf_event *event) -{ - int cpu = raw_smp_processor_id(); - s64 prev; - u64 now; - - now = cpu_clock(cpu); - prev = local64_xchg(&event->hw.prev_count, now); - local64_add(now - prev, &event->count); -} - -static int cpu_clock_perf_event_enable(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - int cpu = raw_smp_processor_id(); - - local64_set(&hwc->prev_count, cpu_clock(cpu)); - perf_swevent_start_hrtimer(event); - - return 0; -} - -static void cpu_clock_perf_event_disable(struct perf_event *event) -{ - perf_swevent_cancel_hrtimer(event); - cpu_clock_perf_event_update(event); -} - -static void cpu_clock_perf_event_read(struct perf_event *event) -{ - cpu_clock_perf_event_update(event); -} - -static const struct pmu perf_ops_cpu_clock = { - .enable = cpu_clock_perf_event_enable, - .disable = cpu_clock_perf_event_disable, - .read = cpu_clock_perf_event_read, -}; - -/* - * Software event: task time clock - */ - -static void task_clock_perf_event_update(struct perf_event *event, u64 now) -{ - u64 prev; - s64 delta; - - prev = local64_xchg(&event->hw.prev_count, now); - delta = now - prev; - local64_add(delta, &event->count); -} - -static int task_clock_perf_event_enable(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - u64 now; - - now = event->ctx->time; - - local64_set(&hwc->prev_count, now); - - perf_swevent_start_hrtimer(event); - - return 0; -} - -static void task_clock_perf_event_disable(struct perf_event *event) -{ - perf_swevent_cancel_hrtimer(event); - task_clock_perf_event_update(event, event->ctx->time); - -} - -static void task_clock_perf_event_read(struct perf_event *event) -{ - u64 time; - - if (!in_nmi()) { - update_context_time(event->ctx); - time = event->ctx->time; - } else { - u64 now = perf_clock(); - u64 delta = now - event->ctx->timestamp; - time = event->ctx->time + delta; - } - - task_clock_perf_event_update(event, time); -} - -static const struct pmu perf_ops_task_clock = { - .enable = task_clock_perf_event_enable, - .disable = task_clock_perf_event_disable, - .read = task_clock_perf_event_read, -}; - /* Deref the hlist from the update side */ static inline struct swevent_hlist * -swevent_hlist_deref(struct perf_cpu_context *cpuctx) +swevent_hlist_deref(struct swevent_htable *swhash) { - return rcu_dereference_protected(cpuctx->swevent_hlist, - lockdep_is_held(&cpuctx->hlist_mutex)); + return rcu_dereference_protected(swhash->swevent_hlist, + lockdep_is_held(&swhash->hlist_mutex)); } static void swevent_hlist_release_rcu(struct rcu_head *rcu_head) @@ -4501,27 +4609,27 @@ static void swevent_hlist_release_rcu(struct rcu_head *rcu_head) kfree(hlist); } -static void swevent_hlist_release(struct perf_cpu_context *cpuctx) +static void swevent_hlist_release(struct swevent_htable *swhash) { - struct swevent_hlist *hlist = swevent_hlist_deref(cpuctx); + struct swevent_hlist *hlist = swevent_hlist_deref(swhash); if (!hlist) return; - rcu_assign_pointer(cpuctx->swevent_hlist, NULL); + rcu_assign_pointer(swhash->swevent_hlist, NULL); call_rcu(&hlist->rcu_head, swevent_hlist_release_rcu); } static void swevent_hlist_put_cpu(struct perf_event *event, int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); - mutex_lock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); - if (!--cpuctx->hlist_refcount) - swevent_hlist_release(cpuctx); + if (!--swhash->hlist_refcount) + swevent_hlist_release(swhash); - mutex_unlock(&cpuctx->hlist_mutex); + mutex_unlock(&swhash->hlist_mutex); } static void swevent_hlist_put(struct perf_event *event) @@ -4539,12 +4647,12 @@ static void swevent_hlist_put(struct perf_event *event) static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); int err = 0; - mutex_lock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); - if (!swevent_hlist_deref(cpuctx) && cpu_online(cpu)) { + if (!swevent_hlist_deref(swhash) && cpu_online(cpu)) { struct swevent_hlist *hlist; hlist = kzalloc(sizeof(*hlist), GFP_KERNEL); @@ -4552,11 +4660,11 @@ static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) err = -ENOMEM; goto exit; } - rcu_assign_pointer(cpuctx->swevent_hlist, hlist); + rcu_assign_pointer(swhash->swevent_hlist, hlist); } - cpuctx->hlist_refcount++; - exit: - mutex_unlock(&cpuctx->hlist_mutex); + swhash->hlist_refcount++; +exit: + mutex_unlock(&swhash->hlist_mutex); return err; } @@ -4580,7 +4688,7 @@ static int swevent_hlist_get(struct perf_event *event) put_online_cpus(); return 0; - fail: +fail: for_each_possible_cpu(cpu) { if (cpu == failed_cpu) break; @@ -4591,17 +4699,64 @@ static int swevent_hlist_get(struct perf_event *event) return err; } -#ifdef CONFIG_EVENT_TRACING +atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX]; + +static void sw_perf_event_destroy(struct perf_event *event) +{ + u64 event_id = event->attr.config; + + WARN_ON(event->parent); + + atomic_dec(&perf_swevent_enabled[event_id]); + swevent_hlist_put(event); +} + +static int perf_swevent_init(struct perf_event *event) +{ + int event_id = event->attr.config; -static const struct pmu perf_ops_tracepoint = { - .enable = perf_trace_enable, - .disable = perf_trace_disable, - .start = perf_swevent_int, - .stop = perf_swevent_void, + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + switch (event_id) { + case PERF_COUNT_SW_CPU_CLOCK: + case PERF_COUNT_SW_TASK_CLOCK: + return -ENOENT; + + default: + break; + } + + if (event_id > PERF_COUNT_SW_MAX) + return -ENOENT; + + if (!event->parent) { + int err; + + err = swevent_hlist_get(event); + if (err) + return err; + + atomic_inc(&perf_swevent_enabled[event_id]); + event->destroy = sw_perf_event_destroy; + } + + return 0; +} + +static struct pmu perf_swevent = { + .task_ctx_nr = perf_sw_context, + + .event_init = perf_swevent_init, + .add = perf_swevent_add, + .del = perf_swevent_del, + .start = perf_swevent_start, + .stop = perf_swevent_stop, .read = perf_swevent_read, - .unthrottle = perf_swevent_void, }; +#ifdef CONFIG_EVENT_TRACING + static int perf_tp_filter_match(struct perf_event *event, struct perf_sample_data *data) { @@ -4645,7 +4800,7 @@ void perf_tp_event(u64 addr, u64 count, void *record, int entry_size, hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_tp_event_match(event, &data, regs)) - perf_swevent_add(event, count, 1, &data, regs); + perf_swevent_event(event, count, 1, &data, regs); } perf_swevent_put_recursion_context(rctx); @@ -4657,10 +4812,13 @@ static void tp_perf_event_destroy(struct perf_event *event) perf_trace_destroy(event); } -static const struct pmu *tp_perf_event_init(struct perf_event *event) +static int perf_tp_event_init(struct perf_event *event) { int err; + if (event->attr.type != PERF_TYPE_TRACEPOINT) + return -ENOENT; + /* * Raw tracepoint data is a severe data leak, only allow root to * have these. @@ -4668,15 +4826,31 @@ static const struct pmu *tp_perf_event_init(struct perf_event *event) if ((event->attr.sample_type & PERF_SAMPLE_RAW) && perf_paranoid_tracepoint_raw() && !capable(CAP_SYS_ADMIN)) - return ERR_PTR(-EPERM); + return -EPERM; err = perf_trace_init(event); if (err) - return NULL; + return err; event->destroy = tp_perf_event_destroy; - return &perf_ops_tracepoint; + return 0; +} + +static struct pmu perf_tracepoint = { + .task_ctx_nr = perf_sw_context, + + .event_init = perf_tp_event_init, + .add = perf_trace_add, + .del = perf_trace_del, + .start = perf_swevent_start, + .stop = perf_swevent_stop, + .read = perf_swevent_read, +}; + +static inline void perf_tp_register(void) +{ + perf_pmu_register(&perf_tracepoint); } static int perf_event_set_filter(struct perf_event *event, void __user *arg) @@ -4704,9 +4878,8 @@ static void perf_event_free_filter(struct perf_event *event) #else -static const struct pmu *tp_perf_event_init(struct perf_event *event) +static inline void perf_tp_register(void) { - return NULL; } static int perf_event_set_filter(struct perf_event *event, void __user *arg) @@ -4721,105 +4894,389 @@ static void perf_event_free_filter(struct perf_event *event) #endif /* CONFIG_EVENT_TRACING */ #ifdef CONFIG_HAVE_HW_BREAKPOINT -static void bp_perf_event_destroy(struct perf_event *event) +void perf_bp_event(struct perf_event *bp, void *data) { - release_bp_slot(event); + struct perf_sample_data sample; + struct pt_regs *regs = data; + + perf_sample_data_init(&sample, bp->attr.bp_addr); + + if (!bp->hw.state && !perf_exclude_event(bp, regs)) + perf_swevent_event(bp, 1, 1, &sample, regs); } +#endif + +/* + * hrtimer based swevent callback + */ -static const struct pmu *bp_perf_event_init(struct perf_event *bp) +static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) { - int err; + enum hrtimer_restart ret = HRTIMER_RESTART; + struct perf_sample_data data; + struct pt_regs *regs; + struct perf_event *event; + u64 period; - err = register_perf_hw_breakpoint(bp); - if (err) - return ERR_PTR(err); + event = container_of(hrtimer, struct perf_event, hw.hrtimer); + event->pmu->read(event); - bp->destroy = bp_perf_event_destroy; + perf_sample_data_init(&data, 0); + data.period = event->hw.last_period; + regs = get_irq_regs(); + + if (regs && !perf_exclude_event(event, regs)) { + if (!(event->attr.exclude_idle && current->pid == 0)) + if (perf_event_overflow(event, 0, &data, regs)) + ret = HRTIMER_NORESTART; + } - return &perf_ops_bp; + period = max_t(u64, 10000, event->hw.sample_period); + hrtimer_forward_now(hrtimer, ns_to_ktime(period)); + + return ret; } -void perf_bp_event(struct perf_event *bp, void *data) +static void perf_swevent_start_hrtimer(struct perf_event *event) { - struct perf_sample_data sample; - struct pt_regs *regs = data; + struct hw_perf_event *hwc = &event->hw; - perf_sample_data_init(&sample, bp->attr.bp_addr); + hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hwc->hrtimer.function = perf_swevent_hrtimer; + if (hwc->sample_period) { + s64 period = local64_read(&hwc->period_left); - if (!perf_exclude_event(bp, regs)) - perf_swevent_add(bp, 1, 1, &sample, regs); + if (period) { + if (period < 0) + period = 10000; + + local64_set(&hwc->period_left, 0); + } else { + period = max_t(u64, 10000, hwc->sample_period); + } + __hrtimer_start_range_ns(&hwc->hrtimer, + ns_to_ktime(period), 0, + HRTIMER_MODE_REL_PINNED, 0); + } } -#else -static const struct pmu *bp_perf_event_init(struct perf_event *bp) + +static void perf_swevent_cancel_hrtimer(struct perf_event *event) { - return NULL; + struct hw_perf_event *hwc = &event->hw; + + if (hwc->sample_period) { + ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer); + local64_set(&hwc->period_left, ktime_to_ns(remaining)); + + hrtimer_cancel(&hwc->hrtimer); + } } -void perf_bp_event(struct perf_event *bp, void *regs) +/* + * Software event: cpu wall time clock + */ + +static void cpu_clock_event_update(struct perf_event *event) { + s64 prev; + u64 now; + + now = local_clock(); + prev = local64_xchg(&event->hw.prev_count, now); + local64_add(now - prev, &event->count); } -#endif -atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX]; +static void cpu_clock_event_start(struct perf_event *event, int flags) +{ + local64_set(&event->hw.prev_count, local_clock()); + perf_swevent_start_hrtimer(event); +} -static void sw_perf_event_destroy(struct perf_event *event) +static void cpu_clock_event_stop(struct perf_event *event, int flags) { - u64 event_id = event->attr.config; + perf_swevent_cancel_hrtimer(event); + cpu_clock_event_update(event); +} - WARN_ON(event->parent); +static int cpu_clock_event_add(struct perf_event *event, int flags) +{ + if (flags & PERF_EF_START) + cpu_clock_event_start(event, flags); - atomic_dec(&perf_swevent_enabled[event_id]); - swevent_hlist_put(event); + return 0; } -static const struct pmu *sw_perf_event_init(struct perf_event *event) +static void cpu_clock_event_del(struct perf_event *event, int flags) { - const struct pmu *pmu = NULL; - u64 event_id = event->attr.config; + cpu_clock_event_stop(event, flags); +} + +static void cpu_clock_event_read(struct perf_event *event) +{ + cpu_clock_event_update(event); +} + +static int cpu_clock_event_init(struct perf_event *event) +{ + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + if (event->attr.config != PERF_COUNT_SW_CPU_CLOCK) + return -ENOENT; + + return 0; +} + +static struct pmu perf_cpu_clock = { + .task_ctx_nr = perf_sw_context, + + .event_init = cpu_clock_event_init, + .add = cpu_clock_event_add, + .del = cpu_clock_event_del, + .start = cpu_clock_event_start, + .stop = cpu_clock_event_stop, + .read = cpu_clock_event_read, +}; + +/* + * Software event: task time clock + */ + +static void task_clock_event_update(struct perf_event *event, u64 now) +{ + u64 prev; + s64 delta; + + prev = local64_xchg(&event->hw.prev_count, now); + delta = now - prev; + local64_add(delta, &event->count); +} + +static void task_clock_event_start(struct perf_event *event, int flags) +{ + local64_set(&event->hw.prev_count, event->ctx->time); + perf_swevent_start_hrtimer(event); +} + +static void task_clock_event_stop(struct perf_event *event, int flags) +{ + perf_swevent_cancel_hrtimer(event); + task_clock_event_update(event, event->ctx->time); +} + +static int task_clock_event_add(struct perf_event *event, int flags) +{ + if (flags & PERF_EF_START) + task_clock_event_start(event, flags); + + return 0; +} + +static void task_clock_event_del(struct perf_event *event, int flags) +{ + task_clock_event_stop(event, PERF_EF_UPDATE); +} + +static void task_clock_event_read(struct perf_event *event) +{ + u64 time; + + if (!in_nmi()) { + update_context_time(event->ctx); + time = event->ctx->time; + } else { + u64 now = perf_clock(); + u64 delta = now - event->ctx->timestamp; + time = event->ctx->time + delta; + } + + task_clock_event_update(event, time); +} + +static int task_clock_event_init(struct perf_event *event) +{ + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + if (event->attr.config != PERF_COUNT_SW_TASK_CLOCK) + return -ENOENT; + + return 0; +} + +static struct pmu perf_task_clock = { + .task_ctx_nr = perf_sw_context, + + .event_init = task_clock_event_init, + .add = task_clock_event_add, + .del = task_clock_event_del, + .start = task_clock_event_start, + .stop = task_clock_event_stop, + .read = task_clock_event_read, +}; + +static void perf_pmu_nop_void(struct pmu *pmu) +{ +} + +static int perf_pmu_nop_int(struct pmu *pmu) +{ + return 0; +} + +static void perf_pmu_start_txn(struct pmu *pmu) +{ + perf_pmu_disable(pmu); +} + +static int perf_pmu_commit_txn(struct pmu *pmu) +{ + perf_pmu_enable(pmu); + return 0; +} +static void perf_pmu_cancel_txn(struct pmu *pmu) +{ + perf_pmu_enable(pmu); +} + +/* + * Ensures all contexts with the same task_ctx_nr have the same + * pmu_cpu_context too. + */ +static void *find_pmu_context(int ctxn) +{ + struct pmu *pmu; + + if (ctxn < 0) + return NULL; + + list_for_each_entry(pmu, &pmus, entry) { + if (pmu->task_ctx_nr == ctxn) + return pmu->pmu_cpu_context; + } + + return NULL; +} + +static void free_pmu_context(void * __percpu cpu_context) +{ + struct pmu *pmu; + + mutex_lock(&pmus_lock); /* - * Software events (currently) can't in general distinguish - * between user, kernel and hypervisor events. - * However, context switches and cpu migrations are considered - * to be kernel events, and page faults are never hypervisor - * events. + * Like a real lame refcount. */ - switch (event_id) { - case PERF_COUNT_SW_CPU_CLOCK: - pmu = &perf_ops_cpu_clock; + list_for_each_entry(pmu, &pmus, entry) { + if (pmu->pmu_cpu_context == cpu_context) + goto out; + } - break; - case PERF_COUNT_SW_TASK_CLOCK: - /* - * If the user instantiates this as a per-cpu event, - * use the cpu_clock event instead. - */ - if (event->ctx->task) - pmu = &perf_ops_task_clock; - else - pmu = &perf_ops_cpu_clock; + free_percpu(cpu_context); +out: + mutex_unlock(&pmus_lock); +} - break; - case PERF_COUNT_SW_PAGE_FAULTS: - case PERF_COUNT_SW_PAGE_FAULTS_MIN: - case PERF_COUNT_SW_PAGE_FAULTS_MAJ: - case PERF_COUNT_SW_CONTEXT_SWITCHES: - case PERF_COUNT_SW_CPU_MIGRATIONS: - case PERF_COUNT_SW_ALIGNMENT_FAULTS: - case PERF_COUNT_SW_EMULATION_FAULTS: - if (!event->parent) { - int err; - - err = swevent_hlist_get(event); - if (err) - return ERR_PTR(err); +int perf_pmu_register(struct pmu *pmu) +{ + int cpu, ret; - atomic_inc(&perf_swevent_enabled[event_id]); - event->destroy = sw_perf_event_destroy; + mutex_lock(&pmus_lock); + ret = -ENOMEM; + pmu->pmu_disable_count = alloc_percpu(int); + if (!pmu->pmu_disable_count) + goto unlock; + + pmu->pmu_cpu_context = find_pmu_context(pmu->task_ctx_nr); + if (pmu->pmu_cpu_context) + goto got_cpu_context; + + pmu->pmu_cpu_context = alloc_percpu(struct perf_cpu_context); + if (!pmu->pmu_cpu_context) + goto free_pdc; + + for_each_possible_cpu(cpu) { + struct perf_cpu_context *cpuctx; + + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); + __perf_event_init_context(&cpuctx->ctx); + cpuctx->ctx.type = cpu_context; + cpuctx->ctx.pmu = pmu; + cpuctx->jiffies_interval = 1; + INIT_LIST_HEAD(&cpuctx->rotation_list); + } + +got_cpu_context: + if (!pmu->start_txn) { + if (pmu->pmu_enable) { + /* + * If we have pmu_enable/pmu_disable calls, install + * transaction stubs that use that to try and batch + * hardware accesses. + */ + pmu->start_txn = perf_pmu_start_txn; + pmu->commit_txn = perf_pmu_commit_txn; + pmu->cancel_txn = perf_pmu_cancel_txn; + } else { + pmu->start_txn = perf_pmu_nop_void; + pmu->commit_txn = perf_pmu_nop_int; + pmu->cancel_txn = perf_pmu_nop_void; + } + } + + if (!pmu->pmu_enable) { + pmu->pmu_enable = perf_pmu_nop_void; + pmu->pmu_disable = perf_pmu_nop_void; + } + + list_add_rcu(&pmu->entry, &pmus); + ret = 0; +unlock: + mutex_unlock(&pmus_lock); + + return ret; + +free_pdc: + free_percpu(pmu->pmu_disable_count); + goto unlock; +} + +void perf_pmu_unregister(struct pmu *pmu) +{ + mutex_lock(&pmus_lock); + list_del_rcu(&pmu->entry); + mutex_unlock(&pmus_lock); + + /* + * We dereference the pmu list under both SRCU and regular RCU, so + * synchronize against both of those. + */ + synchronize_srcu(&pmus_srcu); + synchronize_rcu(); + + free_percpu(pmu->pmu_disable_count); + free_pmu_context(pmu->pmu_cpu_context); +} + +struct pmu *perf_init_event(struct perf_event *event) +{ + struct pmu *pmu = NULL; + int idx; + + idx = srcu_read_lock(&pmus_srcu); + list_for_each_entry_rcu(pmu, &pmus, entry) { + int ret = pmu->event_init(event); + if (!ret) + goto unlock; + + if (ret != -ENOENT) { + pmu = ERR_PTR(ret); + goto unlock; } - pmu = &perf_ops_generic; - break; } + pmu = ERR_PTR(-ENOENT); +unlock: + srcu_read_unlock(&pmus_srcu, idx); return pmu; } @@ -4828,20 +5285,17 @@ static const struct pmu *sw_perf_event_init(struct perf_event *event) * Allocate and initialize a event structure */ static struct perf_event * -perf_event_alloc(struct perf_event_attr *attr, - int cpu, - struct perf_event_context *ctx, +perf_event_alloc(struct perf_event_attr *attr, int cpu, struct perf_event *group_leader, struct perf_event *parent_event, - perf_overflow_handler_t overflow_handler, - gfp_t gfpflags) + perf_overflow_handler_t overflow_handler) { - const struct pmu *pmu; + struct pmu *pmu; struct perf_event *event; struct hw_perf_event *hwc; long err; - event = kzalloc(sizeof(*event), gfpflags); + event = kzalloc(sizeof(*event), GFP_KERNEL); if (!event) return ERR_PTR(-ENOMEM); @@ -4866,7 +5320,6 @@ perf_event_alloc(struct perf_event_attr *attr, event->attr = *attr; event->group_leader = group_leader; event->pmu = NULL; - event->ctx = ctx; event->oncpu = -1; event->parent = parent_event; @@ -4900,29 +5353,8 @@ perf_event_alloc(struct perf_event_attr *attr, if (attr->inherit && (attr->read_format & PERF_FORMAT_GROUP)) goto done; - switch (attr->type) { - case PERF_TYPE_RAW: - case PERF_TYPE_HARDWARE: - case PERF_TYPE_HW_CACHE: - pmu = hw_perf_event_init(event); - break; + pmu = perf_init_event(event); - case PERF_TYPE_SOFTWARE: - pmu = sw_perf_event_init(event); - break; - - case PERF_TYPE_TRACEPOINT: - pmu = tp_perf_event_init(event); - break; - - case PERF_TYPE_BREAKPOINT: - pmu = bp_perf_event_init(event); - break; - - - default: - break; - } done: err = 0; if (!pmu) @@ -4947,6 +5379,13 @@ done: atomic_inc(&nr_comm_events); if (event->attr.task) atomic_inc(&nr_task_events); + if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) { + err = get_callchain_buffers(); + if (err) { + free_event(event); + return ERR_PTR(err); + } + } } return event; @@ -5094,12 +5533,16 @@ SYSCALL_DEFINE5(perf_event_open, struct perf_event_attr __user *, attr_uptr, pid_t, pid, int, cpu, int, group_fd, unsigned long, flags) { - struct perf_event *event, *group_leader = NULL, *output_event = NULL; + struct perf_event *group_leader = NULL, *output_event = NULL; + struct perf_event *event, *sibling; struct perf_event_attr attr; struct perf_event_context *ctx; struct file *event_file = NULL; struct file *group_file = NULL; + struct task_struct *task = NULL; + struct pmu *pmu; int event_fd; + int move_group = 0; int fput_needed = 0; int err; @@ -5125,20 +5568,11 @@ SYSCALL_DEFINE5(perf_event_open, if (event_fd < 0) return event_fd; - /* - * Get the target context (task or percpu): - */ - ctx = find_get_context(pid, cpu); - if (IS_ERR(ctx)) { - err = PTR_ERR(ctx); - goto err_fd; - } - if (group_fd != -1) { group_leader = perf_fget_light(group_fd, &fput_needed); if (IS_ERR(group_leader)) { err = PTR_ERR(group_leader); - goto err_put_context; + goto err_fd; } group_file = group_leader->filp; if (flags & PERF_FLAG_FD_OUTPUT) @@ -5147,6 +5581,58 @@ SYSCALL_DEFINE5(perf_event_open, group_leader = NULL; } + event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); + if (IS_ERR(event)) { + err = PTR_ERR(event); + goto err_fd; + } + + /* + * Special case software events and allow them to be part of + * any hardware group. + */ + pmu = event->pmu; + + if (group_leader && + (is_software_event(event) != is_software_event(group_leader))) { + if (is_software_event(event)) { + /* + * If event and group_leader are not both a software + * event, and event is, then group leader is not. + * + * Allow the addition of software events to !software + * groups, this is safe because software events never + * fail to schedule. + */ + pmu = group_leader->pmu; + } else if (is_software_event(group_leader) && + (group_leader->group_flags & PERF_GROUP_SOFTWARE)) { + /* + * In case the group is a pure software group, and we + * try to add a hardware event, move the whole group to + * the hardware context. + */ + move_group = 1; + } + } + + if (pid != -1) { + task = find_lively_task_by_vpid(pid); + if (IS_ERR(task)) { + err = PTR_ERR(task); + goto err_group_fd; + } + } + + /* + * Get the target context (task or percpu): + */ + ctx = find_get_context(pmu, task, cpu); + if (IS_ERR(ctx)) { + err = PTR_ERR(ctx); + goto err_group_fd; + } + /* * Look up the group leader (we will attach this event to it): */ @@ -5158,42 +5644,66 @@ SYSCALL_DEFINE5(perf_event_open, * becoming part of another group-sibling): */ if (group_leader->group_leader != group_leader) - goto err_put_context; + goto err_context; /* * Do not allow to attach to a group in a different * task or CPU context: */ - if (group_leader->ctx != ctx) - goto err_put_context; + if (move_group) { + if (group_leader->ctx->type != ctx->type) + goto err_context; + } else { + if (group_leader->ctx != ctx) + goto err_context; + } + /* * Only a group leader can be exclusive or pinned */ if (attr.exclusive || attr.pinned) - goto err_put_context; - } - - event = perf_event_alloc(&attr, cpu, ctx, group_leader, - NULL, NULL, GFP_KERNEL); - if (IS_ERR(event)) { - err = PTR_ERR(event); - goto err_put_context; + goto err_context; } if (output_event) { err = perf_event_set_output(event, output_event); if (err) - goto err_free_put_context; + goto err_context; } event_file = anon_inode_getfile("[perf_event]", &perf_fops, event, O_RDWR); if (IS_ERR(event_file)) { err = PTR_ERR(event_file); - goto err_free_put_context; + goto err_context; + } + + if (move_group) { + struct perf_event_context *gctx = group_leader->ctx; + + mutex_lock(&gctx->mutex); + perf_event_remove_from_context(group_leader); + list_for_each_entry(sibling, &group_leader->sibling_list, + group_entry) { + perf_event_remove_from_context(sibling); + put_ctx(gctx); + } + mutex_unlock(&gctx->mutex); + put_ctx(gctx); } event->filp = event_file; WARN_ON_ONCE(ctx->parent_ctx); mutex_lock(&ctx->mutex); + + if (move_group) { + perf_install_in_context(ctx, group_leader, cpu); + get_ctx(ctx); + list_for_each_entry(sibling, &group_leader->sibling_list, + group_entry) { + perf_install_in_context(ctx, sibling, cpu); + get_ctx(ctx); + } + } + perf_install_in_context(ctx, event, cpu); ++ctx->generation; mutex_unlock(&ctx->mutex); @@ -5214,11 +5724,11 @@ SYSCALL_DEFINE5(perf_event_open, fd_install(event_fd, event_file); return event_fd; -err_free_put_context: - free_event(event); -err_put_context: - fput_light(group_file, fput_needed); +err_context: put_ctx(ctx); +err_group_fd: + fput_light(group_file, fput_needed); + free_event(event); err_fd: put_unused_fd(event_fd); return err; @@ -5229,32 +5739,31 @@ err_fd: * * @attr: attributes of the counter to create * @cpu: cpu in which the counter is bound - * @pid: task to profile + * @task: task to profile (NULL for percpu) */ struct perf_event * perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, - pid_t pid, + struct task_struct *task, perf_overflow_handler_t overflow_handler) { - struct perf_event *event; struct perf_event_context *ctx; + struct perf_event *event; int err; /* * Get the target context (task or percpu): */ - ctx = find_get_context(pid, cpu); - if (IS_ERR(ctx)) { - err = PTR_ERR(ctx); - goto err_exit; - } - - event = perf_event_alloc(attr, cpu, ctx, NULL, - NULL, overflow_handler, GFP_KERNEL); + event = perf_event_alloc(attr, cpu, NULL, NULL, overflow_handler); if (IS_ERR(event)) { err = PTR_ERR(event); - goto err_put_context; + goto err; + } + + ctx = find_get_context(event->pmu, task, cpu); + if (IS_ERR(ctx)) { + err = PTR_ERR(ctx); + goto err_free; } event->filp = NULL; @@ -5272,112 +5781,13 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, return event; - err_put_context: - put_ctx(ctx); - err_exit: +err_free: + free_event(event); +err: return ERR_PTR(err); } EXPORT_SYMBOL_GPL(perf_event_create_kernel_counter); -/* - * inherit a event from parent task to child task: - */ -static struct perf_event * -inherit_event(struct perf_event *parent_event, - struct task_struct *parent, - struct perf_event_context *parent_ctx, - struct task_struct *child, - struct perf_event *group_leader, - struct perf_event_context *child_ctx) -{ - struct perf_event *child_event; - - /* - * Instead of creating recursive hierarchies of events, - * we link inherited events back to the original parent, - * which has a filp for sure, which we use as the reference - * count: - */ - if (parent_event->parent) - parent_event = parent_event->parent; - - child_event = perf_event_alloc(&parent_event->attr, - parent_event->cpu, child_ctx, - group_leader, parent_event, - NULL, GFP_KERNEL); - if (IS_ERR(child_event)) - return child_event; - get_ctx(child_ctx); - - /* - * Make the child state follow the state of the parent event, - * not its attr.disabled bit. We hold the parent's mutex, - * so we won't race with perf_event_{en, dis}able_family. - */ - if (parent_event->state >= PERF_EVENT_STATE_INACTIVE) - child_event->state = PERF_EVENT_STATE_INACTIVE; - else - child_event->state = PERF_EVENT_STATE_OFF; - - if (parent_event->attr.freq) { - u64 sample_period = parent_event->hw.sample_period; - struct hw_perf_event *hwc = &child_event->hw; - - hwc->sample_period = sample_period; - hwc->last_period = sample_period; - - local64_set(&hwc->period_left, sample_period); - } - - child_event->overflow_handler = parent_event->overflow_handler; - - /* - * Link it up in the child's context: - */ - add_event_to_ctx(child_event, child_ctx); - - /* - * Get a reference to the parent filp - we will fput it - * when the child event exits. This is safe to do because - * we are in the parent and we know that the filp still - * exists and has a nonzero count: - */ - atomic_long_inc(&parent_event->filp->f_count); - - /* - * Link this into the parent event's child list - */ - WARN_ON_ONCE(parent_event->ctx->parent_ctx); - mutex_lock(&parent_event->child_mutex); - list_add_tail(&child_event->child_list, &parent_event->child_list); - mutex_unlock(&parent_event->child_mutex); - - return child_event; -} - -static int inherit_group(struct perf_event *parent_event, - struct task_struct *parent, - struct perf_event_context *parent_ctx, - struct task_struct *child, - struct perf_event_context *child_ctx) -{ - struct perf_event *leader; - struct perf_event *sub; - struct perf_event *child_ctr; - - leader = inherit_event(parent_event, parent, parent_ctx, - child, NULL, child_ctx); - if (IS_ERR(leader)) - return PTR_ERR(leader); - list_for_each_entry(sub, &parent_event->sibling_list, group_entry) { - child_ctr = inherit_event(sub, parent, parent_ctx, - child, leader, child_ctx); - if (IS_ERR(child_ctr)) - return PTR_ERR(child_ctr); - } - return 0; -} - static void sync_child_event(struct perf_event *child_event, struct task_struct *child) { @@ -5434,16 +5844,13 @@ __perf_event_exit_task(struct perf_event *child_event, } } -/* - * When a child task exits, feed back event values to parent events. - */ -void perf_event_exit_task(struct task_struct *child) +static void perf_event_exit_task_context(struct task_struct *child, int ctxn) { struct perf_event *child_event, *tmp; struct perf_event_context *child_ctx; unsigned long flags; - if (likely(!child->perf_event_ctxp)) { + if (likely(!child->perf_event_ctxp[ctxn])) { perf_event_task(child, NULL, 0); return; } @@ -5455,7 +5862,7 @@ void perf_event_exit_task(struct task_struct *child) * scheduled, so we are now safe from rescheduling changing * our context. */ - child_ctx = child->perf_event_ctxp; + child_ctx = child->perf_event_ctxp[ctxn]; __perf_event_task_sched_out(child_ctx); /* @@ -5464,7 +5871,7 @@ void perf_event_exit_task(struct task_struct *child) * incremented the context's refcount before we do put_ctx below. */ raw_spin_lock(&child_ctx->lock); - child->perf_event_ctxp = NULL; + child->perf_event_ctxp[ctxn] = NULL; /* * If this context is a clone; unclone it so it can't get * swapped to another process while we're removing all @@ -5517,6 +5924,17 @@ again: put_ctx(child_ctx); } +/* + * When a child task exits, feed back event values to parent events. + */ +void perf_event_exit_task(struct task_struct *child) +{ + int ctxn; + + for_each_task_context_nr(ctxn) + perf_event_exit_task_context(child, ctxn); +} + static void perf_free_event(struct perf_event *event, struct perf_event_context *ctx) { @@ -5538,48 +5956,165 @@ static void perf_free_event(struct perf_event *event, /* * free an unexposed, unused context as created by inheritance by - * init_task below, used by fork() in case of fail. + * perf_event_init_task below, used by fork() in case of fail. */ void perf_event_free_task(struct task_struct *task) { - struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_event_context *ctx; struct perf_event *event, *tmp; + int ctxn; - if (!ctx) - return; + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (!ctx) + continue; - mutex_lock(&ctx->mutex); + mutex_lock(&ctx->mutex); again: - list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry) - perf_free_event(event, ctx); + list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, + group_entry) + perf_free_event(event, ctx); - list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, - group_entry) - perf_free_event(event, ctx); + list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, + group_entry) + perf_free_event(event, ctx); - if (!list_empty(&ctx->pinned_groups) || - !list_empty(&ctx->flexible_groups)) - goto again; + if (!list_empty(&ctx->pinned_groups) || + !list_empty(&ctx->flexible_groups)) + goto again; - mutex_unlock(&ctx->mutex); + mutex_unlock(&ctx->mutex); - put_ctx(ctx); + put_ctx(ctx); + } +} + +void perf_event_delayed_put(struct task_struct *task) +{ + int ctxn; + + for_each_task_context_nr(ctxn) + WARN_ON_ONCE(task->perf_event_ctxp[ctxn]); +} + +/* + * inherit a event from parent task to child task: + */ +static struct perf_event * +inherit_event(struct perf_event *parent_event, + struct task_struct *parent, + struct perf_event_context *parent_ctx, + struct task_struct *child, + struct perf_event *group_leader, + struct perf_event_context *child_ctx) +{ + struct perf_event *child_event; + unsigned long flags; + + /* + * Instead of creating recursive hierarchies of events, + * we link inherited events back to the original parent, + * which has a filp for sure, which we use as the reference + * count: + */ + if (parent_event->parent) + parent_event = parent_event->parent; + + child_event = perf_event_alloc(&parent_event->attr, + parent_event->cpu, + group_leader, parent_event, + NULL); + if (IS_ERR(child_event)) + return child_event; + get_ctx(child_ctx); + + /* + * Make the child state follow the state of the parent event, + * not its attr.disabled bit. We hold the parent's mutex, + * so we won't race with perf_event_{en, dis}able_family. + */ + if (parent_event->state >= PERF_EVENT_STATE_INACTIVE) + child_event->state = PERF_EVENT_STATE_INACTIVE; + else + child_event->state = PERF_EVENT_STATE_OFF; + + if (parent_event->attr.freq) { + u64 sample_period = parent_event->hw.sample_period; + struct hw_perf_event *hwc = &child_event->hw; + + hwc->sample_period = sample_period; + hwc->last_period = sample_period; + + local64_set(&hwc->period_left, sample_period); + } + + child_event->ctx = child_ctx; + child_event->overflow_handler = parent_event->overflow_handler; + + /* + * Link it up in the child's context: + */ + raw_spin_lock_irqsave(&child_ctx->lock, flags); + add_event_to_ctx(child_event, child_ctx); + raw_spin_unlock_irqrestore(&child_ctx->lock, flags); + + /* + * Get a reference to the parent filp - we will fput it + * when the child event exits. This is safe to do because + * we are in the parent and we know that the filp still + * exists and has a nonzero count: + */ + atomic_long_inc(&parent_event->filp->f_count); + + /* + * Link this into the parent event's child list + */ + WARN_ON_ONCE(parent_event->ctx->parent_ctx); + mutex_lock(&parent_event->child_mutex); + list_add_tail(&child_event->child_list, &parent_event->child_list); + mutex_unlock(&parent_event->child_mutex); + + return child_event; +} + +static int inherit_group(struct perf_event *parent_event, + struct task_struct *parent, + struct perf_event_context *parent_ctx, + struct task_struct *child, + struct perf_event_context *child_ctx) +{ + struct perf_event *leader; + struct perf_event *sub; + struct perf_event *child_ctr; + + leader = inherit_event(parent_event, parent, parent_ctx, + child, NULL, child_ctx); + if (IS_ERR(leader)) + return PTR_ERR(leader); + list_for_each_entry(sub, &parent_event->sibling_list, group_entry) { + child_ctr = inherit_event(sub, parent, parent_ctx, + child, leader, child_ctx); + if (IS_ERR(child_ctr)) + return PTR_ERR(child_ctr); + } + return 0; } static int inherit_task_group(struct perf_event *event, struct task_struct *parent, struct perf_event_context *parent_ctx, - struct task_struct *child, + struct task_struct *child, int ctxn, int *inherited_all) { int ret; - struct perf_event_context *child_ctx = child->perf_event_ctxp; + struct perf_event_context *child_ctx; if (!event->attr.inherit) { *inherited_all = 0; return 0; } + child_ctx = child->perf_event_ctxp[ctxn]; if (!child_ctx) { /* * This is executed from the parent task context, so @@ -5588,14 +6123,11 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, * child. */ - child_ctx = kzalloc(sizeof(struct perf_event_context), - GFP_KERNEL); + child_ctx = alloc_perf_context(event->pmu, child); if (!child_ctx) return -ENOMEM; - __perf_event_init_context(child_ctx, child); - child->perf_event_ctxp = child_ctx; - get_task_struct(child); + child->perf_event_ctxp[ctxn] = child_ctx; } ret = inherit_group(event, parent, parent_ctx, @@ -5607,11 +6139,10 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, return ret; } - /* * Initialize the perf_event context in task_struct */ -int perf_event_init_task(struct task_struct *child) +int perf_event_init_context(struct task_struct *child, int ctxn) { struct perf_event_context *child_ctx, *parent_ctx; struct perf_event_context *cloned_ctx; @@ -5620,19 +6151,19 @@ int perf_event_init_task(struct task_struct *child) int inherited_all = 1; int ret = 0; - child->perf_event_ctxp = NULL; + child->perf_event_ctxp[ctxn] = NULL; mutex_init(&child->perf_event_mutex); INIT_LIST_HEAD(&child->perf_event_list); - if (likely(!parent->perf_event_ctxp)) + if (likely(!parent->perf_event_ctxp[ctxn])) return 0; /* * If the parent's context is a clone, pin it so it won't get * swapped under us. */ - parent_ctx = perf_pin_task_context(parent); + parent_ctx = perf_pin_task_context(parent, ctxn); /* * No need to check if parent_ctx != NULL here; since we saw @@ -5652,20 +6183,20 @@ int perf_event_init_task(struct task_struct *child) * the list, not manipulating it: */ list_for_each_entry(event, &parent_ctx->pinned_groups, group_entry) { - ret = inherit_task_group(event, parent, parent_ctx, child, - &inherited_all); + ret = inherit_task_group(event, parent, parent_ctx, + child, ctxn, &inherited_all); if (ret) break; } list_for_each_entry(event, &parent_ctx->flexible_groups, group_entry) { - ret = inherit_task_group(event, parent, parent_ctx, child, - &inherited_all); + ret = inherit_task_group(event, parent, parent_ctx, + child, ctxn, &inherited_all); if (ret) break; } - child_ctx = child->perf_event_ctxp; + child_ctx = child->perf_event_ctxp[ctxn]; if (child_ctx && inherited_all) { /* @@ -5694,63 +6225,98 @@ int perf_event_init_task(struct task_struct *child) return ret; } +/* + * Initialize the perf_event context in task_struct + */ +int perf_event_init_task(struct task_struct *child) +{ + int ctxn, ret; + + for_each_task_context_nr(ctxn) { + ret = perf_event_init_context(child, ctxn); + if (ret) + return ret; + } + + return 0; +} + static void __init perf_event_init_all_cpus(void) { + struct swevent_htable *swhash; int cpu; - struct perf_cpu_context *cpuctx; for_each_possible_cpu(cpu) { - cpuctx = &per_cpu(perf_cpu_context, cpu); - mutex_init(&cpuctx->hlist_mutex); - __perf_event_init_context(&cpuctx->ctx, NULL); + swhash = &per_cpu(swevent_htable, cpu); + mutex_init(&swhash->hlist_mutex); + INIT_LIST_HEAD(&per_cpu(rotation_list, cpu)); } } static void __cpuinit perf_event_init_cpu(int cpu) { - struct perf_cpu_context *cpuctx; - - cpuctx = &per_cpu(perf_cpu_context, cpu); - - spin_lock(&perf_resource_lock); - cpuctx->max_pertask = perf_max_events - perf_reserved_percpu; - spin_unlock(&perf_resource_lock); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); - mutex_lock(&cpuctx->hlist_mutex); - if (cpuctx->hlist_refcount > 0) { + mutex_lock(&swhash->hlist_mutex); + if (swhash->hlist_refcount > 0) { struct swevent_hlist *hlist; - hlist = kzalloc(sizeof(*hlist), GFP_KERNEL); - WARN_ON_ONCE(!hlist); - rcu_assign_pointer(cpuctx->swevent_hlist, hlist); + hlist = kzalloc_node(sizeof(*hlist), GFP_KERNEL, cpu_to_node(cpu)); + WARN_ON(!hlist); + rcu_assign_pointer(swhash->swevent_hlist, hlist); } - mutex_unlock(&cpuctx->hlist_mutex); + mutex_unlock(&swhash->hlist_mutex); } #ifdef CONFIG_HOTPLUG_CPU -static void __perf_event_exit_cpu(void *info) +static void perf_pmu_rotate_stop(struct pmu *pmu) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - struct perf_event_context *ctx = &cpuctx->ctx; + struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + + WARN_ON(!irqs_disabled()); + + list_del_init(&cpuctx->rotation_list); +} + +static void __perf_event_exit_context(void *__info) +{ + struct perf_event_context *ctx = __info; struct perf_event *event, *tmp; + perf_pmu_rotate_stop(ctx->pmu); + list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry) __perf_event_remove_from_context(event); list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, group_entry) __perf_event_remove_from_context(event); } + +static void perf_event_exit_cpu_context(int cpu) +{ + struct perf_event_context *ctx; + struct pmu *pmu; + int idx; + + idx = srcu_read_lock(&pmus_srcu); + list_for_each_entry_rcu(pmu, &pmus, entry) { + ctx = &per_cpu_ptr(pmu->pmu_cpu_context, cpu)->ctx; + + mutex_lock(&ctx->mutex); + smp_call_function_single(cpu, __perf_event_exit_context, ctx, 1); + mutex_unlock(&ctx->mutex); + } + srcu_read_unlock(&pmus_srcu, idx); +} + static void perf_event_exit_cpu(int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); - struct perf_event_context *ctx = &cpuctx->ctx; + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); - mutex_lock(&cpuctx->hlist_mutex); - swevent_hlist_release(cpuctx); - mutex_unlock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); + swevent_hlist_release(swhash); + mutex_unlock(&swhash->hlist_mutex); - mutex_lock(&ctx->mutex); - smp_call_function_single(cpu, __perf_event_exit_cpu, NULL, 1); - mutex_unlock(&ctx->mutex); + perf_event_exit_cpu_context(cpu); } #else static inline void perf_event_exit_cpu(int cpu) { } @@ -5780,118 +6346,13 @@ perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) return NOTIFY_OK; } -/* - * This has to have a higher priority than migration_notifier in sched.c. - */ -static struct notifier_block __cpuinitdata perf_cpu_nb = { - .notifier_call = perf_cpu_notify, - .priority = 20, -}; - void __init perf_event_init(void) { perf_event_init_all_cpus(); - perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_UP_PREPARE, - (void *)(long)smp_processor_id()); - perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_ONLINE, - (void *)(long)smp_processor_id()); - register_cpu_notifier(&perf_cpu_nb); -} - -static ssize_t perf_show_reserve_percpu(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - char *buf) -{ - return sprintf(buf, "%d\n", perf_reserved_percpu); -} - -static ssize_t -perf_set_reserve_percpu(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - const char *buf, - size_t count) -{ - struct perf_cpu_context *cpuctx; - unsigned long val; - int err, cpu, mpt; - - err = strict_strtoul(buf, 10, &val); - if (err) - return err; - if (val > perf_max_events) - return -EINVAL; - - spin_lock(&perf_resource_lock); - perf_reserved_percpu = val; - for_each_online_cpu(cpu) { - cpuctx = &per_cpu(perf_cpu_context, cpu); - raw_spin_lock_irq(&cpuctx->ctx.lock); - mpt = min(perf_max_events - cpuctx->ctx.nr_events, - perf_max_events - perf_reserved_percpu); - cpuctx->max_pertask = mpt; - raw_spin_unlock_irq(&cpuctx->ctx.lock); - } - spin_unlock(&perf_resource_lock); - - return count; -} - -static ssize_t perf_show_overcommit(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - char *buf) -{ - return sprintf(buf, "%d\n", perf_overcommit); -} - -static ssize_t -perf_set_overcommit(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - const char *buf, size_t count) -{ - unsigned long val; - int err; - - err = strict_strtoul(buf, 10, &val); - if (err) - return err; - if (val > 1) - return -EINVAL; - - spin_lock(&perf_resource_lock); - perf_overcommit = val; - spin_unlock(&perf_resource_lock); - - return count; -} - -static SYSDEV_CLASS_ATTR( - reserve_percpu, - 0644, - perf_show_reserve_percpu, - perf_set_reserve_percpu - ); - -static SYSDEV_CLASS_ATTR( - overcommit, - 0644, - perf_show_overcommit, - perf_set_overcommit - ); - -static struct attribute *perfclass_attrs[] = { - &attr_reserve_percpu.attr, - &attr_overcommit.attr, - NULL -}; - -static struct attribute_group perfclass_attr_group = { - .attrs = perfclass_attrs, - .name = "perf_events", -}; - -static int __init perf_event_sysfs_init(void) -{ - return sysfs_create_group(&cpu_sysdev_class.kset.kobj, - &perfclass_attr_group); + init_srcu_struct(&pmus_srcu); + perf_pmu_register(&perf_swevent); + perf_pmu_register(&perf_cpu_clock); + perf_pmu_register(&perf_task_clock); + perf_tp_register(); + perf_cpu_notifier(perf_cpu_notify); } -device_initcall(perf_event_sysfs_init); diff --git a/kernel/sched.c b/kernel/sched.c index dc85ceb9083..c0d2067f3e0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3584,7 +3584,7 @@ void scheduler_tick(void) curr->sched_class->task_tick(rq, curr, 0); raw_spin_unlock(&rq->lock); - perf_event_task_tick(curr); + perf_event_task_tick(); #ifdef CONFIG_SMP rq->idle_at_tick = idle_cpu(cpu); diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index fa7ece649fe..65fb077ea79 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -884,10 +884,8 @@ enum { FTRACE_ENABLE_CALLS = (1 << 0), FTRACE_DISABLE_CALLS = (1 << 1), FTRACE_UPDATE_TRACE_FUNC = (1 << 2), - FTRACE_ENABLE_MCOUNT = (1 << 3), - FTRACE_DISABLE_MCOUNT = (1 << 4), - FTRACE_START_FUNC_RET = (1 << 5), - FTRACE_STOP_FUNC_RET = (1 << 6), + FTRACE_START_FUNC_RET = (1 << 3), + FTRACE_STOP_FUNC_RET = (1 << 4), }; static int ftrace_filtered; @@ -1226,8 +1224,6 @@ static void ftrace_shutdown(int command) static void ftrace_startup_sysctl(void) { - int command = FTRACE_ENABLE_MCOUNT; - if (unlikely(ftrace_disabled)) return; @@ -1235,23 +1231,17 @@ static void ftrace_startup_sysctl(void) saved_ftrace_func = NULL; /* ftrace_start_up is true if we want ftrace running */ if (ftrace_start_up) - command |= FTRACE_ENABLE_CALLS; - - ftrace_run_update_code(command); + ftrace_run_update_code(FTRACE_ENABLE_CALLS); } static void ftrace_shutdown_sysctl(void) { - int command = FTRACE_DISABLE_MCOUNT; - if (unlikely(ftrace_disabled)) return; /* ftrace_start_up is true if ftrace is running */ if (ftrace_start_up) - command |= FTRACE_DISABLE_CALLS; - - ftrace_run_update_code(command); + ftrace_run_update_code(FTRACE_DISABLE_CALLS); } static cycle_t ftrace_update_time; @@ -1368,24 +1358,29 @@ enum { #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ struct ftrace_iterator { - struct ftrace_page *pg; - int hidx; - int idx; - unsigned flags; - struct trace_parser parser; + loff_t pos; + loff_t func_pos; + struct ftrace_page *pg; + struct dyn_ftrace *func; + struct ftrace_func_probe *probe; + struct trace_parser parser; + int hidx; + int idx; + unsigned flags; }; static void * -t_hash_next(struct seq_file *m, void *v, loff_t *pos) +t_hash_next(struct seq_file *m, loff_t *pos) { struct ftrace_iterator *iter = m->private; - struct hlist_node *hnd = v; + struct hlist_node *hnd = NULL; struct hlist_head *hhd; - WARN_ON(!(iter->flags & FTRACE_ITER_HASH)); - (*pos)++; + iter->pos = *pos; + if (iter->probe) + hnd = &iter->probe->node; retry: if (iter->hidx >= FTRACE_FUNC_HASHSIZE) return NULL; @@ -1408,7 +1403,12 @@ t_hash_next(struct seq_file *m, void *v, loff_t *pos) } } - return hnd; + if (WARN_ON_ONCE(!hnd)) + return NULL; + + iter->probe = hlist_entry(hnd, struct ftrace_func_probe, node); + + return iter; } static void *t_hash_start(struct seq_file *m, loff_t *pos) @@ -1417,26 +1417,32 @@ static void *t_hash_start(struct seq_file *m, loff_t *pos) void *p = NULL; loff_t l; - if (!(iter->flags & FTRACE_ITER_HASH)) - *pos = 0; - - iter->flags |= FTRACE_ITER_HASH; + if (iter->func_pos > *pos) + return NULL; iter->hidx = 0; - for (l = 0; l <= *pos; ) { - p = t_hash_next(m, p, &l); + for (l = 0; l <= (*pos - iter->func_pos); ) { + p = t_hash_next(m, &l); if (!p) break; } - return p; + if (!p) + return NULL; + + /* Only set this if we have an item */ + iter->flags |= FTRACE_ITER_HASH; + + return iter; } -static int t_hash_show(struct seq_file *m, void *v) +static int +t_hash_show(struct seq_file *m, struct ftrace_iterator *iter) { struct ftrace_func_probe *rec; - struct hlist_node *hnd = v; - rec = hlist_entry(hnd, struct ftrace_func_probe, node); + rec = iter->probe; + if (WARN_ON_ONCE(!rec)) + return -EIO; if (rec->ops->print) return rec->ops->print(m, rec->ip, rec->ops, rec->data); @@ -1457,12 +1463,13 @@ t_next(struct seq_file *m, void *v, loff_t *pos) struct dyn_ftrace *rec = NULL; if (iter->flags & FTRACE_ITER_HASH) - return t_hash_next(m, v, pos); + return t_hash_next(m, pos); (*pos)++; + iter->pos = *pos; if (iter->flags & FTRACE_ITER_PRINTALL) - return NULL; + return t_hash_start(m, pos); retry: if (iter->idx >= iter->pg->index) { @@ -1491,7 +1498,20 @@ t_next(struct seq_file *m, void *v, loff_t *pos) } } - return rec; + if (!rec) + return t_hash_start(m, pos); + + iter->func_pos = *pos; + iter->func = rec; + + return iter; +} + +static void reset_iter_read(struct ftrace_iterator *iter) +{ + iter->pos = 0; + iter->func_pos = 0; + iter->flags &= ~(FTRACE_ITER_PRINTALL & FTRACE_ITER_HASH); } static void *t_start(struct seq_file *m, loff_t *pos) @@ -1502,6 +1522,12 @@ static void *t_start(struct seq_file *m, loff_t *pos) mutex_lock(&ftrace_lock); /* + * If an lseek was done, then reset and start from beginning. + */ + if (*pos < iter->pos) + reset_iter_read(iter); + + /* * For set_ftrace_filter reading, if we have the filter * off, we can short cut and just print out that all * functions are enabled. @@ -1518,6 +1544,11 @@ static void *t_start(struct seq_file *m, loff_t *pos) if (iter->flags & FTRACE_ITER_HASH) return t_hash_start(m, pos); + /* + * Unfortunately, we need to restart at ftrace_pages_start + * every time we let go of the ftrace_mutex. This is because + * those pointers can change without the lock. + */ iter->pg = ftrace_pages_start; iter->idx = 0; for (l = 0; l <= *pos; ) { @@ -1526,10 +1557,14 @@ static void *t_start(struct seq_file *m, loff_t *pos) break; } - if (!p && iter->flags & FTRACE_ITER_FILTER) - return t_hash_start(m, pos); + if (!p) { + if (iter->flags & FTRACE_ITER_FILTER) + return t_hash_start(m, pos); - return p; + return NULL; + } + + return iter; } static void t_stop(struct seq_file *m, void *p) @@ -1540,16 +1575,18 @@ static void t_stop(struct seq_file *m, void *p) static int t_show(struct seq_file *m, void *v) { struct ftrace_iterator *iter = m->private; - struct dyn_ftrace *rec = v; + struct dyn_ftrace *rec; if (iter->flags & FTRACE_ITER_HASH) - return t_hash_show(m, v); + return t_hash_show(m, iter); if (iter->flags & FTRACE_ITER_PRINTALL) { seq_printf(m, "#### all functions enabled ####\n"); return 0; } + rec = iter->func; + if (!rec) return 0; @@ -2418,7 +2455,7 @@ static const struct file_operations ftrace_filter_fops = { .open = ftrace_filter_open, .read = seq_read, .write = ftrace_filter_write, - .llseek = no_llseek, + .llseek = ftrace_regex_lseek, .release = ftrace_filter_release, }; diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 492197e2f86..4e2f0341037 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2606,6 +2606,19 @@ void ring_buffer_record_enable_cpu(struct ring_buffer *buffer, int cpu) } EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu); +/* + * The total entries in the ring buffer is the running counter + * of entries entered into the ring buffer, minus the sum of + * the entries read from the ring buffer and the number of + * entries that were overwritten. + */ +static inline unsigned long +rb_num_of_entries(struct ring_buffer_per_cpu *cpu_buffer) +{ + return local_read(&cpu_buffer->entries) - + (local_read(&cpu_buffer->overrun) + cpu_buffer->read); +} + /** * ring_buffer_entries_cpu - get the number of entries in a cpu buffer * @buffer: The ring buffer @@ -2614,16 +2627,13 @@ EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu); unsigned long ring_buffer_entries_cpu(struct ring_buffer *buffer, int cpu) { struct ring_buffer_per_cpu *cpu_buffer; - unsigned long ret; if (!cpumask_test_cpu(cpu, buffer->cpumask)) return 0; cpu_buffer = buffer->buffers[cpu]; - ret = (local_read(&cpu_buffer->entries) - local_read(&cpu_buffer->overrun)) - - cpu_buffer->read; - return ret; + return rb_num_of_entries(cpu_buffer); } EXPORT_SYMBOL_GPL(ring_buffer_entries_cpu); @@ -2684,8 +2694,7 @@ unsigned long ring_buffer_entries(struct ring_buffer *buffer) /* if you care about this being correct, lock the buffer */ for_each_buffer_cpu(buffer, cpu) { cpu_buffer = buffer->buffers[cpu]; - entries += (local_read(&cpu_buffer->entries) - - local_read(&cpu_buffer->overrun)) - cpu_buffer->read; + entries += rb_num_of_entries(cpu_buffer); } return entries; diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 31cc4cb0dbf..39c059ca670 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -9,7 +9,7 @@ #include <linux/kprobes.h> #include "trace.h" -static char *perf_trace_buf[4]; +static char __percpu *perf_trace_buf[PERF_NR_CONTEXTS]; /* * Force it to be aligned to unsigned long to avoid misaligned accesses @@ -24,7 +24,7 @@ static int total_ref_count; static int perf_trace_event_init(struct ftrace_event_call *tp_event, struct perf_event *p_event) { - struct hlist_head *list; + struct hlist_head __percpu *list; int ret = -ENOMEM; int cpu; @@ -42,11 +42,11 @@ static int perf_trace_event_init(struct ftrace_event_call *tp_event, tp_event->perf_events = list; if (!total_ref_count) { - char *buf; + char __percpu *buf; int i; - for (i = 0; i < 4; i++) { - buf = (char *)alloc_percpu(perf_trace_t); + for (i = 0; i < PERF_NR_CONTEXTS; i++) { + buf = (char __percpu *)alloc_percpu(perf_trace_t); if (!buf) goto fail; @@ -65,7 +65,7 @@ fail: if (!total_ref_count) { int i; - for (i = 0; i < 4; i++) { + for (i = 0; i < PERF_NR_CONTEXTS; i++) { free_percpu(perf_trace_buf[i]); perf_trace_buf[i] = NULL; } @@ -101,22 +101,26 @@ int perf_trace_init(struct perf_event *p_event) return ret; } -int perf_trace_enable(struct perf_event *p_event) +int perf_trace_add(struct perf_event *p_event, int flags) { struct ftrace_event_call *tp_event = p_event->tp_event; + struct hlist_head __percpu *pcpu_list; struct hlist_head *list; - list = tp_event->perf_events; - if (WARN_ON_ONCE(!list)) + pcpu_list = tp_event->perf_events; + if (WARN_ON_ONCE(!pcpu_list)) return -EINVAL; - list = this_cpu_ptr(list); + if (!(flags & PERF_EF_START)) + p_event->hw.state = PERF_HES_STOPPED; + + list = this_cpu_ptr(pcpu_list); hlist_add_head_rcu(&p_event->hlist_entry, list); return 0; } -void perf_trace_disable(struct perf_event *p_event) +void perf_trace_del(struct perf_event *p_event, int flags) { hlist_del_rcu(&p_event->hlist_entry); } @@ -142,7 +146,7 @@ void perf_trace_destroy(struct perf_event *p_event) tp_event->perf_events = NULL; if (!--total_ref_count) { - for (i = 0; i < 4; i++) { + for (i = 0; i < PERF_NR_CONTEXTS; i++) { free_percpu(perf_trace_buf[i]); perf_trace_buf[i] = NULL; } diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 4c758f14632..398c0e8b332 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -600,21 +600,29 @@ out: enum { FORMAT_HEADER = 1, - FORMAT_PRINTFMT = 2, + FORMAT_FIELD_SEPERATOR = 2, + FORMAT_PRINTFMT = 3, }; static void *f_next(struct seq_file *m, void *v, loff_t *pos) { struct ftrace_event_call *call = m->private; struct ftrace_event_field *field; - struct list_head *head; + struct list_head *common_head = &ftrace_common_fields; + struct list_head *head = trace_get_fields(call); (*pos)++; switch ((unsigned long)v) { case FORMAT_HEADER: - head = &ftrace_common_fields; + if (unlikely(list_empty(common_head))) + return NULL; + + field = list_entry(common_head->prev, + struct ftrace_event_field, link); + return field; + case FORMAT_FIELD_SEPERATOR: if (unlikely(list_empty(head))) return NULL; @@ -626,31 +634,10 @@ static void *f_next(struct seq_file *m, void *v, loff_t *pos) return NULL; } - head = trace_get_fields(call); - - /* - * To separate common fields from event fields, the - * LSB is set on the first event field. Clear it in case. - */ - v = (void *)((unsigned long)v & ~1L); - field = v; - /* - * If this is a common field, and at the end of the list, then - * continue with main list. - */ - if (field->link.prev == &ftrace_common_fields) { - if (unlikely(list_empty(head))) - return NULL; - field = list_entry(head->prev, struct ftrace_event_field, link); - /* Set the LSB to notify f_show to print an extra newline */ - field = (struct ftrace_event_field *) - ((unsigned long)field | 1); - return field; - } - - /* If we are done tell f_show to print the format */ - if (field->link.prev == head) + if (field->link.prev == common_head) + return (void *)FORMAT_FIELD_SEPERATOR; + else if (field->link.prev == head) return (void *)FORMAT_PRINTFMT; field = list_entry(field->link.prev, struct ftrace_event_field, link); @@ -688,22 +675,16 @@ static int f_show(struct seq_file *m, void *v) seq_printf(m, "format:\n"); return 0; + case FORMAT_FIELD_SEPERATOR: + seq_putc(m, '\n'); + return 0; + case FORMAT_PRINTFMT: seq_printf(m, "\nprint fmt: %s\n", call->print_fmt); return 0; } - /* - * To separate common fields from event fields, the - * LSB is set on the first event field. Clear it and - * print a newline if it is set. - */ - if ((unsigned long)v & 1) { - seq_putc(m, '\n'); - v = (void *)((unsigned long)v & ~1L); - } - field = v; /* diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 6f233698518..02c708ae0d4 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -15,15 +15,19 @@ #include "trace.h" #include "trace_output.h" +/* When set, irq functions will be ignored */ +static int ftrace_graph_skip_irqs; + struct fgraph_cpu_data { pid_t last_pid; int depth; + int depth_irq; int ignore; unsigned long enter_funcs[FTRACE_RETFUNC_DEPTH]; }; struct fgraph_data { - struct fgraph_cpu_data *cpu_data; + struct fgraph_cpu_data __percpu *cpu_data; /* Place to preserve last processed entry. */ struct ftrace_graph_ent_entry ent; @@ -41,6 +45,7 @@ struct fgraph_data { #define TRACE_GRAPH_PRINT_PROC 0x8 #define TRACE_GRAPH_PRINT_DURATION 0x10 #define TRACE_GRAPH_PRINT_ABS_TIME 0x20 +#define TRACE_GRAPH_PRINT_IRQS 0x40 static struct tracer_opt trace_opts[] = { /* Display overruns? (for self-debug purpose) */ @@ -55,13 +60,15 @@ static struct tracer_opt trace_opts[] = { { TRACER_OPT(funcgraph-duration, TRACE_GRAPH_PRINT_DURATION) }, /* Display absolute time of an entry */ { TRACER_OPT(funcgraph-abstime, TRACE_GRAPH_PRINT_ABS_TIME) }, + /* Display interrupts */ + { TRACER_OPT(funcgraph-irqs, TRACE_GRAPH_PRINT_IRQS) }, { } /* Empty entry */ }; static struct tracer_flags tracer_flags = { /* Don't display overruns and proc by default */ .val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD | - TRACE_GRAPH_PRINT_DURATION, + TRACE_GRAPH_PRINT_DURATION | TRACE_GRAPH_PRINT_IRQS, .opts = trace_opts }; @@ -204,6 +211,14 @@ int __trace_graph_entry(struct trace_array *tr, return 1; } +static inline int ftrace_graph_ignore_irqs(void) +{ + if (!ftrace_graph_skip_irqs) + return 0; + + return in_irq(); +} + int trace_graph_entry(struct ftrace_graph_ent *trace) { struct trace_array *tr = graph_array; @@ -218,7 +233,8 @@ int trace_graph_entry(struct ftrace_graph_ent *trace) return 0; /* trace it when it is-nested-in or is a function enabled. */ - if (!(trace->depth || ftrace_graph_addr(trace->func))) + if (!(trace->depth || ftrace_graph_addr(trace->func)) || + ftrace_graph_ignore_irqs()) return 0; local_irq_save(flags); @@ -855,6 +871,92 @@ print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s, return 0; } +/* + * Entry check for irq code + * + * returns 1 if + * - we are inside irq code + * - we just extered irq code + * + * retunns 0 if + * - funcgraph-interrupts option is set + * - we are not inside irq code + */ +static int +check_irq_entry(struct trace_iterator *iter, u32 flags, + unsigned long addr, int depth) +{ + int cpu = iter->cpu; + struct fgraph_data *data = iter->private; + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + + if (flags & TRACE_GRAPH_PRINT_IRQS) + return 0; + + /* + * We are inside the irq code + */ + if (*depth_irq >= 0) + return 1; + + if ((addr < (unsigned long)__irqentry_text_start) || + (addr >= (unsigned long)__irqentry_text_end)) + return 0; + + /* + * We are entering irq code. + */ + *depth_irq = depth; + return 1; +} + +/* + * Return check for irq code + * + * returns 1 if + * - we are inside irq code + * - we just left irq code + * + * returns 0 if + * - funcgraph-interrupts option is set + * - we are not inside irq code + */ +static int +check_irq_return(struct trace_iterator *iter, u32 flags, int depth) +{ + int cpu = iter->cpu; + struct fgraph_data *data = iter->private; + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + + if (flags & TRACE_GRAPH_PRINT_IRQS) + return 0; + + /* + * We are not inside the irq code. + */ + if (*depth_irq == -1) + return 0; + + /* + * We are inside the irq code, and this is returning entry. + * Let's not trace it and clear the entry depth, since + * we are out of irq code. + * + * This condition ensures that we 'leave the irq code' once + * we are out of the entry depth. Thus protecting us from + * the RETURN entry loss. + */ + if (*depth_irq >= depth) { + *depth_irq = -1; + return 1; + } + + /* + * We are inside the irq code, and this is not the entry. + */ + return 1; +} + static enum print_line_t print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s, struct trace_iterator *iter, u32 flags) @@ -865,6 +967,9 @@ print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s, static enum print_line_t ret; int cpu = iter->cpu; + if (check_irq_entry(iter, flags, call->func, call->depth)) + return TRACE_TYPE_HANDLED; + if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func, flags)) return TRACE_TYPE_PARTIAL_LINE; @@ -902,6 +1007,9 @@ print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s, int ret; int i; + if (check_irq_return(iter, flags, trace->depth)) + return TRACE_TYPE_HANDLED; + if (data) { struct fgraph_cpu_data *cpu_data; int cpu = iter->cpu; @@ -1210,9 +1318,12 @@ void graph_trace_open(struct trace_iterator *iter) pid_t *pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid); int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth); int *ignore = &(per_cpu_ptr(data->cpu_data, cpu)->ignore); + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + *pid = -1; *depth = 0; *ignore = 0; + *depth_irq = -1; } iter->private = data; @@ -1235,6 +1346,14 @@ void graph_trace_close(struct trace_iterator *iter) } } +static int func_graph_set_flag(u32 old_flags, u32 bit, int set) +{ + if (bit == TRACE_GRAPH_PRINT_IRQS) + ftrace_graph_skip_irqs = !set; + + return 0; +} + static struct trace_event_functions graph_functions = { .trace = print_graph_function_event, }; @@ -1261,6 +1380,7 @@ static struct tracer graph_trace __read_mostly = { .print_line = print_graph_function, .print_header = print_graph_headers, .flags = &tracer_flags, + .set_flag = func_graph_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_function_graph, #endif diff --git a/kernel/trace/trace_workqueue.c b/kernel/trace/trace_workqueue.c index a7cc3793baf..209b379a472 100644 --- a/kernel/trace/trace_workqueue.c +++ b/kernel/trace/trace_workqueue.c @@ -263,6 +263,11 @@ int __init trace_workqueue_early_init(void) { int ret, cpu; + for_each_possible_cpu(cpu) { + spin_lock_init(&workqueue_cpu_stat(cpu)->lock); + INIT_LIST_HEAD(&workqueue_cpu_stat(cpu)->list); + } + ret = register_trace_workqueue_insertion(probe_workqueue_insertion, NULL); if (ret) goto out; @@ -279,11 +284,6 @@ int __init trace_workqueue_early_init(void) if (ret) goto no_creation; - for_each_possible_cpu(cpu) { - spin_lock_init(&workqueue_cpu_stat(cpu)->lock); - INIT_LIST_HEAD(&workqueue_cpu_stat(cpu)->list); - } - return 0; no_creation: diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index c77f3eceea2..d6073a50a6c 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -25,6 +25,7 @@ #include <linux/err.h> #include <linux/slab.h> #include <linux/sched.h> +#include <linux/jump_label.h> extern struct tracepoint __start___tracepoints[]; extern struct tracepoint __stop___tracepoints[]; @@ -263,7 +264,13 @@ static void set_tracepoint(struct tracepoint_entry **entry, * is used. */ rcu_assign_pointer(elem->funcs, (*entry)->funcs); - elem->state = active; + if (!elem->state && active) { + enable_jump_label(&elem->state); + elem->state = active; + } else if (elem->state && !active) { + disable_jump_label(&elem->state); + elem->state = active; + } } /* @@ -277,7 +284,10 @@ static void disable_tracepoint(struct tracepoint *elem) if (elem->unregfunc && elem->state) elem->unregfunc(); - elem->state = 0; + if (elem->state) { + disable_jump_label(&elem->state); + elem->state = 0; + } rcu_assign_pointer(elem->funcs, NULL); } diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 7f9c3c52ecc..dc8e16824b5 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -43,7 +43,6 @@ static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved); static DEFINE_PER_CPU(struct perf_event *, watchdog_ev); #endif -static int __read_mostly did_panic; static int __initdata no_watchdog; @@ -187,18 +186,6 @@ static int is_softlockup(unsigned long touch_ts) return 0; } -static int -watchdog_panic(struct notifier_block *this, unsigned long event, void *ptr) -{ - did_panic = 1; - - return NOTIFY_DONE; -} - -static struct notifier_block panic_block = { - .notifier_call = watchdog_panic, -}; - #ifdef CONFIG_HARDLOCKUP_DETECTOR static struct perf_event_attr wd_hw_attr = { .type = PERF_TYPE_HARDWARE, @@ -371,14 +358,14 @@ static int watchdog_nmi_enable(int cpu) /* Try to register using hardware perf events */ wd_attr = &wd_hw_attr; wd_attr->sample_period = hw_nmi_get_sample_period(); - event = perf_event_create_kernel_counter(wd_attr, cpu, -1, watchdog_overflow_callback); + event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback); if (!IS_ERR(event)) { printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n"); goto out_save; } printk(KERN_ERR "NMI watchdog failed to create perf event on cpu%i: %p\n", cpu, event); - return -1; + return PTR_ERR(event); /* success path */ out_save: @@ -422,17 +409,19 @@ static int watchdog_prepare_cpu(int cpu) static int watchdog_enable(int cpu) { struct task_struct *p = per_cpu(softlockup_watchdog, cpu); + int err; /* enable the perf event */ - if (watchdog_nmi_enable(cpu) != 0) - return -1; + err = watchdog_nmi_enable(cpu); + if (err) + return err; /* create the watchdog thread */ if (!p) { p = kthread_create(watchdog, (void *)(unsigned long)cpu, "watchdog/%d", cpu); if (IS_ERR(p)) { printk(KERN_ERR "softlockup watchdog for %i failed\n", cpu); - return -1; + return PTR_ERR(p); } kthread_bind(p, cpu); per_cpu(watchdog_touch_ts, cpu) = 0; @@ -484,6 +473,9 @@ static void watchdog_disable_all_cpus(void) { int cpu; + if (no_watchdog) + return; + for_each_online_cpu(cpu) watchdog_disable(cpu); @@ -526,17 +518,16 @@ static int __cpuinit cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { int hotcpu = (unsigned long)hcpu; + int err = 0; switch (action) { case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: - if (watchdog_prepare_cpu(hotcpu)) - return NOTIFY_BAD; + err = watchdog_prepare_cpu(hotcpu); break; case CPU_ONLINE: case CPU_ONLINE_FROZEN: - if (watchdog_enable(hotcpu)) - return NOTIFY_BAD; + err = watchdog_enable(hotcpu); break; #ifdef CONFIG_HOTPLUG_CPU case CPU_UP_CANCELED: @@ -549,7 +540,7 @@ cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) break; #endif /* CONFIG_HOTPLUG_CPU */ } - return NOTIFY_OK; + return notifier_from_errno(err); } static struct notifier_block __cpuinitdata cpu_nfb = { @@ -565,13 +556,11 @@ static int __init spawn_watchdog_task(void) return 0; err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu); - WARN_ON(err == NOTIFY_BAD); + WARN_ON(notifier_to_errno(err)); cpu_callback(&cpu_nfb, CPU_ONLINE, cpu); register_cpu_notifier(&cpu_nfb); - atomic_notifier_chain_register(&panic_notifier_list, &panic_block); - return 0; } early_initcall(spawn_watchdog_task); |