From bd3bff9e20f454b242d979ec2f9a4dca0d5fa06f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:41 +0200 Subject: sched: add latency tracer callbacks to the scheduler add 3 lightweight callbacks to the tracer backend. zero impact if tracing is turned off. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 5395a6176f4..717cab8a0c8 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2117,6 +2117,32 @@ static inline void arch_pick_mmap_layout(struct mm_struct *mm) } #endif +#ifdef CONFIG_CONTEXT_SWITCH_TRACER +extern void +ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next); +#else +static inline void +ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) +{ +} +#endif + +#ifdef CONFIG_SCHED_TRACER +extern void +ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr); +extern void +ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr); +#else +static inline void +ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +{ +} +static inline void +ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) +{ +} +#endif + extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); extern long sched_getaffinity(pid_t pid, cpumask_t *mask); -- cgit v1.2.3-70-g09d2 From 7c731e0a495e25e79dc1e9e68772a67a55721a65 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:41 +0200 Subject: ftrace: make the task state char-string visible to all The tracer wants to be able to convert the state number into a user visible character. This patch pulls that conversion string out the scheduler into the header. This way if it were to ever change, other parts of the kernel will know. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 2 ++ kernel/sched.c | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 717cab8a0c8..6e26f1fdbfe 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2237,6 +2237,8 @@ static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) } #endif /* CONFIG_MM_OWNER */ +#define TASK_STATE_TO_CHAR_STR "RSDTtZX" + #endif /* __KERNEL__ */ #endif diff --git a/kernel/sched.c b/kernel/sched.c index 463dcdb36ef..73e60085236 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5729,7 +5729,7 @@ out_unlock: return retval; } -static const char stat_nam[] = "RSDTtZX"; +static const char stat_nam[] = TASK_STATE_TO_CHAR_STR; void sched_show_task(struct task_struct *p) { -- cgit v1.2.3-70-g09d2 From 8ac0fca4ccb355ce50471d7aa3f10f5900b28b95 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: sched tracer fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 6 ------ kernel/sched.c | 2 +- kernel/trace/trace_sched_wakeup.c | 13 +++---------- 3 files changed, 4 insertions(+), 17 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 6e26f1fdbfe..05744f9cb09 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2130,17 +2130,11 @@ ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) #ifdef CONFIG_SCHED_TRACER extern void ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr); -extern void -ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr); #else static inline void ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) { } -static inline void -ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) -{ -} #endif extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); diff --git a/kernel/sched.c b/kernel/sched.c index 328494e28df..53ab1174664 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2613,7 +2613,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_new_task(p, rq->curr); + ftrace_wake_up_task(p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 87fa7b253b5..2a012423f9d 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -201,20 +201,13 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) +void wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) { if (likely(!tracer_enabled)) return; - wakeup_check_start(wakeup_trace, wakee, curr); -} - -void -ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) -{ - if (likely(!tracer_enabled)) - return; + tracing_record_cmdline(curr); + tracing_record_cmdline(wakee); wakeup_check_start(wakeup_trace, wakee, curr); } -- cgit v1.2.3-70-g09d2 From 4e65551905fb0300ae7e667cbaa41ee2e3f29a13 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: sched tracer, trace full rbtree Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 32 ++++++++++++++++------- kernel/sched.c | 35 ++++++++++++++++++++++--- kernel/trace/trace.c | 55 ++++++++++++++++----------------------- kernel/trace/trace.h | 14 ++++++++++ kernel/trace/trace_sched_switch.c | 24 +++++++++++------ 5 files changed, 108 insertions(+), 52 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 05744f9cb09..652d380ae56 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2119,20 +2119,34 @@ static inline void arch_pick_mmap_layout(struct mm_struct *mm) #ifdef CONFIG_CONTEXT_SWITCH_TRACER extern void -ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next); +ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next); +extern void +ftrace_wake_up_task(void *rq, struct task_struct *wakee, + struct task_struct *curr); +extern void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data); +extern void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3); #else static inline void -ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) +ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next) +{ +} +static inline void +sched_trace_special(unsigned long p1, unsigned long p2, unsigned long p3) +{ +} +static inline void +ftrace_wake_up_task(void *rq, struct task_struct *wakee, + struct task_struct *curr) +{ +} +static inline void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) { } -#endif - -#ifdef CONFIG_SCHED_TRACER -extern void -ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr); -#else static inline void -ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) { } #endif diff --git a/kernel/sched.c b/kernel/sched.c index 53ab1174664..b9208a0e33a 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2394,6 +2394,35 @@ static int sched_balance_self(int cpu, int flag) #endif /* CONFIG_SMP */ +#ifdef CONFIG_CONTEXT_SWITCH_TRACER + +void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) +{ + struct sched_entity *se; + struct task_struct *p; + struct rb_node *curr; + struct rq *rq = __rq; + + curr = first_fair(&rq->cfs); + if (!curr) + return; + + while (curr) { + se = rb_entry(curr, struct sched_entity, run_node); + if (!entity_is_task(se)) + continue; + + p = task_of(se); + + __trace_special(__tr, __data, + p->pid, p->se.vruntime, p->se.sum_exec_runtime); + + curr = rb_next(curr); + } +} + +#endif + /*** * try_to_wake_up - wake up a thread * @p: the to-be-woken-up thread @@ -2468,7 +2497,7 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) out_activate: #endif /* CONFIG_SMP */ - ftrace_wake_up_task(p, rq->curr); + ftrace_wake_up_task(rq, p, rq->curr); schedstat_inc(p, se.nr_wakeups); if (sync) schedstat_inc(p, se.nr_wakeups_sync); @@ -2613,7 +2642,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_task(p, rq->curr); + ftrace_wake_up_task(rq, p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) @@ -2786,7 +2815,7 @@ context_switch(struct rq *rq, struct task_struct *prev, struct mm_struct *mm, *oldmm; prepare_task_switch(rq, prev, next); - ftrace_ctx_switch(prev, next); + ftrace_ctx_switch(rq, prev, next); mm = next->mm; oldmm = prev->active_mm; /* diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0e4b7119e26..65173b14b91 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -66,7 +66,18 @@ static struct tracer *current_trace __read_mostly; static int max_tracer_type_len; static DEFINE_MUTEX(trace_types_lock); -static DECLARE_WAIT_QUEUE_HEAD (trace_wait); +static DECLARE_WAIT_QUEUE_HEAD(trace_wait); + +unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; + +/* + * FIXME: where should this be called? + */ +void trace_wake_up(void) +{ + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up(&trace_wait); +} #define ENTRIES_PER_PAGE (PAGE_SIZE / sizeof(struct trace_entry)) @@ -103,18 +114,6 @@ enum trace_flag_type { TRACE_FLAG_SOFTIRQ = 0x08, }; -enum trace_iterator_flags { - TRACE_ITER_PRINT_PARENT = 0x01, - TRACE_ITER_SYM_OFFSET = 0x02, - TRACE_ITER_SYM_ADDR = 0x04, - TRACE_ITER_VERBOSE = 0x08, - TRACE_ITER_RAW = 0x10, - TRACE_ITER_HEX = 0x20, - TRACE_ITER_BIN = 0x40, - TRACE_ITER_BLOCK = 0x80, - TRACE_ITER_STACKTRACE = 0x100, -}; - #define TRACE_ITER_SYM_MASK \ (TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR) @@ -132,8 +131,6 @@ static const char *trace_options[] = { NULL }; -static unsigned trace_flags = TRACE_ITER_PRINT_PARENT; - static DEFINE_SPINLOCK(ftrace_max_lock); /* @@ -660,9 +657,6 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } void @@ -673,10 +667,14 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } +#ifdef CONFIG_CONTEXT_SWITCH_TRACER + void -trace_special(struct trace_array *tr, struct trace_array_cpu *data, - unsigned long arg1, unsigned long arg2, unsigned long arg3) +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) { + struct trace_array_cpu *data = __data; + struct trace_array *tr = __tr; struct trace_entry *entry; unsigned long irq_flags; @@ -688,11 +686,10 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, entry->special.arg2 = arg2; entry->special.arg3 = arg3; spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } +#endif + void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, @@ -739,9 +736,6 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.next_prio = next->prio; __trace_stack(tr, data, flags, 4); spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } void @@ -765,9 +759,6 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_prio = wakee->prio; __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } #ifdef CONFIG_FTRACE @@ -1258,7 +1249,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) comm); break; case TRACE_SPECIAL: - trace_seq_printf(s, " %lx %lx %lx\n", + trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1344,7 +1335,7 @@ static int print_trace_fmt(struct trace_iterator *iter) return 0; break; case TRACE_SPECIAL: - ret = trace_seq_printf(s, " %lx %lx %lx\n", + ret = trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1409,7 +1400,7 @@ static int print_raw_fmt(struct trace_iterator *iter) break; case TRACE_SPECIAL: case TRACE_STACK: - ret = trace_seq_printf(s, " %lx %lx %lx\n", + ret = trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 387bdcf45e2..75e23747567 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -274,4 +274,18 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace, extern void *head_page(struct trace_array_cpu *data); +extern unsigned long trace_flags; + +enum trace_iterator_flags { + TRACE_ITER_PRINT_PARENT = 0x01, + TRACE_ITER_SYM_OFFSET = 0x02, + TRACE_ITER_SYM_ADDR = 0x04, + TRACE_ITER_VERBOSE = 0x08, + TRACE_ITER_RAW = 0x10, + TRACE_ITER_HEX = 0x20, + TRACE_ITER_BIN = 0x40, + TRACE_ITER_BLOCK = 0x80, + TRACE_ITER_STACKTRACE = 0x100, +}; + #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 8b1cf1a3aee..12658b3f2b2 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -18,7 +18,7 @@ static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; static void -ctx_switch_func(struct task_struct *prev, struct task_struct *next) +ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) { struct trace_array *tr = ctx_trace; struct trace_array_cpu *data; @@ -34,14 +34,17 @@ ctx_switch_func(struct task_struct *prev, struct task_struct *next) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) + if (likely(disabled == 1)) { tracing_sched_switch_trace(tr, data, prev, next, flags); + ftrace_all_fair_tasks(__rq, tr, data); + } atomic_dec(&data->disabled); local_irq_restore(flags); } -static void wakeup_func(struct task_struct *wakee, struct task_struct *curr) +static void +wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) { struct trace_array *tr = ctx_trace; struct trace_array_cpu *data; @@ -57,14 +60,18 @@ static void wakeup_func(struct task_struct *wakee, struct task_struct *curr) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) + if (likely(disabled == 1)) { tracing_sched_wakeup_trace(tr, data, wakee, curr, flags); + ftrace_all_fair_tasks(__rq, tr, data); + } atomic_dec(&data->disabled); local_irq_restore(flags); } -void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) +void +ftrace_ctx_switch(void *__rq, struct task_struct *prev, + struct task_struct *next) { tracing_record_cmdline(prev); @@ -72,7 +79,7 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) * If tracer_switch_func only points to the local * switch func, it still needs the ptr passed to it. */ - ctx_switch_func(prev, next); + ctx_switch_func(__rq, prev, next); /* * Chain to the wakeup tracer (this is a NOP if disabled): @@ -81,11 +88,12 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) } void -ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +ftrace_wake_up_task(void *__rq, struct task_struct *wakee, + struct task_struct *curr) { tracing_record_cmdline(curr); - wakeup_func(wakee, curr); + wakeup_func(__rq, wakee, curr); /* * Chain to the wakeup tracer (this is a NOP if disabled): -- cgit v1.2.3-70-g09d2 From 017730c11241e26577673eb9d957cfc66172ea91 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: fix wakeups Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 2 ++ kernel/sched.c | 18 ++++++++++++++++++ kernel/trace/trace.c | 15 +++++++++++---- 3 files changed, 31 insertions(+), 4 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 652d380ae56..a3970b56375 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -246,6 +246,8 @@ extern asmlinkage void schedule_tail(struct task_struct *prev); extern void init_idle(struct task_struct *idle, int cpu); extern void init_idle_bootup_task(struct task_struct *idle); +extern int runqueue_is_locked(void); + extern cpumask_t nohz_cpu_mask; #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) extern int select_nohz_load_balancer(int cpu); diff --git a/kernel/sched.c b/kernel/sched.c index 673b588b713..9ca4a2e6a23 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -642,6 +642,24 @@ static inline void update_rq_clock(struct rq *rq) # define const_debug static const #endif +/** + * runqueue_is_locked + * + * Returns true if the current cpu runqueue is locked. + * This interface allows printk to be called with the runqueue lock + * held and know whether or not it is OK to wake up the klogd. + */ +int runqueue_is_locked(void) +{ + int cpu = get_cpu(); + struct rq *rq = cpu_rq(cpu); + int ret; + + ret = spin_is_locked(&rq->lock); + put_cpu(); + return ret; +} + /* * Debugging: various feature bits */ diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 65173b14b91..2ca9d66aa74 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -70,12 +70,13 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; -/* - * FIXME: where should this be called? - */ void trace_wake_up(void) { - if (!(trace_flags & TRACE_ITER_BLOCK)) + /* + * The runqueue_is_locked() can fail, but this is the best we + * have for now: + */ + if (!(trace_flags & TRACE_ITER_BLOCK) && !runqueue_is_locked()) wake_up(&trace_wait); } @@ -657,6 +658,8 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } void @@ -686,6 +689,8 @@ __trace_special(void *__tr, void *__data, entry->special.arg2 = arg2; entry->special.arg3 = arg3; spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } #endif @@ -759,6 +764,8 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_prio = wakee->prio; __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } #ifdef CONFIG_FTRACE -- cgit v1.2.3-70-g09d2 From 1a3c3034336320554a3342572dae98d69e054fc7 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: fix __trace_special() Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 20 ++++++++++++-------- kernel/trace/trace.c | 4 ---- 2 files changed, 12 insertions(+), 12 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index a3970b56375..5b186bed54b 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2119,6 +2119,18 @@ static inline void arch_pick_mmap_layout(struct mm_struct *mm) } #endif +#ifdef CONFIG_TRACING +extern void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3); +#else +static inline void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ +} +#endif + #ifdef CONFIG_CONTEXT_SWITCH_TRACER extern void ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next); @@ -2126,9 +2138,6 @@ extern void ftrace_wake_up_task(void *rq, struct task_struct *wakee, struct task_struct *curr); extern void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data); -extern void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3); #else static inline void ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next) @@ -2146,11 +2155,6 @@ ftrace_wake_up_task(void *rq, struct task_struct *wakee, static inline void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) { } -static inline void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ -} #endif extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 2ca9d66aa74..65d2c0a61ed 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -670,8 +670,6 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } -#ifdef CONFIG_CONTEXT_SWITCH_TRACER - void __trace_special(void *__tr, void *__data, unsigned long arg1, unsigned long arg2, unsigned long arg3) @@ -693,8 +691,6 @@ __trace_special(void *__tr, void *__data, trace_wake_up(); } -#endif - void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, -- cgit v1.2.3-70-g09d2 From 88a4216c3ec4281fc7e6725cc3a3ccd01fb1aa14 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: sched special Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 6 ++++++ kernel/sched_fair.c | 3 +++ kernel/trace/trace.c | 6 +++--- kernel/trace/trace_sched_switch.c | 24 ++++++++++++++++++++++++ 4 files changed, 36 insertions(+), 3 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 5b186bed54b..360ca99033d 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2138,6 +2138,8 @@ extern void ftrace_wake_up_task(void *rq, struct task_struct *wakee, struct task_struct *curr); extern void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data); +extern void +ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3); #else static inline void ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next) @@ -2155,6 +2157,10 @@ ftrace_wake_up_task(void *rq, struct task_struct *wakee, static inline void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) { } +static inline void +ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ +} #endif extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index e24ecd39c4b..dc1856f1079 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1061,6 +1061,8 @@ wake_affine(struct rq *rq, struct sched_domain *this_sd, struct rq *this_rq, if (!(this_sd->flags & SD_WAKE_AFFINE)) return 0; + ftrace_special(__LINE__, curr->se.avg_overlap, sync); + ftrace_special(__LINE__, p->se.avg_overlap, -1); /* * If the currently running task will sleep within * a reasonable amount of time then attract this newly @@ -1238,6 +1240,7 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p) if (unlikely(se == pse)) return; + ftrace_special(__LINE__, p->pid, se->last_wakeup); cfs_rq_of(pse)->next = pse; /* diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3a4032492fc..b87a2641489 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1251,7 +1251,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) comm); break; case TRACE_SPECIAL: - trace_seq_printf(s, " %ld %ld %ld\n", + trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1335,7 +1335,7 @@ static int print_trace_fmt(struct trace_iterator *iter) return 0; break; case TRACE_SPECIAL: - ret = trace_seq_printf(s, " %ld %ld %ld\n", + ret = trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1400,7 +1400,7 @@ static int print_raw_fmt(struct trace_iterator *iter) break; case TRACE_SPECIAL: case TRACE_STACK: - ret = trace_seq_printf(s, " %ld %ld %ld\n", + ret = trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 5a217e86358..bddf676914e 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -103,6 +103,30 @@ ftrace_wake_up_task(void *__rq, struct task_struct *wakee, wakeup_sched_wakeup(wakee, curr); } +void +ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ + struct trace_array *tr = ctx_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (!tracer_enabled) + return; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + __trace_special(tr, data, arg1, arg2, arg3); + + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + static void sched_switch_reset(struct trace_array *tr) { int cpu; -- cgit v1.2.3-70-g09d2 From 5b82a1b08a00b2adca3d9dd9777efff40b7aaaa1 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 12 May 2008 21:21:10 +0200 Subject: Port ftrace to markers Porting ftrace to the marker infrastructure. Don't need to chain to the wakeup tracer from the sched tracer, because markers support multiple probes connected. Signed-off-by: Mathieu Desnoyers CC: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 32 ------- kernel/sched.c | 14 +++- kernel/trace/trace.h | 20 +---- kernel/trace/trace_sched_switch.c | 171 +++++++++++++++++++++++++++++++------- kernel/trace/trace_sched_wakeup.c | 106 +++++++++++++++++++++-- 5 files changed, 255 insertions(+), 88 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 360ca99033d..c0b1c69b55c 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2131,38 +2131,6 @@ __trace_special(void *__tr, void *__data, } #endif -#ifdef CONFIG_CONTEXT_SWITCH_TRACER -extern void -ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next); -extern void -ftrace_wake_up_task(void *rq, struct task_struct *wakee, - struct task_struct *curr); -extern void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data); -extern void -ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3); -#else -static inline void -ftrace_ctx_switch(void *rq, struct task_struct *prev, struct task_struct *next) -{ -} -static inline void -sched_trace_special(unsigned long p1, unsigned long p2, unsigned long p3) -{ -} -static inline void -ftrace_wake_up_task(void *rq, struct task_struct *wakee, - struct task_struct *curr) -{ -} -static inline void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) -{ -} -static inline void -ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ -} -#endif - extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); extern long sched_getaffinity(pid_t pid, cpumask_t *mask); diff --git a/kernel/sched.c b/kernel/sched.c index ad95cca4e42..e2e985eeee7 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2500,7 +2500,9 @@ out_activate: success = 1; out_running: - ftrace_wake_up_task(rq, p, rq->curr); + trace_mark(kernel_sched_wakeup, + "pid %d state %ld ## rq %p task %p rq->curr %p", + p->pid, p->state, rq, p, rq->curr); check_preempt_curr(rq, p); p->state = TASK_RUNNING; @@ -2631,7 +2633,9 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_task(rq, p, rq->curr); + trace_mark(kernel_sched_wakeup_new, + "pid %d state %ld ## rq %p task %p rq->curr %p", + p->pid, p->state, rq, p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) @@ -2804,7 +2808,11 @@ context_switch(struct rq *rq, struct task_struct *prev, struct mm_struct *mm, *oldmm; prepare_task_switch(rq, prev, next); - ftrace_ctx_switch(rq, prev, next); + trace_mark(kernel_sched_schedule, + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + prev->pid, next->pid, prev->state, + rq, prev, next); mm = next->mm; oldmm = prev->active_mm; /* diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 8845033ab49..f5de0601b40 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -234,25 +234,10 @@ void update_max_tr_single(struct trace_array *tr, extern cycle_t ftrace_now(int cpu); -#ifdef CONFIG_SCHED_TRACER -extern void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next); -extern void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr); -#else -static inline void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) -{ -} -static inline void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) -{ -} -#endif - #ifdef CONFIG_CONTEXT_SWITCH_TRACER typedef void (*tracer_switch_func_t)(void *private, + void *__rq, struct task_struct *prev, struct task_struct *next); @@ -262,9 +247,6 @@ struct tracer_switch_ops { struct tracer_switch_ops *next; }; -extern int register_tracer_switch(struct tracer_switch_ops *ops); -extern int unregister_tracer_switch(struct tracer_switch_ops *ops); - #endif /* CONFIG_CONTEXT_SWITCH_TRACER */ #ifdef CONFIG_DYNAMIC_FTRACE diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index a3376478fc2..d25ffa5eaf2 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -16,11 +16,14 @@ static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; +static atomic_t sched_ref; static void -ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) +sched_switch_func(void *private, void *__rq, struct task_struct *prev, + struct task_struct *next) { - struct trace_array *tr = ctx_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; unsigned long flags; long disabled; @@ -41,10 +44,40 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) local_irq_restore(flags); } +static notrace void +sched_switch_callback(void *probe_data, void *call_data, + const char *format, va_list *args) +{ + struct task_struct *prev; + struct task_struct *next; + struct rq *__rq; + + if (!atomic_read(&sched_ref)) + return; + + /* skip prev_pid %d next_pid %d prev_state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, int); + (void)va_arg(*args, long); + __rq = va_arg(*args, typeof(__rq)); + prev = va_arg(*args, typeof(prev)); + next = va_arg(*args, typeof(next)); + + tracing_record_cmdline(prev); + + /* + * If tracer_switch_func only points to the local + * switch func, it still needs the ptr passed to it. + */ + sched_switch_func(probe_data, __rq, prev, next); +} + static void -wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) +wakeup_func(void *private, void *__rq, struct task_struct *wakee, struct + task_struct *curr) { - struct trace_array *tr = ctx_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; unsigned long flags; long disabled; @@ -67,35 +100,29 @@ wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) local_irq_restore(flags); } -void -ftrace_ctx_switch(void *__rq, struct task_struct *prev, - struct task_struct *next) +static notrace void +wake_up_callback(void *probe_data, void *call_data, + const char *format, va_list *args) { - if (unlikely(atomic_read(&trace_record_cmdline_enabled))) - tracing_record_cmdline(prev); + struct task_struct *curr; + struct task_struct *task; + struct rq *__rq; - /* - * If tracer_switch_func only points to the local - * switch func, it still needs the ptr passed to it. - */ - ctx_switch_func(__rq, prev, next); + if (likely(!tracer_enabled)) + return; - /* - * Chain to the wakeup tracer (this is a NOP if disabled): - */ - wakeup_sched_switch(prev, next); -} + /* Skip pid %d state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, long); + /* now get the meat: "rq %p task %p rq->curr %p" */ + __rq = va_arg(*args, typeof(__rq)); + task = va_arg(*args, typeof(task)); + curr = va_arg(*args, typeof(curr)); -void -ftrace_wake_up_task(void *__rq, struct task_struct *wakee, - struct task_struct *curr) -{ - wakeup_func(__rq, wakee, curr); + tracing_record_cmdline(task); + tracing_record_cmdline(curr); - /* - * Chain to the wakeup tracer (this is a NOP if disabled): - */ - wakeup_sched_wakeup(wakee, curr); + wakeup_func(probe_data, __rq, task, curr); } void @@ -132,15 +159,95 @@ static void sched_switch_reset(struct trace_array *tr) tracing_reset(tr->data[cpu]); } +static int tracing_sched_register(void) +{ + int ret; + + ret = marker_probe_register("kernel_sched_wakeup", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &ctx_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup\n"); + return ret; + } + + ret = marker_probe_register("kernel_sched_wakeup_new", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &ctx_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup_new\n"); + goto fail_deprobe; + } + + ret = marker_probe_register("kernel_sched_schedule", + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + sched_switch_callback, + &ctx_trace); + if (ret) { + pr_info("sched trace: Couldn't add marker" + " probe to kernel_sched_schedule\n"); + goto fail_deprobe_wake_new; + } + + return ret; +fail_deprobe_wake_new: + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &ctx_trace); +fail_deprobe: + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &ctx_trace); + return ret; +} + +static void tracing_sched_unregister(void) +{ + marker_probe_unregister("kernel_sched_schedule", + sched_switch_callback, + &ctx_trace); + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &ctx_trace); + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &ctx_trace); +} + +void tracing_start_sched_switch(void) +{ + long ref; + + ref = atomic_inc_return(&sched_ref); + if (ref == 1) + tracing_sched_register(); +} + +void tracing_stop_sched_switch(void) +{ + long ref; + + ref = atomic_dec_and_test(&sched_ref); + if (ref) + tracing_sched_unregister(); +} + static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); atomic_inc(&trace_record_cmdline_enabled); tracer_enabled = 1; + tracing_start_sched_switch(); } static void stop_sched_trace(struct trace_array *tr) { + tracing_stop_sched_switch(); atomic_dec(&trace_record_cmdline_enabled); tracer_enabled = 0; } @@ -181,6 +288,14 @@ static struct tracer sched_switch_trace __read_mostly = __init static int init_sched_switch_trace(void) { + int ret = 0; + + if (atomic_read(&sched_ref)) + ret = tracing_sched_register(); + if (ret) { + pr_info("error registering scheduler trace\n"); + return ret; + } return register_tracer(&sched_switch_trace); } device_initcall(init_sched_switch_trace); diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 5948011006b..5d2fb48e47f 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -15,6 +15,7 @@ #include #include #include +#include #include "trace.h" @@ -44,11 +45,13 @@ static int report_latency(cycle_t delta) return 1; } -void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) +static void notrace +wakeup_sched_switch(void *private, void *rq, struct task_struct *prev, + struct task_struct *next) { unsigned long latency = 0, t0 = 0, t1 = 0; - struct trace_array *tr = wakeup_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; cycle_t T0, T1, delta; unsigned long flags; @@ -113,6 +116,31 @@ out: atomic_dec(&tr->data[cpu]->disabled); } +static notrace void +sched_switch_callback(void *probe_data, void *call_data, + const char *format, va_list *args) +{ + struct task_struct *prev; + struct task_struct *next; + struct rq *__rq; + + /* skip prev_pid %d next_pid %d prev_state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, int); + (void)va_arg(*args, long); + __rq = va_arg(*args, typeof(__rq)); + prev = va_arg(*args, typeof(prev)); + next = va_arg(*args, typeof(next)); + + tracing_record_cmdline(prev); + + /* + * If tracer_switch_func only points to the local + * switch func, it still needs the ptr passed to it. + */ + wakeup_sched_switch(probe_data, __rq, prev, next); +} + static void __wakeup_reset(struct trace_array *tr) { struct trace_array_cpu *data; @@ -188,19 +216,68 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -void wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) +static notrace void +wake_up_callback(void *probe_data, void *call_data, + const char *format, va_list *args) { + struct trace_array **ptr = probe_data; + struct trace_array *tr = *ptr; + struct task_struct *curr; + struct task_struct *task; + struct rq *__rq; + if (likely(!tracer_enabled)) return; + /* Skip pid %d state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, long); + /* now get the meat: "rq %p task %p rq->curr %p" */ + __rq = va_arg(*args, typeof(__rq)); + task = va_arg(*args, typeof(task)); + curr = va_arg(*args, typeof(curr)); + + tracing_record_cmdline(task); tracing_record_cmdline(curr); - tracing_record_cmdline(wakee); - wakeup_check_start(wakeup_trace, wakee, curr); + wakeup_check_start(tr, task, curr); } static void start_wakeup_tracer(struct trace_array *tr) { + int ret; + + ret = marker_probe_register("kernel_sched_wakeup", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &wakeup_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup\n"); + return; + } + + ret = marker_probe_register("kernel_sched_wakeup_new", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &wakeup_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup_new\n"); + goto fail_deprobe; + } + + ret = marker_probe_register("kernel_sched_schedule", + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + sched_switch_callback, + &wakeup_trace); + if (ret) { + pr_info("sched trace: Couldn't add marker" + " probe to kernel_sched_schedule\n"); + goto fail_deprobe_wake_new; + } + wakeup_reset(tr); /* @@ -215,11 +292,28 @@ static void start_wakeup_tracer(struct trace_array *tr) tracer_enabled = 1; return; +fail_deprobe_wake_new: + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &wakeup_trace); +fail_deprobe: + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &wakeup_trace); } static void stop_wakeup_tracer(struct trace_array *tr) { tracer_enabled = 0; + marker_probe_unregister("kernel_sched_schedule", + sched_switch_callback, + &wakeup_trace); + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &wakeup_trace); + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &wakeup_trace); } static void wakeup_tracer_init(struct trace_array *tr) -- cgit v1.2.3-70-g09d2 From 554ec22f075d46e4363520a407d2b7eeb5dfdd43 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:03 +0200 Subject: namespacecheck: more sched.c fixes [ Stephen Rothwell : build fix ] Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 18 ------------------ 1 file changed, 18 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index ae0be3c6237..dc36c3aea01 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -134,7 +134,6 @@ extern unsigned long nr_running(void); extern unsigned long nr_uninterruptible(void); extern unsigned long nr_active(void); extern unsigned long nr_iowait(void); -extern unsigned long weighted_cpuload(const int cpu); struct seq_file; struct cfs_rq; @@ -823,23 +822,6 @@ extern int arch_reinit_sched_domains(void); #endif /* CONFIG_SMP */ -/* - * A runqueue laden with a single nice 0 task scores a weighted_cpuload of - * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a - * task of nice 0 or enough lower priority tasks to bring up the - * weighted_cpuload - */ -static inline int above_background_load(void) -{ - unsigned long cpu; - - for_each_online_cpu(cpu) { - if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE) - return 1; - } - return 0; -} - struct io_context; /* See blkdev.h */ #define NGROUPS_SMALL 32 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) -- cgit v1.2.3-70-g09d2 From c7aceaba042702538b23cf4e0de1b2891ad8e671 Mon Sep 17 00:00:00 2001 From: Richard Kennedy Date: Thu, 15 May 2008 12:09:15 +0100 Subject: sched: reorder task_struct to reduce padding on 64bit builds This patch removes 24 bytes of padding and allows 1 extra object per slab on my fedora based config. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index dc36c3aea01..ea2857b9959 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1021,6 +1021,7 @@ struct task_struct { #endif int prio, static_prio, normal_prio; + unsigned int rt_priority; const struct sched_class *sched_class; struct sched_entity se; struct sched_rt_entity rt; @@ -1104,7 +1105,6 @@ struct task_struct { int __user *set_child_tid; /* CLONE_CHILD_SETTID */ int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ - unsigned int rt_priority; cputime_t utime, stime, utimescaled, stimescaled; cputime_t gtime; cputime_t prev_utime, prev_stime; @@ -1123,12 +1123,12 @@ struct task_struct { gid_t gid,egid,sgid,fsgid; struct group_info *group_info; kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; - unsigned securebits; struct user_struct *user; + unsigned securebits; #ifdef CONFIG_KEYS + unsigned char jit_keyring; /* default keyring to attach requested keys to */ struct key *request_key_auth; /* assumed request_key authority */ struct key *thread_keyring; /* keyring private to this thread */ - unsigned char jit_keyring; /* default keyring to attach requested keys to */ #endif char comm[TASK_COMM_LEN]; /* executable name excluding path - access with [gs]et_task_comm (which lock @@ -1215,8 +1215,8 @@ struct task_struct { # define MAX_LOCK_DEPTH 48UL u64 curr_chain_key; int lockdep_depth; - struct held_lock held_locks[MAX_LOCK_DEPTH]; unsigned int lockdep_recursion; + struct held_lock held_locks[MAX_LOCK_DEPTH]; #endif /* journalling filesystem info */ @@ -1244,10 +1244,6 @@ struct task_struct { u64 acct_vm_mem1; /* accumulated virtual memory usage */ cputime_t acct_stimexpd;/* stime since last update */ #endif -#ifdef CONFIG_NUMA - struct mempolicy *mempolicy; - short il_next; -#endif #ifdef CONFIG_CPUSETS nodemask_t mems_allowed; int cpuset_mems_generation; @@ -1266,6 +1262,10 @@ struct task_struct { #endif struct list_head pi_state_list; struct futex_pi_state *pi_state_cache; +#endif +#ifdef CONFIG_NUMA + struct mempolicy *mempolicy; + short il_next; #endif atomic_t fs_excl; /* holding fs exclusive resources */ struct rcu_head rcu; -- cgit v1.2.3-70-g09d2 From 1f11eb6a8bc92536d9e93ead48fa3ffbd1478571 Mon Sep 17 00:00:00 2001 From: Gregory Haskins Date: Wed, 4 Jun 2008 15:04:05 -0400 Subject: sched: fix cpupri hotplug support The RT folks over at RedHat found an issue w.r.t. hotplug support which was traced to problems with the cpupri infrastructure in the scheduler: https://bugzilla.redhat.com/show_bug.cgi?id=449676 This bug affects 23-rt12+, 24-rtX, 25-rtX, and sched-devel. This patch applies to 25.4-rt4, though it should trivially apply to most cpupri enabled kernels mentioned above. It turned out that the issue was that offline cpus could get inadvertently registered with cpupri so that they were erroneously selected during migration decisions. The end result would be an OOPS as the offline cpu had tasks routed to it. This patch generalizes the old join/leave domain interface into an online/offline interface, and adjusts the root-domain/hotplug code to utilize it. I was able to easily reproduce the issue prior to this patch, and am no longer able to reproduce it after this patch. I can offline cpus indefinately and everything seems to be in working order. Thanks to Arnaldo (acme), Thomas, and Peter for doing the legwork to point me in the right direction. Also thank you to Peter for reviewing the early iterations of this patch. Signed-off-by: Gregory Haskins Cc: Peter Zijlstra Cc: Steven Rostedt Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Signed-off-by: Thomas Gleixner --- include/linux/sched.h | 4 ++-- kernel/sched.c | 54 ++++++++++++++++++++++++++++++++++++++------------- kernel/sched_rt.c | 24 +++++++++++++++++------ 3 files changed, 60 insertions(+), 22 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index ea2857b9959..d25acf600a3 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -903,8 +903,8 @@ struct sched_class { void (*set_cpus_allowed)(struct task_struct *p, const cpumask_t *newmask); - void (*join_domain)(struct rq *rq); - void (*leave_domain)(struct rq *rq); + void (*rq_online)(struct rq *rq); + void (*rq_offline)(struct rq *rq); void (*switched_from) (struct rq *this_rq, struct task_struct *task, int running); diff --git a/kernel/sched.c b/kernel/sched.c index dc0be113f41..f0ed81b7128 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -529,6 +529,7 @@ struct rq { int push_cpu; /* cpu of this runqueue: */ int cpu; + int online; struct task_struct *migration_thread; struct list_head migration_queue; @@ -1498,6 +1499,8 @@ static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) #endif #define sched_class_highest (&rt_sched_class) +#define for_each_class(class) \ + for (class = sched_class_highest; class; class = class->next) static inline void inc_load(struct rq *rq, const struct task_struct *p) { @@ -6065,6 +6068,36 @@ static void unregister_sched_domain_sysctl(void) } #endif +static void set_rq_online(struct rq *rq) +{ + if (!rq->online) { + const struct sched_class *class; + + cpu_set(rq->cpu, rq->rd->online); + rq->online = 1; + + for_each_class(class) { + if (class->rq_online) + class->rq_online(rq); + } + } +} + +static void set_rq_offline(struct rq *rq) +{ + if (rq->online) { + const struct sched_class *class; + + for_each_class(class) { + if (class->rq_offline) + class->rq_offline(rq); + } + + cpu_clear(rq->cpu, rq->rd->online); + rq->online = 0; + } +} + /* * migration_call - callback that gets triggered when a CPU is added. * Here we can start up the necessary migration thread for the new CPU. @@ -6102,7 +6135,8 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { BUG_ON(!cpu_isset(cpu, rq->rd->span)); - cpu_set(cpu, rq->rd->online); + + set_rq_online(rq); } spin_unlock_irqrestore(&rq->lock, flags); break; @@ -6163,7 +6197,7 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu) spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { BUG_ON(!cpu_isset(cpu, rq->rd->span)); - cpu_clear(cpu, rq->rd->online); + set_rq_offline(rq); } spin_unlock_irqrestore(&rq->lock, flags); break; @@ -6385,20 +6419,16 @@ sd_parent_degenerate(struct sched_domain *sd, struct sched_domain *parent) static void rq_attach_root(struct rq *rq, struct root_domain *rd) { unsigned long flags; - const struct sched_class *class; spin_lock_irqsave(&rq->lock, flags); if (rq->rd) { struct root_domain *old_rd = rq->rd; - for (class = sched_class_highest; class; class = class->next) { - if (class->leave_domain) - class->leave_domain(rq); - } + if (cpu_isset(rq->cpu, old_rd->online)) + set_rq_offline(rq); cpu_clear(rq->cpu, old_rd->span); - cpu_clear(rq->cpu, old_rd->online); if (atomic_dec_and_test(&old_rd->refcount)) kfree(old_rd); @@ -6409,12 +6439,7 @@ static void rq_attach_root(struct rq *rq, struct root_domain *rd) cpu_set(rq->cpu, rd->span); if (cpu_isset(rq->cpu, cpu_online_map)) - cpu_set(rq->cpu, rd->online); - - for (class = sched_class_highest; class; class = class->next) { - if (class->join_domain) - class->join_domain(rq); - } + set_rq_online(rq); spin_unlock_irqrestore(&rq->lock, flags); } @@ -7824,6 +7849,7 @@ void __init sched_init(void) rq->next_balance = jiffies; rq->push_cpu = 0; rq->cpu = i; + rq->online = 0; rq->migration_thread = NULL; INIT_LIST_HEAD(&rq->migration_queue); rq_attach_root(rq, &def_root_domain); diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 44b06d75416..e4821593d4d 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -12,6 +12,9 @@ static inline int rt_overloaded(struct rq *rq) static inline void rt_set_overload(struct rq *rq) { + if (!rq->online) + return; + cpu_set(rq->cpu, rq->rd->rto_mask); /* * Make sure the mask is visible before we set @@ -26,6 +29,9 @@ static inline void rt_set_overload(struct rq *rq) static inline void rt_clear_overload(struct rq *rq) { + if (!rq->online) + return; + /* the order here really doesn't matter */ atomic_dec(&rq->rd->rto_count); cpu_clear(rq->cpu, rq->rd->rto_mask); @@ -394,7 +400,10 @@ void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) if (rt_se_prio(rt_se) < rt_rq->highest_prio) { struct rq *rq = rq_of_rt_rq(rt_rq); rt_rq->highest_prio = rt_se_prio(rt_se); - cpupri_set(&rq->rd->cpupri, rq->cpu, rt_se_prio(rt_se)); + + if (rq->online) + cpupri_set(&rq->rd->cpupri, rq->cpu, + rt_se_prio(rt_se)); } #endif #ifdef CONFIG_SMP @@ -448,7 +457,10 @@ void dec_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq) if (rt_rq->highest_prio != highest_prio) { struct rq *rq = rq_of_rt_rq(rt_rq); - cpupri_set(&rq->rd->cpupri, rq->cpu, rt_rq->highest_prio); + + if (rq->online) + cpupri_set(&rq->rd->cpupri, rq->cpu, + rt_rq->highest_prio); } update_rt_migration(rq_of_rt_rq(rt_rq)); @@ -1154,7 +1166,7 @@ static void set_cpus_allowed_rt(struct task_struct *p, } /* Assumes rq->lock is held */ -static void join_domain_rt(struct rq *rq) +static void rq_online_rt(struct rq *rq) { if (rq->rt.overloaded) rt_set_overload(rq); @@ -1163,7 +1175,7 @@ static void join_domain_rt(struct rq *rq) } /* Assumes rq->lock is held */ -static void leave_domain_rt(struct rq *rq) +static void rq_offline_rt(struct rq *rq) { if (rq->rt.overloaded) rt_clear_overload(rq); @@ -1331,8 +1343,8 @@ static const struct sched_class rt_sched_class = { .load_balance = load_balance_rt, .move_one_task = move_one_task_rt, .set_cpus_allowed = set_cpus_allowed_rt, - .join_domain = join_domain_rt, - .leave_domain = leave_domain_rt, + .rq_online = rq_online_rt, + .rq_offline = rq_offline_rt, .pre_schedule = pre_schedule_rt, .post_schedule = post_schedule_rt, .task_wake_up = task_wake_up_rt, -- cgit v1.2.3-70-g09d2 From 9985b0bab332289f14837eff3c6e0bcc658b58f7 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Thu, 5 Jun 2008 12:57:11 -0700 Subject: sched: prevent bound kthreads from changing cpus_allowed Kthreads that have called kthread_bind() are bound to specific cpus, so other tasks should not be able to change their cpus_allowed from under them. Otherwise, it is possible to move kthreads, such as the migration or software watchdog threads, so they are not allowed access to the cpu they work on. Cc: Peter Zijlstra Cc: Paul Menage Cc: Paul Jackson Signed-off-by: David Rientjes Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + kernel/cpuset.c | 14 +++++++++++++- kernel/kthread.c | 1 + kernel/sched.c | 6 ++++++ 4 files changed, 21 insertions(+), 1 deletion(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index d25acf600a3..2db1485f865 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1486,6 +1486,7 @@ static inline void put_task_struct(struct task_struct *t) #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ +#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ diff --git a/kernel/cpuset.c b/kernel/cpuset.c index 6090d18b58a..b84354f4de3 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -1190,6 +1190,15 @@ static int cpuset_can_attach(struct cgroup_subsys *ss, if (cpus_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)) return -ENOSPC; + if (tsk->flags & PF_THREAD_BOUND) { + cpumask_t mask; + + mutex_lock(&callback_mutex); + mask = cs->cpus_allowed; + mutex_unlock(&callback_mutex); + if (!cpus_equal(tsk->cpus_allowed, mask)) + return -EINVAL; + } return security_task_setscheduler(tsk, 0, NULL); } @@ -1203,11 +1212,14 @@ static void cpuset_attach(struct cgroup_subsys *ss, struct mm_struct *mm; struct cpuset *cs = cgroup_cs(cont); struct cpuset *oldcs = cgroup_cs(oldcont); + int err; mutex_lock(&callback_mutex); guarantee_online_cpus(cs, &cpus); - set_cpus_allowed_ptr(tsk, &cpus); + err = set_cpus_allowed_ptr(tsk, &cpus); mutex_unlock(&callback_mutex); + if (err) + return; from = oldcs->mems_allowed; to = cs->mems_allowed; diff --git a/kernel/kthread.c b/kernel/kthread.c index bd1b9ea024e..97747cdd37c 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -180,6 +180,7 @@ void kthread_bind(struct task_struct *k, unsigned int cpu) set_task_cpu(k, cpu); k->cpus_allowed = cpumask_of_cpu(cpu); k->rt.nr_cpus_allowed = 1; + k->flags |= PF_THREAD_BOUND; } EXPORT_SYMBOL(kthread_bind); diff --git a/kernel/sched.c b/kernel/sched.c index e9c24a12865..164fe7fe0d8 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5563,6 +5563,12 @@ int set_cpus_allowed_ptr(struct task_struct *p, const cpumask_t *new_mask) goto out; } + if (unlikely((p->flags & PF_THREAD_BOUND) && p != current && + !cpus_equal(p->cpus_allowed, *new_mask))) { + ret = -EINVAL; + goto out; + } + if (p->sched_class->set_cpus_allowed) p->sched_class->set_cpus_allowed(p, new_mask); else { -- cgit v1.2.3-70-g09d2 From 961ccddd59d627b89bd3dc284b6517833bbdf25d Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Mon, 23 Jun 2008 13:55:38 +1000 Subject: sched: add new API sched_setscheduler_nocheck: add a flag to control access checks Hidehiro Kawai noticed that sched_setscheduler() can fail in stop_machine: it calls sched_setscheduler() from insmod, which can have CAP_SYS_MODULE without CAP_SYS_NICE. Two cases could have failed, so are changed to sched_setscheduler_nocheck: kernel/softirq.c:cpu_callback() - CPU hotplug callback kernel/stop_machine.c:__stop_machine_run() - Called from various places, including modprobe() Signed-off-by: Rusty Russell Cc: Jeremy Fitzhardinge Cc: Hidehiro Kawai Cc: Andrew Morton Cc: linux-mm@kvack.org Cc: sugita Cc: Satoshi OSHIMA Signed-off-by: Ingo Molnar --- include/linux/sched.h | 2 ++ kernel/sched.c | 48 ++++++++++++++++++++++++++++++++++++------------ kernel/softirq.c | 2 +- kernel/stop_machine.c | 2 +- 4 files changed, 40 insertions(+), 14 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index c5d3f847ca8..fe3b9b5d739 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1655,6 +1655,8 @@ extern int can_nice(const struct task_struct *p, const int nice); extern int task_curr(const struct task_struct *p); extern int idle_cpu(int cpu); extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); +extern int sched_setscheduler_nocheck(struct task_struct *, int, + struct sched_param *); extern struct task_struct *idle_task(int cpu); extern struct task_struct *curr_task(int cpu); extern void set_curr_task(int cpu, struct task_struct *p); diff --git a/kernel/sched.c b/kernel/sched.c index b048ad8a11a..8d7c246ab86 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4746,16 +4746,8 @@ __setscheduler(struct rq *rq, struct task_struct *p, int policy, int prio) set_load_weight(p); } -/** - * sched_setscheduler - change the scheduling policy and/or RT priority of a thread. - * @p: the task in question. - * @policy: new policy. - * @param: structure containing the new RT priority. - * - * NOTE that the task may be already dead. - */ -int sched_setscheduler(struct task_struct *p, int policy, - struct sched_param *param) +static int __sched_setscheduler(struct task_struct *p, int policy, + struct sched_param *param, bool user) { int retval, oldprio, oldpolicy = -1, on_rq, running; unsigned long flags; @@ -4787,7 +4779,7 @@ recheck: /* * Allow unprivileged RT tasks to decrease priority: */ - if (!capable(CAP_SYS_NICE)) { + if (user && !capable(CAP_SYS_NICE)) { if (rt_policy(policy)) { unsigned long rlim_rtprio; @@ -4823,7 +4815,8 @@ recheck: * Do not allow realtime tasks into groups that have no runtime * assigned. */ - if (rt_policy(policy) && task_group(p)->rt_bandwidth.rt_runtime == 0) + if (user + && rt_policy(policy) && task_group(p)->rt_bandwidth.rt_runtime == 0) return -EPERM; #endif @@ -4872,8 +4865,39 @@ recheck: return 0; } + +/** + * sched_setscheduler - change the scheduling policy and/or RT priority of a thread. + * @p: the task in question. + * @policy: new policy. + * @param: structure containing the new RT priority. + * + * NOTE that the task may be already dead. + */ +int sched_setscheduler(struct task_struct *p, int policy, + struct sched_param *param) +{ + return __sched_setscheduler(p, policy, param, true); +} EXPORT_SYMBOL_GPL(sched_setscheduler); +/** + * sched_setscheduler_nocheck - change the scheduling policy and/or RT priority of a thread from kernelspace. + * @p: the task in question. + * @policy: new policy. + * @param: structure containing the new RT priority. + * + * Just like sched_setscheduler, only don't bother checking if the + * current context has permission. For example, this is needed in + * stop_machine(): we create temporary high priority worker threads, + * but our caller might not have that capability. + */ +int sched_setscheduler_nocheck(struct task_struct *p, int policy, + struct sched_param *param) +{ + return __sched_setscheduler(p, policy, param, false); +} + static int do_sched_setscheduler(pid_t pid, int policy, struct sched_param __user *param) { diff --git a/kernel/softirq.c b/kernel/softirq.c index 36e06174004..afd9120c2fc 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -645,7 +645,7 @@ static int __cpuinit cpu_callback(struct notifier_block *nfb, p = per_cpu(ksoftirqd, hotcpu); per_cpu(ksoftirqd, hotcpu) = NULL; - sched_setscheduler(p, SCHED_FIFO, ¶m); + sched_setscheduler_nocheck(p, SCHED_FIFO, ¶m); kthread_stop(p); takeover_tasklets(hotcpu); break; diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index b7350bbfb07..ba9b2054ecb 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -187,7 +187,7 @@ struct task_struct *__stop_machine_run(int (*fn)(void *), void *data, struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 }; /* One high-prio thread per cpu. We'll do this one. */ - sched_setscheduler(p, SCHED_FIFO, ¶m); + sched_setscheduler_nocheck(p, SCHED_FIFO, ¶m); kthread_bind(p, cpu); wake_up_process(p); wait_for_completion(&smdata.done); -- cgit v1.2.3-70-g09d2 From c09595f63bb1909c5dc4dca288f4fe818561b5f3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:14 +0200 Subject: sched: revert revert of: fair-group: SMP-nice for group scheduling Try again.. Initial commit: 18d95a2832c1392a2d63227a7a6d433cb9f2037e Revert: 6363ca57c76b7b83639ca8c83fc285fa26a7880e Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 + kernel/sched.c | 430 ++++++++++++++++++++++++++++++++++++++++++++++---- kernel/sched_debug.c | 5 + kernel/sched_fair.c | 124 +++++++++------ kernel/sched_rt.c | 4 + 5 files changed, 489 insertions(+), 75 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index eaf821072db..97a58b622ee 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -765,6 +765,7 @@ struct sched_domain { struct sched_domain *child; /* bottom domain must be null terminated */ struct sched_group *groups; /* the balancing groups of the domain */ cpumask_t span; /* span of all CPUs in this domain */ + int first_cpu; /* cache of the first cpu in this domain */ unsigned long min_interval; /* Minimum balance interval ms */ unsigned long max_interval; /* Maximum balance interval ms */ unsigned int busy_factor; /* less balancing by factor if busy */ diff --git a/kernel/sched.c b/kernel/sched.c index f653af684fb..874b6da1543 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -403,6 +403,43 @@ struct cfs_rq { */ struct list_head leaf_cfs_rq_list; struct task_group *tg; /* group that "owns" this runqueue */ + +#ifdef CONFIG_SMP + unsigned long task_weight; + unsigned long shares; + /* + * We need space to build a sched_domain wide view of the full task + * group tree, in order to avoid depending on dynamic memory allocation + * during the load balancing we place this in the per cpu task group + * hierarchy. This limits the load balancing to one instance per cpu, + * but more should not be needed anyway. + */ + struct aggregate_struct { + /* + * load = weight(cpus) * f(tg) + * + * Where f(tg) is the recursive weight fraction assigned to + * this group. + */ + unsigned long load; + + /* + * part of the group weight distributed to this span. + */ + unsigned long shares; + + /* + * The sum of all runqueue weights within this span. + */ + unsigned long rq_weight; + + /* + * Weight contributed by tasks; this is the part we can + * influence by moving tasks around. + */ + unsigned long task_weight; + } aggregate; +#endif #endif }; @@ -1484,6 +1521,326 @@ static unsigned long source_load(int cpu, int type); static unsigned long target_load(int cpu, int type); static unsigned long cpu_avg_load_per_task(int cpu); static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); + +#ifdef CONFIG_FAIR_GROUP_SCHED + +/* + * Group load balancing. + * + * We calculate a few balance domain wide aggregate numbers; load and weight. + * Given the pictures below, and assuming each item has equal weight: + * + * root 1 - thread + * / | \ A - group + * A 1 B + * /|\ / \ + * C 2 D 3 4 + * | | + * 5 6 + * + * load: + * A and B get 1/3-rd of the total load. C and D get 1/3-rd of A's 1/3-rd, + * which equals 1/9-th of the total load. + * + * shares: + * The weight of this group on the selected cpus. + * + * rq_weight: + * Direct sum of all the cpu's their rq weight, e.g. A would get 3 while + * B would get 2. + * + * task_weight: + * Part of the rq_weight contributed by tasks; all groups except B would + * get 1, B gets 2. + */ + +static inline struct aggregate_struct * +aggregate(struct task_group *tg, struct sched_domain *sd) +{ + return &tg->cfs_rq[sd->first_cpu]->aggregate; +} + +typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); + +/* + * Iterate the full tree, calling @down when first entering a node and @up when + * leaving it for the final time. + */ +static +void aggregate_walk_tree(aggregate_func down, aggregate_func up, + struct sched_domain *sd) +{ + struct task_group *parent, *child; + + rcu_read_lock(); + parent = &root_task_group; +down: + (*down)(parent, sd); + list_for_each_entry_rcu(child, &parent->children, siblings) { + parent = child; + goto down; + +up: + continue; + } + (*up)(parent, sd); + + child = parent; + parent = parent->parent; + if (parent) + goto up; + rcu_read_unlock(); +} + +/* + * Calculate the aggregate runqueue weight. + */ +static +void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long rq_weight = 0; + unsigned long task_weight = 0; + int i; + + for_each_cpu_mask(i, sd->span) { + rq_weight += tg->cfs_rq[i]->load.weight; + task_weight += tg->cfs_rq[i]->task_weight; + } + + aggregate(tg, sd)->rq_weight = rq_weight; + aggregate(tg, sd)->task_weight = task_weight; +} + +/* + * Compute the weight of this group on the given cpus. + */ +static +void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long shares = 0; + int i; + + for_each_cpu_mask(i, sd->span) + shares += tg->cfs_rq[i]->shares; + + if ((!shares && aggregate(tg, sd)->rq_weight) || shares > tg->shares) + shares = tg->shares; + + aggregate(tg, sd)->shares = shares; +} + +/* + * Compute the load fraction assigned to this group, relies on the aggregate + * weight and this group's parent's load, i.e. top-down. + */ +static +void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long load; + + if (!tg->parent) { + int i; + + load = 0; + for_each_cpu_mask(i, sd->span) + load += cpu_rq(i)->load.weight; + + } else { + load = aggregate(tg->parent, sd)->load; + + /* + * shares is our weight in the parent's rq so + * shares/parent->rq_weight gives our fraction of the load + */ + load *= aggregate(tg, sd)->shares; + load /= aggregate(tg->parent, sd)->rq_weight + 1; + } + + aggregate(tg, sd)->load = load; +} + +static void __set_se_shares(struct sched_entity *se, unsigned long shares); + +/* + * Calculate and set the cpu's group shares. + */ +static void +__update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, + int tcpu) +{ + int boost = 0; + unsigned long shares; + unsigned long rq_weight; + + if (!tg->se[tcpu]) + return; + + rq_weight = tg->cfs_rq[tcpu]->load.weight; + + /* + * If there are currently no tasks on the cpu pretend there is one of + * average load so that when a new task gets to run here it will not + * get delayed by group starvation. + */ + if (!rq_weight) { + boost = 1; + rq_weight = NICE_0_LOAD; + } + + /* + * \Sum shares * rq_weight + * shares = ----------------------- + * \Sum rq_weight + * + */ + shares = aggregate(tg, sd)->shares * rq_weight; + shares /= aggregate(tg, sd)->rq_weight + 1; + + /* + * record the actual number of shares, not the boosted amount. + */ + tg->cfs_rq[tcpu]->shares = boost ? 0 : shares; + + if (shares < MIN_SHARES) + shares = MIN_SHARES; + else if (shares > MAX_SHARES) + shares = MAX_SHARES; + + __set_se_shares(tg->se[tcpu], shares); +} + +/* + * Re-adjust the weights on the cpu the task came from and on the cpu the + * task went to. + */ +static void +__move_group_shares(struct task_group *tg, struct sched_domain *sd, + int scpu, int dcpu) +{ + unsigned long shares; + + shares = tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; + + __update_group_shares_cpu(tg, sd, scpu); + __update_group_shares_cpu(tg, sd, dcpu); + + /* + * ensure we never loose shares due to rounding errors in the + * above redistribution. + */ + shares -= tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; + if (shares) + tg->cfs_rq[dcpu]->shares += shares; +} + +/* + * Because changing a group's shares changes the weight of the super-group + * we need to walk up the tree and change all shares until we hit the root. + */ +static void +move_group_shares(struct task_group *tg, struct sched_domain *sd, + int scpu, int dcpu) +{ + while (tg) { + __move_group_shares(tg, sd, scpu, dcpu); + tg = tg->parent; + } +} + +static +void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) +{ + unsigned long shares = aggregate(tg, sd)->shares; + int i; + + for_each_cpu_mask(i, sd->span) { + struct rq *rq = cpu_rq(i); + unsigned long flags; + + spin_lock_irqsave(&rq->lock, flags); + __update_group_shares_cpu(tg, sd, i); + spin_unlock_irqrestore(&rq->lock, flags); + } + + aggregate_group_shares(tg, sd); + + /* + * ensure we never loose shares due to rounding errors in the + * above redistribution. + */ + shares -= aggregate(tg, sd)->shares; + if (shares) { + tg->cfs_rq[sd->first_cpu]->shares += shares; + aggregate(tg, sd)->shares += shares; + } +} + +/* + * Calculate the accumulative weight and recursive load of each task group + * while walking down the tree. + */ +static +void aggregate_get_down(struct task_group *tg, struct sched_domain *sd) +{ + aggregate_group_weight(tg, sd); + aggregate_group_shares(tg, sd); + aggregate_group_load(tg, sd); +} + +/* + * Rebalance the cpu shares while walking back up the tree. + */ +static +void aggregate_get_up(struct task_group *tg, struct sched_domain *sd) +{ + aggregate_group_set_shares(tg, sd); +} + +static DEFINE_PER_CPU(spinlock_t, aggregate_lock); + +static void __init init_aggregate(void) +{ + int i; + + for_each_possible_cpu(i) + spin_lock_init(&per_cpu(aggregate_lock, i)); +} + +static int get_aggregate(struct sched_domain *sd) +{ + if (!spin_trylock(&per_cpu(aggregate_lock, sd->first_cpu))) + return 0; + + aggregate_walk_tree(aggregate_get_down, aggregate_get_up, sd); + return 1; +} + +static void put_aggregate(struct sched_domain *sd) +{ + spin_unlock(&per_cpu(aggregate_lock, sd->first_cpu)); +} + +static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) +{ + cfs_rq->shares = shares; +} + +#else + +static inline void init_aggregate(void) +{ +} + +static inline int get_aggregate(struct sched_domain *sd) +{ + return 0; +} + +static inline void put_aggregate(struct sched_domain *sd) +{ +} +#endif + #endif #include "sched_stats.h" @@ -1498,26 +1855,14 @@ static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); #define for_each_class(class) \ for (class = sched_class_highest; class; class = class->next) -static inline void inc_load(struct rq *rq, const struct task_struct *p) -{ - update_load_add(&rq->load, p->se.load.weight); -} - -static inline void dec_load(struct rq *rq, const struct task_struct *p) -{ - update_load_sub(&rq->load, p->se.load.weight); -} - -static void inc_nr_running(struct task_struct *p, struct rq *rq) +static void inc_nr_running(struct rq *rq) { rq->nr_running++; - inc_load(rq, p); } -static void dec_nr_running(struct task_struct *p, struct rq *rq) +static void dec_nr_running(struct rq *rq) { rq->nr_running--; - dec_load(rq, p); } static void set_load_weight(struct task_struct *p) @@ -1609,7 +1954,7 @@ static void activate_task(struct rq *rq, struct task_struct *p, int wakeup) rq->nr_uninterruptible--; enqueue_task(rq, p, wakeup); - inc_nr_running(p, rq); + inc_nr_running(rq); } /* @@ -1621,7 +1966,7 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep) rq->nr_uninterruptible++; dequeue_task(rq, p, sleep); - dec_nr_running(p, rq); + dec_nr_running(rq); } /** @@ -2274,7 +2619,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) * management (if any): */ p->sched_class->task_new(rq, p); - inc_nr_running(p, rq); + inc_nr_running(rq); } check_preempt_curr(rq, p); #ifdef CONFIG_SMP @@ -3265,9 +3610,12 @@ static int load_balance(int this_cpu, struct rq *this_rq, unsigned long imbalance; struct rq *busiest; unsigned long flags; + int unlock_aggregate; cpus_setall(*cpus); + unlock_aggregate = get_aggregate(sd); + /* * When power savings policy is enabled for the parent domain, idle * sibling can pick up load irrespective of busy siblings. In this case, @@ -3383,8 +3731,9 @@ redo: if (!ld_moved && !sd_idle && sd->flags & SD_SHARE_CPUPOWER && !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE)) - return -1; - return ld_moved; + ld_moved = -1; + + goto out; out_balanced: schedstat_inc(sd, lb_balanced[idle]); @@ -3399,8 +3748,13 @@ out_one_pinned: if (!sd_idle && sd->flags & SD_SHARE_CPUPOWER && !test_sd_parent(sd, SD_POWERSAVINGS_BALANCE)) - return -1; - return 0; + ld_moved = -1; + else + ld_moved = 0; +out: + if (unlock_aggregate) + put_aggregate(sd); + return ld_moved; } /* @@ -4588,10 +4942,8 @@ void set_user_nice(struct task_struct *p, long nice) goto out_unlock; } on_rq = p->se.on_rq; - if (on_rq) { + if (on_rq) dequeue_task(rq, p, 0); - dec_load(rq, p); - } p->static_prio = NICE_TO_PRIO(nice); set_load_weight(p); @@ -4601,7 +4953,6 @@ void set_user_nice(struct task_struct *p, long nice) if (on_rq) { enqueue_task(rq, p, 0); - inc_load(rq, p); /* * If the task increased its priority or is running and * lowered its priority, then reschedule its CPU: @@ -7016,6 +7367,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, ALLNODES); set_domain_attribute(sd, attr); sd->span = *cpu_map; + sd->first_cpu = first_cpu(sd->span); cpu_to_allnodes_group(i, cpu_map, &sd->groups, tmpmask); p = sd; sd_allnodes = 1; @@ -7026,6 +7378,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, NODE); set_domain_attribute(sd, attr); sched_domain_node_span(cpu_to_node(i), &sd->span); + sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7037,6 +7390,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, CPU); set_domain_attribute(sd, attr); sd->span = *nodemask; + sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7048,6 +7402,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, MC); set_domain_attribute(sd, attr); sd->span = cpu_coregroup_map(i); + sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7060,6 +7415,7 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, SIBLING); set_domain_attribute(sd, attr); sd->span = per_cpu(cpu_sibling_map, i); + sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7757,6 +8113,7 @@ void __init sched_init(void) } #ifdef CONFIG_SMP + init_aggregate(); init_defrootdomain(); #endif @@ -8322,14 +8679,11 @@ void sched_move_task(struct task_struct *tsk) #endif /* CONFIG_GROUP_SCHED */ #ifdef CONFIG_FAIR_GROUP_SCHED -static void set_se_shares(struct sched_entity *se, unsigned long shares) +static void __set_se_shares(struct sched_entity *se, unsigned long shares) { struct cfs_rq *cfs_rq = se->cfs_rq; - struct rq *rq = cfs_rq->rq; int on_rq; - spin_lock_irq(&rq->lock); - on_rq = se->on_rq; if (on_rq) dequeue_entity(cfs_rq, se, 0); @@ -8339,8 +8693,17 @@ static void set_se_shares(struct sched_entity *se, unsigned long shares) if (on_rq) enqueue_entity(cfs_rq, se, 0); +} - spin_unlock_irq(&rq->lock); +static void set_se_shares(struct sched_entity *se, unsigned long shares) +{ + struct cfs_rq *cfs_rq = se->cfs_rq; + struct rq *rq = cfs_rq->rq; + unsigned long flags; + + spin_lock_irqsave(&rq->lock, flags); + __set_se_shares(se, shares); + spin_unlock_irqrestore(&rq->lock, flags); } static DEFINE_MUTEX(shares_mutex); @@ -8379,8 +8742,13 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares) * w/o tripping rebalance_share or load_balance_fair. */ tg->shares = shares; - for_each_possible_cpu(i) + for_each_possible_cpu(i) { + /* + * force a rebalance + */ + cfs_rq_set_shares(tg->cfs_rq[i], 0); set_se_shares(tg->se[i], shares); + } /* * Enable load balance activity on this group, by inserting it back on diff --git a/kernel/sched_debug.c b/kernel/sched_debug.c index 8e077b9c91c..04394ccac88 100644 --- a/kernel/sched_debug.c +++ b/kernel/sched_debug.c @@ -167,6 +167,11 @@ void print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) #endif SEQ_printf(m, " .%-30s: %ld\n", "nr_spread_over", cfs_rq->nr_spread_over); +#ifdef CONFIG_FAIR_GROUP_SCHED +#ifdef CONFIG_SMP + SEQ_printf(m, " .%-30s: %lu\n", "shares", cfs_rq->shares); +#endif +#endif } void print_rt_rq(struct seq_file *m, int cpu, struct rt_rq *rt_rq) diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 2e197b8e43f..183388c4dea 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -567,10 +567,27 @@ update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se) * Scheduling class queueing methods: */ +#if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED +static void +add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) +{ + cfs_rq->task_weight += weight; +} +#else +static inline void +add_cfs_task_weight(struct cfs_rq *cfs_rq, unsigned long weight) +{ +} +#endif + static void account_entity_enqueue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_add(&cfs_rq->load, se->load.weight); + if (!parent_entity(se)) + inc_cpu_load(rq_of(cfs_rq), se->load.weight); + if (entity_is_task(se)) + add_cfs_task_weight(cfs_rq, se->load.weight); cfs_rq->nr_running++; se->on_rq = 1; list_add(&se->group_node, &cfs_rq->tasks); @@ -580,6 +597,10 @@ static void account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) { update_load_sub(&cfs_rq->load, se->load.weight); + if (!parent_entity(se)) + dec_cpu_load(rq_of(cfs_rq), se->load.weight); + if (entity_is_task(se)) + add_cfs_task_weight(cfs_rq, -se->load.weight); cfs_rq->nr_running--; se->on_rq = 0; list_del_init(&se->group_node); @@ -1372,75 +1393,90 @@ static struct task_struct *load_balance_next_fair(void *arg) return __load_balance_iterator(cfs_rq, cfs_rq->balance_iterator); } -#ifdef CONFIG_FAIR_GROUP_SCHED -static int cfs_rq_best_prio(struct cfs_rq *cfs_rq) +static unsigned long +__load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, + unsigned long max_load_move, struct sched_domain *sd, + enum cpu_idle_type idle, int *all_pinned, int *this_best_prio, + struct cfs_rq *cfs_rq) { - struct sched_entity *curr; - struct task_struct *p; - - if (!cfs_rq->nr_running || !first_fair(cfs_rq)) - return MAX_PRIO; - - curr = cfs_rq->curr; - if (!curr) - curr = __pick_next_entity(cfs_rq); + struct rq_iterator cfs_rq_iterator; - p = task_of(curr); + cfs_rq_iterator.start = load_balance_start_fair; + cfs_rq_iterator.next = load_balance_next_fair; + cfs_rq_iterator.arg = cfs_rq; - return p->prio; + return balance_tasks(this_rq, this_cpu, busiest, + max_load_move, sd, idle, all_pinned, + this_best_prio, &cfs_rq_iterator); } -#endif +#ifdef CONFIG_FAIR_GROUP_SCHED static unsigned long load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, unsigned long max_load_move, struct sched_domain *sd, enum cpu_idle_type idle, int *all_pinned, int *this_best_prio) { - struct cfs_rq *busy_cfs_rq; long rem_load_move = max_load_move; - struct rq_iterator cfs_rq_iterator; - - cfs_rq_iterator.start = load_balance_start_fair; - cfs_rq_iterator.next = load_balance_next_fair; + int busiest_cpu = cpu_of(busiest); + struct task_group *tg; - for_each_leaf_cfs_rq(busiest, busy_cfs_rq) { -#ifdef CONFIG_FAIR_GROUP_SCHED - struct cfs_rq *this_cfs_rq; + rcu_read_lock(); + list_for_each_entry(tg, &task_groups, list) { long imbalance; - unsigned long maxload; + unsigned long this_weight, busiest_weight; + long rem_load, max_load, moved_load; + + /* + * empty group + */ + if (!aggregate(tg, sd)->task_weight) + continue; + + rem_load = rem_load_move * aggregate(tg, sd)->rq_weight; + rem_load /= aggregate(tg, sd)->load + 1; + + this_weight = tg->cfs_rq[this_cpu]->task_weight; + busiest_weight = tg->cfs_rq[busiest_cpu]->task_weight; - this_cfs_rq = cpu_cfs_rq(busy_cfs_rq, this_cpu); + imbalance = (busiest_weight - this_weight) / 2; - imbalance = busy_cfs_rq->load.weight - this_cfs_rq->load.weight; - /* Don't pull if this_cfs_rq has more load than busy_cfs_rq */ - if (imbalance <= 0) + if (imbalance < 0) + imbalance = busiest_weight; + + max_load = max(rem_load, imbalance); + moved_load = __load_balance_fair(this_rq, this_cpu, busiest, + max_load, sd, idle, all_pinned, this_best_prio, + tg->cfs_rq[busiest_cpu]); + + if (!moved_load) continue; - /* Don't pull more than imbalance/2 */ - imbalance /= 2; - maxload = min(rem_load_move, imbalance); + move_group_shares(tg, sd, busiest_cpu, this_cpu); - *this_best_prio = cfs_rq_best_prio(this_cfs_rq); -#else -# define maxload rem_load_move -#endif - /* - * pass busy_cfs_rq argument into - * load_balance_[start|next]_fair iterators - */ - cfs_rq_iterator.arg = busy_cfs_rq; - rem_load_move -= balance_tasks(this_rq, this_cpu, busiest, - maxload, sd, idle, all_pinned, - this_best_prio, - &cfs_rq_iterator); + moved_load *= aggregate(tg, sd)->load; + moved_load /= aggregate(tg, sd)->rq_weight + 1; - if (rem_load_move <= 0) + rem_load_move -= moved_load; + if (rem_load_move < 0) break; } + rcu_read_unlock(); return max_load_move - rem_load_move; } +#else +static unsigned long +load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, + unsigned long max_load_move, + struct sched_domain *sd, enum cpu_idle_type idle, + int *all_pinned, int *this_best_prio) +{ + return __load_balance_fair(this_rq, this_cpu, busiest, + max_load_move, sd, idle, all_pinned, + this_best_prio, &busiest->cfs); +} +#endif static int move_one_task_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 6b4a6b5a416..765932d0399 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -670,6 +670,8 @@ static void enqueue_task_rt(struct rq *rq, struct task_struct *p, int wakeup) rt_se->timeout = 0; enqueue_rt_entity(rt_se); + + inc_cpu_load(rq, p->se.load.weight); } static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int sleep) @@ -678,6 +680,8 @@ static void dequeue_task_rt(struct rq *rq, struct task_struct *p, int sleep) update_curr_rt(rq); dequeue_rt_entity(rt_se); + + dec_cpu_load(rq, p->se.load.weight); } /* -- cgit v1.2.3-70-g09d2 From b6a86c746f5b708012809958462234d19e9c8177 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:18 +0200 Subject: sched: fix sched_domain aggregation Keeping the aggregate on the first cpu of the sched domain has two problems: - it could collide between different sched domains on different cpus - it could slow things down because of the remote accesses Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 1 - kernel/sched.c | 113 ++++++++++++++++++++++++-------------------------- kernel/sched_fair.c | 12 +++--- 3 files changed, 60 insertions(+), 66 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index 97a58b622ee..eaf821072db 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -765,7 +765,6 @@ struct sched_domain { struct sched_domain *child; /* bottom domain must be null terminated */ struct sched_group *groups; /* the balancing groups of the domain */ cpumask_t span; /* span of all CPUs in this domain */ - int first_cpu; /* cache of the first cpu in this domain */ unsigned long min_interval; /* Minimum balance interval ms */ unsigned long max_interval; /* Maximum balance interval ms */ unsigned int busy_factor; /* less balancing by factor if busy */ diff --git a/kernel/sched.c b/kernel/sched.c index 7d282c52bd4..160d3c209b8 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1480,12 +1480,12 @@ static int task_hot(struct task_struct *p, u64 now, struct sched_domain *sd); */ static inline struct aggregate_struct * -aggregate(struct task_group *tg, struct sched_domain *sd) +aggregate(struct task_group *tg, int cpu) { - return &tg->cfs_rq[sd->first_cpu]->aggregate; + return &tg->cfs_rq[cpu]->aggregate; } -typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); +typedef void (*aggregate_func)(struct task_group *, int, struct sched_domain *); /* * Iterate the full tree, calling @down when first entering a node and @up when @@ -1493,14 +1493,14 @@ typedef void (*aggregate_func)(struct task_group *, struct sched_domain *); */ static void aggregate_walk_tree(aggregate_func down, aggregate_func up, - struct sched_domain *sd) + int cpu, struct sched_domain *sd) { struct task_group *parent, *child; rcu_read_lock(); parent = &root_task_group; down: - (*down)(parent, sd); + (*down)(parent, cpu, sd); list_for_each_entry_rcu(child, &parent->children, siblings) { parent = child; goto down; @@ -1508,7 +1508,7 @@ down: up: continue; } - (*up)(parent, sd); + (*up)(parent, cpu, sd); child = parent; parent = parent->parent; @@ -1520,8 +1520,8 @@ up: /* * Calculate the aggregate runqueue weight. */ -static -void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_weight(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long rq_weight = 0; unsigned long task_weight = 0; @@ -1532,15 +1532,15 @@ void aggregate_group_weight(struct task_group *tg, struct sched_domain *sd) task_weight += tg->cfs_rq[i]->task_weight; } - aggregate(tg, sd)->rq_weight = rq_weight; - aggregate(tg, sd)->task_weight = task_weight; + aggregate(tg, cpu)->rq_weight = rq_weight; + aggregate(tg, cpu)->task_weight = task_weight; } /* * Compute the weight of this group on the given cpus. */ -static -void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long shares = 0; int i; @@ -1548,18 +1548,18 @@ void aggregate_group_shares(struct task_group *tg, struct sched_domain *sd) for_each_cpu_mask(i, sd->span) shares += tg->cfs_rq[i]->shares; - if ((!shares && aggregate(tg, sd)->rq_weight) || shares > tg->shares) + if ((!shares && aggregate(tg, cpu)->rq_weight) || shares > tg->shares) shares = tg->shares; - aggregate(tg, sd)->shares = shares; + aggregate(tg, cpu)->shares = shares; } /* * Compute the load fraction assigned to this group, relies on the aggregate * weight and this group's parent's load, i.e. top-down. */ -static -void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_load(struct task_group *tg, int cpu, struct sched_domain *sd) { unsigned long load; @@ -1571,17 +1571,17 @@ void aggregate_group_load(struct task_group *tg, struct sched_domain *sd) load += cpu_rq(i)->load.weight; } else { - load = aggregate(tg->parent, sd)->load; + load = aggregate(tg->parent, cpu)->load; /* * shares is our weight in the parent's rq so * shares/parent->rq_weight gives our fraction of the load */ - load *= aggregate(tg, sd)->shares; - load /= aggregate(tg->parent, sd)->rq_weight + 1; + load *= aggregate(tg, cpu)->shares; + load /= aggregate(tg->parent, cpu)->rq_weight + 1; } - aggregate(tg, sd)->load = load; + aggregate(tg, cpu)->load = load; } static void __set_se_shares(struct sched_entity *se, unsigned long shares); @@ -1590,8 +1590,8 @@ static void __set_se_shares(struct sched_entity *se, unsigned long shares); * Calculate and set the cpu's group shares. */ static void -__update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, - int tcpu) +__update_group_shares_cpu(struct task_group *tg, int cpu, + struct sched_domain *sd, int tcpu) { int boost = 0; unsigned long shares; @@ -1618,8 +1618,8 @@ __update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, * \Sum rq_weight * */ - shares = aggregate(tg, sd)->shares * rq_weight; - shares /= aggregate(tg, sd)->rq_weight + 1; + shares = aggregate(tg, cpu)->shares * rq_weight; + shares /= aggregate(tg, cpu)->rq_weight + 1; /* * record the actual number of shares, not the boosted amount. @@ -1639,15 +1639,15 @@ __update_group_shares_cpu(struct task_group *tg, struct sched_domain *sd, * task went to. */ static void -__move_group_shares(struct task_group *tg, struct sched_domain *sd, +__move_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd, int scpu, int dcpu) { unsigned long shares; shares = tg->cfs_rq[scpu]->shares + tg->cfs_rq[dcpu]->shares; - __update_group_shares_cpu(tg, sd, scpu); - __update_group_shares_cpu(tg, sd, dcpu); + __update_group_shares_cpu(tg, cpu, sd, scpu); + __update_group_shares_cpu(tg, cpu, sd, dcpu); /* * ensure we never loose shares due to rounding errors in the @@ -1663,19 +1663,19 @@ __move_group_shares(struct task_group *tg, struct sched_domain *sd, * we need to walk up the tree and change all shares until we hit the root. */ static void -move_group_shares(struct task_group *tg, struct sched_domain *sd, +move_group_shares(struct task_group *tg, int cpu, struct sched_domain *sd, int scpu, int dcpu) { while (tg) { - __move_group_shares(tg, sd, scpu, dcpu); + __move_group_shares(tg, cpu, sd, scpu, dcpu); tg = tg->parent; } } -static -void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_group_set_shares(struct task_group *tg, int cpu, struct sched_domain *sd) { - unsigned long shares = aggregate(tg, sd)->shares; + unsigned long shares = aggregate(tg, cpu)->shares; int i; for_each_cpu_mask(i, sd->span) { @@ -1683,20 +1683,20 @@ void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) unsigned long flags; spin_lock_irqsave(&rq->lock, flags); - __update_group_shares_cpu(tg, sd, i); + __update_group_shares_cpu(tg, cpu, sd, i); spin_unlock_irqrestore(&rq->lock, flags); } - aggregate_group_shares(tg, sd); + aggregate_group_shares(tg, cpu, sd); /* * ensure we never loose shares due to rounding errors in the * above redistribution. */ - shares -= aggregate(tg, sd)->shares; + shares -= aggregate(tg, cpu)->shares; if (shares) { - tg->cfs_rq[sd->first_cpu]->shares += shares; - aggregate(tg, sd)->shares += shares; + tg->cfs_rq[cpu]->shares += shares; + aggregate(tg, cpu)->shares += shares; } } @@ -1704,21 +1704,21 @@ void aggregate_group_set_shares(struct task_group *tg, struct sched_domain *sd) * Calculate the accumulative weight and recursive load of each task group * while walking down the tree. */ -static -void aggregate_get_down(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_get_down(struct task_group *tg, int cpu, struct sched_domain *sd) { - aggregate_group_weight(tg, sd); - aggregate_group_shares(tg, sd); - aggregate_group_load(tg, sd); + aggregate_group_weight(tg, cpu, sd); + aggregate_group_shares(tg, cpu, sd); + aggregate_group_load(tg, cpu, sd); } /* * Rebalance the cpu shares while walking back up the tree. */ -static -void aggregate_get_up(struct task_group *tg, struct sched_domain *sd) +static void +aggregate_get_up(struct task_group *tg, int cpu, struct sched_domain *sd) { - aggregate_group_set_shares(tg, sd); + aggregate_group_set_shares(tg, cpu, sd); } static DEFINE_PER_CPU(spinlock_t, aggregate_lock); @@ -1731,18 +1731,18 @@ static void __init init_aggregate(void) spin_lock_init(&per_cpu(aggregate_lock, i)); } -static int get_aggregate(struct sched_domain *sd) +static int get_aggregate(int cpu, struct sched_domain *sd) { - if (!spin_trylock(&per_cpu(aggregate_lock, sd->first_cpu))) + if (!spin_trylock(&per_cpu(aggregate_lock, cpu))) return 0; - aggregate_walk_tree(aggregate_get_down, aggregate_get_up, sd); + aggregate_walk_tree(aggregate_get_down, aggregate_get_up, cpu, sd); return 1; } -static void put_aggregate(struct sched_domain *sd) +static void put_aggregate(int cpu, struct sched_domain *sd) { - spin_unlock(&per_cpu(aggregate_lock, sd->first_cpu)); + spin_unlock(&per_cpu(aggregate_lock, cpu)); } static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares) @@ -1756,12 +1756,12 @@ static inline void init_aggregate(void) { } -static inline int get_aggregate(struct sched_domain *sd) +static inline int get_aggregate(int cpu, struct sched_domain *sd) { return 0; } -static inline void put_aggregate(struct sched_domain *sd) +static inline void put_aggregate(int cpu, struct sched_domain *sd) { } #endif @@ -3539,7 +3539,7 @@ static int load_balance(int this_cpu, struct rq *this_rq, cpus_setall(*cpus); - unlock_aggregate = get_aggregate(sd); + unlock_aggregate = get_aggregate(this_cpu, sd); /* * When power savings policy is enabled for the parent domain, idle @@ -3678,7 +3678,7 @@ out_one_pinned: ld_moved = 0; out: if (unlock_aggregate) - put_aggregate(sd); + put_aggregate(this_cpu, sd); return ld_moved; } @@ -7292,7 +7292,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, ALLNODES); set_domain_attribute(sd, attr); sd->span = *cpu_map; - sd->first_cpu = first_cpu(sd->span); cpu_to_allnodes_group(i, cpu_map, &sd->groups, tmpmask); p = sd; sd_allnodes = 1; @@ -7303,7 +7302,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, NODE); set_domain_attribute(sd, attr); sched_domain_node_span(cpu_to_node(i), &sd->span); - sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7315,7 +7313,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, CPU); set_domain_attribute(sd, attr); sd->span = *nodemask; - sd->first_cpu = first_cpu(sd->span); sd->parent = p; if (p) p->child = sd; @@ -7327,7 +7324,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, MC); set_domain_attribute(sd, attr); sd->span = cpu_coregroup_map(i); - sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; @@ -7340,7 +7336,6 @@ static int __build_sched_domains(const cpumask_t *cpu_map, SD_INIT(sd, SIBLING); set_domain_attribute(sd, attr); sd->span = per_cpu(cpu_sibling_map, i); - sd->first_cpu = first_cpu(sd->span); cpus_and(sd->span, sd->span, *cpu_map); sd->parent = p; p->child = sd; diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 509092af033..40cf24ab4de 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1429,11 +1429,11 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, /* * empty group */ - if (!aggregate(tg, sd)->task_weight) + if (!aggregate(tg, this_cpu)->task_weight) continue; - rem_load = rem_load_move * aggregate(tg, sd)->rq_weight; - rem_load /= aggregate(tg, sd)->load + 1; + rem_load = rem_load_move * aggregate(tg, this_cpu)->rq_weight; + rem_load /= aggregate(tg, this_cpu)->load + 1; this_weight = tg->cfs_rq[this_cpu]->task_weight; busiest_weight = tg->cfs_rq[busiest_cpu]->task_weight; @@ -1451,10 +1451,10 @@ load_balance_fair(struct rq *this_rq, int this_cpu, struct rq *busiest, if (!moved_load) continue; - move_group_shares(tg, sd, busiest_cpu, this_cpu); + move_group_shares(tg, this_cpu, sd, busiest_cpu, this_cpu); - moved_load *= aggregate(tg, sd)->load; - moved_load /= aggregate(tg, sd)->rq_weight + 1; + moved_load *= aggregate(tg, this_cpu)->load; + moved_load /= aggregate(tg, this_cpu)->rq_weight + 1; rem_load_move -= moved_load; if (rem_load_move < 0) -- cgit v1.2.3-70-g09d2 From 2398f2c6d34b43025f274fc42eaca34d23ec2320 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 27 Jun 2008 13:41:35 +0200 Subject: sched: update shares on wakeup We found that the affine wakeup code needs rather accurate load figures to be effective. The trouble is that updating the load figures is fairly expensive with group scheduling. Therefore ratelimit the updating. Signed-off-by: Peter Zijlstra Cc: Srivatsa Vaddagiri Cc: Mike Galbraith Signed-off-by: Ingo Molnar --- include/linux/sched.h | 3 +++ kernel/sched.c | 30 +++++++++++++++++++++++++++++- kernel/sched_features.h | 3 ++- kernel/sysctl.c | 8 ++++++++ 4 files changed, 42 insertions(+), 2 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index eaf821072db..835b6c6fcc5 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -783,6 +783,8 @@ struct sched_domain { unsigned int balance_interval; /* initialise to 1. units in ms. */ unsigned int nr_balance_failed; /* initialise to 0 */ + u64 last_update; + #ifdef CONFIG_SCHEDSTATS /* load_balance() stats */ unsigned int lb_count[CPU_MAX_IDLE_TYPES]; @@ -1605,6 +1607,7 @@ extern unsigned int sysctl_sched_child_runs_first; extern unsigned int sysctl_sched_features; extern unsigned int sysctl_sched_migration_cost; extern unsigned int sysctl_sched_nr_migrate; +extern unsigned int sysctl_sched_shares_ratelimit; int sched_nr_latency_handler(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *length, diff --git a/kernel/sched.c b/kernel/sched.c index 1cff969f664..62db0891025 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -777,6 +777,12 @@ late_initcall(sched_init_debug); */ const_debug unsigned int sysctl_sched_nr_migrate = 32; +/* + * ratelimit for updating the group shares. + * default: 0.5ms + */ +const_debug unsigned int sysctl_sched_shares_ratelimit = 500000; + /* * period over which we measure -rt task cpu usage in us. * default: 1s @@ -1590,7 +1596,13 @@ tg_nop(struct task_group *tg, int cpu, struct sched_domain *sd) static void update_shares(struct sched_domain *sd) { - walk_tg_tree(tg_nop, tg_shares_up, 0, sd); + u64 now = cpu_clock(raw_smp_processor_id()); + s64 elapsed = now - sd->last_update; + + if (elapsed >= (s64)(u64)sysctl_sched_shares_ratelimit) { + sd->last_update = now; + walk_tg_tree(tg_nop, tg_shares_up, 0, sd); + } } static void update_shares_locked(struct rq *rq, struct sched_domain *sd) @@ -2199,6 +2211,22 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) if (!sched_feat(SYNC_WAKEUPS)) sync = 0; +#ifdef CONFIG_SMP + if (sched_feat(LB_WAKEUP_UPDATE)) { + struct sched_domain *sd; + + this_cpu = raw_smp_processor_id(); + cpu = task_cpu(p); + + for_each_domain(this_cpu, sd) { + if (cpu_isset(cpu, sd->span)) { + update_shares(sd); + break; + } + } + } +#endif + smp_wmb(); rq = task_rq_lock(p, &flags); old_state = p->state; diff --git a/kernel/sched_features.h b/kernel/sched_features.h index d56e3053e74..7d616d2a2a3 100644 --- a/kernel/sched_features.h +++ b/kernel/sched_features.h @@ -8,4 +8,5 @@ SCHED_FEAT(SYNC_WAKEUPS, 1) SCHED_FEAT(HRTICK, 1) SCHED_FEAT(DOUBLE_TICK, 0) SCHED_FEAT(ASYM_GRAN, 1) -SCHED_FEAT(LB_BIAS, 0) \ No newline at end of file +SCHED_FEAT(LB_BIAS, 0) +SCHED_FEAT(LB_WAKEUP_UPDATE, 1) diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 29116652dca..fe8cdc80ff0 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -264,6 +264,14 @@ static struct ctl_table kern_table[] = { .extra1 = &min_wakeup_granularity_ns, .extra2 = &max_wakeup_granularity_ns, }, + { + .ctl_name = CTL_UNNUMBERED, + .procname = "sched_shares_ratelimit", + .data = &sysctl_sched_shares_ratelimit, + .maxlen = sizeof(unsigned int), + .mode = 0644, + .proc_handler = &proc_dointvec, + }, { .ctl_name = CTL_UNNUMBERED, .procname = "sched_child_runs_first", -- cgit v1.2.3-70-g09d2 From af52a90a14cdaa54ecbfb6e6982abb13466a4b56 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 7 Jul 2008 14:16:52 -0400 Subject: sched_clock: stop maximum check on NO HZ Working with ftrace I would get large jumps of 11 millisecs or more with the clock tracer. This killed the latencing timings of ftrace and also caused the irqoff self tests to fail. What was happening is with NO_HZ the idle would stop the jiffy counter and before the jiffy counter was updated the sched_clock would have a bad delta jiffies to compare with the gtod with the maximum. The jiffies would stop and the last sched_tick would record the last gtod. On wakeup, the sched clock update would compare the gtod + delta jiffies (which would be zero) and compare it to the TSC. The TSC would have correctly (with a stable TSC) moved forward several jiffies. But because the jiffies has not been updated yet the clock would be prevented from moving forward because it would appear that the TSC jumped too far ahead. The clock would then virtually stop, until the jiffies are updated. Then the next sched clock update would see that the clock was very much behind since the delta jiffies is now correct. This would then jump the clock forward by several jiffies. This caused ftrace to report several milliseconds of interrupts off latency at every resume from NO_HZ idle. This patch adds hooks into the nohz code to disable the checking of the maximum clock update when nohz is in effect. It resumes the max check when nohz has updated the jiffies again. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- include/linux/sched.h | 17 ++++++++++++++++- kernel/sched_clock.c | 39 ++++++++++++++++++++++++++++++++++++++- kernel/time/tick-sched.c | 2 ++ 3 files changed, 56 insertions(+), 2 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/sched.h b/include/linux/sched.h index c5d3f847ca8..33a8f42041f 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1573,13 +1573,28 @@ static inline void sched_clock_idle_sleep_event(void) static inline void sched_clock_idle_wakeup_event(u64 delta_ns) { } -#else + +#ifdef CONFIG_NO_HZ +static inline void sched_clock_tick_stop(int cpu) +{ +} + +static inline void sched_clock_tick_start(int cpu) +{ +} +#endif + +#else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ extern void sched_clock_init(void); extern u64 sched_clock_cpu(int cpu); extern void sched_clock_tick(void); extern void sched_clock_idle_sleep_event(void); extern void sched_clock_idle_wakeup_event(u64 delta_ns); +#ifdef CONFIG_NO_HZ +extern void sched_clock_tick_stop(int cpu); +extern void sched_clock_tick_start(int cpu); #endif +#endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ /* * For kernel-internal use: high-speed (but slightly incorrect) per-cpu diff --git a/kernel/sched_clock.c b/kernel/sched_clock.c index 42b81fa38cb..97159e225a7 100644 --- a/kernel/sched_clock.c +++ b/kernel/sched_clock.c @@ -45,6 +45,9 @@ struct sched_clock_data { u64 tick_raw; u64 tick_gtod; u64 clock; +#ifdef CONFIG_NO_HZ + int check_max; +#endif }; static DEFINE_PER_CPU_SHARED_ALIGNED(struct sched_clock_data, sched_clock_data); @@ -76,11 +79,45 @@ void sched_clock_init(void) scd->tick_raw = 0; scd->tick_gtod = ktime_now; scd->clock = ktime_now; +#ifdef CONFIG_NO_HZ + scd->check_max = 1; +#endif } sched_clock_running = 1; } +#ifdef CONFIG_NO_HZ +/* + * The dynamic ticks makes the delta jiffies inaccurate. This + * prevents us from checking the maximum time update. + * Disable the maximum check during stopped ticks. + */ +void sched_clock_tick_stop(int cpu) +{ + struct sched_clock_data *scd = cpu_sdc(cpu); + + scd->check_max = 0; +} + +void sched_clock_tick_start(int cpu) +{ + struct sched_clock_data *scd = cpu_sdc(cpu); + + scd->check_max = 1; +} + +static int check_max(struct sched_clock_data *scd) +{ + return scd->check_max; +} +#else +static int check_max(struct sched_clock_data *scd) +{ + return 1; +} +#endif /* CONFIG_NO_HZ */ + /* * update the percpu scd from the raw @now value * @@ -112,7 +149,7 @@ static void __update_sched_clock(struct sched_clock_data *scd, u64 now) */ max_clock = scd->tick_gtod + (2 + delta_jiffies) * TICK_NSEC; - if (unlikely(clock + delta > max_clock)) { + if (unlikely(clock + delta > max_clock) && check_max(scd)) { if (clock < max_clock) clock = max_clock; else diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index b854a895591..d63008b09a4 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -276,6 +276,7 @@ void tick_nohz_stop_sched_tick(void) ts->tick_stopped = 1; ts->idle_jiffies = last_jiffies; rcu_enter_nohz(); + sched_clock_tick_stop(cpu); } /* @@ -375,6 +376,7 @@ void tick_nohz_restart_sched_tick(void) select_nohz_load_balancer(0); now = ktime_get(); tick_do_update_jiffies64(now); + sched_clock_tick_start(cpu); cpu_clear(cpu, nohz_cpu_mask); /* -- cgit v1.2.3-70-g09d2 From ebb12db51f6c13b30752fcf506baad4c617b153c Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Wed, 11 Jun 2008 22:04:29 +0200 Subject: Freezer: Introduce PF_FREEZER_NOSIG The freezer currently attempts to distinguish kernel threads from user space tasks by checking if their mm pointer is unset and it does not send fake signals to kernel threads. However, there are kernel threads, mostly related to networking, that behave like user space tasks and may want to be sent a fake signal to be frozen. Introduce the new process flag PF_FREEZER_NOSIG that will be set by default for all kernel threads and make the freezer only send fake signals to the tasks having PF_FREEZER_NOSIG unset. Provide the set_freezable_with_signal() function to be called by the kernel threads that want to be sent a fake signal for freezing. This patch should not change the freezer's observable behavior. Signed-off-by: Rafael J. Wysocki Signed-off-by: Andi Kleen Acked-by: Pavel Machek Signed-off-by: Len Brown --- include/linux/freezer.h | 10 +++++ include/linux/sched.h | 1 + kernel/kthread.c | 2 +- kernel/power/process.c | 97 +++++++++++++++++++++---------------------------- 4 files changed, 54 insertions(+), 56 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/freezer.h b/include/linux/freezer.h index 08934995c7a..deddeedf325 100644 --- a/include/linux/freezer.h +++ b/include/linux/freezer.h @@ -127,6 +127,15 @@ static inline void set_freezable(void) current->flags &= ~PF_NOFREEZE; } +/* + * Tell the freezer that the current task should be frozen by it and that it + * should send a fake signal to the task to freeze it. + */ +static inline void set_freezable_with_signal(void) +{ + current->flags &= ~(PF_NOFREEZE | PF_FREEZER_NOSIG); +} + /* * Freezer-friendly wrappers around wait_event_interruptible() and * wait_event_interruptible_timeout(), originally defined in @@ -174,6 +183,7 @@ static inline void freezer_do_not_count(void) {} static inline void freezer_count(void) {} static inline int freezer_should_skip(struct task_struct *p) { return 0; } static inline void set_freezable(void) {} +static inline void set_freezable_with_signal(void) {} #define wait_event_freezable(wq, condition) \ wait_event_interruptible(wq, condition) diff --git a/include/linux/sched.h b/include/linux/sched.h index 21349173d14..ba2f859c6e4 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1494,6 +1494,7 @@ static inline void put_task_struct(struct task_struct *t) #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ +#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ /* * Only the _current_ task can read/write to tsk->flags, but other diff --git a/kernel/kthread.c b/kernel/kthread.c index 97747cdd37c..ac3fb732664 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -235,7 +235,7 @@ int kthreadd(void *unused) set_user_nice(tsk, KTHREAD_NICE_LEVEL); set_cpus_allowed(tsk, CPU_MASK_ALL); - current->flags |= PF_NOFREEZE; + current->flags |= PF_NOFREEZE | PF_FREEZER_NOSIG; for (;;) { set_current_state(TASK_INTERRUPTIBLE); diff --git a/kernel/power/process.c b/kernel/power/process.c index f1d0b345c9b..5fb87652f21 100644 --- a/kernel/power/process.c +++ b/kernel/power/process.c @@ -19,9 +19,6 @@ */ #define TIMEOUT (20 * HZ) -#define FREEZER_KERNEL_THREADS 0 -#define FREEZER_USER_SPACE 1 - static inline int freezeable(struct task_struct * p) { if ((p == current) || @@ -84,63 +81,53 @@ static void fake_signal_wake_up(struct task_struct *p) spin_unlock_irqrestore(&p->sighand->siglock, flags); } -static int has_mm(struct task_struct *p) +static inline bool should_send_signal(struct task_struct *p) { - return (p->mm && !(p->flags & PF_BORROWED_MM)); + return !(p->flags & PF_FREEZER_NOSIG); } /** * freeze_task - send a freeze request to given task * @p: task to send the request to - * @with_mm_only: if set, the request will only be sent if the task has its - * own mm - * Return value: 0, if @with_mm_only is set and the task has no mm of its - * own or the task is frozen, 1, otherwise + * @sig_only: if set, the request will only be sent if the task has the + * PF_FREEZER_NOSIG flag unset + * Return value: 'false', if @sig_only is set and the task has + * PF_FREEZER_NOSIG set or the task is frozen, 'true', otherwise * - * The freeze request is sent by seting the tasks's TIF_FREEZE flag and + * The freeze request is sent by setting the tasks's TIF_FREEZE flag and * either sending a fake signal to it or waking it up, depending on whether - * or not it has its own mm (ie. it is a user land task). If @with_mm_only - * is set and the task has no mm of its own (ie. it is a kernel thread), - * its TIF_FREEZE flag should not be set. - * - * The task_lock() is necessary to prevent races with exit_mm() or - * use_mm()/unuse_mm() from occuring. + * or not it has PF_FREEZER_NOSIG set. If @sig_only is set and the task + * has PF_FREEZER_NOSIG set (ie. it is a typical kernel thread), its + * TIF_FREEZE flag will not be set. */ -static int freeze_task(struct task_struct *p, int with_mm_only) +static bool freeze_task(struct task_struct *p, bool sig_only) { - int ret = 1; + /* + * We first check if the task is freezing and next if it has already + * been frozen to avoid the race with frozen_process() which first marks + * the task as frozen and next clears its TIF_FREEZE. + */ + if (!freezing(p)) { + rmb(); + if (frozen(p)) + return false; - task_lock(p); - if (freezing(p)) { - if (has_mm(p)) { - if (!signal_pending(p)) - fake_signal_wake_up(p); - } else { - if (with_mm_only) - ret = 0; - else - wake_up_state(p, TASK_INTERRUPTIBLE); - } + if (!sig_only || should_send_signal(p)) + set_freeze_flag(p); + else + return false; + } + + if (should_send_signal(p)) { + if (!signal_pending(p)) + fake_signal_wake_up(p); + } else if (sig_only) { + return false; } else { - rmb(); - if (frozen(p)) { - ret = 0; - } else { - if (has_mm(p)) { - set_freeze_flag(p); - fake_signal_wake_up(p); - } else { - if (with_mm_only) { - ret = 0; - } else { - set_freeze_flag(p); - wake_up_state(p, TASK_INTERRUPTIBLE); - } - } - } + wake_up_state(p, TASK_INTERRUPTIBLE); } - task_unlock(p); - return ret; + + return true; } static void cancel_freezing(struct task_struct *p) @@ -156,7 +143,7 @@ static void cancel_freezing(struct task_struct *p) } } -static int try_to_freeze_tasks(int freeze_user_space) +static int try_to_freeze_tasks(bool sig_only) { struct task_struct *g, *p; unsigned long end_time; @@ -175,7 +162,7 @@ static int try_to_freeze_tasks(int freeze_user_space) if (frozen(p) || !freezeable(p)) continue; - if (!freeze_task(p, freeze_user_space)) + if (!freeze_task(p, sig_only)) continue; /* @@ -235,13 +222,13 @@ int freeze_processes(void) int error; printk("Freezing user space processes ... "); - error = try_to_freeze_tasks(FREEZER_USER_SPACE); + error = try_to_freeze_tasks(true); if (error) goto Exit; printk("done.\n"); printk("Freezing remaining freezable tasks ... "); - error = try_to_freeze_tasks(FREEZER_KERNEL_THREADS); + error = try_to_freeze_tasks(false); if (error) goto Exit; printk("done."); @@ -251,7 +238,7 @@ int freeze_processes(void) return error; } -static void thaw_tasks(int thaw_user_space) +static void thaw_tasks(bool nosig_only) { struct task_struct *g, *p; @@ -260,7 +247,7 @@ static void thaw_tasks(int thaw_user_space) if (!freezeable(p)) continue; - if (!p->mm == thaw_user_space) + if (nosig_only && should_send_signal(p)) continue; thaw_process(p); @@ -271,8 +258,8 @@ static void thaw_tasks(int thaw_user_space) void thaw_processes(void) { printk("Restarting tasks ... "); - thaw_tasks(FREEZER_KERNEL_THREADS); - thaw_tasks(FREEZER_USER_SPACE); + thaw_tasks(true); + thaw_tasks(false); schedule(); printk("done.\n"); } -- cgit v1.2.3-70-g09d2 From f470021adb9190819c03d6d8c5c860a17480aa6d Mon Sep 17 00:00:00 2001 From: Roland McGrath Date: Mon, 24 Mar 2008 18:36:23 -0700 Subject: ptrace children revamp ptrace no longer fiddles with the children/sibling links, and the old ptrace_children list is gone. Now ptrace, whether of one's own children or another's via PTRACE_ATTACH, just uses the new ptraced list instead. There should be no user-visible difference that matters. The only change is the order in which do_wait() sees multiple stopped children and stopped ptrace attachees. Since wait_task_stopped() was changed earlier so it no longer reorders the children list, we already know this won't cause any new problems. Signed-off-by: Roland McGrath --- include/linux/init_task.h | 4 +- include/linux/sched.h | 26 +++--- kernel/exit.c | 226 ++++++++++++++++++++++++---------------------- kernel/fork.c | 6 +- kernel/ptrace.c | 37 +++++--- 5 files changed, 160 insertions(+), 139 deletions(-) (limited to 'include/linux/sched.h') diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 9927a88674a..93c45acf249 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -140,8 +140,8 @@ extern struct group_info init_groups; .nr_cpus_allowed = NR_CPUS, \ }, \ .tasks = LIST_HEAD_INIT(tsk.tasks), \ - .ptrace_children= LIST_HEAD_INIT(tsk.ptrace_children), \ - .ptrace_list = LIST_HEAD_INIT(tsk.ptrace_list), \ + .ptraced = LIST_HEAD_INIT(tsk.ptraced), \ + .ptrace_entry = LIST_HEAD_INIT(tsk.ptrace_entry), \ .real_parent = &tsk, \ .parent = &tsk, \ .children = LIST_HEAD_INIT(tsk.children), \ diff --git a/include/linux/sched.h b/include/linux/sched.h index ba2f859c6e4..1941d8b5cf1 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1062,12 +1062,6 @@ struct task_struct { #endif struct list_head tasks; - /* - * ptrace_list/ptrace_children forms the list of my children - * that were stolen by a ptracer. - */ - struct list_head ptrace_children; - struct list_head ptrace_list; struct mm_struct *mm, *active_mm; @@ -1089,18 +1083,25 @@ struct task_struct { /* * pointers to (original) parent process, youngest child, younger sibling, * older sibling, respectively. (p->father can be replaced with - * p->parent->pid) + * p->real_parent->pid) */ - struct task_struct *real_parent; /* real parent process (when being debugged) */ - struct task_struct *parent; /* parent process */ + struct task_struct *real_parent; /* real parent process */ + struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ /* - * children/sibling forms the list of my children plus the - * tasks I'm ptracing. + * children/sibling forms the list of my natural children */ struct list_head children; /* list of my children */ struct list_head sibling; /* linkage in my parent's children list */ struct task_struct *group_leader; /* threadgroup leader */ + /* + * ptraced is the list of tasks this task is using ptrace on. + * This includes both natural children and PTRACE_ATTACH targets. + * p->ptrace_entry is p's link on the p->parent->ptraced list. + */ + struct list_head ptraced; + struct list_head ptrace_entry; + /* PID/PID hash table linkage. */ struct pid_link pids[PIDTYPE_MAX]; struct list_head thread_group; @@ -1876,9 +1877,6 @@ extern void wait_task_inactive(struct task_struct * p); #define wait_task_inactive(p) do { } while (0) #endif -#define remove_parent(p) list_del_init(&(p)->sibling) -#define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children) - #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) #define for_each_process(p) \ diff --git a/kernel/exit.c b/kernel/exit.c index 7453356a961..1e909826a80 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -71,7 +71,7 @@ static void __unhash_process(struct task_struct *p) __get_cpu_var(process_counts)--; } list_del_rcu(&p->thread_group); - remove_parent(p); + list_del_init(&p->sibling); } /* @@ -152,6 +152,18 @@ static void delayed_put_task_struct(struct rcu_head *rhp) put_task_struct(container_of(rhp, struct task_struct, rcu)); } +/* + * Do final ptrace-related cleanup of a zombie being reaped. + * + * Called with write_lock(&tasklist_lock) held. + */ +static void ptrace_release_task(struct task_struct *p) +{ + BUG_ON(!list_empty(&p->ptraced)); + ptrace_unlink(p); + BUG_ON(!list_empty(&p->ptrace_entry)); +} + void release_task(struct task_struct * p) { struct task_struct *leader; @@ -160,8 +172,7 @@ repeat: atomic_dec(&p->user->processes); proc_flush_task(p); write_lock_irq(&tasklist_lock); - ptrace_unlink(p); - BUG_ON(!list_empty(&p->ptrace_list) || !list_empty(&p->ptrace_children)); + ptrace_release_task(p); __exit_signal(p); /* @@ -315,9 +326,8 @@ static void reparent_to_kthreadd(void) ptrace_unlink(current); /* Reparent to init */ - remove_parent(current); current->real_parent = current->parent = kthreadd_task; - add_parent(current); + list_move_tail(¤t->sibling, ¤t->real_parent->children); /* Set the exit signal to SIGCHLD so we signal init on exit */ current->exit_signal = SIGCHLD; @@ -692,37 +702,71 @@ static void exit_mm(struct task_struct * tsk) mmput(mm); } -static void -reparent_thread(struct task_struct *p, struct task_struct *father, int traced) +/* + * Detach all tasks we were using ptrace on. + * Any that need to be release_task'd are put on the @dead list. + * + * Called with write_lock(&tasklist_lock) held. + */ +static void ptrace_exit(struct task_struct *parent, struct list_head *dead) { - if (p->pdeath_signal) - /* We already hold the tasklist_lock here. */ - group_send_sig_info(p->pdeath_signal, SEND_SIG_NOINFO, p); + struct task_struct *p, *n; - /* Move the child from its dying parent to the new one. */ - if (unlikely(traced)) { - /* Preserve ptrace links if someone else is tracing this child. */ - list_del_init(&p->ptrace_list); - if (ptrace_reparented(p)) - list_add(&p->ptrace_list, &p->real_parent->ptrace_children); - } else { - /* If this child is being traced, then we're the one tracing it - * anyway, so let go of it. + list_for_each_entry_safe(p, n, &parent->ptraced, ptrace_entry) { + __ptrace_unlink(p); + + if (p->exit_state != EXIT_ZOMBIE) + continue; + + /* + * If it's a zombie, our attachedness prevented normal + * parent notification or self-reaping. Do notification + * now if it would have happened earlier. If it should + * reap itself, add it to the @dead list. We can't call + * release_task() here because we already hold tasklist_lock. + * + * If it's our own child, there is no notification to do. */ - p->ptrace = 0; - remove_parent(p); - p->parent = p->real_parent; - add_parent(p); + if (!task_detached(p) && thread_group_empty(p)) { + if (!same_thread_group(p->real_parent, parent)) + do_notify_parent(p, p->exit_signal); + } - if (task_is_traced(p)) { + if (task_detached(p)) { /* - * If it was at a trace stop, turn it into - * a normal stop since it's no longer being - * traced. + * Mark it as in the process of being reaped. */ - ptrace_untrace(p); + p->exit_state = EXIT_DEAD; + list_add(&p->ptrace_entry, dead); } } +} + +/* + * Finish up exit-time ptrace cleanup. + * + * Called without locks. + */ +static void ptrace_exit_finish(struct task_struct *parent, + struct list_head *dead) +{ + struct task_struct *p, *n; + + BUG_ON(!list_empty(&parent->ptraced)); + + list_for_each_entry_safe(p, n, dead, ptrace_entry) { + list_del_init(&p->ptrace_entry); + release_task(p); + } +} + +static void reparent_thread(struct task_struct *p, struct task_struct *father) +{ + if (p->pdeath_signal) + /* We already hold the tasklist_lock here. */ + group_send_sig_info(p->pdeath_signal, SEND_SIG_NOINFO, p); + + list_move_tail(&p->sibling, &p->real_parent->children); /* If this is a threaded reparent there is no need to * notify anyone anything has happened. @@ -737,7 +781,8 @@ reparent_thread(struct task_struct *p, struct task_struct *father, int traced) /* If we'd notified the old parent about this child's death, * also notify the new parent. */ - if (!traced && p->exit_state == EXIT_ZOMBIE && + if (!ptrace_reparented(p) && + p->exit_state == EXIT_ZOMBIE && !task_detached(p) && thread_group_empty(p)) do_notify_parent(p, p->exit_signal); @@ -754,12 +799,15 @@ reparent_thread(struct task_struct *p, struct task_struct *father, int traced) static void forget_original_parent(struct task_struct *father) { struct task_struct *p, *n, *reaper = father; - struct list_head ptrace_dead; - - INIT_LIST_HEAD(&ptrace_dead); + LIST_HEAD(ptrace_dead); write_lock_irq(&tasklist_lock); + /* + * First clean up ptrace if we were using it. + */ + ptrace_exit(father, &ptrace_dead); + do { reaper = next_thread(reaper); if (reaper == father) { @@ -768,58 +816,19 @@ static void forget_original_parent(struct task_struct *father) } } while (reaper->flags & PF_EXITING); - /* - * There are only two places where our children can be: - * - * - in our child list - * - in our ptraced child list - * - * Search them and reparent children. - */ list_for_each_entry_safe(p, n, &father->children, sibling) { - int ptrace; - - ptrace = p->ptrace; - - /* if father isn't the real parent, then ptrace must be enabled */ - BUG_ON(father != p->real_parent && !ptrace); - - if (father == p->real_parent) { - /* reparent with a reaper, real father it's us */ - p->real_parent = reaper; - reparent_thread(p, father, 0); - } else { - /* reparent ptraced task to its real parent */ - __ptrace_unlink (p); - if (p->exit_state == EXIT_ZOMBIE && !task_detached(p) && - thread_group_empty(p)) - do_notify_parent(p, p->exit_signal); - } - - /* - * if the ptraced child is a detached zombie we must collect - * it before we exit, or it will remain zombie forever since - * we prevented it from self-reap itself while it was being - * traced by us, to be able to see it in wait4. - */ - if (unlikely(ptrace && p->exit_state == EXIT_ZOMBIE && task_detached(p))) - list_add(&p->ptrace_list, &ptrace_dead); - } - - list_for_each_entry_safe(p, n, &father->ptrace_children, ptrace_list) { p->real_parent = reaper; - reparent_thread(p, father, 1); + if (p->parent == father) { + BUG_ON(p->ptrace); + p->parent = p->real_parent; + } + reparent_thread(p, father); } write_unlock_irq(&tasklist_lock); BUG_ON(!list_empty(&father->children)); - BUG_ON(!list_empty(&father->ptrace_children)); - - list_for_each_entry_safe(p, n, &ptrace_dead, ptrace_list) { - list_del_init(&p->ptrace_list); - release_task(p); - } + ptrace_exit_finish(father, &ptrace_dead); } /* @@ -1180,13 +1189,6 @@ static int eligible_child(enum pid_type type, struct pid *pid, int options, return 0; } - /* - * Do not consider detached threads that are - * not ptraced: - */ - if (task_detached(p) && !p->ptrace) - return 0; - /* Wait for all children (clone and not) if __WALL is set; * otherwise, wait for clone children *only* if __WCLONE is * set; otherwise, wait for non-clone children *only*. (Note: @@ -1399,7 +1401,7 @@ static int wait_task_zombie(struct task_struct *p, int options, * the lock and this task is uninteresting. If we return nonzero, we have * released the lock and the system call should return. */ -static int wait_task_stopped(struct task_struct *p, +static int wait_task_stopped(int ptrace, struct task_struct *p, int options, struct siginfo __user *infop, int __user *stat_addr, struct rusage __user *ru) { @@ -1407,7 +1409,7 @@ static int wait_task_stopped(struct task_struct *p, uid_t uid = 0; /* unneeded, required by compiler */ pid_t pid; - if (!(p->ptrace & PT_PTRACED) && !(options & WUNTRACED)) + if (!(options & WUNTRACED)) return 0; exit_code = 0; @@ -1416,7 +1418,7 @@ static int wait_task_stopped(struct task_struct *p, if (unlikely(!task_is_stopped_or_traced(p))) goto unlock_sig; - if (!(p->ptrace & PT_PTRACED) && p->signal->group_stop_count > 0) + if (!ptrace && p->signal->group_stop_count > 0) /* * A group stop is in progress and this is the group leader. * We won't report until all threads have stopped. @@ -1445,7 +1447,7 @@ unlock_sig: */ get_task_struct(p); pid = task_pid_vnr(p); - why = (p->ptrace & PT_PTRACED) ? CLD_TRAPPED : CLD_STOPPED; + why = ptrace ? CLD_TRAPPED : CLD_STOPPED; read_unlock(&tasklist_lock); if (unlikely(options & WNOWAIT)) @@ -1536,7 +1538,7 @@ static int wait_task_continued(struct task_struct *p, int options, * Returns zero if the search for a child should continue; * then *@notask_error is 0 if @p is an eligible child, or still -ECHILD. */ -static int wait_consider_task(struct task_struct *parent, +static int wait_consider_task(struct task_struct *parent, int ptrace, struct task_struct *p, int *notask_error, enum pid_type type, struct pid *pid, int options, struct siginfo __user *infop, @@ -1546,6 +1548,15 @@ static int wait_consider_task(struct task_struct *parent, if (ret <= 0) return ret; + if (likely(!ptrace) && unlikely(p->ptrace)) { + /* + * This child is hidden by ptrace. + * We aren't allowed to see it now, but eventually we will. + */ + *notask_error = 0; + return 0; + } + if (p->exit_state == EXIT_DEAD) return 0; @@ -1562,7 +1573,8 @@ static int wait_consider_task(struct task_struct *parent, *notask_error = 0; if (task_is_stopped_or_traced(p)) - return wait_task_stopped(p, options, infop, stat_addr, ru); + return wait_task_stopped(ptrace, p, options, + infop, stat_addr, ru); return wait_task_continued(p, options, infop, stat_addr, ru); } @@ -1583,11 +1595,16 @@ static int do_wait_thread(struct task_struct *tsk, int *notask_error, struct task_struct *p; list_for_each_entry(p, &tsk->children, sibling) { - int ret = wait_consider_task(tsk, p, notask_error, - type, pid, options, - infop, stat_addr, ru); - if (ret) - return ret; + /* + * Do not consider detached threads. + */ + if (!task_detached(p)) { + int ret = wait_consider_task(tsk, 0, p, notask_error, + type, pid, options, + infop, stat_addr, ru); + if (ret) + return ret; + } } return 0; @@ -1601,21 +1618,16 @@ static int ptrace_do_wait(struct task_struct *tsk, int *notask_error, struct task_struct *p; /* - * If we never saw an eligile child, check for children stolen by - * ptrace. We don't leave -ECHILD in *@notask_error if there are any, - * because we will eventually be allowed to wait for them again. + * Traditionally we see ptrace'd stopped tasks regardless of options. */ - if (!*notask_error) - return 0; + options |= WUNTRACED; - list_for_each_entry(p, &tsk->ptrace_children, ptrace_list) { - int ret = eligible_child(type, pid, options, p); - if (unlikely(ret < 0)) + list_for_each_entry(p, &tsk->ptraced, ptrace_entry) { + int ret = wait_consider_task(tsk, 1, p, notask_error, + type, pid, options, + infop, stat_addr, ru); + if (ret) return ret; - if (ret) { - *notask_error = 0; - return 0; - } } return 0; diff --git a/kernel/fork.c b/kernel/fork.c index 4bd2f516401..adefc1131f2 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1125,8 +1125,8 @@ static struct task_struct *copy_process(unsigned long clone_flags, */ p->group_leader = p; INIT_LIST_HEAD(&p->thread_group); - INIT_LIST_HEAD(&p->ptrace_children); - INIT_LIST_HEAD(&p->ptrace_list); + INIT_LIST_HEAD(&p->ptrace_entry); + INIT_LIST_HEAD(&p->ptraced); /* Now that the task is set up, run cgroup callbacks if * necessary. We need to run them before the task is visible @@ -1198,7 +1198,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, } if (likely(p->pid)) { - add_parent(p); + list_add_tail(&p->sibling, &p->real_parent->children); if (unlikely(p->ptrace & PT_PTRACED)) __ptrace_link(p, current->parent); diff --git a/kernel/ptrace.c b/kernel/ptrace.c index e337390fce0..8392a9da645 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -33,13 +33,9 @@ */ void __ptrace_link(struct task_struct *child, struct task_struct *new_parent) { - BUG_ON(!list_empty(&child->ptrace_list)); - if (child->parent == new_parent) - return; - list_add(&child->ptrace_list, &child->parent->ptrace_children); - remove_parent(child); + BUG_ON(!list_empty(&child->ptrace_entry)); + list_add(&child->ptrace_entry, &new_parent->ptraced); child->parent = new_parent; - add_parent(child); } /* @@ -73,12 +69,8 @@ void __ptrace_unlink(struct task_struct *child) BUG_ON(!child->ptrace); child->ptrace = 0; - if (ptrace_reparented(child)) { - list_del_init(&child->ptrace_list); - remove_parent(child); - child->parent = child->real_parent; - add_parent(child); - } + child->parent = child->real_parent; + list_del_init(&child->ptrace_entry); if (task_is_traced(child)) ptrace_untrace(child); @@ -492,15 +484,34 @@ int ptrace_traceme(void) /* * Are we already being traced? */ +repeat: task_lock(current); if (!(current->ptrace & PT_PTRACED)) { + /* + * See ptrace_attach() comments about the locking here. + */ + unsigned long flags; + if (!write_trylock_irqsave(&tasklist_lock, flags)) { + task_unlock(current); + do { + cpu_relax(); + } while (!write_can_lock(&tasklist_lock)); + goto repeat; + } + ret = security_ptrace(current->parent, current, PTRACE_MODE_ATTACH); + /* * Set the ptrace bit in the process ptrace flags. + * Then link us on our parent's ptraced list. */ - if (!ret) + if (!ret) { current->ptrace |= PT_PTRACED; + __ptrace_link(current, current->real_parent); + } + + write_unlock_irqrestore(&tasklist_lock, flags); } task_unlock(current); return ret; -- cgit v1.2.3-70-g09d2