From 49f474331e563a6ecf3b1e87ec27ec5482b3e4f1 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sun, 27 Dec 2009 11:51:52 +0100 Subject: perf events: Remove arg from perf sched hooks Since we only ever schedule the local cpu, there is no need to pass the cpu number to the perf sched hooks. This micro-optimizes things a bit. Signed-off-by: Peter Zijlstra Cc: Mike Galbraith Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Frederic Weisbecker LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index c66b34f75eea..a494e7501292 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -746,10 +746,10 @@ extern int perf_max_events; extern const struct pmu *hw_perf_event_init(struct perf_event *event); -extern void perf_event_task_sched_in(struct task_struct *task, int cpu); +extern void perf_event_task_sched_in(struct task_struct *task); extern void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next, int cpu); -extern void perf_event_task_tick(struct task_struct *task, int cpu); + struct task_struct *next); +extern void perf_event_task_tick(struct task_struct *task); extern int perf_event_init_task(struct task_struct *child); extern void perf_event_exit_task(struct task_struct *child); extern void perf_event_free_task(struct task_struct *task); @@ -870,12 +870,12 @@ extern void perf_event_enable(struct perf_event *event); extern void perf_event_disable(struct perf_event *event); #else static inline void -perf_event_task_sched_in(struct task_struct *task, int cpu) { } +perf_event_task_sched_in(struct task_struct *task) { } static inline void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next, int cpu) { } + struct task_struct *next) { } static inline void -perf_event_task_tick(struct task_struct *task, int cpu) { } +perf_event_task_tick(struct task_struct *task) { } static inline int perf_event_init_task(struct task_struct *child) { return 0; } static inline void perf_event_exit_task(struct task_struct *child) { } static inline void perf_event_free_task(struct task_struct *task) { } -- cgit v1.2.3 From 07b139c8c81b97bbe55c68daf0cbeca8b1c609ca Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 21 Dec 2009 14:27:35 +0800 Subject: perf events: Remove CONFIG_EVENT_PROFILE Quoted from Ingo: | This reminds me - i think we should eliminate CONFIG_EVENT_PROFILE - | it's an unnecessary Kconfig complication. If both PERF_EVENTS and | EVENT_TRACING is enabled we should expose generic tracepoints. | | Nor is it limited to event 'profiling', so it has become a misnomer as | well. Signed-off-by: Li Zefan Cc: Frederic Weisbecker Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <4B2F1557.2050705@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- include/linux/ftrace_event.h | 2 +- include/linux/perf_event.h | 2 +- include/linux/syscalls.h | 4 ++-- include/trace/ftrace.h | 12 ++++++------ include/trace/syscall.h | 4 ++-- 5 files changed, 12 insertions(+), 12 deletions(-) (limited to 'include') diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h index 2233c98d80df..0a09e758c7d3 100644 --- a/include/linux/ftrace_event.h +++ b/include/linux/ftrace_event.h @@ -188,7 +188,7 @@ do { \ __trace_printk(ip, fmt, ##args); \ } while (0) -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_PERF_EVENTS struct perf_event; extern int ftrace_profile_enable(int event_id); extern void ftrace_profile_disable(int event_id); diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index a494e7501292..9a1d276db754 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -658,7 +658,7 @@ struct perf_event { perf_overflow_handler_t overflow_handler; -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_EVENT_TRACING struct event_filter *filter; #endif diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h index 65793e90d6f6..b7c7fcf7790b 100644 --- a/include/linux/syscalls.h +++ b/include/linux/syscalls.h @@ -99,7 +99,7 @@ struct perf_event_attr; #define __SC_TEST5(t5, a5, ...) __SC_TEST(t5); __SC_TEST4(__VA_ARGS__) #define __SC_TEST6(t6, a6, ...) __SC_TEST(t6); __SC_TEST5(__VA_ARGS__) -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_PERF_EVENTS #define TRACE_SYS_ENTER_PROFILE_INIT(sname) \ .profile_enable = prof_sysenter_enable, \ @@ -113,7 +113,7 @@ struct perf_event_attr; #define TRACE_SYS_ENTER_PROFILE_INIT(sname) #define TRACE_SYS_EXIT_PROFILE(sname) #define TRACE_SYS_EXIT_PROFILE_INIT(sname) -#endif +#endif /* CONFIG_PERF_EVENTS */ #ifdef CONFIG_FTRACE_SYSCALLS #define __SC_STR_ADECL1(t, a) #a diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h index 73523151a731..2fdd36df41f6 100644 --- a/include/trace/ftrace.h +++ b/include/trace/ftrace.h @@ -498,7 +498,7 @@ static inline int ftrace_get_offsets_##call( \ #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_PERF_EVENTS /* * Generate the functions needed for tracepoint perf_event support. @@ -541,7 +541,7 @@ static void ftrace_profile_disable_##name(struct ftrace_event_call *unused)\ #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) -#endif +#endif /* CONFIG_PERF_EVENTS */ /* * Stage 4 of the trace events. @@ -626,7 +626,7 @@ static void ftrace_profile_disable_##name(struct ftrace_event_call *unused)\ * */ -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_PERF_EVENTS #define _TRACE_PROFILE_INIT(call) \ .profile_enable = ftrace_profile_enable_##call, \ @@ -634,7 +634,7 @@ static void ftrace_profile_disable_##name(struct ftrace_event_call *unused)\ #else #define _TRACE_PROFILE_INIT(call) -#endif +#endif /* CONFIG_PERF_EVENTS */ #undef __entry #define __entry entry @@ -834,7 +834,7 @@ __attribute__((section("_ftrace_events"))) event_##call = { \ * } */ -#ifdef CONFIG_EVENT_PROFILE +#ifdef CONFIG_PERF_EVENTS #undef __perf_addr #define __perf_addr(a) __addr = (a) @@ -926,7 +926,7 @@ static void ftrace_profile_##call(proto) \ DEFINE_EVENT(template, name, PARAMS(proto), PARAMS(args)) #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) -#endif /* CONFIG_EVENT_PROFILE */ +#endif /* CONFIG_PERF_EVENTS */ #undef _TRACE_PROFILE_INIT diff --git a/include/trace/syscall.h b/include/trace/syscall.h index 961fda3556bb..3d463dcef298 100644 --- a/include/trace/syscall.h +++ b/include/trace/syscall.h @@ -49,12 +49,12 @@ ftrace_format_syscall(struct ftrace_event_call *call, struct trace_seq *s); enum print_line_t print_syscall_enter(struct trace_iterator *iter, int flags); enum print_line_t print_syscall_exit(struct trace_iterator *iter, int flags); #endif -#ifdef CONFIG_EVENT_PROFILE + +#ifdef CONFIG_PERF_EVENTS int prof_sysenter_enable(struct ftrace_event_call *call); void prof_sysenter_disable(struct ftrace_event_call *call); int prof_sysexit_enable(struct ftrace_event_call *call); void prof_sysexit_disable(struct ftrace_event_call *call); - #endif #endif /* _TRACE_SYSCALL_H */ -- cgit v1.2.3 From 889ff0150661512d79484219612b7e2e024b6c07 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 9 Jan 2010 20:04:47 +0100 Subject: perf/core: Split context's event group list into pinned and non-pinned lists Split-up struct perf_event_context::group_list into pinned_groups and flexible_groups (non-pinned). This first appears to be useless as it duplicates various loops around the group list handlings. But it scales better in the fast-path in perf_sched_in(). We don't anymore iterate twice through the entire list to separate pinned and non-pinned scheduling. Instead we interate through two distinct lists. The another desired effect is that it makes easier to define distinct scheduling rules on both. Changes in v2: - Respectively rename pinned_grp_list and volatile_grp_list into pinned_groups and flexible_groups as per Ingo suggestion. - Various cleanups Signed-off-by: Frederic Weisbecker Acked-by: Peter Zijlstra Cc: Paul Mackerras Cc: Ingo Molnar Cc: Arnaldo Carvalho de Melo --- include/linux/perf_event.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 9a1d276db754..cdbc2aa64a0b 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -683,7 +683,8 @@ struct perf_event_context { */ struct mutex mutex; - struct list_head group_list; + struct list_head pinned_groups; + struct list_head flexible_groups; struct list_head event_list; int nr_events; int nr_active; -- cgit v1.2.3 From 5908cdc85eb30f8d07f2cb11d4a62334d7229048 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 9 Jan 2010 20:53:14 +0100 Subject: list: Introduce list_rotate_left() Bring a new list_rotate_left() helper that rotates a list to the left. This is useful for codes that need to round roubin elements which queue priority increases from tail to head. Signed-off-by: Frederic Weisbecker Acked-by: Peter Zijlstra Cc: Paul Mackerras Cc: Ingo Molnar Cc: Arnaldo Carvalho de Melo --- include/linux/list.h | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'include') diff --git a/include/linux/list.h b/include/linux/list.h index 969f6e92d089..5d9c6558e8ab 100644 --- a/include/linux/list.h +++ b/include/linux/list.h @@ -205,6 +205,20 @@ static inline int list_empty_careful(const struct list_head *head) return (next == head) && (next == head->prev); } +/** + * list_rotate_left - rotate the list to the left + * @head: the head of the list + */ +static inline void list_rotate_left(struct list_head *head) +{ + struct list_head *first; + + if (!list_empty(head)) { + first = head->next; + list_move_tail(first, head); + } +} + /** * list_is_singular - tests whether a list has just one entry. * @head: the list to test. -- cgit v1.2.3 From d6f962b57bfaab62891c7abbf1469212a56d6103 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 10 Jan 2010 01:25:51 +0100 Subject: perf: Export software-only event group characteristic as a flag Before scheduling an event group, we first check if a group can go on. We first check if the group is made of software only events first, in which case it is enough to know if the group can be scheduled in. For that purpose, we iterate through the whole group, which is wasteful as we could do this check when we add/delete an event to a group. So we create a group_flags field in perf event that can host characteristics from a group of events, starting with a first PERF_GROUP_SOFTWARE flag that reduces the check on the fast path. Signed-off-by: Frederic Weisbecker Acked-by: Peter Zijlstra Cc: Paul Mackerras Cc: Ingo Molnar Cc: Arnaldo Carvalho de Melo --- include/linux/perf_event.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index cdbc2aa64a0b..c6f812e4d058 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -565,6 +565,10 @@ typedef void (*perf_overflow_handler_t)(struct perf_event *, int, struct perf_sample_data *, struct pt_regs *regs); +enum perf_group_flag { + PERF_GROUP_SOFTWARE = 0x1, +}; + /** * struct perf_event - performance event kernel representation: */ @@ -574,6 +578,7 @@ struct perf_event { struct list_head event_entry; struct list_head sibling_list; int nr_siblings; + int group_flags; struct perf_event *group_leader; struct perf_event *output; const struct pmu *pmu; -- cgit v1.2.3 From abd50713944c8ea9e0af5b7bffa0aacae21cc91a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 26 Jan 2010 18:50:16 +0100 Subject: perf: Reimplement frequency driven sampling There was a bug in the old period code that caused intel_pmu_enable_all() or native_write_msr_safe() to show up quite high in the profiles. In staring at that code it made my head hurt, so I rewrote it in a hopefully simpler fashion. Its now fully symetric between tick and overflow driven adjustments and uses less data to boot. The only complication is that it basically wants to do a u128 division. The code approximates that in a rather simple truncate until it fits fashion, taking care to balance the terms while truncating. This version does not generate that sampling artefact. Signed-off-by: Peter Zijlstra LKML-Reference: Cc: Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index c6f812e4d058..72b2615600d8 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -498,9 +498,8 @@ struct hw_perf_event { atomic64_t period_left; u64 interrupts; - u64 freq_count; - u64 freq_interrupts; - u64 freq_stamp; + u64 freq_time_stamp; + u64 freq_count_stamp; #endif }; -- cgit v1.2.3 From 430ad5a600a83956749307b13257c464c3826b55 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Thu, 28 Jan 2010 09:32:29 +0800 Subject: perf: Factorize trace events raw sample buffer operations Introduce ftrace_perf_buf_prepare() and ftrace_perf_buf_submit() to gather the common code that operates on raw events sampling buffer. This cleans up redundant code between regular trace events, syscall events and kprobe events. Changelog v1->v2: - Rename function name as per Masami and Frederic's suggestion - Add __kprobes for ftrace_perf_buf_prepare() and make ftrace_perf_buf_submit() inline as per Masami's suggestion - Export ftrace_perf_buf_prepare since modules will use it Signed-off-by: Xiao Guangrong Acked-by: Masami Hiramatsu Cc: Ingo Molnar Cc: Steven Rostedt Cc: Paul Mackerras Cc: Jason Baron Cc: Peter Zijlstra LKML-Reference: <4B60E92D.9000808@cn.fujitsu.com> Signed-off-by: Frederic Weisbecker --- include/linux/ftrace_event.h | 18 ++++++++++++++--- include/trace/ftrace.h | 48 ++++++-------------------------------------- 2 files changed, 21 insertions(+), 45 deletions(-) (limited to 'include') diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h index 0a09e758c7d3..cd95919d9ff3 100644 --- a/include/linux/ftrace_event.h +++ b/include/linux/ftrace_event.h @@ -5,6 +5,7 @@ #include #include #include +#include struct trace_array; struct tracer; @@ -138,9 +139,6 @@ struct ftrace_event_call { #define FTRACE_MAX_PROFILE_SIZE 2048 -extern char *perf_trace_buf; -extern char *perf_trace_buf_nmi; - #define MAX_FILTER_PRED 32 #define MAX_FILTER_STR_VAL 256 /* Should handle KSYM_SYMBOL_LEN */ @@ -195,6 +193,20 @@ extern void ftrace_profile_disable(int event_id); extern int ftrace_profile_set_filter(struct perf_event *event, int event_id, char *filter_str); extern void ftrace_profile_free_filter(struct perf_event *event); +extern void * +ftrace_perf_buf_prepare(int size, unsigned short type, int *rctxp, + unsigned long *irq_flags); + +static inline void +ftrace_perf_buf_submit(void *raw_data, int size, int rctx, u64 addr, + u64 count, unsigned long irq_flags) +{ + struct trace_entry *entry = raw_data; + + perf_tp_event(entry->type, addr, count, raw_data, size); + perf_swevent_put_recursion_context(rctx); + local_irq_restore(irq_flags); +} #endif #endif /* _LINUX_FTRACE_EVENT_H */ diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h index 4a46a60c2077..f2c09e4d656c 100644 --- a/include/trace/ftrace.h +++ b/include/trace/ftrace.h @@ -850,22 +850,12 @@ ftrace_profile_templ_##call(struct ftrace_event_call *event_call, \ proto) \ { \ struct ftrace_data_offsets_##call __maybe_unused __data_offsets;\ - extern int perf_swevent_get_recursion_context(void); \ - extern void perf_swevent_put_recursion_context(int rctx); \ - extern void perf_tp_event(int, u64, u64, void *, int); \ struct ftrace_raw_##call *entry; \ u64 __addr = 0, __count = 1; \ unsigned long irq_flags; \ - struct trace_entry *ent; \ int __entry_size; \ int __data_size; \ - char *trace_buf; \ - char *raw_data; \ - int __cpu; \ int rctx; \ - int pc; \ - \ - pc = preempt_count(); \ \ __data_size = ftrace_get_offsets_##call(&__data_offsets, args); \ __entry_size = ALIGN(__data_size + sizeof(*entry) + sizeof(u32),\ @@ -875,42 +865,16 @@ ftrace_profile_templ_##call(struct ftrace_event_call *event_call, \ if (WARN_ONCE(__entry_size > FTRACE_MAX_PROFILE_SIZE, \ "profile buffer not large enough")) \ return; \ - \ - local_irq_save(irq_flags); \ - \ - rctx = perf_swevent_get_recursion_context(); \ - if (rctx < 0) \ - goto end_recursion; \ - \ - __cpu = smp_processor_id(); \ - \ - if (in_nmi()) \ - trace_buf = rcu_dereference(perf_trace_buf_nmi); \ - else \ - trace_buf = rcu_dereference(perf_trace_buf); \ - \ - if (!trace_buf) \ - goto end; \ - \ - raw_data = per_cpu_ptr(trace_buf, __cpu); \ - \ - *(u64 *)(&raw_data[__entry_size - sizeof(u64)]) = 0ULL; \ - entry = (struct ftrace_raw_##call *)raw_data; \ - ent = &entry->ent; \ - tracing_generic_entry_update(ent, irq_flags, pc); \ - ent->type = event_call->id; \ - \ + entry = (struct ftrace_raw_##call *)ftrace_perf_buf_prepare( \ + __entry_size, event_call->id, &rctx, &irq_flags); \ + if (!entry) \ + return; \ tstruct \ \ { assign; } \ \ - perf_tp_event(event_call->id, __addr, __count, entry, \ - __entry_size); \ - \ -end: \ - perf_swevent_put_recursion_context(rctx); \ -end_recursion: \ - local_irq_restore(irq_flags); \ + ftrace_perf_buf_submit(entry, __entry_size, rctx, __addr, \ + __count, irq_flags); \ } #undef DEFINE_EVENT -- cgit v1.2.3 From 9f41699ed067fa695faff8e2e9981b2550abec62 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 22 Jan 2010 15:59:29 +0100 Subject: bitops: Provide compile time HWEIGHT{8,16,32,64} Provide compile time versions of hweight. Signed-off-by: Peter Zijlstra Cc: Stephane Eranian Cc: Linus Torvalds Cc: Andrew Morton Cc: Thomas Gleixner LKML-Reference: <20100122155535.797688466@chello.nl> [ Remove some whitespace damage while we are at it ] Signed-off-by: Ingo Molnar --- include/linux/bitops.h | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/bitops.h b/include/linux/bitops.h index c05a29cb9bb2..ba0fd1eb4af7 100644 --- a/include/linux/bitops.h +++ b/include/linux/bitops.h @@ -25,7 +25,7 @@ static __inline__ int get_bitmask_order(unsigned int count) { int order; - + order = fls(count); return order; /* We could be slightly more clever with -1 here... */ } @@ -33,7 +33,7 @@ static __inline__ int get_bitmask_order(unsigned int count) static __inline__ int get_count_order(unsigned int count) { int order; - + order = fls(count) - 1; if (count & (count - 1)) order++; @@ -45,6 +45,20 @@ static inline unsigned long hweight_long(unsigned long w) return sizeof(w) == 4 ? hweight32(w) : hweight64(w); } +#define HWEIGHT8(w) \ + ( (!!((w) & (1ULL << 0))) + \ + (!!((w) & (1ULL << 1))) + \ + (!!((w) & (1ULL << 2))) + \ + (!!((w) & (1ULL << 3))) + \ + (!!((w) & (1ULL << 4))) + \ + (!!((w) & (1ULL << 5))) + \ + (!!((w) & (1ULL << 6))) + \ + (!!((w) & (1ULL << 7))) ) + +#define HWEIGHT16(w) (HWEIGHT8(w) + HWEIGHT8(w >> 8)) +#define HWEIGHT32(w) (HWEIGHT16(w) + HWEIGHT16(w >> 16)) +#define HWEIGHT64(w) (HWEIGHT32(w) + HWEIGHT32(w >> 32)) + /** * rol32 - rotate a 32-bit value left * @word: value to rotate -- cgit v1.2.3 From 184f412c3341cd24fbd26604634a5800b83dbdc3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 27 Jan 2010 08:39:39 +0100 Subject: perf, x86: Clean up event constraints code a bit - Remove stray debug code - Improve ugly macros a bit - Remove some whitespace damage - (Also fix up some accumulated damage in perf_event.h) Signed-off-by: Ingo Molnar Cc: Stephane Eranian Cc: Peter Zijlstra LKML-Reference: --- include/linux/perf_event.h | 24 +++++++++++------------- 1 file changed, 11 insertions(+), 13 deletions(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 72b2615600d8..953c17731e0d 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -290,7 +290,7 @@ struct perf_event_mmap_page { }; #define PERF_RECORD_MISC_CPUMODE_MASK (3 << 0) -#define PERF_RECORD_MISC_CPUMODE_UNKNOWN (0 << 0) +#define PERF_RECORD_MISC_CPUMODE_UNKNOWN (0 << 0) #define PERF_RECORD_MISC_KERNEL (1 << 0) #define PERF_RECORD_MISC_USER (2 << 0) #define PERF_RECORD_MISC_HYPERVISOR (3 << 0) @@ -356,8 +356,8 @@ enum perf_event_type { * u64 stream_id; * }; */ - PERF_RECORD_THROTTLE = 5, - PERF_RECORD_UNTHROTTLE = 6, + PERF_RECORD_THROTTLE = 5, + PERF_RECORD_UNTHROTTLE = 6, /* * struct { @@ -371,10 +371,10 @@ enum perf_event_type { /* * struct { - * struct perf_event_header header; - * u32 pid, tid; + * struct perf_event_header header; + * u32 pid, tid; * - * struct read_format values; + * struct read_format values; * }; */ PERF_RECORD_READ = 8, @@ -412,7 +412,7 @@ enum perf_event_type { * char data[size];}&& PERF_SAMPLE_RAW * }; */ - PERF_RECORD_SAMPLE = 9, + PERF_RECORD_SAMPLE = 9, PERF_RECORD_MAX, /* non-ABI */ }; @@ -752,8 +752,7 @@ extern int perf_max_events; extern const struct pmu *hw_perf_event_init(struct perf_event *event); extern void perf_event_task_sched_in(struct task_struct *task); -extern void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next); +extern void perf_event_task_sched_out(struct task_struct *task, struct task_struct *next); extern void perf_event_task_tick(struct task_struct *task); extern int perf_event_init_task(struct task_struct *child); extern void perf_event_exit_task(struct task_struct *child); @@ -853,8 +852,7 @@ extern int sysctl_perf_event_mlock; extern int sysctl_perf_event_sample_rate; extern void perf_event_init(void); -extern void perf_tp_event(int event_id, u64 addr, u64 count, - void *record, int entry_size); +extern void perf_tp_event(int event_id, u64 addr, u64 count, void *record, int entry_size); extern void perf_bp_event(struct perf_event *event, void *data); #ifndef perf_misc_flags @@ -895,13 +893,13 @@ static inline void perf_sw_event(u32 event_id, u64 nr, int nmi, struct pt_regs *regs, u64 addr) { } static inline void -perf_bp_event(struct perf_event *event, void *data) { } +perf_bp_event(struct perf_event *event, void *data) { } static inline void perf_event_mmap(struct vm_area_struct *vma) { } static inline void perf_event_comm(struct task_struct *tsk) { } static inline void perf_event_fork(struct task_struct *tsk) { } static inline void perf_event_init(void) { } -static inline int perf_swevent_get_recursion_context(void) { return -1; } +static inline int perf_swevent_get_recursion_context(void) { return -1; } static inline void perf_swevent_put_recursion_context(int rctx) { } static inline void perf_event_enable(struct perf_event *event) { } static inline void perf_event_disable(struct perf_event *event) { } -- cgit v1.2.3 From c965be10ca3cb0bdd04016c852764afaf8e647c8 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Sat, 30 Jan 2010 20:43:32 +0900 Subject: perf lock: Enhance information of lock trace events Add wait time and lock identification details. Signed-off-by: Hitoshi Mitake Cc: Peter Zijlstra Cc: Paul Mackerras Cc: Frederic Weisbecker LKML-Reference: <1264851813-8413-11-git-send-email-mitake@dcl.info.waseda.ac.jp> [ removed the file/line bits as we can do that better via IPs ] Signed-off-by: Ingo Molnar --- include/trace/events/lock.h | 29 ++++++++++++++++++++--------- 1 file changed, 20 insertions(+), 9 deletions(-) (limited to 'include') diff --git a/include/trace/events/lock.h b/include/trace/events/lock.h index a870ba125aa8..5c1dcfc16c60 100644 --- a/include/trace/events/lock.h +++ b/include/trace/events/lock.h @@ -20,14 +20,17 @@ TRACE_EVENT(lock_acquire, TP_STRUCT__entry( __field(unsigned int, flags) __string(name, lock->name) + __field(void *, lockdep_addr) ), TP_fast_assign( __entry->flags = (trylock ? 1 : 0) | (read ? 2 : 0); __assign_str(name, lock->name); + __entry->lockdep_addr = lock; ), - TP_printk("%s%s%s", (__entry->flags & 1) ? "try " : "", + TP_printk("%p %s%s%s", __entry->lockdep_addr, + (__entry->flags & 1) ? "try " : "", (__entry->flags & 2) ? "read " : "", __get_str(name)) ); @@ -40,13 +43,16 @@ TRACE_EVENT(lock_release, TP_STRUCT__entry( __string(name, lock->name) + __field(void *, lockdep_addr) ), TP_fast_assign( __assign_str(name, lock->name); + __entry->lockdep_addr = lock; ), - TP_printk("%s", __get_str(name)) + TP_printk("%p %s", + __entry->lockdep_addr, __get_str(name)) ); #ifdef CONFIG_LOCK_STAT @@ -59,13 +65,16 @@ TRACE_EVENT(lock_contended, TP_STRUCT__entry( __string(name, lock->name) + __field(void *, lockdep_addr) ), TP_fast_assign( __assign_str(name, lock->name); + __entry->lockdep_addr = lock; ), - TP_printk("%s", __get_str(name)) + TP_printk("%p %s", + __entry->lockdep_addr, __get_str(name)) ); TRACE_EVENT(lock_acquired, @@ -75,16 +84,18 @@ TRACE_EVENT(lock_acquired, TP_STRUCT__entry( __string(name, lock->name) - __field(unsigned long, wait_usec) - __field(unsigned long, wait_nsec_rem) + __field(s64, wait_nsec) + __field(void *, lockdep_addr) ), + TP_fast_assign( __assign_str(name, lock->name); - __entry->wait_nsec_rem = do_div(waittime, NSEC_PER_USEC); - __entry->wait_usec = (unsigned long) waittime; + __entry->wait_nsec = waittime; + __entry->lockdep_addr = lock; ), - TP_printk("%s (%lu.%03lu us)", __get_str(name), __entry->wait_usec, - __entry->wait_nsec_rem) + TP_printk("%p %s (%llu ns)", __entry->lockdep_addr, + __get_str(name), + __entry->wait_nsec) ); #endif -- cgit v1.2.3 From 2cfa19780d61740f65790c5bae363b759d7c96fa Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Tue, 2 Feb 2010 16:49:11 -0500 Subject: ftrace/alternatives: Introducing *_text_reserved functions Introducing *_text_reserved functions for checking the text address range is partially reserved or not. This patch provides checking routines for x86 smp alternatives and dynamic ftrace. Since both functions modify fixed pieces of kernel text, they should reserve and protect those from other dynamic text modifier, like kprobes. This will also be extended when introducing other subsystems which modify fixed pieces of kernel text. Dynamic text modifiers should avoid those. Signed-off-by: Masami Hiramatsu Cc: systemtap Cc: DLE Cc: Steven Rostedt Cc: przemyslaw@pawelczyk.it Cc: Frederic Weisbecker Cc: Ananth N Mavinakayanahalli Cc: Jim Keniston Cc: Mathieu Desnoyers Cc: Jason Baron LKML-Reference: <20100202214911.4694.16587.stgit@dhcp-100-2-132.bos.redhat.com> Signed-off-by: Ingo Molnar --- include/linux/ftrace.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'include') diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 0b4f97d24d7f..9d127efed43c 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -134,6 +134,8 @@ extern void unregister_ftrace_function_probe_func(char *glob, struct ftrace_probe_ops *ops); extern void unregister_ftrace_function_probe_all(char *glob); +extern int ftrace_text_reserved(void *start, void *end); + enum { FTRACE_FL_FREE = (1 << 0), FTRACE_FL_FAILED = (1 << 1), @@ -250,6 +252,10 @@ static inline int unregister_ftrace_command(char *cmd_name) { return -EINVAL; } +static inline int ftrace_text_reserved(void *start, void *end) +{ + return 0; +} #endif /* CONFIG_DYNAMIC_FTRACE */ /* totally disable ftrace - can not re-enable after this */ -- cgit v1.2.3 From f24bb999d2b9f2950e5cac5b69bffedf73c24ea4 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Tue, 2 Feb 2010 16:49:25 -0500 Subject: ftrace: Remove record freezing Remove record freezing. Because kprobes never puts probe on ftrace's mcount call anymore, it doesn't need ftrace to check whether kprobes on it. Signed-off-by: Masami Hiramatsu Cc: systemtap Cc: DLE Cc: Steven Rostedt Cc: przemyslaw@pawelczyk.it Cc: Frederic Weisbecker LKML-Reference: <20100202214925.4694.73469.stgit@dhcp-100-2-132.bos.redhat.com> Signed-off-by: Ingo Molnar --- include/linux/ftrace.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include') diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 9d127efed43c..eb054ae95605 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -143,7 +143,6 @@ enum { FTRACE_FL_ENABLED = (1 << 3), FTRACE_FL_NOTRACE = (1 << 4), FTRACE_FL_CONVERTED = (1 << 5), - FTRACE_FL_FROZEN = (1 << 6), }; struct dyn_ftrace { -- cgit v1.2.3 From fce877e3a429940a986e085a41e8b57f2d922e36 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 29 Jan 2010 13:25:12 +0100 Subject: bitops: Ensure the compile time HWEIGHT is only used for such Avoid accidental misuse by failing to compile things Suggested-by: Andrew Morton Signed-off-by: Peter Zijlstra Cc: Linus Torvalds LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/bitops.h | 33 ++++++++++++++++++++++----------- 1 file changed, 22 insertions(+), 11 deletions(-) (limited to 'include') diff --git a/include/linux/bitops.h b/include/linux/bitops.h index ba0fd1eb4af7..25b8b2f33ae9 100644 --- a/include/linux/bitops.h +++ b/include/linux/bitops.h @@ -45,19 +45,30 @@ static inline unsigned long hweight_long(unsigned long w) return sizeof(w) == 4 ? hweight32(w) : hweight64(w); } -#define HWEIGHT8(w) \ - ( (!!((w) & (1ULL << 0))) + \ - (!!((w) & (1ULL << 1))) + \ - (!!((w) & (1ULL << 2))) + \ - (!!((w) & (1ULL << 3))) + \ - (!!((w) & (1ULL << 4))) + \ - (!!((w) & (1ULL << 5))) + \ - (!!((w) & (1ULL << 6))) + \ +/* + * Clearly slow versions of the hweightN() functions, their benefit is + * of course compile time evaluation of constant arguments. + */ +#define HWEIGHT8(w) \ + ( BUILD_BUG_ON_ZERO(!__builtin_constant_p(w)) + \ + (!!((w) & (1ULL << 0))) + \ + (!!((w) & (1ULL << 1))) + \ + (!!((w) & (1ULL << 2))) + \ + (!!((w) & (1ULL << 3))) + \ + (!!((w) & (1ULL << 4))) + \ + (!!((w) & (1ULL << 5))) + \ + (!!((w) & (1ULL << 6))) + \ (!!((w) & (1ULL << 7))) ) -#define HWEIGHT16(w) (HWEIGHT8(w) + HWEIGHT8(w >> 8)) -#define HWEIGHT32(w) (HWEIGHT16(w) + HWEIGHT16(w >> 16)) -#define HWEIGHT64(w) (HWEIGHT32(w) + HWEIGHT32(w >> 32)) +#define HWEIGHT16(w) (HWEIGHT8(w) + HWEIGHT8((w) >> 8)) +#define HWEIGHT32(w) (HWEIGHT16(w) + HWEIGHT16((w) >> 16)) +#define HWEIGHT64(w) (HWEIGHT32(w) + HWEIGHT32((w) >> 32)) + +/* + * Type invariant version that simply casts things to the + * largest type. + */ +#define HWEIGHT(w) HWEIGHT64((u64)(w)) /** * rol32 - rotate a 32-bit value left -- cgit v1.2.3 From 447a194b393f32699607fd99617a40abd6a95114 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Mon, 1 Feb 2010 14:50:01 +0200 Subject: perf_events, x86: Fix bug in hw_perf_enable() We cannot assume that because hwc->idx == assign[i], we can avoid reprogramming the counter in hw_perf_enable(). The event may have been scheduled out and another event may have been programmed into this counter. Thus, we need a more robust way of verifying if the counter still contains config/data related to an event. This patch adds a generation number to each counter on each cpu. Using this mechanism we can verify reliabilty whether the content of a counter corresponds to an event. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: Mike Galbraith Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Frederic Weisbecker LKML-Reference: <4b66dc67.0b38560a.1635.ffffae18@mx.google.com> Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 556b0f4a668e..071a7db52549 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -478,9 +478,11 @@ struct hw_perf_event { union { struct { /* hardware */ u64 config; + u64 last_tag; unsigned long config_base; unsigned long event_base; int idx; + int last_cpu; }; struct { /* software */ s64 remaining; -- cgit v1.2.3 From d76a0812ac4139ceb54daab3cc70e1bd8bd9d43a Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Mon, 8 Feb 2010 17:06:01 +0200 Subject: perf_events: Add new start/stop PMU callbacks In certain situations, the kernel may need to stop and start the same event rapidly. The current PMU callbacks do not distinguish between stop and release (i.e., stop + free the resource). Thus, a counter may be released, then it will be immediately re-acquired. Event scheduling will again take place with no guarantee to assign the same counter. On some processors, this may event yield to failure to assign the event back due to competion between cores. This patch is adding a new pair of callback to stop and restart a counter without actually release the underlying counter resource. On stop, the counter is stopped, its values saved and that's it. On start, the value is reloaded and counter is restarted (on x86, actual restart is delayed until perf_enable()). Signed-off-by: Stephane Eranian [ added fallback to ->enable/->disable for all other PMUs fixed x86_pmu_start() to call x86_pmu.enable() merged __x86_pmu_disable into x86_pmu_stop() ] Signed-off-by: Peter Zijlstra LKML-Reference: <4b703875.0a04d00a.7896.ffffb824@mx.google.com> Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 071a7db52549..b08dfdad08cb 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -513,6 +513,8 @@ struct perf_event; struct pmu { int (*enable) (struct perf_event *event); void (*disable) (struct perf_event *event); + int (*start) (struct perf_event *event); + void (*stop) (struct perf_event *event); void (*read) (struct perf_event *event); void (*unthrottle) (struct perf_event *event); }; -- cgit v1.2.3 From 6e37738a2fac964583debe91099bc3248554f6e5 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 11 Feb 2010 13:21:58 +0100 Subject: perf_events: Simplify code by removing cpu argument to hw_perf_group_sched_in() Since the cpu argument to hw_perf_group_sched_in() is always smp_processor_id(), simplify the code a little by removing this argument and using the current cpu where needed. Signed-off-by: Peter Zijlstra Cc: David Miller Cc: Paul Mackerras Cc: Benjamin Herrenschmidt Cc: Frederic Weisbecker LKML-Reference: <1265890918.5396.3.camel@laptop> Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index b08dfdad08cb..d0e072c5b58a 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -772,7 +772,7 @@ extern int perf_event_task_disable(void); extern int perf_event_task_enable(void); extern int hw_perf_group_sched_in(struct perf_event *group_leader, struct perf_cpu_context *cpuctx, - struct perf_event_context *ctx, int cpu); + struct perf_event_context *ctx); extern void perf_event_update_userpage(struct perf_event *event); extern int perf_event_release_kernel(struct perf_event *event); extern struct perf_event * -- cgit v1.2.3