From 9f498cc5be7e013d8d6e4c616980ed0ffc8680d2 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 23 Jul 2009 14:46:33 +0200 Subject: perf_counter: Full task tracing In order to be able to distinguish between no samples due to inactivity and no samples due to task ended, Arjan asked for PERF_EVENT_EXIT events. This is useful to the boot delay instrumentation (bootchart) app. This patch changes the PERF_EVENT_FORK to be emitted on every clone, and adds PERF_EVENT_EXIT to be emitted on task exit, after the task's counters have been closed. This task tracing is controlled through: attr.comm || attr.mmap and through the new attr.task field. Suggested-by: Arjan van de Ven Cc: Paul Mackerras Cc: Anton Blanchard Signed-off-by: Peter Zijlstra [ cleaned up perf_counter.h a bit ] Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index bd15d7a5f5c..e604e6ef72d 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -181,8 +181,9 @@ struct perf_counter_attr { freq : 1, /* use freq, not period */ inherit_stat : 1, /* per task counts */ enable_on_exec : 1, /* next exec enables */ + task : 1, /* trace fork/exit */ - __reserved_1 : 51; + __reserved_1 : 50; __u32 wakeup_events; /* wakeup every n events */ __u32 __reserved_2; @@ -308,6 +309,15 @@ enum perf_event_type { */ PERF_EVENT_COMM = 3, + /* + * struct { + * struct perf_event_header header; + * u32 pid, ppid; + * u32 tid, ptid; + * }; + */ + PERF_EVENT_EXIT = 4, + /* * struct { * struct perf_event_header header; @@ -323,6 +333,7 @@ enum perf_event_type { * struct { * struct perf_event_header header; * u32 pid, ppid; + * u32 tid, ptid; * }; */ PERF_EVENT_FORK = 7, -- cgit v1.2.3 From f413cdb80ce00ec1a4d0ab949b5d96c81cae7f75 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 7 Aug 2009 01:25:54 +0200 Subject: perf_counter: Fix/complete ftrace event records sampling This patch implements the kernel side support for ftrace event record sampling. A new counter sampling attribute is added: PERF_SAMPLE_TP_RECORD which requests ftrace events record sampling. In this case if a PERF_TYPE_TRACEPOINT counter is active and a tracepoint fires, we emit the tracepoint binary record to the perfcounter event buffer, as a sample. Result, after setting PERF_SAMPLE_TP_RECORD attribute from perf record: perf record -f -F 1 -a -e workqueue:workqueue_execution perf report -D 0x21e18 [0x48]: event: 9 . . ... raw event: size 72 bytes . 0000: 09 00 00 00 01 00 48 00 d0 c7 00 81 ff ff ff ff ......H........ . 0010: 0a 00 00 00 0a 00 00 00 21 00 00 00 00 00 00 00 ........!...... . 0020: 2b 00 01 02 0a 00 00 00 0a 00 00 00 65 76 65 6e +...........eve . 0030: 74 73 2f 31 00 00 00 00 00 00 00 00 0a 00 00 00 ts/1........... . 0040: e0 b1 31 81 ff ff ff ff ....... . 0x21e18 [0x48]: PERF_EVENT_SAMPLE (IP, 1): 10: 0xffffffff8100c7d0 period: 33 The raw ftrace binary record starts at offset 0020. Translation: struct trace_entry { type = 0x2b = 43; flags = 1; preempt_count = 2; pid = 0xa = 10; tgid = 0xa = 10; } thread_comm = "events/1" thread_pid = 0xa = 10; func = 0xffffffff8131b1e0 = flush_to_ldisc() What will come next? - Userspace support ('perf trace'), 'flight data recorder' mode for perf trace, etc. - The unconditional copy from the profiling callback brings some costs however if someone wants no such sampling to occur, and needs to be fixed in the future. For that we need to have an instant access to the perf counter attribute. This is a matter of a flag to add in the struct ftrace_event. - Take care of the events recursivity! Don't ever try to record a lock event for example, it seems some locking is used in the profiling fast path and lead to a tracing recursivity. That will be fixed using raw spinlock or recursivity protection. - [...] - Profit! :-) Signed-off-by: Frederic Weisbecker Cc: Li Zefan Cc: Tom Zanussi Cc: Arnaldo Carvalho de Melo Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Steven Rostedt Cc: Paul Mackerras Cc: Pekka Enberg Cc: Gabriel Munteanu Cc: Lai Jiangshan Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index e604e6ef72d..a67dd5c5b6d 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -121,8 +121,9 @@ enum perf_counter_sample_format { PERF_SAMPLE_CPU = 1U << 7, PERF_SAMPLE_PERIOD = 1U << 8, PERF_SAMPLE_STREAM_ID = 1U << 9, + PERF_SAMPLE_TP_RECORD = 1U << 10, - PERF_SAMPLE_MAX = 1U << 10, /* non-ABI */ + PERF_SAMPLE_MAX = 1U << 11, /* non-ABI */ }; /* @@ -413,6 +414,11 @@ struct perf_callchain_entry { __u64 ip[PERF_MAX_STACK_DEPTH]; }; +struct perf_tracepoint_record { + int size; + char *record; +}; + struct task_struct; /** @@ -681,6 +687,7 @@ struct perf_sample_data { struct pt_regs *regs; u64 addr; u64 period; + void *private; }; extern int perf_counter_overflow(struct perf_counter *counter, int nmi, -- cgit v1.2.3 From 3a43ce68ae1758fa6a839386025ef45acb6baa22 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 8 Aug 2009 04:26:37 +0200 Subject: perf_counter: Fix tracepoint sampling to be part of generic sampling Based on Peter's comments, make tracepoint sampling generic just like all the other sampling bits are. This is a rename with no code changes: - PERF_SAMPLE_TP_RECORD to PERF_SAMPLE_RAW - struct perf_tracepoint_record to perf_raw_record We want the system in place that transport tracepoints raw samples events into the perf ring buffer to be generalized and usable by any type of counter. Reported-by; Peter Zijlstra Signed-off-by: Frederic Weisbecker Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Mike Galbraith Cc: Paul Mackerras LKML-Reference: <1249698400-5441-4-git-send-email-fweisbec@gmail.com> Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index a67dd5c5b6d..2aabe43c1d0 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -121,7 +121,7 @@ enum perf_counter_sample_format { PERF_SAMPLE_CPU = 1U << 7, PERF_SAMPLE_PERIOD = 1U << 8, PERF_SAMPLE_STREAM_ID = 1U << 9, - PERF_SAMPLE_TP_RECORD = 1U << 10, + PERF_SAMPLE_RAW = 1U << 10, PERF_SAMPLE_MAX = 1U << 11, /* non-ABI */ }; @@ -414,9 +414,9 @@ struct perf_callchain_entry { __u64 ip[PERF_MAX_STACK_DEPTH]; }; -struct perf_tracepoint_record { - int size; - char *record; +struct perf_raw_record { + u32 size; + void *data; }; struct task_struct; @@ -687,7 +687,7 @@ struct perf_sample_data { struct pt_regs *regs; u64 addr; u64 period; - void *private; + struct perf_raw_record *raw; }; extern int perf_counter_overflow(struct perf_counter *counter, int nmi, -- cgit v1.2.3 From a044560c3a1f0ad75ce685c1ed7604820b9ed319 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 10 Aug 2009 11:16:52 +0200 Subject: perf_counter: Correct PERF_SAMPLE_RAW output PERF_SAMPLE_* output switches should unconditionally output the correct format, as they are the only way to unambiguously parse the PERF_EVENT_SAMPLE data. Signed-off-by: Peter Zijlstra Acked-by: Frederic Weisbecker Cc: Arnaldo Carvalho de Melo Cc: Mike Galbraith Cc: Paul Mackerras LKML-Reference: <1249896447.17467.74.camel@twins> Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index 2aabe43c1d0..a9d823a93fe 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -369,6 +369,8 @@ enum perf_event_type { * * { u64 nr, * u64 ips[nr]; } && PERF_SAMPLE_CALLCHAIN + * { u32 size; + * char data[size];}&& PERF_SAMPLE_RAW * }; */ PERF_EVENT_SAMPLE = 9, -- cgit v1.2.3 From 28402971d869e26271b25301011f667d3a5640c3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 13 Aug 2009 10:13:22 +0200 Subject: perf_counter: Provide hw_perf_counter_setup_online() APIs Provide weak aliases for hw_perf_counter_setup_online(). This is used by the BTS patches (for v2.6.32), but it interacts with fixes so propagate this upstream. (it has no effect as of yet) Also export perf_counter_output() to architecture code. Cc: Peter Zijlstra Cc: Paul Mackerras Cc: Frederic Weisbecker LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index a9d823a93fe..2b36afe6ae0 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -694,6 +694,8 @@ struct perf_sample_data { extern int perf_counter_overflow(struct perf_counter *counter, int nmi, struct perf_sample_data *data); +extern void perf_counter_output(struct perf_counter *counter, int nmi, + struct perf_sample_data *data); /* * Return 1 for a software counter, 0 for a hardware counter -- cgit v1.2.3 From 3dab77fb1bf89664bb1c9544607159dcab6f7d57 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 13 Aug 2009 11:47:53 +0200 Subject: perf: Rework/fix the whole read vs group stuff Replace PERF_SAMPLE_GROUP with PERF_SAMPLE_READ and introduce PERF_FORMAT_GROUP to deal with group reads in a more generic way. This allows you to get group reads out of read() as well. Signed-off-by: Peter Zijlstra Cc: Corey J Ashford Cc: Paul Mackerras Cc: stephane eranian LKML-Reference: <20090813103655.117411814@chello.nl> Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 47 +++++++++++++++++++++++++++++++++----------- 1 file changed, 36 insertions(+), 11 deletions(-) (limited to 'include/linux/perf_counter.h') diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index 2b36afe6ae0..b53f7006cc4 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -115,7 +115,7 @@ enum perf_counter_sample_format { PERF_SAMPLE_TID = 1U << 1, PERF_SAMPLE_TIME = 1U << 2, PERF_SAMPLE_ADDR = 1U << 3, - PERF_SAMPLE_GROUP = 1U << 4, + PERF_SAMPLE_READ = 1U << 4, PERF_SAMPLE_CALLCHAIN = 1U << 5, PERF_SAMPLE_ID = 1U << 6, PERF_SAMPLE_CPU = 1U << 7, @@ -127,16 +127,32 @@ enum perf_counter_sample_format { }; /* - * Bits that can be set in attr.read_format to request that - * reads on the counter should return the indicated quantities, - * in increasing order of bit value, after the counter value. + * The format of the data returned by read() on a perf counter fd, + * as specified by attr.read_format: + * + * struct read_format { + * { u64 value; + * { u64 time_enabled; } && PERF_FORMAT_ENABLED + * { u64 time_running; } && PERF_FORMAT_RUNNING + * { u64 id; } && PERF_FORMAT_ID + * } && !PERF_FORMAT_GROUP + * + * { u64 nr; + * { u64 time_enabled; } && PERF_FORMAT_ENABLED + * { u64 time_running; } && PERF_FORMAT_RUNNING + * { u64 value; + * { u64 id; } && PERF_FORMAT_ID + * } cntr[nr]; + * } && PERF_FORMAT_GROUP + * }; */ enum perf_counter_read_format { PERF_FORMAT_TOTAL_TIME_ENABLED = 1U << 0, PERF_FORMAT_TOTAL_TIME_RUNNING = 1U << 1, PERF_FORMAT_ID = 1U << 2, + PERF_FORMAT_GROUP = 1U << 3, - PERF_FORMAT_MAX = 1U << 3, /* non-ABI */ + PERF_FORMAT_MAX = 1U << 4, /* non-ABI */ }; #define PERF_ATTR_SIZE_VER0 64 /* sizeof first published struct */ @@ -343,10 +359,8 @@ enum perf_event_type { * struct { * struct perf_event_header header; * u32 pid, tid; - * u64 value; - * { u64 time_enabled; } && PERF_FORMAT_ENABLED - * { u64 time_running; } && PERF_FORMAT_RUNNING - * { u64 parent_id; } && PERF_FORMAT_ID + * + * struct read_format values; * }; */ PERF_EVENT_READ = 8, @@ -364,11 +378,22 @@ enum perf_event_type { * { u32 cpu, res; } && PERF_SAMPLE_CPU * { u64 period; } && PERF_SAMPLE_PERIOD * - * { u64 nr; - * { u64 id, val; } cnt[nr]; } && PERF_SAMPLE_GROUP + * { struct read_format values; } && PERF_SAMPLE_READ * * { u64 nr, * u64 ips[nr]; } && PERF_SAMPLE_CALLCHAIN + * + * # + * # The RAW record below is opaque data wrt the ABI + * # + * # That is, the ABI doesn't make any promises wrt to + * # the stability of its content, it may vary depending + * # on event, hardware, kernel version and phase of + * # the moon. + * # + * # In other words, PERF_SAMPLE_RAW contents are not an ABI. + * # + * * { u32 size; * char data[size];}&& PERF_SAMPLE_RAW * }; -- cgit v1.2.3