diff options
author | Peter Zijlstra <a.p.zijlstra@chello.nl> | 2009-03-25 19:39:37 +0100 |
---|---|---|
committer | Ingo Molnar <mingo@elte.hu> | 2009-06-18 14:46:11 +0200 |
commit | 43a21ea81a2400992561146327c4785ce7f7be38 (patch) | |
tree | d4974c0ff9d7f40291515c5c0cf7e0d51abccb66 /include/linux/perf_counter.h | |
parent | d3a9262e59f7fb83c6d44df3b2b1460ed57d3ea1 (diff) |
perf_counter: Add event overlow handling
Alternative method of mmap() data output handling that provides
better overflow management and a more reliable data stream.
Unlike the previous method, that didn't have any user->kernel
feedback and relied on userspace keeping up, this method relies on
userspace writing its last read position into the control page.
It will ensure new output doesn't overwrite not-yet read events,
new events for which there is no space left are lost and the
overflow counter is incremented, providing exact event loss
numbers.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Diffstat (limited to 'include/linux/perf_counter.h')
-rw-r--r-- | include/linux/perf_counter.h | 40 |
1 files changed, 28 insertions, 12 deletions
diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index a7d3a61a59b..0765e8e6984 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -236,10 +236,16 @@ struct perf_counter_mmap_page { /* * Control data for the mmap() data buffer. * - * User-space reading this value should issue an rmb(), on SMP capable - * platforms, after reading this value -- see perf_counter_wakeup(). + * User-space reading the @data_head value should issue an rmb(), on + * SMP capable platforms, after reading this value -- see + * perf_counter_wakeup(). + * + * When the mapping is PROT_WRITE the @data_tail value should be + * written by userspace to reflect the last read data. In this case + * the kernel will not over-write unread data. */ __u64 data_head; /* head in the data section */ + __u64 data_tail; /* user-space written tail */ }; #define PERF_EVENT_MISC_CPUMODE_MASK (3 << 0) @@ -275,6 +281,15 @@ enum perf_event_type { /* * struct { + * struct perf_event_header header; + * u64 id; + * u64 lost; + * }; + */ + PERF_EVENT_LOST = 2, + + /* + * struct { * struct perf_event_header header; * * u32 pid, tid; @@ -313,26 +328,26 @@ enum perf_event_type { /* * When header.misc & PERF_EVENT_MISC_OVERFLOW the event_type field - * will be PERF_RECORD_* + * will be PERF_SAMPLE_* * * struct { * struct perf_event_header header; * - * { u64 ip; } && PERF_RECORD_IP - * { u32 pid, tid; } && PERF_RECORD_TID - * { u64 time; } && PERF_RECORD_TIME - * { u64 addr; } && PERF_RECORD_ADDR - * { u64 config; } && PERF_RECORD_CONFIG - * { u32 cpu, res; } && PERF_RECORD_CPU + * { u64 ip; } && PERF_SAMPLE_IP + * { u32 pid, tid; } && PERF_SAMPLE_TID + * { u64 time; } && PERF_SAMPLE_TIME + * { u64 addr; } && PERF_SAMPLE_ADDR + * { u64 config; } && PERF_SAMPLE_CONFIG + * { u32 cpu, res; } && PERF_SAMPLE_CPU * * { u64 nr; - * { u64 id, val; } cnt[nr]; } && PERF_RECORD_GROUP + * { u64 id, val; } cnt[nr]; } && PERF_SAMPLE_GROUP * * { u16 nr, * hv, * kernel, * user; - * u64 ips[nr]; } && PERF_RECORD_CALLCHAIN + * u64 ips[nr]; } && PERF_SAMPLE_CALLCHAIN * }; */ }; @@ -424,6 +439,7 @@ struct file; struct perf_mmap_data { struct rcu_head rcu_head; int nr_pages; /* nr of data pages */ + int writable; /* are we writable */ int nr_locked; /* nr pages mlocked */ atomic_t poll; /* POLL_ for wakeups */ @@ -433,8 +449,8 @@ struct perf_mmap_data { atomic_long_t done_head; /* completed head */ atomic_t lock; /* concurrent writes */ - atomic_t wakeup; /* needs a wakeup */ + atomic_t lost; /* nr records lost */ struct perf_counter_mmap_page *user_page; void *data_pages[0]; |