perf, x86: Improve the PEBS ABI
authorPeter Zijlstra <peterz@infradead.org>
Thu, 8 Apr 2010 21:03:20 +0000 (23:03 +0200)
committerIngo Molnar <mingo@elte.hu>
Fri, 7 May 2010 09:31:02 +0000 (11:31 +0200)
commitab608344bcbde4f55ec4cd911b686b0ce3eae076
treeebd38efabfaab59d6de11a24143d70e1eec36fae
parent2b0b5c6fe9b383f3cf35a0a6371c9d577bd523ff
perf, x86: Improve the PEBS ABI

Rename perf_event_attr::precise to perf_event_attr::precise_ip and
widen it to 2 bits. This new field describes the required precision of
the PERF_SAMPLE_IP field:

  0 - SAMPLE_IP can have arbitrary skid
  1 - SAMPLE_IP must have constant skid
  2 - SAMPLE_IP requested to have 0 skid
  3 - SAMPLE_IP must have 0 skid

And modify the Intel PEBS code accordingly. The PEBS implementation
now supports up to precise_ip == 2, where we perform the IP fixup.

Also s/PERF_RECORD_MISC_EXACT/&_IP/ to clarify its meaning, this bit
should be set for each PERF_SAMPLE_IP field known to match the actual
instruction triggering the event.

This new scheme allows for a PEBS mode that uses the buffer for more
than a single event.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/kernel/cpu/perf_event.c
arch/x86/kernel/cpu/perf_event_intel.c
arch/x86/kernel/cpu/perf_event_intel_ds.c
include/linux/perf_event.h
tools/perf/builtin-top.c
tools/perf/util/parse-events.c