perf/x86: Remove PERF_X86_EVENT_COMMITTED
authorPeter Zijlstra <peterz@infradead.org>
Thu, 14 Mar 2019 11:58:52 +0000 (12:58 +0100)
committerIngo Molnar <mingo@kernel.org>
Wed, 3 Apr 2019 07:25:31 +0000 (09:25 +0200)
commit1f6a1e2d7d7135280125418b25d35aebd6fa0952
tree7abf92c2fdea80d7c2487fa058ddc1d91eaa8cf3
parent21d65555cd878c8d6ff2b2902e6c16b2ca73f33d
perf/x86: Remove PERF_X86_EVENT_COMMITTED

The flag PERF_X86_EVENT_COMMITTED is used to find uncommitted events
for which to call put_event_constraint() when scheduling fails.

These are the newly added events to the list, and must form, per
definition, the tail of cpuc->event_list[]. By computing the list
index of the last successfull schedule, then iteration can start there
and the flag is redundant.

There are only 3 callers of x86_schedule_events(), notably:

 - x86_pmu_add()
 - x86_pmu_commit_txn()
 - validate_group()

For x86_pmu_add(), cpuc->n_events isn't updated until after
schedule_events() succeeds, therefore cpuc->n_events points to the
desired index.

For x86_pmu_commit_txn(), cpuc->n_events is updated, but we can
trivially compute the desired value with cpuc->n_txn -- the number of
events added in this transaction.

For validate_group(), we can make the rule for x86_pmu_add() work by
simply setting cpuc->n_events to 0 before calling schedule_events().

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Stephane Eranian <eranian@google.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/events/core.c
arch/x86/events/perf_event.h