perf: Fix running time accounting

commit 00a2916f7f82c348a2a94dbb572874173bc308a3 upstream.

A recent fix to the shadow timestamp inadvertly broke the running time
accounting.

We must not update the running timestamp if we fail to schedule the
event, the event will not have ran. This can (and did) result in
negative total runtime because the stopped timestamp was before the
running timestamp (we 'started' but never stopped the event -- because
it never really started we didn't have to stop it either).

Reported-and-Tested-by: Vince Weaver <vincent.weaver@maine.edu>
Fixes: 72f669c0086f ("perf: Update shadow timestamp before add event")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Shaohua Li <shli@fb.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 130a5f637c0ba9f4d8e00d94e83ad30e5ab62266..26f387f243702c44d76481a732d13e2d89747dfb 100644 (file)
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1886,8 +1886,6 @@ event_sched_in(struct perf_event *event,
 
        perf_pmu_disable(event->pmu);
 
-       event->tstamp_running += tstamp - event->tstamp_stopped;
-
        perf_set_shadow_time(event, ctx, tstamp);
 
        perf_log_itrace_start(event);
@@ -1899,6 +1897,8 @@ event_sched_in(struct perf_event *event,
                goto out;
        }
 
+       event->tstamp_running += tstamp - event->tstamp_stopped;
+
        if (!is_software_event(event))
                cpuctx->active_oncpu++;
        if (!ctx->nr_active++)