perf/x86/intel/bts: Fix BTS PMI detection

Since BTS doesn't have a dedicated PMI status bit, the driver needs to
take extra care to check for the condition that triggers it to avoid
spurious NMI warnings.

Regardless of the local BTS context state, the only way of knowing that
the NMI is ours is to compare the write pointer against the interrupt
threshold.

Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
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: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: vince@deater.net
Link: http://lkml.kernel.org/r/20160906132353.19887-5-alexander.shishkin@linux.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This commit is contained in:
Alexander Shishkin 2016-09-06 16:23:52 +03:00 committed by Ingo Molnar
parent a9a94401c2
commit 4d4c474124
1 changed files with 15 additions and 4 deletions

View File

@ -446,26 +446,37 @@ bts_buffer_reset(struct bts_buffer *buf, struct perf_output_handle *handle)
int intel_bts_interrupt(void)
{
struct debug_store *ds = this_cpu_ptr(&cpu_hw_events)->ds;
struct bts_ctx *bts = this_cpu_ptr(&bts_ctx);
struct perf_event *event = bts->handle.event;
struct bts_buffer *buf;
s64 old_head;
int err = -ENOSPC;
int err = -ENOSPC, handled = 0;
/*
* The only surefire way of knowing if this NMI is ours is by checking
* the write ptr against the PMI threshold.
*/
if (ds->bts_index >= ds->bts_interrupt_threshold)
handled = 1;
/*
* this is wrapped in intel_bts_enable_local/intel_bts_disable_local,
* so we can only be INACTIVE or STOPPED
*/
if (READ_ONCE(bts->state) == BTS_STATE_STOPPED)
return 0;
return handled;
buf = perf_get_aux(&bts->handle);
if (!buf)
return handled;
/*
* Skip snapshot counters: they don't use the interrupt, but
* there's no other way of telling, because the pointer will
* keep moving
*/
if (!buf || buf->snapshot)
if (buf->snapshot)
return 0;
old_head = local_read(&buf->head);
@ -473,7 +484,7 @@ int intel_bts_interrupt(void)
/* no new data */
if (old_head == local_read(&buf->head))
return 0;
return handled;
perf_aux_output_end(&bts->handle, local_xchg(&buf->data_size, 0),
!!local_xchg(&buf->lost, 0));