4.14-stable review patch. If anyone has any objections, please let me know.
------------------
From: Song Liu <songliubrav...@fb.com>
commit bd903afeb504db5655a45bb4cf86f38be5b1bf62 upstream.
In ctx_resched(), EVENT_FLEXIBLE should be sched_out when EVENT_PINNED is
added. However, ctx_resched() calculates ctx_event_type before checking
this condition. As a result, pinned events will NOT get higher priority
than flexible events.
The following shows this issue on an Intel CPU (where ref-cycles can
only use one hardware counter).
1. First start:
perf stat -C 0 -e ref-cycles -I 1000
2. Then, in the second console, run:
perf stat -C 0 -e ref-cycles:D -I 1000
The second perf uses pinned events, which is expected to have higher
priority. However, because it failed in ctx_resched(). It is never
run.
This patch fixes this by calculating ctx_event_type after re-evaluating
event_type.
Reported-by: Ephraim Park <ephiep...@fb.com>
Signed-off-by: Song Liu <songliubrav...@fb.com>
Signed-off-by: Peter Zijlstra (Intel) <pet...@infradead.org>
Cc: <jo...@redhat.com>
Cc: <kernel-t...@fb.com>
Cc: Alexander Shishkin <alexander.shish...@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <a...@redhat.com>
Cc: Jiri Olsa <jo...@redhat.com>
Cc: Linus Torvalds <torva...@linux-foundation.org>
Cc: Stephane Eranian <eran...@google.com>
Cc: Thomas Gleixner <t...@linutronix.de>
Cc: Vince Weaver <vincent.wea...@maine.edu>
Fixes: 487f05e18aa4 ("perf/core: Optimize event rescheduling on active
contexts")
Link: http://lkml.kernel.org/r/20180306055504.3283731-1-songliubrav...@fb.com
Signed-off-by: Ingo Molnar <mi...@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gre...@linuxfoundation.org>
---
kernel/events/core.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -2322,7 +2322,7 @@ static void ctx_resched(struct perf_cpu_
struct perf_event_context *task_ctx,
enum event_type_t event_type)
{
- enum event_type_t ctx_event_type = event_type & EVENT_ALL;
+ enum event_type_t ctx_event_type;
bool cpu_event = !!(event_type & EVENT_CPU);
/*
@@ -2332,6 +2332,8 @@ static void ctx_resched(struct perf_cpu_
if (event_type & EVENT_PINNED)
event_type |= EVENT_FLEXIBLE;
+ ctx_event_type = event_type & EVENT_ALL;
+
perf_pmu_disable(cpuctx->ctx.pmu);
if (task_ctx)
task_ctx_sched_out(cpuctx, task_ctx, event_type);
