Revert "perf/x86/intel, watchdog: Switch NMI watchdog to ref cycles on x86"
authorIngo Molnar <mingo@kernel.org>
Sun, 10 Jul 2016 18:58:36 +0000 (20:58 +0200)
committerIngo Molnar <mingo@kernel.org>
Sun, 10 Jul 2016 18:58:36 +0000 (20:58 +0200)
commit44530d588e142a96cf0cd345a7cb8911c4f88720
treef1b2a9d46904fd3479c320267691ee89d172b3dc
parent46866b59dfbe9bf99bb1323ce1f3fd2073a81aa3
Revert "perf/x86/intel, watchdog: Switch NMI watchdog to ref cycles on x86"

This reverts commit 2c95afc1e83d93fac3be6923465e1753c2c53b0a.

Stephane reported the following regression:

 > Since Andi added:
 >
 > commit 2c95afc1e83d93fac3be6923465e1753c2c53b0a
 > Author: Andi Kleen <ak@linux.intel.com>
 > Date:   Thu Jun 9 06:14:38 2016 -0700
 >
 >    perf/x86/intel, watchdog: Switch NMI watchdog to ref cycles on x86
 >
 > $ perf stat -e ref-cycles ls
 >   <not counted> ....
 >
 > fails systematically because the ref-cycles is now used by the
 > watchdog and given this is a system-wide pinned event, it monopolizes
 > the fixed counter 2 which is the only counter able to measure this event.

Since the next merge window is near, fix the regression for now
by reverting the commit.

Reported-by: Stephane Eranian <eranian@google.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/apic/hw_nmi.c
include/linux/nmi.h
kernel/watchdog.c