perf,x86: fix kernel crash with PEBS/BTS after suspend/resume
authorStephane Eranian <eranian@google.com>
Fri, 15 Mar 2013 13:26:07 +0000 (14:26 +0100)
committerBen Hutchings <ben@decadent.org.uk>
Wed, 27 Mar 2013 02:40:54 +0000 (02:40 +0000)
commit8da1a618f37700141f1cf6e3a9c16dc1b1e261c9
tree93c90365323a03b14ef17bafb64aa0a39cf2d852
parentdc71dd5c5bd96038a90f4efc0c127e21ea86bba1
perf,x86: fix kernel crash with PEBS/BTS after suspend/resume

commit 1d9d8639c063caf6efc2447f5f26aa637f844ff6 upstream.

This patch fixes a kernel crash when using precise sampling (PEBS)
after a suspend/resume. Turns out the CPU notifier code is not invoked
on CPU0 (BP). Therefore, the DS_AREA (used by PEBS) is not restored properly
by the kernel and keeps it power-on/resume value of 0 causing any PEBS
measurement to crash when running on CPU0.

The workaround is to add a hook in the actual resume code to restore
the DS Area MSR value. It is invoked for all CPUS. So for all but CPU0,
the DS_AREA will be restored twice but this is harmless.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
arch/x86/kernel/cpu/perf_event_intel_ds.c
arch/x86/power/cpu.c
include/linux/perf_event.h