Fix mcelog regression
Sorry for bypassing normal channels, but this looks like a trivial regression fix to me, but I'm getting pushback from my co-maintainer and from Ingo.
1. This used to work
2. Now it doesn't
3. People have complained
Previous threads:
https://lkml.org/lkml/2015/1/30/641
https://lkml.org/lkml/2015/2/11/437
The following changes since commit bfa76d49576599a4b9f9b7a71f23d73d6dcff735:
Linux 3.19 (2015-02-08 18:54:22 -0800)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras.git tags/please-pull-fixmcelog
for you to fetch changes up to a2413d8b2982fb6823cdb66704938a9a345d0a7d:
x86/mce: Fix regression. All error records should report via /dev/mcelog (2015-02-09 09:36:53 -0800)
----------------------------------------------------------------
Fix regression - functions on the mce notifier chain should not be able to decide that an event should not be logged
----------------------------------------------------------------
Tony Luck (1):
x86/mce: Fix regression. All error records should report via /dev/mcelog
arch/x86/kernel/cpu/mcheck/mce.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)
Quoted 11 lines · Luck, Tony
Sorry for bypassing normal channels, but this looks like a trivial regression fix to me, but I'm getting pushback from my co-maintainer and from Ingo.
1. This used to work
2. Now it doesn't
3. People have complainedPrevious threads:
https://lkml.org/lkml/2015/1/30/641
https://lkml.org/lkml/2015/2/11/437
So this fix is perfectly OK to me. The primary thing I objected to (which objection you did not quote!!), was the initial approach, the ugly vendor-specific switch you added:
https://lkml.org/lkml/2015/2/9/237
The latest replacement commit you added looks clean - and /dev/mcelog is certainly a legacy we have to keep.
Thanks,
Ingo