x86, MCE, AMD: Drop software-defined bank in error thresholding
authorBorislav Petkov <bp@suse.de>
Tue, 21 Oct 2014 20:19:59 +0000 (22:19 +0200)
committerBorislav Petkov <bp@suse.de>
Tue, 21 Oct 2014 20:28:48 +0000 (22:28 +0200)
commita3a529d104ec5149fb9a667dce988635941be1ed
tree7be73a235ce379c859e47710b5192316a51de89e
parent69b957583580bf40624553c64d802fefb54199cb
x86, MCE, AMD: Drop software-defined bank in error thresholding

Aravind had the good question about why we're assigning a
software-defined bank when reporting error thresholding errors instead
of simply using the bank which reports the last error causing the
overflow.

Digging through git history, it pointed to

95268664390b ("[PATCH] x86_64: mce_amd support for family 0x10 processors")

which added that functionality. The problem with this, however, is that
tools don't know about software-defined banks and get puzzled. So drop
that K8_MCE_THRESHOLD_BASE and simply use the hw bank reporting the
thresholding interrupt.

Save us a couple of MSR reads while at it.

Reported-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>
Link: https://lkml.kernel.org/r/5435B206.60402@amd.com
Signed-off-by: Borislav Petkov <bp@suse.de>
arch/x86/include/asm/mce.h
arch/x86/kernel/cpu/mcheck/mce_amd.c