diff options
author | Borislav Petkov <bp@suse.de> | 2013-04-08 15:57:45 (GMT) |
---|---|---|
committer | Borislav Petkov <bp@suse.de> | 2013-04-16 09:50:51 (GMT) |
commit | 682469a5db6fade318a72406935b5000186e5643 (patch) | |
tree | 0578a2dd19ecf2f3349905ddb5b69991fe3e6d72 /drivers/infiniband/hw/qib/qib_rc.c | |
parent | 55a36b65ee7107d6bb557c96fd202c4e90164542 (diff) | |
download | linux-fsl-qoriq-682469a5db6fade318a72406935b5000186e5643.tar.xz |
x86, AMD: Correct {rd,wr}msr_amd_safe warnings
The idea with those routines is to slowly phase them out and not call
them on anything else besides K8. They even have a check for that which,
when called too early, fails. Let me explain:
It gets the cpuinfo_x86 pointer from the per_cpu array and when this
happens for cpu0, before its boot_cpu_data has been copied back to the
per_cpu array in smp_store_boot_cpu_info(), we get an empty struct and
thus the check fails.
Use boot_cpu_data directly instead.
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: http://lkml.kernel.org/r/1365436666-9837-4-git-send-email-bp@alien8.de
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Diffstat (limited to 'drivers/infiniband/hw/qib/qib_rc.c')
0 files changed, 0 insertions, 0 deletions