]> git.baikalelectronics.ru Git - kernel.git/commit
perf/x86/kvm: Fix Broadwell Xeon stepping in isolation_ucodes[]
authorJim Mattson <jmattson@google.com>
Thu, 22 Apr 2021 00:18:34 +0000 (17:18 -0700)
committerPeter Zijlstra <peterz@infradead.org>
Thu, 22 Apr 2021 12:36:01 +0000 (14:36 +0200)
commitc488efb479c9bc15b5ca2a53a82c1057746b547c
treef53b83638635ae507f58751932438200e8898a2a
parent8f2d1d459f79fe72035cf54c79c9c06ddabfe734
perf/x86/kvm: Fix Broadwell Xeon stepping in isolation_ucodes[]

The only stepping of Broadwell Xeon parts is stepping 1. Fix the
relevant isolation_ucodes[] entry, which previously enumerated
stepping 2.

Although the original commit was characterized as an optimization, it
is also a workaround for a correctness issue.

If a PMI arrives between kvm's call to perf_guest_get_msrs() and the
subsequent VM-entry, a stale value for the IA32_PEBS_ENABLE MSR may be
restored at the next VM-exit. This is because, unbeknownst to kvm, PMI
throttling may clear bits in the IA32_PEBS_ENABLE MSR. CPUs with "PEBS
isolation" don't suffer from this issue, because perf_guest_get_msrs()
doesn't report the IA32_PEBS_ENABLE value.

Fixes: 7257c4c5cba20 ("perf/x86/kvm: Avoid unnecessary work in guest filtering")
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Peter Shier <pshier@google.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Link: https://lkml.kernel.org/r/20210422001834.1748319-1-jmattson@google.com
arch/x86/events/intel/core.c