Message ID | 20250117010718.2328467-5-seanjc@google.com (mailing list archive) |
---|---|
State | New |
Headers | show |
Series | [GIT,PULL] KVM: Selftests changes for 6.14 | expand |
On Fri, Jan 17, 2025 at 2:07 AM Sean Christopherson <seanjc@google.com> wrote: > > FYI, the "LLC references/misses" patch exposed a latent failure on SKX/CLX/CPL[*] > (who's brilliant idea was it to use "CPL" for a CPU code name on x86?). Dapeng > is following up with the uarch folks to understand what's going on. If -rc1 is > immiment and we don't have a fix, my plan is to have the test only assert that > the count is non-zero, and then go with a more precise fix if one arises. So based on the thread there is a root cause and fix---the test is just counting on an unrelated event. Paolo
On Mon, Jan 20, 2025, Paolo Bonzini wrote: > On Fri, Jan 17, 2025 at 2:07 AM Sean Christopherson <seanjc@google.com> wrote: > > > > FYI, the "LLC references/misses" patch exposed a latent failure on SKX/CLX/CPL[*] > > (who's brilliant idea was it to use "CPL" for a CPU code name on x86?). Dapeng > > is following up with the uarch folks to understand what's going on. If -rc1 is > > immiment and we don't have a fix, my plan is to have the test only assert that > > the count is non-zero, and then go with a more precise fix if one arises. > > So based on the thread there is a root cause and fix---the test is > just counting on an unrelated event. Oh, yeah. Sorry, forgot to follow-up here.