On Wed, Jan 03, 2024 at 02:22:23PM -0800, Paul E. McKenney wrote: > Hello! > > Since some time between v5.19 and v6.4, long-running rcutorture tests > would (rarely but intolerably often) have all guests on a given host die > simultaneously with something like an instruction fault or a segmentation > violation. > > Each bisection step required 20 hosts running 10 hours each, and > this eventually fingered commit c59a1f106f5c ("KVM: x86/pmu: Add > IA32_PEBS_ENABLE MSR emulation for extended PEBS"). Although this commit > is certainly messing with things that could possibly cause all manner > of mischief, I don't immediately see a smoking gun. Except that the > commit prior to this one is rock solid. > > Just to make things a bit more exciting, bisection in mainline proved > to be problematic due to bugs of various kinds that hid this one. I was > therefore forced to bisect among the commits backported to the internal > v5.19-based kernel, which fingered the backported version of the patch > called out above. Ah, and so why do I believe that this is a problem in mainline rather than just (say) a backporting mistake? Because this issue was first located in v6.4, which already has this commit included. Thanx, Paul > Please note that this is not (yet) an emergency. I will just continue > to run rcutorture on v5.19-based hypervisors in the meantime. > > Any suggestions for debugging or fixing? > > Thanx, Paul