On Sat, Sep 28, 2024 at 6:30 PM Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Sat, 28 Sept 2024 at 08:33, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
> >
> > Apologize for the late pull request; all the traveling made things a
> > bit messy. Also, we have a known regression here on ancient processors
> > and will fix it next week.
>
> Gaah. Don't leave it hanging like that. When somebody reports a
> problem, I need to know if it's this known one.
> I've pulled it, but you really need to add a pointer to "look, this is
> the known one, we have a fix in the works"
If that's what you mean, it was not reported by users (and it's very
unlikely that it will, unless they run selftests on pre-2008
processors or with non-standard module parameters). It's a NULL
pointer dereference on VM shutdown, caused by the selftests added by
commit b4ed2c67d275 ("KVM: selftests: Test slot move/delete with slot
zap quirk enabled/disabled").
It's also not reproducible yet outside selftests since the bug is in a
new API; which is also why we didn't revert with prejudice and didn't
go too much into detail above.
Paolo
