Re: [PATCH 5.15 000/923] 5.15.3-rc3 review

[Kees Cook <keescook@chromium.org>]

On Thu, Nov 18, 2021 at 09:12:59AM +0100, Greg Kroah-Hartman wrote:
> On Wed, Nov 17, 2021 at 04:16:51PM -0800, Kees Cook wrote:
> > On Wed, Nov 17, 2021 at 03:50:17PM -0800, Linus Torvalds wrote:
> > > Sorry for top-posting and quoting this all, but the actual people
> > > involved with the wchan changes don't seem to be on the participant
> > > list.
> > 
> > Adding more folks from a private report and
> > https://bugzilla.kernel.org/show_bug.cgi?id=215031
> > 
> > and for the new people, here's a lore link for this thread:
> > https://lore.kernel.org/stable/YZV02RCRVHIa144u@fedora64.linuxtx.org/
> > 
> > 
> > FWIW, earlier bisection pointed to the stable backport of
> > 5d1ceb3969b6b2e47e2df6d17790a7c5a20fcbb4 being the primary culprit.
> > At first glance it seems to me that the problem with -stable is that an
> > unvetted subset of the wchan refactoring series landed in -stable.
> 
> What would be the vetted subset?  :)

The ones with "Cc: stable" ;)

> Anyway, I have now dropped the following patches that were in the
> 5.15.3-rc tree:
> 	bc9bbb81730e ("x86: Fix get_wchan() to support the ORC unwinder")
> 	42a20f86dc19 ("sched: Add wrapper for get_wchan() to keep task blocked")
> 	5d1ceb3969b6 ("x86: Fix __get_wchan() for !STACKTRACE")
> 
> And will push out another -rc release to let people test.
> 
> If there are any other commits I should have also dropped, please let me
> know.

That looks right to me. These three were part of Peter's much larger
wchan refactoring (of which 42a20f86dc19 was a core part).

Thanks!

-- 
Kees Cook