[PATCH v2] powerpc: Don't clobber fr0/vs0 during fp|altivec register save

Timothy Pearson tpearson at raptorengineering.com
Fri Dec 1 03:29:26 AEDT 2023

----- Original Message -----
> From: "Michael Ellerman" <mpe at ellerman.id.au>
> To: "Timothy Pearson" <tpearson at raptorengineering.com>
> Cc: "Jens Axboe" <axboe at kernel.dk>, "regressions" <regressions at lists.linux.dev>, "npiggin" <npiggin at gmail.com>,
> "christophe leroy" <christophe.leroy at csgroup.eu>, "linuxppc-dev" <linuxppc-dev at lists.ozlabs.org>
> Sent: Tuesday, November 28, 2023 6:57:01 AM
> Subject: Re: [PATCH v2] powerpc: Don't clobber fr0/vs0 during fp|altivec register  save

> Michael Ellerman <mpe at ellerman.id.au> writes:
>> Timothy Pearson <tpearson at raptorengineering.com> writes:
>>> Just wanted to check back and see if this patch was going to be queued
>>> up soon?  We're still having to work around / advertise the data
>>> destruction issues the underlying bug is causing on e.g. Debian
>>> Stable.
>> Yeah I'll apply it this week, so it will be in rc4.
> I reworked the change log to include the exact call path I identified
> instead of the more high level description you had. And tweaked a few
> other bits of wording and so on, apparently fr0 is a kernelism, the ABI
> and binutils calls it f0.
> I'm not sure how wedded you were to your change log, so if you dislike
> my edits let me know and we can come up with a joint one.
> The actual patch is unchanged.
> cheers

The commit message looks OK to me.  I've also seen application crashes as a result of the register corruption, but that may be a minor detail that isn't really worth updating things over at this point -- those come from e.g. glibc using vs0 as part of a path that processes pointer information, typically seen where there's a need to replicate the same pointer to adjacent fields in a data struct.

More information about the Linuxppc-dev mailing list