[PATCH 3/3] ppc64le: Revert "powerpc: Optimized strcmp for power10" (CVE-2025-5702)
Carlos O'Donell
carlos@redhat.com
Mon Jun 16 20:27:22 GMT 2025
On 6/16/25 9:46 AM, Florian Weimer wrote:
> * Carlos O'Donell:
>
>> This reverts commit 3367d8e180848030d1646f088759f02b8dfe0d6f
>>
>> Reason for revert: Power10 strcmp clobbers non-volatile vector
>> registers (Bug 33056)
>> ---
>> sysdeps/powerpc/powerpc64/le/power10/strcmp.S | 233 ------------------
>> sysdeps/powerpc/powerpc64/multiarch/Makefile | 2 +-
>> .../powerpc64/multiarch/ifunc-impl-list.c | 4 -
>> .../powerpc64/multiarch/strcmp-power10.S | 26 --
>> sysdeps/powerpc/powerpc64/multiarch/strcmp.c | 4 -
>> 5 files changed, 1 insertion(+), 268 deletions(-)
>> delete mode 100644 sysdeps/powerpc/powerpc64/le/power10/strcmp.S
>> delete mode 100644 sysdeps/powerpc/powerpc64/multiarch/strcmp-power10.S
>
> Should this one be reverted as well?
Yes, I did not include that commit as I removed the file entirely.
Let me expand it out for completeness.
Posted v2: https://inbox.sourceware.org/libc-alpha/20250616202221.1162240-1-carlos@redhat.com/T/#t
I double checked again for any missing commits that were not just
copyright year changes.
My intent was to capture a revert of the substantive changes in
reverse order.
> commit 90bcc8721ef82b7378d2b080141228660e862d56
> Author: Amrita H S <amritahs@linux.vnet.ibm.com>
> Date: Fri Dec 15 11:48:17 2023 -0500
>
> powerpc: Fix performance issues of strcmp power10
>
> Current implementation of strcmp for power10 has
> performance regression for multiple small sizes
> and alignment combination.
>
> Most of these performance issues are fixed by this
> patch. The compare loop is unrolled and page crosses
> of unrolled loop is handled.
>
> Thanks to Paul E. Murphy for helping in fixing the
> performance issues.
>
> Signed-off-by: Amrita H S <amritahs@linux.vnet.ibm.com>
> Co-Authored-By: Paul E. Murphy <murphyp@linux.ibm.com>
> Reviewed-by: Rajalakshmi Srinivasaraghavan <rajis@linux.ibm.com>
>
> Thanks,
> Florian
>
--
Cheers,
Carlos.
More information about the Libc-alpha
mailing list