[PATCH v2 1/4] ppc64le: Revert "powerpc: Optimized strncmp for power10" (CVE-2025-5745)

[Sachin Monga]

Thank you all for closing the reverts.


On 17/06/25 3:36 am, Carlos O'Donell wrote:
> On 6/16/25 5:58 PM, Siddhesh Poyarekar wrote:
>> I verified that the reverts are clean by applying these commits and 
>> then running this command:
>>
>> git log --pretty=oneline HEAD~4..HEAD | awk '{print $1}' |
>> while read ref; do
>>    diff -u \
>>      <(git show -R --color=never \
>>        $(git show $ref | sed -n 's/This reverts commit \(.*\)/\1/p')) \
>>      <(git show --color=never $ref) | less;
>> done
>>
>> and confirmed that barring minor differences in Makefile contexts 
>> (which is expected), they are all straight reverts.
>>
>> Thanks,
>> Sid
>>
>> On 2025-06-16 16:20, Carlos O'Donell wrote:
>>> This reverts commit 23f0d81608d0ca6379894ef81670cf30af7fd081
>>>
>>> Reason for revert: Power10 strncmp clobbers non-volatile vector
>>> registers (Bug 33060)
>>>
>>> Tested on ppc64le with no regressions.
>>> ---
>>
>> OK.
>>
>> Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
>
> Sorry, I was *so* excited I pushed these without adding your RBs into 
> the commits.
>
> I'll add your RBs into the backports to active branches.
>
> All 4 pushed and both CVEs now have fixed commits.
>
> I need to update the advisory.
>