Re: [PATCH] dma-fence: Make dma_fence_dedup_array() robust against 0-count input

[Christian König]

On 6/29/26 12:16, Philipp Stanner wrote:
> On Mon, 2026-06-29 at 11:52 +0200, Christian König wrote:
>> On 6/29/26 11:06, Philipp Stanner wrote:
>>> On Mon, 2026-06-29 at 10:52 +0200, Christian König wrote:
>>>> On 6/29/26 10:49, Philipp Stanner wrote:
>>>>>>
>>>>>> I will add a CC stable before pushing to drm-misc-fixes.
>>>>>
>>>>> No offense intended or taken, but don't the DRM rules say that
>>>>> things
>>>>> do not get merged while there are outstanding concerns or
>>>>> significant
>>>>> points in review feedback?
>>>>
>>>> I haven't seen that before writing the response.
>>>>
>>>> I usually go over my mails till the end and wait a couple of
>>>> hours before pushing anything.
>>>>
>>>>> What about my comments?
>>>>
>>>> Looks valid to me as well, but I think that is a separate issue.
>>>
>>> But if we keep it an integer for now, and if that check is added,
>>> and
>>> it most certainly should also catch negative integers, shouldn't
>>> it?
>>
>> Maybe with a WARN_ON(), but not as regular code path.
>>
>> The thing is I think we need to backport this fix to stable kernels,
>> but switching from signed to unsigned is only a minor cleanup when no
>> real users are currently affected.
> 
> Fine by me.

Thanks, I updated the commit message a bit and added Fixes and CC stable tags and pushed the result to drm-misc-fixes.

@Baineng it would be cool if you could write a test cases for this as well. E.g. just ~10 line in drivers/dma-buf/st-dma-fence-unwrap.c.

Thanks in advance,
Christian.

> 
> P.