[PATCH v2] media: verisilicon: Fix crash when probing encoder

[Marek Szyprowski]

On 23.05.2023 16:54, Michael Tretter wrote:
> On Tue, 23 May 2023 12:50:42 +0200, Linux regression tracking (Thorsten Leemhuis) wrote:
>> CCing the Regression list and a bunch of other people that were CCed in
>> threads that look related:
> Thanks!
>
>> On 23.05.23 00:38, Diederik de Haas wrote:
>>> On Monday, 22 May 2023 18:17:39 CEST Benjamin Gaignard wrote:
>>>> Le 20/05/2023 à 00:34, Diederik de Haas a écrit :
>>>>> On Thursday, 13 April 2023 21:52:50 CEST Nicolas Dufresne wrote:
>>> [...]
>>>>> When I booted into my 6.4-rc1 (but also rc2) kernel on my
>>>>> Pine64 Quartz64 Model A, I noticed a crash which seems the same as
>>>>> above, but I didn't have such a crash with my 6.3 kernel.
>>>>> Searching for 'hantro' led me to this commit as the most likely culprit
>>>>> but when I build a new 6.4-rcX kernel with this commit reverted,
>>>>> I still had this crash.
>>>>> Do you have suggestions which commit would then be the likely culprit?
>>>> This patch fix the crash at boot time, revert it doesn't seem to be the
>>>> solution. Maybe this proposal from Marek can help you ?
>>>>
>>>> https://patchwork.kernel.org/project/linux-media/patch/20230421104759.2236463-1-m.szyprowski@samsung.com/
>>> That helped :) After applying that patch I no longer have the crash.
>>> Thanks!
>> That regression fix is now a month old, but not yet merged afaics --
>> guess due to Nicolas comment that wasn't addressed yet and likely
>> requires a updated patch.
> I agree with Nicolas comment on that patch and it needs to be updated.
>
>> Michael afaics a week ago posted a patch that to my *very limited
>> understanding of things* (I hope I don't confuse matters here!) seems to
>> address the same problem, but slightly differently:
>> https://lore.kernel.org/all/20230516091209.3098262-1-m.tretter@pengutronix.de/
> Correct, my patch addresses the same problem.
>
>> No reply yet.
>>
>> That's all a bit unfortunate, as it's not how regression fixes should be
>> dealt with -- and caused multiple people headaches that could have been
>> avoided. :-/
>>
>> But well, things happen. But it leads to the question:
>>
>> How can we finally address the issue quickly now to ensure is doesn't
>> cause headaches for even more people?
>>
>> Marek, Michael, could you work on a patch together that we then get
>> somewhat fast-tracked to Linus to avoid him getting even more unhappy
>> about the state of things[1]?
> Marek, if you have an updated patch, I will happily test and review it.
> Otherwise, please take a look at my patch.

Go ahead with your, my was just a blind try eliminating the oops during 
driver probe.


Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland