Re: [PATCH v2 4/6] Revert "media: iris: Add sanity check for stop streaming"

[Bryan O'Donoghue <bryan.odonoghue@linaro.org>]

On 05/01/2026 09:51, Dikshita Agarwal wrote:
> 
> 
> On 12/30/2025 3:55 PM, Bryan O'Donoghue wrote:
>> On 29/12/2025 06:31, Dikshita Agarwal wrote:
>>> Revert the check that skipped stop_streaming when the instance was in
>>> IRIS_INST_ERROR, as it caused multiple regressions:
>>>
>>> 1. Buffers were not returned to vb2 when the instance was already in
>>>      error state, triggering warnings in the vb2 core because buffer
>>>      completion was skipped.
>>>
>>> 2. If a session failed early (e.g. unsupported configuration), the
>>>      instance transitioned to IRIS_INST_ERROR. When userspace attempted
>>>      to stop streaming for cleanup, stop_streaming was skipped due to the
>>>      added check, preventing proper teardown and leaving the firmware
>>>      in an inconsistent state.
>>>
>>> Fixes: ad699fa78b59 ("media: iris: Add sanity check for stop streaming")
>>> Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
>>> ---
>>>    drivers/media/platform/qcom/iris/iris_vb2.c | 8 +++-----
>>>    1 file changed, 3 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/drivers/media/platform/qcom/iris/iris_vb2.c
>>> b/drivers/media/platform/qcom/iris/iris_vb2.c
>>> index
>>> db8768d8a8f61c9ceb04e423d0a769d35114e20e..139b821f7952feb33b21a7045aef9e8a4782aa3c 100644
>>> --- a/drivers/media/platform/qcom/iris/iris_vb2.c
>>> +++ b/drivers/media/platform/qcom/iris/iris_vb2.c
>>> @@ -231,8 +231,6 @@ void iris_vb2_stop_streaming(struct vb2_queue *q)
>>>            return;
>>>          mutex_lock(&inst->lock);
>>> -    if (inst->state == IRIS_INST_ERROR)
>>> -        goto exit;
>>>          if (!V4L2_TYPE_IS_OUTPUT(q->type) &&
>>>            !V4L2_TYPE_IS_CAPTURE(q->type))
>>> @@ -243,10 +241,10 @@ void iris_vb2_stop_streaming(struct vb2_queue *q)
>>>            goto exit;
>>>      exit:
>>> -    if (ret) {
>>> -        iris_helper_buffers_done(inst, q->type, VB2_BUF_STATE_ERROR);
>>> +    iris_helper_buffers_done(inst, q->type, VB2_BUF_STATE_ERROR);
>>> +    if (ret)
>>>            iris_inst_change_state(inst, IRIS_INST_ERROR);
>>> -    }
>>> +
>>>        mutex_unlock(&inst->lock);
>>>    }
>>>   
>>
>> This revert looks strange, should be something like:
>>
>> commit 9b6b11d31918722b4522b8982141d7b9646c0e48 (HEAD -> next-6.19-camss-v2)
>> Author: Bryan O'Donoghue <bod@kernel.org>
>> Date:   Tue Dec 30 10:20:01 2025 +0000
>>
>>      Revert "media: iris: Add sanity check for stop streaming"
>>
>>      This reverts commit ad699fa78b59241c9d71a8cafb51525f3dab04d4.
>>
>>      Everything is broken I give up.
>>
>>      Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
>>
>> diff --git a/drivers/media/platform/qcom/iris/iris_vb2.c
>> b/drivers/media/platform/qcom/iris/iris_vb2.c
>> index db8768d8a8f61..139b821f7952f 100644
>> --- a/drivers/media/platform/qcom/iris/iris_vb2.c
>> +++ b/drivers/media/platform/qcom/iris/iris_vb2.c
>> @@ -231,8 +231,6 @@ void iris_vb2_stop_streaming(struct vb2_queue *q)
>>                  return;
>>
>>          mutex_lock(&inst->lock);
>> -       if (inst->state == IRIS_INST_ERROR)
>> -               goto exit;
>>
>>          if (!V4L2_TYPE_IS_OUTPUT(q->type) &&
>>              !V4L2_TYPE_IS_CAPTURE(q->type))
>> @@ -243,10 +241,10 @@ void iris_vb2_stop_streaming(struct vb2_queue *q)
>>                  goto exit;
>>
>>   exit:
>> -       if (ret) {
>> -               iris_helper_buffers_done(inst, q->type, VB2_BUF_STATE_ERROR);
>> +       iris_helper_buffers_done(inst, q->type, VB2_BUF_STATE_ERROR);
>> +       if (ret)
>>                  iris_inst_change_state(inst, IRIS_INST_ERROR);
>> -       }
>> +
>>          mutex_unlock(&inst->lock);
>>   }
>>
>> Just `git revert ad699fa78b59241c9d71a8cafb51525f3dab04d4` and add your
>> commit log ?!
> 
> Yeah I did the same, revert and changed the commit message.
> 
> BTW, I don't see any difference in my change and your commit, anything I am
> missing here?

Take this example, I believe the "This reverts commit xxx" is added by 
the revert command and its best practice to include it.

commit afb9917d9b374ecb77d478c2a052e20875c6e232
Author: Christian Brauner <brauner@kernel.org>
Date:   Fri Dec 5 13:50:31 2025 +0100

     Revert "net/socket: convert sock_map_fd() to FD_ADD()"

     This reverts commit 245f0d1c622b0183ce4f44b3e39aeacf78fae594.

I can just add it back in with the PR though its NBD.

---
bod