Re: [PATCH v2 1/2] media: imx: csi: Disable CSI immediately after last EOF

[Steve Longerbeam <slongerbeam@gmail.com>]

On 1/18/19 11:01 AM, Steve Longerbeam wrote:
>
>
> On 1/18/19 2:24 AM, Philipp Zabel wrote:
>> On Thu, 2019-01-17 at 12:49 -0800, Steve Longerbeam wrote:
>>> Disable the CSI immediately after receiving the last EOF before stream
>>> off (and thus before disabling the IDMA channel).
>>>
>>> This fixes a complete system hard lockup on the SabreAuto when 
>>> streaming
>>> from the ADV7180, by repeatedly sending a stream off immediately 
>>> followed
>>> by stream on:
>>>
>>> while true; do v4l2-ctl  -d4 --stream-mmap --stream-count=3; done
>>>
>>> Eventually this either causes the system lockup or EOF timeouts at all
>>> subsequent stream on, until a system reset.
>>>
>>> The lockup occurs when disabling the IDMA channel at stream off. 
>>> Disabling
>>> the CSI before disabling the IDMA channel appears to be a reliable 
>>> fix for
>>> the hard lockup.
>>>
>>> Fixes: 4a34ec8e470cb ("[media] media: imx: Add CSI subdev driver")
>>>
>>> Reported-by: Gaël PORTAY <gael.portay@collabora.com>
>>> Signed-off-by: Steve Longerbeam <slongerbeam@gmail.com>
>>> Cc: stable@vger.kernel.org
>>> ---
>>> Changes in v2:
>>> - restore an empty line
>>> - Add Fixes: and Cc: stable
>>> ---
>>>   drivers/staging/media/imx/imx-media-csi.c | 6 ++++--
>>>   1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/staging/media/imx/imx-media-csi.c 
>>> b/drivers/staging/media/imx/imx-media-csi.c
>>> index e18f58f56dfb..e0f6f88e2e70 100644
>>> --- a/drivers/staging/media/imx/imx-media-csi.c
>>> +++ b/drivers/staging/media/imx/imx-media-csi.c
>>> @@ -681,6 +681,8 @@ static void csi_idmac_stop(struct csi_priv *priv)
>>>       if (ret == 0)
>>>           v4l2_warn(&priv->sd, "wait last EOF timeout\n");
>>>   +    ipu_csi_disable(priv->csi);
>>> +
>> Can you add a short comment why this call is here? Since now
>> csi_idmac_stop is kind of a misnomer and symmetry with csi(_idmac)_start
>> is broken, I think this is a bit un-obvious.
>
> Yeah. I think a cleaner, more symmetric solution would be to split up 
> csi_idmac_stop.
>
>>
>> Also note that now the error path of csi_start() will now call
>> ipu_csi_disable() while the CSI is disabled. This happens to work
>> because that just calls ipu_module_disable(), which is not refcounted.
>
> Thanks for catching. Splitting up csi_idmac_stop will fix this. 
> Working on that.


Well turns out Peter Seiderer kindly provided a solution they have been 
using for some time on custom imx6 hardware. His solution is to disable 
the SMFC before the IDMA channel, which I have tested on the SabreAuto 
and also works. This is a much simpler change, so I will post v3 that 
does this instead.

Steve