Re: notifier is skipped in some situations

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Nov 24, 2017 at 6:17 PM, Sakari Ailus
<sakari.ailus@xxxxxxxxxxxxxxx> wrote:
> Hi Tomasz,
>
> On Fri, Nov 24, 2017 at 06:03:26PM +0900, Tomasz Figa wrote:
>> Hi Sakari,
>>
>> We have the following graph:
>>
>>     ISP (registers notifier for v4l2_dev)
>>
>>      |
>>
>>     PHY (registers notifier for v4l2_subdev, just like sensors for
>> flash/focuser)
>>
>>   |       \
>>
>> sensor0   sensor1
>>
>> ...
>>
>> Both ISP and PHY are completely separate drivers not directly aware of
>> each other, since we have several different PHY IP blocks that we need
>> to support and some of them are multi-functional, such as CSI+DSI PHY
>> and need to be supported by drivers independent from the ISP driver.
>
> That should work fine. In the above case there are two notifiers, indeed,
> but they're not expecting the *same* sub-devices.

Got it.

Jacob, could you make sure there are no mistakes in the Device Tree source?

Best regards,
Tomasz

>
> What this could be about is that in some version of the set I disabled the
> complete callback on the sub-notifiers for two reasons: there was no need
> seen for them and the complete callback is problematic in general (there's
> been discussion on that, mostly related to earlier versions of the fwnode
> parsing patchset, on #v4l and along the Renesas rcar-csi2 patchsets).
>
>>
>> Best regards,
>> Tomasz
>>
>>
>> On Fri, Nov 24, 2017 at 5:55 PM, Sakari Ailus
>> <sakari.ailus@xxxxxxxxxxxxxxx> wrote:
>> > Hi Jacob,
>> >
>> > On Fri, Nov 24, 2017 at 09:00:14AM +0800, Jacob Chen wrote:
>> >> Hi Sakari,
>> >>
>> >> I encountered a problem when using async sub-notifiers.
>> >>
>> >> It's like that:
>> >>     There are two notifiers, and they are waiting for one subdev.
>> >>     When this subdev is probing, only one notifier is completed and
>> >> the other one is skipped.
>> >
>> > Do you have a graph that has two master drivers (that register the
>> > notifier) and both are connected to the same sub-device? Could you provide
>> > exact graph you have?
>> >
>> >>
>> >> I found that in v15 of patch "v4l: async: Allow binding notifiers to
>> >> sub-devices", "v4l2_async_notifier_complete" is replaced by
>> >> v4l2_async_notifier_call_complete, which make it only complete one
>> >> notifier.
>> >>
>> >> Why is it changed? Can this be fixed?
>> >
>> > --
>> > Sakari Ailus
>> > sakari.ailus@xxxxxxxxxxxxxxx
>
> --
> Sakari Ailus
> sakari.ailus@xxxxxxxxxxxxxxx



[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]
  Powered by Linux