Re: [PATCH v2 05/12] media: ipu-bridge: Use v4l2_fwnode for unknown rotations

[Hans de Goede <hansg@kernel.org>]

Hi,

On 8-Jul-25 16:58, Ricardo Ribalda wrote:
> On Tue, 8 Jul 2025 at 14:21, Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
>>
>> Hi Ricardo,
>>
>> On Tue, Jul 08, 2025 at 02:09:28PM +0200, Ricardo Ribalda wrote:
>>> On Tue, 8 Jul 2025 at 11:22, Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
>>>>
>>>> Hi Ricardo,
>>>>
>>>> On Tue, Jul 08, 2025 at 11:16:25AM +0200, Ricardo Ribalda wrote:
>>>>> Hi Sakari
>>>>>
>>>>> Thanks for your review
>>>>>
>>>>> On Mon, 7 Jul 2025 at 23:45, Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
>>>>>>
>>>>>> Hi Ricardo,
>>>>>>
>>>>>> On Thu, Jun 05, 2025 at 05:52:58PM +0000, Ricardo Ribalda wrote:
>>>>>>> The v4l2_fwnode_device_properties contains information about the
>>>>>>> rotation. Use it if the ssdb data is inconclusive.
>>>>>>
>>>>>> As SSDB and _PLD provide the same information, are they always aligned? Do
>>>>>> you have any experience on how is this actually in firmware?
>>>>>
>>>>> Not really, in ChromeOS we are pretty lucky to control the firmware.
>>>>>
>>>>> @HdG Do you have some experience/opinion here?
>>>>>
>>>>>>
>>>>>> _PLD is standardised so it would seem reasonable to stick to that -- if it
>>>>>> exists. Another approach could be to pick the one that doesn't translate to
>>>>>> a sane default (0°).
>>>>>
>>>>> I'd rather stick to the current prioritization unless there is a
>>>>> strong argument against it. Otherwise there is a chance that we will
>>>>> have regressions (outside CrOS)
>>>>
>>>> My point was rather there are no such rules currently for rotation: only
>>>> SSDB was being used by the IPU bridge to obtain the rotation value,
>>>> similarly only _PLD is consulted when it comes to orientation.
>>>
>>> So something like this:?
>>>
>>> static u32 ipu_bridge_parse_rotation(struct acpi_device *adev,
>>>                                      struct ipu_sensor_ssdb *ssdb,
>>>                                      struct
>>> v4l2_fwnode_device_properties *props)
>>> {
>>>         if (props->rotation != V4L2_FWNODE_PROPERTY_UNSET)
>>>                 return props->rotation;
>>>
>>>         switch (ssdb->degree) {
>>>         case IPU_SENSOR_ROTATION_NORMAL:
>>>                 return 0;
>>>         case IPU_SENSOR_ROTATION_INVERTED:
>>>                 return 180;
>>>         }
>>>
>>>         dev_warn(ADEV_DEV(adev),
>>>                  "Unknown rotation %d. Assume 0 degree rotation\n",
>>>                  ssdb->degree);
>>
>> Maybe:
>>
>>         acpi_handle_warn(acpi_device_handle(adev), ...);
>>
>> ?
>>
>>>         return 0;
>>> }
>>
>> Looks good to me. Maybe something similar for orientation?
> 
> Do you mean using ssdb also for orientation or using acpi_handle_warn?
> 
> 
> I cannot find anything related to orientation for SSDB
> https://github.com/coreboot/coreboot/blob/main/src/drivers/intel/mipi_camera/chip.h#L150
> 
> Am I looking in the right place?

I believe that orientation is only available in the PLD,
so for orientation we can just use the value returned in
v4l2_fwnode_device_properties defaulting to front when
it is not set.

Otherwise I agree with what has been discussed in this
thread (for this patch) so far.

Regards,

Hans