Re: [PATCH 0/8] media: imx: Miscalleanous format-related cleanups

[Steve Longerbeam <slongerbeam@gmail.com>]

Hi Laurent,

On 3/11/20 5:47 PM, Laurent Pinchart wrote:
> Hi Steve,
>
> On Wed, Mar 11, 2020 at 05:16:49PM -0700, Steve Longerbeam wrote:
>> Hi Laurent,
>>
>> I agree that the find/enum format code in imx-utils needs cleanup, it's
>> too confusing. I will be ready to give my ack to the imx-utils patches
>> once I do some smoke testing on a sabre target when I return from vacation.
>>
>> Note that Phillip also submitted a fixup patch to the find/enum format
>> code, IIRC it did more consolidating of the imx_media_pixfmt tables. I
>> can't find it and it has gotten lost, but I tested and provided a
>> reviewed-by at the time.
> I've found them in the mail archive. There were 3 patches, Hans said he
> would take the first two, but only the first one got merged. I'll take
> the two others and build on top of them, fixing the issues pointed out
> by the kbuild robot and addressing Hans concerns.

I found the thread finally. In fact I did some work on this set already, 
I fixed the kbuild warnings as well as added some function headers 
describing all the format search criteria arguments. Also changed a 
local var in enum_formats() to better document the enum algorithm.

Unless you've started on this work already, I will be returning from 
vaca tomorrow and can pick this up again, merging in your patches, as 
well as splitting up find|enum_formats() into find|enum_formats() and 
find|enum_codes(), which is the remaining suggestion from Hans.


Steve

>
>> On 3/10/20 9:18 AM, Laurent Pinchart wrote:
>>> Hello,
>>>
>>> This patch series started as an attempt to fix the format get and set
>>> subdev operations on the i.MX7 CSI-2 receiver subdev, which it does in
>>> patch 1/8. Patch 2/8 further cleans up the format-related code in that
>>> subdev.
>>>
>>> Patches 3/8 to 8/8 pushes the cleanups further as I was attempting to
>>> fix the format enumeration on the video node at the end of the pipeline.
>>> I realized as part of that effort that there's more work than I
>>> anticipated, and I'm currently evaluating the possible options.
>>> Nonetheless, I think the cleanups make sense even without what I wanted
>>> to build on top of them, so I'm sending them out already.
>>>
>>> Laurent Pinchart (8):
>>>     media: imx: imx7-mipi-csis: Cleanup and fix subdev pad format handling
>>>     media: imx: imx7-mipi-csis: Centralize initialization of pad formats
>>>     media: imx: utils: Inline init_mbus_colorimetry() in its caller
>>>     media: imx: utils: Handle Bayer format lookup through a selection flag
>>>     media: imx: utils: Simplify IPU format lookup and enumeration
>>>     media: imx: utils: Make imx_media_pixfmt handle variable number of
>>>       codes
>>>     media: imx: utils: Remove unneeded argument to (find|enum)_format()
>>>     media: imx: utils: Rename format lookup and enumeration functions
>>>
>>>    drivers/staging/media/imx/imx-ic-prp.c        |   8 +-
>>>    drivers/staging/media/imx/imx-ic-prpencvf.c   |   6 +-
>>>    drivers/staging/media/imx/imx-media-capture.c |  22 +-
>>>    .../staging/media/imx/imx-media-csc-scaler.c  |   2 +-
>>>    drivers/staging/media/imx/imx-media-csi.c     |  26 +-
>>>    drivers/staging/media/imx/imx-media-utils.c   | 313 ++++++++----------
>>>    drivers/staging/media/imx/imx-media-vdic.c    |   6 +-
>>>    drivers/staging/media/imx/imx-media.h         |  24 +-
>>>    drivers/staging/media/imx/imx7-media-csi.c    |  15 +-
>>>    drivers/staging/media/imx/imx7-mipi-csis.c    | 138 ++++----
>>>    10 files changed, 271 insertions(+), 289 deletions(-)