Hi Ruben, hi Thomas, I’ll pick up the drm_simple_encoder_init() TODO first. Once that’s in a good shape, I can look into the logging conversion in GUD as a follow-up.
I’ll send a patch for the encoder cleanup once I have something ready. Thanks for the pointers! regards, Hardik On Wednesday, 25 February 2026 at 17:33, Ruben Wauters <[email protected]> wrote: > Hi > > On Wed, 2026-02-25 at 09:48 +0100, Thomas Zimmermann wrote: > > Hi > > > > Am 22.02.26 um 07:45 schrieb Hardik Phalet: > > > Hi Ruben, > > > > > > I am looking to work on a small starter task in DRM and was considering > > > converting logging in drivers/gpu/drm/gud from dev_info/dev_warn/dev_err > > > to the drm_*() helpers where a struct drm_device is already (or readily) > > > available. > > My apologies, I seem to have missed this message in the first place, I > think this is mainly fine. > > > > > > Since GUD is a USB-backed DRM driver and can have multiple active > > > instances, using drm_*() would make log messages instance-aware and > > > consistent with DRM drivers. > > > > > > Before preparing a patch series, I wanted to check whether this kind of > > > conversion is welcome for GUD, and whether there are any parts of the > > > driver > > > you would prefer to keep using dev_*() instead of drm_*(). > > In my opinion, logging that relates specifically to USB should use dev > rather than logging that relates to drm, if it does relate to drm then > it should be fine to convert them. I'm happy to accept a patch. > > > > > > If this is fine, I plan to keep the changes mechanical and scoped only to > > > call sites where drm_device is already in scope, preserving logging > > > semantics. > > > > I'm not Ruben, but generally speaking that conversion of the logging is > > ok. But there's not much in gud to convert and the early calls in > > gud_probe() should rather not be converted. > > I agree with Thomas here, gud_probe() as a function is still a bit of a > mess, but since it relates to early init, and the early parts are > mainly to do with USB and capabilities assessment, they should remain > dev > > > > If you're looking for an easy task, maybe try open-coding > > drm_simple_encoder_init(). There's an upcoming TODO item at [1]. > > I'm happy for you to do conversions if you want, but this also seems to > be a good task to do. If not I probably will end up looking at it > myself eventually. > > Ruben > > > > Best regards > > Thomas > > > > [1] > > https://lore.kernel.org/dri-devel/[email protected]/T/#Z2e.:..:20260224153656.261351-17-tzimmermann::40suse.de:1Documentation:gpu:todo.rst > > > > > > > Thanks, > > > Hardik > > > >
