00:27DemiMarie: Is there a way to efficiently undo premultiplied alpha in a shader?
12:37jfalempe: tzimmermann: can you help me solve this build error: https://lists.freedesktop.org/archives/dri-devel/2024-November/477101.html ? should DRM_CLIENT_LIB depends (or select) DRM_CLIENT ?
13:07tzimmermann: jfalempe, you building DRM_CLIENT_LIB without DRM_CLIENT_SETUP
13:07tzimmermann: 'you are'
13:08tzimmermann: try to do 'select DRM_CLIENT_LIB if DRM_LOG' at https://github.com/intel-lab-lkp/linux/commit/627ef6c7f28c30d7b4ab7f88761f1c74df87091f#diff-f00f9dddc98c59c08ed70bca42328592bfe71aec11cfa92e04b72b9010f08daeR237
13:08tzimmermann: you cannot select DRM_CLIENT_LIB unconditionally
13:09tzimmermann: there needs to be a single unditional select line for each client ala 'select DRM_CLIENT_LIB if <client token>'
13:10tzimmermann: 'unconditional'
13:12jfalempe: tzimmermann: ok, so all clients should add a line here, I will do that
13:14tzimmermann: jfalempe, yeah. the complexity in these kconfig rules is a concession to making everything flexible and modular. otherwise, we'd have large modules with everything, which isn't great either
13:15tzimmermann: jfalempe, and please name DRM_LOG to DRM_CLIENT_LOG. so that it's clear that it's a client and not a DRM core/helper feature
13:16tzimmermann: i cannot rename DRM_FBDEV_EMULATION obviouslym but new code should be clear
13:17jfalempe: tzimmermann: ok that make sense.
13:19tzimmermann: jfalempe, DRM_CLIENT_LOG should also select DRM_CLIENT, as it's build on top of these interfaces
13:20jfalempe: tzimmermann: Yes I'm adding that too, I just noticed that difference with fbdev emulation.
13:34jfalempe: tzimmermann: I have a drm/log v8 ready with this changes, but if you have other comments, please send them to me.
13:34tzimmermann: i just sent one
13:35tzimmermann: :)
13:36jfalempe: ok thanks, I will send v8 on Tuesday then, so you have time to review (Monday is off here).
13:40tzimmermann: jfalempe, i'm thinking that we could put the client code into a subdirectory
13:40tzimmermann: ?
13:41jfalempe: tzimmermann: Yes all the files built in drm_client_lib can go in a subdir.
13:42tzimmermann: i'm about to make patch that moves everything into clients/
13:42tzimmermann: except the core client code and seom fbdev helpers that require a rework
13:43jfalempe: tzimmermann: agreed, I will wait a bit for drm/log then ;)
13:43tzimmermann: no code changes intented
13:43tzimmermann: great
14:03Ermine:looks for someone to review https://lore.kernel.org/dri-devel/20241107141133.13624-1-mustela@erminea.space/T/#u
17:48alyssa: could I solicit a review on the common code pieces of https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31964 ?
17:49alyssa: It is significantly simpler than the first version, ~30 lines of nir changes iirc
17:49alyssa: thank u