06:17 mlankhorst: pixelcluster: You broke the docs!
06:54 wens: MoeIcenowy: I'm looking at the verisilicon display driver, and the plane scanout coordinates vs buffer address seem a bit iffy. they seem wrong if clipping is involved
11:10 melissawen: hey, looks like this bug-fix commit (https://gitlab.freedesktop.org/drm/misc/kernel/-/commit/d7971640) disappeared in a merge conflict resolution. Should I resend it or can a drm-misc maintainer just cherry-pick/reapply?
12:50 dviola: is virgl going away?
14:06 mareko: unlikely
14:20 FireBurn: So I've got HDR working with the Vino driver on the newer dock. Everything is still a bit fradgile (epecally with ports 3 & 4) but it's promising
14:21 FireBurn: The really old dock arrives on saturday and hopefully that covers all the docks, there's USB monitors but they seem pretty rare and expensive
14:31 mareko: image clear/copy/blit/resolve perf tests finished: https://gitlab.freedesktop.org/mesa/gpu-ratemeter#imgbw-framebuffer-clear-image-clearcopyblit-and-msaa-image-clearcopyblitresolve-bandwidth-gbs
20:13 mlankhorst: Lyude: hey mind ack if I fixup a patch from thomas hellström to not break with nouveau support?
20:14 mlankhorst: ba9fdfabe28c ("drm/nouveau: Wire up dmem cgroups")
20:14 mlankhorst: breaks when I apply 'cgroup/dmem: Introduce struct dmem_cgroup_init for region initialization'
20:14 mlankhorst: https://patch.msgid.link/20260725100036.2372-3-thomas.hellstrom@linux.intel.com
20:16 mlankhorst: dakr: or ack from you^
20:18 airlied: mlankhorst: ack
20:19 mlankhorst: thanks, should I also add eviction?
20:23 airlied: if you feel like it
20:26 airlied: agd5f, hwentlan__ : with all the code motion in drm-next in amdgpu_dm.c, I'm quite weary some changes from fixes are going to get lost in drm-tip or worse when Linus merges
20:27 airlied: just want to make sure you can keep any eye on it
20:27 airlied: (also can we get amd trees into dim :-)
20:29 mlankhorst: Yeah can't test, but I'll see if eviction still works with pixelcluster's changes
20:36 mlankhorst: I think it might be a bit too much to write nouveau code for eviction too, but I'll test on xe, I'll happily allow nouveau eviction through drm-misc-next-fixes if required
21:37 mlankhorst: firefox really doesn't like suddenly being restricted to less vram and crashes, but the dmemcg eviction code appears to work
21:40 mlankhorst: neither does ut2004
21:44 dwfreed: oh man, I miss ut'04
21:47 mlankhorst: it's updated. :-)
21:47 mlankhorst: new linux patch too
21:48 dwfreed: Yeah, I saw that
21:49 dwfreed: I think there was a PCMag article about it a month or two ago
21:54 mlankhorst: I may have broken drm-tip, but can't fix it
21:55 mlankhorst: got to sleep now
21:55 mlankhorst: If no-one else fixes it, I'll do so tomorrow morning ASAP
22:12 airlied: mlankhorst: I do hope we can someday make firefox do the thing chromebooks can do, where the memory pressure cycle works