02:17 kode54: [Jan30 15:33] amdgpu 0000:03:00.0: [drm] REG_WAIT timeout 1us * 100 tries - dcn32_program_compbuf_size line:139
02:17 kode54: this appears to happen every time the displays are powered off using wlopm
02:17 kode54: would amdgpu.runpm=0 fix that?
02:36 johnny0: it might, especially if the displays are connected via DP
05:15 kode54: johnny0: yes, DP
05:15 kode54: same on both machines
05:16 kode54: same model/vendor card, same motherboard model
08:14 kode54: johnny0: runpm=0 did not work, at least not when I removed the ppfeaturemask override
13:03 johnny0: kode54: bummer, was worth a shot
13:45 karolherbst: pepp: 666a6eb871d5dec79362bdc5d16f15915eb52f96 is causing flickering in davinci resolve at least with radeonsi + rusticl. Not really sure why though
13:49 karolherbst: looks fine with ROCm, so maybe something changing on the CL side of things with that
13:49 karolherbst: it's pre GFX12 hardware
13:51 pepp: karolherbst: can you file a ticket please? I'll take a look
13:53 karolherbst: done
13:55 pepp: thanks
14:04 soreau: I've been getting flickering textures that shows frame buffers from other clients, outputs or black, on mouse cursors and firefox especially
14:06 soreau: seems it might be triggered when I do something like start wf-recorder (with h264_vaapi) or some wayfire GL effect
14:07 soreau: currently the mouse cursor is black when mousing over geany
14:07 soreau: latest mesa from a couple days ago, RX580
14:07 soreau: and 6.8 kernel
14:08 soreau: if I could reliably reproduce, I would bisect, but it's largely intermittent
14:09 soreau: I updated mesa and reset the gpu manually a few times, then it started happening
14:09 soreau: so idk if resetting caused the hw to go bad somehow or if it's mesa
14:41 karolherbst: soreau: just filed a similar issue, might be related? https://gitlab.freedesktop.org/mesa/mesa/-/issues/12552
14:47 soreau: karolherbst: I suppose I could try the commit before that but since I can't cause it reliably, not sure
14:47 soreau: so I'm reluctant to offer a 'me too' there
14:48 soreau: hopefully it is the same issue and it will be resolved when your issue is fixed
14:49 karolherbst: soreau: yeah.. it was super reliable for me either, but stating that the commit in question is showing it, might be good enough even if you can't claim the previous commit is not affected
14:50 soreau: karolherbst: ok, will investigate further when I have the gumption