15:57KungFuJesus: anything I can do to get more visibility to my MR? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20468
16:19KungFuJesus: it fixes a pretty big show stopper on nv30
19:13karolherbst: KungFuJesus: it was new year and you can't really expect people to notice within a day
20:08KungFuJesus: fair enough - it's just odd that it's not listed under the MRs for whatever reason
20:10KungFuJesus: I've never known gitlab put MRs in an unlisted state like that before but maybe you guys have it configured differently than we do at work
20:11karolherbst: what do you mean by "unlisted"?
20:11KungFuJesus: ah, it is listed, it's just using the original MR's subject line
20:12KungFuJesus: was looking for "nv30"
20:12karolherbst: you have to update the MRs title separately
20:13KungFuJesus: gotcha, updated
20:15KungFuJesus: Seeing some other weird regressions of nv30 I hope to try to chase down - though I'm not 100% sure a successful bisect will help with some of them. Launching mythtv launches to a blank screen unless an I have the window manager quickly switch out focus, then _sometimes_ I'll get a properly rendered GUI
20:16KungFuJesus: and even then, if the deinterlacer kicks in (which may or may not be using glsl, I'm not entirely sure these days), I get really weird flickering focus problems
20:18KitsuWhooa: KungFuJesus: out of curiosity, what big endian platform are you using?
20:18KungFuJesus: karolherbst: there are still a couple of other places applying that start offset that I'm not 100% sure should be there
20:18KungFuJesus: KitsuWhooa: Quad powermac G5 - de-waterpumped
20:18KitsuWhooa: I see
20:20karolherbst: KungFuJesus: right... https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10752 could also break stuff randomly, so maybe worth a shot checking if I broke something there as I'm the least confident about the nv30 changes
20:21KungFuJesus: Not 100% sure if this should be mapped at a 0 offset as well: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/gallium/drivers/nouveau/nv30/nv30_push.c#L220
20:23KungFuJesus: I can check if your fencing chances are what cause that weird flicker problem - is there anyway to revert an entire list of hashes with git?
20:23karolherbst: good question.. no idea
20:24karolherbst: but you could just try out the commit before or something.. unless something else is required to observe the broken behavior
20:24KungFuJesus: yeah maybe I can peel them off in reverse order and see if it fixes things
20:58KungFuJesus: SDL2 is also giving me glxbadfbconfig during some attempts to set the mode to 1920x1080x32, perhaps that's related
21:34KungFuJesus: karolherbst: mind adding a Cc: mesa-stable to the comment so it gets picked for stable branches?
21:34KungFuJesus: did you mean commit message or comment on gitlab?
21:36KungFuJesus: ahh, adjacent commit messages answers my question
21:57KungFuJesus: heading out for tonight but I've started a build from 5c2c8652d16b56c0bb5c4e6ab368ca8215e5ca1f, hopefully it's a valuable bisect position, at least
22:11KungFuJesus: managed to test that commit before I left. It does not make the problem go away. Hoping it's an issue in mesa and not in something like xcb. I really don't want to be touching xcb