06:24 hakzsam: alyssa: it really doesn't matter for radv/aco because nir_jump_halt isn't used for anything else, so it's not incorrect for us.
06:28 hakzsam: I'm changing it for you right now
07:15 hakzsam: alyssa: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/42162
10:31 karolherbst: ping on https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/41853 (shader_fma for anv)
11:13 javierm: /q tzimmermann
11:13 javierm: ups, sorry about
12:57 ella-0: Hi. Would anyone be opposed to adding a new label yacc for yacc related changes?
13:47 MrCooper: pendingchaos: bisected piglit spec@arb_shader_image_load_store@shader-mem-barrier regressions with radeonsi on Navi 14 to 3676c3860e7b ("aco: only assume load/store with semantic_atomic is atomic")
13:57 pendingchaos: after a quick look at the source, this test looks racy
13:59 pendingchaos: in vulkan, two invocations can't write->read a location using non-atomic accesses without a control barrier or atomics synchronizing them
13:59 pendingchaos: not sure if that's the same in opengl
14:21 MrCooper: k, thanks for looking into it
17:14 lucaceresoli: There is no DRM/display/multimedia microconference at Linux Plumbers this year? :-(
17:27 lucaceresoli: lumag: mripard: will you be at plumbers? We may find a slot to discuss DRM bridge hotplug even without a MC. That would be valuable and very appreciated.
17:41 anonymix007[m]: Does ANV really need LLVM at runtime? I'm trying to cross-compile it and LLVM doesn't seem to like this
18:17 alyssa: anonymix007[m]: Not at runtime no
18:17 alyssa: compile-time dependency
18:17 alyssa: Although cross-compiling is annoying
18:18 AidenIsik: Are there any known segfaulting issues with kms_swrast_dri.so in Mesa 24.0.7?
18:18 alyssa: there's a trick with -Dmesa-clc=system but idk if i'ts docuemnted
18:18 AidenIsik: Sorry if this is the wrong place for this, but this is the only Mesa-related IRC channel I can find.
18:18 AidenIsik: I'm trying to force software rendering to a simpledrm device
18:19 AidenIsik: And I'm not sure if my problem is with Mesa or somewhere else
18:35 eric_engestrom: AidenIsik: unless that was a typo, 24.0.7 is 2+ years old and your first step should be to update to 26.1.x, as it's unlikely you'll find someone interested in debugging something that old when it might have already been fixed :)
18:38 AidenIsik: eric_engestrom: It is indeed 24.0.7, I wasn't expecting any sort of extended debugging help, just if anyone knows off the top of their head about any issues.
18:39 AidenIsik: I would update but my situation is a bit unique. The OS I am working with is currently on an old version of Yocto and updating is quite a big task.
18:40 AidenIsik: Thanks anyways
18:40 eric_engestrom: I see; I don't remember any, but good luck, because it sounds like you're in an uncomfortable situation :(
18:41 AidenIsik: Very much so aha, I also only have one-way communication with this device at runtime over UART lol so I need to reflash to run a single command xD
18:42 AidenIsik: Debugging is a pain.
18:42 AidenIsik: I think I'll just comb through the relevant source files' commit history and see if I can see any segfault-fixing commits
18:44 eric_engestrom: yeah that sounds like a good strategy!
18:45 eric_engestrom: I'm off now, best of luck!
18:45 AidenIsik: thanks!
18:45 AidenIsik: o/
19:16 anonymix007[m]: alyssa: thanks, -Dmesa-clc=system worked
19:19 alyssa: anonymix007[m]: i'm glad one of us knows how to build my employer's driver ;-)
19:36 anonymix007[m]: Too bad it doesn't build here yet because of "-Werror=padded" all over the src/Intel/compiler ;-)
20:20 alyssa: karolherbst: "NIR validation failed after spirv cfg' on c11_atomics atomic_flag
20:20 alyssa: seems.. bad
20:46 karolherbst: alyssa: mhh, new issue?
20:46 alyssa: xI think so
20:49 lumag: lucaceresoli: I hope to visit
21:51 lucaceresoli: lumag: good, we can meet there hopefully!
23:48 karolherbst: alyssa: "(12-Jun 01:47:33) Testing complete. 0 failures for 74 tests."