04:24 mareko: why isn't the sanity job starting? https://gitlab.freedesktop.org/mesa/mesa/-/pipelines/1428803
09:08 emersion: any thoughts on this? https://gitlab.freedesktop.org/mesa/drm/-/issues/74
09:13 MrCooper: would there be an alias for the old name?
09:13 K900: Gitlab always creates redirects if you rename the repo
09:13 K900: SO yes
09:13 K900: * So
09:14 MrCooper: sounds good to me then
09:15 MrCooper: (FWIW, the "drm" name is for hysterical raisins, once upon a time this repo also hosted the kernel drivers, as can still be seen in Git history)
09:21 emersion: yeah, there will be redirects
11:00 rburton: should shared-llvm=disabled work? it appears to forget to actually link to libllvm.a for me
11:01 rburton: eg mesa-25.0.5/src/amd/vulkan/radv_nir_to_llvm.c:380:(.text.ac_diagnostic_handler+0x17): undefined reference to `LLVMGetDiagInfoSeverity'
11:42 psykose: i think llvm doesn't have a mere libLLVM.a anymore so it needs the explicit .a list of the smaller libs like the clang_modules entries
14:04 tonyk: should this patch go through drm or drm-misc tree? https://lore.kernel.org/lkml/20250509142627.639419-1-andrealmeid@igalia.com/
14:20 dianders: "No space left on device" on ssh.gitlab.freedesktop.org?
14:20 dianders: https://www.irccloud.com/pastebin/SyHU0ot8/
14:27 MrCooper: dianders: from #freedesktop: <bentiss> FWIW, gitaly-2 is running out of space. That means pushes to drm repos are not working properly :(
14:37 dianders: MrCooper: Ah, thanks. Somehow I wasn't on that channel. :-P
17:38 karolherbst: did somebody work on phi node related optimizations? I have a pattern where the phi sources are a constant and u2u32, the result u2u16, and I wonder if somebody already worked on an optimization like that to reduce the phi node from 32 to 16 bit and get rid of the u2u16/u2u32
17:39 karolherbst: or is that nir_opt_phi_precision?
17:39 karolherbst: ohh yeah.. looks like that's what I'm looking for
17:41 karolherbst: doesn't work tho...
17:42 karolherbst: ohh.. it just misses things...
17:43 karolherbst: it doesn't handle u2u32/u2u16...
17:52 alyssa: karolherbst: yeah probably just need to fix that pass
17:52 karolherbst: already got a patch
17:52 karolherbst: but yeah.. could be a bit improved :D
17:52 alyssa: r-b
17:52 karolherbst: just needed to add nir_op_u2u16 and nir_op_u2u32 in those switches