#asahi-gpu · 2026-01-29 09:22
olivialI was *gonna* test it against a version of panvk hacked to have the same bug, but panvk doesn't have ESO and I don't have hw to test against real hk :(#asahi-gpu · 2026-01-29 09:22
olivial(binding pipelines normally, the application will unbind the GS automatically)#asahi-gpu · 2026-01-29 09:21
olivialoof, putting together a CTS test for the xfb bug is kinda annoying cause it's only possible to actually trigger it with ESO#asahi-gpu · 2026-01-29 02:49
olivialhahaha :)#asahi-gpu · 2026-01-29 02:40
olivialand recompute the passthrough GS every time GS/VS are dirty and a passthrough is needed#asahi-gpu · 2026-01-29 02:40
olivialthe way I'm handling this in panvk is to just not have a fast path for passthrough-gs-already-bound#asahi-gpu · 2026-01-29 02:39
olivialyeah#asahi-gpu · 2026-01-29 02:39
olivialcan send a patch in a bit#asahi-gpu · 2026-01-29 02:39
olivialI thought *maybe* the xfb fuzz tests would catch this, but I guess not :)#asahi-gpu · 2026-01-28 23:26
olivial[1]: https://gitlab.freedesktop.org/mesa/mesa/-/blob/2d38da94d4829b339f580e3170744936c3d8a0dd/src/asahi/vulkan/hk_cmd_draw.c#L3225#asahi-gpu · 2026-01-28 23:26
olivialbecause otherwise, I think there is a bug here[1], where the application binds multiple VS with different XFB configurations consectitively#asahi-gpu · 2026-01-28 23:25
olivialfor passthrough GS, is there something that causes hk to unbind the passthrough GS when a new application GS is bound?#asahi-gpu · 2025-10-11 18:11
olivialjannau: oh yeah sorry, allocation was the only thing I was concerned about#asahi-gpu · 2025-10-11 07:01
olivialand I didn't find any synchronization mechanism at the call sites#asahi-gpu · 2025-10-11 07:01
olivialit seems like the code assumed concurrent use is not possible, but I'm not sure that's correct, because this function is called during command buffer recording and concurrent command buffer recording is allowed#asahi-gpu · 2025-10-11 07:00
olivialI don't have any apple hardware to test on#asahi-gpu · 2025-10-11 07:00
olivialno, just noticed it when reading the code#asahi-gpu · 2025-10-11 03:55
olivialwanna be sure I'm not missing anything before submitting a patch :)#asahi-gpu · 2025-10-11 03:54
olivial*data race#asahi-gpu · 2025-10-11 03:53
olivial<https://gitlab.freedesktop.org/mesa/mesa/-/blob/74859c19fb601c12e6dacd4552095a56a7e19309/src/asahi/vulkan/hk_cmd_draw.c#L1004> is a race condition, right?