00:03gfxstrand[d]: I haven't tried but I can
00:03gfxstrand[d]: Could be an Ada issue more than a laptop one
00:39airlied[d]: Let me know, I can probably do some tests next week if I can down the TODOS I've been ignoring all this week 🙂
02:52mrmx450: is the flag NOUVEAU_USE_ZINK=1 still preferred for people running piglit tests on nouveau when submitting results?
02:53mrmx450: on "piglit run -x glx -x streaming-texture-leak -x max-texture-size -1 --dmesg tests/gpu.py"
06:38asdqueerfromeu[d]: asdqueerfromeu[d]: I hacked around this issue by setting a TARGET variable in my NVK package :ferris:
14:11gfxstrand[d]: I tagged Dylan on the Mesa issue (he probably already saw the meson issue).
14:12gfxstrand[d]: My device is generating commands. It's also executing commands. They are not the right commands....
14:16gfxstrand[d]: Ugh... How do I get GDB to look at mapped memory. I know there's a way
14:18soreau: gfxstrand[d]: https://stackoverflow.com/questions/5691193/gdb-listing-all-mapped-memory-regions-for-a-crashed-process ?
14:23ahuillet[d]: gfxstrand[d]: are you asking about the "x" command?
14:23ahuillet[d]: as for the mappings themselves, /proc/<pid>/maps or info proc mappings work
14:32gfxstrand[d]: The problem is that nouveau doesn't implement the TTM access_memory hook so anything that lives in VRAM isn't accessible via ptrace
14:32gfxstrand[d]: airlied[d]: skeggsb9778[d] ^^
14:33gfxstrand[d]: Shouldn't be hard for someone who knows their way around nouveau. I'm not that someone. I'll just pin to system RAM for now
14:34karolherbst[d]: interesting... that would also allow us to inspect memory if the GPU goes to sleep in the meantime...
17:29gfxstrand[d]: karolherbst[d]: Do you know of any way to tell a specific shader to ignore conditional rendering?
17:29gfxstrand[d]: Not being able to flip on/off with the MME is really annoying
17:30gfxstrand[d]: I can engineer my way around this one but it's gonna be a little insane
17:30gfxstrand[d]: There's a GLOBAL render mode
17:31gfxstrand[d]: Which is somehow not the same as the other one
17:31gfxstrand[d]: Ooh, there's a RENDER_ENABLE_OVERRIDE
18:18karolherbst[d]: gfxstrand[d]: is that what you needed?
18:21gfxstrand[d]: Yeah
18:30karolherbst[d]: cool