07:43 hakzsam: Venemo: 6.11+ is the minimum AFAIK
14:53 agd5f: Venemo, 6.11 or newer
14:54 Venemo: I guess then I was just unlucky with 6.13.5
14:54 Venemo: it seems that 6.13.6 is somewhat better
19:43 hakzsam: agd5f: https://pastebin.com/raw/rS1w683E when running vkcts on gfx1201, built from f54ff4ad3811526540eec148ea41cbc11ee54957 in your branch
19:55 agd5f: hakzsam, should be fixed with this patch: https://patchwork.freedesktop.org/patch/642158/
21:32 tonyk: agd5f: the app name that appears in dmeasg as "Process information: process %s pid %d thread %s pid %d\n" after a GPU hang is guaranteed to be related to the hang?
21:32 tonyk: the context for this question: https://lore.kernel.org/lkml/Z8KjZfLYjH6ehYwy@black.fi.intel.com/
21:41 agd5f: not guaranteed, it's the app on whose behalf the hang was detected. It's the first job that happens to timeout from a hang.
21:47 tonyk: ok, do you think it would be OK to use it as the potential guilty one? as is done in this dmesg message and in devcoredump anyways?
21:49 agd5f: it's usually the guilty one. we more or less do that already for GL robustness, etc.
21:51 tonyk: ok, I will reply that to the mail thread