14:07 zmike: kusma: did you have a fix for that recursion?
14:57 kusma: No
15:03 kusma: Not today, will debug more tomorrow
15:04 zmike: cool
16:24 eric_engestrom: so... mesa ci runs test jobs as root. we select the driver by setting `VK_DRIVER_FILES`. https://vulkan.lunarg.com/doc/view/latest/linux/LoaderInterfaceArchitecture.html#elevated-privilege-caveats explicitly says that's not allowed and doesn't work. and yet it works right now (in main), but changing an unrelated thing (b2c version) makes it not work anymore (as it should never have). I'm so confused 😩
16:32 mupuf: eric_engestrom: do you have an example job that fails?
16:32 eric_engestrom: not on hand, let me re-create one
16:35 daniels: eric_engestrom: check man secure_getenv - it only changes behaviour if geteuid() != getuid() i.e. you are suid, or if you have something very weird with caps
16:39 eric_engestrom: oh, so if you're actually logged in as root, secure_getenv lets vars through? I forgot/misremembered that
16:39 eric_engestrom: hmm, thanks daniels that might actually be a good direction to investigate
16:41 eric_engestrom: mupuf: https://gitlab.freedesktop.org/eric/mesa/-/jobs/75499928 hasn't started yet but it should show the issue
16:41 eric_engestrom: (if it doesn't and it was a pebcak I'll be pretty embarrassed xD)
16:43 eric_engestrom: hahaha it works, it's not that
16:43 eric_engestrom:goes to find a brown paper bag xD
16:45 eric_engestrom: ok, so I completely mis-bisected/diagnosed the problem, and it's in the ci-tron change I've been working on
16:45 eric_engestrom: I guess that means I should probably stop working for today :P
16:48 pinchartl: sima: airlied: could one of you ack https://lore.kernel.org/dri-devel/2e50a5bd-cf72-4a85-8a99-d8837acfda00@ideasonboard.com/T/#u for merge through the V4L2 tree ? I don't expect any conflicting changes for v6.16 for the DU driver
16:50 mupuf: eric_engestrom: well, a brown bag is a good problem to have!
16:50 mupuf: means you don't need to convince anyone else to change their code ;)
17:15 eric_engestrom: mupuf: hehe indeed, it's always better when the mistake is in my own code :P
17:15 eric_engestrom: (well, unless it's already merged ^^)
19:25 ukleinek: We (= Debian kernel team) got a bug report for the amdgpu driver at https://bugs.debian.org/1102175. Does someone in here has a quick idea how to debug that further?
19:34 ukleinek: and there is also https://bugs.debian.org/1104165
19:34 ukleinek: amdgpu is quite present in our current list of bugs, there are one or two more ...