00:41karolherbst: I should come back to me "llvm is broken" assumption...
03:07karolherbst: dcbaker: what do you say to this suggestion? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22484#note_1869831
12:17kode54: are there known GPU reset bugs with AMD Ryzen motherboards?
12:17kode54: or at least B450 boards
12:21kode54: https://gitlab.freedesktop.org/drm/intel/-/issues/8365
12:21kode54: also, is it normal that those 01.org links return 403 errors?
12:35kisak: hmm ... https://web.archive.org/web/20220916035804/https://01.org/linuxgraphics/documentation/development/how-dump-video-bios
12:42kisak: kode54: there's a contact form on that site, and https://github.com/01org also points to webadmin at linux.intel.com
13:14zamundaaa[m]: ChaosPrincess: for rotated screens afaik you have no choice but to let userspace handle it
13:16zamundaaa[m]: The "panel orientation" property not doing anything is most likely a result of whatever drm master you're using for testing ignoring the property
14:36karolherbst: mhhh "ERROR: Scanline 0 did not verify for image size 3,2,5 pitch 48 (extra 0 bytes)\nFailed at column: 0 *{0xbe2b3eff, 0xb9e5f651, 0x8fc4c9b2, 0x3e395020} vs. {0xbe2b3eff, 0xb9e5f651, 0x8fc4c9b2, 0x3e395020}"
14:39jannau: zamundaaa[m]: related question. does kde have a concept like auxiliary display? just as I tried ChaosPrincess' driver it became obvious that moving the panel to the touchbar display might be neat: see https://social.treehouse.systems/@janne/110203289175541411
14:40jannau: it's right now rather broken since the touchbar is handled as second display with a resolution of 2008x60
14:45zamundaaa[m]: I don't think we have anything like that, no
14:47jannau: I don't think it's worth investing time into that based on my guess of at most a low 3 digit number of users
15:34Company: is there a way to disable all extensions with Mesa?
15:34Company: I'm using epoxy and epoxy smartly loads all the symbols without me doing anything, but I want to make sure I check for all the necessary extensions at startup
15:49soreau: Company: have you looked at https://docs.mesa3d.org/envvars.html#envvar-MESA_EXTENSION_OVERRIDE ?
15:55Company: soreau: I did - that seems to require listing them all individually
16:00soreau: Company: I imagine you could put the extension list in a text editor or file and use a text replacement tool to do what you need, then feed it back through the env var?
16:01Company: yeah, though that's not safe against future extensions
16:01Company: and I was thinking about adding that to CI
16:03Company: I had hoped for something like MESA_EXTENSION_OVERRIDE=-* +EXT_i_check_for +EXT_i_also_check
16:03soreau: if you got the extension list from the driver to modify, wouldn't it report all/new extensions?
16:03Company: yeah, if I run that every time
16:05emersion: maybe a new allow-list env var?
16:05emersion: could be introduced?
16:07soreau: looks like the existing code is here https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/mesa/main/extensions.c#L162
17:47edman007: Hey, I'm trying to debug a fairly massive memory leak in I think mesa, maybe opencv, and it's related to opencl, I think possibly related to not having functining opencl drivers, anyways, can someone take a look at my test program and valgrind backtrace, any idea what is going on here: https://gist.github.com/edman007/3e9e943251bd2a11fea421d27db0eb17
17:47edman007: basically seems like cv::UMat.release() ends up allocating and leaking the memory (is it getting dumped into an opencl queue that never runs?)
22:32kode54: kisak: thanks
22:32kode54: but I'll have to use the email address from the github repository
22:32kode54: the contact form on 01.org says unable to send email
22:32kode54: maybe the site is somewhat broken right now