02:42mareko: why does uploading libdrm tarballs doesn't work? Error: the path "/srv/xorg.freedesktop.org/archive/individual/drm" on the web server does not exist.
02:44mareko: release.sh fails
03:00mareko: I need to author the announce message to describe libdrm versions that shouldn't be used because they contain an ABI breakage
08:47emersion: mareko: you need to pull latest modular release script
08:47emersion: hm what do you mean?
08:48emersion: drmDevice breakage has been announced already
08:48emersion: oh, some other breakage
09:16mareko: emersion: I have
09:18mareko: emersion: the path doesn't exit on the server
09:18mareko: *exist
09:21emersion: the path is wrong and seems to indicate that this is missing https://gitlab.freedesktop.org/xorg/util/modular/-/commit/aa2e257ffb7103931cb4cc24e36329172780ae1c
09:22mareko: emersion: I have that commit
09:23mareko: emersion: reverting the commit fixed release.sh for me
09:23mareko: thanks for pointing it out
09:24emersion: maybe your repo is still name "drm" instead of the new "libdrm" name
09:24emersion: named*
09:24emersion: or remote?
09:24mareko: the directory is drm, but I use: ../modular/release.sh [options] .
09:25mareko: the directory shouldn't matter
09:26mareko: if the directory was called flying_pigs, it should still work
09:26emersion: remote URL probably does
09:28mareko: like I said, reverting the commit fixes the upload, so it has no effect on the remote URL
09:28mareko: it just breaks the script
09:29emersion: i mean your Git remote URL is probably git@gitlab.freedesktop.org:mesa/drm.git
09:29emersion: whereas the new one is git@gitlab.freedesktop.org:mesa/libdrm.git
09:30emersion: without the commit I can't upload releases
09:30emersion: btw still missign docs on AMD modifiers
09:32mareko: what docs?
09:39emersion: https://lore.kernel.org/dri-devel/vahBbQHXGpyFcIwzIVTPHRnphiAma3_wNbTftk7O3I6gN4gToIj3zIJrIkO263Ly61q2HArlyB1lvyKM1FFyqkqAdLH195Y41xK8GWL4ZBg=@emersion.fr/
09:59mareko: I'd like to leave that to somebody else
18:54macromorgan: quick question... is there some boilerplate code that polls for EDID data on an HDMI port without a working HPD? I've got some hardware that did things in a... let's say non-standard way and I need to figure out a workaround to support it.
18:56macromorgan: I have a device that uses this HDMI port here: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/rockchip/dw_hdmi_qp-rockchip.c but doesn't follow the guidelines and repurposed the HPD pin as a button input. Was going to add a devicetree parameter of "no-hpd" or some such existing parameter, but didn't want to reinvent the wheel if I didn't have to when it comes to setting up the periodic polling.