07:14hakzsam: MrCooper: regression should be fixed in main now
08:06MrCooper: hakzsam: confirmed, thanks
13:11dj-death: Lynne: what version of ffmpeg or compile options do I need to use https://github.com/cyanreg/rc_vk_test ?
13:14Lynne: someone tested it on the latest stable version (7.1), so anything at least more recent than a year
13:17dj-death: Thread 1 "rc_vk_test" received signal SIGSEGV, Segmentation fault.
13:17dj-death: #0 ff_vk_init (s=s@entry=0x7fffffffa140, log_parent=log_parent@entry=0x0, device_ref=<optimized out>, frames_ref=frames_ref@entry=0x0) at ../vulkan.c:2914
13:17dj-death: not sure what I need to make this work
13:17dj-death: looks like it's detecting the GPU alright (printing out a bunch of info)
13:18dj-death: maybe I'm linking against the wrong ffmpeg
13:18dj-death: libavutil.so.58 => /lib/x86_64-linux-gnu/libavutil.so.58
13:22Lynne: dj-death: what command line args do you give it?
13:23dj-death: none
13:23Lynne: should work
13:23Lynne: could you try using git master ffmpeg?
13:24Lynne: you don't need to install it or anything, just LD_PRELOAD libavutil.so
13:24dj-death: okay
13:25dj-death: ah...
13:25dj-death: here is the problem
13:25dj-death: the ffmpeg compile didn't produce any .so
13:25dj-death: just .a
13:26dj-death: ah no ./libavutil/libavutil.so.60
13:26dj-death: same result
13:27dj-death: LD_PRELOAD=/home/djdeath/mesa/src/FFmpeg/libavutil/libavutil.so.60 gdb --args ./build/rc_vk_test
13:27dj-death: Thread 1 "rc_vk_test" received signal SIGSEGV, Segmentation fault.
13:31dj-death: actually reading this code I just don't get how it can work
13:32dj-death: memset(s, 0, sizeof(*s));
13:32dj-death: ah no s->device
13:33dj-death: yeah no...
13:33dj-death: s->device_ref->data = NULL
13:33dj-death: I just don't know what to do
13:39Lynne: dj-death: try running with arguments 0 1
13:39Lynne: to enable validation layers
13:40Lynne: the only way s->device_ref->data can be NULL is if the device did not initialize properly
13:40Lynne: (for which we already check)
13:40Lynne: you'll also need to recompile using the new headers, since we just had an ABI bump in ffmpeg
13:41Lynne: ABI issues seem the most likely
14:26alyssa: eric_engestrom: sorry for how much i'm backporting >.>
14:26alyssa: how do i write so many bugs? who knows!
14:34zmike: pq: I was in the area and came out with https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34610
14:34zmike: pls test
15:44eric_engestrom: alyssa: welcome to being upstream :P
15:44eric_engestrom: can't just tag a new commit on main and ship it to users anymore, you gotta backport to regular releases now :P
15:45alyssa: eric_engestrom: It's a nice change :)
15:46alyssa: we'll probably keep tagging downstream for a while to transition things over, not sure what the long term plan is
15:46alyssa: but having upstream releases be competent matters for flatpak etc anyway :)
15:46eric_engestrom: I assume as much, at least until 25.1 has shipped to fedora
15:47alyssa: yeah, among other things
15:47alyssa: we also need the virglrenderer support landed
15:47eric_engestrom: (/me has no clue how long that would be actually, I don't know how/when they update)
15:47alyssa: current plan is shipping next week 25.1rc2 with a trivial patch to enable virglrenderer
17:37dj-death: Lynne: no luck, anv_CreateDevice was called and return VK_SUCCESS
20:07Lynne: dj-death: did you check with git master plus using the git master's header?