07:26 wlb: wayland Issue #391 opened by Karl Fleischmann (fleischie) Unable to set event queue https://gitlab.freedesktop.org/wayland/wayland/-/issues/391
08:09 jadahl: emersion: fwiw, been nagging flatpak maintainers for some input this week
08:12 emersion: thanks!
08:31 wlb: wayland Issue #391 closed \o/ (Unable to set event queue https://gitlab.freedesktop.org/wayland/wayland/-/issues/391)
11:57 cu-kai: hello, i have a very specific issue with wayland that my google-fu is failing to solve. i have a dual AMD GPU laptop (AMD Zephyrus G14, model GA402RK) and i am trying to run a SDL2 game (in this case, dhewm3) on the dedicated GPU in wayland mode. i use the following environment variables:
11:57 cu-kai: DRI_PRIME=1 SDL_VIDEODRIVER=wayland
11:57 cu-kai: however, this causes a segfault. the backtrace suggests this is a wayland or EGL related problem as no matter the game, the backtrace always starts with the following: https://p.mort.coffee/yoT.png
11:58 pq: I'm guessing libEGL is calling wl_proxy_get_version(NULL), so not Wayland's fault.
11:59 pq: you'll need Mesa debug files to get a better backtrace
12:00 cu-kai: i see. would you know the best place for me to go and report / ask aboutb this? it's a fairly new crash and this used to work at least to some extent around 2-3 months ago.
12:00 pq: you'll need to refer to your distribution on how to get the debug information installed or downloaded
12:00 emersion: zmike: have you seen this issue? https://gitlab.freedesktop.org/wayland/wayland-protocols/-/issues/146
12:01 pq: cu-kai, once you have function names listed in libEGL_mesa.so, you can file an issue with Mesa in gitlab.freedesktop.org.
12:02 cu-kai: thanks pq
12:04 pq: cu-kai, since you know it used to work, if you are up to building Mesa yourself, you could git-bisect Mesa to find the change that broke it.
12:04 pq: assuming it was a Mesa change
12:04 cu-kai: looks like i'll need to build mesa myself anyway as arch doesn't provide a debug package, but i think trying to bisect or resolve this myself is a little beyond my ability
12:05 cu-kai: anyway: that dll belongs to the mesa package
12:06 pq: cu-kai, does https://wiki.archlinux.org/title/Debuginfod help? I've never used arch.
12:06 cu-kai: oh huh, that could help
12:07 cu-kai: lol, i forgot this was a thing. thanks
12:07 pq: cu-kai, https://wiki.archlinux.org/title/Debugging/Getting_traces looks more to the point.
12:07 cu-kai: yeah, i was on that page before#
12:08 cu-kai: https://p.mort.coffee/QeT.png
12:08 cu-kai: this is a lot better
12:09 pq: yup
12:09 cu-kai: thanks for your help pq
12:09 cu-kai: i think i'll report this on the mesa gitlab
12:10 pq: you're welcome
12:10 pq: make it a text paste, not a picture, please
12:10 cu-kai: ofc, i just did not want to flood this channel
12:16 wlb: weston Merge request !1287 opened by Derek Foreman (derekf) Build view list for all outputs at once https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1287 [libweston API]
14:03 wlb: weston Issue #766 opened by Link Mauve (linkmauve) Segfault when clicking on a fullscreen surface after coming back from another TTY https://gitlab.freedesktop.org/wayland/weston/-/issues/766
15:24 wlb: wayland-protocols Issue #149 opened by Tait Hoyem (TTWNO) Accessibility Input Protocol (`a11y_input_v1`) https://gitlab.freedesktop.org/wayland/wayland-protocols/-/issues/149
16:18 wlb: weston Merge request !1288 opened by Derek Foreman (derekf) tests: Don't wait for frame callbacks when a surface is on no outputs https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1288 [Testing]
16:43 wlb: weston Merge request !1289 opened by Daniel Stone (daniels) Fix surface/view handling for test-surface https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1289 [Testing]
17:47 wlb: weston Merge request !1288 merged \o/ (tests: Don't wait for frame callbacks when a surface is on no outputs https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1288)
17:47 wlb: weston/main: Derek Foreman * tests: Don't wait for frame callbacks when a surface is on no outputs https://gitlab.freedesktop.org/wayland/weston/commit/14220a5f80f8 tests/ event-test.c weston-test-client-helper.c weston-test-client-helper.h
18:09 Max1: Quick question about Weston
18:09 Max1: In weston_pointer_start_drag (https://gitlab.freedesktop.org/wayland/weston/-/blob/main/libweston/data-device.c#L938), it seems ok for icon to be NULL
18:09 Max1: As there's an if checking for it
18:10 Max1: But later on, there's a call to weston_coord_surface(0, 0, icon), which will throw an assertion if icon is NULL
18:11 Max1: So either icon mustn't be NULL, or that call shouldn't happen unconditionally
18:11 daniels: ManMower: ^
18:15 ManMower: ouch. yeah, nice
18:15 ManMower: that's definitely my bug
18:18 ManMower: I'll look into that a bit later, I don't have an off hand fix. may have to make base.offset into a pointer so it can be legitimately NULL
18:19 ManMower: which is... kinda nasty
18:22 Max1: Do we need it to be a weston_coord_surface?
18:22 Max1: I think we only ever use the weston_coord member, can't we just make it that directly?
18:24 ManMower: I guess that's a quick way out of this hole, but I think it really should be a surface coordinate
18:32 Max1: Ah, I missed this: https://gitlab.freedesktop.org/wayland/weston/-/blob/main/libweston/data-device.c#L438-439
18:36 ManMower: does that actually happen when icon is NULL? no icon view would exist?
18:38 daniels: Max1: maybe more to the point - do you have a testcase here?
18:39 Max1: You should be able to trigger the assertion by running Chromium under Weston and trying to drag a tab out of a window
18:40 ManMower: yeah that sounds like it'd do it. thanks
18:40 Max1: I'm able to do so with Chromium 114
18:41 Max1: I've changed it to use just weston_coord instead of weston_coord_surface and it looks like it'll do as a temporary fix for me
18:42 ManMower: I'd have just // out the assert() if I was looking for a quick fix.
18:42 ManMower: I'll try to fix this up properly soon though
18:42 daniels: thanks both!
18:43 Max1: Ah that'll work as well I guess
18:43 Max1: Thanks for looking into a real fix :) happy to drop my temp fix in favour of that once it's available
18:50 ManMower: Max1: I think you had the right answer all along though... < Max1> So either icon mustn't be NULL, or that call shouldn't happen unconditionally
18:50 ManMower: I think the call should just be conditional :)
18:51 daniels: yeah, similar to https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1289/diffs?commit_id=edb9d83c15bbfeeb9f1fcd8cf54ffd618a859e46
18:51 daniels: if you've landed in ->committed() and thus configure() with a !weston_surface_has_content(), don't go remapping it or ... doing anything
18:51 ManMower: :)
18:52 Max1: That sounds sensible
18:54 daniels: start_drag is silent on what should happen if a surface is subsequently unmapped, but is very clear that a drag surface doesn't need to be provided in the first place
18:56 wlb: weston Merge request !1290 opened by Derek Foreman (derekf) data-device: Don't make a weston_coord with no valid space https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1290 [libweston API]
19:00 Max1: That was quick! Thanks, looks very good on my end
19:00 ManMower: \o/
19:02 Max1: I'm not too familiar with how weston handles releases, how long will it approximately take until this fix is in some 12.* branch?
19:03 ManMower:looks at his shoes quietly
19:04 ManMower: in some branch will be really soon, but in an actual release could be a while.
19:06 Max1: "in a branch" is good enough for me, I think
19:07 Max1: Chromium has its own mirror of Weston, as long as the commit is somewhere in there it should be fine
19:07 ManMower: nice
19:18 Max1: Weston has a test suite, right? Does it include DnD tests? If yes, should there be a test that doesn't use a drag surface?
19:19 ManMower: yes, no :(, yes
19:20 ManMower: I don't think the omission of DnD tests is intentional, just that nobody has had time to spend on it
19:23 Max1: If I weren't busy fixing Chromium's tab dragging tests, I'd offer to help out :)
19:24 ManMower: :)
19:25 Max1: But I'll keep this in mind, maybe I get lucky and do find time to help out after I'm done with Chromium's tests
19:26 ManMower: feel free to ping me directly for review
19:51 wlb: weston/main: Derek Foreman * data-device: Don't make a weston_coord with no valid space https://gitlab.freedesktop.org/wayland/weston/commit/815a560dd497 libweston/data-device.c
19:51 wlb: weston Merge request !1290 merged \o/ (data-device: Don't make a weston_coord with no valid space https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1290)
19:53 daniels: Max1: ^ it's in main now, tagged for backport to 12.0.x so that will happen whenever the next stable release is; currently we have no firm date for that due to people being at conferences / on holiday / moving house / etc
22:52 wlb: weston Merge request !1291 opened by Derek Foreman (derekf) Allow overlapping outputs https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/1291 [libweston API], [GL renderer], [Pixman renderer], [VNC Backend], [DRM/KMS backend]