| View previous topic :: View next topic |
| Author |
Message |
kgroombr
n00b
Joined: 23 Jul 2007
Posts: 26
|
 Posted: Thu Jul 11, 2024 4:01 pm Post subject: >=mesa-24.1.2 breaks wayland |
 |
|
I have been fighting this on my system since mesa-24.1.2 was released. I noticed after logging in after upgrading to mesa-24.1.2 that my desktop effects were super choppy. mesa-24.1.3 didn't resolve it. Been trying different things, but downgrading to mesa-24.1.1 fixes it. I have been using plasma with wayland since late last year and it has always worked well.
I added QT_LOGGING_RULES="kwin_wayland_*.debug=true" to /etc/environment
I get these log entries that don''t output when using mesa-24.1.1:
| Code: |
XXXXXXX kwin_wayland_wrapper[866]: kwin_wayland_drm: Atomic test failed! Permission denied
XXXXXXX kwin_wayland_wrapper[866]: kwin_wayland_drm: Presentation failed! Permission denied
XXXXXXX kwin_wayland_wrapper[866]: kwin_wayland_drm: Atomic test failed! Permission denied
|
Googling the above didn't help.
I have a laptop with wayland and it is working fine. Major difference is that it has an nvidia card and this affected system has an AMD Radeon 6800XT card.
I created a UID0/GID0 test user and it works correctly as that user so it does appear to be a permissions issue, but the user on the affected system belongs to the same groups as the user on the laptop that has no issues: uid=1000(XXXXXX) gid=1000(XXXXXX) groups=1000(XXXXXX),10(wheel),18(audio),19(cdrom),27(video),47(pcscd),85(usb),272(plugdev),394(openvpn),460(crontab),509(pipewire),512(libreoffice)
I don't want to just paste a bunch of logs and such here as I think most of it is irrelevant.
Hope someone can lead me in the right direction.
Thanks,
Ken |
|
| Back to top |
|
 |
kgroombr
n00b
Joined: 23 Jul 2007
Posts: 26
|
| Posted: Sat Jul 13, 2024 12:10 am Post subject: |
|
|
In the process of trying to fix this, I rebooted and it was working correctly with mesa-24.1.3. Checked the logs and all looked good. I changed nothing and rebooting again and then it started failing again. This is driving me nuts. I did find a couple entries I missed before that are likely the root issue. I googled and tried a few things but nothing panned out.
Here are the entries which I believe may help determine the issue:
| Code: |
kwin_wayland_wrapper[857]: _amdgpu_device_initialize: amdgpu_query_info(ACCEL_WORKING) failed (-13)
kwin_wayland_wrapper[857]: amdgpu: amdgpu_device_initialize failed.
|
Here is a more complete log:
| Code: |
kernel: amdgpu 0000:03:00.0: [drm] fb0: amdgpudrmfb frame buffer device
systemd[1]: Starting Load Kernel Module drm...
systemd[1]: modprobe@drm.service: Deactivated successfully.
systemd[1]: Finished Load Kernel Module drm.
dbus-daemon[610]: [system] Activating via systemd: service name='org.freedesktop.locale1' unit='dbus-org.freedesktop.locale1.service' requested by ':1.23' (uid=1000 pid=828 comm="/usr/bin/startplasma-wayland")
dbus-daemon[823]: [session uid=1000 pid=823 pidfd=5] Activating service name='org.kde.KSplash' requested by ':1.5' (uid=1000 pid=828 comm="/usr/bin/startplasma-wayland")
kwin_wayland_wrapper[857]: No backend specified, automatically choosing drm
kwin_wayland_wrapper[857]: _amdgpu_device_initialize: amdgpu_query_info(ACCEL_WORKING) failed (-13)
kwin_wayland_wrapper[857]: amdgpu: amdgpu_device_initialize failed.
kwin_wayland_wrapper[857]: kwin_wayland_drm: drmModeAddFB2WithModifiers is supported on GPU "/dev/dri/card0"
kwin_wayland_wrapper[857]: kwin_wayland_drm: Using Atomic Mode Setting on gpu "/dev/dri/card0"
kwin_wayland_wrapper[857]: kwin_wayland_drm: Number of planes on GPU "/dev/dri/card0" : 14
kwin_wayland_wrapper[857]: kwin_wayland_drm: adding GPU /dev/dri/card0
kwin_wayland_wrapper[857]: kwin_xkbcommon: XKB: couldn't find a Compose file for locale "C.UTF8" (mapped to "C.UTF8")
kwin_wayland_wrapper[857]: kwin_wayland_drm: New output on GPU /dev/dri/card0: LG ULTRAWIDE/0x0004BE01
kwin_wayland_wrapper[1062]: The XKEYBOARD keymap compiler (xkbcomp) reports:
kwin_wayland_wrapper[1062]: > Warning: Unsupported maximum keycode 708, clipping.
kwin_wayland_wrapper[1062]: > X11 cannot support keycodes above 255.
kwin_wayland_wrapper[1062]: > Warning: Could not resolve keysym XF86KbdInputAssistPrevgrou
kwin_wayland_wrapper[1062]: > Warning: Could not resolve keysym XF86KbdInputAssistNextgrou
kwin_wayland_wrapper[1062]: Errors from xkbcomp are not fatal to the X server
kwin_wayland_wrapper[1060]: Initializing "/usr/lib64/qt6/plugins/plasma/kcms/systemsettings/kcm_fonts.so"
kwin_wayland_wrapper[1060]: Initializing "/usr/lib64/qt6/plugins/plasma/kcms/systemsettings/kcm_style.so"
org_kde_powerdevil[1196]: kscreen.kwayland: Connection to Wayland server timed out.
systemd[806]: Reached target plasma-workspace-wayland.target.
systemd[806]: Starting Xwayland Video Bridge...
systemd[806]: Started Xwayland Video Bridge.
kwin_wayland_wrapper[857]: kf.windowsystem: static bool KX11Extras::mapViewport() may only be used on X11
org_kde_powerdevil[1196]: kscreen.kwayland: Connection to Wayland server timed out.
|
|
|
| Back to top |
|
|
kgroombr
n00b
Joined: 23 Jul 2007
Posts: 26
|
| Posted: Fri Jul 26, 2024 5:53 pm Post subject: |
|
|
I used git bisect to track down the commit that started my issue.
| Code: |
commit 92a43a3f473436225a4b9ef5e968d817731f91df
Author: Lucas Fryzek <lfryzek@igalia.com>
Date: Mon Jun 3 11:39:27 2024 +0100
llvmpipe: query winsys support for dmabuf mapping
Fixes #11257 by ensuring winsys mapping functions is only called
if its supported by the winsys, which should prevent llvmpipe from
crashing with kmswast.
If the winsys is kms_swrast then this method will be null, but on
drisw it will be available.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
(cherry picked from commit db38a4913ec6fde264facf95077e241c64573b27)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29721>
diff --git a/.pick_status.json b/.pick_status.json
index e3a2a88f472..f9034c72c48 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -5334,7 +5334,7 @@
"description": "llvmpipe: query winsys support for dmabuf mapping",
"nominated": false,
"nomination_type": 3,
- "resolution": 4,
+ "resolution": 1,
"main_sha": null,
"because_sha": null,
"notes": null
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index 9bad41a9836..6e86a6da06c 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -664,7 +664,13 @@ llvmpipe_resource_from_handle(struct pipe_screen *_screen,
#ifdef HAVE_LINUX_UDMABUF_H
struct llvmpipe_memory_fd_alloc *alloc;
uint64_t size;
- if(_screen->import_memory_fd(_screen, whandle->handle, (struct pipe_memory_allocation**)&alloc, &size, true)) {
+ /* Not all winsys implement displaytarget_create_mapped so we need to check
+ * that is available (not null).
+ */
+ if (winsys->displaytarget_create_mapped &&
+ _screen->import_memory_fd(_screen, whandle->handle,
+ (struct pipe_memory_allocation**)&alloc,
+ &size, true)) {
data = alloc->data;
lpr->dt = winsys->displaytarget_create_mapped(winsys, template->bind,
template->format, template->width0, template->height0,
|
I removed this commit from mesa-2.1.4 and wayland functions perfectly. I am a bit befuddled on why this commit would cause this issue. Any guesses? |
|
| Back to top |
|
|
sam_
Developer
Joined: 14 Aug 2020
Posts: 1949
|
| Posted: Sun Aug 04, 2024 6:59 pm Post subject: |
|
|
| I'd just file a bug upstream with Mesa, letting them know you bisected, share any logs you have, and CC the commit author. |
|
| Back to top |
|
|
kgroombr
n00b
Joined: 23 Jul 2007
Posts: 26
|
| Posted: Tue Aug 13, 2024 6:34 pm Post subject: |
|
|
| sam_ wrote: | | I'd just file a bug upstream with Mesa, letting them know you bisected, share any logs you have, and CC the commit author. |
Thank you, already done. The person I am working with is just as perplexed as I am. Just find it hard to believe I am the only one affected by this. |
|
| Back to top |
|
|
Obi1
n00b
Joined: 09 Dec 2013
Posts: 7
|
| Posted: Thu Aug 15, 2024 11:33 am Post subject: |
|
|
| I had a similar problem with mesa 24.2 and my AMD 7900 GRE card on Sway. Recompiling mesa with both radeonsi and radeon VIDEO_CARDS USE flags fixed it for me, with older mesa versions I only had radeonsi set. |
|
| Back to top |
|
|
kgroombr
n00b
Joined: 23 Jul 2007
Posts: 26
|
| Posted: Mon Aug 19, 2024 7:46 pm Post subject: |
|
|
| Obi1 wrote: | | I had a similar problem with mesa 24.2 and my AMD 7900 GRE card on Sway. Recompiling mesa with both radeonsi and radeon VIDEO_CARDS USE flags fixed it for me, with older mesa versions I only had radeonsi set. |
Thanks for taking the time to try to help me on this.
I tinkered with the VIDEO_CARDS settings to no avail. My current settings are:
VIDEO_CARDS="amdgpu radeon radeonsi"
I tried removing "radeon", but that didn't change the result.
Ken |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Desktop Environments
