Bug 521653 - kwin_wayland 6.7.0 crashes on startup after amdgpu dcn32_program_compbuf_size timeout and atomic commit EBUSY
Summary: kwin_wayland 6.7.0 crashes on startup after amdgpu dcn32_program_compbuf_size...
Status: RESOLVED DUPLICATE of bug 522618
Alias: None
Product: kwin
Classification: Plasma
Component: wayland-generic (other bugs)
Version First Reported In: 6.7.0
Platform: Fedora RPMs Linux
: NOR crash
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2026-06-18 16:04 UTC by alex zhang
Modified: 2026-07-08 18:28 UTC (History)
3 users (show)

See Also:
Latest Commit:
Version Fixed/Implemented In:
Sentry Crash Report:


Attachments
Logs, coredump info, display configuration, PCI GPU info, and package versions collected from the affected Fedora KDE system. (17.68 KB, application/zip)
2026-06-18 16:04 UTC, alex zhang
Details
Full KWin Wayland crash report from 2026-07-05: gdb thread apply all bt full with debuginfod symbols, relevant journal lines, and system/package/GPU summary. (37.48 KB, application/gzip)
2026-07-05 00:48 UTC, alex zhang
Details

Note You need to log in before you can comment on or make changes to this bug.
Description alex zhang 2026-06-18 16:04:52 UTC
Created attachment 193297 [details]
Logs, coredump info, display configuration, PCI GPU info, and package versions collected from the affected Fedora KDE system.

System:
Fedora Linux 44 KDE Plasma
KWin: 6.7.0-1.fc44
Plasma Workspace: 6.7.0-1.fc44
Qt: qt6-qtbase-6.11.1-1.fc44
Mesa: mesa-dri-drivers-26.0.8-1.fc44
libdrm: 2.4.134-1.fc44
Kernel: 7.0.12-201.fc44.x86_64
Firmware: linux-firmware-20260519-1.fc44, amd-gpu-firmware-20260519-1.fc44

GPU:
AMD Navi 32 [Radeon RX 7700 XT / 7800 XT] [1002:747e]
Subsystem: Sapphire Technology Limited [1da2:475f]
Driver: amdgpu

Display setup:
Two 4K displays:
- HDMI-A-1, 3840x2160@60, scale 2, logical geometry 1920x1080
- DP-1, 3840x2160@60, scale 2, logical geometry 1920x1080
KDE Wayland session.

Problem:
After reboot/login, kwin_wayland consistently crashes once during startup. It then restarts, but fcitx5 exits because the Wayland connection is removed, leaving input method unavailable.

Relevant log:
amdgpu 0000:0a:00.0: [drm] REG_WAIT timeout 1us * 100 tries - dcn32_program_compbuf_size line:147
kwin_wayland[2389]: atomic commit failed: Device or resource busy
kwin_wayland_wrapper[2389]: KCrash: Application 'kwin_wayland' crashing
systemd-coredump: Process 2389 (kwin_wayland) dumped core
fcitx5: waylandmodule.cpp:367] Connection removed

Reproducibility:
Reproduced on two consecutive reboots:
- 2026-06-18 23:15:58, kwin_wayland PID 2420, SIGSEGV
- 2026-06-18 23:52:11, kwin_wayland PID 2389, SIGSEGV

Workaround:
Setting KWIN_DRM_NO_AMS=1 for plasma-kwin_wayland.service appears to be the current workaround being tested.
Comment 1 thesword 2026-06-18 18:05:29 UTC Comment hidden (spam)
Comment 2 TraceyC 2026-06-18 18:11:37 UTC
Searchable backtrace of the crashing process


Message: Process 2420 (kwin_wayland) of user 1000 dumped core.

Stack trace of thread 2420:
#0  0x00007f614fc7bdcc __pthread_kill_implementation (libc.so.6 + 0x74dcc)
#1  0x00007f614fc20f8e raise (libc.so.6 + 0x19f8e)
#2  0x00007f6153d171ca _ZN6KCrash19defaultCrashHandlerEi (libKF6Crash.so.6 + 0x51ca)
#3  0x00007f614fc210b0 __restore_rt (libc.so.6 + 0x1a0b0)
#4  0x00007f61535e4431 _ZN4KWin6DrmGpu15pageFlipHandlerEijjjjPv (libkwin.so.6 + 0x3e4431)
#5  0x00007f61509c6795 drmHandleEvent (libdrm.so.2 + 0xa795)
#6  0x00007f61535e3b0c _ZN4KWin6DrmGpu14dispatchEventsEv (libkwin.so.6 + 0x3e3b0c)
#7  0x00007f6150384eb7 _Z10doActivateILb0EEvP7QObjectiPPv (libQt6Core.so.6 + 0x184eb7)
#8  0x00007f6150393cb3 _ZN15QSocketNotifier9activatedE17QSocketDescriptorNS_4TypeENS_14QPrivateSignalE (libQt6Core.so.6 + 0x193cb3)
#9  0x00007f61503943ef _ZN15QSocketNotifier5eventEP6QEvent (libQt6Core.so.6 + 0x1943ef)
#10 0x00007f615163f92e _ZN19QApplicationPrivate13notify_helperEP7QObjectP6QEvent (libQt6Widgets.so.6 + 0x3f92e)
#11 0x00007f615030f7b8 _ZN16QCoreApplication15notifyInternal2EP7QObjectP6QEvent (libQt6Core.so.6 + 0x10f7b8)
#12 0x00007f615050f315 _ZN27QEventDispatcherUNIXPrivate23activateSocketNotifiersEv (libQt6Core.so.6 + 0x30f315)
#13 0x00007f615050fc83 _ZN20QEventDispatcherUNIX13processEventsE6QFlagsIN10QEventLoop17ProcessEventsFlagEE (libQt6Core.so.6 + 0x30fc83)
#14 0x00007f6151161f22 _ZN23QUnixEventDispatcherQPA13processEventsE6QFlagsIN10QEventLoop17ProcessEventsFlagEE (libQt6Gui.so.6 + 0x761f22)
#15 0x00007f615031d57b _ZN10QEventLoop4execE6QFlagsINS_17ProcessEventsFlagEE (libQt6Core.so.6 + 0x11d57b)
#16 0x00007f6150318992 _ZN16QCoreApplication4execEv (libQt6Core.so.6 + 0x118992)
#17 0x000055b18fae1421 main (kwin_wayland + 0xf421)
#18 0x00007f614fc0a681 __libc_start_call_main (libc.so.6 + 0x3681)
#19 0x00007f614fc0a798 __libc_start_main@@GLIBC_2.34 (libc.so.6 + 0x3798)
#20 0x000055b18fae5955 _start (kwin_wayland + 0x13955)
Comment 3 TraceyC 2026-06-18 18:12:56 UTC
(In reply to thesword from comment #1)
> I also observed kwin 6.7.0 crash but not at startup:
> GPU: RTX 2080 SUPER
> Nvidia drivers: 610.43.02
> 

Your backtrace is different. Please open a new bug report for yours. Thanks.
Comment 5 alex zhang 2026-07-05 00:45:31 UTC
I hit the crash again on the previous boot and collected a full gdb backtrace with debuginfod-enabled debug symbols.

Crash summary:
- Process: kwin_wayland
- Signal: SIGSEGV
- Time: 2026-07-05 08:02:46 CST
- Boot ID: 5a5f796f2c6143928c37c2459b992b18
- Package: kwin-6.7.1-2.fc44.x86_64
- KWin version: 6.7.1
- KDE Frameworks: 6.27.0
- Qt: 6.11.1
- DrKonqi Sentry event ID: 92e91eb8ef904396b2ecec6e85b3fb54

System:
- Fedora Linux 44 KDE Plasma Desktop Edition
- Kernel: 7.0.14-201.fc44.x86_64
- Session: Plasma Wayland
- GPU: AMD Navi 32 / Radeon RX 7700 XT / 7800 XT [1002:747e]
- Driver: amdgpu
- Mesa: 26.1.3-1.fc44
- libdrm: 2.4.134-1.fc44
- amd-gpu-firmware: 20260622-1.fc44

The crash happened right after login/session startup. Relevant journal lines just before the crash:

```
2026-07-05 08:02:43 kernel: amdgpu 0000:0a:00.0: [drm] REG_WAIT timeout 1us * 100 tries - dcn32_program_compbuf_size line:148
2026-07-05 08:02:45 systemd[2349]: Started plasma-kwin_wayland.service - KDE Wayland Compositor.
2026-07-05 08:02:45 kwin_wayland[2560]: No backend specified, automatically choosing drm
2026-07-05 08:02:46 kwin_wayland[2560]: Failed to load cursor theme "Layan-white-cursors"
2026-07-05 08:02:46 kwin_wayland[2560]: atomic commit failed: Device or resource busy
2026-07-05 08:02:46 kwin_wayland[2560]: atomic commit failed: Device or resource busy
2026-07-05 08:02:46 kwin_wayland[2560]: atomic commit failed: Device or resource busy
2026-07-05 08:02:46 kwin_wayland[2560]: atomic commit failed: Device or resource busy
2026-07-05 08:02:46 kwin_wayland[2560]: atomic commit failed: Device or resource busy
2026-07-05 08:02:46 kwin_wayland_wrapper[2560]: KCrash: Application 'kwin_wayland' crashing... crashRecursionCounter = 2
2026-07-05 08:02:46 systemd-coredump[2625]: Process 2560 (kwin_wayland) of user 1000 terminated abnormally with signal 11/SEGV, processing...
2026-07-05 08:02:47 systemd-coredump[2626]: Process 2560 (kwin_wayland) of user 1000 dumped core.
2026-07-05 08:02:47 kernel: amdgpu 0000:0a:00.0: [drm] REG_WAIT timeout 1us * 100 tries - dcn32_program_compbuf_size line:148
```

The full backtrace points at:

```
Thread 1 (Thread 0x7f14afc86480 (LWP 2560)):
#0  __pthread_kill_implementation (...) at pthread_kill.c:44
#1  __pthread_kill_internal (...) at pthread_kill.c:89
#2  __GI_raise (sig=11) at ../sysdeps/posix/raise.c:26
#3  KCrash::defaultCrashHandler (sig=11) at /usr/src/debug/kf6-kcrash-6.27.0-1.fc44.x86_64/src/kcrash.cpp:630
#4  <signal handler called>
#5  std::vector<std::unique_ptr<KWin::DrmCommit>>::size(...) at /usr/include/c++/16/bits/stl_vector.h:1109
#6  std::__detail::__erase_if<...> at /usr/include/c++/16/bits/erase_if.h:55
#7  std::erase_if<...> at /usr/include/c++/16/vector:118
#8  KWin::DrmGpu::pageFlipHandler(fd=31, sequence=509, sec=22, usec=466207, crtc_id=363, user_data=0x7f1434001370)
    at /usr/src/debug/kwin-6.7.1-2.fc44.x86_64/src/backends/drm/drm_gpu.cpp:559
#9  drmHandleEvent(fd=31, evctx=0x7ffd858536b0) at ../xf86drmMode.c:1070
#10 KWin::DrmGpu::dispatchEvents(...) at /usr/src/debug/kwin-6.7.1-2.fc44.x86_64/src/backends/drm/drm_gpu.cpp:594
```

I am attaching:
- `kwin_wayland-2560-gdb-bt-full.txt` - full gdb `thread apply all bt full` output with debuginfod enabled
- `kwin_wayland-2560-relevant-journal.txt` - journal lines around the crash
- `system-summary.txt` - package, kernel, session, and GPU summary

The full gdb run emitted these warnings, but still completed and produced the backtrace:

```
warning: Can't open file /memfd:JSGCHeap:QtQml (deleted) during file-backed mapping note processing
warning: Can't open file /memfd:JSVMStack:QtQml (deleted) during file-backed mapping note processing
warning: .gdb_index address table has a range (0x6e02a0 - 0x6e02a1) that overlaps with an earlier range, ignoring .gdb_index
```
Comment 6 alex zhang 2026-07-05 00:48:05 UTC
Created attachment 194019 [details]
Full KWin Wayland crash report from 2026-07-05: gdb thread apply all bt full with debuginfod symbols, relevant journal lines, and system/package/GPU summary.

Attached an updated crash report bundle from the 2026-07-05 occurrence. It includes a full gdb `thread apply all bt full` backtrace generated with debuginfod symbols, relevant journal lines around the crash, and a system/GPU/package summary.

The backtrace now points to `KWin::DrmGpu::pageFlipHandler()` in `src/backends/drm/drm_gpu.cpp:559`. The journal also shows amdgpu REG_WAIT timeouts and repeated `atomic commit failed: Device or resource busy` messages immediately before the SIGSEGV.
Comment 7 TraceyC 2026-07-08 17:22:04 UTC
Searchable backtrace of the crashing process

Stack trace of thread 2560:
#0  0x00007f14b6c7bdcc __pthread_kill_implementation (libc.so.6 + 0x74dcc)
#1  0x00007f14b6c20f8e raise (libc.so.6 + 0x19f8e)
#2  0x00007f14bac261ca _ZN6KCrash19defaultCrashHandlerEi (libKF6Crash.so.6 + 0x51ca)
#3  0x00007f14b6c210b0 __restore_rt (libc.so.6 + 0x1a0b0)
#4  0x00007f14ba5e4771 _ZN4KWin6DrmGpu15pageFlipHandlerEijjjjPv (libkwin.so.6 + 0x3e4771)
#5  0x00007f14b70c8795 drmHandleEvent (libdrm.so.2 + 0xa795)
#6  0x00007f14ba5e3e4c _ZN4KWin6DrmGpu14dispatchEventsEv (libkwin.so.6 + 0x3e3e4c)
#7  0x00007f14b7384eb7 _Z10doActivateILb0EEvP7QObjectiPPv (libQt6Core.so.6 + 0x184eb7)
#8  0x00007f14b7393cb3 _ZN15QSocketNotifier9activatedE17QSocketDescriptorNS_4TypeENS_14QPrivateSignalE (libQt6Core.so.6 + 0x193cb3)
#9  0x00007f14b73943ef _ZN15QSocketNotifier5eventEP6QEvent (libQt6Core.so.6 + 0x1943ef)
#10 0x00007f14b863f92e _ZN19QApplicationPrivate13notify_helperEP7QObjectP6QEvent (libQt6Widgets.so.6 + 0x3f92e)
#11 0x00007f14b730f7b8 _ZN16QCoreApplication15notifyInternal2EP7QObjectP6QEvent (libQt6Core.so.6 + 0x10f7b8)
#12 0x00007f14b750f315 _ZN27QEventDispatcherUNIXPrivate23activateSocketNotifiersEv (libQt6Core.so.6 + 0x30f315)
#13 0x00007f14b750fc83 _ZN20QEventDispatcherUNIX13processEventsE6QFlagsIN10QEventLoop17ProcessEventsFlagEE (libQt6Core.so.6 + 0x30fc83)
#14 0x00007f14b8161f22 _ZN23QUnixEventDispatcherQPA13processEventsE6QFlagsIN10QEventLoop17ProcessEventsFlagEE (libQt6Gui.so.6 + 0x761f22)
#15 0x00007f14b731d57b _ZN10QEventLoop4execE6QFlagsINS_17ProcessEventsFlagEE (libQt6Core.so.6 + 0x11d57b)
#16 0x00007f14b7318992 _ZN16QCoreApplication4execEv (libQt6Core.so.6 + 0x118992)
#17 0x0000558723402421 main (kwin_wayland + 0xf421)
#18 0x00007f14b6c0a681 __libc_start_call_main (libc.so.6 + 0x3681)
#19 0x00007f14b6c0a798 __libc_start_main@@GLIBC_2.34 (libc.so.6 + 0x3798)
#20 0x0000558723406955 _start (kwin_wayland + 0x13955)
Comment 8 TraceyC 2026-07-08 17:23:44 UTC
This is probably a duplicate of bug 522618, based on the backtrace
Comment 9 Zamundaaa 2026-07-08 18:28:22 UTC
Yeah

*** This bug has been marked as a duplicate of bug 522618 ***