This was generated by AI, but I’ve read over it and everything looks accurate. I’ve hit this multiple times in the past few days. It consistently happens when browsing websites that play video (YouTube, X, etc.).
Summary
On 2026-08-26, the graphical session became unusable after the NVIDIA driver
started failing GPU virtual-address/BAR mappings. The driver repeatedly logged
that the GPU was probably locked. Brave crashed, then Hyprland and the
Wayland desktop graphics context failed. The machine was subsequently
restarted without a clean shutdown.
This same driver failure pattern occurred multiple times in the same morning.
Reproduction / timeline
- 10:03:46: first observed
dmaAllocMapping_GM107/NV_ERR_NO_MEMORYerrors
in boot session -2. - 10:04:26 and 10:05:06: Brave processes dumped core with
SIGILL. - 10:21:51: the NVIDIA mapping failures repeated.
- 10:22:00: Brave dumped core with
SIGILL. - 10:22:52: mapping failures repeated, including failed atomic modeset.
- 10:23:05 onward: NVIDIA repeatedly logged
GPU is probably locked!. - 10:23:21: Hyprland crashed in
Render::GL::CHyprOpenGLImpl::begin. - 10:23:25: NVIDIA logged
Flip event timeout on head 0. - 12:02:29: the same mapping failures happened again in the next boot
session. - 12:02:29: Brave process 50916 terminated with
SIGILL. - 12:02:30: atomic modeset failed with error
-11. - 12:02:40 through 12:03:53: repeated NVIDIA GPU watchdog messages.
- 12:03:48: Omarchy Shell reported
QRhiGles2: Context is lostand failed
to initialize its OpenGL backend. - 12:03:53: last journal entry before the unclean restart.
- 12:22:45: system booted again; there was no normal shutdown record.
Relevant kernel log
2026-08-26T12:02:29-05:00 computer kernel: NVRM: dmaAllocMapping_GM107: can't alloc VA space for mapping.
2026-08-26T12:02:29-05:00 computer kernel: NVRM: nvAssertOkFailedNoLog: Assertion failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from pReuseMappingDb->pMapCb(pReuseMappingDb->pGlobalCtx, pAllocCtx, range, cachingFlags, &token, _reusemappingdbAddMappingCallback) @ mapping_reuse.c:273
2026-08-26T12:02:29-05:00 computer kernel: resource: resource sanity check: requesting [mem 0x00000000dfd50000-0x00000000e007ffff], which spans more than 0000:05:00.0 [mem 0xd0000000-0xdfffffff 64bit pref]
2026-08-26T12:02:29-05:00 computer kernel: caller __nv_drm_gem_nvkms_map+0xc3/0x130 [nvidia_drm] mapping multiple BARs
2026-08-26T12:02:29-05:00 computer kernel: [drm:__nv_drm_gem_nvkms_map [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000500] Failed to map NvKmsKapiMemory 0x000000002cc5852d
2026-08-26T12:02:30-05:00 computer kernel: [drm:nv_drm_atomic_apply_modeset_config [nvidia_drm]] *ERROR* [nvidia_drm] [GPU ID 0x00000500] Failed to initialize semaphore for plane fence
2026-08-26T12:02:30-05:00 computer kernel: [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia_drm] [GPU ID 0x00000500] Failed to apply atomic modeset. Error code: -11
2026-08-26T12:02:40-05:00 computer kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked! Notify Timeout Seconds: 7
2026-08-26T12:03:25-05:00 computer kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked! Notify Timeout Seconds: 7
2026-08-26T12:03:45-05:00 computer kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked! Notify Timeout Seconds: 7
2026-08-26T12:03:53-05:00 computer kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked! Notify Timeout Seconds: 7
2026-08-26T12:03:48-05:00 computer omarchy-shell[1258]: WARN: QRhiGles2: Context is lost.
2026-08-26T12:03:48-05:00 computer omarchy-shell[1258]: ERROR: Failed to create graphics context for qs::wayland::layershell::WlrLayershell: Failed to initialize graphics backend for OpenGL.
The same messages occurred at 10:03:46, 10:04:15, 10:21:51, and 10:22:52.
Hardware and software
GPU: NVIDIA GeForce RTX 5060 Ti, GB206 / Blackwell
PCI ID: 10de:2d04, revision a1
Subsystem: Gigabyte Technology Co., Ltd. 1458:418f
PCI address: 0000:05:00.0
Kernel driver: nvidia
VBIOS: 98.06.1F.00.D3
Driver: NVIDIA open kernel module 610.57.04
Kernel module: nvidia-open-dkms 610.57.04-1
Kernel: 7.1.9-arch1-2, x86_64, PREEMPT_DYNAMIC
Distribution: Omarchy 4.0.1 (Arch-based)
Desktop/compositor: Hyprland 0.56.2-1
Session: Wayland
Mesa: 26.2.1-1
Wayland: 1.26.0-1
Brave: 1.93.138-1 (brave-origin-bin)
At the time of the report, after reboot:
VRAM total: 16311 MiB
VRAM used: approximately 712 MiB
BAR1 total: 256 MiB
BAR1 used: 20 MiB
GPU temperature: 30 C
GPU recovery action: None
ECC: not applicable / unavailable
System RAM available: approximately 56 GiB
Swap used: 0 B
The NV_ERR_NO_MEMORY messages appear to concern NVIDIA GPU virtual address
or BAR mapping, not exhaustion of system RAM or VRAM. There were no kernel OOM
kills in the relevant log window.
Related crashes
Brave: SIGILL at 2026-08-26 10:04:26
Brave: SIGILL at 2026-08-26 10:05:06
Brave: SIGILL at 2026-08-26 10:22:00
Hyprland: SIGABRT at 2026-08-26 10:23:22
xdg-desktop-portal-hyprland: SIGSEGV at 2026-08-26 10:23:22
Brave: SIGILL at 2026-08-26 12:03:46
The Hyprland backtrace pointed into its OpenGL renderer:
Render::GL::CHyprOpenGLImpl::begin
Render::GL::CHyprGLRenderer::beginRenderInternal
Render::IHyprRenderer::beginRender
Render::IHyprRenderer::renderMonitor
Monitor::CMonitorFrameScheduler::onFrame