Public bug reported:
# Summary
Mini PC (Ryzen 7 6800H / Radeon 680M) locks up with a kernel NULL pointer
dereference in amdgpu_hmm_invalidate_gfx. It started while Jellyfin was
transcoding on the GPU, but it keeps happening with CPU-only transcoding too.
After the crash the machine answers ping but not SSH, and I have to force it
off. I couldn't find an existing Ubuntu bug for this.
# Details
My machine keeps locking up with a NULL pointer dereference in
amdgpu_hmm_invalidate_gfx. It's always kcompactd0 that hits it. After it
happens the box still answers ping, but SSH is dead and I have to force it off.
kdump caught every crash, and the traces are identical each time:
RIP: amdgpu_hmm_invalidate_gfx+0x42/0xe0 [amdgpu]
__mmu_notifier_invalidate_range_start
try_to_migrate_one
rmap_walk
compact_zone
It has happened five times, all on 7.0.0-31-generic:
2026-09-14 12:42 and 13:16 - Jellyfin transcoding with VAAPI
2026-09-19 15:12 - after I turned Jellyfin hardware acceleration off
2026-09-20 17:31 - after I switched to multi-user.target (no desktop)
2026-09-22 20:02 - same, still headless
So CPU-only transcoding and running headless didn't stop it. During all
of these the GPU was still passed into the Jellyfin container
(/dev/dri/renderD128). I've removed that since the last crash, so
nothing should be using the GPU now. I'll update here if it crashes
again.
It crashes every 1 to 5 days. In one of those boots the machine also ran
out of memory about a day in (OOM kill), which I guess is what gets
kcompactd going.
Hardware: GEEKOM A6, Ryzen 7 6800H, Radeon 680M iGPU, BIOS 2.41.
As far as I can tell this is the regression from 91250893cbaa
("drm/amdgpu: fix waiting for all submissions for userptrs"), and it's
already fixed upstream and in stable:
Mainline fix (52f650963d88):
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=52f650963d8825e97a0ccdd2b616f8a01d9d3d38
Stable backport (be354ea7261c, applied 2026-08-03):
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=be354ea7261c2fa43d2c78fc1192ddae08bdbffc
Upstream bug reports:
https://gitlab.freedesktop.org/drm/amd/-/work_items/5399 and
https://bugzilla.kernel.org/show_bug.cgi?id=221679
I just upgraded to 7.0.0-34-generic and its changelog doesn't mention
this fix, so I assume it's still affected. Could the stable fix be
picked up for the 26.04 kernel?
I've attached dmesg from the most recent crash (2026-09-22). I also have dmesg
and vmcores for the two crashes before it and can add them if they're useful.
Thank you in advance.
ProblemType: Bug
DistroRelease: Ubuntu 26.04
Package: linux-image-7.0.0-34-generic 7.0.0-34.34
ProcVersionSignature: Ubuntu 7.0.0-34.34-generic 7.0.14
Uname: Linux 7.0.0-34-generic x86_64
ApportVersion: 2.34.1-0ubuntu0.1
Architecture: amd64
AudioDevicesInUse: Error: command ['fuser', '-v', '/dev/snd/controlC1',
'/dev/snd/hwC1D0', '/dev/snd/pcmC1D0c', '/dev/snd/pcmC1D0p',
'/dev/snd/by-path', '/dev/snd/controlC0', '/dev/snd/hwC0D0',
'/dev/snd/pcmC0D9p', '/dev/snd/pcmC0D8p', '/dev/snd/pcmC0D7p',
'/dev/snd/pcmC0D3p', '/dev/snd/seq', '/dev/snd/timer'] failed with exit code 1:
CasperMD5CheckResult: pass
Date: Wed Sep 23 14:00:21 2026
InstallationDate: Installed on 2026-07-22 (63 days ago)
InstallationMedia: Ubuntu 26.04 "Resolute Raccoon" - Release amd64 (20260423.1)
IwDevWlp2s0Link: Not connected.
MachineType: GEEKOM A6
ProcEnviron:
LANG=en_US.UTF-8
PATH=(custom, no user)
SHELL=/bin/bash
TERM=xterm-256color
XDG_RUNTIME_DIR=<set>
ProcFB:
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-7.0.0-34-generic
root=UUID=08f2db2e-de1c-46e8-8e06-872ff108f74a ro quiet splash
crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M
SourcePackage: linux
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 12/09/2025
dmi.bios.release: 2.41
dmi.bios.vendor: American Megatrends International, LLC.
dmi.bios.version: 2.41
dmi.board.asset.tag: Default string
dmi.board.name: A6
dmi.board.vendor: Default string
dmi.board.version: Default string
dmi.chassis.asset.tag: Default string
dmi.chassis.type: 3
dmi.chassis.vendor: Default string
dmi.chassis.version: Default string
dmi.ec.firmware.release: 0.36
dmi.modalias:
dmi:bvnAmericanMegatrendsInternational,LLC.:bvr2.41:bd12/09/2025:br2.41:efr0.36:svnGEEKOM:pnA6:pvrDefaultstring:rvnDefaultstring:rnA6:rvrDefaultstring:cvnDefaultstring:ct3:cvrDefaultstring:skuGEEKOMA6:pfaASeries:
dmi.product.family: A Series
dmi.product.name: A6
dmi.product.sku: GEEKOM A6
dmi.product.version: Default string
dmi.sys.vendor: GEEKOM
** Affects: linux (Ubuntu)
Importance: Undecided
Status: New
** Tags: amd64 amdgpu resolute
** Attachment added: "dmesg-202609222002.txt"
https://bugs.launchpad.net/bugs/2168045/+attachment/6001934/+files/dmesg-202609222002.txt
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2168045
Title:
amdgpu_hmm_invalidate_gfx NULL pointer dereference from kcompactd on
26.04 (7.0.0-31)
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2168045/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
