Hello,

I see the same bug on my server and can add some information.

Kernel: linux-image-6.1.0-55-cloud-amd64 (6.1.192-1)
Same oops: RIP ext4_mb_load_buddy_gfp+0xbc/0x380, process fstrim.

Regression confirmed: 6.1.0-54 (6.1.190-1) works. After booting back
into it, with the same configuration and the same cron jobs, the
problem is gone and the hourly run completes normally.

Trigger on this machine: an hourly cron job runs, in this order:
    1. echo 1 > /proc/sys/vm/drop_caches
    2. fstrim /
The oops happens right after drop_caches, in the fstrim process.
It happened 3 times on 2026-11-16, each time at the first hourly run
after boot. Not tested yet: fstrim alone, without drop_caches.

After the oops all filesystem I/O blocks: processes stay in D state,
systemd-journald hits its start timeout, services stop answering.
Only a hard reset recovers the system.

Environment:
- Debian 12 (bookworm), KVM/QEMU virtual machine (i440FX), 8 vCPU, 23 GiB RAM - Single virtual SCSI disk, 200 GB, discard supported (DISC-GRAN 4K, DISC-MAX 1G)
- Root filesystem: ext4, mounted rw,relatime,discard,errors=remount-ro

Kernel log. The call trace was not captured: the filesystem was
already blocked and journald could not write the rest to disk.

[  383.672829] BUG: kernel NULL pointer dereference, address: 0000000000000000
[  383.674450] #PF: supervisor read access in kernel mode
[  383.675785] #PF: error_code(0x0000) - not-present page
[  383.676986] PGD 0 P4D 0
[  383.677710] Oops: 0000 [#1] PREEMPT SMP NOPTI
[  383.678806] CPU: 5 PID: 4257 Comm: fstrim Not tainted 6.1.0-55-cloud-amd64 #1  Debian 6.1.192-1 [  383.680712] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
[  383.683178] RIP: 0010:ext4_mb_load_buddy_gfp+0xbc/0x380
[  383.684441] Code: 89 44 24 08 f7 3c 24 48 63 d8 89 54 24 0c ba 01 00 00 00 48 89 de e8 33 d8 e3 ff 49 89 c7 48 3d 00 f0 ff ff 0f 87 28 01 00 00 <48> 8b 00 a8 04 0f 84 12 01 00 00 49 8b 07 a8 01 0f 85 07 01 00 00
[  383.688310] RSP: 0018:ffffcf9503daba60 EFLAGS: 00010207
[  383.689562] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 [  383.691126] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff8cf244326500 [  383.692716] RBP: ffff8cf24665a800 R08: ffffcf9503dabadc R09: 0000000000008000

Bug #1154151 has the same subject and may be a duplicate of this one;
#1154101 (system freeze after update to 6.1.0-55) may be related too.

Possibly related upstream: a NULL pointer dereference in
ext4_mb_load_buddy_gfp was reported during the 6.1.167-rc1 stable
review and bisected to the backport of "ext4: convert bd_buddy_page
to bd_buddy_folio": https://lkml.iu.edu/2603.3/01511.html

Regards,
Mauro

Reply via email to