Hello,

One more confirmation on amd64 (cloud flavour), with the full call trace,
since the original report only had the oops tail.

- Debian 12 on Google Compute Engine (e2-highmem-2, 2 vCPU, 16 GB RAM),
linux-image-6.1.0-55-cloud-amd64 6.1.192-1, not tainted.
- The affected filesystem is ext4 directly on a whole-disk data volume (not
the root filesystem), used by a database running in Docker.
- The oops came about 19 minutes after booting 6.1.192-1, while the
database's diagnostics thread (ftdc) renamed a file on that filesystem.
- No memory pressure: about 8 GB of 16 GB was available shortly before and
after the oops, swap unused.
- After the oops the filesystem hung; systemd-journald and
systemd-timesyncd hit their watchdog timeouts and could not be killed, and
sudo hung. A hard reset was needed.
- A second oops at the same RIP followed about a minute later in a kworker
(kworker/u4:0, "exited with irqs disabled").
- Back on 6.1.0-54-cloud-amd64 (6.1.190-1), held, and stable since.

BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 0 PID: 4422 Comm: ftdc Not tainted 6.1.0-55-cloud-amd64 #1  Debian
6.1.192-1
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 03/18/2026
RIP: 0010:ext4_mb_load_buddy_gfp+0xbc/0x380
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
RSP: 0018:ffffd07a01e27398 EFLAGS: 00010207
RAX: 0000000000000000 RBX: 00000000000000ec RCX: 0000000000000001
RDX: 000000000000002c RSI: ffff8ee3441716d0 RDI: ffff8ee3f98ae500
Call Trace:
 <TASK>
 ext4_mb_regular_allocator+0x29e/0xce0
 ext4_mb_new_blocks+0x820/0x1240
 ext4_ext_map_blocks+0x3c7/0x1990
 ext4_map_blocks+0x1b6/0x5f0
 ext4_writepages+0x88a/0x1190
 do_writepages+0xc0/0x1c0
 filemap_fdatawrite_wbc+0x5f/0x80
 __filemap_fdatawrite_range+0x58/0x80
 ext4_rename+0x77e/0xbd0
 vfs_rename+0x814/0x10d0
 do_renameat2+0x57f/0x5f0
 __x64_sys_rename+0x3f/0x50
 do_syscall_64+0x5d/0x90
 entry_SYSCALL_64_after_hwframe+0x6e/0xd8

The trace matches the analysis in message #10 (page-cache miss in the
fast-path folio lookup returning NULL, tested with IS_ERR()); in this case
the miss happened without memory pressure, on a filesystem whose page cache
was still cold after boot.

We can test the 6.1.193 package on a non-production copy when it is
available.

Regards,
Scitus Solutions

Reply via email to