Package: src:linux
Version: 6.1.192-1
Severity: important

Dear Maintainer,

After unattended-upgrades moved a server from linux-image-6.1.0-54-amd64 
(6.1.190-1) to linux-image-6.1.0-55-amd64 (6.1.192-1), the system crashes 2-4 
minutes after every boot (5 out of 5 boots) with a NULL pointer dereference in 
ext4_mb_load_buddy_gfp(), called from ext4_free_blocks() during unlinkat(). 
Booting 6.1.0-54-amd64 is stable (ran 14 days before the upgrade, stable again 
since rollback).

Setup:
- Debian 12 (bookworm), amd64, OpenStack Nova / KVM guest, 4 vCPU, 8 GB RAM
- Root filesystem: ext4, 158 GB, on virtio-scsi
- Docker (overlay2) with /var/lib/docker on that ext4 root filesystem, ~34 
containers; some restart repeatedly right after boot, so there is heavy file 
creation/deletion in the first minutes

A second server on the same kernel with Docker data on XFS (ext4 only for a 
mostly idle 20 GB root) has not crashed in 12 hours.

Without panic_on_oops the system half-hangs (processes stuck in D state, shell 
hangs on file reads) until a hard reset, and only the first BUG: line reaches 
the journal. Full oops captured with kdump (kernel.panic_on_oops=1):

[  168.267572] BUG: kernel NULL pointer dereference, address: 0000000000000000
[  168.269937] #PF: supervisor read access in kernel mode
[  168.271618] #PF: error_code(0x0000) - not-present page
[  168.273015] PGD 105cfe067 P4D 105cfe067 PUD 105cfd067 PMD 0 
[  168.274825] Oops: 0000 [#1] PREEMPT SMP NOPTI
[  168.276112] CPU: 1 PID: 2164 Comm: containerd-shim Kdump: loaded Not tainted 
6.1.0-55-amd64 #1  Debian 6.1.192-1
[  168.278957] Hardware name: OpenStack Foundation OpenStack Nova, BIOS 
1.14.0-2 04/01/2014
[  168.280952] RIP: 0010:ext4_mb_load_buddy_gfp+0xbc/0x380 [ext4]
[  168.282533] Code: 89 44 24 08 f7 3c 24 48 63 d8 89 54 24 0c ba 01 00 00 00 
48 89 de e8 d3 8e ba d9 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
[  168.286830] RSP: 0018:ffffd34a04b57b28 EFLAGS: 00010207
[  168.288110] RAX: 0000000000000000 RBX: 0000000000000026 RCX: 0000000000000003
[  168.289870] RDX: 0000000000000026 RSI: ffff8c6dd183cdb0 RDI: ffff8c6dc77c3300
[  168.291725] RBP: ffff8c6dd24ab800 R08: ffff8c6dc0501f70 R09: 0000000000000000
[  168.293534] R10: ffffd34a04b57bc0 R11: 0000000000000000 R12: ffffd34a04b57bc0
[  168.295294] R13: 0000000000008c40 R14: ffff8c6dd218c258 R15: 0000000000000000
[  168.296980] FS:  0000280882380090(0000) GS:ffff8c6ef7c80000(0000) 
knlGS:0000000000000000
[  168.298900] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  168.300409] CR2: 0000000000000000 CR3: 000000010283c000 CR4: 0000000000350ee0
[  168.302424] Call Trace:
[  168.303156]  <TASK>
[  168.303852]  ext4_free_blocks+0x410/0xbb0 [ext4]
[  168.305008]  ext4_ext_remove_space+0x7ff/0x15d0 [ext4]
[  168.306383]  ext4_ext_truncate+0xb4/0xd0 [ext4]
[  168.307526]  ext4_truncate+0x314/0x490 [ext4]
[  168.308719]  ? __ext4_journal_start_sb+0x105/0x120 [ext4]
[  168.310033]  ext4_evict_inode+0x2f8/0x710 [ext4]
[  168.311348]  evict+0x112/0x2c0
[  168.312318]  ? iput+0x6b/0x220
[  168.313226]  ? _atomic_dec_and_lock+0x35/0x50
[  168.314428]  do_unlinkat+0x173/0x320
[  168.315397]  __x64_sys_unlinkat+0x33/0x60
[  168.316504]  do_syscall_64+0x5d/0xb0
[  168.317485]  ? do_syscall_64+0x69/0xb0
[  168.318663]  ? do_syscall_64+0x69/0xb0
[  168.319656]  ? __flush_smp_call_function_queue+0x11a/0x1a0
[  168.321012]  ? exit_to_user_mode_prepare+0x44/0x240
[  168.322388]  entry_SYSCALL_64_after_hwframe+0x6e/0xd8
[  168.323707] RIP: 0033:0x410dae
[  168.329164] RSP: 002b:00002808823bb368 EFLAGS: 00000202 ORIG_RAX: 
0000000000000107
[  168.339536]  </TASK>
[  168.340175] Modules linked in: nf_conntrack_netlink xt_nat veth 
xt_MASQUERADE bridge stp llc xfrm_user xfrm_algo xt_set ip_set nft_chain_nat 
nf_nat overlay cfg80211 rfkill ip6t_REJECT nf_reject_ipv6 xt_hl ip6_tables 
ip6t_rt ipt_REJECT nf_reject_ipv4 xt_LOG nf_log_syslog nft_limit xt_limit 
xt_addrtype xt_tcpudp xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 
nft_compat nf_tables libcrc32c nfnetlink binfmt_misc nls_ascii nls_cp437 
intel_rapl_msr vfat intel_rapl_common fat ghash_clmulni_intel sha512_ssse3 
sha512_generic sha256_ssse3 sha1_ssse3 aesni_intel crypto_simd cryptd cirrus 
drm_shmem_helper joydev sg drm_kms_helper evdev virtio_console virtio_balloon 
serio_raw button fuse loop drm efi_pstore dm_mod configfs qemu_fw_cfg 
virtio_rng ip_tables x_tables autofs4 ext4 crc16 mbcache jbd2 crc32c_generic 
hid_generic usbhid hid sd_mod t10_pi crc64_rocksoft crc64 crc_t10dif 
crct10dif_generic sr_mod cdrom virtio_net virtio_scsi net_failover uhci_hcd 
failover ata_generic ehci_hcd ata_piix libata virtio_pci virtio_pci_legacy_dev 
psmouse scsi_mod crct10dif_pclmul usbcore crct10dif_common crc32_pclmul 
virtio_pci_modern_dev i2c_piix4 crc32c_intel virtio virtio_ring usb_common 
scsi_common
[  168.363936] CR2: 0000000000000000

The faulting instruction dereferences the return value of the page cache lookup 
right after an IS_ERR() check (cmp rax,-4096; ja ...; mov rax,[rax]), so the 
lookup appears to return NULL where the code only expects an ERR_PTR.

Possibly related: during the 6.1.167-rc1 review (March 2026) a NULL pointer 
dereference in ext4_mb_load_buddy_gfp was bisected to the backport of "ext4: 
convert bd_buddy_page to bd_buddy_folio":
https://lkml.iu.edu/2603.3/01511.html

The full vmcore and dmesg are available if useful. Workaround: boot 
linux-image-6.1.0-54-amd64 and hold linux-image-amd64.

Regards,
Thomas

Reply via email to