[syzbot] [kernfs?] BUG: unable to handle kernel paging request in kernfs_unlink_sibling

[syzbot] [kernfs?] BUG: unable to handle kernel paging request in kernfs_unlink_sibling

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit:    58c1eb84e977 arm64: sdei: Preserve the interrupted context..
git tree:       git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci
console output: https://syzkaller.appspot.com/x/log.txt?x=10ee7f8d580000
kernel config:  https://syzkaller.appspot.com/x/.config?x=a58b6b4d667e7a61
dashboard link: https://syzkaller.appspot.com/bug?extid=02ab1258b7fa66eb1f7f
compiler:       Debian clang version 22.1.8 (1~deb13u4), Debian LLD 22.1.8
userspace arch: arm64
syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=17119f8d580000

Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/b1ec47537e12/disk-58c1eb84.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/3bca0c2b1169/vmlinux-58c1eb84.xz
kernel image: https://storage.googleapis.com/syzbot-assets/0a7c49594d5a/Image-58c1eb84.gz.xz

AI-generated root cause summary:
The crash manifests during network namespace teardown (cleanup_net ->
rdma_dev_exit_net -> ib_free_port_attrs -> destroy_gid_attrs ->
kernfs_unlink_sibling -> rb_erase) when ____rb_erase_color encounters a NULL
sibling pointer and dereferences address 0x0 (detected by KASAN at shadow
address dfff800000000000). In a valid red-black tree, ____rb_erase_color is only
entered after deleting a black node, so the sibling subtree must contain at
least one black node and sibling can never be NULL; furthermore, all kernfs
directory operations are strictly protected by root->kernfs_rwsem. This rbtree
corruption is induced externally: userspace mmaps the PCI BAR0 MMIO window
exposed via sysfs (/sys/bus/pci/devices/.../resource0 in
drivers/pci/pci-sysfs.c) and writes arbitrary data into device MMIO space,
reprogramming hardware DMA queues and causing device DMA to overwrite kernel
memory structures.

full analysis:  https://syzkaller.appspot.com/x/ai_root_cause?x=02ab1258b7fa66eb1f7f

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+02ab1258b7fa66eb1f7f@syzkaller.appspotmail.com

Unable to handle kernel paging request at virtual address dfff800000000000
KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
Mem abort info:
  ESR = 0x0000000096000005
  EC = 0x25: DABT (current EL), IL = 32 bits
  SET = 0, FnV = 0
  EA = 0, S1PTW = 0
  FSC = 0x05: level 1 translation fault
Data abort info:
  ISV = 0, ISS = 0x00000005, ISS2 = 0x00000000
  CM = 0, WnR = 0, TnD = 0, TagAccess = 0
  GCS = 0, Overlay = 0, DirtyBit = 0
[dfff800000000000] address between user and kernel address ranges
Internal error: Oops: 0000000096000005 [#1]  SMP
Modules linked in:
CPU: 1 UID: 0 PID: 1195 Comm: kworker/u8:6 Not tainted syzkaller #0 PREEMPT 
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 11/06/2026
Workqueue: netns cleanup_net
pstate: 63400005 (nZCv daif +PAN -UAO +TCO +DIT -SSBS BTYPE=--)
pc : ____rb_erase_color lib/rbtree.c:354 [inline]
pc : rb_erase+0x4e4/0xba0 lib/rbtree.c:445
lr : kernfs_unlink_sibling+0x190/0x218 fs/kernfs/dir.c:449
sp : ffff800094d775f0
x29: ffff800094d77600 x28: ffff0000d985c408 x27: ffff0000d985c410
x26: 1fffe0001b30b882 x25: dfff800000000000 x24: ffff0000c6b58ca0
x23: 0000000000000000 x22: 0000000000000000 x21: ffff0000d98545e0
x20: ffff0000d985c400 x19: 0000000000000000 x18: 1fffe000359d3e28
x17: 0000000000000003 x16: ffff8000898b7000 x15: 0000000000000000
x14: 0000000000287419 x13: 0000000000000001 x12: 0000000000000003
x11: 0000000000000000 x10: 0000000000000001 x9 : 0000000000000000
x8 : 0000000000000000 x7 : ffff800080f6a0dc x6 : 0000000000000000
x5 : 0000000000000000 x4 : 0000000000000000 x3 : ffff80008046e414
x2 : 0000000000000001 x1 : ffff0000c6b58ca0 x0 : ffff0000d98545e0
Call trace:
 ____rb_erase_color lib/rbtree.c:354 [inline] (P)
 rb_erase+0x4e4/0xba0 lib/rbtree.c:445 (P)
 kernfs_unlink_sibling+0x190/0x218 fs/kernfs/dir.c:449
 __kernfs_remove+0x36c/0x748 fs/kernfs/dir.c:1584
 kernfs_remove_by_name_ns+0x70/0xd0 fs/kernfs/dir.c:1796
 kernfs_remove_by_name include/linux/kernfs.h:644 [inline]
 remove_files+0x90/0x184 fs/sysfs/group.c:28
 sysfs_remove_group+0x98/0x160 fs/sysfs/group.c:328
 sysfs_remove_groups+0x5c/0xb4 fs/sysfs/group.c:352
 destroy_gid_attrs drivers/infiniband/core/sysfs.c:1143 [inline]
 ib_free_port_attrs+0xa8/0x340 drivers/infiniband/core/sysfs.c:1367
 remove_one_compat_dev+0x54/0x7c drivers/infiniband/core/device.c:1035
 rdma_dev_exit_net+0x188/0x388 drivers/infiniband/core/device.c:1173
 ops_exit_list net/core/net_namespace.c:200 [inline]
 ops_undo_list+0x3cc/0x81c net/core/net_namespace.c:253
 cleanup_net+0x474/0x6c4 net/core/net_namespace.c:708
 process_one_work kernel/workqueue.c:3495 [inline]
 process_scheduled_works+0x91c/0x1250 kernel/workqueue.c:3578
 worker_thread+0x798/0xbd0 kernel/workqueue.c:3659
 kthread+0x304/0x3d4 kernel/kthread.c:436
 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:854
Code: 91004280 968b3531 f9400376 d343fed7 (38796ae8) 
---[ end trace 0000000000000000 ]---
----------------
Code disassembly (best guess):
   0:	91004280 	add	x0, x20, #0x10
   4:	968b3531 	bl	0xfffffffffa2cd4c8
   8:	f9400376 	ldr	x22, [x27]
   c:	d343fed7 	lsr	x23, x22, #3
* 10:	38796ae8 	ldrb	w8, [x23, x25] <-- trapping instruction


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

If the report is already addressed, let syzbot know by replying with:
#syz fix: exact-commit-title

If you want to overwrite report's subsystems, reply with:
#syz set subsystems: new-subsystem
(See the list of subsystem names on the web dashboard)

If the report is a duplicate of another one, reply with:
#syz dup: exact-subject-of-another-report

If you want to undo deduplication, reply with:
#syz undup