possible deadlock in md_alloc

Status: upstream: reported C repro on 2026/10/31 09:26
Subsystems: block
Labels: prio:low
[Documentation on labels]
Reported-by: syzbot+cd254889291ef8666036@syzkaller.appspotmail.com
First crash: 6d11h, last: 15m
✨ AI-generated root cause summary:

A circular locking dependency exists between major_names_lock, md's internal disks_mutex, and mddev->reconfig_mutex. In blk_probe_dev(), major_names_lock is held while invoking the driver's autoload probe callback md_probe(), which calls md_alloc() and acquires disks_mutex followed by mddev->reconfig_mutex. Concurrently, md ioctls (such as ADD_NEW_DISK) hold mddev->reconfig_mutex while calling bdev_file_open_by_dev(), which triggers block device autoloading and attempts to acquire major_names_lock via blk_request_module() / blk_probe_dev(), resulting in a potential ABBA deadlock.

▶ ▼ ✨ Full AI root cause analysis
Based on kernel commit 151a134a4e6e:

Root Cause Analysis

A circular locking dependency (potential deadlock) exists between major_names_lock (in block/genhd.c), disks_mutex (in drivers/md/md.c), and &mddev->reconfig_mutex (in drivers/md/md.c).

The Circular Dependency Chain

  1. disks_mutex $\rightarrow$ &mddev->reconfig_mutex: In drivers/md/md.c:md_alloc() (e.g., invoked when creating an md array via /sys/module/md_mod/parameters/new_array $\rightarrow$ add_named_array() $\rightarrow$ md_alloc_and_put()):

  2. &mddev->reconfig_mutex $\rightarrow$ major_names_lock: In drivers/md/md.c:md_ioctl():

    • When executing ioctl operations like ADD_NEW_DISK or HOT_ADD_DISK, mddev_lock(mddev) is held (drivers/md/md.c:8426), holding &mddev->reconfig_mutex.
    • In md_add_new_disk(), the kernel calls md_import_device(dev, ...) (drivers/md/md.c:7681).
    • md_import_device() calls bdev_file_open_by_dev() (drivers/md/md.c:3861).
    • bdev_file_open_by_dev() calls blkdev_get_no_open(dev, true) (block/bdev.c:1091).
    • With legacy block device autoloading enabled (CONFIG_BLOCK_LEGACY_AUTOLOAD), blkdev_get_no_open() calls blk_request_module() (block/bdev.c:870), which in turn calls blk_probe_dev() (block/genhd.c:901).
    • blk_probe_dev() acquires mutex_lock(&major_names_lock) (block/genhd.c:885).

  3. major_names_lock $\rightarrow$ disks_mutex: When an unopened block device node is opened via userspace (sys_openat $\rightarrow$ blkdev_open() $\rightarrow$ blkdev_get_no_open(dev, true)):

    • blk_request_module() calls blk_probe_dev() (block/genhd.c:885).
    • blk_probe_dev() acquires mutex_lock(&major_names_lock) and iterates over major_names.
    • While still holding major_names_lock, it invokes the driver's registered .probe callback ((*n)->probe(devt)) (block/genhd.c:888).
    • For MD devices, this calls md_probe() (drivers/md/md.c:6463).
    • When create_on_open is true (default), md_probe() calls md_alloc_and_put(), which calls md_alloc().
    • md_alloc() calls mutex_lock(&disks_mutex) (drivers/md/md.c:6355).

Consequence in a Production Kernel

In a production kernel without lockdep, if one thread runs an ioctl(ADD_NEW_DISK) on an existing array (holding &mddev->reconfig_mutex and attempting to acquire major_names_lock during component device probe/open), and a concurrent thread accesses/opens an MD device node (holding major_names_lock in blk_probe_dev() and attempting to acquire disks_mutex or &mddev->reconfig_mutex), both threads will deadlock, hanging all subsequent block device probing and MD administration indefinitely.

Classification

  • RootCauseFiles: block/genhd.c, drivers/md/md.c
  • MemoryCorruption: false (locking order violation causing deadlocks; no memory safety violation or corrupted pointers)
  • InfoLeak: "none" (no kernel memory disclosure to userspace)
  • RawDenialOfService: true (mutual deadlock between kernel threads resulting in indefinite hanging of tasks and block device subsystem operations)
  • IsNetCandidate: false (block layer / md raid subsystem)
  • IsKVMCandidate: false (not KVM / virtualization related)
▶ ▼ ✨ AI Jobs (2)
ID Workflow Result Correct Bug Created Started Finished Revision
d2ad509d repro-c Reproduced: ✅ ❓ possible deadlock in md_alloc 2026/10/30 00:03 2026/10/30 00:03 2026/10/30 02:22 2dbcf159
67859728 assessment-security DenialOfService: ✅ Exploitable: ❌ FilesystemTrigger: ❌ FuzzingArtifact: ❌ MemoryCorruption: ❌ NetworkNeedsSetup: ❌ NetworkTrigger: ❌ PeripheralTrigger: ❌ RemoteTrigger: ❌ Unprivileged: ❌ UserNamespace: ❌ VMGuestTrigger: ❌ VMHostTrigger: ❌ ❓ possible deadlock in md_alloc 2026/10/29 01:32 2026/10/29 01:32 2026/10/29 02:30 b1d00f76
▶ ▼ Discussions (1)
Title Replies (including bot) Last reply
[syzbot] [block?] possible deadlock in md_alloc 0 (1) 2026/10/31 09:26
▶ ▼ Similar bugs (1)
Kernel Title Rank 🛈 Repro Cause bisect Fix bisect Count Last Reported Patched Status
linux-5.15 possible deadlock in md_alloc 4 1 27d 27d 0/3 upstream: reported on 2026/10/06 03:16
▶ ▼ Last patch testing requests (1)
Created Duration User Patch Repo Result
2026/10/30 03:23 18m retest repro git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git a39c950806213f6861d00f90729acaaa21f2d95a report job log log

Sample crash report:
md: could not open device unknown-block(8,200).
md: error, md_import_device() returned -6
md: could not open device unknown-block(8,250).
md: error, md_import_device() returned -6
======================================================
WARNING: possible circular locking dependency detected
syzkaller #0 Not tainted
------------------------------------------------------
syz-executor140/6002 is trying to acquire lock:
ffffffff91329200 (disks_mutex){+.+.}-{4:4}, at: md_alloc+0x40/0xf40 drivers/md/md.c:6355

but task is already holding lock:
ffffffff9096c060 (major_names_lock){+.+.}-{4:4}, at: blk_probe_dev block/genhd.c:885 [inline]
ffffffff9096c060 (major_names_lock){+.+.}-{4:4}, at: blk_request_module+0x35/0x2a0 block/genhd.c:901

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #2 (major_names_lock){+.+.}-{4:4}:
       __mutex_lock_common kernel/locking/mutex.c:646 [inline]
       __mutex_lock+0x197/0x15a0 kernel/locking/mutex.c:821
       blk_probe_dev block/genhd.c:885 [inline]
       blk_request_module+0x35/0x2a0 block/genhd.c:901
       blkdev_get_no_open+0x3f/0xe0 block/bdev.c:870
       bdev_file_open_by_dev+0xa0/0x240 block/bdev.c:1091
       md_import_device+0x2e9/0x650 drivers/md/md.c:3861
       md_add_new_disk+0x522/0x1850 drivers/md/md.c:7681
       md_ioctl+0x2027/0x2630 drivers/md/md.c:8525
       blkdev_ioctl+0x5e3/0x740 block/ioctl.c:798
       vfs_ioctl fs/ioctl.c:51 [inline]
       __do_sys_ioctl fs/ioctl.c:597 [inline]
       __se_sys_ioctl+0xfc/0x170 fs/ioctl.c:583
       do_syscall_x64 arch/x86/entry/syscall_64.c:61 [inline]
       do_syscall_64+0x166/0x520 arch/x86/entry/syscall_64.c:84
       entry_SYSCALL_64_after_hwframe+0x77/0x7f

-> #1 (&mddev->reconfig_mutex){+.+.}-{4:4}:
       __mutex_lock_common kernel/locking/mutex.c:646 [inline]
       __mutex_lock+0x197/0x15a0 kernel/locking/mutex.c:821
       mddev_lock_nointr drivers/md/md.h:734 [inline]
       md_alloc+0x81e/0xf40 drivers/md/md.c:6413
       md_alloc_and_put+0x1a/0x1f0 drivers/md/md.c:6452
       add_named_array+0x240/0x350 drivers/md/md.c:-1
       param_attr_store+0x214/0x310 kernel/params.c:591
       module_attr_store+0x65/0x90 kernel/params.c:906
       kernfs_fop_write_iter+0x3a4/0x540 fs/kernfs/file.c:345
       new_sync_write fs/read_write.c:595 [inline]
       vfs_write+0x612/0xba0 fs/read_write.c:687
       ksys_write+0x150/0x270 fs/read_write.c:1036
       do_syscall_x64 arch/x86/entry/syscall_64.c:61 [inline]
       do_syscall_64+0x166/0x520 arch/x86/entry/syscall_64.c:84
       entry_SYSCALL_64_after_hwframe+0x77/0x7f

-> #0 (disks_mutex){+.+.}-{4:4}:
       check_prev_add kernel/locking/lockdep.c:3209 [inline]
       check_prevs_add kernel/locking/lockdep.c:3328 [inline]
       validate_chain kernel/locking/lockdep.c:3952 [inline]
       __lock_acquire+0x164c/0x2de0 kernel/locking/lockdep.c:5288
       lock_acquire+0x115/0x350 kernel/locking/lockdep.c:5942
       __mutex_lock_common kernel/locking/mutex.c:646 [inline]
       __mutex_lock+0x197/0x15a0 kernel/locking/mutex.c:821
       md_alloc+0x40/0xf40 drivers/md/md.c:6355
       md_alloc_and_put+0x1a/0x1f0 drivers/md/md.c:6452
       blk_probe_dev block/genhd.c:888 [inline]
       blk_request_module+0x27d/0x2a0 block/genhd.c:-1
       blkdev_get_no_open+0x3f/0xe0 block/bdev.c:870
       blkdev_open+0x1d7/0x600 block/fops.c:678
       do_dentry_open+0x80e/0x1360 fs/open.c:1000
       vfs_open_consume+0x4a/0x350 fs/open.c:1131
       do_open fs/namei.c:4845 [inline]
       path_openat+0x16fc/0x20f0 fs/namei.c:5010
       do_file_open+0x23e/0x4a0 fs/namei.c:5039
       do_sys_openat2+0x115/0x200 fs/open.c:1442
       do_sys_open fs/open.c:1448 [inline]
       __do_sys_openat fs/open.c:1788 [inline]
       __se_sys_openat fs/open.c:1783 [inline]
       __x64_sys_openat+0x138/0x170 fs/open.c:1783
       do_syscall_x64 arch/x86/entry/syscall_64.c:61 [inline]
       do_syscall_64+0x166/0x520 arch/x86/entry/syscall_64.c:84
       entry_SYSCALL_64_after_hwframe+0x77/0x7f

other info that might help us debug this:

Chain exists of:
  disks_mutex --> &mddev->reconfig_mutex --> major_names_lock

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(major_names_lock);
                               lock(&mddev->reconfig_mutex);
                               lock(major_names_lock);
  lock(disks_mutex);

 *** DEADLOCK ***

locks held by syz-executor140/6002: 1, last CPU#1:
 #0: ffffffff9096c060 (major_names_lock){+.+.}-{4:4}, at: blk_probe_dev block/genhd.c:885 [inline]
 #0: ffffffff9096c060 (major_names_lock){+.+.}-{4:4}, at: blk_request_module+0x35/0x2a0 block/genhd.c:901

stack backtrace:
CPU: 1 UID: 0 PID: 6002 Comm: syz-executor140 Not tainted syzkaller #0 PREEMPT(full) 
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/07/2026
Call Trace:
 <TASK>
 dump_stack_lvl+0xe8/0x150 lib/dump_stack.c:120
 print_circular_bug+0x2e2/0x300 kernel/locking/lockdep.c:2087
 check_noncircular+0x12f/0x150 kernel/locking/lockdep.c:2219
 check_prev_add kernel/locking/lockdep.c:3209 [inline]
 check_prevs_add kernel/locking/lockdep.c:3328 [inline]
 validate_chain kernel/locking/lockdep.c:3952 [inline]
 __lock_acquire+0x164c/0x2de0 kernel/locking/lockdep.c:5288
 lock_acquire+0x115/0x350 kernel/locking/lockdep.c:5942
 __mutex_lock_common kernel/locking/mutex.c:646 [inline]
 __mutex_lock+0x197/0x15a0 kernel/locking/mutex.c:821
 md_alloc+0x40/0xf40 drivers/md/md.c:6355
 md_alloc_and_put+0x1a/0x1f0 drivers/md/md.c:6452
 blk_probe_dev block/genhd.c:888 [inline]
 blk_request_module+0x27d/0x2a0 block/genhd.c:-1
 blkdev_get_no_open+0x3f/0xe0 block/bdev.c:870
 blkdev_open+0x1d7/0x600 block/fops.c:678
 do_dentry_open+0x80e/0x1360 fs/open.c:1000
 vfs_open_consume+0x4a/0x350 fs/open.c:1131
 do_open fs/namei.c:4845 [inline]
 path_openat+0x16fc/0x20f0 fs/namei.c:5010
 do_file_open+0x23e/0x4a0 fs/namei.c:5039
 do_sys_openat2+0x115/0x200 fs/open.c:1442
 do_sys_open fs/open.c:1448 [inline]
 __do_sys_openat fs/open.c:1788 [inline]
 __se_sys_openat fs/open.c:1783 [inline]
 __x64_sys_openat+0x138/0x170 fs/open.c:1783
 do_syscall_x64 arch/x86/entry/syscall_64.c:61 [inline]
 do_syscall_64+0x166/0x520 arch/x86/entry/syscall_64.c:84
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f9cb93efb37
Code: 48 89 fa 4c 89 df e8 98 1d 00 00 8b 93 08 03 00 00 59 5e 48 83 f8 fc 74 1a 5b c3 0f 1f 84 00 00 00 00 00 48 8b 44 24 10 0f 05 <5b> c3 0f 1f 80 00 00 00 00 83 e2 39 83 fa 08 75 de e8 23 ff ff ff
RSP: 002b:00007ffd0a7a14d0 EFLAGS: 00000202 ORIG_RAX: 0000000000000101
RAX: ffffffffffffffda RBX: 000055558698d400 RCX: 00007f9cb93efb37
RDX: 0000000000000002 RSI: 00007ffd0a7a15f0 RDI: ffffffffffffff9c
RBP: 00007ffd0a7a15b0 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000202 R12: 00007ffd0a7a15f0
R13: 00000000ffffffff R14: 00007f9cb942909a R15: 0000000000000002
 </TASK>
md: async del_gendisk mode will be removed in future, please upgrade to mdadm-4.5+
block device autoloading is deprecated and will be removed.
md: async del_gendisk mode will be removed in future, please upgrade to mdadm-4.5+
block device autoloading is deprecated and will be removed.

Crashes (19):
Time Kernel Commit Syzkaller Config Log Report Syz repro C repro VM info Assets (help?) Manager Title
2026/10/30 03:41 git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git a39c95080621 b1d00f76 .config console log report C ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/11/02 20:51 upstream 6628e7d4c6b2 0a94d799 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-kasan-badwrites-root possible deadlock in md_alloc
2026/11/02 03:12 upstream 46fec1c3029b 25a0ae84 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-kasan-gce-root possible deadlock in md_alloc
2026/11/01 12:17 upstream cbbddb6b08cc 25a0ae84 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-kasan-gce-root possible deadlock in md_alloc
2026/10/31 18:21 upstream 36aad21d50d1 25a0ae84 .config console log report info [disk image (non-bootable)] [vmlinux] [kernel image] ci-qemu-upstream possible deadlock in md_alloc
2026/10/31 12:28 upstream abf8c33470f7 2041d3f8 .config console log report info [disk image (non-bootable)] [vmlinux] [kernel image] ci-qemu-upstream possible deadlock in md_alloc
2026/10/31 12:28 upstream abf8c33470f7 2041d3f8 .config console log report info [disk image (non-bootable)] [vmlinux] [kernel image] ci-qemu-upstream possible deadlock in md_alloc
2026/10/31 13:32 linux-next 7649ba73f9e7 2041d3f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-rust-kasan-gce possible deadlock in md_alloc
2026/10/31 09:23 linux-next 7649ba73f9e7 0ec395f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/10/31 08:17 linux-next 7649ba73f9e7 0ec395f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/10/31 07:26 linux-next 7649ba73f9e7 0ec395f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/10/31 07:20 linux-next 7649ba73f9e7 0ec395f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/10/30 20:42 linux-next 7649ba73f9e7 0ec395f8 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-rust-kasan-gce possible deadlock in md_alloc
2026/10/29 09:38 linux-next 6d0933ae9d46 b1d00f76 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-rust-kasan-gce possible deadlock in md_alloc
2026/10/29 08:50 linux-next 6d0933ae9d46 b1d00f76 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-rust-kasan-gce possible deadlock in md_alloc
2026/10/28 02:11 linux-next 151a134a4e6e 96923c60 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/10/27 09:16 linux-next f1ac44c32cba 168cb7a2 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-linux-next-kasan-gce-root possible deadlock in md_alloc
2026/11/02 19:42 git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci 6628e7d4c6b2 0a94d799 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-gce-arm64 possible deadlock in md_alloc
2026/11/02 17:16 git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci 6628e7d4c6b2 0a94d799 .config console log report info [disk image] [vmlinux] [kernel image] ci-upstream-gce-arm64 possible deadlock in md_alloc
* Struck through repros no longer work on HEAD.