CVE-2026-64112

rbd: eliminate a race in lock_dwork draining on unmap

Published Jul 19, 2026 CVSS 3.1 · 7.8 High EPSS < 1% KEV · Not listed
Description
In the Linux kernel, the following vulnerability has been resolved:

rbd: eliminate a race in lock_dwork draining on unmap

Given how rbd_lock_add_request() and rbd_img_exclusive_lock() are
written, lock_dwork may be (re)queued more than it's actually needed:
for example in case a new I/O request comes in while we are in the
middle of rbd_acquire_lock() on behalf of another I/O request. This is
expected and with rbd_release_lock() preemptively canceling lock_dwork
is benign under normal operation.

A more problematic example is maybe_kick_acquire():

if (have_requests || delayed_work_pending(&rbd_dev->lock_dwork)) {
dout("%s rbd_dev %p kicking lock_dwork\n", __func__, rbd_dev);
mod_delayed_work(rbd_dev->task_wq, &rbd_dev->lock_dwork, 0);
}

It's not unrealistic for lock_dwork to get canceled right after
delayed_work_pending() returns true and for mod_delayed_work() to
requeue it right there anyway. This is a classic TOCTOU race.

When it comes to unmapping the image, there is an implicit assumption
of no self-initiated exclusive lock activity past the point of return
from rbd_dev_image_unlock() which unlocks the lock if it happens to be
held. This unlock is assumed to be final and lock_dwork (as well as
all other exclusive lock tasks, really) isn't expected to get queued
again. However, lock_dwork is canceled only in cancel_tasks_sync()
(i.e. later in the unmap sequence) and on top of that the cancellation
can get in effect nullified by maybe_kick_acquire(). This may result
in rbd_acquire_lock() executing after rbd_dev_device_release() and
rbd_dev_image_release() run and free and/or reset a bunch of things.
One of the possible failure modes then is a violated

rbd_assert(rbd_image_format_valid(rbd_dev->image_format));

in rbd_dev_header_info() which is called via rbd_dev_refresh() from
rbd_post_acquire_action().

Redo exclusive lock task draining to provide saner semantics and try
to meet the assumptions around rbd_dev_image_unlock().

Analysis

AI-generated

Technical impact

The flaw lies in the RADOS Block Device (RBD) subsystem of the Linux kernel, where a TOCTOU race is introduced between the cancellation and re‑queueing of the exclusive‑lock work item (lock_dwork). When a device is unmapped, the code assumes that the lock work has already finished, but the race can allow lock_dwork to run after rbd_dev_device_release() has freed critical kernel structures. This leads to memory corruption and an assertion failure in rbd_dev_header_info(), potentially causing a kernel panic or other severe corruption. The weaknesses are identified as CWE‑362 and CWE‑367.

Affected Systems

All Linux kernel builds that include the RBD subsystem are potentially affected until the patch (commit 3427d7ae from the stable branch) is applied. No specific version range is supplied in the CNA data; therefore, any kernel with RBD enabled is at risk.

Risk and Exploitability

With a CVSS score of 7.8 the severity is high, yet the EPSS score of less than 1% indicates a very low likelihood of exploitation. The vulnerability requires local or privileged interaction with RBD I/O and unmap operations, making reliable exploitation challenging. It is not listed in the CISA KEV catalog, implying no publicly known exploits. Nevertheless, because a kernel crash can be catastrophic, the risk warrants prompt remediation.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Apply the latest stable Linux kernel that incorporates commit 3427d7ae or patch your kernel with that commit directly.
  • If an immediate kernel upgrade is not possible, limit RBD device access to trusted users, avoid performing unmap operations while the device is active, and disable RBD locking features if they are not required.
  • Continuously monitor kernel logs for lock or assertion failures and validate system stability after any changes.

Generated by OpenCVE AI on October 20, 2026 at 09:33 UTC.

Tracking

Sign in to view the affected projects.

Advisories

Source ID Title
Ubuntu USN Ubuntu USN USN-8593-1 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-8603-1 Linux kernel (Azure) vulnerabilities
Ubuntu USN Ubuntu USN USN-8618-1 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-8663-1 Linux kernel (NVIDIA) vulnerabilities
Ubuntu USN Ubuntu USN USN-8664-1 Linux kernel (NVIDIA BaseOS) vulnerabilities
Ubuntu USN Ubuntu USN USN-8728-1 Linux kernel (GCP) vulnerabilities
Ubuntu USN Ubuntu USN USN-8781-1 Linux kernel (NVIDIA Tegra) vulnerabilities
Ubuntu USN Ubuntu USN USN-8728-2 Linux kernel (Azure) vulnerabilities
Ubuntu USN Ubuntu USN USN-8728-3 Linux kernel (Oracle) vulnerabilities
History
Type Values Removed Values Added
Weaknesses CWE-362
Type Values Removed Values Added
Metrics ssvc

{'options': {'Automatable': 'no', 'Exploitation': 'none', 'Technical Impact': 'total'}, 'version': '2.0.3'}

Type Values Removed Values Added
CPEs cpe:2.3:o:linux:linux_kernel:7.1:rc1:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:7.1:rc2:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:7.1:rc3:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:7.1:rc4:*:*:*:*:*:*
Type Values Removed Values Added
Metrics cvssV3_1

{'score': 7.8, 'vector': 'CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H'}

Type Values Removed Values Added
Description In the Linux kernel, the following vulnerability has been resolved: rbd: eliminate a race in lock_dwork draining on unmap Given how rbd_lock_add_request() and rbd_img_exclusive_lock() are written, lock_dwork may be (re)queued more than it's actually needed: for example in case a new I/O request comes in while we are in the middle of rbd_acquire_lock() on behalf of another I/O request. This is expected and with rbd_release_lock() preemptively canceling lock_dwork is benign under normal operation. A more problematic example is maybe_kick_acquire(): if (have_requests || delayed_work_pending(&rbd_dev->lock_dwork)) { dout("%s rbd_dev %p kicking lock_dwork\n", __func__, rbd_dev); mod_delayed_work(rbd_dev->task_wq, &rbd_dev->lock_dwork, 0); } It's not unrealistic for lock_dwork to get canceled right after delayed_work_pending() returns true and for mod_delayed_work() to requeue it right there anyway. This is a classic TOCTOU race. When it comes to unmapping the image, there is an implicit assumption of no self-initiated exclusive lock activity past the point of return from rbd_dev_image_unlock() which unlocks the lock if it happens to be held. This unlock is assumed to be final and lock_dwork (as well as all other exclusive lock tasks, really) isn't expected to get queued again. However, lock_dwork is canceled only in cancel_tasks_sync() (i.e. later in the unmap sequence) and on top of that the cancellation can get in effect nullified by maybe_kick_acquire(). This may result in rbd_acquire_lock() executing after rbd_dev_device_release() and rbd_dev_image_release() run and free and/or reset a bunch of things. One of the possible failure modes then is a violated rbd_assert(rbd_image_format_valid(rbd_dev->image_format)); in rbd_dev_header_info() which is called via rbd_dev_refresh() from rbd_post_acquire_action(). Redo exclusive lock task draining to provide saner semantics and try to meet the assumptions around rbd_dev_image_unlock().
Title rbd: eliminate a race in lock_dwork draining on unmap
First Time appeared Linux
Linux linux Kernel
CPEs cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*
Vendors & Products Linux
Linux linux Kernel
References
Weaknesses
  • CWE-362

    Concurrent Execution using Shared Resource with Improper Synchronization ('Race Condition')

  • CWE-367

    Time-of-check Time-of-use (TOCTOU) Race Condition

Sources

Status
PUBLISHED
Assigner
Linux
Published
Updated
Oct 19, 2026, 8:28 PM
Reserved
Jul 19, 2026, 7:54 AM
Status
Modified
Published
Jul 19, 2026, 4:17 PM
Modified
Oct 19, 2026, 9:17 PM
Severity
Moderate
Public date
Jul 19, 2026, 12:00 AM