CVE-2025-38349

eventpoll: don't decrement ep refcount while still holding the ep mutex

Published Jul 18, 2025 CVSS 3.1 · 7.8 High EPSS < 1% KEV · Not listed
Description
In the Linux kernel, the following vulnerability has been resolved:

eventpoll: don't decrement ep refcount while still holding the ep mutex

Jann Horn points out that epoll is decrementing the ep refcount and then
doing a

mutex_unlock(&ep->mtx);

afterwards. That's very wrong, because it can lead to a use-after-free.

That pattern is actually fine for the very last reference, because the
code in question will delay the actual call to "ep_free(ep)" until after
it has unlocked the mutex.

But it's wrong for the much subtler "next to last" case when somebody
*else* may also be dropping their reference and free the ep while we're
still using the mutex.

Note that this is true even if that other user is also using the same ep
mutex: mutexes, unlike spinlocks, can not be used for object ownership,
even if they guarantee mutual exclusion.

A mutex "unlock" operation is not atomic, and as one user is still
accessing the mutex as part of unlocking it, another user can come in
and get the now released mutex and free the data structure while the
first user is still cleaning up.

See our mutex documentation in Documentation/locking/mutex-design.rst,
in particular the section [1] about semantics:

"mutex_unlock() may access the mutex structure even after it has
internally released the lock already - so it's not safe for
another context to acquire the mutex and assume that the
mutex_unlock() context is not using the structure anymore"

So if we drop our ep ref before the mutex unlock, but we weren't the
last one, we may then unlock the mutex, another user comes in, drops
_their_ reference and releases the 'ep' as it now has no users - all
while the mutex_unlock() is still accessing it.

Fix this by simply moving the ep refcount dropping to outside the mutex:
the refcount itself is atomic, and doesn't need mutex protection (that's
the whole _point_ of refcounts: unlike mutexes, they are inherently
about object lifetimes).

Remediation

No solution or workaround provided in the CVE record

Tracking

Sign in to view the affected projects.

Advisories

Source ID Title
Debian DSA Debian DSA DSA-5975-1 linux security update
EUVD EUVD EUVD-2025-21860 In the Linux kernel, the following vulnerability has been resolved: eventpoll: don't decrement ep refcount while still holding the ep mutex Jann Horn points out that epoll is decrementing the ep refcount and then doing a mutex_unlock(&ep->mtx); afterwards. That's very wrong, because it can lead to a use-after-free. That pattern is actually fine for the very last reference, because the code in question will delay the actual call to "ep_free(ep)" until after it has unlocked the mutex. But it's wrong for the much subtler "next to last" case when somebody *else* may also be dropping their reference and free the ep while we're still using the mutex. Note that this is true even if that other user is also using the same ep mutex: mutexes, unlike spinlocks, can not be used for object ownership, even if they guarantee mutual exclusion. A mutex "unlock" operation is not atomic, and as one user is still accessing the mutex as part of unlocking it, another user can come in and get the now released mutex and free the data structure while the first user is still cleaning up. See our mutex documentation in Documentation/locking/mutex-design.rst, in particular the section [1] about semantics: "mutex_unlock() may access the mutex structure even after it has internally released the lock already - so it's not safe for another context to acquire the mutex and assume that the mutex_unlock() context is not using the structure anymore" So if we drop our ep ref before the mutex unlock, but we weren't the last one, we may then unlock the mutex, another user comes in, drops _their_ reference and releases the 'ep' as it now has no users - all while the mutex_unlock() is still accessing it. Fix this by simply moving the ep refcount dropping to outside the mutex: the refcount itself is atomic, and doesn't need mutex protection (that's the whole _point_ of refcounts: unlike mutexes, they are inherently about object lifetimes).
Ubuntu USN Ubuntu USN USN-7879-1 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-7879-2 Linux kernel (Real-time) vulnerabilities
Ubuntu USN Ubuntu USN USN-7880-1 Linux kernel (OEM) vulnerabilities
Ubuntu USN Ubuntu USN USN-7879-3 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-7879-4 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-7934-1 Linux kernel (Azure) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-1 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-2 Linux kernel (Real-time) vulnerabilities
Ubuntu USN Ubuntu USN USN-8031-1 Linux kernel (GCP) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-3 Linux kernel (Real-time) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-4 Linux kernel (FIPS) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-5 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-8031-2 Linux kernel (GCP FIPS) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-6 Linux kernel (HWE) vulnerabilities
Ubuntu USN Ubuntu USN USN-8031-3 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-8052-1 Linux kernel (Low Latency) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-7 Linux kernel (Low Latency NVIDIA) vulnerabilities
Ubuntu USN Ubuntu USN USN-8028-8 Linux kernel (IBM) vulnerabilities
Ubuntu USN Ubuntu USN USN-8052-2 Linux kernel (Xilinx) vulnerabilities
Ubuntu USN Ubuntu USN USN-8074-1 Linux kernel (Azure) vulnerabilities
Ubuntu USN Ubuntu USN USN-8074-2 Linux kernel (Azure FIPS) vulnerabilities
Ubuntu USN Ubuntu USN USN-8126-1 Linux kernel (Azure) vulnerabilities
Ubuntu USN Ubuntu USN USN-8961-1 Linux kernel vulnerabilities
Ubuntu USN Ubuntu USN USN-9029-1 Linux kernel (Raspberry Pi) vulnerabilities
History
Type Values Removed Values Added
Metrics ssvc

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

Type Values Removed Values Added
Weaknesses CWE-416
CPEs cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.16:rc1:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.16:rc2:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.16:rc3:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.16:rc4:*:*:*:*:*:*
cpe:2.3:o:linux:linux_kernel:6.16:rc5:*:*:*:*:*:*
Metrics cvssV3_1

{'score': 7.0, 'vector': 'CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H'}

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
References
Metrics threat_severity

None

cvssV3_1

{'score': 7.0, 'vector': 'CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H'}

threat_severity

Moderate

Type Values Removed Values Added
Description In the Linux kernel, the following vulnerability has been resolved: eventpoll: don't decrement ep refcount while still holding the ep mutex Jann Horn points out that epoll is decrementing the ep refcount and then doing a mutex_unlock(&ep->mtx); afterwards. That's very wrong, because it can lead to a use-after-free. That pattern is actually fine for the very last reference, because the code in question will delay the actual call to "ep_free(ep)" until after it has unlocked the mutex. But it's wrong for the much subtler "next to last" case when somebody *else* may also be dropping their reference and free the ep while we're still using the mutex. Note that this is true even if that other user is also using the same ep mutex: mutexes, unlike spinlocks, can not be used for object ownership, even if they guarantee mutual exclusion. A mutex "unlock" operation is not atomic, and as one user is still accessing the mutex as part of unlocking it, another user can come in and get the now released mutex and free the data structure while the first user is still cleaning up. See our mutex documentation in Documentation/locking/mutex-design.rst, in particular the section [1] about semantics: "mutex_unlock() may access the mutex structure even after it has internally released the lock already - so it's not safe for another context to acquire the mutex and assume that the mutex_unlock() context is not using the structure anymore" So if we drop our ep ref before the mutex unlock, but we weren't the last one, we may then unlock the mutex, another user comes in, drops _their_ reference and releases the 'ep' as it now has no users - all while the mutex_unlock() is still accessing it. Fix this by simply moving the ep refcount dropping to outside the mutex: the refcount itself is atomic, and doesn't need mutex protection (that's the whole _point_ of refcounts: unlike mutexes, they are inherently about object lifetimes).
Title eventpoll: don't decrement ep refcount while still holding the ep mutex
References
Weaknesses

Sources

Status
PUBLISHED
Assigner
Linux
Published
Updated
Oct 3, 2026, 10:55 AM
Reserved
Apr 16, 2025, 4:51 AM
Status
Modified
Published
Jul 18, 2025, 8:15 AM
Modified
Oct 3, 2026, 11:17 AM
Severity
Moderate
Public date
Jul 18, 2025, 12:00 AM