data built 2026-10-19T13:16:53Z

Look up another CVE · open CVE-2026-64541 in the interactive tracker

CVE-2026-64541

net/smc: fix UAF in smc_cdc_rx_handler() by pinning the socket

9.8 EPSS 0.6% AV:N

Published 2026-07-27 (3 months ago) by the Linux kernel CNA. Subsystem net/smc. Debian urgency: not yet assigned.

What to do, per Debian release

Debian's own status for CVE-2026-64541 in every release this site tracks, and what each one means in practice. Every status below is the one the Debian Security Team publishes for this CVE. source

ReleaseStatus VersionWhat this means
Debian 12bookwormfixedDLA-4720-1 · 2026-08-05 · text6.1.180-1Update the kernel and reboot. DLA-4720-1 shipped the fix on 2026-08-05, in 6.1.180-1. A running kernel keeps the old code until the machine restarts.
Debian 13trixiefixedDSA-6405-1 · 2026-07-31 · text6.12.100-1Update the kernel and reboot. DSA-6405-1 shipped the fix on 2026-07-31, in 6.12.100-1. A running kernel keeps the old code until the machine restarts.
Debian 14forkyfixed7.1.5-1Update the kernel and reboot. Fixed in 7.1.5-1. A running kernel keeps the old code until the machine restarts.
Debian unstablesidfixed7.1.5-1Update the kernel and reboot. Fixed in 7.1.5-1. A running kernel keeps the old code until the machine restarts.
Debian 12bookworm · linux-6.12fixedDLA-4724-1 · 2026-08-07 · text6.12.100-1~deb12u1Update the kernel and reboot. DLA-4724-1 shipped the fix on 2026-08-07, in 6.12.100-1~deb12u1. A running kernel keeps the old code until the machine restarts.

The vulnerability

What the bug is

smc_cdc_rx_handler() looks up the connection by token under the link group's conns_lock, drops the lock, and then dereferences conn and the smc_sock derived from it, ending in sock_hold(&smc->sk) inside smc_cdc_msg_recv(). No reference is held across the lock release. The only reference pinning the socket while the connection is discoverable in the link group is taken in smc_lgr_register_conn() (sock_hold) and dropped in __smc_lgr_unregister_conn() (sock_put), both under conns_lock. Once the handler drops conns_lock, a concurrent close() -> smc_release() -> smc_conn_free() -> smc_lgr_unregister_conn() can drop that reference and free the smc_sock, so the handler's later sock_hold() runs on freed memory: WARNING: lib/refcount.c:25 at refcount_warn_saturate Workqueue: rxe_wq do_work refcount_warn_saturate (lib/refcount.c:25) smc_cdc_msg_recv (net/smc/smc_cdc.c:430) smc_cdc_rx_handler (net/smc/smc_cdc.c:502) smc_wr_rx_tasklet_fn (net/smc/smc_wr.c:445) tasklet_action_common (kernel/softirq.c:938) handle_softirqs (kernel/softirq.c:622) Kernel panic - not syncing: panic_on_warn set Only SMC-R is affected. The SMC-D receive tasklet is stopped by tasklet_kill(&conn->rx_tsklet) in smc_conn_free() before the connection is unregistered, so it cannot run concurrently with the free. Take the socket reference while still holding conns_lock, so the registration reference can no longer be the last one, and drop it once the handler is done.

Where it lives

The fix touches these files. If the subsystem is one you do not use, the practical exposure is lower, though the package is still the vulnerable one.

net/smc/smc_cdc.c

What this is
SHARED MEMORY COMMUNICATIONS (SMC) SOCKETSMAINTAINERS section matched on net/smc/ · linux-rdma@vger.kernel.org · Supported
Kernel module
smc

Check it on the machine

grep -w CONFIG_SMC /boot/config-$(uname -r)

Reads the booted kernel's own configuration, on the machine and nowhere else. =y is built in, =m is a module, and an is not set line means the code is not there.

lsmod | grep -w smc

Prints a line if that module is loaded right now, and nothing if it is not. One that is not loaded can still be loaded later by anything that needs it.

Show whether Debian builds this, per release

Debian does not build this code in 2 releases. It does build it in 3 releases, so the answer depends on which release the machine runs.

  • bookworm, trixiem
  • forky, sidn
  • bookworm · linux-6.12m
m
a loadable module on every flavour - check lsmod, can be blacklisted
n
not enabled on any flavour - this release does not build the code
Build switch
CONFIG_SMC

From the configuration Debian publishes for every kernel flavour it builds. It covers only the architectures Debian ships a linux-config package for, and nothing about a kernel you built yourself. Debian's own status above stays authoritative about the source package.MAINTAINERS kbuild Makefiles Debian linux-config packages

Signals

Published
2026-07-27 (3 months ago)source
CVSS
9.8 critical
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:Hsource recompute it
Attack vector
network (AV:N)source
EPSS
0.56%, higher than 45.8% of all CVEssource
CISA KEV
not listedthe KEV catalogue
Debian urgency
not yet assignedsource
Subsystem
net/smc

Upstream fixes

The versions the CNA records as introducing and fixing this, and the commit for each. source

IntroducedFixed inCommit
4.185.10.2618de4f665d0fe
4.185.15.2128145b4321362
4.186.1.1781951bffbc649
4.186.6.145647b19e5cc14
4.186.12.97472e9d7c0d5b
4.186.18.403bfb96d9bc6a
4.187.1.5ce5aa8084329
4.187.29d160b35cc34

Debian advisories

AdvisoryDateShipped toDebian
DLA-4724-12026-08-07bookworm 6.12.100-1~deb12u1tracker
DLA-4723-12026-08-07bullseye 6.1.180-1~deb11u1tracker
DLA-4720-12026-08-05bookworm 6.1.180-1tracker
DLA-4717-12026-08-05bullseye 5.10.262-1tracker
DSA-6405-12026-07-31trixie 6.12.100-1tracker

Why the CNA scored it this way

Quoted from the kernel CNA's own scoring file, word for word. A highlighted metric carries the most severe value CVSS v3.1 defines for it, which is what pushes the score up. source

AV:NAttack vector, network
The vulnerable handler is driven entirely by a CDC control message sent by the remote SMC-R peer over the RDMA link, with an attacker-chosen `token` field and no local syscall involvement. SMC-R runs over RoCEv2 (IP/UDP-routable, port 4791) with the CLC handshake over plain TCP, and SMC-Rv2 explicitly supports non-local/routed peers, so the attacker need not share an L2 segment.
AC:LAttack complexity, low
The attacker controls both sides of the race: it chooses when to emit CDC messages and which connection to target (the alert token is disclosed to it in the CLC accept/confirm), and it can drive the teardown side itself via CDC close/abort flags into `smc_close_passive_work()` or a stale seqno with `failover_validation` into `abort_work` -> `smc_conn_kill()`. The race can be retried indefinitely across many connections in the same link group, and was reproduced in practice with soft-RoCE.
PR:NPrivileges required, none
Establishing an SMC-R connection to a listening SMC socket requires no credentials — SMC has no authentication layer of its own, exactly like plain TCP. Everything after connection establishment (sending CDC messages with a chosen token, signalling close/abort) is available to an unauthenticated peer.
UI:NUser interaction, none
The CDC message is processed in a softirq tasklet from the RDMA completion queue with no involvement from any local user or application action. No victim interaction of any kind is needed.
S:UScope, unchanged
The corruption is confined to kernel memory within the same security authority; there is no VM, IOMMU, or sandbox boundary crossed. This is a standard in-kernel use-after-free.
C:HConfidentiality impact, high
The freed `smc_sock` is reused type-safely as another live socket, so attacker-supplied cursors applied by `smc_cdc_msg_recv_action()` break the `0 <= bytes_to_rcv <= rmb_desc->len` invariant and make `smc_rx_recvmsg()` hand stale RMB contents to userspace, while the urgent-data path reads `*(base + conn->urg_curs.count - 1)` at an attacker-chosen offset. Once the slab page is reclaimed after the RCU grace period, reuse is unconstrained, giving a general kernel-memory read primitive.
I:HIntegrity impact, high
`sock_hold()` on an already-zero refcount followed by `sock_put()` re-enters `sk_free()`, yielding a double free, and the handler additionally writes `sk_err`/`sk_shutdown`, makes indirect calls through `sk_data_ready()`/`sk_write_space()`, and `queue_work()`s a freed `work_struct`. These are classic heap-spray and control-flow-hijack primitives leading to arbitrary kernel write.
A:HAvailability impact, high
The reported symptom is `refcount_warn_saturate` from `smc_cdc_msg_recv()`, which is an immediate kernel panic on `panic_on_warn` systems, and dereferencing freed memory in softirq context otherwise oopses. The attacker can trigger it repeatedly without authentication.