CVE-2026-72464
- Source
- https://cve.org/CVERecord?id=CVE-2026-72464
- Import Source
- https://storage.googleapis.com/cve-osv-conversion/osv-output/CVE-2026-72464.json
- JSON Data
- https://api.osv.dev/v1/vulns/CVE-2026-72464
- Downstream
- Published
- 2026-08-15T05:57:08.080Z
- Modified
- 2026-08-18T03:56:56.517250695Z
- Severity
-
- 7.5 (High) CVSS_V3 - CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H CVSS Calculator
- Summary
- xprtrdma: Repost Receive buffers for malformed replies
- Details
-
In the Linux kernel, the following vulnerability has been resolved:
xprtrdma: Repost Receive buffers for malformed replies
rpcrdmawcreceive() decrements the transport's Receive count for every completion before it dispatches a successful Receive to rpcrdmareplyhandler(). The handler must post a replacement Receive WR before returning unless ownership of the rep has moved elsewhere, as on the backchannel path.
Commit 2ae50ad68cd7 ("xprtrdma: Close window between waking RPC senders and posting Receives") moved the Receive refill out of rpcrdmawcreceive(), where it had run ahead of every reply, into rpcrdmareplyhandler() so that the responder's credit grant could be parsed before reposting. The bad-version and short-reply exits never reach that refill: they recycle the rep and return without calling rpcrdmapostrecvs().
A remote peer can therefore drain the client's posted Receive queue by sending a sustained stream of replies that are shorter than the fixed transport header or that carry an unrecognized RPC/RDMA version. Each such reply consumes one posted Receive without replacing it. Once the queue empties, the peer's next Send finds no posted Receive and the transport stalls until reconnect.
Route both malformed-reply exits through the shared repost tail after recycling the rep, refilling against buf->rbcredits, the most recent accepted credit grant. Neither exit updates the congestion window, so RPCs admitted under the previous grant remain in flight awaiting replies. A smaller refill target would let a stream of malformed replies ratchet the posted Receive count down to the batch floor while the congestion window still admits rbcredits RPCs; a burst of valid replies to those RPCs could then overrun the posted Receives, and because the client connects with rnrretrycount of zero, a single RNR NAK terminates the connection. Refilling against rbcredits also restores the target that applied to malformed replies before commit 2ae50ad68cd7 ("xprtrdma: Close window between waking RPC senders and posting Receives") when rpcrdmapostrecvs() computed it from rbcredits internally. rb_credits is at least one from connection establishment onward, so the repost path always keeps Receives posted.
- Database specific
{ "cna_assigner": "Linux", "osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/72xxx/CVE-2026-72464.json" }- References
-
- https://git.kernel.org/stable/c/007b4da2f38dcc16a13265416f4ca9f179bab610
- https://git.kernel.org/stable/c/19fae02b272ee4bcdfb5db57f402d28f1697167a
- https://git.kernel.org/stable/c/4322fd9645ee769ad29ce5caea74a1cd9b17269d
- https://git.kernel.org/stable/c/abc011ddaf1617e3e82d8a1e87daa7ddbfb9bac5
- https://git.kernel.org/stable/c/d7c531ab477ae94fd03771d707fd29c787408039
- https://git.kernel.org/stable/c/ef6fb8a5c521f1a07f85202d13e8f2898f247362
- https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/72xxx/CVE-2026-72464.json
- https://nvd.nist.gov/vuln/detail/CVE-2026-72464
- https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
Affected packages
Git / git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
Affected ranges
- Type
- GIT
- Repo
- https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
- Events
-
Introduced2ae50ad68cd79224198b525f7bd645c9da98b6ffFixedef6fb8a5c521f1a07f85202d13e8f2898f247362Fixed4322fd9645ee769ad29ce5caea74a1cd9b17269dFixed19fae02b272ee4bcdfb5db57f402d28f1697167aFixed007b4da2f38dcc16a13265416f4ca9f179bab610Fixedd7c531ab477ae94fd03771d707fd29c787408039Fixedabc011ddaf1617e3e82d8a1e87daa7ddbfb9bac5
- Type
- GIT
- Repo
- https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
- Events
-
Introduced0 Unknown introduced commit / All previous commits are affectedLast affected3791c5982ba1eebf2900ee7ca7b9a89619c26d54
- Type
- GIT
- Repo
- https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
- Events
-
Introduced5.4.13Fixed5.5