CVE-2026-89650

In the Linux kernel, the following vulnerability has been resolved:

ceph: bound num export targets array for mds info v2/v3

ceph mdsmap decode() in fs/ceph/mdsmap.c reads num export targets from each per-mds info record and advances the decode cursor by num export targets * sizeof(u32) without first checking that many bytes remain. The only upper-bound check that catches a runaway cursor (*p > info end) is gated on info v >= 4, because info end is left NULL for info v 2 and 3. When the monitor sends an MDS map whose per-mds info version is 2 or 3 with an oversized num export targets, the cursor moves past the message front buffer and the later export-targets loop calls the unchecked ceph decode 32() on out-of-bounds memory.

A kernel client processes CEPH MSG MDS MAP from its monitor session (net/ceph/mon client.c dispatches it; fs/ceph/super.c routes it to ceph mdsc handle mdsmap(), which sets end to the front buffer bound and calls ceph mdsmap decode()). A malicious or compromised monitor, or an on-path attacker on an unsigned/unencrypted messenger session, can therefore drive an out-of-bounds read in the client kernel; on x86 64 with KASAN it is reported as a slab-out-of-bounds read in ceph mdsmap decode(). The decoded values land in the internal info->export targets[] array, so the consequence is a kernel out-of-bounds read, not an information leak to the attacker.

Impact: a malicious or compromised Ceph monitor sending an MDS map with a per-mds info version of 2 or 3 and an oversized num export targets field triggers an out-of-bounds read in the CephFS client kernel.

Add a ceph decode need() for the export-targets array before advancing the cursor, so the bound is enforced for every info v >= 2, not only info v >= 4. This mirrors the count-then-need idiom already used for m data pg pools later in the same function.

Compute the export-targets byte count with size mul() and reuse that checked length when advancing the cursor, so the attacker-controlled num export targets multiplication fails closed on overflow rather than relying on the later kcalloc() guard.