---
id: "CVE-2026-99306"
page: "https://infrastack.ai/live/CVE-2026-99306"
kind: "vulnerability"
aliases: ["CVE-2026-99306"]
title: "Linux kernel Ceph client: malformed MDS map reads past a stack buffer"
severity: "high"
cvss: null
epss: 0.00209
exploited: false
status: "open"
category: "runtime"
products: [{"name": "linux-kernel", "fixed": ["6.18.63"], "affected": "kernels with the Ceph client containing d517b3983dd3 and without the fix; < 6.18.63 on 6.18", "ecosystem": "linux"}]
ecosystems: ["linux"]
tags: ["ceph", "cephfs", "linux-kernel", "runtime", "self-hosted"]
not_affected_if: ["no-cephfs-mount", "userspace-ceph-client"]
published_at: "2026-11-16T10:17:05+00:00"
first_seen_at: "2026-11-16T10:39:27+00:00"
updated_at: "2026-11-16T18:53:18+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-99306", "first_seen_at": "2026-11-16T10:39:27+00:00"}, {"source": "analyst", "url": "https://lore.landau.one/lkml/20260915051924.737109-1-xmei5@asu.edu/T/", "first_seen_at": "2026-11-16T10:40:59+00:00", "cites": ["https://lore.landau.one/lkml/20260915051924.737109-1-xmei5@asu.edu/T/", "https://lore.landau.one/lkml/20261108-rc-reply-03-xfs-6.18.63-rc1@kernel.org/t/"]}, {"source": "epss", "url": "https://api.first.org/data/v1/epss?cve=CVE-2026-99306", "first_seen_at": "2026-11-16T18:53:18+00:00"}]
exploit_sources: [{"type": "epss", "score": 0.00209, "url": "https://api.first.org/data/v1/epss?cve=CVE-2026-99306"}]
---

# Linux kernel Ceph client: malformed MDS map reads past a stack buffer

Link for people: https://infrastack.ai/live/CVE-2026-99306 (this .md is for agents)

A Ceph monitor can send an MDS map claiming more ranks than the kernel's Ceph client supports, and the client reads past the end of a stack array while checking it. An attacker who can send messages to the monitor connection, which is unauthenticated in legacy messenger mode, can crash the kernel or possibly read adjacent kernel memory.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- linux:linux-kernel affected kernels with the Ceph client containing d517b3983dd3 and without the fix; < 6.18.63 on 6.18 fixed 6.18.63

## How to check

- Run uname -r; on 6.18 it is fixed from 6.18.63. For other series, check the distro kernel changelog for 'reject mdsmaps that claim more than CEPH_MAX_MDS ranks'.
- Check whether the ceph module is loaded: lsmod | grep ceph.
- Check for CephFS kernel mounts: mount -t ceph. No CephFS kernel mount means you can skip it.

## Fix

Update to a kernel carrying the patch 'ceph: reject mdsmaps that claim more than CEPH_MAX_MDS ranks': 6.18.63 or later on the 6.18 series (its rc1 shortlog lists it). Other stable branches and the mainline commit id not confirmed; check your distro's changelog for this CVE.

## Timeline

- 2026-11-16T10:39:27+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-99306
- 2026-11-16T10:40:59+00:00 analyst https://lore.landau.one/lkml/20260915051924.737109-1-xmei5@asu.edu/T/ — Fix patch by Xiang Mei (2026-09-15, Cc stable, Fixes d517b3983dd3) adds CEPH_MAX_MDS check in ceph_mdsmap_decode(); listed in 6.18.63-rc1 shortlog, 6.18.63 released 2026-11-10. Upstream commit hash and other stable branches (6.12.117 changelog truncated) not confirmed; not marked verified. Sibling CVE-2026-89653 is a different bug (export_targets ranks), not merged.
- 2026-11-16T18:53:18+00:00 epss https://api.first.org/data/v1/epss?cve=CVE-2026-99306

## Documents

- [fix] [PATCH] ceph: reject mdsmaps that claim more than CEPH_MAX_MDS ranks (Xiang Mei, 2026-09-15) https://lore.landau.one/lkml/20260915051924.737109-1-xmei5@asu.edu/T/
- [fix] 6.18.63-rc1 review (shortlog lists the ceph fix) https://lore.landau.one/lkml/20261108-rc-reply-03-xfs-6.18.63-rc1@kernel.org/t/

## From the sources

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

ceph: reject mdsmaps that claim more than CEPH_MAX_MDS ranks

check_new_map() sizes its export-target bitmap from CEPH_MAX_MDS but walks
it with newmap->possible_max_rank, which comes straight off the wire:
ceph_mdsmap_decode() sets it to max(m_num_active_mds, m_max_mds), then
overwrites it with the size of the "in" set.  Neither is bounded by
CEPH_MAX_MDS, only by m_info allocating, so a monitor announcing more ranks
than the bitmap holds makes test_bit() read past the end of targets[].  The
clear_bit() loop above indexes the same bitmap, though it is additionally
bounded by mdsc->max_sessions.  The map arrives on the monitor connection,
which is unauthenticated in legacy msgr mode.

CEPH_MAX_MDS is a protocol constant from src/mds/mdstypes.h, and the
reserved inode ranges in super.h are laid out one rank apart over that
span, so a map claiming more ranks is corrupt.  Reject it where the rank
space is established rather than letting the count reach check_new_map().

  BUG: KASAN: stack-out-of-bounds in check_new_map (fs/ceph/mds_client.c:5968)
  Read of size 8 at addr ffffc90000ac7798 by task kworker/1:2/127
  Workqueue: ceph-msgr ceph_con_workfn
  Call Trace:
   <TASK>
   dump_stack_lvl (lib/dump_stack.c:94 lib/dump_stack.c:120)
   print_report (mm/kasan/report.c:378 mm/kasan/report.c:482)
   kasan_report (mm/kasan/report.c:595)
   kasan_check_range (mm/kasan/generic.c:186 mm/kasan/generic.c:200)
   check_new_map (fs/ceph/mds_client.c:5968)
   ceph_mdsc_handle_mdsmap (fs/ceph/mds_client.c:7060)
   extra_mon_dispatch (fs/ceph/super.c:802)
   mon_dispatch (net/ceph/mon_client.c:1483)
   ceph_con_process_message (net/ceph/messenger.c:1424)
   ceph_con_v1_try_read (net/ceph/messenger_v1.c:1430)
   ceph_con_workfn (net/ceph/messenger.c:1576)
   process_one_work (kernel/workqueue.c:3396)
   worker_thread (kernel/workqueue.c:3479 kernel/workqueue.c:3560)
   kthread (kernel/kthread.c:436)
   ret_from_fork (arch/x86/kernel/process.c:158)
   ret_from_fork_asm (arch/x86/entry/entry_64.S:245)
   </TASK>
  The buggy address belongs to stack of task kworker/1:2/127
   and is located at offset 288 in frame:
   check_new_map (fs/ceph/mds_client.c:5846)
  This frame has 1 object:
   [32, 288) 'targets'

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_
