llamaCVEVulnerability Intelligence
/
Linux kernel

CVE-2026-99306

Linux kernel out-of-bounds read in Ceph MDS map handling

An attacker who can send a crafted metadata server map can trigger a memory-safety fault in the kernel.

Crítica · CVSS alto en tecnología listadaDefault setupsSin parche publicadoceph
Descargar PNG JSON
CVSS 3.1
9.8
Crítica
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H Fuente: CNA
ExplotaciónSin evidenciaNo figura en CISA KEV
Exploits públicos0Ninguno conocido
ParcheSin parcheEl fabricante no ha publicado corrección
Exposición3 dDesde la divulgación, 16 nov 2026, 4:59 a. m.
Detección8 minDesde la divulgación hasta su ingreso al índice
Timeline

Ciclo de vida de la vulnerabilidad

Hoy
Divulgación
Publicada
Modificada
Explotación
Remediación
Monitoreo
Detectada +2
Verificada
16 nov 26
19 nov 26
16 nov 26
17 nov 26
18 nov 26
19 nov 26
19 nov 26
FechaFaseEventoDetalle
Divulgación Publicada CVE-2026-99306 divulgada · CVSS 9.8 (CNA)
Monitoreo Detectada Ingresó al índice
Monitoreo Fuente: NIST NVD Registro visto por primera vez en NIST NVD
Monitoreo Fuente: ENISA EUVD Registro visto por primera vez en ENISA EUVD
Divulgación Modificada Última actualización del registro
Monitoreo Verificada Última verificación contra las fuentes
Resumen
Resumen generado por IA · CC-BY 4.0 · 16 nov 2026

The Linux kernel's Ceph filesystem client can read past the end of a stack buffer when handling a metadata server map from a monitor. An attacker who can send a crafted map claiming more ranks than the buffer holds could cause a memory-safety fault in the kernel.

Affects the Linux kernel ceph client (fs/ceph) when it processes MDS maps from a monitor, notably over the unauthenticated legacy msgr mode. The facts name no fixed version or patch.

¿Qué significa?The ceph client trusts the number of MDS ranks in a map from the monitor. If a monitor announces more ranks than the CEPH_MAX_MDS-sized bitmap holds, the kernel reads and writes past a stack bitmap. This can panic the ceph-msgr worker (denial of service) and corrupt kernel stack memory.
¿Quién está afectado?The facts list the Linux kernel ceph client (fs/ceph) without specific version ranges. The issue arises when it handles MDS maps from a monitor, and the connection is unauthenticated in legacy msgr mode. Readers can check whether their kernel has the ceph client (fs/ceph) in use and whether legacy msgr mode is used for monitor connections.
¿Qué hacer?The facts give no fixed version or patch id. The fix is described as rejecting MDS maps that claim more than CEPH_MAX_MDS ranks, so check the Linux kernel vendor or distribution for an update containing that change.
no-patchThe facts give no workaround. Limiting use of the ceph kernel client, particularly against monitors that cannot be trusted or over legacy unauthenticated connections, would reduce exposure to the affected code path.
Descripción original del registro (Linux)

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'

Contexto

Condiciones y alcance

Condiciones necesarias para que el fallo sea explotable
01
Linux kernel ceph client (fs/ceph) processing MDS maps received from a monitor“The map arrives on the monitor connection, which is unauthenticated in legacy msgr mode.”
02
The monitor announces more MDS ranks than the CEPH_MAX_MDS-sized bitmap holds“a monitor announcing more ranks than the bitmap holds makes test_bit() read past the end of targets[]”
¿Quién debe revisar?Teams running Linux kernels with the CephFS client (fs/ceph) mounting filesystems from Ceph monitors, especially over legacy msgr v1 connections.
Vector CVSS 3.1

Desglose de métricas

Opción resaltada = valor del vector · izquierda = peor caso · CNA
Vector de ataqueAV:N
RedAdyacenteLocalFísica
ComplejidadAC:L
BajaAlta
PrivilegiosPR:N
NingunoBajosAltos
InteracciónUI:N
NingunaRequerida
AlcanceS:U
CambiadoSin cambio
ConfidencialidadC:H
AltoBajoNinguno
IntegridadI:H
AltoBajoNinguno
DisponibilidadA:H
AltoBajoNinguno
Versiones

Versiones afectadas

5.15
Remediación

Estado del parche por paquete

Paquete / productoEstadoCorregida enDesde divulgaciónÚltima versión
Linux
Sin parche — — —
Explotación

Arsenal público

Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.

Registro
EscenarioAV:N - The rank count that overruns the bitmap is a field of the MDS map inside a CEPH_MSG_MDS_MAP on the monitor's TCP connection. ceph_con_v1_try_read() hands it to mon_dispatch(), extra_mon_dispatch() calls ceph_mdsc_handle_mdsmap(), and ceph_mdsmap_decode() stores m_max_mds, m_num_active_mds, or the "in"-set length in possible_max_rank. AC:L - That peer picks the epoch and those fields. The zeroed placeholder map from ceph_mdsc_init() makes the first CEPH_MSG_MDS_MAP call check_new_map(); a second map, after open_root_dentry() has registered the high rank, is what reaches clear_bit(). A few thousand m_info entries fit in a normal GFP_NOFS allocation, and legacy msgr is the default. PR:N - mon_con_ops has neither get_authorizer nor check_message_signature, so v1 process_connect() enters CEPH_CON_S_OPEN on TAG_READY without a peer authorizer, and mon_dispatch() accepts CEPH_MSG_MDS_MAP with no ceph_auth_is_authenticated() test. No client key and no local privilege are required. UI:N - The monitor pushes CEPH_MSG_MDS_MAP itself on the socket opened by ceph_monc_open_session(), and ceph_mdsc_handle_mdsmap() decodes it in the ceph-msgr workqueue. The session clear_bit() later uses is created by open_root_dentry() inside ceph_real_mount(), as part of that same mount. S:U - check_new_map() overruns the stack of the ceph-msgr worker in the host kernel that contains the ceph client. The corrupted frame and the impact stay in that same kernel. C:H - check_new_map() calls test_bit(i, targets) for every i below possible_max_rank. targets[] holds 2048 bits (32 unsigned longs), and m_info can be allocated large enough for the walk to continue through a few kilobytes of adjacent stack. Setting those ranks to CEPH_MDS_STATE_RECONNECT makes the client connect when the spilled bit is set. I:H - clear_bit(i, targets) runs for a session whose rank is below oldmap->possible_max_rank and mdsc->max_sessions once an address change sets CEPH_MDS_SESSION_RESTARTING and the new state is at least CEPH_MDS_STATE_RECONNECT. open_root_dentry() registers that rank through ceph_mdsmap_get_random_mds(), so the bit clear is past the 2048-bit array. A:H - clear_bit() writes the word just past targets[] in check_new_map(), the slot a stack-protected frame places next to that array. Flipping it panics the ceph-msgr worker in __stack_chk_fail when check_new_map() returns. The reported KASAN stack-out-of-bounds read is that same access on the worker.Linux
Referencias
Otrahttps://git.kernel.org/stable/c/7931e6798f419bda9ac8cf0c095323eff90a7f52
Otrahttps://git.kernel.org/stable/c/e9d72cf5d1bf5b4ddfe86707ada614973d75bf64
Otrahttps://git.kernel.org/stable/c/0574e4aff8e4b28e0d5636204ed4ba2350b8701a
Otrahttps://git.kernel.org/stable/c/c1742590d2b2fc5fb217e1a9b520108f1fe282e5

Las referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.