linux-stable.git/net/ceph, branch v7.2

libceph: Avoid using invalid osd indices from primary_temp

2026-08-12T19:21:41+00:00

A corrupted osdmap received from a Ceph monitor or OSD may contain osd
indices in its pg_temp, primary_temp, pg_upmap, and pg_upmap_items parts
that don't exist, i.e., that are greater than max_osd or smaller than
CEPH_HOMELESS_OSD (-1). These indices are used to create the up and
acting set in ceph_pg_to_up_acting_osds(), called from calc_target().
While most of these osd indices are checked, the one from primary_temp
is not. Subsequently, this may lead to calc_target() returning this
(potentially invalid) index as target osd for a (linger) request.
Because the osd_state, osd_weight, and osd_addr arrays only contain
max_osd entries (with indices 0 to max_osd -1), this leads to
out-of-bounds accesses when trying to read values from these arrays.

This patch fixes the issue by adding a check to get_temp_osds(), so that
only valid osd indices from primary_temp are used, and it falls back to
using the primary from pg_temp or the up set if it is invalid.

[ idryomov: changelog ]

Cc: stable@vger.kernel.org
Fixes: 5e8d4d36bf23 ("libceph: add support for primary_temp mappings")
Signed-off-by: Raphael Zimmer 
Reviewed-by: Ilya Dryomov 
Signed-off-by: Ilya Dryomov

libceph: fix OOB read in decode_watchers() via missing bounds check

2026-08-12T19:21:41+00:00

ceph_start_decoding() validates that struct_len bytes remain in the
buffer after the encoding header, but accepts struct_len=0 as valid:
ceph_decode_need(p, end, 0, bad) always passes. When a malicious or
compromised OSD sends an obj_list_watch_response_t reply with
struct_len=0, ceph_start_decoding() returns success with p == end,
leaving zero bytes guaranteed for subsequent reads.

The immediately following ceph_decode_32(p) in decode_watchers() has
no preceding bounds check. With p == end this is a 4-byte read past
the validated buffer boundary. The garbage value is then passed
directly to kzalloc_objs() as the watcher count.

The sibling function decode_watcher() already uses the safe variants
(ceph_decode_copy_safe, ceph_decode_64_safe, ceph_decode_skip_32)
after its own ceph_start_decoding() call. decode_watchers() is the
only site that uses the bare variant, confirming an oversight.

Fix by replacing ceph_decode_32(p) with ceph_decode_32_safe(p, end,
*num_watchers, bad), consistent with the established pattern.

Attacker model: a malicious or compromised OSD in a multi-tenant Ceph
deployment (e.g. cloud) can trigger this against any kernel client
that calls CEPH_OSD_OP_LIST_WATCHERS, without any further privileges
beyond OSD session establishment.

[ idryomov: trim changelog ]

Cc: stable@vger.kernel.org
Fixes: a4ed38d7a180 ("libceph: support for CEPH_OSD_OP_LIST_WATCHERS")
Signed-off-by: Pavitra Jha 
Reviewed-by: Viacheslav Dubeyko 
Signed-off-by: Ilya Dryomov

libceph: fix multiple unsafe decodes in decode_locker()

2026-08-12T19:21:41+00:00

decode_locker() in cls_lock_client.c contains three unsafe decode
operations that allow a malicious or compromised OSD to trigger
slab-out-of-bounds reads:

1. ceph_decode_copy() at the locker_id_t name field has no preceding
   bounds check. With p == end after ceph_start_decoding() accepts
   struct_len=0, this reads sizeof(ceph_entity_name) = 9 bytes past
   the validated buffer boundary.

2. *p += sizeof(struct ceph_timespec) after the locker_info_t header
   is an unchecked pointer advance. A malicious OSD can position p
   past end, causing all subsequent _safe checks to pass against a
   bogus boundary.

3. len = ceph_decode_32(p) has no preceding bounds check, and the
   immediately following *p += len is uncapped. A malicious OSD can
   send len=0xffffffff, advancing p gigabytes past end and escaping
   the decode window entirely.

Fix all three by replacing bare operations with their safe variants:
  ceph_decode_copy   -> ceph_decode_copy_safe
  *p += sizeof(...)  -> ceph_decode_skip_n
  ceph_decode_32(p)  -> ceph_decode_32_safe
  *p += len          -> ceph_decode_skip_n

A new label is added to return -EINVAL on any bounds violation.
-EINVAL is appropriate here: the data received from the OSD
is structurally malformed, which is an invalid argument to the decode
contract regardless of whether the caller or the wire is at fault.

Attacker model: a malicious or compromised OSD in a multi-tenant Ceph
deployment can trigger this against any kernel client that issues the
lock.get_info class method (e.g. during RBD exclusive lock acquisition)
without any further privileges beyond OSD session establishment.

[ idryomov: use ceph_decode_skip_string() to skip description, trim
  changelog ]

Cc: stable@vger.kernel.org
Fixes: d4ed4a530562 ("libceph: support for lock.lock_info")
Signed-off-by: Pavitra Jha 
Reviewed-by: Ilya Dryomov 
Signed-off-by: Ilya Dryomov

libceph: tolerate addrvecs with multiple entries of the same type

2026-08-12T19:20:54+00:00

ceph_decode_entity_addrvec() rejects any addrvec containing more than
one entry that matches the requested msgr type (LEGACY or MSGR2),
logging "another match of type N in addrvec" and returning -EINVAL.

Some admin tooling (e.g. pveceph mon create from Proxmox VE) generates
addrvecs with multiple same-type entries when public_network lists more
than one CIDR: it picks one local IP per subnet and emits both a v2 and
a v1 entry for each IP.  Monmaps shaped this way cause:

  libceph: mon0 (1)10.10.10.15:6789 session established
  libceph: another match of type 1 in addrvec
  libceph: problem decoding monmap, -22

No Ceph code uses the extra entries: since Nautilus, the userspace
messenger (AsyncMessenger) unconditionally picks the first address of
the requested type and ignores any subsequent matches.

Match that behavior: use the first matching entry and silently skip any
subsequent ones.  This is a compatibility fix for existing deployments
and does not enable dual-stack or multi-subnet address selection.

[ idryomov: tweak ceph_decode_entity_addrvec() comment ]

Cc: stable@vger.kernel.org
Fixes: a5cbd5fc22d5 ("libceph, ceph: get and handle cluster maps with addrvecs")
Link: https://bugzilla.proxmox.com/show_bug.cgi?id=7518
Signed-off-by: Kefu Chai 
Reviewed-by: Ilya Dryomov 
Signed-off-by: Ilya Dryomov

libceph: bound pg_{temp,upmap,upmap_items} length to CEPH_PG_MAX_SIZE

2026-07-23T18:29:42+00:00

__decode_pg_temp() decodes an user-controlled length but only rejects
values large enough to overflow the allocation; it does not bound it to
CEPH_PG_MAX_SIZE. The helper backs both pg_temp and pg_upmap decoding, and
apply_upmap()/get_temp_osds() later copy the decoded list into the fixed-size
on-stack array struct ceph_osds.osds[CEPH_PG_MAX_SIZE]. A monitor that sends
an OSDMap with a pg_temp/pg_upmap entry longer than 32 thus causes a stack
out-of-bounds write.

An OSD set for a single PG can never exceed CEPH_PG_MAX_SIZE, so reject longer
entries at decode time. The bound is well below the old overflow threshold, so
it also covers the allocation-size overflow the previous check guarded against.

  BUG: KASAN: stack-out-of-bounds in ceph_pg_to_up_acting_osds
  Write of size 4 ... by task exploit
   kasan_report (mm/kasan/report.c:595)
   ceph_pg_to_up_acting_osds (net/ceph/osdmap.c:2617 net/ceph/osdmap.c:2833)
   calc_target (net/ceph/osd_client.c:1638)
   __submit_request (net/ceph/osd_client.c:2394)
   ceph_osdc_start_request (net/ceph/osd_client.c:2490)
   ceph_osdc_call (net/ceph/osd_client.c:5164)
   rbd_dev_image_probe (drivers/block/rbd.c:6899)
   do_rbd_add (drivers/block/rbd.c:7138)
   ...
  kernel BUG at net/ceph/osdmap.c:2670!

[ idryomov: do the same in __decode_pg_upmap_items() ]

Cc: stable@vger.kernel.org
Fixes: a303bb0e5834 ("libceph: introduce and switch to decode_pg_mapping()")
Reported-by: Weiming Shi 
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Xiang Mei 
Reviewed-by: Alex Markuze 
Signed-off-by: Ilya Dryomov

libceph: refresh auth->authorizer_buf{,_len} after authorizer update

2026-07-23T18:29:41+00:00

ceph_x_create_authorizer() caches au->buf->vec.iov_base and
au->buf->vec.iov_len in struct ceph_auth_handshake.  These
cached values are then used by the messenger connect code when
sending the authorizer.

ceph_x_update_authorizer() can rebuild the authorizer when a newer
service ticket is available.  If the rebuilt authorizer no longer
fits in the existing buffer, ceph_x_build_authorizer() drops its
reference to au->buf and allocates a new one.  If this is the final
reference, ceph_buffer_put() frees the old ceph_buffer and its
vec.iov_base, but auth->authorizer_buf still points at that freed
memory.

A subsequent msgr1 reconnect can therefore queue the stale pointer
and trigger a KASAN slab-use-after-free in _copy_from_iter() while
tcp_sendmsg() copies the authorizer.

Refresh auth->authorizer_buf and auth->authorizer_buf_len after a
successful authorizer rebuild so the messenger sends the current
buffer.

Cc: stable@vger.kernel.org
Fixes: 0bed9b5c523d ("libceph: add update_authorizer auth method")
Closes: https://lore.kernel.org/all/E378850E-106C-427B-A241-970EB2D054D7@gmail.com/
Signed-off-by: Shuangpeng Bai 
Reviewed-by: Alex Markuze 
Signed-off-by: Ilya Dryomov

libceph: guard missing CRUSH type name lookup

2026-07-23T18:29:41+00:00

Localized read selection can walk a parent bucket whose name exists in
the CRUSH map while its type has no matching entry in type_names.
get_immediate_parent() then dereferences a NULL type_cn and passes an
invalid pointer into strcmp(), causing a null-ptr-deref.

Skip such malformed parent buckets unless both the bucket name and type
name metadata are present. This keeps malformed hierarchy data from
crashing locality lookup and safely falls back to "not local".

[ idryomov: add WARN_ON_ONCE ]

Cc: stable@vger.kernel.org
Fixes: 117d96a04f00 ("libceph: support for balanced and localized reads")
Reported-by: Yuan Tan 
Reported-by: Zhengchuan Liang 
Reported-by: Xin Liu 
Assisted-by: Codex:GPT-5.4
Signed-off-by: Zhao Zhang 
Signed-off-by: Ren Wei 
Reviewed-by: Viacheslav Dubeyko 
Signed-off-by: Ilya Dryomov

libceph: remove debugfs files before client teardown

2026-07-23T18:29:41+00:00

ceph_destroy_client() tears down the monitor client before removing
the per-client debugfs files. A concurrent read of the monmap debugfs
file can enter monmap_show() after ceph_monc_stop() has freed
monc->monmap, triggering a use-after-free.

Remove the debugfs files before stopping the OSD and monitor clients.
debugfs_remove() drains active handlers and prevents new accesses, so
the debugfs callbacks can no longer race the rest of client teardown.

Cc: stable@vger.kernel.org
Fixes: 76aa844d5b2f ("ceph: debugfs")
Reported-by: Yuan Tan 
Reported-by: Zhengchuan Liang 
Reported-by: Xin Liu 
Assisted-by: Codex:GPT-5.4
Signed-off-by: Douya Le 
Signed-off-by: Ren Wei 
Reviewed-by: Viacheslav Dubeyko 
Signed-off-by: Ilya Dryomov

libceph: bound get_version reply decode to front len

2026-07-23T18:29:41+00:00

handle_get_version_reply() uses msg->front_alloc_len as the decode
boundary for MON_GET_VERSION_REPLY.  That is the size of the reused
reply buffer, not the number of bytes actually received.

A truncated reply can therefore pass ceph_decode_need() and decode the
second u64 from stale tail bytes left in the buffer by an earlier
message, causing an uninitialized memory read.

Use msg->front.iov_len as the receive-side decode boundary, matching
other libceph reply handlers and limiting decoding to the bytes that
were actually read from the wire.

Cc: stable@vger.kernel.org
Fixes: 513a8243d67f ("libceph: mon_get_version request infrastructure")
Reported-by: Yuan Tan 
Reported-by: Zhengchuan Liang 
Reported-by: Xin Liu 
Assisted-by: Codex:GPT-5.4
Signed-off-by: Douya Le 
Signed-off-by: Ren Wei 
Reviewed-by: Viacheslav Dubeyko 
Signed-off-by: Ilya Dryomov

libceph: fix two unsafe bare decodes in decode_lockers()

2026-07-23T18:29:41+00:00

decode_lockers() in cls_lock_client.c contains two bare decode operations
that allow a malicious or compromised OSD to trigger slab-out-of-bounds
reads:

1. ceph_decode_32(p) at the num_lockers field has no preceding bounds
   check. ceph_start_decoding() accepts struct_len=0 as valid -- the
   internal ceph_decode_need(p, end, 0, bad) always passes -- so when an
   OSD sends struct_len=0, ceph_start_decoding() returns success with
   p == end. The immediately following bare ceph_decode_32(p) then reads
   4 bytes past the validated buffer boundary. The garbage value is
   passed directly to kzalloc_objs() as the locker count.

   The sibling function decode_watchers() in osd_client.c already uses
   ceph_decode_32_safe() after its own ceph_start_decoding() call.
   decode_lockers() was the only site using the bare variant.

2. ceph_decode_8(p) after the decode_locker() loop has no preceding
   bounds check. If an OSD crafts num_lockers such that the loop
   advances p exactly to end, the subsequent bare ceph_decode_8(p) reads
   one byte past the validated buffer boundary. The result is passed
   directly into *type, which is used as a lock type discriminator by
   callers, giving an OSD-controlled one-byte OOB read with direct
   influence over the lock type field.

Fix both by replacing bare operations with their safe variants:
  ceph_decode_32(p) -> ceph_decode_32_safe(p, end, *num_lockers,
                                           err_inval)
  ceph_decode_8(p)  -> ceph_decode_8_safe(p, end, *type,
                                          err_free_lockers)

The goto targets differ intentionally:
  err_inval: is a new label returning -EINVAL directly. It is used for
  the pre-allocation failure path where *lockers is not yet allocated
  and must not be passed to ceph_free_lockers().

  err_free_lockers: is the existing label. It is used for the
  post-allocation failure path where *lockers is allocated and must
  be freed.

ret is set to -EINVAL before ceph_decode_8_safe() so that
err_free_lockers returns the correct error code on bounds violation.
Without this, err_free_lockers would return a stale ret value (0 from
the successful decode_locker() loop), silently swallowing the error.

-EINVAL is correct for both failure paths. The data received from the
OSD is structurally malformed. -ENOMEM would misrepresent the failure
class to callers and to stable@ backporters triaging error paths.

Attacker model: a malicious or compromised OSD in a multi-tenant Ceph
deployment can trigger this against any kernel client that issues the
lock.get_info class method (e.g. during RBD exclusive lock acquisition).

[ idryomov: trim changelog, formatting ]

Cc: stable@vger.kernel.org
Fixes: d4ed4a530562 ("libceph: support for lock.lock_info")
Signed-off-by: Pavitra Jha 
Reviewed-by: Viacheslav Dubeyko 
Signed-off-by: Ilya Dryomov