[PATCH] libceph: reject mismatched CRUSH weight-set sizes

From: Felix Hoffmann

Date: Thu Sep 10 2026 - 07:48:39 EST


A CEPH_MSG_OSD_MAP message can contain CRUSH choose args with replacement
weight vectors for straw2 buckets. decode_choose_args() validates the
optional IDs vector against the referenced bucket, but does not validate
the size of any replacement-weight vector.

The CRUSH mapper assumes that each replacement vector has exactly as many
entries as the bucket. A zero-length vector for a nonempty bucket makes
bucket_straw2_choose() dereference NULL, while a shorter nonzero vector
causes an out-of-bounds read. An authenticated monitor or OSD can deliver
such a map and crash a kernel client when it performs placement for an
ordinary request.

Reject a nonempty collection of weight sets for an absent bucket and
require every replacement-weight vector to match its bucket size.

Fixes: 5cf9c4a9959b ("libceph, crush: per-pool crush_choose_arg_map for crush_do_rule()")
Cc: stable@xxxxxxxxxxxxxxx
Assisted-by: Codex:gpt-5
Signed-off-by: Felix Hoffmann <f3lix.dev@xxxxxx>
---
Tested on x86-64 with KASAN enabled:

- a KUnit reproducer containing a complete serialized CRUSH map with a
two-item straw2 bucket and a zero-length replacement-weight vector
panicked in crush_bucket_choose() through crush_do_rule()
- with this change, the same map is rejected with -EINVAL and the KUnit
regression passes without a KASAN report
- a valid replacement vector of the required size maps successfully
- the KASAN-enabled x86-64 kernel builds successfully

The reproducer and complete console logs are available privately on
request.

net/ceph/osdmap.c | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/net/ceph/osdmap.c b/net/ceph/osdmap.c
index 558c7c3a4a8..51833dba999 100644
--- a/net/ceph/osdmap.c
+++ b/net/ceph/osdmap.c
@@ -376,7 +376,9 @@ static int decode_choose_args(void **p, void *end, struct crush_map *c)
ceph_decode_32_safe(p, end, num_buckets, e_inval);
while (num_buckets--) {
struct crush_choose_arg *arg;
+ struct crush_bucket *bucket;
u32 bucket_index;
+ u32 i;

ceph_decode_32_safe(p, end, bucket_index, e_inval);
if (bucket_index >= arg_map->size)
@@ -387,10 +389,18 @@ static int decode_choose_args(void **p, void *end, struct crush_map *c)
if (ret)
goto fail;

+ bucket = c->buckets[bucket_index];
if (arg->ids_size &&
- (!c->buckets[bucket_index] ||
- arg->ids_size != c->buckets[bucket_index]->size))
+ (!bucket || arg->ids_size != bucket->size))
goto e_inval;
+
+ if (arg->weight_set_size && !bucket)
+ goto e_inval;
+
+ for (i = 0; i < arg->weight_set_size; i++) {
+ if (arg->weight_set[i].size != bucket->size)
+ goto e_inval;
+ }
}

if (!__insert_choose_arg_map(&c->choose_args, arg_map)) {
--
2.43.0