[PATCH bpf-next v2 1/2] bpf: Reject offloaded maps as map-in-map inner maps
From: Shihuang Liu
Date: Wed Nov 18 2026 - 11:16:58 EST
Offloaded ARRAY/HASH maps share map_type with the software maps but use
bpf_map_offload_ops and a different object layout. bpf_map_meta_equal()
compares map_type, key/value size, flags and the btf record only, so an
offloaded map can be inserted into an outer map created from a software
template. The verifier and JIT then use the software layout on the
offloaded object: an ARRAY lookup writes to bpf_offloaded_map.dev_priv
and a HASH lookup reads bpf_htab fields past the 320-byte allocation.
An offloaded map used as inner_map_fd is accepted too and leads to a
NULL map_lookup_elem call.
Drop map_meta_equal from bpf_map_offload_ops so the existing check
rejects offloaded templates, and compare ops in bpf_map_meta_equal() so
an offloaded candidate no longer matches a software template.
Fixes: a38845729ea398 ("bpf: offload: add map offload infrastructure")
Link: https://lore.kernel.org/all/20261112105620.57520-1-shlomojune6@xxxxxxxxx/
Signed-off-by: Shihuang Liu <shlomojune6@xxxxxxxxx>
---
kernel/bpf/map_in_map.c | 7 ++++---
kernel/bpf/syscall.c | 1 -
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/bpf/map_in_map.c b/kernel/bpf/map_in_map.c
index 0fa3f2ac0f992f..9ec3bbfb5d4e3c 100644
--- a/kernel/bpf/map_in_map.c
+++ b/kernel/bpf/map_in_map.c
@@ -38,6 +38,7 @@ struct bpf_map *bpf_map_meta_alloc(int inner_map_ufd)
if (!inner_map_meta)
return ERR_PTR(-ENOMEM);
+ inner_map_meta->ops = inner_map->ops;
inner_map_meta->map_type = inner_map->map_type;
inner_map_meta->key_size = inner_map->key_size;
inner_map_meta->value_size = inner_map->value_size;
@@ -65,7 +66,6 @@ struct bpf_map *bpf_map_meta_alloc(int inner_map_ufd)
}
/* Misc members not needed in bpf_map_meta_equal() check. */
- inner_map_meta->ops = inner_map->ops;
if (inner_map->ops == &array_map_ops || inner_map->ops == &percpu_array_map_ops) {
struct bpf_array *inner_array_meta =
container_of(inner_map_meta, struct bpf_array, map);
@@ -88,8 +88,9 @@ void bpf_map_meta_free(struct bpf_map *map_meta)
bool bpf_map_meta_equal(const struct bpf_map *meta0,
const struct bpf_map *meta1)
{
- /* No need to compare ops because it is covered by map_type */
- return meta0->map_type == meta1->map_type &&
+ /* An offloaded map has the map_type of a host map and its own ops */
+ return meta0->ops == meta1->ops &&
+ meta0->map_type == meta1->map_type &&
meta0->key_size == meta1->key_size &&
meta0->value_size == meta1->value_size &&
meta0->map_flags == meta1->map_flags &&
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 60c4d84b9aba6f..0cae12c3f70849 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -112,7 +112,6 @@ int bpf_check_uarg_tail_zero(bpfptr_t uaddr,
}
const struct bpf_map_ops bpf_map_offload_ops = {
- .map_meta_equal = bpf_map_meta_equal,
.map_alloc = bpf_map_offload_map_alloc,
.map_free = bpf_map_offload_map_free,
.map_check_btf = map_check_no_btf,
--
2.43.0