[PATCH bpf v2] bpf: Add nelems overflow check in btf_find_field_one()

Himanshu Anand posted 1 patch 1 day, 10 hours ago
kernel/bpf/btf.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
[PATCH bpf v2] bpf: Add nelems overflow check in btf_find_field_one()
Posted by Himanshu Anand 1 day, 10 hours ago
btf_find_field_one() multiplies a u32 nelems accumulator by each
nested array level's element count without checking for overflow.
The sibling function __btf_resolve_size() already guards against
the same overflow pattern (btf.c line 2110).

Currently the BTF array verifier (btf_array_resolve) rejects BTF
blobs whose total array size would overflow u32, so this code path
is not reachable with crafted BTF input on kernels that include
that check.  Add check_mul_overflow() anyway to keep
btf_find_field_one() self-consistent with __btf_resolve_size()
and to guard against future changes in the validation ordering.

Fixes: 994796c0256c ("bpf: create repeated fields for arrays.")
Signed-off-by: Himanshu Anand <anand.himanshu17@gmail.com>
---
 kernel/bpf/btf.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index a62d78581207..b767a9fcf095 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -3765,7 +3765,8 @@ static int btf_find_field_one(const struct btf *btf,
 	 */
 	for (i = 0; i < MAX_RESOLVE_DEPTH && btf_type_is_array(var_type); i++) {
 		array = btf_array(var_type);
-		nelems *= array->nelems;
+		if (check_mul_overflow(nelems, array->nelems, &nelems))
+			return -E2BIG;
 		var_type = btf_type_by_id(btf, array->type);
 	}
 	if (i == MAX_RESOLVE_DEPTH)
-- 
2.34.1
Re: [PATCH bpf v2] bpf: Add nelems overflow check in btf_find_field_one()
Posted by Alexei Starovoitov 1 day, 10 hours ago
On Sat, May 23, 2026 at 9:47 AM Himanshu Anand
<anand.himanshu17@gmail.com> wrote:
>
> btf_find_field_one() multiplies a u32 nelems accumulator by each
> nested array level's element count without checking for overflow.
> The sibling function __btf_resolve_size() already guards against
> the same overflow pattern (btf.c line 2110).
>
> Currently the BTF array verifier (btf_array_resolve) rejects BTF
> blobs whose total array size would overflow u32, so this code path
> is not reachable with crafted BTF input on kernels that include
> that check.  Add check_mul_overflow() anyway to keep
> btf_find_field_one() self-consistent with __btf_resolve_size()
> and to guard against future changes in the validation ordering.
>
> Fixes: 994796c0256c ("bpf: create repeated fields for arrays.")
> Signed-off-by: Himanshu Anand <anand.himanshu17@gmail.com>
> ---
>  kernel/bpf/btf.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> index a62d78581207..b767a9fcf095 100644
> --- a/kernel/bpf/btf.c
> +++ b/kernel/bpf/btf.c
> @@ -3765,7 +3765,8 @@ static int btf_find_field_one(const struct btf *btf,
>          */
>         for (i = 0; i < MAX_RESOLVE_DEPTH && btf_type_is_array(var_type); i++) {
>                 array = btf_array(var_type);
> -               nelems *= array->nelems;
> +               if (check_mul_overflow(nelems, array->nelems, &nelems))
> +                       return -E2BIG;

It's unnecessary.
see __btf_resolve_size() and btf_array_resolve().

pw-bot: cr