[PATCH net-next] selftests/net/ipsec: Fix variable size type not at the end of struct

Ankit Khushwaha posted 1 patch 5 days, 17 hours ago
tools/testing/selftests/net/ipsec.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
[PATCH net-next] selftests/net/ipsec: Fix variable size type not at the end of struct
Posted by Ankit Khushwaha 5 days, 17 hours ago
The "struct alg" object contains a union of 3 xfrm structures:

	union {
		struct xfrm_algo;
		struct xfrm_algo_aead;
		struct xfrm_algo_auth;
	}

All of them end with a flexible array member used to store key material,
but the flexible array appears at *different offsets* in each struct.
bcz of this, union itself is of variable-sized & Placing it above
char buf[...] triggers:

ipsec.c:835:5: warning: field 'u' with variable sized type 'union
(unnamed union at ipsec.c:831:3)' not at the end of a struct or class
is a GNU extension [-Wgnu-variable-sized-type-not-at-end]
  835 |                 } u;
      |                   ^

one fix is to use "TRAILING_OVERLAP()" which works with one flexible
array member only.

But In "struct alg" flexible array member exists in all union members,
but not at the same offset, so TRAILING_OVERLAP cannot be applied.

so the fix is to explicitly overlay the key buffer at the correct offset
for the largest union member (xfrm_algo_auth). This ensures that the
flexible-array region and the fixed buffer line up.

No functional change.

Signed-off-by: Ankit Khushwaha <ankitkhushwaha.linux@gmail.com>
---
 tools/testing/selftests/net/ipsec.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/net/ipsec.c b/tools/testing/selftests/net/ipsec.c
index 0ccf484b1d9d..f4afef51b930 100644
--- a/tools/testing/selftests/net/ipsec.c
+++ b/tools/testing/selftests/net/ipsec.c
@@ -43,6 +43,10 @@

 #define BUILD_BUG_ON(condition) ((void)sizeof(char[1 - 2*!!(condition)]))

+#ifndef offsetof
+#define offsetof(TYPE, MEMBER)	__builtin_offsetof(TYPE, MEMBER)
+#endif
+
 #define IPV4_STR_SZ	16	/* xxx.xxx.xxx.xxx is longest + \0 */
 #define MAX_PAYLOAD	2048
 #define XFRM_ALGO_KEY_BUF_SIZE	512
@@ -827,13 +831,16 @@ static int xfrm_fill_key(char *name, char *buf,
 static int xfrm_state_pack_algo(struct nlmsghdr *nh, size_t req_sz,
 		struct xfrm_desc *desc)
 {
-	struct {
+	union {
 		union {
 			struct xfrm_algo	alg;
 			struct xfrm_algo_aead	aead;
 			struct xfrm_algo_auth	auth;
 		} u;
-		char buf[XFRM_ALGO_KEY_BUF_SIZE];
+		struct {
+			unsigned char __offset_to_FAM[offsetof(struct xfrm_algo_auth, alg_key)];
+			char buf[XFRM_ALGO_KEY_BUF_SIZE];
+		};
 	} alg = {};
 	size_t alen, elen, clen, aelen;
 	unsigned short type;
--
2.52.0
Re: [PATCH net-next] selftests/net/ipsec: Fix variable size type not at the end of struct
Posted by Simon Horman 4 days, 4 hours ago
On Wed, Nov 26, 2025 at 11:17:11AM +0530, Ankit Khushwaha wrote:
> The "struct alg" object contains a union of 3 xfrm structures:
> 
> 	union {
> 		struct xfrm_algo;
> 		struct xfrm_algo_aead;
> 		struct xfrm_algo_auth;
> 	}
> 
> All of them end with a flexible array member used to store key material,
> but the flexible array appears at *different offsets* in each struct.
> bcz of this, union itself is of variable-sized & Placing it above
> char buf[...] triggers:
> 
> ipsec.c:835:5: warning: field 'u' with variable sized type 'union
> (unnamed union at ipsec.c:831:3)' not at the end of a struct or class
> is a GNU extension [-Wgnu-variable-sized-type-not-at-end]
>   835 |                 } u;
>       |                   ^

FWIIW, I was able to reproduce this using the
-Wflex-array-member-not-at-end (n.b. different option name to above)
option with the GCC 15.2.0 toolchain here [1].

[1] https://mirrors.edge.kernel.org/pub/tools/crosstool/

> 
> one fix is to use "TRAILING_OVERLAP()" which works with one flexible
> array member only.
> 
> But In "struct alg" flexible array member exists in all union members,
> but not at the same offset, so TRAILING_OVERLAP cannot be applied.
> 
> so the fix is to explicitly overlay the key buffer at the correct offset
> for the largest union member (xfrm_algo_auth). This ensures that the
> flexible-array region and the fixed buffer line up.
> 
> No functional change.

I verified this does not change the layout of the union (now structure),
by copying it outside the function and giving it a name.
And then analysing the binary using pahole.

So I agree this should not have any run-time effect.

But I've CCed Gustavo and linux-hardening, as their experience
seems relevant here.

> 
> Signed-off-by: Ankit Khushwaha <ankitkhushwaha.linux@gmail.com>

Reviewed-by: Simon Horman <horms@kernel.org>

> ---
>  tools/testing/selftests/net/ipsec.c | 11 +++++++++--
>  1 file changed, 9 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/testing/selftests/net/ipsec.c b/tools/testing/selftests/net/ipsec.c
> index 0ccf484b1d9d..f4afef51b930 100644
> --- a/tools/testing/selftests/net/ipsec.c
> +++ b/tools/testing/selftests/net/ipsec.c
> @@ -43,6 +43,10 @@
> 
>  #define BUILD_BUG_ON(condition) ((void)sizeof(char[1 - 2*!!(condition)]))
> 
> +#ifndef offsetof
> +#define offsetof(TYPE, MEMBER)	__builtin_offsetof(TYPE, MEMBER)
> +#endif
> +
>  #define IPV4_STR_SZ	16	/* xxx.xxx.xxx.xxx is longest + \0 */
>  #define MAX_PAYLOAD	2048
>  #define XFRM_ALGO_KEY_BUF_SIZE	512
> @@ -827,13 +831,16 @@ static int xfrm_fill_key(char *name, char *buf,
>  static int xfrm_state_pack_algo(struct nlmsghdr *nh, size_t req_sz,
>  		struct xfrm_desc *desc)
>  {
> -	struct {
> +	union {
>  		union {
>  			struct xfrm_algo	alg;
>  			struct xfrm_algo_aead	aead;
>  			struct xfrm_algo_auth	auth;
>  		} u;
> -		char buf[XFRM_ALGO_KEY_BUF_SIZE];
> +		struct {
> +			unsigned char __offset_to_FAM[offsetof(struct xfrm_algo_auth, alg_key)];
> +			char buf[XFRM_ALGO_KEY_BUF_SIZE];
> +		};
>  	} alg = {};
>  	size_t alen, elen, clen, aelen;
>  	unsigned short type;