diff mbox series

crypto: chacha20poly1305 - Annotate struct chachapoly_ctx with __counted_by()

Message ID 20240805175237.63098-2-thorsten.blum@toblux.com (mailing list archive)
State Superseded
Delegated to: Herbert Xu
Headers show
Series crypto: chacha20poly1305 - Annotate struct chachapoly_ctx with __counted_by() | expand

Commit Message

Thorsten Blum Aug. 5, 2024, 5:52 p.m. UTC
Add the __counted_by compiler attribute to the flexible array member
salt to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
CONFIG_FORTIFY_SOURCE.

Use struct_size_t() instead of manually calculating the struct's size.

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
---
 crypto/chacha20poly1305.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Comments

Eric Biggers Aug. 5, 2024, 5:59 p.m. UTC | #1
On Mon, Aug 05, 2024 at 07:52:38PM +0200, Thorsten Blum wrote:
>  struct poly_req {
> @@ -611,8 +611,8 @@ static int chachapoly_create(struct crypto_template *tmpl, struct rtattr **tb,
>  				       poly->base.cra_priority) / 2;
>  	inst->alg.base.cra_blocksize = 1;
>  	inst->alg.base.cra_alignmask = chacha->base.cra_alignmask;
> -	inst->alg.base.cra_ctxsize = sizeof(struct chachapoly_ctx) +
> -				     ctx->saltlen;
> +	inst->alg.base.cra_ctxsize = struct_size_t(struct chachapoly_ctx, salt,
> +						   ctx->saltlen);

What was wrong with the more straightforward code it had before?

- Eric
Thorsten Blum Aug. 5, 2024, 8:32 p.m. UTC | #2
On 5. Aug 2024, at 19:59, Eric Biggers <ebiggers@kernel.org> wrote:
> On Mon, Aug 05, 2024 at 07:52:38PM +0200, Thorsten Blum wrote:
>> struct poly_req {
>> @@ -611,8 +611,8 @@ static int chachapoly_create(struct crypto_template *tmpl, struct rtattr **tb,
>>       poly->base.cra_priority) / 2;
>> inst->alg.base.cra_blocksize = 1;
>> inst->alg.base.cra_alignmask = chacha->base.cra_alignmask;
>> - inst->alg.base.cra_ctxsize = sizeof(struct chachapoly_ctx) +
>> -     ctx->saltlen;
>> + inst->alg.base.cra_ctxsize = struct_size_t(struct chachapoly_ctx, salt,
>> +   ctx->saltlen);
> 
> What was wrong with the more straightforward code it had before?

There's nothing wrong with it, but I find using the helper macro
struct_size_t() more straightforward. It's just a refactoring; happy to
take it out if there's a preference for the open coded version.

Thanks,
Thorsten
Kees Cook Aug. 5, 2024, 8:49 p.m. UTC | #3
On Mon, Aug 05, 2024 at 07:52:38PM +0200, Thorsten Blum wrote:
> Add the __counted_by compiler attribute to the flexible array member
> salt to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
> CONFIG_FORTIFY_SOURCE.
> 
> Use struct_size_t() instead of manually calculating the struct's size.
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
> ---
>  crypto/chacha20poly1305.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/crypto/chacha20poly1305.c b/crypto/chacha20poly1305.c
> index 9e4651330852..b37f59a8280a 100644
> --- a/crypto/chacha20poly1305.c
> +++ b/crypto/chacha20poly1305.c
> @@ -27,7 +27,7 @@ struct chachapoly_ctx {
>  	struct crypto_ahash *poly;
>  	/* key bytes we use for the ChaCha20 IV */
>  	unsigned int saltlen;
> -	u8 salt[];
> +	u8 salt[] __counted_by(saltlen);
>  };

AFAICT, all the allocations of struct chachapoly_ctx set "saltlen" before
using "salt".

Reviewed-by: Kees Cook <kees@kernel.org>

>  
>  struct poly_req {
> @@ -611,8 +611,8 @@ static int chachapoly_create(struct crypto_template *tmpl, struct rtattr **tb,
>  				       poly->base.cra_priority) / 2;
>  	inst->alg.base.cra_blocksize = 1;
>  	inst->alg.base.cra_alignmask = chacha->base.cra_alignmask;
> -	inst->alg.base.cra_ctxsize = sizeof(struct chachapoly_ctx) +
> -				     ctx->saltlen;
> +	inst->alg.base.cra_ctxsize = struct_size_t(struct chachapoly_ctx, salt,
> +						   ctx->saltlen);
>  	inst->alg.ivsize = ivsize;
>  	inst->alg.chunksize = chacha->chunksize;
>  	inst->alg.maxauthsize = POLY1305_DIGEST_SIZE;

My instinct is to use struct_size_t() as little as possible (compared
to normal struct_size), since I prefer to use it where a thing is being
allocated. (In this case, it's "far away".) So given Eric already thinks
this part should be dropped, I would agree: leave this as is.
diff mbox series

Patch

diff --git a/crypto/chacha20poly1305.c b/crypto/chacha20poly1305.c
index 9e4651330852..b37f59a8280a 100644
--- a/crypto/chacha20poly1305.c
+++ b/crypto/chacha20poly1305.c
@@ -27,7 +27,7 @@  struct chachapoly_ctx {
 	struct crypto_ahash *poly;
 	/* key bytes we use for the ChaCha20 IV */
 	unsigned int saltlen;
-	u8 salt[];
+	u8 salt[] __counted_by(saltlen);
 };
 
 struct poly_req {
@@ -611,8 +611,8 @@  static int chachapoly_create(struct crypto_template *tmpl, struct rtattr **tb,
 				       poly->base.cra_priority) / 2;
 	inst->alg.base.cra_blocksize = 1;
 	inst->alg.base.cra_alignmask = chacha->base.cra_alignmask;
-	inst->alg.base.cra_ctxsize = sizeof(struct chachapoly_ctx) +
-				     ctx->saltlen;
+	inst->alg.base.cra_ctxsize = struct_size_t(struct chachapoly_ctx, salt,
+						   ctx->saltlen);
 	inst->alg.ivsize = ivsize;
 	inst->alg.chunksize = chacha->chunksize;
 	inst->alg.maxauthsize = POLY1305_DIGEST_SIZE;