[PATCH] bcachefs: Annotate struct bucket_array with __counted_by()

Thorsten Blum posted 1 patch 1 year, 5 months ago
fs/bcachefs/buckets_types.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
[PATCH] bcachefs: Annotate struct bucket_array with __counted_by()
Posted by Thorsten Blum 1 year, 5 months ago
Add the __counted_by compiler attribute to the flexible array member
bucket to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
CONFIG_FORTIFY_SOURCE.

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
---
 fs/bcachefs/buckets_types.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/bcachefs/buckets_types.h b/fs/bcachefs/buckets_types.h
index c9698cdf866f..a19460a1bb7c 100644
--- a/fs/bcachefs/buckets_types.h
+++ b/fs/bcachefs/buckets_types.h
@@ -24,7 +24,7 @@ struct bucket_array {
 	u16			first_bucket;
 	size_t			nbuckets;
 	size_t			nbuckets_minus_first;
-	struct bucket		b[];
+	struct bucket		b[] __counted_by(nbuckets);
 };
 
 struct bucket_gens {
-- 
2.46.0
Re: [PATCH] bcachefs: Annotate struct bucket_array with __counted_by()
Posted by Kees Cook 1 year, 5 months ago
On Wed, Aug 21, 2024 at 06:29:22PM +0200, Thorsten Blum wrote:
> Add the __counted_by compiler attribute to the flexible array member
> bucket to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
> CONFIG_FORTIFY_SOURCE.
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
> ---
>  fs/bcachefs/buckets_types.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/fs/bcachefs/buckets_types.h b/fs/bcachefs/buckets_types.h
> index c9698cdf866f..a19460a1bb7c 100644
> --- a/fs/bcachefs/buckets_types.h
> +++ b/fs/bcachefs/buckets_types.h
> @@ -24,7 +24,7 @@ struct bucket_array {
>  	u16			first_bucket;
>  	size_t			nbuckets;
>  	size_t			nbuckets_minus_first;
> -	struct bucket		b[];
> +	struct bucket		b[] __counted_by(nbuckets);
>  };
>  
>  struct bucket_gens {

I only see this being allocated in one place and nbuckets is immediately
assigned to the count of "b" elements, so this looks correct to me.

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

-- 
Kees Cook