Re: [PATCH] fscrypt: don't load ->i_crypt_info before it's known to be valid

[Eric Biggers <ebiggers@kernel.org>]

On Mon, Jul 27, 2020 at 10:41:58AM -0700, Eric Biggers wrote:
> From: Eric Biggers <ebiggers@google.com>
> 
> In fscrypt_set_bio_crypt_ctx(), ->i_crypt_info isn't known to be
> non-NULL until we check fscrypt_inode_uses_inline_crypto().  So, load
> ->i_crypt_info after the check rather than before.  This makes no
> difference currently, but it prevents people from introducing bugs where
> the pointer is dereferenced when it may be NULL.
> 
> Suggested-by: Dave Chinner <david@fromorbit.com>
> Cc: Satya Tangirala <satyat@google.com>
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> ---
>  fs/crypto/inline_crypt.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/crypto/inline_crypt.c b/fs/crypto/inline_crypt.c
> index dfb06375099ae..b6b8574caa13c 100644
> --- a/fs/crypto/inline_crypt.c
> +++ b/fs/crypto/inline_crypt.c
> @@ -244,11 +244,12 @@ static void fscrypt_generate_dun(const struct fscrypt_info *ci, u64 lblk_num,
>  void fscrypt_set_bio_crypt_ctx(struct bio *bio, const struct inode *inode,
>  			       u64 first_lblk, gfp_t gfp_mask)
>  {
> -	const struct fscrypt_info *ci = inode->i_crypt_info;
> +	const struct fscrypt_info *ci;
>  	u64 dun[BLK_CRYPTO_DUN_ARRAY_SIZE];
>  
>  	if (!fscrypt_inode_uses_inline_crypto(inode))
>  		return;
> +	ci = inode->i_crypt_info;
>  
>  	fscrypt_generate_dun(ci, first_lblk, dun);
>  	bio_crypt_set_ctx(bio, &ci->ci_enc_key.blk_key->base, dun, gfp_mask);

Applied to fscrypt.git#master for 5.9.

- Eric