[PATCH] fs: ecryptfs: initialize payload_len in keystore.c

[PATCH] fs: ecryptfs: initialize payload_len in keystore.c

[Simon Que]

Signed-off-by: Simon Que <sque@chromium.org>
---
 fs/ecryptfs/keystore.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ecryptfs/keystore.c b/fs/ecryptfs/keystore.c
index 2333203..6154cde 100644
--- a/fs/ecryptfs/keystore.c
+++ b/fs/ecryptfs/keystore.c
@@ -1150,7 +1150,7 @@ decrypt_pki_encrypted_session_key(struct ecryptfs_auth_tok *auth_tok,
 	struct ecryptfs_message *msg = NULL;
 	char *auth_tok_sig;
 	char *payload;
-	size_t payload_len;
+	size_t payload_len = 0;
 	int rc;
 
 	rc = ecryptfs_get_auth_tok_sig(&auth_tok_sig, auth_tok);
-- 
1.7.8.6

Re: [PATCH] fs: ecryptfs: initialize payload_len in keystore.c

[Tyler Hicks]

[-- Attachment #1: Type: text/plain, Size: 1566 bytes --]

On 2013-01-16 18:08:14, Simon Que wrote:
> Signed-off-by: Simon Que <sque@chromium.org>

Hi Simon - Thanks for the patch but please resubmit it with a
description. Even with a one-liner as simple as this, there needs to be
some background information conveyed with the change.

Is this to just quiet a useless compiler warning or is there a place
were we actually use payload_len without initializing it? If we use it
uninitialized, does it cause a crash or do we "handle" it in error
paths? These details are factored in to determine if I need to send this
to the stable team and if distros need to backport it. It also keeps me
from scratching my head a year down the road when someone asks me about
the patch with no commit message. :)

FWIW, I took a quick look and don't see how it can be used uninitialized
but maybe I'm missing something. I also don't get a compiler warning
about it when building with `make W=12 M=fs/ecryptfs`.

Tyler

> ---
>  fs/ecryptfs/keystore.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/ecryptfs/keystore.c b/fs/ecryptfs/keystore.c
> index 2333203..6154cde 100644
> --- a/fs/ecryptfs/keystore.c
> +++ b/fs/ecryptfs/keystore.c
> @@ -1150,7 +1150,7 @@ decrypt_pki_encrypted_session_key(struct ecryptfs_auth_tok *auth_tok,
>  	struct ecryptfs_message *msg = NULL;
>  	char *auth_tok_sig;
>  	char *payload;
> -	size_t payload_len;
> +	size_t payload_len = 0;
>  	int rc;
>  
>  	rc = ecryptfs_get_auth_tok_sig(&auth_tok_sig, auth_tok);
> -- 
> 1.7.8.6
> 

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] fs: ecryptfs: initialize payload_len in keystore.c

[Simon Que]

Hi Tyler --

> Is this to just quiet a useless compiler warning or is there a place
> were we actually use payload_len without initializing it? If we use it
> uninitialized, does it cause a crash or do we "handle" it in error
> paths? These details are factored in to determine if I need to send this
> to the stable team and if distros need to backport it. It also keeps me
> from scratching my head a year down the road when someone asks me about
> the patch with no commit message. :)

This is just to squash a compiler warning.  Assuming
write_tag_64_packet() works as intended, if it fails to initialize
payload_len, it should return an error and the function should skip to
the end.

I'll re-send the patch with a comment explaining this.