[PATCH] aio: fix potential leak in aio_run_iocb().

[PATCH] aio: fix potential leak in aio_run_iocb().

[Leon Yu]

iovec should be reclaimed whenever caller of rw_copy_check_uvector() returns,
but it doesn't hold when failure happens right after aio_setup_vectored_rw().

Fix that in a such way to avoid hairy goto.

Signed-off-by: Leon Yu <chianglungyu@gmail.com>
---
 fs/aio.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/fs/aio.c b/fs/aio.c
index 12a3de0e..04cd768 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1299,10 +1299,8 @@ rw_common:
 						&iovec, compat)
 			: aio_setup_single_vector(req, rw, buf, &nr_segs,
 						  iovec);
-		if (ret)
-			return ret;
-
-		ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);
+		if (!ret)
+			ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);
 		if (ret < 0) {
 			if (iovec != &inline_vec)
 				kfree(iovec);
-- 
1.9.2

Re: [PATCH] aio: fix potential leak in aio_run_iocb().

[Mateusz Guzik]

On Thu, May 01, 2014 at 03:31:28AM +0000, Leon Yu wrote:
> iovec should be reclaimed whenever caller of rw_copy_check_uvector() returns,
> but it doesn't hold when failure happens right after aio_setup_vectored_rw().
> 

There is a proposal (wich a patch) to modify semantics of rw_copy_check_uvector
so that it frees stuff on error, taking care of this case as well:

https://lkml.org/lkml/2014/4/25/778

-- 
Mateusz Guzik

Re: [PATCH] aio: fix potential leak in aio_run_iocb().

[Benjamin LaHaise]

On Thu, May 01, 2014 at 03:31:28AM +0000, Leon Yu wrote:
> iovec should be reclaimed whenever caller of rw_copy_check_uvector() returns,
> but it doesn't hold when failure happens right after aio_setup_vectored_rw().
> 
> Fix that in a such way to avoid hairy goto.

Good catch -- applied.

		-ben

> Signed-off-by: Leon Yu <chianglungyu@gmail.com>
> ---
>  fs/aio.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/aio.c b/fs/aio.c
> index 12a3de0e..04cd768 100644
> --- a/fs/aio.c
> +++ b/fs/aio.c
> @@ -1299,10 +1299,8 @@ rw_common:
>  						&iovec, compat)
>  			: aio_setup_single_vector(req, rw, buf, &nr_segs,
>  						  iovec);
> -		if (ret)
> -			return ret;
> -
> -		ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);
> +		if (!ret)
> +			ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);
>  		if (ret < 0) {
>  			if (iovec != &inline_vec)
>  				kfree(iovec);
> -- 
> 1.9.2

-- 
"Thought is the essence of where you are now."

Re: [PATCH] aio: fix potential leak in aio_run_iocb().

[Lukáš Czerner]

On Thu, 1 May 2014, Leon Yu wrote:

> Date: Thu,  1 May 2014 03:31:28 +0000
> From: Leon Yu <chianglungyu@gmail.com>
> To: Benjamin LaHaise <bcrl@kvack.org>,
>     Alexander Viro <viro@zeniv.linux.org.uk>
> Cc: linux-aio@kvack.org, linux-fsdevel@vger.kernel.org,
>     linux-kernel@vger.kernel.org, Leon Yu <chianglungyu@gmail.com>
> Subject: [PATCH] aio: fix potential leak in aio_run_iocb().
> 
> iovec should be reclaimed whenever caller of rw_copy_check_uvector() returns,
> but it doesn't hold when failure happens right after aio_setup_vectored_rw().
> 
> Fix that in a such way to avoid hairy goto.


Hi,

this does not seem right, see bellow

> 
> Signed-off-by: Leon Yu <chianglungyu@gmail.com>
> ---
>  fs/aio.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/aio.c b/fs/aio.c
> index 12a3de0e..04cd768 100644
> --- a/fs/aio.c
> +++ b/fs/aio.c
> @@ -1299,10 +1299,8 @@ rw_common:
>  						&iovec, compat)
>  			: aio_setup_single_vector(req, rw, buf, &nr_segs,
>  						  iovec);
> -		if (ret)
> -			return ret;
> -
> -		ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);

here ret could be possibly set to a positive number.

> +		if (!ret)
> +			ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes);
>  		if (ret < 0) {

but here we're checking for negative and bail out. However this
changes the way it worked before this patch and the iovec would not
be reclaimed anyway as you mentioned in the commit description.

Thanks!
-Lukas

>  			if (iovec != &inline_vec)
>  				kfree(iovec);