[PATCH] drivers/staging/hv/blkvsc_drv.c: eliminate NULL pointer dereference

[PATCH] drivers/staging/hv/blkvsc_drv.c: eliminate NULL pointer dereference

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

In this code, blkvsc_req is allocated in the cache blkdev->request_pool,
but freed in the first case to the cache blkvsc_req->dev->request_pool.
blkvsc_req->dev is subsequently initialized to blkdev, making these the
same at the second call to kmem_cache_free.  But at the point of the first
call, blkvsc_req->dev is NULL.  The second call is changed too, for
uniformity.

The semantic patch that fixes this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression x,e,e1,e2,e3;
@@

x = \(kmem_cache_alloc\|kmem_cache_zalloc\)(e1,e2)
... when != x = e
(
kmem_cache_free(e1,x);
|
?-kmem_cache_free(e3,x);
+kmem_cache_free(e1,x);
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
 drivers/staging/hv/blkvsc_drv.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff -u -p a/drivers/staging/hv/blkvsc_drv.c b/drivers/staging/hv/blkvsc_drv.c
--- a/drivers/staging/hv/blkvsc_drv.c
+++ b/drivers/staging/hv/blkvsc_drv.c
@@ -325,7 +325,7 @@ static int blkvsc_do_operation(struct bl
 
 	page_buf = alloc_page(GFP_KERNEL);
 	if (!page_buf) {
-		kmem_cache_free(blkvsc_req->dev->request_pool, blkvsc_req);
+		kmem_cache_free(blkdev->request_pool, blkvsc_req);
 		return -ENOMEM;
 	}
 
@@ -422,7 +422,7 @@ cleanup:
 
 	__free_page(page_buf);
 
-	kmem_cache_free(blkvsc_req->dev->request_pool, blkvsc_req);
+	kmem_cache_free(blkdev->request_pool, blkvsc_req);
 
 	return ret;
 }

RE: [PATCH] drivers/staging/hv/blkvsc_drv.c: eliminate NULL pointer dereference

[KY Srinivasan]

> -----Original Message-----
> From: Julia Lawall [mailto:julia@diku.dk]
> Sent: Saturday, July 09, 2011 3:23 PM
> To: Hank Janssen
> Cc: kernel-janitors@vger.kernel.org; Haiyang Zhang; Greg Kroah-Hartman; KY
> Srinivasan; Abhishek Kane (Mindtree Consulting PVT LTD);
> devel@driverdev.osuosl.org; linux-kernel@vger.kernel.org
> Subject: [PATCH] drivers/staging/hv/blkvsc_drv.c: eliminate NULL pointer
> dereference
> 
> From: Julia Lawall <julia@diku.dk>
> 
> In this code, blkvsc_req is allocated in the cache blkdev->request_pool,
> but freed in the first case to the cache blkvsc_req->dev->request_pool.
> blkvsc_req->dev is subsequently initialized to blkdev, making these the
> same at the second call to kmem_cache_free.  But at the point of the first
> call, blkvsc_req->dev is NULL.  The second call is changed too, for
> uniformity.
> 
> The semantic patch that fixes this problem is as follows:
> (http://coccinelle.lip6.fr/)
> 
> // <smpl>
> @@
> expression x,e,e1,e2,e3;
> @@
> 
> x = \(kmem_cache_alloc\|kmem_cache_zalloc\)(e1,e2)
> ... when != x = e
> (
> kmem_cache_free(e1,x);
> |
> ?-kmem_cache_free(e3,x);
> +kmem_cache_free(e1,x);
> )
> // </smpl>
> 
> Signed-off-by: Julia Lawall <julia@diku.dk>
> 
> ---
>  drivers/staging/hv/blkvsc_drv.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff -u -p a/drivers/staging/hv/blkvsc_drv.c b/drivers/staging/hv/blkvsc_drv.c
> --- a/drivers/staging/hv/blkvsc_drv.c
> +++ b/drivers/staging/hv/blkvsc_drv.c
> @@ -325,7 +325,7 @@ static int blkvsc_do_operation(struct bl
> 
>  	page_buf = alloc_page(GFP_KERNEL);
>  	if (!page_buf) {
> -		kmem_cache_free(blkvsc_req->dev->request_pool,
> blkvsc_req);
> +		kmem_cache_free(blkdev->request_pool, blkvsc_req);
>  		return -ENOMEM;
>  	}
> 
> @@ -422,7 +422,7 @@ cleanup:
> 
>  	__free_page(page_buf);
> 
> -	kmem_cache_free(blkvsc_req->dev->request_pool, blkvsc_req);
> +	kmem_cache_free(blkdev->request_pool, blkvsc_req);
> 
>  	return ret;
>  }
> 
Thank you for catching this. As you know, this driver is going away. Some weeks ago,
I had submitted patches to handle the IDE devices (the ones currently handled by this 
driver) with the stor driver. Greg, I am in the process of re-spinning all the patches I had
sent in June, based on your comments last week. If you are going to be applying this patch,
please let me know, I can locally apply this and correctly re-base my patches. 

Regards,

K. Y