Re: [PATCH] Fix leak in PNP interface code.

[Dave Jones <davej@redhat.com>]

On Wed, Sep 01, 2004 at 04:51:20PM +0100, Dave Jones wrote:
 > Spotted with the source checker from Coverity.com.
 > 
 > Signed-off-by: Dave Jones <davej@redhat.com>
 > 
 > 
 > diff -urpN --exclude-from=/home/davej/.exclude bk-linus/drivers/pnp/interface.c linux-2.6/drivers/pnp/interface.c
 > --- bk-linus/drivers/pnp/interface.c	2004-06-03 13:40:04.000000000 +0100
 > +++ linux-2.6/drivers/pnp/interface.c	2004-06-03 13:42:25.000000000 +0100
 > @@ -244,8 +244,10 @@ static ssize_t pnp_show_current_resource
 >  				pnp_alloc(sizeof(pnp_info_buffer_t));
 >  	if (!buffer)
 >  		return -ENOMEM;
 > -	if (!dev)
 > +	if (!dev) {
 > +		kfree(buffer);
 >  		return -EINVAL;
 > +	}
 >  	buffer->len = PAGE_SIZE;
 >  	buffer->buffer = buf;
 >  	buffer->curr = buffer->buffer;
 > -

Joe Perches pointed out that it may be better to just avoid the
allocation completely if !dev, which makes perfect sense to me.

Signed-off-by: Dave Jones <davej@redhat.com>


--- 1/drivers/pnp/interface.c~	2004-09-01 21:37:37.572983736 +0100
+++ 2/drivers/pnp/interface.c	2004-09-01 21:38:41.056332800 +0100
@@ -240,12 +240,14 @@
 {
 	struct pnp_dev *dev = to_pnp_dev(dmdev);
 	int i, ret;
-	pnp_info_buffer_t *buffer = (pnp_info_buffer_t *)
-				pnp_alloc(sizeof(pnp_info_buffer_t));
-	if (!buffer)
-		return -ENOMEM;
+	pnp_info_buffer_t *buffer;
+
 	if (!dev)
 		return -EINVAL;
+
+	buffer = (pnp_info_buffer_t *) pnp_alloc(sizeof(pnp_info_buffer_t));
+	if (!buffer)
+		return -ENOMEM;
 	buffer->len = PAGE_SIZE;
 	buffer->buffer = buf;
 	buffer->curr = buffer->buffer;