[PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[Micah Gruber]

This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.

Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
---

--- a/drivers/usb/storage/shuttle_usbat.c
+++ b/drivers/usb/storage/shuttle_usbat.c
@@ -187,12 +187,14 @@
  */
 static int usbat_check_status(struct us_data *us)
 {
-	unsigned char *reply = us->iobuf;
+	unsigned char *reply;
 	int rc;
 
 	if (!us)
 		return USB_STOR_TRANSPORT_ERROR;
 
+	reply = us->iobuf;
+
 	rc = usbat_get_status(us, reply);
 	if (rc != USB_STOR_XFER_GOOD)
 		return USB_STOR_TRANSPORT_FAILED;

Re: [PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[Jens Axboe]

On Tue, Sep 04 2007, Micah Gruber wrote:
> This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> 
> Signed-off-by: Micah Gruber <micah.gruber@gmail.com>

Be careful with stuff like that, if you actually look at the code, a us
== NULL doesn't seem to be possible (or usbat_flash_transport() would
have oopsed before).

-- 
Jens Axboe

Re: [PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[Simon Holm Thøgersen]

tir, 04 09 2007 kl. 13:06 +0200, skrev Jens Axboe:
> On Tue, Sep 04 2007, Micah Gruber wrote:
> > This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> > 
> > Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
> 
> Be careful with stuff like that, if you actually look at the code, a us
> == NULL doesn't seem to be possible (or usbat_flash_transport() would
> have oopsed before).
> 
If that is true, then
        if (!us)
                return USB_STOR_TRANSPORT_ERROR;
is utterly pointless.


Simon Holm Thøgersen

Re: [linux-usb-devel] [PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[Alan Stern]

On Tue, 4 Sep 2007, Simon Holm Thøgersen wrote:

> > tir, 04 09 2007 kl. 13:06 +0200, skrev Jens Axboe:
> > On Tue, Sep 04 2007, Micah Gruber wrote:
> > > This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> > > 
> > > Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
> > 
> > Be careful with stuff like that, if you actually look at the code, a us
> > == NULL doesn't seem to be possible (or usbat_flash_transport() would
> > have oopsed before).
> > 
> If that is true, then
>         if (!us)
>                 return USB_STOR_TRANSPORT_ERROR;
> is utterly pointless.

Indeed, so it is.

Alan Stern

Re: [PATCH] Fix a potential NULL pointer dereference in usbat_check_status() in drivers/usb/storage/shuttle_usbat.c

[Jens Axboe]

On Tue, Sep 04 2007, Simon Holm Thøgersen wrote:
> tir, 04 09 2007 kl. 13:06 +0200, skrev Jens Axboe:
> > On Tue, Sep 04 2007, Micah Gruber wrote:
> > > This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> > > 
> > > Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
> > 
> > Be careful with stuff like that, if you actually look at the code, a us
> > == NULL doesn't seem to be possible (or usbat_flash_transport() would
> > have oopsed before).
> > 
> If that is true, then
>         if (!us)
>                 return USB_STOR_TRANSPORT_ERROR;
> is utterly pointless.

Well that was the point I was trying to make, that test and return
should be deleted instead.

-- 
Jens Axboe

[PATCH] Remove pointless NULL pointer check in drivers/usb/storage/shuttle_usbat.c.

[Simon Holm Thøgersen]

tir, 04 09 2007 kl. 23:06 +0200, skrev Jens Axboe:
> On Tue, Sep 04 2007, Simon Holm Thøgersen wrote:
> > tir, 04 09 2007 kl. 13:06 +0200, skrev Jens Axboe:
> > > On Tue, Sep 04 2007, Micah Gruber wrote:
> > > > This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> > > > 
> > > > Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
> > > 
> > > Be careful with stuff like that, if you actually look at the code, a us
> > > == NULL doesn't seem to be possible (or usbat_flash_transport() would
> > > have oopsed before).
> > > 
> > If that is true, then
> >         if (!us)
> >                 return USB_STOR_TRANSPORT_ERROR;
> > is utterly pointless.
> 
> Well that was the point I was trying to make, that test and return
> should be deleted instead.
> 
I guess we agree that we want the following then.


If us would ever be NULL, the function would have oopsed already before
the check.

Signed-off-by: Simon Holm Thøgersen <odie@cs.aau.dk>
---

--- a/drivers/usb/storage/shuttle_usbat.c
+++ b/drivers/usb/storage/shuttle_usbat.c
@@ -190,9 +190,6 @@ static int usbat_check_status(struct us_data *us)
 	unsigned char *reply = us->iobuf;
 	int rc;
 
-	if (!us)
-		return USB_STOR_TRANSPORT_ERROR;
-
 	rc = usbat_get_status(us, reply);
 	if (rc != USB_STOR_XFER_GOOD)
 		return USB_STOR_TRANSPORT_FAILED;

Re: [PATCH] Remove pointless NULL pointer check in drivers/usb/storage/shuttle_usbat.c.

[Jens Axboe]

On Fri, Sep 07 2007, Simon Holm Thøgersen wrote:
> tir, 04 09 2007 kl. 23:06 +0200, skrev Jens Axboe:
> > On Tue, Sep 04 2007, Simon Holm Thøgersen wrote:
> > > tir, 04 09 2007 kl. 13:06 +0200, skrev Jens Axboe:
> > > > On Tue, Sep 04 2007, Micah Gruber wrote:
> > > > > This patch fixes a potential null dereference bug where we dereference us before a null check. This patch simply moves the dereferencing after the null check.
> > > > > 
> > > > > Signed-off-by: Micah Gruber <micah.gruber@gmail.com>
> > > > 
> > > > Be careful with stuff like that, if you actually look at the code, a us
> > > > == NULL doesn't seem to be possible (or usbat_flash_transport() would
> > > > have oopsed before).
> > > > 
> > > If that is true, then
> > >         if (!us)
> > >                 return USB_STOR_TRANSPORT_ERROR;
> > > is utterly pointless.
> > 
> > Well that was the point I was trying to make, that test and return
> > should be deleted instead.
> > 
> I guess we agree that we want the following then.
> 
> 
> If us would ever be NULL, the function would have oopsed already before
> the check.

Yep, looks much better.

Acked-by: Jens Axboe <jens.axboe@oracle.com>

> 
> Signed-off-by: Simon Holm Thøgersen <odie@cs.aau.dk>
> ---
> 
> --- a/drivers/usb/storage/shuttle_usbat.c
> +++ b/drivers/usb/storage/shuttle_usbat.c
> @@ -190,9 +190,6 @@ static int usbat_check_status(struct us_data *us)
>  	unsigned char *reply = us->iobuf;
>  	int rc;
>  
> -	if (!us)
> -		return USB_STOR_TRANSPORT_ERROR;
> -
>  	rc = usbat_get_status(us, reply);
>  	if (rc != USB_STOR_XFER_GOOD)
>  		return USB_STOR_TRANSPORT_FAILED;
> 
> 

-- 
Jens Axboe