[PATCH] USB: serial: Fix use after free in debug printk

[PATCH] USB: serial: Fix use after free in debug printk

[Dan Carpenter]

The dev_dbg() call dereferences "urb" but it was already freed on the
previous line.  Move the debug output earlier in the function.

Fixes: 984f68683298 ("USB: serial: io_edgeport.c: remove dbg() usage")
Cc: stable@vger.kernel.org
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
---
 drivers/usb/serial/io_edgeport.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/drivers/usb/serial/io_edgeport.c b/drivers/usb/serial/io_edgeport.c
index c7d6b5e3f898..b8f1bd41fb24 100644
--- a/drivers/usb/serial/io_edgeport.c
+++ b/drivers/usb/serial/io_edgeport.c
@@ -775,7 +775,10 @@ static void edge_bulk_out_cmd_callback(struct urb *urb)
 	atomic_dec(&CmdUrbs);
 	dev_dbg(&urb->dev->dev, "%s - FREE URB %p (outstanding %d)\n",
 		__func__, urb, atomic_read(&CmdUrbs));
-
+	if (status)
+		dev_dbg(&urb->dev->dev,
+			"%s - nonzero write bulk status received: %d\n",
+			__func__, status);
 
 	/* clean up the transfer buffer */
 	kfree(urb->transfer_buffer);
@@ -783,12 +786,8 @@ static void edge_bulk_out_cmd_callback(struct urb *urb)
 	/* Free the command urb */
 	usb_free_urb(urb);
 
-	if (status) {
-		dev_dbg(&urb->dev->dev,
-			"%s - nonzero write bulk status received: %d\n",
-			__func__, status);
+	if (status)
 		return;
-	}
 
 	/* tell the tty driver that something has changed */
 	if (edge_port->open)
-- 
2.45.2

Re: [PATCH] USB: serial: Fix use after free in debug printk

[Johan Hovold]

On Thu, Oct 31, 2024 at 09:59:10AM +0300, Dan Carpenter wrote:
> The dev_dbg() call dereferences "urb" but it was already freed on the
> previous line.  Move the debug output earlier in the function.

Thanks for catching this, but please use a temporary variable for the
struct device pointer instead of changing the flow.

Also make sure to include the driver name in the patch summary prefix
(i.e. "USB: serial: io_edgeport: ..."):

> Fixes: 984f68683298 ("USB: serial: io_edgeport.c: remove dbg() usage")
> Cc: stable@vger.kernel.org
> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>

Johan

Re: [PATCH] USB: serial: Fix use after free in debug printk

[Dan Carpenter]

On Thu, Oct 31, 2024 at 10:07:42AM +0100, Johan Hovold wrote:
> On Thu, Oct 31, 2024 at 09:59:10AM +0300, Dan Carpenter wrote:
> > The dev_dbg() call dereferences "urb" but it was already freed on the
> > previous line.  Move the debug output earlier in the function.
> 
> Thanks for catching this, but please use a temporary variable for the
> struct device pointer instead of changing the flow.
> 

Why?  The output is the same either way and this way is cleaner code.

> Also make sure to include the driver name in the patch summary prefix
> (i.e. "USB: serial: io_edgeport: ..."):

Sure.

regards,
dan carpenter

Re: [PATCH] USB: serial: Fix use after free in debug printk

[Dan Carpenter]

On Thu, Oct 31, 2024 at 12:35:31PM +0300, Dan Carpenter wrote:
> On Thu, Oct 31, 2024 at 10:07:42AM +0100, Johan Hovold wrote:
> > On Thu, Oct 31, 2024 at 09:59:10AM +0300, Dan Carpenter wrote:
> > > The dev_dbg() call dereferences "urb" but it was already freed on the
> > > previous line.  Move the debug output earlier in the function.
> > 
> > Thanks for catching this, but please use a temporary variable for the
> > struct device pointer instead of changing the flow.
> > 
> 
> Why?  The output is the same either way and this way is cleaner code.
> 

Nah, you're right.  A temporary variable is nicer.  It avoids having two if
statements.

regards,
dan carpenter

Re: [PATCH] USB: serial: Fix use after free in debug printk

[Johan Hovold]

On Thu, Oct 31, 2024 at 12:39:10PM +0300, Dan Carpenter wrote:
> On Thu, Oct 31, 2024 at 12:35:31PM +0300, Dan Carpenter wrote:
> > On Thu, Oct 31, 2024 at 10:07:42AM +0100, Johan Hovold wrote:
> > > On Thu, Oct 31, 2024 at 09:59:10AM +0300, Dan Carpenter wrote:
> > > > The dev_dbg() call dereferences "urb" but it was already freed on the
> > > > previous line.  Move the debug output earlier in the function.
> > > 
> > > Thanks for catching this, but please use a temporary variable for the
> > > struct device pointer instead of changing the flow.
> > 
> > Why?  The output is the same either way and this way is cleaner code.
> 
> Nah, you're right.  A temporary variable is nicer.  It avoids having two if
> statements.

Yeah, and the debug printk belongs with the return.

v2 now applied, thanks.

Johan