Skip to contentSkip to conversation

Fix array overflow in parport_serial.c

PATCH In review Takashi Iwai 2 participants 3 messages Nov 20, 2008 – Nov 22, 2008 Parallel port

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix array overflow in parport_serial.c

The netmos_9xx5_combo type assumes that PCI SSID provides always the correct value for the number of parallel and serial ports, but there are indeed broken devices with wrong numbers, which may result in Oops.

This patch simply adds the check of the array range.

Reference: Novell bnc#447067

	https://bugzilla.novell.com/show_bug.cgi?id=447067

---

drivers/parport/parport_serial.c +2 -0
index e2e95b3..101ed49 100644
@@ -70,6 +70,8 @@ static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc
 	 * parallel ports and <S> is the number of serial ports.
 	 */
 	card->numports = (dev->subsystem_device & 0xf0) >> 4;
+	if (card->numports > ARRAY_SIZE(card->addr))
+		card->numports = ARRAY_SIZE(card->addr);
 	return 0;
 }
 
#1 lore raw
Quoted 1 line · On Thu, 20 Nov 2008 17:35:20 +0100 Takashi Iwai <[email protected]> wrote:

Subject: [PATCH] Fix array overflow in parport_serial.c

Please prefer titles in the form

subsystem identifer: what was done to it

I renamed this one to

parport_serial: fix array overflow

Quoted 23 lines

Date: Thu, 20 Nov 2008 17:35:20 +0100
User-Agent: Wanderlust/2.12.0 (Your Wildest Dreams) SEMI/1.14.6 (Maruoka)
FLIM/1.14.7 (Sanj__) APEL/10.6 Emacs/22.3 (x86_64-suse-linux-gnu) MULE/5.0 (SAKAKI)

The netmos_9xx5_combo type assumes that PCI SSID provides always the correct value for the number of parallel and serial ports, but there are indeed broken devices with wrong numbers, which may result in Oops.

This patch simply adds the check of the array range.

Reference: Novell bnc#447067

	https://bugzilla.novell.com/show_bug.cgi?id=447067

---

drivers/parport/parport_serial.c +0 -0
index e2e95b3..101ed49 100644

drivers/parport/parport_serial.c static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc · line 73

Quoted 4 lines
drivers/parport/parport_serial.c +1 -0
@@ -70,6 +70,8 @@ static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc
 	 * parallel ports and <S> is the number of serial ports.
 	 */
 	card->numports = (dev->subsystem_device & 0xf0) >> 4;
+	if (card->numports > ARRAY_SIZE(card->addr))

hm. ARRAY_SIZE returns an unsigned type so we don't have to worry about negative values when doing comparisons like this. Not that card->numports could be negative anyway, but it's always nice to set readers' minds at rest..

drivers/parport/parport_serial.c static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc · line 76

Quoted 3 lines
drivers/parport/parport_serial.c +1 -0
+		card->numports = ARRAY_SIZE(card->addr);
 	return 0;
 }

Should we emit some kind of warning when this is detected? I guess not, if we're sure that there will never be a situation in which users find that some of their ports don't work?

#2 lore raw

At Fri, 21 Nov 2008 14:16:21 -0800,

Quoted 6 lines · Andrew Morton
Quoted 1 line · On Thu, 20 Nov 2008 17:35:20 +0100 Takashi Iwai <[email protected]> wrote:

Subject: [PATCH] Fix array overflow in parport_serial.c

Quoted 7 lines

Please prefer titles in the form

subsystem identifer: what was done to it

I renamed this one to

parport_serial: fix array overflow

Yep, better. Thanks.

Quoted 17 lines
Quoted 17 lines

Date: Thu, 20 Nov 2008 17:35:20 +0100
User-Agent: Wanderlust/2.12.0 (Your Wildest Dreams) SEMI/1.14.6 (Maruoka)
FLIM/1.14.7 (Sanj__) APEL/10.6 Emacs/22.3 (x86_64-suse-linux-gnu) MULE/5.0 (SAKAKI)

The netmos_9xx5_combo type assumes that PCI SSID provides always the correct value for the number of parallel and serial ports, but there are indeed broken devices with wrong numbers, which may result in Oops.

This patch simply adds the check of the array range.

Reference: Novell bnc#447067

	https://bugzilla.novell.com/show_bug.cgi?id=447067
Quoted 15 lines

drivers/parport/parport_serial.c static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc · line 73

Quoted 10 lines

---

drivers/parport/parport_serial.c +1 -0
index e2e95b3..101ed49 100644
@@ -70,6 +70,8 @@ static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc
 	 * parallel ports and <S> is the number of serial ports.
 	 */
 	card->numports = (dev->subsystem_device & 0xf0) >> 4;
+	if (card->numports > ARRAY_SIZE(card->addr))

hm. ARRAY_SIZE returns an unsigned type so we don't have to worry about negative values when doing comparisons like this. Not that card->numports could be negative anyway, but it's always nice to set readers' minds at rest..

I think changing numports to unsigned is the easiest way.

Quoted 8 lines

drivers/parport/parport_serial.c static int __devinit netmos_parallel_init(struct pci_dev *dev, struct parport_pc · line 76

Quoted 3 lines
drivers/parport/parport_serial.c +1 -0
+		card->numports = ARRAY_SIZE(card->addr);
 	return 0;
 }

Should we emit some kind of warning when this is detected? I guess not, if we're sure that there will never be a situation in which users find that some of their ports don't work?

I thought of that, too, but wanted to make the change minimum.

Besides that, I'm not sure whether netmos_parallel_init() really works as its description. It changes numports, but it doesn't change the addr[]. In cards[], only one item is defined, thus all of the reset point BAR 0.

I made the patch just to fix the oops, but this code should be fixed in a better way, anyway.

thanks,

Takashi