Re: [Patch] Off by one in drivers/usb/serial/usb-serial.c

[Eric Sesterhenn / Snakebyte]

* Mikael Pettersson (mikpe@xxxxxxxx) wrote:
> On Wed, 21 Jun 2006 23:28:17 +0200, Eric Sesterhenn wrote:
> > this fixes coverity id #554. since serial table
> > is defines as serial_table[SERIAL_TTY_MINORS] we
> > should make sure we dont acess with an index
> > of SERIAL_TTY_MINORS.
> > 
> > Signed-off-by: Eric Sesterhenn <snakebyte@xxxxxx>
> > 
> > --- linux-2.6.17-git2/drivers/usb/serial/usb-serial.c.orig	2006-06-21 23:24:07.000000000 +0200
> > +++ linux-2.6.17-git2/drivers/usb/serial/usb-serial.c	2006-06-21 23:25:12.000000000 +0200
> > @@ -83,7 +83,7 @@ static struct usb_serial *get_free_seria
> >  
> >  		good_spot = 1;
> >  		for (j = 1; j <= num_ports-1; ++j)
> > -			if ((i+j >= SERIAL_TTY_MINORS) || (serial_table[i+j])) {
> > +			if ((i+j >= SERIAL_TTY_MINORS-1)||(serial_table[i+j])) {
> >  				good_spot = 0;
> >  				i += j;
> >  				break;
> 
> Where is the access coverity complained about? If it's the serial_table[i+j]
> quoted above, then the original code is OK since i+j < SERIAL_TTY_MINORS is
> an invariant in that subexpression.
> 
> And the other accesses to serial_table[] in get_free_serial() are also only
> done when the index is < SERIAL_TTY_MINORS.

guess i was too quick on that one, sorry. Here is the coverity
report for completeness.

Event assignment: Assigning "1" to "j"
Also see events: [overrun-local]
At conditional (11): "j <= (num_ports - 1)" taking true path
At conditional (16): "j <= (num_ports - 1)" taking true path

85   			for (j = 1; j <= num_ports-1; ++j)

Event overrun-local: Overrun of static array "serial_table" of size 255
at position 255 with index variable "(i + j)"
Also see events: [assignment]
At conditional (12): "(i + j) >= 255" taking true path
At conditional (17): "(i + j) >= 255" taking false path

86   				if ((i+j >= SERIAL_TTY_MINORS) ||
(serial_table[i+j])) {
87   					good_spot = 0;
88   					i += j;
89   					break;
90   				}




greetings, Eric

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/