On Fri, Oct 6, 2017 at 12:00 PM, Paul Bolle <pebo...@tiscali.nl> wrote:
> On Thu, 2017-10-05 at 12:31 -0700, Kees Cook wrote:
>> --- a/drivers/isdn/gigaset/bas-gigaset.c
>> +++ b/drivers/isdn/gigaset/bas-gigaset.c
>
>> -static void cmd_in_timeout(unsigned long data)
>> +static void cmd_in_timeout(struct timer_list *t)
>> {
>> - struct cardstate *cs = (struct cardstate *) data;
>> - struct bas_cardstate *ucs = cs->hw.bas;
>> + struct bas_cardstate *ucs = from_timer(ucs, t, timer_cmd_in);
>> + struct urb *urb = ucs->urb_int_in;
>> + struct cardstate *cs = urb->context;
>
> This makes me nervous. Are you sure urb->context points to a struct cardstate
> here and in the other two places this patch changes?
>
> Anyhow, I'd like to have some time to do my review. So what's your timeframe
> here? I do hope I have at least a few weeks. (In other words: I hope gigaset
> isn't the only driver where the ability to use random pointers in these timer
> callbacks is removed.)
Hi!
I'm in no rush for any specific change. There are about 900 call sites
I'm making my way through, about 2/3rd are pretty trivial, and the
less obvious is what I've started sending out now, since I expect some
will need some more careful review.
Thanks for taking a look at it!
-Kees
--
Kees Cook
Pixel Security
