Re: Coverity: ucsi_check_cable(): Null pointer dereferences

[Kees Cook]

On Thu, Mar 07, 2024 at 09:16:00PM +0100, Christian A. Ehrhardt wrote:
> 
> Hi,
> 
> On Thu, Mar 07, 2024 at 11:34:21AM -0800, coverity-bot wrote:
> > Hello!
> > 
> > This is an experimental semi-automated report about issues detected by
> > Coverity from a scan of next-20240307 as part of the linux-next scan project:
> > https://scan.coverity.com/projects/linux-next-weekly-scan
> > 
> > You're getting this email because you were associated with the identified
> > lines of code (noted below) that were touched by commits:
> > 
> >   Tue Mar 5 13:11:08 2024 +0000
> >     f896d5e8726c ("usb: typec: ucsi: Register SOP/SOP' Discover Identity Responses")
> > 
> > Coverity reported the following:
> > 
> > *** CID 1584245:  Null pointer dereferences  (FORWARD_NULL)
> > drivers/usb/typec/ucsi/ucsi.c:1136 in ucsi_check_cable()
> > 1130     	}
> > 1131
> > 1132     	ret = ucsi_register_cable(con);
> > 1133     	if (ret < 0)
> > 1134     		return ret;
> > 1135
> > vvv     CID 1584245:  Null pointer dereferences  (FORWARD_NULL)
> > vvv     Passing "con" to "ucsi_get_cable_identity", which dereferences null "con->cable".
> > 1136     	ret = ucsi_get_cable_identity(con);
> > 1137     	if (ret < 0)
> > 1138     		return ret;
> > 1139
> > 1140     	ret = ucsi_register_plug(con);
> > 1141     	if (ret < 0)
> > 
> > If this is a false positive, please let us know so we can mark it as
> > such, or teach the Coverity rules to be smarter. If not, please make
> > sure fixes get into linux-next. :) For patches fixing this, please
> > include these lines (but double-check the "Fixes" first):
> 
> This looks like a false positive to me. The code looks like this:
> 
> 	if (con->cable)
> 		return 0;
> 	[ ... ]
> 	ret = ucsi_register_cable(con)
> 	if (ret < 0)
> 		return ret;
> 	ret = ucsi_get_cable_identity(con);
> 	[ ... ]
> 
> From the con->cable check coverity concludes that con->cable is
> initially NULL. Later ucsi_register_cable() initializes con->cable
> if successful. Coverity seems to miss this and still thinks that
> con->cable is NULL. Then converity correctly notices that
> ucsi_get_cable_identity() dereferences con->cable and complains.

Ah-ha! Yes, the ucsi_register_cable() check seems to have been missed.
I think it's confused by:

        cable = typec_register_cable(con->port, &desc);
        if (IS_ERR(cable)) {

This isn't IS_ERR_OR_NULL, so it thinks cable might still be NULL, but
there's no path through typec_register_cable() where that can be true.

Thanks for taking a look!

-Kees

-- 
Kees Cook