Re: [PATCH net] net: ethtool: pse-pd: Fix possible null-deref

[Kory Maincent <kory.maincent@bootlin.com>]

On Tue, 9 Jul 2024 08:52:05 -0700
Jakub Kicinski <kuba@kernel.org> wrote:

> On Tue, 9 Jul 2024 16:43:05 +0200 Kory Maincent wrote:
> > > Normal ethtool flow is to first fill in the data with a ->get() then
> > > modify what user wants to change.
> > > 
> > > Either we need:
> > >  - an explanation in the commit message how this keeps old config; or
> > >  - a ->get() to keep the previous values; or
> > >  - just reject setting one value but not the other in
> > >    ethnl_set_pse_validate() (assuming it never worked, anyway).    
> > 
> > In fact it is the contrary we can't set both value at the same time because
> > a PSE port can't be a PoE and a PoDL power interface at the same time.  
> 
> In that case maybe we should have an inverse condition in validate, too?
> Something like:
> 
> 	if ((pse_has_podl(phydev->psec) &&
> 	     GENL_REQ_ATTR_CHECK(info, ETHTOOL_A_PODL_PSE_ADMIN_CONTROL)) ||
> 	    (pse_has_c33(phydev->psec) &&
> 	     GENL_REQ_ATTR_CHECK(info, ETHTOOL_A_C33_PSE_ADMIN_CONTROL)))
> 		return -EINVAL;
> 
> GENL_REQ_ATTR_CHECK will set the extack for us.

I don't think it will work.
In a case a PSE controller have one PoDL port and one PoE port.
Your code will always return EINVAL if we try to set the config of only one of
those ports. 

I think the patch I sent plus a change in pse-pd core to do nothing in case of
an empty config will do.

Regards,
-- 
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com