Re: [PATCH net v2] net: ethtool: netlink: Allow NULL nlattrs when getting a phy_device

[Maxime Chevallier]

On Mon, 3 Mar 2025 16:10:24 -0800
Jakub Kicinski <kuba@xxxxxxxxxx> wrote:

> On Sat,  1 Mar 2025 15:11:13 +0100 Maxime Chevallier wrote:
> > ethnl_req_get_phydev() is used to lookup a phy_device, in the case an
> > ethtool netlink command targets a specific phydev within a netdev's
> > topology.
> > 
> > It takes as a parameter a const struct nlattr *header that's used for
> > error handling :
> > 
> >        if (!phydev) {
> >                NL_SET_ERR_MSG_ATTR(extack, header,
> >                                    "no phy matching phyindex");
> >                return ERR_PTR(-ENODEV);
> >        }
> > 
> > In the notify path after a ->set operation however, there's no request
> > attributes available.
> > 
> > The typical callsite for the above function looks like:
> > 
> > 	phydev = ethnl_req_get_phydev(req_base, tb[ETHTOOL_A_XXX_HEADER],
> > 				      info->extack);
> > 
> > So, when tb is NULL (such as in the ethnl notify path), we have a nice
> > crash.
> > 
> > It turns out that there's only the PLCA command that is in that case, as
> > the other phydev-specific commands don't have a notification.
> > 
> > This commit fixes the crash by passing the cmd index and the nlattr
> > array separately, allowing NULL-checking it directly inside the helper.
> > 
> > Fixes: c15e065b46dc ("net: ethtool: Allow passing a phy index for some commands")
> > Signed-off-by: Maxime Chevallier <maxime.chevallier@xxxxxxxxxxx>  
> 
> Well, this alone doesn't look too bad.. :) Hopefully we can address
> adding more suitable handlers for phy ops in net-next.

Yeah I'm cooking something to improve on that, and I have also dusted
off a netdevsim patch I had written back when working on the
phy_link_topology that adds very very basic PHY support so that we can
start covering all these commands with proper tests. I'll hopefully send
something in the coming week or so.

> Didn't someone report this, tho? I vaguely remember seeing an email,
> unless they said they don't want to be credited maybe we should add
> a Reported-by tag? You can post it in reply, no need to repost 
> the patch.

Parthiban reported this without CC: netdev, but I think it's fair to
add :

Reported-by: Parthiban Veerasooran <parthiban.veerasooran@xxxxxxxxxxxxx>

I didn't include it in the first place because checkpatch complained
about a reported-by tag without a "Closes:", which we don't have
because of the private reporting :)

Thanks Jakub,

Maxime