Re: [PATCH net-next v14 12/13] net: ethtool: strset: Allow querying phy stats by index

From: Maxime Chevallier
Date: Wed Jul 03 2024 - 02:55:31 EST


Hello Simon,

On Tue, 2 Jul 2024 11:54:11 +0100
Simon Horman <horms@xxxxxxxxxx> wrote:

> On Mon, Jul 01, 2024 at 03:17:58PM +0200, Maxime Chevallier wrote:
> > The ETH_SS_PHY_STATS command gets PHY statistics. Use the phydev pointer
> > from the ethnl request to allow query phy stats from each PHY on the
> > link.
> >
> > Signed-off-by: Maxime Chevallier <maxime.chevallier@xxxxxxxxxxx>
> > ---
> > net/ethtool/strset.c | 24 +++++++++++++++++-------
> > 1 file changed, 17 insertions(+), 7 deletions(-)
> >
> > diff --git a/net/ethtool/strset.c b/net/ethtool/strset.c
>
> ...
>
> > @@ -279,6 +280,8 @@ static int strset_prepare_data(const struct ethnl_req_info *req_base,
> > const struct strset_req_info *req_info = STRSET_REQINFO(req_base);
> > struct strset_reply_data *data = STRSET_REPDATA(reply_base);
> > struct net_device *dev = reply_base->dev;
> > + struct nlattr **tb = info->attrs;
>
> Hi Maxime,
>
> Elsewhere in this function it is assumed that info may be NULL.
> But here it is dereferenced unconditionally.

Hmm in almst all netlink commands we do dereference the genl_info *info
pointer without checks.

I've looked into net/netlink/genetlink.c to backtrack call-sites and it
looks to be that indeed info can't be NULL (either populated from
genl_start() or genl_family_rcv_msg_doit(). Maybe Jakub can confirm
this ?

If what I say above is correct, I can include a small patch to remove
the un-necessary check that makes smatch think the genl_info pointer can
be NULL.

Thanks for the report,

Maxime