[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20211025153418.GA6853@asgard.redhat.com>
Date: Mon, 25 Oct 2021 17:34:18 +0200
From: Eugene Syromiatnikov <esyr@...hat.com>
To: David Miller <davem@...emloft.net>
Cc: jk@...econstruct.com.au, netdev@...r.kernel.org, kuba@...nel.org,
matt@...econstruct.com.au
Subject: Re: [PATCH net-next v5] mctp: Implement extended addressing
On Mon, Oct 25, 2021 at 04:15:49PM +0100, David Miller wrote:
> From: Jeremy Kerr <jk@...econstruct.com.au>
> Date: Mon, 25 Oct 2021 11:27:57 +0800
>
> > @@ -152,10 +155,16 @@ struct mctp_sk_key {
> >
> > struct mctp_skb_cb {
> > unsigned int magic;
> > - unsigned int net;
> > + int net;
> > + int ifindex; /* extended/direct addressing if set */
> > + unsigned char halen;
> > mctp_eid_t src;
> > + unsigned char haddr[];
> > };
> >
> putting a variably sized type in the skb control blocxk is not a good idea.
> Overruns will be silent, nothing in the typing protects you from udsing more space
> han exists in skb->cb.
>
> Plrease find another way, thank you.
haddr[MAX_ADDR_LEN]? It is defined to 32 bytes.
Powered by blists - more mailing lists