Re: [RFC 1/1] whitelisting UDP GSO and GRO cmsgs

[Victor Stewart <v@nametag.social>]

On Mon, Nov 23, 2020 at 8:37 PM Pavel Begunkov <asml.silence@gmail.com> wrote:
>
> On 23/11/2020 15:35, Victor Stewart wrote:
> > add __sys_whitelisted_cmsghdrs() and configure __sys_recvmsg_sock and
> > __sys_sendmsg_sock to use it.
>
> They haven't been disabled without a reason, and that's not only
> because of creds. Did you verify that it's safe to allow those?

conceptually i don't see how it would not be safe? but maybe someone
who knows better can make such an argument.

>
> >
> > Signed-off by: Victor Stewart <v@nametag.social>
> > ---
> >  net/socket.c | 15 ++++++++++++---
> >  1 file changed, 12 insertions(+), 3 deletions(-)
> >
> > diff --git a/net/socket.c b/net/socket.c
> > index 6e6cccc2104f..44e28bb08bbe 100644
> > --- a/net/socket.c
> > +++ b/net/socket.c
> > @@ -2416,9 +2416,9 @@ static int ___sys_sendmsg(struct socket *sock,
> > struct user_msghdr __user *msg,
> >  long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg,
> >                         unsigned int flags)
> >  {
> > -       /* disallow ancillary data requests from this path */
> >         if (msg->msg_control || msg->msg_controllen)
> > -               return -EINVAL;
> > +               if (!__sys_whitelisted_cmsghdrs(msr))
>
> Its definition below and I don't see a forward declaration anywhere.
> Did you even compile this?

i knew it would be a hotly contested topic, so no i did not compile
it, that's why i put RFC not patch. Just to broach a potential
solution for discussion.

>
> > +                       return -EINVAL;
> >
> >         return ____sys_sendmsg(sock, msg, flags, NULL, 0);
> >  }
> > @@ -2620,6 +2620,15 @@ static int ___sys_recvmsg(struct socket *sock,
> > struct user_msghdr __user *msg,
> >         return err;
> >  }
> >
> > +static bool __sys_whitelisted_cmsghdrs(struct msghdr *msg)
>
> Don't call it __sys*

kk

>
> > +{
> > +       for (struct cmsghdr *cmsg = CMSG_FIRSTHDR(msg); cmsg != NULL;
> > cmsg = CMSG_NXTHDR(message, cmsg))
>
> no var declarations in for, run checkpatch.pl first

kk

>
> > +               if (cmsg->cmsg_level != SOL_UDP || (cmsg->cmsg_type !=
> > UDP_GRO && cmsg->cmsg_type != UDP_SEGMENT))
> > +                       return false;
> > +
> > +       return true;
> > +}
> > +
> >  /*
> >   *     BSD recvmsg interface
> >   */
> > @@ -2630,7 +2639,7 @@ long __sys_recvmsg_sock(struct socket *sock,
> > struct msghdr *msg,
> >  {
> >         if (msg->msg_control || msg->msg_controllen) {
> >                 /* disallow ancillary data reqs unless cmsg is plain data */
> > -               if (!(sock->ops->flags & PROTO_CMSG_DATA_ONLY))
> > +               if (!( sock->ops->flags & PROTO_CMSG_DATA_ONLY ||
>
> extra space after "("
>
> don't forget about brackets when mixing bitwise and logical and/or
>
> It would look better if you'd get rid of outer brackets and propagate !
>
> > __sys_whitelisted_cmsghdrs(msr) ))
> >                         return -EINVAL;
> >         }
> >
>
> --
> Pavel Begunkov