Re: [PATCH 1/1] net: ioctl: Use kernel memory on protocol ioctl callbacks

[Breno Leitao <leitao@debian.org>]

On Fri, May 19, 2023 at 11:09:29AM -0400, Willem de Bruijn wrote:
> On Fri, May 19, 2023 at 9:59 AM Breno Leitao <leitao@debian.org> wrote:
> >
> > Most of the ioctls to net protocols operates directly on userspace
> > argument (arg). Usually doing get_user()/put_user() directly in the
> > ioctl callback.  This is not flexible, because it is hard to reuse these
> > functions without passing userspace buffers.
> >
> > Change the "struct proto" ioctls to avoid touching userspace memory and
> > operate on kernel buffers, i.e., all protocol's ioctl callbacks is
> > adapted to operate on a kernel memory other than on userspace (so, no
> > more {put,get}_user() and friends being called in the ioctl callback).
> >
> > This changes the "struct proto" ioctl format in the following way:
> >
> >     int                     (*ioctl)(struct sock *sk, int cmd,
> > -                                        unsigned long arg);
> > +                                        int *karg);
> >
> > So, the "karg" argument, which is passed to the ioctl callback, is a
> > pointer allocated to kernel space memory (inside a function wrapper -
> > sock_skprot_ioctl()). This buffer (karg) may contain input argument
> > (copied from userspace in a prep function) and it might return a
> > value/buffer, which is copied back to userspace if necessary. There is
> > not one-size-fits-all format (that is I am using 'may' above), but
> > basically, there are three type of ioctls:
> >
> > 1) Do not read from userspace, returns a result to userspace
> > 2) Read an input parameter from userspace, and does not return anything
> >   to userspace
> > 3) Read an input from userspace, and return a buffer to userspace.
> >
> > The default case (1) (where no input parameter is given, and an "int" is
> > returned to userspace) encompasses more than 90% of the cases, but there
> > are two other exceptions. Here is a list of exceptions:
> >
> > * Protocol RAW:
> >    * cmd = SIOCGETVIFCNT:
> >      * input and output = struct sioc_vif_req
> >    * cmd = SIOCGETSGCNT
> >      * input and output = struct sioc_sg_req
> >    * Explanation: for the SIOCGETVIFCNT case, userspace passes the input
> >      argument, which is struct sioc_vif_req. Then the callback populates
> >      the struct, which is copied back to userspace.
> >
> > * Protocol RAW6:
> >    * cmd = SIOCGETMIFCNT_IN6
> >      * input and output = struct sioc_mif_req6
> >    * cmd = SIOCGETSGCNT_IN6
> >      * input and output = struct sioc_sg_req6
> >
> > * Protocol PHONET:
> >   * cmd == SIOCPNADDRESOURCE | SIOCPNDELRESOURCE
> >      * input int (4 bytes)
> >   * Nothing is copied back to userspace.
> >
> > For the exception cases, functions sock_skproto_ioctl_in{out}() will
> > copy the userspace input, and copy it back to kernel space.
> >
> > The wrapper that prepares the buffer and puts the buffer back to user is
> > sock_skprot_ioctl(), so, instead of calling sk->sk_prot->ioctl(), the
> > callee now calls sock_skprot_ioctl().
> >
> > Signed-off-by: Breno Leitao <leitao@debian.org>
> 
> Overall this looks great to me.

Thanks for the guidance and quick review!

> 
> Thanks for the detailed commit message that lists all exceptions, Bruno.
> 
> Since that is a limited well understood list, I'm not in favor of the
> suggestion to add an explicit length argument that then needs to be
> checked in each callee.
> 
> > +/* Copy 'size' bytes from userspace and return `size` back to userspace */
> > +int sock_skproto_ioctl_inout(struct sock *sk, unsigned int cmd,
> > +                            void __user *arg, size_t size)
> > +{
> > +       void *ptr;
> > +       int ret;
> > +
> > +       ptr = kmalloc(size, GFP_KERNEL);
> > +       if (!ptr)
> > +               return -ENOMEM;
> 
> > +/* A wrapper around sock ioctls, which copies the data from userspace
> > + * (depending on the protocol/ioctl), and copies back the result to userspace.
> > + * The main motivation for this function is to pass kernel memory to the
> > + * protocol ioctl callsback, instead of userspace memory.
> > + */
> > +int sock_skprot_ioctl(struct sock *sk, unsigned int cmd,
> > +                     void __user *arg)
> > +{
> > +#ifdef CONFIG_IP_MROUTE
> > +       if (!strcmp(sk->sk_prot->name, "RAW")) {
> 
> This must check both sk_family and sk_protocol. That is preferable
> over string match.
> 
> For these exception cases, instead of having sock_skproto_ioctl_inout
> dynamically allocate the struct, how about stack allocating them here
> and passing to the function?

Should I stack allocate all the 4 structures sock_skprot_ioctl and pass
them to sock_skproto_ioctl_inout() together with the size? (using the
original name to avoid confusion - will rename in V2)

I mean, writing something as:

int sock_skprot_ioctl(struct sock *sk, unsigned int cmd
                     void __user *arg`
{
	struct sioc_vif_req sioc_vif_req_arg;
	struct sioc_sg_req sioc_sg_req_arg;
	struct sioc_mif_req6 sioc_mif_req6_arg;
	struct sioc_sg_req6 sioc_sg_req6_arg;

	..

	if (!strcmp(sk->sk_prot->name, "RAW6")) {
        switch (cmd) {
               case SIOCGETMIFCNT_IN6:
                       return sock_skproto_ioctl_inout(sk, cmd,
                               arg, &sioc_mif_req6_arg, sizeof(sioc_mif_req6_arg);
               case SIOCGETSGCNT_IN6:
                       return sock_skproto_ioctl_inout(sk, cmd,
                               arg, &sioc_sg_req6_arg, sizeof(sioc_sg_req6_arg));
               }
       }
       ...
}