Re: [RFC] sandbox: prevent segfault in tap_alloc()

[Antony Pavlov <antonynpavlov@gmail.com>]

On Wed, 26 Jun 2019 09:12:02 +0200
Sascha Hauer <s.hauer@pengutronix.de> wrote:

> Hi Antony,
> 
> On Wed, Jun 19, 2019 at 01:49:16PM +0300, Antony Pavlov wrote:
> > Tap network interface initialization in sandbox
> > barebox leads to segfault under Debian Buster/Sid.
> > 
> > The problem is that strcpy(dev, ifr.ifr_name) inside
> > tap_alloc() tries to alter read-only data passed
> > by tap_probe() and barebox receives SIGSEGV.
> > 
> > Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
> > ---
> >  drivers/net/tap.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/net/tap.c b/drivers/net/tap.c
> > index 1fbfa085b1..d7e32f4875 100644
> > --- a/drivers/net/tap.c
> > +++ b/drivers/net/tap.c
> > @@ -14,7 +14,7 @@
> >  
> >  struct tap_priv {
> >  	int fd;
> > -	char *name;
> > +	char name[128];
> >  };
> >  
> >  static int tap_eth_send(struct eth_device *edev, void *packet, int length)
> > @@ -65,7 +65,7 @@ static int tap_probe(struct device_d *dev)
> >  	int ret = 0;
> >  
> >  	priv = xzalloc(sizeof(struct tap_priv));
> > -	priv->name = "barebox";
> > +	strncpy(priv->name, "barebox", sizeof(priv->name));
> >  
> >  	priv->fd = tap_alloc(priv->name);
> 
> Can we change the prototype of tap_alloc() to something like this:
> 
> int tap_alloc(const char *name, int *fd, char **outname);
> 
> outname would be an allocated string to be freed by the caller.

There is one problem.
tap_alloc works in the sandbox "os domain" (glibc *alloc&free etc),
the caller works in the "barebox domain" (barebox *alloc&free).

Can we just drop this outname?
nobody actually use it at the moment.


-- 
Best regards,
  Antony Pavlov

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox