Thanks, pushed to master, branch-1.4, branch-1.3.
On Thu, Dec 01, 2011 at 02:17:03PM +0900, Jari Sundell wrote:
> Confirmed that this patch fixes the bug for me.
>
> Jari Sundell
>
> On Thu, Dec 1, 2011 at 4:40 AM, Ben Pfaff <b...@nicira.com> wrote:
> > Reported-by: Jari Sundell <sundell.softw...@gmail.com>
> > ---
> > ?lib/learn.c | ? 11 +++++++++++
> > ?1 files changed, 11 insertions(+), 0 deletions(-)
> >
> > diff --git a/lib/learn.c b/lib/learn.c
> > index 19a0e00..9d97cb3 100644
> > --- a/lib/learn.c
> > +++ b/lib/learn.c
> > @@ -621,6 +621,17 @@ learn_format(const struct nx_action_learn *learn,
> > struct ds *s)
> > ? ? ? ? ? ? ? ? union mf_value value;
> > ? ? ? ? ? ? ? ? uint8_t *bytes = (uint8_t *) &value;
> >
> > + ? ? ? ? ? ? ? ?if (src_value_bytes > dst_field->n_bytes) {
> > + ? ? ? ? ? ? ? ? ? ?/* The destination field is an odd number of bytes,
> > which
> > + ? ? ? ? ? ? ? ? ? ? * got rounded up to a multiple of 2 to be put into the
> > + ? ? ? ? ? ? ? ? ? ? * learning action. ?Skip over the leading byte, which
> > + ? ? ? ? ? ? ? ? ? ? * should be zero anyway. ?Otherwise the memcpy() below
> > + ? ? ? ? ? ? ? ? ? ? * will overrun the start of 'value'. */
> > + ? ? ? ? ? ? ? ? ? ?int diff = src_value_bytes - dst_field->n_bytes;
> > + ? ? ? ? ? ? ? ? ? ?src_value += diff;
> > + ? ? ? ? ? ? ? ? ? ?src_value_bytes -= diff;
> > + ? ? ? ? ? ? ? ?}
> > +
> > ? ? ? ? ? ? ? ? memset(&value, 0, sizeof value);
> > ? ? ? ? ? ? ? ? memcpy(&bytes[dst_field->n_bytes - src_value_bytes],
> > ? ? ? ? ? ? ? ? ? ? ? ?src_value, src_value_bytes);
> > --
> > 1.7.2.5
> >
> >
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
