On Fri, Jan 04, 2019 at 07:24:08AM -0500, Hal Rosenstock wrote:
> On 1/2/2019 8:13 AM, Honggang Li wrote:
> > node->sw->fdb[to] is uint8_t. So, (node->sw->fdb[to] < 0) is always
> > false.
> >
> > Issue was detected by Coverity.
> >
> > Signed-off-by: Honggang Li <honli@xxxxxxxxxx>
> > ---
> > ibsim/sim_cmd.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/ibsim/sim_cmd.c b/ibsim/sim_cmd.c
> > index e2fe6ba6dddb..1d36c600814f 100644
> > --- a/ibsim/sim_cmd.c
> > +++ b/ibsim/sim_cmd.c
> > @@ -751,8 +751,8 @@ static int dump_route(FILE * f, char *line)
> > break; // found
> > outport = portnum;
> > if (node->type == SWITCH_NODE) {
> > - if ((outport = node->sw->fdb[to]) < 0
> > - || to > node->sw->linearFDBtop)
> > + outport = node->sw->fdb[to];
> > + if (to > node->sw->linearFDBtop)
>
> This is not the proper fix. I think that the check should be == 0xff
> rather than < 0 or eliminated as in this patch since fdb is initialized
> to 0xff (no port for LID).
Yes, 0xff indicates that an invalid port number was used. It should be:
outport = node->sw->fdb[to]) == 0xff
as you said.
Please fix it.
