Hi Simon, On Fri, Apr 21, 2023 at 5:57 PM Simon Horman <[email protected]> wrote: > > 1. I see that some users of dp_packet_data() are not modified by this patch. > I did check all of them, but one example is in lib/bfd.c:bfd_process_packet > > Is that because they were not flagged by Coverity in your investigation? > Or for some other reason (I often miss obvious points). >
To be honest, the reason was that they were not flagged by Coverity when I perform my investigation. If you would like me to add the `ovs_assert()` to the `bfd_process_packet()` function, I can do that since that particular case seems to make sense. I am not sure about other users of `dp_packet_data()` though. > 2. Did you consider moving the ovs_assert() into dp_packet_data(). > Perhaps that is more risky because perahps sometimes it can > legitimately be NULL. > > But, OTOH, if it is correct for dp_packet_data() to ever return NULL > then it would nicely cover all cases. > > Unless I am missing something. Which often happens... Unfortunately, I don't think this can be done as it seems that `dp_packet_data()` can legitimately return NULL in some cases. I tested this by simply adding the `ovs_assert()` into `dp_packet_data()` and 676 tests fail. Best regards, James Raphael Tiovalen _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
