On Mon, Jun 15, 2020 at 5:56 PM Tobias Brunner <tob...@strongswan.org> wrote:
>
> Hi Xin,
>
> > To fix this duplicated policies issue, and also fix the issue in
> > commit ed17b8d377ea ("xfrm: fix a warning in xfrm_policy_insert_list"),
> > when doing add/del/get/update on user interfaces, this patch is to change
> > to look up a policy with both mark and mask by doing:
> >
> > mark.v == pol->mark.v && mark.m == pol->mark.m
>
> Looks good, thanks a lot for your work on this. All tests in our
> regression test suite complete successfully with this patch applied.
>
> Tested-by: Tobias Brunner <tob...@strongswan.org>
>
> > and leave the check:
> >
> > ((mark.v & mark.m) & pol->mark.m) == pol->mark.v.
> >
> > for tx/rx path only.
>
> If you are referring to the check in xfrm_policy_match() it's actually:
>
> (fl->flowi_mark & pol->mark.m) != pol->mark.v
>
> Or more generically something like:
>
> (mark & pol->mark.m) == pol->mark.v
>
> As we only have the mark on the packets/flow (no mask) to match against.
>
> > -static bool xfrm_policy_mark_match(struct xfrm_policy *policy,
> > +static bool xfrm_policy_mark_match(const struct xfrm_mark *mark,
> > struct xfrm_policy *pol)
> > {
> > - if (policy->mark.v == pol->mark.v &&
> > - policy->priority == pol->priority)
> > - return true;
> > -
> > - return false;
> > + return mark->v == pol->mark.v && mark->m == pol->mark.m;
> > }
>
> I guess you could make that function `static inline`.
>
Thanks, Tobias, I will post v2 with your suggestion.
Just note that I have another patch similar to this one,
but for xfrm_state's mark. I will post it later too.
Please also check if it may cause any regression.
