Hi,
On 14/07/2019 22:02, Gert Doering wrote:
> It *must not* set the ON_LINK flag for routes that are behind a gatway
> ("route to gateway" vs. "route to interface"), but it still does.
> [SKIP SKIP] > > as you can see, it now insists on "everything is ON_LINK". > interesting to note that on my test machine I don't see this behaviour. ON_LINK is returned only when appropriate. > > Now... I think the issue might be just that you removed the > > - CLEAR(*rgi6); > > bit, so "there is cruft in that structure, and it is flagged as > 'the ON_LINK bit is set'" - but it's too late for me to do a full > code review, so can't say for sure. If I just add it back, it > behaves correctly, but it would be good to give this function another > long and hard stare... > This makes a lot of sense: that CLEAR() statement has been removed by accident. And honestly this missing initialization is somewhat compatible with not always seeing the problem (initialization of stack variables may differ on different platforms based on compiler version, kernel options, etc..). Actually I also realized that the 'gw' variable introduced next to this line was not used, so I have got rid of it. > However - as it is today, I need to NAK this. Sorry. (And I should > have been much quicker) No worries. v5 incoming!! Cheers, > > gert > > (PS: FreeBSD is, as expected, showing the same things before/after, with > ON_LINK only for test 3+4 - but that's a totally different code path, so > no surprises here) > -- Antonio Quartulli
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Openvpn-devel mailing list Openvpn-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openvpn-devel
