On Wed, Jun 05, 2019 at 10:52:14PM +0200, Phil Sutter wrote:
> On Wed, Jun 05, 2019 at 07:34:51PM +0200, Pablo Neira Ayuso wrote:
> > Fix bogus error message:
> >
> > # nft monitor
> > Cannot set up netlink socket buffer size to 16777216 bytes, falling back
> > to 16777216 bytes
> >
> > Fixes: bcf60fb819bf ("mnl: add mnl_set_rcvbuffer() and use it")
> > Signed-off-by: Pablo Neira Ayuso <[email protected]>
> > ---
> > src/mnl.c | 2 --
> > 1 file changed, 2 deletions(-)
> >
> > diff --git a/src/mnl.c b/src/mnl.c
> > index c0df2c941d88..a7693ef1de30 100644
> > --- a/src/mnl.c
> > +++ b/src/mnl.c
> > @@ -1433,8 +1433,6 @@ int mnl_nft_event_listener(struct mnl_socket
> > *nf_sock, unsigned int debug_mask,
> >
> > ret = mnl_set_rcvbuffer(nf_sock, bufsiz);
> > if (ret < 0)
> > - nft_print(octx, "# Cannot increase netlink socket buffer size,
> > expect message loss\n");
> > - else
> > nft_print(octx, "# Cannot set up netlink socket buffer size to
> > %u bytes, falling back to %u bytes\n",
> > NFTABLES_NLEVENT_BUFSIZ, bufsiz);
>
> This error message is not correct: If mnl_set_rcvbuffer() returned
> non-zero, both setsockopt() calls failed. The removed message would be
> more appropriate for that situation.
setsockopt() should not fail, I think. So this is unlikely to show up.
events are unreliable anyway, no matter if buffer is set or not.
> BTW: While being at it, maybe s/socket buffer/socket receive buffer/?
Yes I can update this.
Thanks for reviewing.
