svn commit: r264987 - head/sys/net
Ian Lepore
ian at FreeBSD.org
Sun Apr 27 02:38:48 UTC 2014
On Sun, 2014-04-27 at 06:23 +0400, Alexander V. Chernikov wrote:
> On 27.04.2014 03:16, Ian Lepore wrote:
> > On Sat, 2014-04-26 at 22:32 +0000, Alexander V. Chernikov wrote:
> >> Author: melifaro
> >> Date: Sat Apr 26 22:32:04 2014
> >> New Revision: 264987
> >> URL: http://svnweb.freebsd.org/changeset/base/264987
> >>
> >> Log:
> >> Determine fibnum once in the beginning of route_output().
> >>
> >> MFC after: 1 month
> >>
> >> Modified:
> >> head/sys/net/rtsock.c
> >
> >> Modified: head/sys/net/rtsock.c
> >> ==============================================================================
> >> --- head/sys/net/rtsock.c Sat Apr 26 21:03:41 2014 (r264986)
> >> +++ head/sys/net/rtsock.c Sat Apr 26 22:32:04 2014 (r264987)
> >> @@ -526,7 +526,7 @@ route_output(struct mbuf *m, struct sock
> >> struct sockaddr_in6 *sin6;
> >> int i, rti_need_deembed = 0;
> >> #endif
> >> - int len, error = 0;
> >> + int len, error = 0, fibnum;
> >> struct ifnet *ifp = NULL;
> >> union sockaddr_union saun;
> >> sa_family_t saf = AF_UNSPEC;
> >> @@ -582,6 +582,8 @@ route_output(struct mbuf *m, struct sock
> >> senderr(error);
> >> }
> >>
> >> + fibnum = so->so_fibnum;
> >> +
> >> /*
> >> * The given gateway address may be an interface address.
> >> * For example, issuing a "route change" command on a route
> >> @@ -596,7 +598,7 @@ route_output(struct mbuf *m, struct sock
> >>
> >> bzero(&gw_ro, sizeof(gw_ro));
> >> gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY];
> >> - rtalloc_ign_fib(&gw_ro, 0, so->so_fibnum);
> >> + rtalloc_ign_fib(&gw_ro, 0, fibnum);
> >> /*
> >> * A host route through the loopback interface is
> >> * installed for each interface adddress. In pre 8.0
> >> @@ -637,7 +639,7 @@ route_output(struct mbuf *m, struct sock
> >> break;
> >> }
> >> error = rtrequest1_fib(rtm->rtm_type, &info, &saved_nrt,
> >> - so->so_fibnum);
> >> + fibnum);
> >> if (error == 0 && saved_nrt != NULL) {
> >> #ifdef INET6
> >> rti_need_deembed = (V_deembed_scopeid) ? 1 : 0;
> >> @@ -663,8 +665,7 @@ route_output(struct mbuf *m, struct sock
> >> #endif
> >> break;
> >> }
> >> - error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt,
> >> - so->so_fibnum);
> >> + error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt, fibnum);
> >> if (error == 0) {
> >> RT_LOCK(saved_nrt);
> >> rt = saved_nrt;
> >> @@ -677,8 +678,7 @@ route_output(struct mbuf *m, struct sock
> >> break;
> >>
> >> case RTM_GET:
> >> - rnh = rt_tables_get_rnh(so->so_fibnum,
> >> - info.rti_info[RTAX_DST]->sa_family);
> >> + rnh = rt_tables_get_rnh(fibnum, saf);
> >> if (rnh == NULL)
> >> senderr(EAFNOSUPPORT);
> >>
> >> @@ -867,7 +867,7 @@ flush:
> >> m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
> >> }
> >> if (m) {
> >> - M_SETFIB(m, so->so_fibnum);
> >> + M_SETFIB(m, fibnum);
> >> m->m_flags |= RTS_FILTER_FIB;
> >> if (rp) {
> >> /*
> >>
> >
> > I've got build breakage that I think may be from this change...
> >
> > /local/build/staging/freebsd/head/src/sys/net/rtsock.c: In function 'route_output':
> > /local/build/staging/freebsd/head/src/sys/net/rtsock.c:529: warning: 'fibnum' may be used uninitialized in this function
> Fixed in r264998, thank you!
> Twice per night is a kind of record for me :(
> I'm wondering why clang hasn't seen any issue here..
>
I've been doing a bunch of arm universe kernel builds that bring both
clang and gcc into play; they do sometimes whine about different things.
-- Ian
More information about the svn-src-head
mailing list