svn commit: r264987 - head/sys/net

[Alexander V. Chernikov]

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..

> 
> -- Ian
> 
> 
> 
>