bin/187526: [patch] traceroute -a breaks of "whois" socket timeout

[Eugene Grosbein]

>Number:         187526
>Category:       bin
>Synopsis:       [patch] traceroute -a breaks of "whois" socket timeout
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Mar 13 11:00:00 UTC 2014
>Closed-Date:
>Last-Modified:
>Originator:     Eugene Grosbein
>Release:        FreeBSD 8.4-STABLE amd64
>Organization:
RDTC JSC
>Environment:
System: FreeBSD eg.sd.rdtc.ru 8.4-STABLE FreeBSD 8.4-STABLE #3 r251830M: Mon Sep 30 18:01:03 NOVT 2013 root at eg.sd.rdtc.ru:/usr/local/obj/usr/local/src/sys/EG amd64

>Description:
	"traceroute -a" opens TCP socket to whois server and fails to check
	if this socket stays valid during (sometimes long) runtime.
	If whois server closes this connection, traceroute gets EOF at next
	hop (and shows AS0) and is killed with SIGPIPE later.

>How-To-Repeat:

$ traceroute -Ia 194.150.149.239
traceroute to 194.150.149.239 (194.150.149.239), 64 hops max, 72 byte packets
 1  [AS29072] 62.231.161.217 (62.231.161.217)  0.276 ms  0.268 ms  0.189 ms
 2  [AS29072] k-45.rdtc.ru (62.231.160.42)  1.169 ms  0.496 ms  0.706 ms
 3  [AS29072] asbr.rdtc.ru (62.231.191.137)  0.467 ms  0.548 ms  0.498 ms
 4  [AS31133] 37.29.3.49 (37.29.3.49)  1.423 ms  1.650 ms  1.374 ms
 5  * * *
 6  * * *
 7  * * *
 8  * * *
 9  * * *
10  * * *
11  [AS0] 37.29.107.202 (37.29.107.202)  70.641 ms  70.599 ms  70.525 ms

	Here traceroute is killed. Instead, it should show complete trace:

11  [AS31133] 37.29.107.202 (37.29.107.202)  104.550 ms  70.603 ms  72.888 ms
12  [AS12389] 95.167.91.93 (95.167.91.93)  166.594 ms  65.702 ms  65.790 ms
13  [AS12389] customer-as41440.xe-0-2-0.brnl-rgr3.sib.ip.rostelecom.ru (188.254.36.146)  68.412 ms  68.322 ms  68.287 ms
14  [AS41440] ge-0-0-0-v501.bar-csr1.ncc.sibirtelecom.ru (213.228.118.6)  85.369 ms  101.501 ms  85.300 ms
15  [AS12846] 212.94.98.54 (212.94.98.54)  85.021 ms  85.116 ms  85.004 ms
16  [AS12846] Bsk-ATS24-c7301.mss.ab.ru (212.94.101.129)  74.240 ms  74.504 ms  74.214 ms
17  [AS12846] adsl-149-239.biysk.ru (194.150.149.239)  103.236 ms  103.358 ms  103.013 ms

>Fix:

	A workaround is to lower wait time for proble timeout using '-w 1' option.

	The following patch fixes the problem introducing proper error checking
	while reading/writing "whois" connection. In case of error it reconnects.

--- as.h.orig	2013-06-17 11:18:23.000000000 +0700
+++ as.h	2014-03-13 17:13:48.000000000 +0700
@@ -31,5 +31,5 @@
  */
 
 void *as_setup(const char *);
-unsigned int as_lookup(void *, char *, sa_family_t);
+unsigned int as_lookup(void *, char *, sa_family_t, int *);
 void as_shutdown(void *);
--- as.c.orig	2013-06-17 11:18:23.000000000 +0700
+++ as.c	2014-03-13 17:37:51.000000000 +0700
@@ -119,7 +119,7 @@ as_setup(const char *server)
 }
 
 unsigned int
-as_lookup(void *_asn, char *addr, sa_family_t family)
+as_lookup(void *_asn, char *addr, sa_family_t family, int *status)
 {
 	struct aslookup *asn = _asn;
 	char buf[1024];
@@ -129,8 +129,17 @@ as_lookup(void *_asn, char *addr, sa_fam
 	as = 0;
 	rc = dlen = 0;
 	plen = (family == AF_INET6) ? 128 : 32;
-	(void)fprintf(asn->as_f, "!r%s/%d,l\n", addr, plen);
-	(void)fflush(asn->as_f);
+	*status = fprintf(asn->as_f, "!r%s/%d,l\n", addr, plen);
+	if (*status < 0) {
+		*status = errno;
+		return 0;
+	}
+	*status = fflush(asn->as_f);
+	if (*status == EOF) {
+		*status = errno;
+		return 0;
+	}
+	*status = 0;
 
 #ifdef AS_DEBUG_FILE
 	if (asn->as_debug) {
@@ -139,7 +148,14 @@ as_lookup(void *_asn, char *addr, sa_fam
 	}
 #endif /* AS_DEBUG_FILE */
 
-	while (fgets(buf, sizeof(buf), asn->as_f) != NULL) {
+	while (1) {
+		if (fgets(buf, sizeof(buf), asn->as_f) == NULL) {
+			if(feof(asn->as_f) || ferror(asn->as_f)) {
+				*status = EIO;
+				return 0;
+			}
+			break;
+		}
 		buf[sizeof(buf) - 1] = '\0';
 
 #ifdef AS_DEBUG_FILE
--- traceroute.c.orig	2013-06-17 11:18:23.000000000 +0700
+++ traceroute.c	2014-03-13 17:27:14.000000000 +0700
@@ -931,6 +931,8 @@
 			as_path = 0;
 		}
 	}
+	if (as_path)
+		signal(SIGPIPE, SIG_IGN);
 	
 #if	defined(IPSEC) && defined(IPSEC_POLICY_IPSEC)
 	if (setpolicy(sndsock, "in bypass") < 0)
@@ -1471,6 +1473,7 @@
 {
 	register struct ip *ip;
 	register int hlen;
+	int as, status;
 	char addr[INET_ADDRSTRLEN];
 
 	ip = (struct ip *) buf;
@@ -1479,8 +1482,24 @@
 
 	strlcpy(addr, inet_ntoa(from->sin_addr), sizeof(addr));
 
-	if (as_path)
-		Printf(" [AS%u]", as_lookup(asn, addr, AF_INET));
+	while(as_path) {
+		as = as_lookup(asn, addr, AF_INET, &status);
+		if (status) {
+			as_shutdown(asn);
+			asn = as_setup(as_server);
+			if (asn == NULL) {
+				Fprintf(stderr, "%s: as_setup failed, AS# lookups"
+						" disabled\n", prog);
+				(void)fflush(stderr);
+				as_path = 0;
+				break;
+			}
+			else
+				continue;
+		}
+		Printf(" [AS%u]", as);
+		break;
+	}
 
 	if (nflag)
 		Printf(" %s", addr);
>Release-Note:
>Audit-Trail:
>Unformatted: