panic: page fault

Matthew Dillon dillon at apollo.backplane.com
Fri Nov 25 08:27:19 PST 2005


:
:Core is being uploaded to leaf.
:
:It looks like this was caused by tcpdump.  From a kernel that's about a month old:
:
:#20 0xc020bd9d in syncache_add (inc=0xdaaa8c08, to=0xdaaa8c78, th=0x0, sop=0xdaaa8c04, m=0xeec30f00) at /usr/src/sys/netinet/tcp_syncache.c:999
:...

     It looks like the zinit for tcp_syncache.bucket_limit was not
     taking into account the number of cpus.   Try the below patch.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>

Index: netinet/tcp_syncache.c
===================================================================
RCS file: /cvs/src/sys/netinet/tcp_syncache.c,v
retrieving revision 1.22
diff -u -r1.22 tcp_syncache.c
--- netinet/tcp_syncache.c	4 Mar 2005 05:57:50 -0000	1.22
+++ netinet/tcp_syncache.c	25 Nov 2005 16:21:28 -0000
@@ -363,7 +363,7 @@
 	 * older one.
 	 */
 	tcp_syncache.zone = zinit("syncache", sizeof(struct syncache),
-	    tcp_syncache.cache_limit, ZONE_INTERRUPT, 0);
+	    tcp_syncache.cache_limit * ncpus2, ZONE_INTERRUPT, 0);
 	tcp_syncache.cache_limit -= 1;
 }
 





More information about the Bugs mailing list