paul@hyperlink.net.au writes:
>Firstly - I am running a Slackware 3.1 distribution of Linux with kernel
>2.0.27. This distribution has gcc 2.7.2, and libc-5.3.12. This proxy server
>server services relatively light load, our squid.conf was pretty much stock
>except acl's and other local stuff. The proxy server ran without a hickup
>for 6 days an then started coredumping and dying about twice a day. Cache
You may be seeing a coredump bug from the IP cache. This bug was introduced
by me in version 1.1.2. There is an extra call to ipcache_release()
which causes free-memory-reads. Here's a patch:
Index: ipcache.c
===================================================================
RCS file: /surf1/CVS/squid/src/ipcache.c,v
retrieving revision 1.92
retrieving revision 1.94
diff -w -u -r1.92 -r1.94
--- ipcache.c 1996/12/21 00:43:52 1.92
+++ ipcache.c 1997/01/03 22:45:49 1.94
@@ -1,5 +1,5 @@
/*
- * $Id: ipcache.c,v 1.92 1996/12/21 00:43:52 wessels Exp $
+ * $Id: ipcache.c,v 1.94 1997/01/03 22:45:49 wessels Exp $
*
* DEBUG: section 14 IP Cache
* AUTHOR: Harvest Derived
@@ -686,8 +686,7 @@
ipcacheAddPending(i, fd, handler, handlerData);
if (squid_curtime - i->expires > 60) {
i->status = IP_NEGATIVE_CACHED;
- ipcache_call_pending(i);
- ipcache_release(i);
+ ipcache_call_pending(i); /* will also release it */
}
return;
} else {
Duane W.
Received on Sun Jan 05 1997 - 20:21:59 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:59 MST