segfaults with 2.1.11

[Andreas Hasenack]

On Mon, Feb 14, 2005 at 11:06:39AM -0200, Andreas Hasenack wrote:
> After upgrading to 2.1.11 this morning I started to get random segfaults when
> switching mailboxes (closing one and opening the other):
> 
> #0  index_fetchreply (mailbox=0x8130f80, msgno=28, rock=0xbfffc410) at index.c:2371
> #1  0x080600ad in index_forsequence (mailbox=0x8130f80, sequence=0x81450d5 "", usinguid=0, proc=0x80632e0 <index_fetchreply>,
>     rock=0xbfffc410, fetchedsomething=0xbfffc390) at index.c:1524
> #2  0x080514a0 in cmd_fetch (tag=0x8144fa0 "a0008", sequence=0x2ad0fcb0 <Address 0x2ad0fcb0 out of bounds>, usinguid=0) at imapd.c:3092
> #3  0x0805cf06 in cmdloop () at imapd.c:1081
> #4  0x0805f69b in service_main (argc=1, argv=0x813b008, envp=0xbfffec0c) at imapd.c:689
> #5  0x0804c8f7 in main (argc=1, argv=0x5c93f, envp=0xbfffec0c) at service.c:530
> 
> I'm using GSSAPI over TLS and mutt as a client.

FYI, reverting the change from 2.2.10 -> 2.2.11 in index.c stopped the segfaults here (for
now at least, I didn't check all the implications yet):

--- cyrus-imapd-2.2.11/imap/index.c~	2005-01-24 01:06:51.000000000 -0200
+++ cyrus-imapd-2.2.11/imap/index.c	2005-02-14 11:19:53.515226784 -0200
@@ -207,7 +207,8 @@
 	seendb = 0;
     }
     if (index_len) {
-	/* Let caller call mailbox_close() to unmap */
+	map_free(&index_base, &index_len);
+	map_free(&cache_base, &cache_len);
 	index_len = cache_end = 0;
     }
 }

---
Cyrus Home Page: http://asg.web.cmu.edu/cyrus
Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html