Milter-greylist

Hello,

I am using 3.1.4, but I got crashes earlier too. Decided to troubleshoot it.
How can I run milter-greylist from gdb? I tried gdb ./milter-greylist,
r -P /var/milter-greylist/milter-greylist.pid -p /var/milter-greylist/milter-greylist.sock -D,
but it stops after the first email (when it wants to print "(unknown id): addr
superafu.de[81.169.170.70] from <xxx@...> to <xxx@...> delayed
for 00:02:00 (ACL 103)" I assume). So I just ran
./milter-greylist -P /var/milter-greylist/milter-greylist.pid -p /var/milter-greylist/milter-greylist.sock -D
After about half a day it crashed:
*** glibc detected *** double free or corruption (!prev): 0x0a5441a8 ***
Aborted

How can I troubleshoot it?

Regards,
Nerijus

Nerijus Baliunas <nerijus@...> wrote:

> I am using 3.1.4, but I got crashes earlier too. Decided to troubleshoot
> it. How can I run milter-greylist from gdb? I tried gdb ./milter-greylist,
> r -P /var/milter-greylist/milter-greylist.pid -p
> /var/milter-greylist/milter-greylist.sock -D, but it stops after the first
> email (when it wants to print "(unknown id): addr
> superafu.de[81.169.170.70] from <xxx@...> to <xxx@...> delayed for
> 00:02:00 (ACL 103)" I assume).

How does it stops? 

> So I just ran ./milter-greylist -P
> /var/milter-greylist/milter-greylist.pid -p
> /var/milter-greylist/milter-greylist.sock -D After about half a day it
> crashed: *** glibc detected *** double free or corruption (!prev):
> 0x0a5441a8 *** Aborted
> 
> How can I troubleshoot it?

Do you have a core dump? If you do, you can open it from gdb. Issue help
target to discover the command to open a core. On my favourite system, I
do this like this:

$ gdb milter-greylist
(gdb) target netbsd-core milter-greylist.core

And then I can run bt to see where it crashed.

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu@...

On Tue, 23 Jan 2007 23:40:52 +0100 manu@... wrote:

> > I am using 3.1.4, but I got crashes earlier too. Decided to troubleshoot
> > it. How can I run milter-greylist from gdb? I tried gdb ./milter-greylist,
> > r -P /var/milter-greylist/milter-greylist.pid -p
> > /var/milter-greylist/milter-greylist.sock -D, but it stops after the first
> > email (when it wants to print "(unknown id): addr
> > superafu.de[81.169.170.70] from <xxx@...> to <xxx@...> delayed for
> > 00:02:00 (ACL 103)" I assume).
> 
> How does it stops? 

Program received signal SIGTTOU, Stopped (tty output). But I found out that it
happens only when I run it from mc shell, it works when I run it from pure bash.

> > So I just ran ./milter-greylist -P
> > /var/milter-greylist/milter-greylist.pid -p
> > /var/milter-greylist/milter-greylist.sock -D After about half a day it
> > crashed: *** glibc detected *** double free or corruption (!prev):
> > 0x0a5441a8 *** Aborted
> > 
> > How can I troubleshoot it?
> 
> Do you have a core dump?

No, even after ulimit -c unlimited it didn't produce core file. But as now
I was able to run milter-greylist under gdb:

[Thread -1239950416 (LWP 7915) exited]
[New Thread -1239950416 (LWP 7920)]
(unknown id): addr 69.90.101.137 from <journal@...> rcpt <gintaras@...>: autowhitelisted for 72:00:00
*** glibc detected *** double free or corruption (!prev): 0x099f3430 ***

Program received signal SIGABRT, Aborted.
[Switching to Thread -1239950416 (LWP 7920)]
0x00b8a7a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
(gdb) bt
#0  0x00b8a7a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
#1  0x00bca7a5 in raise () from /lib/tls/libc.so.6
#2  0x00bcc209 in abort () from /lib/tls/libc.so.6
#3  0x00bfe71a in __libc_message () from /lib/tls/libc.so.6
#4  0x00c04fbf in _int_free () from /lib/tls/libc.so.6
#5  0x00c0533a in free () from /lib/tls/libc.so.6
#6  0x0804a829 in mlfi_close (ctx=0x99f3430) at milter-greylist.c:1036
#7  0x0805e788 in mi_engine ()
#8  0x0805ccd6 in mi_handle_session ()
#9  0x0805c394 in mi_thread_handle_wrapper ()
#10 0x00d00371 in start_thread () from /lib/tls/libpthread.so.0
#11 0x00c6affe in clone () from /lib/tls/libc.so.6

Anything else I should check? I didn't exit from gdb yet.

Regards,
Nerijus

Nerijus Baliunas <nerijus@...> wrote:

> (gdb) bt
> #0  0x00b8a7a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
> #1  0x00bca7a5 in raise () from /lib/tls/libc.so.6
> #2  0x00bcc209 in abort () from /lib/tls/libc.so.6
> #3  0x00bfe71a in __libc_message () from /lib/tls/libc.so.6
> #4  0x00c04fbf in _int_free () from /lib/tls/libc.so.6
> #5  0x00c0533a in free () from /lib/tls/libc.so.6
> #6  0x0804a829 in mlfi_close (ctx=0x99f3430) at milter-greylist.c:1036
> #7  0x0805e788 in mi_engine ()
> #8  0x0805ccd6 in mi_handle_session ()
> #9  0x0805c394 in mi_thread_handle_wrapper ()
> #10 0x00d00371 in start_thread () from /lib/tls/libpthread.so.0
> #11 0x00c6affe in clone () from /lib/tls/libc.so.6
> 
> Anything else I should check? I didn't exit from gdb yet.

frame 6
You should see you crashed on:
                free(priv->priv_buf);

You are using content filtering (body or header clause), right?

Then apply this patch and tell me that it does not happen anymore.

Index: milter-greylist.c
===================================================================
RCS file: /milter-greylist/milter-greylist/milter-greylist.c,v
retrieving revision 1.156
diff -U4 -r1.156 milter-greylist.c
--- milter-greylist.c   18 Jan 2007 15:04:28 -0000      1.156
+++ milter-greylist.c   26 Jan 2007 20:27:12 -0000
@@ -727,8 +727,9 @@
        /* Gather data saved from a previous call */
        if (priv->priv_buf) {
                memcpy(b->b_lines, priv->priv_buf, priv->priv_buflen);
                free(priv->priv_buf);
+               priv->priv_buf = NULL;
        }
        memcpy(b->b_lines + priv->priv_buflen, chunk, i + 1);
        b->b_lines[linelen] = '\0';
        priv->priv_buflen = 0;


-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu@...

On Fri, 26 Jan 2007 21:30:14 +0100 manu@... wrote:

> > Anything else I should check? I didn't exit from gdb yet.
> 
> frame 6
> You should see you crashed on:
>                 free(priv->priv_buf);

Yes.

> You are using content filtering (body or header clause), right?

I use MailScanner, which modifies Postfix queue files.

> Then apply this patch and tell me that it does not happen anymore.

Yes, it didn't crash for 2 days, so it seems it is fixed, thanks!

Regards,
Nerijus