postfix-users October 2010 archive
Main Archive Page > Month Archives  > postfix-users archives
postfix-users: Re: postscreen segfault

Re: postscreen segfault

From: Vernon A. Fort <vfort_at_nospam>
Date: Wed Oct 06 2010 - 18:40:41 GMT
To: Postfix users <postfix-users@postfix.org>

On Wed, 2010-10-06 at 14:32 -0400, Wietse Venema wrote:
> Vernon A. Fort:
> > I recompiled postfix: unstripped and with -ggdb. I ran gdb postscreen
> > core file - new binary but old core file. May be still useful but will
> > have to wait on another segfault.
> >
> > attached is the backtrace using new binary with old core file.
>
> Assuming that the generated program code/data is the same compiling
> with -ggdb:
>
> #3 0x0805f827 in msg_info (fmt=0x80683bb "%s: notify %s:%s") at msg.c:179
> #4 0x0804c39a in ps_early_dnsbl_event (unused_event=0,
> context=0x88d1348 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\
> 377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\
> 377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\
> 377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\
> 377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\
> 377\377p")
>
> This suggests that ps_early_dnsbl_event() is called with a free()-d
> pointer. The context argument points to 96 bytes with 0xff values,
> which is very close to the 86 bytes for a postscreen PS_STATE data
> structure on 32-bit systems, plus the 8 bytes that the Postfix
> mymalloc() function uses on 32-bit systems for safety checks.
>
> Were you seeing the signal 11 errors before turning on verbose logging?
>
> Wietse

yes, as noted with my previous post. just not as frequent, say one
every 4-5 days. I have two servers using this version. The amd64
install did display a segfault (once) but not since I updated to the
latest postfix-2.8-20100923 version.

The amd64 install is NOT using the -v option - no segfaults for 8 days.
Both server have approximatively the same mail flow.

Also - looking at the backtrace, what is the "vbuf_print.c: No such file
or directory." mean. I initial review (not a programmer) of the
backtrace saw this is a problem.

Vernon