Re: can't read SMFIC_BODY reply packet header: Success

From: Murray S. Kucherawy <msk_at_blackops.org>
Date: Tue, 29 Nov 2011 06:51:04 -0800 (PST)

On Tue, 29 Nov 2011, Rolf E. Sonneveld wrote:
> In the 'screen' display I got neither a 'Segfault' nor a 'Segfault (core
> dumped)'. Instead, all I saw was:
>
> opendkim: dkim.c:6482: dkim_body: Assertion `dkim != ((void *)0)' failed.

What version of opendkim is this? If I know that, I can get the source
for that version and match up with the assertion. As of the current code,
line 6482 of libopendkim/dkim.c isn't this assertion check.

But this is a useful first step at any rate.

> I used gdb from within another window on the same system. Not sure whether
> this makes any difference from using it from within the screen display?

Nope, should be the same thing.

> #3 0x00007f3d31d82cdf in dkim_body (dkim=<value optimized out>, buf=<value
> optimized out>, buflen=<value optimized out>) at dkim.c:6482

Looks like this was not compiled with "--enable-debug". But we already
know the first value is 0 (NULL) since that's what tripped the assertion.

> #4 0x0000000000409b8c in dkimf_msr_body (ctx=0x17332f0, bodyp=0x7f3d28002840
> "test\r\n", bodylen=6) at opendkim.c:4954

This is very strange, because dkimf_msr_body() has never had "ctx" as a
parameter as far back as I can remember. It looks like gdb is reporting
the same parameter list here as it is for mlfi_body().

> Does this reveal a possible cause of the problem?

It's a start. Did you get this binary from a packaged RPM or did you
build it yourself?

If you attach your configuration, the output of "opendkim -V", and a
sample message that causes the problem, I can try to reproduce it in my
environment. That might get us closer to a solution.

-MSK
Received on Tue Nov 29 2011 - 14:51:24 PST