Everything we compile is typically compiled 64-bit. We were running 4.52
(64-bit) and it was fine. This is compiled with Sun Studio 11. It's not
that EXIM doesn't work, because it does (mail flows through as long as
HELO/EHLO is not used.. It's only when a system HELO/EHLO's that it
crumbles. I'm thinking maybe it has to do with libspf2 but then again, I
don't have any ACL's setup... but EXIM still has refernces so that's the
only thing I can think of. We are recompiling right now without libspf2 to
see what happens.
Justin
>From: Markus Mayer <mmayer@???>
>To: Justin Ewing <jl_ewing@???>
>CC: exim-dev@???
>Subject: Re: [exim-dev] Exim 4.62 errors on Solaris
>Date: Fri, 30 Jun 2006 09:07:07 -0700
>
>Hi Justin,
>
>I am maintaining Exim for the Blastwave project. I am actually just
>preparing an Exim 4.62 package for Blastwave.
>
> > I've compiled Exim 4.62 on Solaris using the following flags:
> >
> > -D_REENTRANT -xtarget=ultra -xarch=v9 -xO3
>
>I am suspecting that the problems are related to you compiling Exim as
>64 bit application. Has that worked before (with older versions) or is
>this the first time you are trying?
>
>I am wondering if anybody else has tried 64 bit Exim binaries on other
>platforms/OSes and what their results were.
>
>Which version of Sun Studio are you using to compile Exim?
>
>Do you really *need* 64 bit address space? If it's just about taking
>advantage of UltraSPARC extensions, you could always compile with
>-xarch=v8plusa.
>
> > 3510: sysinfo(SI_HOSTNAME, "baladi", 255) = 7
> > 3510: Incurred fault #6, FLTBOUNDS %pc = 0xFFFFFFFF7E538890
> > 3510: siginfo: SIGSEGV SEGV_MAPERR addr=0xFFFFFFFF80000000
> > 3510: Received signal #11, SIGSEGV [default]
> > 3510: siginfo: SIGSEGV SEGV_MAPERR addr=0xFFFFFFFF80000000
> > 3502: Received signal #18, SIGCLD, in pollsys() [caught]
> > 3502: siginfo: SIGCLD CLD_DUMPED pid=3510 status=0x000B
> > 3502: pollsys(0xFFFFFFFF7FFFD1E0, 3, 0x00000000, 0x00000000) Err#4
>EINTR
>
>The segfault may very well indicate some 64 bit memory access issues. I
>suggest you try 32 bit to see if it goes away.
>
>If it's a 32 bit vs. 64 bit issue, the code obviously should still be
>fixed, but running 32 bit for now might be a good work-around.
>
>Regards,
>Markus
>
>--
>## List details at http://www.exim.org/mailman/listinfo/exim-dev Exim
>details at http://www.exim.org/ ##
_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
