----- Original Message -----
From: "Mike Pumford" <mpumford@black-star.demon.co.uk>
To: "Mike Pumford" <mpumford@black-star.demon.co.uk>
Cc: <port-arm@netbsd.org>; <mpumford@black-star.demon.co.uk>
Sent: Sunday, April 07, 2002 12:37 AM
Subject: Re: acorn32, cats, shark ports switch to ELF


>
> > I did have to restore sendmail to an a.out binary as the ELF one
repeatedly
> > failed to send mail reporting rewrite problems. This occured both with
my
> > original sendmail.cf and with a new one generated from the same .mc
file.
> > An a.out sendmail from a couple of weeks ago (which reports as the same
> > version) was perfectly happy with both config files. Anyone got any
ideas
> > what is going on here?

> Oops forgot to put in the exact error messages
> Apr  6 14:27:13 black-star sendmail[813]: g36DRDN00813: SYSERR(root):
rewrite: r
> uleset check_relay: replacement $1 out of bounds
> Apr  6 14:27:13 black-star sendmail[814]: g36DRDN00814: SYSERR(root):
rewrite: r
> uleset canonify: replacement $1 out of bounds
> Apr  6 14:27:14 black-star sendmail[815]: g36DRDN00815: SYSERR(root):
rewrite: r
> uleset canonify: replacement $1 out of bounds
> Apr  6 14:27:14 black-star sendmail[816]: g36DREN00816: SYSERR(root):
rewrite: r
> uleset canonify: replacement $1 out of bounds
>
> The same config file with an a.out sendmail is fine as is an ELF
NetBSD/i386
> sendmail built from the same tree as the ELF arm build.

Snap, got the same errors here as well.  However recompiling with -O instead
of -O2 cured it.  Looks liked there's a codegen bug in the new compiler.  I
believe someone mentioned lint suffers as well.

I may try to narrow this one down at some point this week (IE which file is
the breakage in kinda thing)

Chris