>Date: Tue, 26 Nov 1996 16:24:53 -0500 (EST)
>From: "Jim Spath (Webmaster Jim)" <jspath@mail.bcpl.lib.md.us>

>----  quote ----
>Something to the effect that C-Kermit 6.0, which has been built successfully
>on hundreds of different UNIX and non-UNIX platforms (i.e. combinations of
>OS, OS version, hardware, C compiler and version, etc) has this problem with
>NetBSD (version xxx) using gcc (version xxx), in which, if it is built with
>the -O (optimize) compiler switch, then the call stack is mangled, and after
>a few function calls, a routine returns to never-never land and dumps core.
>In the version I (i.e. you) built, this happens in the setlin() routine in
>ckuus7.c, and the core dump occurs when you give a command like:
>
>  set line hostname:socket
>
>The makefile has been modified to omit optimization, and unoptimized binaries
>-- which seem to work fine -- are being distributed.
>
>- Frank
>---- endquote ----
>
>I'm on NetBSD 1.2, which has gcc V2.72.  This bug also got me on a FreeBSD
>2.0.5 system, which probably has gcc 2.4.6(?).

I have not experienced any problems like this with kermit 6.0; I'm
running one of the recent beta versions.  I use several of kermit's
code-shrinking options (NOCSETS, NOSPL, NODEBUG, &c.), which may hide
the problem.

I am using a -current/i386 system and gcc 2.7.2.1, updated from 2.7.2
using the patches I submitted as PR bin/2827.

>Anyone have suggestions how to isolate an optimization-type bug?

gcc has options (-d[MNDyrxjsLtfcSlgRJdkamp]) to dump output at
whatever stage of the compilation process you like; see gcc(1).  If I
were you, I'd first patch gcc up to 2.7.2.1 and see if the problem
Goes Away.
-- 
Mike Long <mike.long@analog.com>     <URL:http://www.shore.net/~mikel>
VLSI Design Engineer         finger mikel@shore.net for PGP public key
Analog Devices, CPD Division          CCBF225E7D3F7ECB2C8F7ABB15D9BE7B
Norwood, MA 02062 USA       (eq (opinion 'ADI) (opinion 'mike)) -> nil