*BSD News Article 25164


Return to BSD News archive

Xref: sserve comp.os.386bsd.misc:1737 comp.os.386bsd.questions:7528
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!howland.reston.ans.net!gatech!udel!news.sprintlink.net!demon!awfulhak.demon.co.uk!awfulhak!brian
From: brian@awfulhak.demon.co.uk (Brian Somers)
Newsgroups: comp.os.386bsd.misc,comp.os.386bsd.questions
Subject: Gcc-2.4.5
Date: 17 Dec 93 22:00:27
Organization: Coverform Ltd.
Lines: 24
Message-ID: <BRIAN.93Dec17220027@awfulhak.demon.co.uk>
NNTP-Posting-Host: awfulhak.demon.co.uk
In-reply-to: storm@cs.mcgill.ca's message of 14 Dec 1993 04:28:03 GMT


I know this isn't really a 386BSD thing - but -

I'm putting gcc-2.4.5 (I know 2.4.6 is now around) on a Sequent machine
running ptx.  It worked fairly easily with one problem.

I can't find somewhere to call __do_global_ctors() !

__main() isn't called from crt0.o, and _start() is defined in crt0.o.
There are two more functions - __ppinit and __init_locale, but ppinit is
used when loading from the parallel processing library and init_locale
gets called and core-dumps when I try to set a static variable (initialized).

I think I'm going to have to resort to binary-editing a copy of crt0.o,
changing the 'main' symbol to Main, and writing Main into crtbegin so
that it calls __do_global_ctors and then exit(main(argc,argv)).

I don't want to do this !

Any suggestions ?

Brian.
--
Brian <brian@awfulhak.demon.co.uk>