www.delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/1998/08/24/04:33:02

Date: Mon, 24 Aug 1998 10:49:11 +0200 (WET)
From: Andris Pavenis <pavenis AT lanet DOT lv>
To: Nate Eldredge <nate AT cartsys DOT com>
cc: Bill Currie <bill AT taniwha DOT tssc DOT co DOT nz>, djgpp AT delorie DOT com
Subject: Re: Linux and GNU C
In-Reply-To: <35DF0C4C.7AF2ADA1@cartsys.com>
Message-ID: <Pine.A32.3.91.980824103925.40130C-100000@ieva01.lanet.lv>
MIME-Version: 1.0


On Sat, 22 Aug 1998, Nate Eldredge wrote:

> Bill Currie wrote:
> > 
> > Thibaut Murez wrote:
> > > Hi I'm a newbie in Linux' world and i'd like to know why one must not
> > > compile the kernel using gcc 2.8 or higher ?
> > 
> > There's a bug in one of the kernel header files that works with
> > gcc-2.7.2.x but not with gcc-2.8+.  This only applies to 2.0.x kernels
> > and 2.1.x below 90 or so (can't remember the number, but I rembeber the
> > fuss:).
> 
> You can easily work around it by adding `volatile' to the declaration of
> the `eflags' arg of `sys_iopl' in `arch/i386/kernel/ioport.c'.  I did
> that to my 2.0.x kernel and never looked back.  In fact, I don't know
> why they don't add it to the source tree-- it's well known.
> -- 

Yes. The first impression is that kernel works Ok after such change.
egcs-1.0.3 does not require even this (however to avoid many warnings
one include file should be edditted). However are You sure there 
are no other more hidden problems (Broken iopl() is very well visible).

Even if You don't have problems (and I didn't have) there may be something
with things we don't use. The question here is if we want to live
dangerously: if so then use of new compilers is acceptable. Otherwise
as I think one should stay with version used to develop 2.0.XX and it is
gcc-2.7.2.X.

Andris

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019