This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: egcs-1.1 and linux-2.0.35
- To: Carlo Wood <carlo at runaway dot xs4all dot nl>
- Subject: Re: egcs-1.1 and linux-2.0.35
- From: Toon Moene <toon at moene dot indiv dot nluug dot nl>
- Date: Mon, 31 Aug 98 12:04:56 +0200
- Cc: egcs at cygnus dot com (egcs at cygnus dot com)
- Organization: Moene Computational Physics, Maartensdijk, The Netherlands
- References: <199808310113.DAA11061@jolan.ppro>
> Today I compiled linux kernel 2.0.35 on a i686 with
> egcs-2.91.54, but for my i486: Configured for a 486, most
> notably, using -m486 thus. Nevertheless, XF86_SVGA
> segfaulted.
>
> I then compiled with gcc-2.7.3 the same kernel (on the
> 486, but using the same source tree nfs mounted), and
> then it worked.
[ You probably mean gcc-2.7.2.3 ]
> Does this mean:
>
> 1) A 2.0 kernel compiled with egcs never works.
>
> or
>
> 2) A 2.0 kernel compiled with egcs on a pentiumpro (but
> for a 486)
> doesn't work, but using egcs on the 486 itself might
> still work correctly. (I didn't try this).
Alan Cox, who does the Linux 2.0.x release management, basically
said that he doesn't want to spend time on searching and eradicating
all places where the 2.0 kernels are wrong w.r.t. either egcs or
gcc-2.8.
So your best bet is to keep a gcc-2.7.2.3 around for compiling 2.0
kernels.
On the 2.1.x series, egcs _should_ be doing fine. If a kernel in
that series doesn't compile, or doesn't work, there are three
possibilities:
1. There's a bug or a shortcoming in egcs (like the regparm issue).
2. There's a bug in an asm used by the kernel.
3. There's a mismatch between the documentation of egcs and its
real behaviour (e.g. the "asm-without-explicit-output-registers
-is-supposed-to-be-volatile" issue December last year).
Hoop dat je hier wat aan hebt.
Toon.