linux-mips-fnet
[Top] [All Lists]

Re: Low-res problem idea

To: linux-mips@fnet.fr
Subject: Re: Low-res problem idea
From: Systemkennung Linux <linux@mailhost.uni-koblenz.de>
Date: Sat, 25 Nov 1995 00:59:20 +0100 (MET)
In-reply-to: <199511240650.BAA26737@huahaga.rutgers.edu> from "David S. Miller" at Nov 24, 95 01:50:17 am
Hi all,

>    From: zaitcev@ithil.mcst.ru
>    Date: Fri, 24 Nov 1995 09:36:52 +0300
> 
>    > Linux doesn't yet handle big endian machines very good.  Especially
>    > the filesystem made serious trouble.  The worst thing is that you cannot
>    > easily build filesystem images on your PC to boot your Linux machine.
> 
>    This is not the case for a two or three days already. The hottest news is
>    that gcc compiles object code on SPARC with i386-compatible ext2 
> filesystem.
>    But I have not done this myself yet and rely on i386 with a cross-compiler.
> 
> 
> Peter is correct, I wrote the necessary byte swapping / bit-swapping
> code for the ext2 filesystem sources in the kernel.  The code
> basically turns into nops on little endian machines, but does the
> correct thing on big endian architectures.
> 
> A copy of my tools which create little endian ext2 filesystems on big
> endian sparc machines can be gotten from:
> 
> vger.rutgers.edu:/pub/linux/Sparc/ext2fs/e2fsprogs.tar.gz
> 
> I have built two kernels under sparclinux on these filesystems, so
> things seem ok.

Thanks for the info; that's really interesting news.  I guess this will not
only help the Sparc and m68k people, but also people that intend like Larry
to port Linux/MIPS to big endian machines.

How about setting up a new mailing list exclusivly dedicated to the purpose
of porting Linux to other architectures?  I think in general there is often
a bit a lack of comunication between the Linux peoples about such issues.
Recently the were a few email discussions about portability topics.
Unfortunately not all the people that should had interest in the discussion
were involved and I think a mailing list would be a way to solve the
problem.  I think we could setup such a list with *fast* answer times on
either fnet.fr (Stoned, Luc?) or on phil.uni-sb.de.

  Ralf

<Prev in Thread] Current Thread [Next in Thread>