[uClibc]Re: Strange busybox/uClibc problem

Paul Komarek komarek at andrew.cmu.edu
Thu Jun 27 08:36:35 UTC 2002


On Tue, 25 Jun 2002, Robert Schwebel wrote:

> On Tue, Jun 25, 2002 at 02:26:35PM -0400, Paul Komarek wrote:
> > 1) the Soekris net4501 uses an AMD Elan SC520.  The Elan target in kernel
> > the recent 2.4 kernels doesn't seem to work with it, and Configure.help
> > states that the Elan target is for the SC400/SC410.
>
> Unfortunately I don't have a SC520 based system for tests, so it was just
> tested with an SC410. I was told by other users that the timer 0 difference
> and the serial port bug are also there on the SC520; the A20 gate code
> should also work on SC520, so I don't see a reason why it shouldn't work
> there. However, if somebody can do more tests I'm very interested. I have
> some other things in the pipeline, but one of the most important ones is to
> find out on runtime if the CPU is an SC400/410 or SC520 (which I wasn't
> able to implement yet).

I'd be happy to do some testing for you.  Feel free to contact me
off-list.  I don't have the knowledge and time to hunt down why the Elan
kernel didn't boot for me while the i386 did, but I can certainly try out
your ideas for you.

> Chosing the normal i386 architecture should always work (although you have
> to live with the wrong clock then or apply only that hunk of the patch).
>
> > 3) My uClibc toolchain-wrappers (the DEVEL directory) is name
> > i386-linux-uclibc, which I'd hoped indicated that I was targeting i386.
> > However, I didn't set CROSS to anything in Config, and my host machine
> > uses gcc-i686.
>
> You don't need a cross compiler for the Elans.
>
> Robert

As it turned out (maybe you saw the post), specifying -mcpu=i386 made
everything work, finally.  As you stated, I don't need a cross compiler in
the sense of generating assembly that cannot be executed by the host;
however, it was necessary to specify a different architecture than the
default for my compiler (i686-pc-linux-gnu).

It seems to me that Rules.mak or Config.i386 ought to specify -mcpu=i386
somewhere; Rules.mak specifies -march=i386 in the OPTIMIZATION variable,
but nobody specifies -mcpu=i386.  In the end, I grudgingly added it to the
WARNINGS variable.

-Paul Komarek




More information about the uClibc mailing list