Re: [ECOS] Recommended gcc version for 1.0?

[Grant Edwards <grante at visi dot com>]

On Wed, Aug 06, 2003 at 04:18:22PM +0100, Nick Garnett wrote:

> > When did big-endian ARM gcc users become second-class citizens?
> > I swear that for 2.95, doing a "confgure --target=arm-elf" used
> > to work just fine.
> 
> This was done to keep the number of different versions of the
> libraries down.

Here is where I get confused.

What libraries?

When I build 2.95.2 for use with eCos, the only library that is
built is libgcc.a:

$ rpm -q -l arm-elf-gcc | grep '\.a'
/usr/local/lib/gcc-lib/arm-elf/2.95.2/be/libgcc.a
/usr/local/lib/gcc-lib/arm-elf/2.95.2/be/nofmult/libgcc.a
/usr/local/lib/gcc-lib/arm-elf/2.95.2/libgcc.a
/usr/local/lib/gcc-lib/arm-elf/2.95.2/nofmult/libgcc.a

Eliminating big-endian would cut that from four files to two.

There are now 21 different libgcc.a files in my 3.2.1
toolchain. Big endian support constitutes 8 of those files.

However, building those files didn't take a noticable amount of
time. When I built 3.2.1, it looked like it spent most of its
time building general libc and libm type stuff.  Doesn't eCos
provides it's own libc and libm libraries?

> Each new option doubles the number that have to be built. If
> you are not careful you get an exponential explosion. Since
> very few of our supported ARM boards are big-endian, we elected
> to leave that option out of the standard build.

Ah. I thought this was something done by the GCC guys. The
default --target=arm-elf now seems to build only little-endian
libgcc.a, when it used to build both.

> At least that is my understanding. I expect a toolchain expert
> will be along in a minute to provide any corrections.

-- 
Grant Edwards
grante@visi.com

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss