Re: ct-ng x86_64-host, but want ia32 binaries

["Yann E. MORIN" <yann dot morin dot 1998 at anciens dot enib dot fr>]

Doug,
All,

On Thursday 29 January 2009 20:17:03 Doug Reiland wrote:
> > From: yann.morin.1998@anciens.enib.fr
> I understand. I was hoping others have done something like this since is
> x86_64 has -m32.

x86_64 and x86 are too different architectures. You can't simply see
x86_64 as a superset of x86, even if x86_64 can run x86 binaries, and
the kernel makes this easy.

> I can't depend on actual "host" system being x86_64 so I want
> host=ia32(i386/i686..). If the actual host system is x86_64 or ia32
> the binaries will still work.

Yes, but it would be much simpler and esier to build four toolchains:
build   host    target
x86     x86     x86
x86_64  x86_64  x86
x86     x86     x86_64
x86_64  x86_64  x86_64

The two first toolchains are built with the same .config (with x86 as the
target), but on two different hosts, one is x86, the other x86_64.
The next two toolchains are also built using the same .config (most
probably the same as above, but with x86_64 as the target), and again on
two different hosts.

You'd then have two sets of two toolchains, but that's not quite a problem,
is it? When you make a change to the toolchain configuration, it is quite
easy to do it for both x86 and x86_64 targets, and then build each toolchain
twice, once on a x86 host, and once on a x86_64 host.

The only drawback is is your x86 host is much slower than your x86_64 host.
But then, you could run one of the x86_64 host in x86, just for the sake of
building the toolchains.

> > What is "ABI"? A GMP-specific variable?
[--SNIP--]
> First, I nothing of GMP, and I should test attempting to do a build
> without it. It is the one thing that broke when I forced -m32 during
> the configure step and the error messages directed me to it's ABI
> environment variable.

It should not be needed, as the ./configure of GMP should detect it
automatically. If not, then it should be set conditionally (as I said
previously). Or better yet, GMP should be fixed.

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +0/33 662376056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| --==< ^_^ >==-- `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
`------------------------------^-------^------------------^--------------------'


--
For unsubscribe information see http://sourceware.org/lists.html#faq