[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: interpreting variants
From: |
Rob Savoye |
Subject: |
Re: interpreting variants |
Date: |
Tue, 14 May 2002 10:01:25 -0600 |
User-agent: |
Mutt/1.2i |
On Mon, May 13, 2002 at 02:22:04PM -0400, Ben Elliston wrote:
> variants exposed to individual test drivers so that they can intepret
> the variant strings and act accordingly? Suppose my test driver
> compiles programs with an ACME compiler that requires me to compile
> little endian object files using --endian=little. What is the correct
> way of making the mapping from "-mel" to "--endian-little"? I
> couldn't find anything in the documentation that covered this.
The code dealing with variants is purely a GCC specific feature that
I don't know a whole lot about, as I didn't write it, (Doug Evans did)
and as a result there is zero docs I know of on this feature. Testing
variants is buried in the multilib support. If you just want to
change this to "--endian-little" you could do this in the base board file
for your target, or change $CFLAGS.
I daydream of rewriting the mulitlib test code by moving out of DejaGnu
proper, and into gcc/testsuite... and then cleaning it up so it's more
clear how to use it.
- rob -