Re: [PATCH, MIPS] Better diagnostic on incorrect size for .reginfo

[Adam Nemet <anemet at caviumnetworks dot com>]

Maciej W. Rozycki writes:
>  I think we have other tests that fail if the default emulation is neither 
> of {o32,n32} and I think the reason is different -m options would have to 
> be used based on what o32 for the given target actually means, so I am 
> leaving the decision to the maintainers.

So just to summarize for the maintainers.  The last version of the
patch (http://sourceware.org/ml/binutils/2007-08/msg00230.html)
performed the test with -mabi=n32 on $newabi targets and with default
options otherwise.  The idea is that we want to explicitly avoid n64.

This wouldn't test o32 under linux64 for example which is the issue
that Maciej raised.  The question is whether the latest version of the
patch is sufficient or I should add a third test for $newabi with
forcing -mabi=32 and the corresponding emulation.

> I do not seem to see any attempt to address the issue in a
> systematic way in the test suite at the moment and certainly I would
> not expect you to be requested to implement it as a prerequisite for
> your change (unless you actually want to, that is!).

Probably not for this small test but maybe later for something more
involved.

Adam