Re: IA32 FP

[jtc@redback.com (J.T. Conklin)]

>>>>> "Jim" == Jim Blandy <jimb@cygnus.com> writes:
Jim> That's exactly what I'm trying to do.  My immediate interest is
Jim> making Linux work, but if we IA32 maintainers talk about it a
Jim> bit, we can all use the same solution.
Jim>
Jim> There's a lot of gratuitous duplication among the gdb/config/i386
Jim> headers.  (For fun, diff nm-i386v.h and nm-i386sco.h.)  So I
Jim> don't think there's any technical reason we couldn't do a great
Jim> deal of unification.

Excellent!  I was unclear on how much upheaval was acceptable in this
effort to clean up the IA32 target configs.  

I'm all in favour of making whatever changes are necessary, then
fixing all of the common targets, and finally repairing obsolete
target configs as best we can.  Once your baseline changes are
integrated, I'll see what I can contribute for embedded (i386-aout /
i386-coff / i386-elf) and NetBSD targets.

	--jtc

P.S.: 

There has been a lot of discussion about the format of the "info
float" command.  I have no strong feelings about this, because once
all the FP registers are supported, a user should be able to create a
GDB user defined function to present the information in whatever form
desired.  IMO "info float" is somewhat to blame for the poor state of
FP support in the IA32 ports --- it made it all too easy to treat FP
as second class registers.

-- 
J.T. Conklin
RedBack Networks