binutils patches for Cirrus/arm9e/maverick support

[Richard Earnshaw]

> In message < 3BC33655.47390D36@redhat.com >, Jonathan Larmour writes:
> >Actually the Maverick is a family of CPUs from Cirrus, of which only one is
> >at present an ARM9: http://beamaverick.com/sub.cfm?CategoryID=2
> 
> Mmm.  Well, they seem to include CLPS7500FE in the "Maverick" family,
> and that pretty clearly doesn't support the DSP extensions that are at
> issue here.  But if Cirrus haven't given the actual instruction set a
> name, using "maverick" to describe it is probably the best you can do.
> 
> So: maverick-elf?  Well, it might be nicer if all ARM-derived targets
> had names generally in the pattern arm*-*, and that would avoid any
> future embarrassment if Cirrus decide to expand the Maverick marketing
> name to, say, MIPS-based processors.  But "armmaverick" is, if
> anything, even more ugly a name than "beamaverick".
> 
> It doesn't really matter all that much anyway.  Even something like
> "arm9m-elf" would probably be okay.

No, I think that sounds too much like the arm7[d]m, which indicated the 
presence of the multiplier feature.

I'd go for maverick (in the same way as xscale or strongarm); but I'd 
prefer to see something in the config files that noted that this was 
simply an arm with some defaulted set of extensions TE_? (rather than 
having to have a lot of additional config files lying around).

R.

Hm, now I'm going to have to port all that new code to my new parsing 
routines...