RFC: Mips option parsing patch

[Richard Sandiford]

Daniel Jacobowitz <drow@mvista.com> writes:
> Consider it done.
> 
> Binaries built with -march=r8000 -mips2 are now marked:
>   Flags:                             0x30000100, mips4 UNKNOWN
> 
> The mips4 marking is wrong, but that's a BFD bug - it sets mips_arch
> unconditionally ignoring what GAS may have set.  The UNKNOWN is a
> readelf bug, and actually E_MIPS_32BITMODE - perfect.

Is this patch for mainline or the upcoming release branch?  Any chance
it could just be the branch, if that wasn't the idea all along?  I can
see the need for the next release to be compatible with gcc 3.0, but
I'm not sure why CVS mainline has to be, especially since CVS gcc
ought to work with the new scheme.

Richard