>> I defined a type mips_fpreg_t in types.h so people could choose the size
>> of registers to save, but forgot to implement this in the usual places.
>>
>> I thought it would be better to leave those developing embedded systems
>> with the choice of sizes since they might be more memory constrained.

I'm seriously doubt to have mips_fpreg_t exposed to user land.

What application programs see for floating point caliculus in C
language, are 'float' or 'double', and their implementation is hidden
inside depending on how C compilers handle FP operatations.  Explicit
operations of FPU (like some of libc routines) would require to how
large FP registers are, and would NEED to know how FPU registers
implemented, that is, contracdicting 'to hide' FPU register size with
new type 'mips_fpregs_t'. 

I'd propose FP register type is to be defined with mips_reg_t to well
establish MIPS programming models.  Please keep in mind to distinguish
standpoints between, compiler/library maintainer, kernel programmer
and what application programmers see.  They all are different things,
which sometimes happen to belong to overlapped person(s).

Tohru Nishimura
Information Technology Centre
Nara Institute of Science and Technology