2004-06-03 03:52:34

by ndiamond

[permalink] [raw]
Subject: MIPS, How to use floating point in a module?

Now I am told that our next target will be a MIPS-based CPU.
Looking at files under arch and asm includes for MIPS, I don't see
any equivalent of the x86 (x87, 686, etc.) functions and macros
kernel_fpu_begin, init_fpu, kernel_fpu_end, etc. Is it safe to
just barge ahead and use floating-point arithmetic operators when
the driver needs to use them?

This CPU has no opcodes for log2, exp2, sin, and cos, so it looks
like I'll have to buy one of the books that some posters here kindly
recommended, and do polynomial interpolations.


2004-06-03 15:14:52

by Yoichi Yuasa

[permalink] [raw]
Subject: Re: MIPS, How to use floating point in a module?

On Wed, 2 Jun 2004 22:39:17 -0500 (CDT)
[email protected] wrote:

> Now I am told that our next target will be a MIPS-based CPU.
> Looking at files under arch and asm includes for MIPS, I don't see
> any equivalent of the x86 (x87, 686, etc.) functions and macros
> kernel_fpu_begin, init_fpu, kernel_fpu_end, etc. Is it safe to
> just barge ahead and use floating-point arithmetic operators when
> the driver needs to use them?

We cannot use the FPU instruction in a MIPS kernel.

Yoichi