Re: kernel/microcode.c error from new 64bit code

["H. Peter Anvin" <hpa@zytor.com>]

Linus Torvalds wrote:
> 
> No. Because on x86-64 it is NOT zero. Because there "unsigned long" is
> 64-bit, and it results in the high 32 bits. Which is, again, exactly what
> we want.
> 

Ah yes, dual-mode code.  Should have figured.

> Guys, give it up. The code is not only already committed, it's simply the 
> best way to do what it does.

Perhaps an even better thing to have would be a wrmsr64() and rdmsr64() 
routines, for the MSRs which genuinely are a 64-bit item.  Splitting 
them up is rather ugly when it's a real 64-bit value.

Then the code would just be:

	/* 32 bits on x64, 64 bits on x86-64 */
	wrmsr64(MSR_NUMBER, (unsigned long)value);

I think the comment would be justified.

If you agree I'll send a patch.

	-hpa