Re: [RFC 00/15] x86_64: Optimize percpu accesses

[H. Peter Anvin]

Eric W. Biederman wrote:
> Another alternative that almost fares better then a segment with
> a base of zero is a base of -32K or so.  Only trouble that would get us
> manually managing the per cpu area size again.

Yes, an extra link pass would be better than that.  I have tried, and 
none of the clever things I tried actually works, since GNU ld has 
pretty much no way to get it to reveal its information ahead of time.

However, I want to explore the details of the supposed toolchain issue; 
it might be just a simple tweak to the way things are done now to fix it.

Clearly, given the stack protector ABI, we want %gs:40 to be usable and 
free.

	-hpa
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/