On Mon, Mar 21, 2005 at 05:41:51PM -0800, David Daney wrote:
> Kumba wrote:
> >Daniel Jacobowitz wrote:
> >Doesn't this need the glibc side of things to be effective?, or is it
> >testable w/o that component?
It is testable independently. Also, I posted the glibc bits last week.
> I think the main point is that it should not break existing code.
Of course. It doesn't. The only thing it could possibly break would
be four-argument clone (it's supposed to be five argument, and the
missing argument conventionally goes in the middle... oops). But
I strongly believe nothing is yet using the four-argument form so I
synced MIPS with the rest of the world.
> We need NPTL support in all three of GCC, Linux kernel and glibc before
> it can be tested. If it doesn't break existing code, I think it should
> go in the kernel so that we have something on which to test gcc and glibc.
GCC support was committed two weeks ago, BTW.