Re: [patch] futex requeueing feature, futex-requeue-2.5.69-D3

[Ingo Oeser <ingo.oeser@informatik.tu-chemnitz.de>]

Hi,

On Tue, May 20, 2003 at 07:56:27AM +0100, Christoph Hellwig wrote:
> On Tue, May 20, 2003 at 08:27:03AM +0200, Ingo Molnar wrote:
> > yes, but the damage has been done already, and now we've got to start the
> > slow wait for the old syscall to flush out of our tree.
> 
> Actually it should go away before 2.6.0.  sys_futex never was part of a
> released stable kernel so having the old_ version around is silly.  I
> Think it's enough time until 2.6 hits the roads for people to have those
> vendor libc flushed out that use it.  (and sys_futex still isn't used
> in the glibc CVS, only in the addon nptl package with pre-1 release
> numbers.)

This sounds reasonable. The people who shipped that already to
customers have so heavily patched kernels, that this simple patch
for the old sys_futex shouldn't really matter.

But cluttering the kernel with an API/ABI that was born in the
same development cycle, where it has been obsoleted sounds not
worth the bytes it consumes. And so we have a syscall slot
available for the next development cycle.

Or did anybody promise that this was final already? Usally this
promise comes with x.even.y not with x.odd.y .

Regards

Ingo Oeser