Re: [patch] lockless, scalable get_pid(), for_each_process() elimination, 2.5.35-BK

[Cort Dougan]

} of course, and it should also be massively-threaded, LSB-compliant,
} enterprise-ready, secure, cluster-aware, power-saving and self-healing. I
} admit that there's still lots of work to be done, but there's just so many
} hours in a day.

get_pid() is simple, I don't see why it's taking you so long to add those
features.  I'll sketch out a plan for an improvement in getpid_MARK02(c)
numbers once you have the prototype done.

} actually, on-the-fly O(log(N)) multiprocessor garbage collection is
} already integrated into its high-end modular OO design.
} 
} > Perhaps a get_pid() that solves the Turning Halting Problem should be on
} > the todo list for 2.6.

Given my misspelling of Turing, I think it's clear that get_pid() needs a
spellchecker.  This would be an opportunity to begin work on get_pid .NET.

} the first small mystery to solve are certain perturbations in Alan
} Turing's name. But, yes, it's definitely a goal of the PID allocator to be
} an answer to all, but also for it to avoid infinite loops for every
} possible input value, while yielding slightly more subtle output than the
} numeric value of 42. Patch in a few minutes.

Please be sure to implement a BK replacement, open-source it, switch the
kernel over to it and supply the patch in that format.  It would be
appreciated.