Re: [PATCH] cpumask 5/10 rewrite cpumask.h - single bitmap based implementation

[Paul Jackson <pj@sgi.com>]

> Three things are required.

Thank-you, Rusty.  That makes more sense.  The first step in designing
good API's is to be clear what information it is essential to pass.


> In this case, though, the early example programs for setaffinity all
> used "unsigned long mask; sys_sched_setaffinity(...&mask,
> sizeof(mask))", which was both simple and wrong. 

Yeah ... several layers of suckage are in the kernel bitmap layout,
kernel/user API for passing bitmaps, and low level glibc sched set and
get affinity API.  You identify another one.  Not felony (major)
suckage, just petty (minor).  Annoying none the less.  And the source of
quite a few hours of "software maintenance" work.  This margin is too
small to contain the details ;).

Oh well ... I've done my stint for now (hopeful past tense, again)
trying to leave things a little neater than when I arrived.

I hope to turn now back to the cpuset work that I've been assisting
Simon Derr (Bull) on.

Thanks, Rusty, William, Matthew, Nick and others.  

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373