On Thu, Sep 21, 2006 at 10:21:48AM +0200, Martin Husemann
wrote:
> On Wed, Sep 20, 2006 at 07:55:40PM -0400, Jed Davis
wrote:
> > Uh... don't we already have some 32-bit per-CPU
bitsets in various places?
>
> Well, that is pretty easy: just follow what Chuq and
Andrey did for sparc64,
> have a look at sys/arch/sparc64/include/cpuset.h.
> (For the lazy ones: it typedefs a MD cpuset_t, and
provides access macros.
> No suprise, on sparc64 it is defined as uint64_t)
Can we also have a MAXPROCS/MAXCPUS whatever define for CPU
local data
structures? For UP, it can be hard-wired to 1 and be an
upper limit for
the others. I expect it to be used mostly for arrays of
pointers or
small datastructures, so having a 32 or 64 for now wouldn't
hurt.
Joerg
|