Re: PMC strawman

[Matt Thomas <matt%3am-software.com@localhost>]

At 08:44 PM 2/14/2003, Allen Briggs wrote:
> On Fri, Feb 14, 2003 at 07:42:44PM -0800, Matt Thomas wrote:
> > There would be a bit to note whether that blob has info.  I'd
> > make the blob explicitly opaque to userspace.
>
> Hmm...  Will that work with libpthread?  If you're doing a voluntary
> thread switch, the kernel's not really involved in setting or restoring
> the context.  Well, it is involved in restoring the context if you're
> switching to a thread that was preempted.  Unless I'm missing something.

the ppc code has control of task switching so it can make sure the
right thing happens (by calling setcontext(2) if required).

> > Of course it would conditional.  I'd also prefer to have
> > separate contents for interrupt / non-interrupt code.
>
> That would be nice.
>
> > Right now the pmc counters are per proc, not lwp, which needs to be
> > fixed.  I was thinking about creating a pmc context that a process
> > could establish and share among it's children (or all of its threads).
> > Make it refcnt'ed and on last reference, free it.
>
> I like the idea.  How would you un/bind threads/procs from/to it?

By default you be bound.  pmc_configure_counter would be used to
force a new context (perhaps a flag in pcm_evid_t).  For now, you
can't bind it to it.  it must be inherited.


--
Matt Thomas               Internet:   matt%3am-software.com@localhost
3am Software Foundry      WWW URL:    http://www.3am-software.com/bio/matt/
Cupertino, CA             Disclaimer: I avow all knowledge of this message