[coyotos-dev] Status and roadmap

[Sam Mason]

On Fri, Jan 19, 2007 at 10:48:08AM -0500, Jonathan S. Shapiro wrote:
> On Fri, 2007-01-19 at 10:29 -0500, Jonathan Adams wrote:
> > On Fri, 2007-01-19 at 14:28 +0000, Sam Mason wrote:
> > > On Thu, Jan 18, 2007 at 04:57:41AM -0500, Jonathan S. Shapiro wrote:
> > > > The kernel specification has been public for quite some time, and the
> > > > implementation is already underway.
> > > 
> > > The work is showing, section 3.3.2 (or was it 3.3.3) only documented
> > > what happened when the Activated field was in one state---I see this has
> > > been filled out to both now.
> > > 
> > > One thing that worries me now is where the kernel gets the memory to
> > > "synthesize a message to be delivered" when the userland scheduler is
> > > activated.  In fact where does this memory come from in the general
> > > case?
> > 
> > The message is a "short" message, which any non-blocked FCRB can receive
> > without needing additional storage.
> 
> Oh. Perhaps I misunderstood Sam's question.
> 
> Sam: can you say which place in the manual that came from? It sounds
> like I need to clarify that text.

I posted my response to Jonathan Adams before I got your messages.  I'm
still confused by this passage:

    If the Activated field is non-zero, then the exception has occurred
    in the activation handler itself. In this case the kernel will
    synthesize a message to be delivered as if by invoking the handler
    capability stored in the process structure. The fault code and
    auxiliary fault information will be passed as arguments to this
    invocation in w0 and w1 respectively. The ex field of the FCRB named
    by the handler capability will be set to 0.

I think it could be because I don't understand how run-in works properly
yet.  I'll read more about how it works and wait for your changes in the
spec to go in before responding again.


Thanks,
  Sam