On Fri, Apr 26, 2013 at 5:15 PM, Michael Haberler <mai...@mah.priv.at>wrote:

> Mark,
>
> > I was thinking more of hardware resources, such as interrupts.  What
> > happens when both instances require the same interrupt at the same time?
>
> Any instance usually will have a HAL driver to talk to the outside world
> exactly like the single instance which we have now, and that driver will
> attach to a hardware device and will be responsible to allocate all
> required resources during startup or bark if isnt possible - there is no
> difference at all from a single instance which loads several drivers
> talking to different devices
>
> this would include an interrupt if there were one - you probably wont find
> any in LinuxCNC
>

Isn't that presuming that all the hardware devices will be relegated to
that singular instance?  That would be okay if there were multiples of each
hardware device, but in cases where there aren't is where my concern lies.


>
>
> > Unlike shared memory, interrupts are a bit harder to share, or have
> > concurrent use.  Can the separate instances somehow share the IO bus?
>
> no, separation is at the HAL driver level as it always was, no new
> concepts here
>
> an example:
>
> assume you have 2 parports in a PC:
>
> it is exactly the same I/O usage if you do this in a single instance:
>
> loadrt hal_parport cfg="0x278 0x378"
>
> or do this in instance #1
> loadrt hal_parport cfg="0x278"
>
> and this in instance #2:
> loadrt hal_parport cfg="0x378"
>
>
> Remember: instances are ships in the night _except_ if there are
> cross-linked signals, in which case they can share signal values
>
> you're thinking too complicated ;)
>
> - Michael
>

I probably am overthinking it, and making it too complicated, but my degree
is computer science, and I've always been intrigued by RTOS systems.  Now
that folks are looking into multiple instances of RTOS running on the same
machine, I'm very curious how they get around resources that can't be
shared, or how they schedule them so that a lock on the resource by one
instance affects the other instance which might need that resource at the
same time.

Mark
------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to