Phil,

I wonder what the significance of being interruptible is if not garbling
results. I imagined that MVC was "atomic" in nature meaning that the second
processor couldn't "garble" things until the MVC had run it's course.

Well, this probably isn't of much interest to anyone here but thanks for
discussing anyway.

Paul

> You can rely on MVCL to operate the way the POP says it will. My 
> impression
from reading the POP is that the MVCL is an interruptible instruction so if
you're coding first level on a multi processor configuration without an
operating system to manager things for you the results of the MVCL can be
erractic in appearance even if the instruction has done exactly what it was
supposed to since another thread on another processor was updating the
storage the MVCL was operating on.

I haven't dug this piece of PoP out for some time, but memory suggests that
the interruptibility of MVCL has little to do with it, apart from increasing
the window.  Even instructions like MVC can cause garbling if two sharing
processors hit the same piece of storage at the same time.  Older machines
used to store a doubleword at a time - do modern designs do a conceptual
full-length store at one time?  ISTR PoP saying the results of two
processors trying to update the same storage at the same time would always
be unpredictable.

--
  Phil Payne
  http://www.isham-research.com
  +44 7785 302 803

Reply via email to