Thu, 26 Mar 2015, 00:49


After today's lecture, my TAs pointed out to me that my choice
of notation was unnecessary confusing.  They are absolutely correct.
Therefore this email.

In particular, I used subscript i to indicate 1 or 2 when talking about
BRi and BGi and to indicate IN when talking about BBSYi.  Shame on me!  
I should know better than to use confusing notation.

So, I ask you to correct your notes, as follows:

BBSY: 
=====

There is one signal BBSY which is asserted by one device controller as
BBSY_OUT and read by the other device controllers as BBSY_IN.  The signal is 
asserted by the bus master upon taking control of the bus to start the bus 
cycle, and remains asserted by the bus master for the duration of the bus
cycle.

Suppose we are somewhere in the middle of bus cycle n, device controller X 
is the bus master, so it is asserting BBSY_OUT.  All other device controllers 
are receiving BBSY as BBSY_IN.  ...in particular, device controller Y (which is
in SACK state waiting for bus cycle n to end so it can be the bus master for 
bus cycle n+1.  Device controller Y is reading the BBSY signal (i.e., BBSY_IN).
As long as BBSY_IN is asserted, we remain in bus cycle n, and Y remains in 
SACK state.  When bus cycle n ends, the bus master for cycle n (device 
controller X) negates BBSY (i.e., NOT-BBSY_OUT).  Device Controller Y sees 
BBSY_IN as negated, and immediately asserts BBSY itself (BBSY_OUT), taking 
control of the bus, i.e., initiating bus cycle n+1.

BR and BG: 
==========

Each priority level has its own BR and BG signals.  BRi can mean
BR1 or BR2, depending on which priority level is associated with a particular 
device controller.  BGi can mean BG1 or BG2, depending on which priority level 
the PAU wants to grant the next bus cycle to.  The device controller closest
to the PAU will receive BGi first.  If it wants the next bus cycle, it will
assert SACK.  If it does not want the next bus cycle, it will enter a state
where it continues to transmit BGi.  This will enable a device controller
further away from the PAU to accept the BGi signal, and assert SACK.
 
OK?  Again, I apologize.

Yale Patt