This message is from the T13 list server.


Yes, I am confusing the low active signal (a dash character at the end of a
signal name) with figures standard shown.
Many thanks for your kindly and clear explaination.

-----Original Message-----
From: Mcgrath, Jim [mailto:[EMAIL PROTECTED]]
Sent: Friday, January 04, 2002 11:13 AM
To: Mcgrath, Jim; '[EMAIL PROTECTED]'; [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query



Actually, the more I think about it the more clear that it is the sign that
threw you off.  DSTROBE and DDMARDY- are used, and since they map to IORDY,
they all must have the "PIO state" when the UDMA burst has ended.  But
clearly this means that DSTROBE and DDMARDY (if it existed) would have
opposite states.

In general just remember that all signals must end up in the state they
would be in before a PIO operation just after a UDMA burst ends (i.e. DMACK-
is deasserted).  Everything flows from there.

Jim


-----Original Message-----
From: Mcgrath, Jim [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 03, 2002 7:05 PM
To: '[EMAIL PROTECTED]'; [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query


This message is from the T13 list server.




Actually, you never asked me about DDMARDY, but instead about DSTROBE.  They
map to the same physical line, but the two are quite different in the UDMA
protocol (the former is used in DATA OUT, the latter in DATA IN).

Secondly, as I just stated in my previous reply, DDMARDY- is deasserted
before DMACK- is deasserted, as the standard clearly indicates.  That is,
there is no problem here.  Perhaps you are confusing DDMARDY (which you use
as a term, but which is not used in ATA) with DDMARDY- (which is used in the
definitions and figures for UDMA).  i.e. the standard uses the reversed
state of the signal, just like it does for DMACK- (as opposed to DMACK).

Jim

PS ATA has standard conventions on how signals are defined, what asserted
means, and the like.  e.g. section 3.2.5 Signal conventions:
        Signal names are shown in all uppercase letters.
        All signals are either high active or low active signals. A dash
character ( - ) at the end of a signal name
        indicates the signal is a low active signal. A low active signal is
true when the signal is below ViL, and is false
        when the signal is above ViH. No dash at the end of a signal name
indicates the signal is a high active signal.
        A high active signal is true when the signal is above ViH, and is
false when the signal is below ViL.
        Asserted means that the signal is driven by an active circuit to the
true state. Negated means that the signal is
        driven by an active circuit to the false state. Released means that
the signal is not actively driven to any state
        (see 4.2.1). Some signals have bias circuitry that pull the signal
to either a true state or false state when no
        signal driver is actively asserting or negating the signal.
        Control signals that may be used for more than one mutually
exclusive functions are identified with their
        function names separated by a colon (e.g., DIOW-:STOP).



-----Original Message-----
From: coolaron [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 03, 2002 6:52 PM
To: 'Mcgrath, Jim'; [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query



As you said, "if you did not assert it at this point the legacy PIO
operations after getting out of UDMA might be strange".
So, if DDMARDY (IORDY at PIO mode) doesn't return to high level before the
host has deasserted DMACK-, isn't it strange for PIO?

-----Original Message-----
From: Mcgrath, Jim [mailto:[EMAIL PROTECTED]]
Sent: Friday, January 04, 2002 10:32 AM
To: '[EMAIL PROTECTED]'; [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query



It does (i.e. see figures 57 and 58).  The deassertion of DMACK- is always
the last thing done, since it clocks the CRC data from the host to the
device and defines the end of the DMA burst (after which all the signals
revert to their earlier meaning).  Indeed, the very question of "changing
DDMARDY after DMACK-s has been deasserted" has no meaning since the DDMARDY
signal is no longer defined after DMACK- is deasserted!

Jim


-----Original Message-----
From: coolaron [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 03, 2002 6:07 PM
To: [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query


This message is from the T13 list server.



Many thanks for your answers.

But why DDMARDY doesn't return to high level before the host has deasserted
DMACK- in UDMA data-out mode?

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
Mcgrath, Jim
Sent: Thursday, January 03, 2002 4:43 AM
To: '[EMAIL PROTECTED]'; [EMAIL PROTECTED]
Subject: RE: [t13] Ultra DMA query


This message is from the T13 list server.



The key design principle of UDMA is to have the signals be in the exactly
same state after the DMA burst as it was before the DMA burst.  Remember
that outside of the burst you have to play be old PIO rules, which do not
allow for double transitions clocking.  It is this careful attention to
backward compatibility that makes it so successful.

In particular, DSTOBE does not exist outside of a DMA burst.  It reverts to
the old definition of the signal.  HSTROBE is DIOR-, while DSTROBE is IORDY.
To make this all work they are in the ASSERTED state when you start a UDMA
burst and ASSERTED when you leave the burst.

I think 1) was placed there so that if for some reason DSTROBE had not been
asserted, the device can assert it.  It might be possible that it was
deasserted if the last PIO before the UDMA burst was using IORDY.  I'd just
make sure it is asserted at the start of the burst.

In 2) the issue is more straightforward, since if you did not assert it at
this point the legacy PIO operations after getting out of UDMA might be
strange.

Jim







-----Original Message-----
From: aronlin [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, January 02, 2002 1:08 AM
To: [EMAIL PROTECTED]
Subject: [t13] Ultra DMA query


This message is from the T13 list server.


Please help me to resolve questions about ultra DMA,

1. During initiating an ulta DMA data-in burst, why need the action, "The
device may assert DSTROBE tZIORDY after the host has asserted DMACK-"?
does it mean that first strobe should be from high to low? Why?

2. I cannot understand why "if STROBE is not in the asserted state, the
sender shall return STROBE to the asserted state" ?
STROBE is active on edge not on level, why we have to care its state on high
or low?

Thanks in advance.

Subscribe/Unsubscribe instructions can be found at www.t13.org.
Subscribe/Unsubscribe instructions can be found at www.t13.org.

Subscribe/Unsubscribe instructions can be found at www.t13.org.

Subscribe/Unsubscribe instructions can be found at www.t13.org.

Subscribe/Unsubscribe instructions can be found at www.t13.org.

Reply via email to