This message is from the T13 list server.

> Harlan Andrews <[EMAIL PROTECTED]>
> these questions ...
> about Device0/Device1 (Master/Slave) detection and operation.
> I've trouble finding them in the spec.

I think the printed answers were as follows in Ata/pi 4.  I'm posting quickly in the 
hope of a free, good, quick education on how wrong I am about Ata/pi 4 and how much 
more wrong Ata/pi 7 makes me.

(I hear the real answers are in some dispute, especially when people are working to 
minimise boot/reset delays.  But I trust we all want to begin the discussion by 
establishing consensus over what the printed answers are.)

> 1.  Is there a spec on how long the Device1 should assert DASP ?
> I.E.  Is there a minimum assertion time for DASP ?

Device0 should wait 750ms after a reset that requires DASP- before giving up on seeing 
DASP- ... so Device1 had better get it asserted by then.  In the printed text, how 
agreed Device0 and Device1 are about when a power-on reset begins used to be fuzzy.

Device1 should keep DASP- asserted lo until told to release it: by a write of x1F7 
Command when Device1 is not BSY and is selected by DEV, or by the expiry of a 
specified timeout (30-ish seconds for most resets).

Some Windows Oem's require that Device1 also release DASP- when a command is written 
to Device0, to turn off the HDD light between when the Bios hands control to Windows 
and when Windows gets around to talking to Device1.

> 2.  Is there a spec on how long or how often
> Device0 should check DASP to determine
> if a Device1 is present ?
> Is DASP allowed to move after initial handshake?

Grin.  Excellent question.  No spec.  Rumour says plug 'n pray goes better if Device0 
is a little slow to believe DASP- is lo.  Some people think Best Practice demands a 
debounce like at least two samples in a row that agree.

> 3.  What is the reason for the 1 MS max spec
> on releasing PDIAG after RESET ?
> 4.  What would be the consequences
> of being late releasing PDIAG after RESET ?

PDIAG- could not then be easily used for other things, like 80 pin cable detection, in 
which PDIAG- is helpfully named CBLID-.

> 5.  Does method of determining
> Device0 vs Device1 have ANY effect
> upon the requirements for DASP and PDIAG ?

Not specified.

> I.E.  Is JumperSelect same as CableSelect
> with respect to DASP and PDIAG ?

Not specified.  Often true.  Lately the once-reserved "word 93" of the op xEC/A1 
Identify device has been given meaning to report if the device knows whether it was 
jumper-selected or cable-selected.

> I'm posting quickly in the hope of a free, good, quick education

Let the games begin.

Pat LaVarre


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

Reply via email to