My questions:
1. Is there really a requirement for such a direct relationship between DMARQ negate and DDMARDY- negate?
2. if so, why? Especially when the relationship does NOT apply between HDMARDY- and DMARQ.
See below for more detailed reasoning.
Cheers
Keith
________________________
>From the standard:
"9.14.4.1 Host terminating an Ultra DMA data-out burst
The following steps shall occur in the order they are listed unless otherwise specified......
:
c) The device shall negate DMARQ within tLI after host asserts STOP......
d) The device shall negate DDMARDY- within tLI after host asserts STOP......"
Because of the covering statement, c) must occur before d), but both are the consequence of the same event (STOP assert).
. So, is this saying that DDMARDY- negate must always occur BEFORE DMARQ negate (for host terminate)?
. What about AFTER?
Implies in practice tLI(DDMARDY) > tLI(DMARQ) to guarantee sequence is met. (If =, skew may cause violation at host).
The timing diagram implies that these are independent processes, with no such interdependency (simply dependent on STOP assert).
Compare this with 9.13.4.1 Device terminating an Ultra DMA data-in burst: d) and e) describe a similar case with DMARQ and HDMARDY- but because e) is qualified with the statement "Steps (d) and (e) may occur at the same time.", this seems to allow for tLI(HDMARDY) > tLI(DMARQ) AND tLI(HDMARDY)=< tLI(DMARQ).
-- Keith Clausen, STMicroelectronics, Mail: 1000 Aztec West, Bristol BS32 4SQ, UK Email: [EMAIL PROTECTED] Phone: +44 1454 462457 Fax: +44 1454 617910
