This message is from the T13 list server.
I agree with a lot of what you say here, Hale. I'd like to get some feedback on
changes I'd like to propose to this document:
A. Move the new bits in ID word 63 up to some unused (reserved) field. I'd suggest
offset 62, and the data out there would include the DMADIR bit, plus a complete map of
the UDMA support bits (not just a highest UDMA number). This allows devices that have
spotty support for UDMA speeds to function properly if host software steps down UDMA
modes upon errors.
B. A sub-section to the Packet command section in both the "General Operational
Requirements" sections should describe this feature, and also described in the
"Command Descriptions" section. This includes what happens when the ATAPI device
finds an incorrect DMADIR bit setting in the Packet command. This agrees with what
Hale suggests.
C. The overlapped command support is disabled for a bridged ATAPI device, and
Multi-Word (Single-Word too?) is basically disabled for a bridged ATAPI device. I
think this is fine. Does anyone disagree?
Looking forward to the flurry of emails this will generate. Regards, MKE.
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Hale
Landis
Sent: Monday, November 10, 2003 11:46 AM
To: T13 List Server
Subject: [t13] e03132r2 DMADIR bit for PACKET command
This message is from the T13 list server.
I have some comments/questions about E03132R2...
Note: The T13 web site is down (for several days now) so I can't get
the "official" E03132R2 PDF. But I was sent the MS Word version that
is *very* difficult to read.
However...
My first comment is: This sure is a kludge.
My second comment is: Since the ATAPI device designers (mostly in the
Far East) DO NOT pay any attention to T13 or the ATA/ATAPI-x
documents, why do you think you will ever see any ATAPI devices that
implement this kludge?
My technical comments are:
1) What error does a device report if the value of the DMADIR bit
does not match the direction of the current command?
2) Do not mess with the definition of the bits in ID word 63. If a
device supports this DMADIR bit thing then ID word 63 should be 0.
3) Do not mess with the definition of the bits in ID word 88. If a
device supports this DMADIR bit thing then ID word 88 should be 0.
4) General question: What proposal changes the definition of ID word
49 bit 15 to "Bit 15 of word 49 is used to indicate that the device
supports interleaved DMA data transfer for overlapped DMA commands."
?
5) Define two entirely new words to take the place of ID words 63 and
88. These words should have the exact same format as ID words 63 and
88 respectively.
6) There is no reason to change the definition of the IO and CD bits.
No change is required here. In fact if the value of the IO and CD
bits does not match the host's value of DMADIR then that would be
some kind of error condition that a host device driver should detect.
7) This proposal needs a new sub-section in clause 6.x of the
ATA/ATAPI-7 Vol 1, probably new text in the "PACKET Command Feature
Set" clause.
Hale
*** Hale Landis *** www.ata-atapi.com ***
