This message is from the T13 list server.


See embedded responses
 
-------------------------------------------------
Curtis E. Stevens
20511 Lake Forest Drive #C-214D
Lake Forest, California 92630
Phone: 949-672-7933
Cell: 949-307-5050
E-Mail: [EMAIL PROTECTED]
Ambition is a poor excuse for not having enough sense to be lazy.
 
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
[EMAIL PROTECTED]
Sent: Monday, August 15, 2005 5:21 PM
To: [email protected]
Subject: Re: [t13] Question on Streaming

This message is from the T13 list server.


Curtis, 

I would contend that there are only 8 stream ID's 
and that R/W is advisory information like AU 
size. There is no error condition for writing to 
a read stream number or reading from a 
write stream number, so the R/W has to be 
advisory. 

> CES: I Agree, but the standard is silent on the point.  I think we need to
clarify this at the next meeting.

The only hard constraint imposed 
by the configure stream command is to set 
the CCTL for the stream number so that stream 
number may be subsituted for setting the CCTL
in each command. 

> CES: Depending on how you read the text, you could set different CCTL
numbers for read and write.  I am in agreement with your interpretation, I
think the standard needs a little more clarity.

On the time-out during write I think that is what the 
DWE (Deferred Write Error) bit is for. If you get 
all of the data in the buffer before the timer runs out
and the flush bit is off then failure to get the data 
to the media would cause a DWE and an error 
entry in the Write Error log.  The time limit is just
there to get the interface freed up for the next command. 
Having data in the buffer should not count against 
the CCTL.

> CES: I agree and the spec is fairly clear

If the flush bit is on then you have to wait until the 
buffer is flushed to report command completion. 
If the buffer is not flushed by the CCTL then it is 
a CCTL error and you log the error (no DWE). 

> CES: I agree, but I am seeing differing implementations.  I just wanted to
see if anyone wanted to speak up in defense of their implementation.

Best Regards,
Daniel J. Colegrove
Hitachi Global Storage Technologies

[EMAIL PROTECTED]
(702) 614-6119



"Curtis Stevens" <[EMAIL PROTECTED]> 
Sent by: [EMAIL PROTECTED]
08/15/2005 04:27 PM

To
<[email protected]>
cc

Subject
[t13] Question on Streaming






Are there 8 or 16 stream ID’s?  Is a stream a bi-directional entity, or a 
single direction?  In ATA8-ACS, I am looking at clause 7.9 (Configure 
Stream).  This can get a bit touchy for error checking.  If you configure 
stream to add a stream with ID 5h as a write stream, and there is already 
an ID 5h that is a read stream, do you error or do you consider it a 
separate stream.  If it is a separate stream, it will make turning a 
stream off a little more tedious.
 
In the streaming command set, when you have a command completion timeout, 
how is the error reported if you have received all the data, but the flush 
to disk is still in process?  We see varying implementations…
 
 
-------------------------------------------------
Curtis E. Stevens
20511 Lake Forest Drive #C-214D
Lake Forest, California 92630
Phone: 949-672-7933
Cell: 949-307-5050
E-Mail: [EMAIL PROTECTED]
Ambition is a poor excuse for not having enough sense to be lazy.
 

Reply via email to