Obsoleting Terminate Task

Gene_Milligan at notes.seagate.com Gene_Milligan at notes.seagate.com
Wed Nov 19 08:48:03 PST 1997


* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* Gene_Milligan at notes.seagate.com
*
--0__=zOUFMuoASI9HhATnNIyq8Eh2el4tNhagi7L0NkuQnaYCjsxBP1pksbuQ
Content-type: text/plain; charset=us-ascii

Questions below the answers:

I waited to see the answers of others but if there were answers I did not
see them on the reflector. So:

1) First of all Obsolete in the standard has a special meaning. It is
"obsolete: A keyword indicating that an item was defined in prior SCSI
standards but has been removed from this standard."

     This in itself does not prevent it from being used but any
implimentation is based upon a prior standard. It does send a message that
it is probably not widely supported. The motivation is normally that the
item is excess baggage that did not become popular.

     In this case, from my veiwpoint, the motivation is that this is a
rather complex function that is  no longer implimented and used by the
combination of initiator and target. That is not to say that no targets or
no iniitators have the capability. But it is to say that the function is
not exercised. This means that as we evolve the protocol we are making
changes without an exercised confirmation that this function still works.
Rather than closing our eyes to this situation I think we are better off
sending the warning "obsolete".

2) No and Yes. No there is, to my memory no direct way. Yes there are
indirect ways. Enable write caching and the commands will receive
completions and all will be subsequently written provided power is not
removed. In this case even a reset will still result in all being written
provided there is not a deferred error. If write caching is disabled
commands can be ended with aborts. In this case it is assumed that the host
saves a copy of the write data until a safe completion is received. For the
aborted commands the host can then read the data and compare it to the
saved write data. If different the write did not complete. If the data is
the same the write did complete or it doesn't matter. However if the data
is different, it does not guarantee that none of it has been written - just
that not all of it has been written. Some devices will write all of a
command if they started the writing and have all of the data but others do
not. Obviously if the host has not sent all of the data it was not all been
written.

Gene






bmckean at mylexboulder.com (Brian McKean) on 11/12/97 09:46:14 AM

To:   t10 at Symbios.COM @ INTERNET
cc:    (bcc: Gene Milligan)
Subject:  Obsoleting Terminate Task

I do not object to obsoleting Terminate Task but would like to know...

1. What are the reasons to obsolete the Terminate Task task management
function?


2. Is there another method to allow an initiator to end a task and know
exactly what was completed?  Specifically, for a write command is there
a way to stop it and know exactly how many blocks were written?


--

Brian McKean                            Voice: (303)449-2228 x4246

Mylex Corporation                       Fax:   (303)413-0464
4900 Pearl East Circle Suite 104        brianm at mylexboulder.com
Boulder, CO 80301-6108



--0__=zOUFMuoASI9HhATnNIyq8Eh2el4tNhagi7L0NkuQnaYCjsxBP1pksbuQ
Content-type: application/octet-stream; 
	name="Card for Brian McKean"
Content-transfer-encoding: base64

YmVnaW46ICAgICAgICAgIHZjYXJkDQpmbjogICAgICAgICAgICAgQnJpYW4gTWNLZWFuDQpuOiAg
ICAgICAgICAgICAgTWNLZWFuO0JyaWFuDQpvcmc6ICAgICAgICAgICAgTXlsZXggRXh0ZXJuYWwg
UHJvZHVjdHMNCmFkcjogICAgICAgICAgICA0OTAwIFBlYXJsIENpcmNsZSBFYXN0IDs7U3VpdGUg
MTA0O0JvdWxkZXI7Q29sb3JhZG87ODAzMDE7VVNBDQplbWFpbDtpbnRlcm5ldDogYm1ja2VhbkBt
eWxleGJvdWxkZXIuY29tDQp0ZWw7d29yazogICAgICAgMzAzLTQ0OS0yMjI4IHg0MjQ2DQp0ZWw7
ZmF4OiAgICAgICAgMzAzLTQxMy0wNDY0DQp4LW1vemlsbGEtY3B0OiAgOzANCngtbW96aWxsYS1o
dG1sOiBUUlVFDQp2ZXJzaW9uOiAgICAgICAgMi4xDQplbmQ6ICAgICAgICAgICAgdmNhcmQNCg0K
DQo=

--0__=zOUFMuoASI9HhATnNIyq8Eh2el4tNhagi7L0NkuQnaYCjsxBP1pksbuQ--

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com




More information about the T10 mailing list