72120.1106 at compuserve.com
Sun Jan 2 14:49:58 PST 1994
Cc: Gerald Marazas, Scott Smyers, and Lawrence Lamers, Technical
Subject: Comments on Working Draft SCSI-3 Serial Bus Protocol
(No Document Number) Revision 14
I did not want to submit comments on one of the several X3T9.2
documents being "forwarded" simultaneously until I had an opportunity
to review them all. Having completed the review of the set, this
is the third of a series of comments on the documents.
1) A minor nit, on the front page when forwarded "The contents
are being actively modified by the X3T9.2 Task Group." should
be deleted. It may be true they are doing this but the procedures
contemplates that dpANS are forwarded because they have been completed
by the committee and not sent out as trial balloons.
2) Although not an early implementor of the SBP I find the statement
"Any commercial or for-profit use is strictly prohibited." counter
productive. Please delete it. I presume the early implementors
have aspirations of making a profit from SBP.
3) It would be a service to the general community to include the
acronym SBP in the title.
4) For publication the temporary designation P1394 should be replaced
with the permanent version 1394 as a global change.
5) It had been alleged at one point that SBP had applications
beyond 1394. If that is true the last sentence of the first paragraph
of the Introduction is inappropriate. As a free standing layered
standard it may be inappropriate in any case. As a minimum change
"shall meet all requirements" should be changed to "shall meet
all mandatory requirements". But if they are mandatory in 1394
why does SBP need to make this comment. Are mandatory IEEE requirements
ignored? This comment also applies to the next to last paragraph
of section 2, 12 and to 13.
6) The statement that SBP "attempts to solve some of the problems
associated with Parallel SCSI protocol" has previously been agreed
to be deleted; is inflammatory; may be libelous; is obnoxious;
is beneath the elegance of 1394; and should be deleted as previously
7) Queue full conditions are not "arbitrary" and usually result
|from the queue being full. A reading of SBP indicates that queue
full will result from the same condition. The principal difference
is the assumption that the host will have huge memory with a humongous
amount available for the queue while the target will have only
a modest amount. Why has the committee surrendered there technical
integrity to the marketeers?
8) Market statement 2 needs an additional "the" and wrongly considers
only a single initiator case since the lack of interruption is
the fact that queuing per se is maintained in the host. Depending
upon the target horsepower additional I/Os from other initiator
may still cause the pause.
9) As far as I know the committee has not seen any data to support
10) In aspect 1, 2, and the first paragraph after 3 "Parallel"
should be deleted. In 1 "Block" should be plural.
11) In aspect 3 the "less packet overhead" is as compared to what?
How much less is it on a percentage basis?
12) I believe "isoch" is slang and consequently should be included
in the 3.1 definitions. Another alternative would be to replace
it globally with "isochronous".
13) "The present approach" is inappropriate terminology. Replace
it with "SBP".
14) The next paragraph should be deleted as "discussions in the
SCSI 3 community" are inappropriate for a standard.
15) In the third item 2 (I assume the ISO editors will object
to the mixed methods of identifying or highlighting paragraphs
and/or items) "It is desired to maintain some level of commonality
for similar functions ... transport." should be replaced with
"The SCSI-3 Architectural Model (SAM) is adhered to maximizing
commonality between the various SCSI transport alternatives."
16) Change the fourth bullet of the Scope from "in conformity
to task set concept of" to "in conformance with the requirements
17) Without an explanation of the difference between "queue" and
"overlapped" bullets six and seven appear to be in conflict.
18) Delete "parallel" from bullet eight.
19) The Normative references may need expansion and addition of
X3 document numbers. I think SCSI-2 should be enumerated.
20) In 3.1.1 I think it is "The result of transferring" rather
than "The act of transferring".
21) Since "asynch" is slang, a definition should be added for
it or it should be changed to asynchronous globally.
22) In 3.1.2-4 isn't a FIFO more than an address? The command
FIFO description comes closer.
23) What makes the command FIFO write only? Surely it performs
some useful work by being read.
24) Regarding 3.1.9 are there no word or byte displacements? (I
think the answer is no just offsets.)
25) In 3.1.10 are elements entered "onto" or "into" a queue? What
is an "element"? Should the term be "queue" or "task set"?
26) In 3.1.10 why is it "enter policy" rather than "enqueue policy"?
27) In 3.1.14 I think an identifier returned from the target to
the initiator would identify the target. Does this definition
tie to 3.1.22?
28) In 3.1.19 what is "normal execution"?
29) Does 3.1.26 indicate that untagged commands are not supported
30) Since there is only one baseline portion of a CDS 3.1.28 should
be "hold the baseline".
31) In 3.1.29 "unit dependant directory" needs to have a definition
32) A section should be added for acronyms.
33) Should section 4 address 1394 services?
34) The second paragraph should mention where SBP requires the
ROM to be implemented.
35) In the third paragraph why not delete "node offset" and delete
the parenthesis around "unique address on the bus".
36) In the next to last paragraph replace "limits" with "determines".
37) The second sentence is nothing but marketing hype with no
technical basis. Delete it.
38) Why are the Conventions not in section 3?
39) Why is upper case not explained?
40) Why is it a named flag rather than a named bit?
41) Table 1 includes the distinction between 1394 and SCSI. Why
doesn't Table 2?
42) Table 1 and 2 should have a note indicating which is transferred
43) What does the second sentence of Section 5 mean? SBP is a
SCSI-3 standard. It would be expected to use SCSI-3 terminology.
Is it only used in section 5?
44) After the standard is published for a while the CDS will not
be new. The second paragraph should be recast from new to different
or neither with CDS just explained. The standard should be written
as a lasting standard not a newspaper.
45) Replace "pointers into initiator" with "pointers to initiator".
46) In the fourth paragraph replace "usage" with ""use".
47) Does the third sentence of this paragraph awkwardly say that
FIFOs are only defined in the TARGET? If so a brief clause in
the second sentence would convey this. In either case the third
sentence should be deleted.
48) In the last paragraph are "The most significant elements"
a superset of mandatory items or is there something else that
makes them most significant? What are the least significant elements?
What bearing does this have on the standard? Why must it be emphasized?
It probably will always improve a standard to search out all instances
of "thus" and eliminate them.
49) What is the exact meaning of "largely equally applicable"?
50) Replace "Such differences as do exist" with "Differences".
51) Replace "are cited in the given topic under discussion." with
"are cited in the appropriate clauses." (A determination of whether
"clause" or "section" should be used is and the document should
be made consistent with this determination. I will usually use
the American section.) (Parenthetical statement also applies to
annex and appendix.)
52) I think an additional phrase is needed in the second paragraph
of 5.1. "once set, it can not be cleared" makes it sound like
the slot is lost forever.
53) Is the first sentence of the third paragraph needed in view
of the second sentence?
54) What is "disgard"? It could be disregard or perhaps discard.
55) What are "task codes"?
56) The note is confusing since it seems contrary that a command
data structure does not contain commands.
57) Change the first portion of the first paragraph after the
note to "The target shall conform to the requirements of the SCSI
58) The last sentence of the section adds no additional information
and should be deleted.
59) I think the first sentence of 5.2 should be turned around.
"The initiator shall set the CDS A-flag in the initiator memory
to perform an abort task function and then send a task management
CDS with the abort task function set" Is that last portion redundant?
Is the A-flag and the abort task function the same thing? If so
delete the latter portion. Same comment for 5.3.
60) In the second paragraph shouldn't the first subline be an
"and" statement with the first full line? Same comment for the
next two paragraphs.
61) The note should not contain a requirement. The best solution
is to make the note a main portion and to reword it to "The target
shall continue processing the chain beyond the CDS which has been
aborted." But what if it was the last CDS in the chain?
62) In the second paragraph of 5.3 delete "operation".
63) In the second paragraph of 5.4 what does "For all identifiers"
64) Does "acquired at the time the clear task set function was
received" substitute for "with a clear task set function" or
does it imply that two CDSs can be acquired simultaneously?
65) Delete the last sentence of 5.5.
66) In 5.5.1 delete "the possibly" and "implemented at the given
67) All CDSs within a chain are for the same target. However can
they be for a random selection of LUNs?
68) Since the appendix (needs decision on title for ISO commonality)
is informative, I presume no further details are necessary and
suggest "details" in 5.5.2 be replaced with "information".
69) In 5.6 replace "A means is provided to enable" with "Autosense
enables". Delete the second sentence of 5.6 as it is not relevant
even though true.
70) The last paragraph should be translated from "baby talk".
71) In the third paragraph of 5.7, and perhaps globally, replace
"has value one" with "equals one".
72) In 5.8 replace "As a part of" with "During".
73) In 5.9 each of the paragraphs seem to be somewhat contradictory
to the others.
74) The editors can not be blamed having been rendered senseless
by SCSI-2's "an SCSI" but "a available" should be "an available".
75) In 5.11 the second sentence should be "An initiator shall
provide a queued CDS".
76) The third sentence should be "The delivery is paced by the
77) The fourth sentence should be "This helps to avoid ...".
78) In the second paragraph replace "Provision is made for the
initiator to specify the manner in which" with "The initiator
may optionally control the manner in which".
79) In the third paragraph replace "In all target environments,
the initiator must be prepared to accept a response" with "The
initiator shall accept a response".
80) Either delete the second sentence or replace it with "The
tap slot architecture should reduce the probability of a tap not
being accepted by a target." if the claim is true rather than
just more marketing hype.
81) In the last sentence delete the first "allowed".
82) In the last sentence of the first paragraph of 5.12 delete
"It is emphasized that".
83) In the second paragraph what does "having indicated to it
there is work to be performed on their behalf" mean? An acceptable
alternative to making it clear, is to delete the statement.
84) The third paragraph adds more confusion than benefit. I suggest
85) The first sentence of the fourth paragraph is redundant to
5.11. Delete it.
86) The second sentence of the next to last paragraph is redundant
and confused. Delete it. The last sentence adds no value. Delete
87) In 5.12.1 I think "relation" should be "relationship".
88) In the third paragraph what does "A useful and expected to
be common value is 64" mean? Is this a disguised mandatory requirement
or a disguised advertisement for a particular vendor?
89) Delete "adopted from FCP".
90) Assuming much of SBP is important, in 5.12.2 delete "It is
important to recognize".
91) Change the second sentence of the second paragraph to "The
target shall accept an implementation specific, minimum number
of taps ...". Having made the statement correct from a standards
standpoint what does it mean? To me it means the target is free
to accept any old number of taps so why bother saying anything?
92) In the third paragraph delete "It is convenient to use the
term tap slot when dealing with the obligations associated with
accepting a tap sent in a given command FIFO." Also delete the
rest of the paragraph.
93) In the last paragraph replace "from" with "in". Also replace
"overlap" with "overlapped".
94) In 5.12.3 replace "decides to manage" with "manages".
95) The second paragraph needs major work to translate it to something
for a standard.
96) In the third paragraph replace "Observe, the given tap" with
97) Replace the "The very latest point in time for the target
to release a tap slot is when" with "The target shall release
the tap slot no later than when".
98) Delete the last two sentences of the paragraph.
99) Delete the next paragraph.
100) Delete the first two sentences of the fifth paragraph and
change the remaining sentence to "The target shall provide an
101) In the following paragraph combine the second and third sentences
to "The notice is given in only one Status block by setting the
TSA flag to one."
102) Replace "is given permission by the Serial Bus Protocol to"
103) In the next paragraph what does "is the latest time at which
the target that the initiator that the tap slot has entered the
"available for use" state" mean?
104) The next paragraph is too confused to keep.
105) What is "TAP SLOTS ALLOCATED FOR USE BY AN INITIATOR:"? Why
is it in all caps? Can the rest of the paragraph be put into acceptable
style for a standard?
106) In the next paragraph replace "Given that the initiator gets
both an identifier and an allocation of tap slots, it is strongly
recommended then the initiator decide to be a "good neighbor"
and limit its" with "The initiator shall limit its".
107) In the last line of that paragraph delete "so as".
108) Delete the fourth from last paragraph.
109) What is "USE OF TAP SLOTS FROM THE GENERAL POOL OF TAP SLOTS:"
and why is it capitalized?
110) Delete the first sentence if (109) was not the first sentence.
111) This paragraph should be translated from free form discussion
to something appropriate for a standard.
112) Consider deleting the last paragraph covering one of an infinite
number of things that are not requirements.
113) Section 5.12.4 and 5.12.5 require major editorial work.
114) In Figure 1 it is not clear what the M_flag is equal to.
115) In the next paragraph replace "a available" with "an available".
116) Decide whether "acknowledgment" should be lower case or upper
case and make it consistent.
117) In the fourth paragraph after Figure 1 delete "Before continuing
with Example 1, it should be mentioned that" and "In this spirit,
the explanation continues as to command delivery." Also delete
"As a major emphasis point,".
118) Delete the first two sentences of the third from last paragraph.
Having done that decide if the balance of the paragraph should
119) In the next paragraph delete "between the situations depicted
between the two examples".
120) Delete or salvage the last paragraph.
121) Change section 6 to "Figure 3 is an illustration".
122) In the last sentence of the first paragraph delete the redundant
"The illustration shows a single task set;".
123) Figure 3 appears to be a block diagram of more than SBP.
I can also see 1394.
124) In the first paragraph of 6.1 what is the definition of "units"?
125) Figure X should be Figure 4.
126) In 6.1 where does the number of quadlets in the directory
entry start? Does it include quadlet one?
127) How does the key (12h) indicate that the low order 24 bits
of this quadlet contain the vendor ID of the entity defining the
architectural interface of this node? What is "the architectural
128) How does the key (D4h) indicate the offset and which item
is the offset from?
129) In the first paragraph after Table 4, the "base address"
is which base address?
130) In the last note replace "may implemented' with "may be implemented".
131) In 6.2.1 replace "This tap" with "This FIFO".
132) The nomenclature in Table 5, 6, and A-1 do not match the
conventions of 4.1.
133) In the first paragraph after Table 5 replace "must" with
134) In 6.2.2 replace "The initiator is expected to" with "The
135) Replace the last sentence of 6.2.3 and 6.2.4 with the equivalent
of the last sentence of 6.2.2.
136) Replace "This tap" in the first line of 6.3.1 with "This
137) Repair 6.3.4's "to anasync FIFO based upon existance based
on the existance" somehow.
138) Also repair "The number of isoch ACA FIFOs available one
per stream identifier."
139) In 7 replace "as result" with "as a result".
140) In the next to last paragraph change "The initiator is expected
to" to "The initiator shall" and "can associated with" to "can
be associated with".
141) Change "definde" in 7.1 to "defined" and run a spelling checker
on the document.
142) Some tables use Byte, some Bytes, and some Displacement.
Determine which should be used.
143) What does "The CDS address is the value contained in the
address of this CDS field" mean?
144) Is there an essential difference between rules 3 and 6?
145) What does "No consideration is give with regard to availability
of tap slots from the general pool" mean?
146) Is rule 7 "Available means the target considers that it has
the resource available to accept a tap" helpful? To me it sounds
like available means available.
147) In section 8 a more specific reference than "in this standard"
should be given.
148) In the second paragraph of 8 replace "address contained there"
with "sense buffer" and "must be" with "is".
149) In the last sentence what is "codes field"? Is it the CDS
codes field of Table 15?
150) After Table 14 what field does the note apply to?
151) In the next to last paragraph FIFO should be capitalized.
152) Where is Table x?
153) Should the description of value 1h in Table 16 be Format
specified in Table 14?
154) In Table 18 what is a "Unit management CDS"?
155) Where are the values 1h through 3h of Table 20 defined?
156) After Table 24 replace "is allowed to modify" with "shall
modify". Change "the chain contain the CDS" to "the chain that
contains the CDS".
157) What are we to learn from the second sentence of the Note?
158) Regarding the "O-flag" in addition to replacing "must' with
"shall" how can data over the 1394 be transferred in other than
sequential order since it is a serial I/O?
159) Where is the scatter / gather function defined?
160) For the case where the M-flag is cleared and the link bit
is not, shouldn't the additional sense codes be specified?
161) Replace "The current CDS marks" with "A CDS with the SCE-flag
162) Why is the S-flag described out of sequence? In item b change
"the data" to "data". Change "This data address" to "The data
163) In Table 26 are Bytes 56 through 71 for Login only?
164) What are the units of the division?
165) In the seventh paragraph after Table 27 replace "must' with
166) In the next paragraph I appreciate the plug, even if it is
spelled with a "k" but doesn't it apply to targets which are not
disc drives also?
167) I think the note should be up near the definition of R.
168) The three definitions following Table 28 are shams. "Black
is black" does not pass as a definition. If nothing more is to
be said delete them.
169) The fields noted after Table 29 are not contained in Table
29 so why are they mentioned as not used if they don't exist?
170) Does Table 31 apply only to isochronous? Should this be mentioned?
171) It would be clearer to replace "tag value field" with "tag
172) In 10 how much should the fetch exceed 72 bytes by?
173) Where is "transaction time-out time" defined?
174) Replace "rejects a target fetch" with "rejects a target's
175) In the last portion of 11.1 there is a conflict. It is not
possible to perform an or statement in order since only one item
176) In the next to last paragraph of 11.2 where is the "pending
acknowledge code" and the "resp_normal rcode" specified?
177) In 12 and 13 what does "possibly a write a response packet"
178) In 14 what are the protocols in addition to? Change the last
paragraph to "Targets shall not respond to SBP ... identifier
for use ... identifier for use ...".
179) In 14.1 delete the extra space in the title or define "p"
180) Please accept the "an/a prize" and then change "obtains an
identifier or a identifier and the addresses of command FIFO"
to "obtains an identifier or an identifier and the
addresses of the command FIFO". However should address be singular
or FIFO plural? Is the address a fixed address?
181) Change the second paragraph to "obtain an identifier".
182) The location of the Async login FIFO should be moved from
14.2.1 to 14.1.1.
183) In the next to last paragraph of 14.1.1 if the login is not
accomplished how does the target know where to return sense data
and status? If the paragraph is correct otherwise, change "protocol,
then" to "protocol, and then". These comments also apply to 14.1.2.
184) In the last paragraph of 14.1.2 what does "using the protocols"
mean? Is it something like "taking the waters"?
185) In 14.2 I presume the Initiator resources should also be
186) Replace "isochronous data transfer on the given stream" with
"isochronous data transfer of the given stream". Is it necessary
for the first two sentences of the second paragraph to be so redundant?
187) In 14.2.1 I think the initiator identifier should not be
released until after the status is returned. Otherwise how is
it known who to return it to?
188) Add the phrase "without performing another login" to the
end of 14.2.1 and 14.2.2.
189) In 14.3 change "request the target assign it" to "request
that the target assign it". Change "In the alternative," to "Optionally".
Delete "wish to". Change "In either situation, initiator may"
to "In either case, the initiator may".
190) Change "optional to " to "optional for" and "mandatory to"
to "mandatory for". In the next to last paragraph the first sentence
states tap slot allocation is mandatory for all targets but the
third states it is only mandatory for targets that are supporting
asynch transfer. If the first sentence is correct reduce the paragraph
to only the first sentence. If the first sentence is wrong correct
it and reduce the redundancy it the rest of the sentences and
convert them to "standard" language.
191) In the last paragraph change "must" to "shall". Is it intentional
that initiators that don't request allocation of tap slots can
hog all tap slots?
192) In 14.4. replace "at any time thereafter" with "later".
193) In the second sentence delete "at the target". Change "In
the alternative," to "Optionally". Delete "wish to gain". In the
second paragraph delete "one and".
194) In the next to last paragraph change "in the AE sense length"
to "if the AE sense length".
195) What takes place if the AE sense length is greater than zero
but less than required?
196) Shouldn't there be a section for a Sign-out procedure?
197) In the scope of Annex A replace "carries with it" with "has".
198) In the last paragraph of A.1 the first sentence states the
ROM is an example. The second sentence states that again and adds
that it is not definitive. What does this mean? Is the example
wrong? Why not delete the second sentence?
199) In the third sentence replace "ROM must be parsed for pointers"
with "ROM is parsed for pointers". ("Must be" is as objectionable
in an informative annex as is "shall" even if it is not correct
200) Adjust the column sizes of Table A-1 to avoid the odd word
wrap in the first column.
201) In addition to not following the hex convention, Table A-1
does not follow the binary convention of 4.1.
202) In the table and in A.3 there appears to be room for "Capability"
rather than "Capab.".
203) It is nice to know that the "lost" bit has been found but
probably "last" was intended.
204) In the table and in A.3 there appears to be room for "unit
directory offset" rather than "unit directory off".
205) In the unique_id_hi delete "is".
206) The description of async login FIFO key states in effect
that the lack of the async login FIFO entry in the unit dependant
directory indicates that the target does not implement all mandatory
elements of the async data transfer protocol. Is this allowed?
An analogous comment applies to the isoch login FIFO key.
207) In A.7 replace "this FIFO" with "the Async FIFO".
208) The SCSI-3 queuing model is not an appropriate reference
several places in Annex B. It should cite SAM.
209) What about multiple LUNs with regard to linked commands?
210) Delete "As a further restriction". Delete both "shalls" in
the second full paragraph and change the second "take" to "takes".
211) ACA is not a queue type. It is a task attribute.
212) In the last paragraph replace "accept a modestly reduced
level of control over the time sequence of processing" with "accept
reduced control of the processing sequence".
213) In B.1 replace "If there exists a sequence" with "If a sequence"
and "commands which are to be completed" with "commands to be
completed". Delete "interested in the situation".
214) In the second paragraph change "The first alternative is
probably best suited to the case in which only one initiator is
relevant to the situation." with "The first alterative is best
suited to single initiator task sets."
215) Should "in front" be replaced with "at a lower address" or
"at a higher address"?
216) In the third paragraph delete "This is to say, the initiator
constructing the given chain has it within its own ability to
control fetch policy by the target so that the given chain is
processed in the manner considered appropriate by that initiator."
217) In the last paragraph delete "given". Make "relative to commands"
singular unless "command blocks" in this case should be "CDS".
Delete "interested". Move "time" from "sequence" to "stamps".
Delete "desired relative". Delete "and therefore". Delete "in
the desired order" after "stamps". The net result is "Taps for
these several chains are sent in an appropriate sequence such
that the order of arrival (implied time stamp) is achieved at
218) The first paragraph of B.2 is confusing. The next two need
to be worked on.
219) In B.3 delete "subject". Delete "value" twice. Better yet
reduce the paragraph to just the first sentence. If the other
sentences are not deleted Change "For this final" to "The final"
and "must indicate" to "indicates". But why are the linking requirements
being repeated here?
220) Unless this Annex should be made Normative, in the Scope
replace "necessary to implement" with "on". Make "formats" singular
or change the related section name.
221) In C.2 change "some terms which are used freely throughout
this document" to "special terms used in this Annex."
222) Does 1394 have "targets"?
223) Should "sy" be upper case?
224) Change the paragraph after Table C-1 from "that listener
treats such a packet as an error" to "the listener declares an
error." In the e.g. statement should "halt" be "report and stop"?
225) In the next paragraph should it be "synchPeriod field" or
"synchPoint field"? (The Normative Annex uses neither term.)
226) In the last paragraph in two places should it be "this standard"
or "this Annex"?
227) I think the last paragraph of C.4.1 and the third paragraph
plus two bullets of C.4.2 should be moved into C.4. ahead of C.4.1.
Adjust "above" when this is done.
228) Why aren't potential talker errors included?
229) Three paragraphs ahead of Table C-2 change "Notice that a
stream gap" to "A stream gap".
230) Unless the Annex is changed to Normative, change the line
before table C-2 from "defined" to "shown".
231) What are the units of the seconds count?
232) In C.4.3 what constitutes execution? If the OK to continue
bit is a one, when does the
233) Why is STREAM CONTROL a normative Annex rather than included
in the body of the standard?
234) Change "must have already obtained" to "shall have obtained".
235) In the third paragraph change "contents" to "Operation Code"
and "determine" to "designate" or "specify".
236) Where are the Operation Codes and Control Field codes specified?
237) Change the fourth paragraph to "talker and the command is
pause, the target shall wait for".
238) Change the fifth paragraph to "If the function code is start,
the target shall either ...".
239) Where is the "identifier field"?
240) What are the units of the synch period field and the seconds
241) In the paragraph before Table D-2 delete "the following table
and described in".
242) Should there be definitions for abort handling?
243) It appears the two paragraphs after the table should have
244) The parenthetical statement under the table seems wrong since
this is the normative section. Where are they defined elsewhere?
245) For the stream event field what about the twelve other codes?
246) For the Action field what about the thirteen other codes?
247) Do all devices have to support a byte offset of 2 to the
248) Why is only the target required to set the reserved field
249) In the last paragraph change "the target shall treat this
as an error" to "the target shall declare an error". Feel free
to choose other than "declare" but don't make it a non-error treated
as an error.
250) Error handling is promised to be in a later section of this
appendix (SIC) in the last line of the Annex. Who is planning
to make good later on this promise?
251) Change the title of Annex E to "Read examples".
252) Should the latter portion of the first paragraph be "optional
split transactions are not recommended"? Why has the option been
included if it degrades performance?
253) Replace the second and third paragraph with "Although SBP
is defined to facilitate commonality with other SCSI transport
mechanisms, the remaining sections include some additions required
to support the 1394 transport."
More information about the T10