This message is from the T13 list server.
David F. wrote:
This message is from the T13 list server.
To summarize:
I'm suggesting that the "Command Protocol" sections of the spec be
"tightened" meaning "more clearly defined". That is, taking the various
protocol tidbits that are spread out throughout 3 volumes be placed or at
least referenced in their correct location in the "Command Protocol".
There is a growing problem, started some time ago with the "invention"
of serial ATA, that T13 is unable to unwilling to deal with.
First, serial ATA is not ATA. Calling it ATA may be good marking hype
but this thing we call serial ATA is really a very different interface -
just about as different as parallel ATA is to parallel SCSI.
Second, the protocol descriptions in ATA/ATAPI-x for traditional
parallel ATA are not the protocol descriptions for serial ATA. This
thing called serial ATA should have never been merged into ATA/ATAPI-x.
ATA/ATAPI-7 is a mess - it is confusing and often incorrect because in
many places it fails to recognize there are differences between
traditional parallel ATA operation and a similar operation in serial
ATA. Anyone working with parallel ATA should not use ATA/ATAPI-7 -
ATA/ATAPI-7 should not be considered a valid description of parallel ATA.
Third, based on the public and private emails with David F., I am sure
the problem he is having is the result of a failure in his serial ATA
host controller to correctly emulate traditional parallel ATA. To
isolate this problem, David would probably need to get PCI bus traces
and serial ATA traces to see what is causing the problem. Please
remember that David has a serial ATAPI device. I have made the following
guesses:
a) it is a design bug in the host controller.
b) the host software is doing something unusual that is confusing the
host controller.
c) the host controller really doesn't support a serial ATAPI device.
d) the device is doing something unusual that may not fail on other
serial ATA host controllers but fails on this serial ATA host controller.
But the real problem here is that most serial ATA host controllers
emulate parallel ATA operation to be backward compatible with old
parallel ATA host software. The fact that this is happening is not known
to many people. This emulation is no different than many other interface
emulations that exist today (example, USB based SCSI that is really ATA
that is used by most external disk drives). The descriptions of how a
serial ATA host controller emulates parallel ATA host controller/device
operation are not very good in ATA/ATAPI-7. But then the T13
descriptions of any host controller operation are not very good (or
don't exist). Because these emulations are so poorly defined we should
expect more of these problems as long as the myth that serial ATA is ATA
continues. It is time to resolve this problem. Let the SATA secret
society publish the serial ATA standards and have T13 either continue to
publish the parallel ATA standards (why?) or disband T13. Or maybe T13
should take a new direction and publish only standards for how interface
emulations and conversions operate?
Hale
--
++ Hale Landis ++ www.ata-atapi.com ++
