Section 6.10
EXECUTE DIAGNOSTICS: Host Shall assert Reset (SRST) prior to issuing.
Section 6.2 SoftReset (SRST)
Device side State Diagram: D0SR3 "Set_Status_State"
second paragraph
"All actions required in this state shall be completed within 31 s"
Page 86 Volume 3 ATA/ATAPI 7 rev 0, 5 November 2001.
Sorry for the old reference it was the quickest hard copy I could find.
The world of the HOST driver does not live in software alone.
You have to read the device side rules to obtain the rest of the solution.
Yeah this stinks, but very few of us are nutty enough to get the entire
process.
What I need most is people like yourself who are skilled in the kernel api
to help me get that part correct. This stuff is easy and can drive you
insane. Since I am there, you are welcome to come but I don't think you
are weird enough yet :-). If you want to visit my world and stay, jump in
and bring a barf bag for the first part of the ride. Also borrow Linus's
crack pipe too, mine is in use.
Cheers,
Andre Hedrick
LAD Storage Consulting Group
On Fri, 23 Aug 2002, Adam J. Richter wrote:
> On Fri, 23 Aug 2002, Andre Hedrick wrote:
> >On Thu, 22 Aug 2002, Adam J. Richter wrote:
>
> >> 1. Regardless of whatever specification you are referring to
> >> or Andre's "31 second rule of [Power On Self Test]", it is genuinely
> >> useful to boot faster by overlapping some other kernel work before the
> >> drive is. Specifications ultimately exist only to serve this
> >> usefulness. When a specification impedes usefulness, sometimes it's
> >> the right decision to violate it. Of course, we're not talking about
>
> >Listen to yourself, and understand why 2.5 failed.
>
> >"When a specification impedes usefulness, sometimes it's the right
> >decision to violate it."
>
> >"Gee there is no traffic in the on coming lanes, maybe I should use them."
>
> >There are rules for how the hardware works, and if everything out there
> >comes up in 4 seconds great. If everything returns faster than the worst
> >case great. You start assuming everything behaves that way and you repeat
> >history.
>
> >You guys in 2.5 walked away from the rules because you thought you knew
> >better, where did it get you? Lost interrupts, PIO command block
> >exectution failures, dropping EOT on PRD's because reading something into
> >the published documents which is not there, etc ...
>
>
> Those are not examples of cases where the specification impedes
> usefulness. Those are examples where a specification helped usefulness
> by helping compatability (not driving in oncoming lanes, not having timing
> problems), and a decision made either out of recklessness (driving into
> oncoming lanes) or, I infer, out of a misunderstanding of some document.
>
> However, your point is well taken in that when I said
> "sometimes it's the right decision to violate [a specification]," I
> did not mean by "sometimes" that this is a random event, like
> "sometimes it rains." I meant that it's a very careful decision, the
> details of which differ from case to case, and that the result is not
> always against.
>
>
> >> your IDE code violating such a specification, but rather not relying
> >> on this particular guarantee.
> >>
> >> 2. Besides, if this code is supposed to be a generic IDE core,
> >> it many need to run on platforms that do not provide that guarantee or
> >> where the boot code is not even capable of finding where all of the
> >> IDE controllers.
>
> >It is a means for probing signatures w/o identify to test for presence.
> >It to has a 31 second rule. Break the worst case and device get lost.
>
> Can you provide a reference for this "31 second rule?" If
> your reference does not directly discuss how it would impede the test
> that you refer to, then you might want to explain that too. Thanks in
> advance.
>
> >Please do not take this personal, because it is a technical arguemnet.
>
> Of couse, likewise.
>
> >We do it by the books and then we cheat when we can, but only after we
> >have all the proper stuff in place for compliance.
>
> Let's keep in perspective that I am talking adding a test that
> would cause a delay until it is safe to proceed (unless you're saying
> that waiting for the busy bit to clear is insufficient, in which case
> I'd like to know how, but it would still be no more dangerous), and
> you are advocating skipping that test.
>
> Adam J. Richter __ ______________ 575 Oroville Road
> adam@yggdrasil.com \ / Milpitas, California 95035
> +1 408 309-6081 | g g d r a s i l United States of America
> "Free Software For The Rest Of Us."
>
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Aug 23 2002 - 22:00:27 EST