Ultra 160 Support in Redhat 7.1 (aic7xxx 6.1.7)

[Justin T. Gibbs]

>Hello Justin,
>
>My question was: Is there Ultra 160 support in version
>6.1.7 of AIC7XXX driver?

And I answered that question.

>Advising to upgrade to a newer version does not really
>help me at this moment. This is a production server
>with AIC driver built into kernel and upgrading it is
>an option only if there is no U160 support from
>developers. 

Whom are you referring to with "developers"?  I wrote and
maintain the aic7xxx driver.

>I am trying to track down the problem with U160 15K
>Seagate Cheetah. I am assuming that the problem
>originates from the AIC driver. But I have to get a
>word from developers if this is the reality.

I have yet to see a reported parity error that was not
really a parity error.  Parity errors are caused by a flaw
in your configuration (cabling, termination, etc. etc.).

>The same system boots just fine from DOS (floppy boot)
>with 2 drives and we have copied 9 GB IBM Ultra2 drive
>to 36 GB 15K Ultra 160 Seagate drive without a hitch.
>I mean, both drives are visible and work fine from
>DOS.
>
>So, this definitely is not a termination/bad
>pin/cabling issue.

Ha!  If you are relying on BIOS services, you will get perhaps 3 or 4
concurrent commands running on the bus.  Compare this with the 253
concurrent commands that Linux will attempt.  Note also that the BIOS
will perform domain validation and may have reduced your bus speed,
without you even knowing it, to make the bus function correctly.  The
Linux driver does not include domain validation.  Parity errors are
never reported by the BIOS unless they cause a catastrophic (retries
exhausted after falling back to asnyc/narrow) error.  I can't recall if
the BIOS will even performed tagged queing.  You are comparing apples
to bacon (not remotely close enough to say oranges).  I'm sure there are
lots of QA engineers that will tell you that testing something at 10%
load does not guarantee that it will function properly at full load.

--
Justin

To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message