Anthony's drive issues.Re: ssh password delay
Anthony Atkielski
atkielski.anthony at wanadoo.fr
Tue Mar 29 11:01:49 PST 2005
Ted Mittelstaedt writes:
> He is saying that the microcode was modified and that we speculate that
> the mods contain a bug proprietary to the HP implementation of that
> controller.
What makes it a _bug_? Why would the modified firmware contain a bug
... but not FreeBSD?
> Or had whatever extra code was needed for the microcode mods.
Yes, or approached the hardware in a way that made the modifications
irrelevant.
> Yes, they do - I've got a Compaq professional workstation on my desk
> at work which has a modded microcode in an Adaptec 2940U adapter card
> (I know it's modded because the card will not work in any other
> non-Compaq system, even where non-Compaq-branded 2940U cards will
> work) that displays similar disk strangeness (although it doesen't
> spew errors) This is the same scsi chipset as Anthonys Vectra.
> (aic7880)
And what does Compaq give you in exchange for the headache of a
non-standard adapter card?
Can you replace Compaq's distorted adapter with a standard one, or is it
theirs or nothing?
> This incidentally is WHY I am speculating it's a microcode mod (and it
> was I that started this line of discussion regarding the microcode on
> his SCSI chipset) because I have proof positive that modded microcode
> in other manufacturer's aic7880-based SCSI adapters has problems with
> the ahc driver.
How did you resolve the problem?
> He doesen't want to run Windows (on this system at least)
Correct. It's a more or less spare system and I'm more interesting in
getting more experience with UNIX than with getting more experience with
Windows. I already know plenty about Windows.
> He wants the FreeBSD ahc driver modded so that it won't generate
> errors and SCSI bus resets anymore under FreeBSD.
That would be nice, if it's a legitimate bug in the FreeBSD code (which
I suspect it is). If it's a regression (i.e., a change that would break
the behavior with standard hardware), then the utility of changing it is
debatable (although I still wouldn't object to a version that would run
on my hardware).
In any case, this wonderfully fun experience is pushing me more and more
in the direction of home-built hardware, and further and further away
from brand-name machines. I'm glad I decided to build my own server
instead of buying that IBM eSeries machine. Who knows what problems I
might have had with it?
> Unfortunately, Anthony won't do the least bit of troubleshooting (such
> as pulling the Quantum disk and just running on the Seagate disk in
> this system to see if perhaps the problem is execerbated by one or the
> other implementations of SCSI in one or the other of the disks -
> granted that is a long shot, but it's within the realm of possibility
> it might fix it) so I doubt he would do anything that the ahc driver
> (who most likely isn't even subscribed to freebsd-questions) tells him
> to do in the way of troubleshooting either.
Anything isn't going to do anything until someone can tell him what the
existing messages are saying. I don't go pulling boards every time I
see a message that I don't recognize.
> Also long forgotten in this discussion is Anthony stated once on this
> list that Mandrake Linux wouldn't even install on this Vectra system
> either.
It stops after the splash screen, but I think that is related to the
same problem that prevents FreeBSD from booting directly from disk.
> I am not sure why he's trying to hold FreeBSD up to the driver support
> of Windows NT when Linux won't even talk to the card in his system.
I don't know what Linux will or won't do, and unlike you, I'm not
prepared to make wild guesses. I know only that Mandrake Linux will
stall after displaying a splash screen, and that's that.
> ... and we all know that Windows has far better support for the
> oddest-ballist modifications of standard computer components such as
> SCSI adapters than FreeBSD does since they have unlimited money to buy
> oddball samples of hardware to experiment with ...
I suspect they just ask the vendor for information on the hardware.
Even Microsoft has neither the time nor the money to test every
conceivable hardware configuration.
A more likely scenario is that the vendor itself writes the driver and
then has Microsoft certify it. The certification is pretty rudimentary,
IIRC; essentially MS ensures that the system doesn't melt or spew acrid
smoke when the driver is invoked and that's about it.
--
Anthony
More information about the freebsd-questions
mailing list