RE: 2410SA controller in non-intel system?

It is fit for use, even if you want to keep your data.

This software is provided `as is' by Adaptec and any express or implied
warranties, including, but not limited to, the implied warranties of
merchantability and fitness for a particular purpose are disclaimed. In no
event shall Adaptec be liable for any direct, indirect, incidental, special,
exemplary or consequential damages (including, but not limited to,
procurement of substitute goods or services; loss of use, data, or profits;
or business interruptions) however caused and on any theory of liability,
whether in contract, strict liability, or tort (including negligence or
otherwise) arising in any way out of the use of this driver software, even
if advised of the possibility of such damage. Yadda Yadda Yadda.

The driver is *now* 64 bit clean (around April 2003, for IA64), but has done
the PCI accesses the proper way for some time.

Once the array is configured, the firmware will activate it (ie, when the
build has started or zeroing has completed). *The* experiment that needs to
be performed on your machine is *if* the BIOS can be run properly. You may
use almost *any* aacraid based card to perform this experiment, it need not
be the 2410A specifically.

As for array failure notification without the management applications,
sorry, nothing will show until it is *failed* and data is lost, degraded
status will have to be checked using the BIOS periodically. It is *possible*
for the driver to be enhanced to report status, but there is resistance both
on the space utilization in the driver and the proc filesystem being shunned
going into the future. Ideas on mechanism? There is also the possibility to
explore a *simple* application that communicates with the ioctl to pick up
said status; this would be a skunkwork. I will check if there is something
internally that has been done that can be released, no promises. Dell has
their own management application that communicates directly to the ioctl
too. The benefit of asking for this is beyond this specific issue of an
Alpha box.

The utility source code is proprietary at the moment. With the I2O based
cards, we released the management applications into the public domain, I
have pressed to do so with the AAC based cards recently and the answer was a
firm no at this instant. NDA's can be signed, usually a pressing business
case drives them though. Can I say `no' any more softly?

Sincerely -- Mark Salyzyn

-----Original Message-----
From: Måns Rullgård [mailto:[email protected]]
Sent: Tuesday, September 30, 2003 10:34 AM
To: Salyzyn, Mark
Cc: [email protected]; 'Mark Haverkamp'
Subject: Re: 2410SA controller in non-intel system?


"Salyzyn, Mark" <[email protected]> writes:

> The aacraid driver in the 2.6 Kernel is being brought up-to-date. It is
> stable (IMHO), as the changes are mainly centered around feature
> enhancements. There are some corner case instabilities that have been
> addressed.

I take it that it's fit for use, even if you want to keep your data.
Is that correct?

> The driver has, in the distant past, been tested on Alpha machines running
> Linux. I have the DS10 I believe it was tested on in my Cube, handed down
> through the ages, but it is collecting dust and I have never turned it on
> myself. I do not believe the testing was done with any business case in
> mind. I am afraid to turn it on, only because my plate is full and I need
> not add one more to it ... Someone here at Dell may be able to give a more
> definitive history.

Well, at least someone thought about it.  I've seen drivers that would
never work on Alphas because of the way PCI memory is mapped.  Is it
safe to assume that the aacraid driver is 64-bit clean and does PCI
accesses the proper way?

> The driver can benefit by the native management applications, but I fear
> that the only means you will have to communicate will be the BIOS. Yes,
the
> BIOS (under Intel, and hopefully under the ARC x86 emulation) can
configure

I don't mind firing up ARC once or twice.  Once the array is
configured, will the BIOS need to be run to activate it, or will it be
possible to use SRM?  SRM apparently doesn't run BIOS code from all
PCI cards.

> arrays. Array failures may require you to take the system down to take
> action since the native management applications may not be available. 

I could live with that, as long as I can find out about the failure in
a portable manner.

> I was the BIOS author for the 2400A, and had to go through hoops do deal
> with the protected memory (32 bit) access required by the BIOS when in the
> ARC environment. There were enhancements made (specifically in the int 15
> protected mode functions) in the ARC BIOS made just for the I2O based
> adapters such as the 2400A. The 2410S uses a totally different BIOS and
> interface, but still has the protected memory access issues. I had talked
> with the author of that BIOS at length about what had to be done (nearly 4
> years ago), and accesses were done with similar intents, but have no idea
if
> it was ever tested. No doubt it was done on the DS10 in my cube if it was.
I
> would be interested in your overall success.

I'd be happy to try it out, but if it fails it's $400 down the drain.

I don't suppose there's any possibility to get access to the source
code of the utilities, is there?

-- 
Måns Rullgård
[email protected]

_______________________________________________
Linux-aacraid-devel mailing list
[email protected]
http://lists.us.dell.com/mailman/listinfo/linux-aacraid-devel
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/