Okay, to me it sounded as the name of the supplier. REminded me of Broadcom a bit. But anyway, to the customer it is totally uninteresting whose fault it was if he got a device that does not work correctly. He just wants the manufacturer to stand for it and correct no matter if he made those errors himself or any of the suppliers be it Genesi, Freescale or whoever. So that "except" is no meaningful workaround? To the customers it would be if he got more speed and responsiveness by it, I am sure.
Just to understand everything right: AFAIR there was some statement somewhere that only Linux distribution on EFIKA could make use of DMA (I think "real" DMA instead of what you call DMA-assisted PIO). I do not remember the pointer to it, nor do I remember which distro it was. Is it THIS distro that uses your workaround and all the others do not (including commercial MorphOS) or are the others using it (incl. MorphOS) but this special distro invented something else?

Here is the last round of communication with respect to the BestComm engine and the Efika 5200B...

On Dec 11, 2007, at 3:59 PM, XXXX wrote:


Bill,

The file of documents is quite large (130MB) so I have made it available to you on the Freescale web site. You will receive a mail from the system with instructions on how to get the files. You will also need the pass code included in this mail.

Please let me know if you have any questions in getting the files down.

Thanks,
XXXXX


From: Raquel Velasco and Bill Buck
Sent: Wednesday, December 12, 2007 7:10 AM
To: XXXXX
Cc: ZZZZZ; Dave Mothersole; etc...
Subject: Re: Agreement(s)

Hi, we have been through the documentation. In the "5200b" folder we find another called "bestcomm_2.2.1." Inside that folder there is another called "training" -- herein is a PowerPoint presentation, BestComm-tutorial.ppt. XYZ123 saved the PowerPoint presentation the last time in May 2004. Is XYZ still there? The SmartDMA Assembler or SAS is what we need. Doesn't Freescale have anywhere the real documentation of the BestComm device, the compile engine/sources or the chip internal information?! :) Even to have the known dependencies between modules would be helpful.

Best regards,
R&B

From: "ZZZZZ" <ZZZZZ@freescale.com>
Date: December 12, 2007 7:22:51 AM CST
To: Raquel Velasco and Bill Buck, "XXXXX" <XXXXX@freescale.com>
Cc: Dave Mothersole, etc.
Subject: RE: Agreement(s)


R&B,

We have given you what we have. The design team for the MPC5200B was disbanded (most of those folks left the company before I joined Freescale) and you have all of the documentation and tools that we have. We can also provide technical support for the pre-defined tasks.


Best Regards,

ZZZZZ

From: Raquel Velasco and Bill Buck
Date: December 12, 2007 9:02:39 AM CST
To: ZZZZZ <ZZZZZ@freescale.com
Cc: Dave Mothersole
Subject: Re: Agreement(s)

OK, ZZZZZ, now we understand why you abandoned the 5200B before. :(

We sympathize with your situation, but from a Freescale perspective it is nearly criminal. XXXXX, YYYYY and probably others were there before you and are still working at Freescale. Surely, in their files they must have something. We are not chastising you. It is an awful dilemma.

R&B

===========

ZZZZZ ending up being laid off at Freescale and the Division he managed was shut down. This was the same individual that later denied having cache-coherency problems with the 5121e. He also contributed significantly to the THTF fiasco.

Digital DNA was a great concept (background link). BestComm was part of that marketing package. Unfortunately, we could never get enough information about the design itself. The ATA state machine should have been designed properly in the beginning to support higher system loads.

The fact is, in the end, this is why it is called 'development'. For $99, the Efika was an inexpensive, flexible piece of hardware that enabled some great projects.

R&B

_________________
http://bbrv.blogspot.com