[Coco] SuperIDE backup/restore

[Gustavo Ranaur Schoenaker]

Ed,

Can you replicate the test I made with your cards? Just transfer a disk
from both cards (the original and the copy) and compare the result. If you
find the same pattern of erros, at least we are on the same boat.

Timing problems can be a good theory. Maybe some cards send data to the
controller with different timings and SuperIde can't handle all of them in
a reliable form. If this is true, is something useful to know. It won't be
a bug, it will be a documented feature. :)

Any ideas on how to test it? I may try to write disks with specific
patterns to see which of them cause errors ...

On thing is certain, the bytes are "swallowed" in pairs. And after
swalloing a pair, it doesn't swallows the next pair of bytes. I never found
a 4 byte swallow, but it may be just coincidence.
I know nothing about the IDE/CF standard. The data bus width is 16 bits?
(could be ... they have 40 pins ...)

BTW, my image copy from good cards (I have 3 good cards!) works flawlessy.

After my vacations I'll take a look on HDB-DOS source.

On Mon, Aug 12, 2013 at 7:53 AM, Mark Marlette <mmarlette at frontiernet.net>wrote:

> Mark
> I don't recall the mode ATM. 2904 long time ago.
> I did have Ed runn a test and he expanded on it.
> Basically. ..VERIFYON DSKINI
> He did this for all drives incrementing up down sideways, etc. No problems
> detected.
> The problem in Ed's case is when he uses an external program on CF to
> image or transfer.
>
> My results was that sidewalk was not creating a clone. Aaron gave an
> explanation but as I recall it acted just like Ed had noted and I
> observed.  Ed then tried Winhex, which jis what I use and is still having
> issues.
>
> I had Ed's CFs once before, imaged them did hist test that he just
> described and had no issues here in the lab.
> He has sent in his color code CFs and his sIDE and I am going to test
> again when I get some free time towards this next weekend.
>
> Regards
> Mark
> Sent from Yahoo! Mail on Android
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco
>