[Coco] Connect CoCo Floppy drive to Windows PC

[Gene Heskett]

On Monday 19 January 2009, Steven Hirsch wrote:
>On Mon, 19 Jan 2009, Gene Heskett wrote:
>>> I'll add to that.  It's something I've noticed when moving disks back and
>>> forth.  Once it's been formatted on a CoCo, my Linux box chokes on it.
>>> Other way around works fine.
>>
>> And as you'll note, my experience has been the opposite.  What I think is
>> wrong is that the linux formatter doesn't like sector 0's, on any track. 
>> But that's just what I call a SWAG too.  I have not put a scope on it and
>> compared them in that manner.
>
>Are you quite sure that the CoCo writes a sector id of 0?

Having walked around a little in the floppy drivers code a decade ago, and 
played with the docs on the chip using pokes, it would appear to.  Or my 
aging wet ram is playing up.  IIRC the last, or 18th sector of a track is not 
$12, but $11. Std base zero numbering.

>>>> FWIW Bill, I find that when I am doing nitros9 disk images here in this
>>>> linux box, Fedora 8, generic kernel local build, that I must first
>>>> format the disk on the coco in order for it to work correctly, otherwise
>>>> dd gets a tummy ache a few sectors into the write and dumps out.  If
>>>> anyone knows the secret of properly formatting a coco disk on linux, I
>>>> would appreciate a hand. Software to use, command line to use, etc,
>>>> please.
>>>
>>> Exactly the opposite from my experience.  I _can_ tell you that the disk
>>> parms are incorrect on many systems.  Here's what I had to do on a
>>> Xubuntu 7.04 box:
>>>
>>> Install 'fdutils'
>>
>> And what version does that install?   5.4 works mostly here, 5.5 has
>> something wrong in its mediaprm parser from what I can get it to spit out.
>>  It also ships with a broken mediaprm file, and an error in that file
>> causes it to skip all entries below it.
>
>I don't use Fedora, so I cannot comment one way or the other.  On a Ubuntu
>box, 'apt-get install fdutils' gets me a working copy.
>
>>> Check the media descriptor in /etc/mediaprm.  On my box, COCO360 and
>>> COCO720 used an extraneous 'zero-based' flag.  This is certainly
>>> incorrect for 5.25" diskettes (haven't tried the 3.5 yet).
>>
>> That is another problem, that is not incorrect, but the parser thinks it
>> is,, the coco (1793 family fdc's) are in fact a base zero sector numbering
>> system, one of the very few, and only by beating a 765 based fdc about the
>> brow forcefully, can you get it to accept 'track zero sector zero'
>> numbering for write functions.
>
>That comment puzzles me a bit.  In all my experience with FDC chips (and I
>had considerable exposure to this back in the day), the sector (and track)
>ids were purely a software notion.  I wrote code to build a complete image
>of the track in memory (complete with gap bytes and magic values per spec
>sheet) and used a chip intrinsic (write-track) to blast that on to the
>media.  The software then modifed the track id, stepped the heads (or
>switched sides) and proceeded.

Which is as I understand it, exactly how the coco os9's format command works, 
and it fails if there is insufficient memory to hold this track image.

>I have heard that the NEC 765 chips have 
>problems with SD diskettes, but that's the only incompatibility I've
>personally encountered besides the start-of-track timing already mentioned
>in this thread.
>
>Are you claiming that the hardware itself has a fixation for zero origined
>sector numbers?  I just don't recall ever running into that.

Some 'pc' fdc's appear to act like it.  The 'SuperIO' chipsets seem more 
amenable to the concept.

>Also, how would you explain the fact that 360k DSDD diskettes generated
>with 'dd' read correctly on the CoCo and appear uncorrupted?  This doesn't
>seem to support a zero-based sectoring scheme.  But, I have not
>specifically dug into it and perhaps you are correct.  Stranger things
>have happened.

And will continue to happen.  I could be full of it, but thats what I can 
recall from some experimentation's with the original disk basic, which is 
pretty well documented.  15+ years ago though.

>> The repos have been cleaned out, this is an F8 system.  EOL for F8 support
>> was a week ago. :(
>
>~
>If it were me, I would look for a new distro.  I've had my fill of RPM
>based package systems.  What a PITA.  After SuSE jumped into bed with the
>evil empire of Redmond a few years ago I switched to Ubuntu and never
>looked back.

2 of my 3 boxes here are ubuntu now, and I'm beginning to get the feel for 
8.04 LTS on a spare box, so I expect that will be the next step.  I don't 
think that will stop me from using a bleeding kernel like I do now, presently 
2.6.28 final, locally built of course.

>> It may be on a dvd, I haven't looked.  Busy putting out other fires, like
>> verizon's dns servers became inaccessible to anything but a ping about 7Pm
>> & I've been battling with the typical morons in tech support on a Sunday
>> since. They are incapable of groking that my OS of choice has absolutely
>> zip to do with their failed dns setup. Finally a friend called and he got
>> me setup to use the opendns servers.  And they are 10x faster than vz's
>> have ever been. I won't convert back.
>
>Been there.  Have the T-Shirt!

Does it fit? :)

>I had to do the same thing with Comcast. 
>You must keep in mind that their technical support people are measured on
>the number of issues they close rather than how many folks actually
>receive help with problems.  Mentioning Linux gives them all the excuse
>they need to blow you off and dismiss the case.
>
>Fortunately we are one of the 2% of cities in the US with municipal
>fiber-to-the-premises.  I now get IP TV + 5Mbs symmetrical internet
>service for about 2/3 the price of Comcast, so they and their support
>morons are a fading memory.  Burlington Telecom, while not "Linux
>friendly" will at least listen to polite reasoning and take action.

So will vz, once you escalate it about 3 to 4 levels.  But that is a genuine, 
full priced, Pain In The Ass.

>>> Adjust names to suit your system.  The important issue to always execute
>>> setfdprm with the target disk in the drive.
>>
>> Yes, cuz all those config changes are nulled when the drive says the disk
>> is gone.
>
>We've got to get you a working copy of fdutils.  Have you tried building
>from source?

Once, that made 5.4 work back about Fedora 2, which has been moldering in its 
grave for about 4 years now.

More recently, I've been using minicom over a 232 cable to the deluxe 232 
pack, and it has rzsz support, so for smaller transfers I've been doing that.  
But I may have had an emp blow something cuz it hasn't worked recently, 
rather like minicom lost its configuration or ??? That, and other things 
getting in the way like bios problems on this @#*&^$% ASUS motherboard I paid 
$285 for. ASUS used to be considered the gold std, but not anymore.  Waaay 
down the list I keep.

>Steve

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Fortune's current rates:

	Answers				.10
	Long answers			.25
	Answers requiring thought	.50
	Correct answers			$1.00

	Dumb looks are still free.