Geneve OS development discussion

[RickyDean]

Got one, haven't used it any on a Geneve as of yet, got to get it going first. Been working 10 hour days with a 2 hour each way drive, not much free time.

[+Ksarul]

19 hours ago, dhe said:

How many folks on here are using a scsi2sd drive on their geneve?

I'm working up to it and trying to create a howto to try to make it easier for others.

I've got one, but I still need to connect it. . .one of my backburner hobby projects.

[+dhe]

Tim, is of course helping me on this little project of mine.

So far, I think we might need three sections on drive setup - scsi2sd 5, 5.5 and 6 - each seems to have a bit different options.

 

[rant] - like everything I touch it ends up being a lot harder then I'd have thunk.

case 1: goteks - existing cabling all wrong, pbox holes in wrong places.

   sunk about 18 hours in to figuring out custom cables, making new holes, figuring out how to copy files.

case 2: scsi2sd 5.5 - tried to communicate on usb, couldn't with cable provider by vendor - samplerzone.com

              tried another cable I had on hand - nada...

            finally ordered another new cable from amazon and I'm connected!

case 3: ide controller - about 10 hour sink in getting this going, and that's with a lot of help from Shift838.

 

So I've got a couple of hours in to hooking up the scsi2sd - and the only thing accomplished is hooked up USB cable.  ?

 

What I have learned so far from Tim, is 7.3 has special goodness for interrupts to handle scsi2sd - that wasn't needed for traditional drives.

mys and power will tell you the id your controller is set for and scsmap will tell you what id's you other drives are set to.

 

In the case of scsi2sd 5.5 - it's a 25 pin connector and the terminator is always on, so it will need to go at the end of the chain.

[+InsaneMultitasker]

Screenshots from the PC are easy    POWER and MYS successfully compile with TIC 1.67 and I was able to assemble the code with @mizapf's updated TASM v3.41.   Took me a bit of time to realize that the allegedly common code between the two programs was located in two different files. Silly me.  I've since consolidated and cleaned up the dependencies, now both programs leverage the same routines.  I'm a little wary of MYS though, because it uses a simple gets() input routine to capture format information.  And until a few minutes ago the format confirmation proceeded unless you typed 'N' instead of requiring positive confirmation.  Here's a screenshot of POWER in action.

 

 

[+InsaneMultitasker]

The OS Dev thread is a nice catch-all but we probably need a separate topic for Geneve program and utility updates.... Regardless, I put a few things together for release thanks in part to the "extra" hour from the time change this morning  

 

(1) BLOB - Geneve command line unarchiver/decompressor, may be used standalone or invoked by another program such as Directory Manager.

(2) TIAV2 - Geneve command line viewer for Myart, TI-Artist, and Instance files. Meant to be invoked by Directory Manager but works as a standalone program. I used it as a splashscreen program in my autoexec batch file for a while.

(3) Directory Manager 2.65t - invoke BLOB with 'Y', invoke TIAV2 with 'S'.  Updated help text. Added output path option to decompressor selection. Added some readme text.

 

I have two folders on my devices that contain programs like BLOB, TIAV2, EDIT, QDE, FED, ASM, LINK, and others.   The two folders are part of my "PATH" so that I can invoke the contents from "anywhere" -and- so that other programs can launch/invoke them.  I strongly recommend that if you have the ability to do so, locate your common utilities and programs in a folder in the search path. 

BLOB 1.1 Command Line Unarchiver.zip Directory Manager 2.65t.zip TIAV2 2.0 Viewer.zip

[+dhe]

I was looking at the new MDOS documentation that Tim and Beery released.

Technically, not a MDOS/GeneveOS related question, but are there certain CRU addresses that either the Boot EPROM or MDOS expects particular cards to be located at?

[+9640News]

Look at the SCSMAP and REMAP notes in the documentation and that will identify the supported CRU's for hardware selectable CRU devices.

 

Beery

[+dhe]

Right..

I see mention of

>1000   - RAMDISK

>1100   - Myarc HFDC   --- Can you have a floppy and an HFDC?

>1200   - SCSI

>1400,  >1600 and >1700 for RAMDISK

>1800   - TIPI

>1900   - IDE

 

This is scatter though out the DOCS, it might be helpful to have this consolidated at the beginning of the docs, in one place.  Is the question of, are these addresses fixed in MDOS (and/or bootrom) or can they be moved around?

 

Also left to the imagination, can you have two RS232's?

 

I'm trying to think of the person that just spent $900 dollars on a Geneve, where would they start?

 

Michael / Tim - do either of you know the answer(s)?

 

 

 

[+9640News]

47 minutes ago, dhe said:

Right..

I see mention of

>1000   - RAMDISK

>1100   - Myarc HFDC   --- Can you have a floppy and an HFDC?

>1200   - SCSI

>1400,  >1600 and >1700 for RAMDISK

>1800   - TIPI

>1900   - IDE

 

This is scatter though out the DOCS, it might be helpful to have this consolidated at the beginning of the docs, in one place.  Is the question of, are these addresses fixed in MDOS (and/or bootrom) or can they be moved around?

 

Also left to the imagination, can you have two RS232's?

 

I'm trying to think of the person that just spent $900 dollars on a Geneve, where would they start?

 

Michael / Tim - do either of you know the answer(s)?

 

Yes, you can have a HFDC and a separate floppy controller.  Generally, you would want the HFDC to be at >1000 so it is the bootable device.  I think the DSR will permit above >1100, but it has probably been ages since anyone has had it at anything other than >1000 or >1100.

 

CRU's are fixed at present. CRU's >1300 (RS232 1/2) and >1500 (RS232 3/4).  I do not know of any present reason to have two RS232 cards in a system unless someone is trying to do some debug code on a running program.

 

Two RS232's should be possible if the second has been modified.

 

The person that spent $900 on a Geneve hopefully knows what a Geneve is, and has been to Atariage and followed past and/or present developments.  They will have to start somewhere to acquire software to boot the Geneve unless they have a bootable system when it arrives at their doorstep.  If they are not aware of this website and/or do not find links to this website to ask questions, then they surely are not going to find much to do with their new system and any tweaks to docs will never find them anyways.

 

 

 

 

[+9640News]

59 minutes ago, dhe said:

 

This is scatter though out the DOCS, it might be helpful to have this consolidated at the beginning of the docs, in one place.  Is the question of, are these addresses fixed in MDOS (and/or bootrom) or can they be moved around?

 

Dan,

 

You are more than free to update the present docs and are encouraged to do so.  It would not hurt to have a fresh set of eyes providing updates to the docs!!!

[+InsaneMultitasker]

Addresses are fixed in certain cases for code simplicity and because "that's how it was for years" - oh, and for the native OS, CRU order doesn't matter.  The Geneve OS doesn't scan each card for every DSR call, it simply uses the routine corresponding to the device name and mappings. This is true in both native mode for all operations and GPL mode for level 3 operations.  The only times CRU order come into play are at startup (boot EPROM), the DSR powerup (to determine certain mappings), and ROMPAGE operations.

 

If you intend to boot from the RAMdisk, that card should live at CRU 0x1400.  This has to do more with the EPROM and OS dependencies.  The Geneve can still run older OS revisions though with recent enhancements and bug fixes, there aren't many "good" reasons to do so.  The TIMODE2 command uses 64K less RAM than TIMODE, so with MDOS 7.30 you actually have more free memory (when TIMODE is invoked) than with earlier OS iterations.  Native OS mode has less memory available but for most users, this won't be a noticeable concern.

 

One remaining "requirement" that I'd like to eliminate is the need for the extra 32K to run OS versions 2.50 or higher. The limitation comes from the SCSI buffer mechanism (and anything tied into it, like the Horizon and IDE) being hard-coded to SRAM.  If the extra 32K is not present, the OS should reserve a page in DRAM instead.  Changing the SRAM dependency isn't high on my list but is certainly kept in mind as other work is completed.

 

[+dhe]

>1000   - RAMDISK
>1100   - Myarc HFDC 
>1200   - SCSI
>1300   - (RS232 1/2)
>1400   - RAMDISK - Bootable
>1500   - (RS232 3/4) - but card needs modified.
>1600   - RAMDISK
>1700   - RAMDISK
>1800   - TIPI
>1900   - IDE

or

(If HFDC & FDC)

>1000   - Myarc HFDC
>1100   - FDC

 

Edited November 12, 2021 by dhe
need to update >1500

[+dhe]

Just a stray thought, does the PFM or Memex have a CRU address that you need to be respectful of?

[+9640News]

10 hours ago, dhe said:

Just a stray thought, does the PFM or Memex have a CRU address that you need to be respectful of?

No

 

 

[+dhe]

It's been a little while.

chkdsk was suppose to be like chkdsk for ms-dos.

 

What does chkdsk actually check for?

Has it been tested of IDE and TIPI also?

 

[+InsaneMultitasker]

My personal opinion is the CHKDSK is fairly worthless, more so now that MEM reports the available memory. And a DIRectory provides more TI/Geneve relevant information anyway.  The /F option is only valid for floppy disk structures, so no IDE, SCSI,TIPI, HFDC, RamHD support now or in the future.   

[+dhe]

Is it possible to hook up a serial console and run cli software?

[+InsaneMultitasker]

1 hour ago, dhe said:

Is it possible to hook up a serial console and run cli software?

Console output and console-based program output was tested during MDOS 7.0-7.30 development.  Remotely controlling the console's keystrokes was problematic, in part due to the RS232 circular buffer and system interrupt handler. Another problem was the use of direct video in many of the native programs, which left display gaps or worse - no output to the console after such a program loaded.  I didn't pursue this any further as there were many other bigger fish to fry; still, it might be nice to revisit the code some day.

[+dhe]

How does one get to the point of testing (playing) with that?

[+9640News]

11 hours ago, dhe said:

How does one get to the point of testing (playing) with that?

The easiest..... said with a sly grin???? Have someone with the time to write the code.  What's your free time like Dan??

 

As Tim mentioned, anything with direct video writes is not going to work.  I did have my ANSI driver writing out to the RS232 terminal to another computer displaying ANSI graphics as they were also being displayed on the Geneve video output while attached to a BBS via the TIPI through a telnet connection.  Keyboard input is going to be problematic for any program as the attached RS232 terminal would not know what keyboard mode the Geneve would be using.  Not saying it could not be overcome, but it would need a remote keyboard interpreter to patch things in.  Easier said than done.

 

 

 

 

[+dhe]

Trying to boot from SCS6 with MDOS 7.3.

 

I have a copy of System/sys - MDOS 7.3.

 

Can someone point me to the paired scsi/sys?

 

At one time I think I might have had to call scsi/sys >  scs6/sys?

 

 

 

PS.. I did pull down all the files here: https://www.9640news.com/mdos/mdos-7-30-released/ - scsi/sys or scsi-sys not included.

I also peeked over here: https://www.9640news.com/device-loader/

and here: https://github.com/BeeryMiller/LOAD-SYS#readme

 

 

[+dhe]

Because at my current boot stage and lacking tools I copied "SYSTEM/SYS" from B to A.

A: didn't have enough room. Copy finished with out error.

This left the system in a state where you would get the GeneveOS 7.3 Pop-up screen then the system would lock up.

  Ctrl-C would get terminate batch job, but nothing would work.

 

If that failure mode isn't just me, we might want to do a check before copy (if possible) to make sure there is enough room on the target device.

 

I was able to get everything built on the PC with ti99dir - using a usb drive in the GOTEK. I now just need to find scsi-sys!

[+InsaneMultitasker]

3 hours ago, dhe said:

If that failure mode isn't just me, we might want to do a check before copy (if possible) to make sure there is enough room on the target device.

Similar to the file size question, we'd have to have a way to reproduce your conditions and test with v7.3.  I'm assuming you copied using the command line w/MDOS 6.50?  There have been quite a few fixes since then though it doesn't mean an old bug isn't lurking.  A good practice is to run the CRCOS tool after any copy (and on the source if necessary) to confirm the file is valid.

 

I thought the -SYS files were part of the EPROM update release.  SCSI/SYS 2.5 should be in the forum somewhere from when I released [Edit: I could not find it in the forum; some of the messages about it are from 2019, which is when the now-resolved SCSI corruption issue bit me during IDE/SCS testing, so maybe it never made it into the wild, however, Tony's EPROM came with a working scsi/sys, disk/sys, etc back in the day] but I don't recall any of the load routines having the ability to select the ID as they would default to SCS1 (id 0).  Do Tony's boot eprom docs specify anything about these options?  I feel we are missing something important.

[+9640News]

2 hours ago, dhe said:

Because at my current boot stage and lacking tools I copied "SYSTEM/SYS" from B to A.

A: didn't have enough room. Copy finished with out error.

This left the system in a state where you would get the GeneveOS 7.3 Pop-up screen then the system would lock up.

  Ctrl-C would get terminate batch job, but nothing would work.

 

If that failure mode isn't just me, we might want to do a check before copy (if possible) to make sure there is enough room on the target device.

 

I was able to get everything built on the PC with ti99dir - using a usb drive in the GOTEK. I now just need to find scsi-sys!

Dan,

 

The loader files for SCSI are on 9640news.com at Geneve V2.00 BIOS Update – 9640 News with the Eprom 2.0 updates.

 

As Tim mentioned, it is going to use SCS1 (id 0) unless you had a special loader.  If you are using Tony's eprom, you will need to rename LOAD-SCS to whatever name Tony was using with his Eprom.  For all practical purposes, they should be identical.  Keep in mind the SCSMAP command to map SCSI devices around is not active until after MDOS is loaded.

 

Beery

 

 

 

[+dhe]

I need to do a bit more investigation but "SCSI/SYS" and "SCS6/SYS" have the differences you would expect.

 

The label on the EPROM has a really nice label, is it was NOT a one off Dan custom. If I made the EPROM, it probably would have just had a floppy read/write protection tab over the erase window. It makes sense this would have existed, because in the time frame that Tony released 1.0 - 2002, many people where using Iomega Zip Drives which are locked into SCSI ID 5 and 6. I just need to take a photo of the boot sequence.

 

I also am going to order an EPROM from Bob to simplify life going forward.

 

dhe@DESKTOP-TTA01OD:/mnt/d/ti/geneve/scsisys$ strings scs6.sys | grep SCS
SCS6/SYS
No FDR table found&Cannot locate SYSTEM/SYS file on SCS6.
Cannot locate SCSI controller
             SCSI/SYS
  Loading SYSTEM/SYS from SCS6 ...

dhe@DESKTOP-TTA01OD:/mnt/d/ti/geneve/scsisys$ strings scsi.sys | grep SCS
SCSI/SYS
No FDR table found&Cannot locate SYSTEM/SYS file on SCS1.
Cannot locate SCSI controller
             SCSI/SYS
  Loading SYSTEM/SYS from SCSI ...

 

dhe@DESKTOP-TTA01OD:/mnt/d/ti/geneve/scsisys$