Geneve OS development discussion

[+9640News]

OK, I think I have the suggestion box worked out.  There was a setting that did not allow users not signed in to post.  Now, any suggestions entered will be reviewed prior to posting for the public.

 

Go the URL at MDOS Ideas – 9640 News

 

 

[+Schmitzi]

 

Is it possible to make it work that "CD.." (without space) will work like "CD .."  (the standard command, with space)

 

 

 

[+Schmitzi]

 

 

Is it possible to let CHKDSK show the different RAM-types ?

So that one can see how much base RAM, SRAM, PFM-FlashRAM a.s.o is mounted ?

thx

 

[+InsaneMultitasker]

The period is used as a delimiter with the TI and Geneve, so people should be typing the expected space    There might be a way to make it work.  Did you add these two items to Beery's site?  Post 151.

 

[mizapf]

38 minutes ago, Schmitzi said:

Is it possible to make it work that "CD.." (without space) will work like "CD .."  (the standard command, with space)

I remember my father always wanted to type it this way, but I never got the sense behind leaving away just that one space.

 

Feature request: "alias"

 

alias cd.. = cd ..

[mizapf]

4 hours ago, 9640News said:

OK, I think I have the suggestion box worked out.  There was a setting that did not allow users not signed in to post.  Now, any suggestions entered will be reviewed prior to posting for the public.

 

Go the URL at MDOS Ideas – 9640 News

 

But I still need a login, right? What about a CAPTCHA?

[+9640News]

1 minute ago, mizapf said:

 

But I still need a login, right? What about a CAPTCHA?

It takes your name/email during the submission and creates an account.  All post to any topics are moderated, and since it is not really a forum, anything replied to will be carefully moderated.  At least that is my hope unless I overlooked something.

[+9640News]

8 hours ago, Schmitzi said:

 

 

Is it possible to let CHKDSK show the different RAM-types ?

So that one can see how much base RAM, SRAM, PFM-FlashRAM a.s.o is mounted ?

thx

 

The new MDOS command MEM shows how much slow/fast memory is reported.  @InsaneMultitasker has written a separate program for reading the devices in the PEBox and reporting memory for the HRD.  PFM memory, no tool to actually report the details I am aware at the moment.  Post something on www.9640News.com for a request so we can track things.

[+Schmitzi]

3 hours ago, 9640News said:

The new MDOS command MEM shows how much slow/fast memory is reported.  @InsaneMultitasker has written a separate program for reading the devices in the PEBox and reporting memory for the HRD.  PFM memory, no tool to actually report the details I am aware at the moment.  Post something on www.9640News.com for a request so we can track things.

 

yes, I have entered my 4 requests now, to have them all there

 

1. VERSION without CLS (already solved IIRC)

2. DIR should not show subdirectories on errative command input or empty list

3. CD.. for CD .. (no space alias)

4. CHKDSK to show detailed RAM infos

 

 

 

[mizapf]

As for the alias, I thought of a command "alias" to define such aliases as needed, not only for cd; this is a feature of Unix shells.

[+InsaneMultitasker]

On 4/2/2021 at 11:19 AM, Schmitzi said:

 

 

Is it possible to let CHKDSK show the different RAM-types ?

So that one can see how much base RAM, SRAM, PFM-FlashRAM a.s.o is mounted ?

thx

 

Maybe this will help.  Loading GenCFG with a parameter now displays current devices then returns to the command prompt.  If you really want to see what's in your box, in a batch file, etc, you can run this program.  (Edit: I just noticed that yesterday was a "countdown" day... at least in the MM-DD-YY data configuration.. and here we are w/7.30 about to "launch"    Thanks, Major Tom. )

 

[+9640News]

Geneve MDOS V7.30 has been released.  See message #1 in this topic:

 

MDOS V7.30 - TI-99/4A Computers - AtariAge Forums

[+InsaneMultitasker]

Noting a few reference items for future development:

1) Pending reports of successful PFM software updates, add power-up routines for SCSI, TIPI, IDE.  This may also pave the way to allow the TIPI and other cards to be placed at different CRU addresses for more optimal /4a mode usage. 

2) Validate and/or add interrupt handlers for SCSI (/4a mode) and IDE (/4a and 9640 mode)

3) Consider adding IDEx device to /9640 mode in addition to the mappings to SCSx; /4a mode of the additional device names is not yet possible due to limited DSR space.

4) RS232 mode is reset by the powerups; review whether or not the settings can (and should) be persistent.

[+dhe]

I was paging through the Boot EPROM source, TAT did a heck of a job of 'reconciling' Original Code, Tony Code, Jim Code, etc to update everything.

 

One of the comments was - a horizon can act just like a Hard Drive - with large horizon, I think that is the way most people would like to go.

 

In an other thread, I think it was Beery who outlined the different ways to format different devices. Maybe it's time to pull format out of Geneve OS and make it a stand alone program - while that wouldn't by itself make it a universal formater, it might give room in the persistent parts of Geneve OS?

Edited Wednesday at 10:50 PM by dhe
Correct Attribute To Source Code Reconcilliation.

[+InsaneMultitasker]

Adding relevant development notes from 6.70 and 7.00 testing for future documentation.  (some items overlap prior release notes).

 

@9640News and @mizapf  - should we remove MDOS 6.70RC1, 6.70RC2, and any stray test files/images of 7.00 from WHT and other repositories?  These versions propagated the SCSI Cache bug that corrupts devices; the issue has since been fixed.   I recommend that we retain the release disks for versions 6.50, 5.00, 2.21 etc. for historical reasons and we may want to pull forward some of the old files, if relevant.  Documentation of the current feature set is still an outstanding task for future completion.

 

Updates:
V7.00   SCSI Cache error corrected; cache was not invalidated properly
V7.00   TIPI Disk drive remaps added
V7.00   TIPI support for TIP1, URIx, PI, and DSKx added
V7.00   Remap table boundary issue resolved
V7.00   TIMODE2 command
V7.00   Corrected hard-coded references to SCSI buffer page
V7.00   IDE Low level support cleanup and preparation for multiple devices via direct sector IO/TI environment emulation
V7.00    TIPI Passthrough tested; deprecated in favor of adding support directly to the OS
V7.00    SCSI Interrupt/9640 mode - interrupt handler added to capture stray interrupts and SCSI2SD startup.
V7.00    SCSI Powerup updated
V7.00   Added IDE device support; tested by building MDOS on IDE device.
V7.00   Added Horizon Ramdisk hard drive support and updated boot stub; eliminated old ROS 6.00 code
V7.00    DSK REMAPs for Horizon ramdisk changed to 16-bit devices 1400/1600
V7.00    Ramdisk IO speed improvements via unrolled loop method and MOV instructions
V7.00    Removed RAVE ramdisk routines
V7.00   Removed WDS/Winchester Personality code; retained powerup stubs for old software
V7.00    Renamed HARD command to HFDC        
v7.00    PFM Restart code in L6\KEY2S
6.7 RC2    SCSI card PDMA Detection updated; default PDMA on for modified SCSI cards (detection routine validated)
6.7 RC2    TYPE Command updated; scrollback, incorrect screen display when record size>screen width, More command, buffer cleanup
6.7 RC2    Fix SCSI Error Passing; incorrect shift    
6.7 RC2    SCSI Read Caching implemented and enabled (later identified as cause of corruption)
6.7 RC2    SCSI Directory Rename bug corrected        
6.7 RC1    SCSI/DISK device conflict resolved
6.7 RC1    SCSI EOF failure fixed
6.7 RC1    HFDC Disk relative record bug to fix record IO problem w/>1 AU/sector (i.e., quad/high density)
6.7 RC1    SCSI Read/Write optimization; achieved 2x speed increase
6.7 RC1    PDMA Override implemented        
6.7 RC1    CYA Step rate Override implemented
6.7 RC1    SCSI transfer loop optimizations     
6.7 RC1    HOME/END added to key scan  (END=8A, HOME=8C)
6.7 RC1    HOME added to CLI command line during data entry

 

Potential Bugs:
Issue   Hard drive cluster allocation bug; reproducable by interleaving two files.  create test scenario            
Issue    CLI Command Stack error while editing    
Issue    HFDC .dsk vol access not functional

 

[+InsaneMultitasker]

Anyone out there familiar with SCSI low level access?   In MDOS 2.50-6.50 (maybe even 6.70) the SCSI low level routines poll the device for ready status.  The snip of code shown below has no way to exit when a fault condition is encountered, i.e., R8=error=endless loop.  Tonight I was trying to figure out why my SCSI card was suspending the OS for 30+ seconds if no media is found in the SCSI drive e.g., EZ135 platter.  After some RS232 tracing and examining the log in Excel, I noticed a pattern of 1000 iterations.  That led me to code I introduced and subsequently forgot - circa 2009? - in an attempt to recover gracefully instead of forcing the user to power down the Geneve.

 

I've been looking through my old SCSI reference books but I can't determine how many times the device should be polled before falling through to an error condition.  What I don't want to do is set the threshold too low or induce data loss if there is some minimum requirement here.  I am probably being too conservative anyway, so for now I have lowered the polling count from 1000 to 200, or roughly 6 second delay.  I think the SCSI peripheral is "smart" enough to also know that there is no media as subsequent accesses do not poll the device.

 

I'm thinking there might be some error flag I can isolate instead of testing the device indefinitely (or for xxx iterations)?  Suggestions? 

 

PAUSER11  DATA 0

PAUSE
 MOV     R11,@PAUSER11
L49
 BL      @TEST

 MOV     R8,R4   <-- endless polling loop and hang if in error state
 JNE     L49
 MOV     @PAUSER11,R11
 RT

 

(Also, it appears the low level routines incorrectly access device ID #0 before the target device under certain conditions, which I think explains why I see lockups attempting to access SCSI ID#5 without media in ID#0)

[+InsaneMultitasker]

Past my bed time but I at least found the "parenthetical problem" and it dates back to the original code from ~1994. 

 

The SCSI start device subroutine contains this code snippet: 

 MOVB   R1,@SCUNIT

 CLR    @SCFLAG

 

The flag is defined as: 

{{EVEN boundary here}}

SCUNIT  BYTE 0           *UNIT NUMBER (SET BY C HOST)
SCFLAG  BYTE 0           *READ/WRITE FLAG (SET BY C HOST, 0 = WRITE )

 

Whenever the device startup routine was called, the SCSI ID/unit# was being RESET to 0 by "CLR", a 16-bit operation, and passed incorrectly to the low level routine.

 

Changing the CLR to a MOVB operation resolved the problem and now the correct SCSI ID is started.  SCSI ID#0 can finally get some 'rest'.

 

SCSI ID #2 Before fix: 

DSEL: 0002
Start#0201
SCSICP; UNIT: 0000  <-- bug-- SCSI ID#0 requested
SCSICP; UNIT: 0201

 

SCSI ID #2 After fix:

DSEL: 0002
Start#0201
SCSICP; UNIT: 0200  <- SCSI ID#2

SCSICP; UNIT: 0201

 

[+Schmitzi]

wow   🛠️ 👀

[+Schmitzi]

 

I will enter another request, hope it´s easy

 

5. Pressing <F3> at the command prompt should bring up the last command entered by user (like <arrow-up>)

    This request is due to have similarity to MS-DOS, because I press F3 on the Geneve about 20x per hour

 

PS:  Sorry I cannot enter it at @9640News.com because my browser did not save the password....

       (Can you reset it ?)

 

 

[+FarmerPotato]

1 hour ago, Schmitzi said:

 

I will enter another request, hope it´s easy

 

5. Pressing <F3> at the command prompt should bring up the last command entered by user (like <arrow-up>)

    This request is due to have similarity to MS-DOS, because I press F3 on the Geneve about 20x per hour

 

PS:  Sorry I cannot enter it at @9640News.com because my browser did not save the password....

       (Can you reset it ?)

 

 

I expect F3 to mean Erase. 

[+9640News]

46 minutes ago, FarmerPotato said:

I expect F3 to mean Erase. 

F3 will remain as Erase.

[+Schmitzi]

5 hours ago, 9640News said:

F3 will remain as Erase.

 

ah OK, I did not know 

[+retroclouds]

4 hours ago, 9640News said:

F3 will remain as Erase.

 

Why not make this configurable? I know it’s not easy as it sounds.

 

I have been thinking a lot about keyboard mapping lately. Especially because of my Stevie editor. And there is simply no option that fits all. You have the real keyboard, you have emulators with a PC keyboard (where for example certain keys are already “occupied” by the PC OS (or browser). I didn’t come to a proper solution for a keyboard mapping that fits all and in the end, I think the only way out is for the user to reconfigure the mapping as wanted/required. 

[+Schmitzi]

7 minutes ago, retroclouds said:

 

Why not make this configurable? I know it’s not easy as it sounds.

 

I have been thinking a lot about keyboard mapping lately. Especially because of my Stevie editor. And there is simply no option that fits all. You have the real keyboard, you have emulators with a PC keyboard (where for example certain keys are already “occupied” by the PC OS (or browser). I didn’t come to a proper solution for a keyboard mapping that fits all and in the end, I think the only way out is for the user to reconfigure the mapping as wanted/required. 

 

haaa, I was just typing this here when the notification about your post came in:   😃

but here it is:

 

(@Beery) Do you think there is a possibility to "patch" the keyboard ?

We had this littly tool in early MS-DOS, to swap the "," with the "." on the number-block on the right side of the keyboard, i.e. for faster input of bills. And later software just accepted the "," as "." as a workaround.

 

 

 

[+Schmitzi]

 

OK, I have entered it as follows to https://www.9640news.com/mdos-ideas/ :

 

Request: KEYDEF.COM

 

" The request here is to have a small tool (or command?) were you can assign keys into another way as usual. The origin background for this request was that the user (me) wanted to change the function of the "F3"-key into the MS-DOS-like "ReDo"-function on the commandline (like the "arrow-up"-key on MS-DOS, what is an actual funtion on the Geneve too. This request is for very MS-DOS-affine users. And as there are maybe more keys that are waiting to be changed for some reason, why not having a small tool to do that on all (possible) keys ? For me, this request has a low priority, but I like the idea very much. "

 

 

@9640News thanks a lot for the quick pwd-reset, it´s saved now