U1MB FJC Alternative Bios woes

[ivop]

Hi,

 

Until now, I have used a U1MB/SIDE2 combination with all the stock firmware. Everything worked fine, but I thought it would be nice to try out Flashjazzcat's alternative BIOS. Flashing went all well, but then the trouble started

 

First, about half the time it wouldn't boot of SIDE2.

Secondly, I copied one file of about 9kB to the CF card and it totally screwed up my partition! It turned a directory named 'AS6' to a file '<heart><heart>AS6' and the file I copied was nowhere to be found!

Thirdly, the latest UFlash I downloaded did not recognize my SIDE2 cartridge being inserted. It crashes with all sorts of garbage on the screen.

Fourthly, reverting back to candle's PBI ROM, the boot problem mostly went away, but not 100%. Sometimes it still didn't boot of SIDE2.

Fifthly, so I reverted to all of candle's ROMs, used latest UFlash to flash them back. Reboot ---> U1MB is bricked!

 

Currently I am trying to unbrick it by flashing U1MBv2.rom to an 8Mbit AtariMax cart (twice, to fill 8Mbit) and swap the IC. Hope this'll work!

 

Anybody else suffering similar problems?

 

Edit: unbricking worked! So glad I got a MaxFlash cart recently. The PLCC to DIP adapter I have is not wide enough to fit a SIC! Cart

Edited June 25, 2017 by ivop

[flashjazzcat]

Firsty, the PBI ROM which drives your hard disk is largely unchanged between the original BIOS and the "alt" version, so if your partitions are screwed up, you have hardware issues. The new firmware is sufficiently mature now that if it was routinely blowing up, I think we'd know about it. The fact you're still having issues following a downgrade (despite the thread title) also suggests software is not to blame, although now that you've unbricked the board, you neglect to update us on whether things are working again.

 

A few further questions:

 

1) When you say "it would not boot off SIDE2", how exactly do you mean? Are you talking about booting mounted ATRs?

2) You say UFLASH would not recognize your SIDE2 cart being inserted. It sounds as if you're plugging the cart in mid-session. This will not work, and nor is it meant to.

 

It's not especially helpful to leave the open-ended implication that all problems went away when the original firmware was reinstated, at least without some evidence or further information which might help figure out why your particular setup was so problematic.

[ivop]

Sorry if I did not supply enough information. I did still experience issues after downgrading the PBI Bios only, but you explicitly state that Candle's PBI ROM should only be used with Candle's BIOS. After downgrading the full 4MBit, it works like before, which means no problems at all. Just copied several megabytes over Hi-speed SIO to my SIDE2 partition and back again. Diff'd them and they all are fine.

 

Answers to your further questions:

1) By booting of SIDE2, I mean U1MB's SDX 4.47 boots of D1:, which is an APT partition. It gets its autoexec.bat and config.sys from the CF card.

2) I did not hot-plug the cart. I boot from SIDE2, run uflash, either over SIO2PC or from the CF card itself (same behaviour). uflash did recognize U1MB straight away. Selecting SIDE2 made it crash. Selecting SIDE it did see the Flash ROM.

 

Now that I'm back at SDX 4.45 (which came with U1MBv2.rom from 2012), uflash does not recognize my hardware automatically. I can select U1MB manually though, and then it sees all the banks. SIDE2 still doesn't work, but does not crash anymore! It shows an empty dialog with a single OK button. After that it says No Device.

 

If it's not a software problem, what kind of hardware problem are you thinking of?

This is an 800XL with Sophia and a U1MB+SIDE2. Nothing else.

Thanks for your response!

[flashjazzcat]

...you explicitly state that Candle's PBI ROM should only be used with Candle's BIOS.

That's correct. I was referring to the code which drives the HDD hardware, reads and writes sectors, etc. This code wasn't subject to redesign. However, the main BIOS and PBI BIOS now share PBI RAM, and for this reason you should not mix and match versions.

 

After downgrading the full 4MBit, it works like before, which means no problems at all. Just copied several megabytes over Hi-speed SIO to my SIDE2 partition and back again. Diff'd them and they all are fine.

Again: IO code didn't change so I have no idea why you would get transmission errors with one revision and not the other.

 

2) I did not hot-plug the cart. I boot from SIDE2, run uflash, either over SIO2PC or from the CF card itself (same behaviour). uflash did recognize U1MB straight away. Selecting SIDE2 made it crash. Selecting SIDE it did see the Flash ROM.

Make sure everything is turned off on the U1MB (SDX, PBI, the lot). Can you tell me the version number of UFLASH (Help->About)?

 

If it's not a software problem, what kind of hardware problem are you thinking of?

This is an 800XL with Sophia and a U1MB+SIDE2. Nothing else.

Unless you flashed the single 64KB "firmware" update, I'm wondering if there was some serious mismatch of firmware components at work here. Otherwise, I'm not sure, which is why it would be nice to investigate further. However, if you're not equipped to recover easily from issues which result in a non-working machine, you may not be well positioned to perform further tests.

 

Honestly, in the two years since the alt firmware was released, I've pretty much seen it all, and - with a few non-crtitical exceptions - every single issue reported was traced somehow back to hardware. Reported issues include:

 

1. Garbage in the U1MB BIOS menu. This was usually caused by noise on the reset line, and fixed by a cap to GND. Probably was most common on - but not exclusive to - the 1200XL. Other causes of screen garbage were traced to problems elsewhere in the system (address lines, connectivity issues, etc). Sometimes the original firmware flattered to deceive (in as much as it happened not to rely on some memory location needed by the alt firmware, although usually the issues could be reproduced in the older firmware if one tried hard enough).

2. Immediate bricking of the machine. This happened when flashing the new firmware to an U1MB with the ancient v1 CPLD code which did not have RAM at $D1xx and $D5xx. I subsequently placed a warning in red letters on the download page stating that the board should be upgraded to the PBI JED before updating the ROMs.

3. SIDE loader not recognizing CF cards or errors during FAT reads. The new loader is different to the old one, but the approach to IO (using the IDE data buffer as a sector buffer) is pretty much the same. However, some setups don't like it and using a different CF card or replacing 74LS08 on the machine is a confirmed fix. Lotharek documents another fix on his website which is pertinent to Rapidus.

4. Stereo Pokey toggling appearing not to work. In every case, it turned out that the original firmware had fooled the user into thinking stereo toggling worked when in fact it had never worked in the first place. Since the new firmware will not allow toggling of stereo unless a second Pokey appears on the bus after stereo is turned on, the fault immediately became apparent. Either U-Switch was faulty, it was wired up wrongly, or there was some other hardware issue.

5. MegaSpeedy appearing to encounter IO errors using the high-speed SIO code built into the PBI BIOS. This turned out to be an NTSC bug in the MegaSpeedy firmware, but also exposed a small issue with the retry code in the PBI HSIO code (fall-back to default speed was not working).

6. Flashing issues. Cause: using old versions of UFLASH instead of those supplied with the firmware (the versions supplied with the firmware are used to update the four test machines on the desk here).

7. PBI BIOS hanging during initialisation on Rapidus machines when the 65C816 OS was used. This turned out to be a bug in the Rapidus firmware.

8. Inability to mount ATRs on Rapidus machines in 65C816 mode. Turned out to be an issue in the Rapidus firmware.

 

Confirmed bugs included some boot drive issues (long ago fixed, but perhaps pertinent to you depending on the firmware revision in use), some minor keyboard issues flagged up by TK2, a bug in the Z: handler. But I do not recall (unless I forget) a single firmware release containing critical bugs which could be blamed for bricking machines.

 

Because of this - while I welcome bug reports which provide an opportinity to fix something long overlooked (like the SIO speed fallback bug, for example) - experience leads me to suppose there are usually factors at work other than software. But the aim is to make things as near perfect as can be. However, if performing further diagnostics is liable to interrupt your use of your A8 and things work well using the original firmware, feel free to leave it be.

Edited June 25, 2017 by flashjazzcat

[ivop]

Thanks for the elaborate answer!

 

Make sure everything is turned off on the U1MB (SDX, PBI, the lot). Can you tell me the version number of UFLASH (Help->About)?

v.1.28

However, if you're not equipped to recover easily from issues which result in a non-working machine, you may not be well positioned to perform further tests.

...

However, if performing further diagnostics is liable to interrupt your use of your A8 and things work well using the original firmware, feel free to leave it be.

At this very moment, I leave it be, but I will return to this shortly. It is a bit of a pain to put the OS and MMU back in, flash AtariMax cartridge, swap PLCC's and reïnstall U1MB, but I'm willing to. Hmm, probably get another 800XL out of storage to do the flashing if I need to do this more often

 

[flashjazzcat]

v.1.28

OK - thanks. I'm on v.1.30 here but I'm not sure off-hand whether I put that version out in the wild yet. UFLASH is a right PITA to maintain owing to the plethora of hardware and banking schemes it has to support, but I'll get to run some tests when I flash the forthcoming loader update to a real SIDE2 here.

 

At this very moment, I leave it be, but I will return to this shortly. It is a bit of a pain to put the OS and MMU back in, flash AtariMax cartridge, swap PLCC's and reïnstall U1MB, but I'm willing to. Hmm, probably get another 800XL out of storage to do the flashing if I need to do this more often

Seriously, if a BIOS update bricks the machine at this stage in the game, steer clear of changing anything. I'm flashing firmware day in, day out with UFLASH here sometimes on a 600XL, 800XL, 130XE and 1200XL and I cannot recall the last time I had to remove the PLCC flash ROM from any of them. On the very rare occasions I have, it's invariably down to some flashing glitch or power blip which resulted in a corrupted ROM. Lots of bad flashes or a good flash resulting in a bricked machine with the new firmware is a red flag signifying something is seriously wrong elsewhere.

[ivop]

I'll keep that in mind.

 

Plan for next session:

 

1. Test Power Supply (under load)

2. Test RAM w/ SysCheck II

3. Flash full firmware.rom

4. Test test test

5. If fails, swap CF card, goto 4.

 

I'll keep you posted.

 

By swapping the 74LS08, do you mean by a faster version? (F, ACT?)

Cap to ground on the reset line, what value did you use? 100n or larger?

 

Regards,

Ivo

Edited June 25, 2017 by ivop

[flashjazzcat]

Please do keep me posted! Even if there's a software issue, I'd rather it was discovered and fixed than missed.

 

Lotharek says the "F" '08s are fine, but I have none of them here and I usually find a like-for-like swap from neighbouring machine rectifies the O2 timing issues.

 

Good luck!

[Roydea6]

My Uflash (Help->About)? reads uFlash v.2.00B

[flashjazzcat]

My Uflash (Help->About)? reads uFlash v.2.00B

It was a test version (beta, if you like, signified by "B") Roy. You can replace it with the version in last December's firmware pack.

[Roydea6]

OK I replaced it and it seems to be the same as the one I have.

UFLASH XEX 22813 9-12-16 2:38p

v1.28 does not seem to have a FIRMWARE slot.

Edited June 25, 2017 by rdea6

[flashjazzcat]

Must have released the beta with the last pack, then. Keep using that one.

[ivop]

Update:

 

Recently I have quite extensively "hacked" this same Atari 800XL. Added second SIO port in place of RF modulator, added simple stereo and u-switch, added 6.3mm stereo audio out and... just now I replaced the 74LS08 with a 74F08.

 

All problems seem to have vanished. I have copied over several megabytes of data between SIO2PC and SIDE2 and have not experienced any corruption. This was all with your alternative BIOS and PBI drivers.

 

As an added bonus, my MyIDE (1) has resumed working. It didn't work on any machine anymore, even machines it used to work on (like this one) and not even on Mr. Atari's own computer. We have added several post-production fixes to the cart, but the IDE part of it remained dead. And now it works again

Edited September 30, 2017 by ivop

[+mytek]

Although I have never experienced any serious issues with re-flashing the Alt BIOS, which I have done so many times I lost count of, whenever something weird did happen during the flashing it was attributable to not using the latest version of Uflash. So now I've made it standard practice to always use the latest greatest that comes packaged with the new BIOS version.

 

As for switching over to the 74F08, that is also a good practice to follow since it appears to take away the last little bit of timing fussyness in the hardware. It was discovered during beta testing of the 1088XEL, that this was a must have in order to obtain 100% reliability in all operational aspects.

 

Glad to see that you got everything working ivop, because if you had to stay with the original BIOS it would be truly sad, since FJC's is truly remarkable.

 

- Michael

[ivop]

Glad to see that you got everything working ivop, because if you had to stay with the original BIOS it would be truly sad, since FJC's is truly remarkable.

 

Yes, I'm really glad it works now Have to get used to the keys used to navigate the menus. I expected TAB to change tabs and left/right arrows to change values, but the arrows change tabs and return changes the values. But I'll probably get used to it after a few days.

 

Also, I noticed that the Motor Control line is active during the setup and even during the splash screen if no setup is entered. I wonder how cassette decks would react to that

 

And is there a way to set the duration of the splash screen? I like having it shown, but it takes a little too long for my taste. 1 second would be enough.

 

Anyway, not meant to criticize. Really remarkable piece of software!

[flashjazzcat]

Hit space to skip the splash screen. Any faster and it'll be too fleeting, IMO.

 

I'll look into the motor control line.

 

Regarding navigation: tab moves through sub-fields in time/date dialogs. Open to suggestions, although changes made now may prove disruptive.

[ivop]

Yeah, no need to change the navigation. Those used to the current key setup are going to be upset!

 

Didn't know about the space bar. That really helps! Maybe add a notice on the splash screen in a future revision? I guess it's in the manual, but nobody reads manuals

[flashjazzcat]

You can use ESC to clear it in upcoming revision too. Quite a few secret shortcuts too numerous to describe on screen.

[flashjazzcat]

Motor control line issue fixed: thanks to ivop for reminding me about that (ProWizard originally spotted it too).

[+Stephen]

You can use ESC to clear it in upcoming revision too. Quite a few secret shortcuts too numerous to describe on screen.

It's almost like you need to write a user manual for this thing

[+mytek]

And is there a way to set the duration of the splash screen? I like having it shown, but it takes a little too long for my taste. 1 second would be enough.

 

I wish my Linux box would boot in a couple of seconds . Or for that matter, my HDTV.

 

Michael