Print

[AlexTheCat123]

Unless ESProFile is somehow getting reset (which shouldn't be possible unless the firmware is crashing and restarting itself), that error message can't be coming from the Selector because LOS will still be the selected disk image even after the reboot. So I guess the bootloader might display messages in LOS 1 after all. Interesting!

Let's try connecting to ESProFile over serial next; getting that debug data is probably the best next step! And if that doesn't reveal much, then it might be time to break out the logic analyzer...

[ried]

Here's what it says over serial when Error 85 is generated:

Code: [Select]

Read  Block: 000000 - Read phase 2: Host didn't respond with a 55!

Edit: Okay, this is weird. The ESProFile is still being powered by my laptop's USB port (serial communication still open) and I powered up the Lisa again and it worked, booting straight into LOS 1.2 as expected. Huh.

After further attempts while powering the ESProFile from my laptop's USB port (serial communication open or closed) and from the wall outlet USB port... sometimes it works, sometimes it doesn't - power source and serial comm status don't seem to affect the issue. After plugging in the ESProFile, the first boot into Selector and then LOS 1.2 always works fine. Subsequently powering down the Lisa, then powering it up again, results in Error 85 about 75% of the time, while 25% of the time it boots as normal.

So, it seems that the host (is that Lisa? or is that something in the ESProFile?) frequently doesn't "respond with a 55" in the expected manner or timeframe which breaks things. Note that this issue is unique to the ESProFile and I haven't encountered Error 85 with a 5MB ProFile, X/ProFile + voltage regulator in a ProFile case, or the ArcaneByte Cameo/Aphid emulator on the same machine using its built-in parallel port.

[stepleton]

Clearly I was wrong... is the error text coming from "The Selector"?

It weren't me, guv!

I couldn't remember whether that was an error message in the Selector, my Lisa I/O routines, or my ProFile bootloader (all of which are present here), so I grepped my entire project directory.

It didn't show up in any of my source code. But it did match in the binary data of some Office System 1 disk images:

$ strings lisa_disks/bitsavers/twiggy/t001_Office_System_I.dc42 | grep -i 'profile read failed'
NuPROFILE READ FAILED

(ignore the "Nu", that's just noise.)

I'm as swamped as ever, but I can see if I can find a time this weekend to try on my machine. Can you spell out a basic replication procedure? (sorry!)
My bet is that this is as likely to be an operating system bug as anything else.

[ried]

Can you spell out a basic replication procedure? (sorry!)

• Connect the ESProFile to the built-in parallel port, then connect a USB power source to it.
• Power on your Lisa.
• Boot into Selector, then select your LOS 1.2 image, then press "B" for boot. (This should work fine)
• After successfully booting LOS 1.2, press the Lisa's power button again to shut down.
• Then press the Lisa's power button to start up again. This usually generates Error 85, and choosing CONTINUE will result in PROFILE READ FAILED.

[AlexTheCat123]

After further attempts while powering the ESProFile from my laptop's USB port (serial communication open or closed) and from the wall outlet USB port... sometimes it works, sometimes it doesn't - power source and serial comm status don't seem to affect the issue. After plugging in the ESProFile, the first boot into Selector and then LOS 1.2 always works fine. Subsequently powering down the Lisa, then powering it up again, results in Error 85 about 75% of the time, while 25% of the time it boots as normal.

So, it seems that the host (is that Lisa? or is that something in the ESProFile?) frequently doesn't "respond with a 55" in the expected manner or timeframe which breaks things. Note that this issue is unique to the ESProFile and I haven't encountered Error 85 with a 5MB ProFile, X/ProFile + voltage regulator in a ProFile case, or the ArcaneByte Cameo/Aphid emulator on the same machine using its built-in parallel port.

Weird that it works sometimes but not others; I'm guessing it's a really subtle timing issue or something! The "not responding with 0x55" thing means that the host (Lisa) didn't send an 0x55 back to your ESProFile as part of the command confirmation phase of the handshake. The Lisa successfully sent the command bytes (as evidenced by ESProFile saying that it intends to read from block 0), at which point the Lisa is supposed to assert /CMD, followed by ESProFile sending the Lisa an 0x02 and asserting /BSY. When the Lisa sees this, it's supposed to send the 0x55. But something's going wrong here.

Could you send me your LOS 1 disk image? I'll try to fire it up on a Lisa 2/5 and see if I can reproduce things.

[jamesdenton]

@reid - What ROM version are you running? Wonder if there's a difference between how something like the D ROM operates vs F/H in this particular scenario.

[ried]

What ROM version are you running? Wonder if there's a difference between how something like the D ROM operates vs F/H in this particular scenario.

Now THAT is a very good question! This is an earlier Lisa 1 and it has C ROMs.

Alex, zipped image file attached.

[jamesdenton]

What ROM version are you running? Wonder if there's a difference between how something like the D ROM operates vs F/H in this particular scenario.

Now THAT is a very good question! This is an earlier Lisa 1 and it has C ROMs.

Ahh, now that's even better. A peek thru the ROM notes mentions some Profile-related changes between C->D->F->H.

5/12/83 - Create and release rev D of boot ROM
9/1/83 - Add retry for hard disk booting. (CHG016)
...
10/20/83 - Release as rev E for Lisa and Pepsi systems.
12/15/83 - 3)Extend timeout for hard disk ready. (CHG031)
...
1/25/84 - 1)Add code to properly initialize Profile-reset and parity-reset lines for Profile booting (CHG036)

Needless to say, whatever you're experiencing might've been addressed by the time H and 3A came along.

https://www.applefritter.com/content/lisa-boot-rom-modification-history

Edit: Pure speculation that might not amount to anything.

[ried]

Makes perfect sense! Can the ESProFile extend its compatibility (and behave the way other drive emulators do) to a Rev. C machine with a minor software update?

[stepleton]

• Connect the ESProFile to the built-in parallel port, then connect a USB power source to it.
• Power on your Lisa.
• Boot into Selector, then select your LOS 1.2 image, then press "B" for boot. (This should work fine)
• After successfully booting LOS 1.2, press the Lisa's power button again to shut down.
• Then press the Lisa's power button to start up again. This usually generates Error 85, and choosing CONTINUE will result in PROFILE READ FAILED.

I found a spare moment. All my attempts to replicate the problem on a Lisa 1 with D boot roms have failed. Everything worked fine.

[ried]

There are at least two or three people in the universe who could possibly use this fix, I’m sure 

[sigma7]

This is an earlier Lisa 1

Does the I/O Board have R41 intact? This is a pullup on the /BSY signal of the parallel port and its removal was directed for the Lisa 1 - 2 conversion.

I'm not aware of the symptoms that removing it was supposed to address, but perhaps you've discovered one.

R41 is above UF5

Similarly, C55 was added on Rev D of the schematic, and on the last revision of the boards, C55 is replaced with a pull-up resistor with an insulating sleeve. This is connected to the /Strobe signal of the parallel port VIA.

C55 is above UC5

If you have additional I/O Boards with different R41/C55 configurations, you could see if they make a difference with this issue. Perhaps disconnect the twiggy drives while fiddling with this.

[ried]

Thank you, sigma7. While R41 and R47 are both present, I do not see C55 on the board at all so perhaps this is an earlier layout?

[sigma7]

R41 and R47 are both present, I do not see C55 on the board at all so perhaps this is an earlier layout?

Yes, that's an earlier layout. In your picture, C55 is the capacitor tacked-on to pin 6 of UF4.

[ried]

Looks like we have found an intermittent issue that is likely dependent upon having Rev. C ROMs and perhaps also a specific I/O board revision. Given the tiny universe of people who have these and might want to use an ESProFile, this is probably an academic exercise that's going to fall pretty low on Alex's priority list 

Thank you for the insight, everyone.