Print

[kgraaf]

Hey everyone,

Trying to get back into the Lisa world. I've noticed that the most recent version of LisaEm appears to be broken. For refererence, I'm building on Ubuntu 22.04 with WxWidgets 3.1.5 per Ray's (RIP) instructions. The build succeeds, but, any attempt to boot from a floppy (I've tried several from many different sources) results in a segfault. I tried master, unstable, and livedev branches all without success. Anyone know what's up with this? I've been wanting to do some Clascal development now that "the bug" has been fixed.

[jamesdenton]

Hi -

I haven't tried a build recently, but the latest MacOS X downloadable release seems to work on MacOS 13 (though it was last built against MacOS 10.11). I did a MacWorks install without issue, so it's entirely possible there's a bug w/ Linux that wasn't discovered.

[jamesdenton]

I did just compile the latest "master' branch against MacOS X 12.5.1 and got the following error attempting to boot from floppy:

"Sorry, the emulation aborted due to a fatal error... ipc is null..." etc etc

Can't say exactly why this is happening, but I'll have to look at previous correspondence with Ray to see if there's anything helpful there.

Which version of wxWidgets did you compile against?

[kgraaf]

I did just compile the latest "master' branch against MacOS X 12.5.1 and got the following error attempting to boot from floppy:

"Sorry, the emulation aborted due to a fatal error... ipc is null..." etc etc

Can't say exactly why this is happening, but I'll have to look at previous correspondence with Ray to see if there's anything helpful there.

Which version of wxWidgets did you compile against?

This is precisely the error I get. I get different versions with the other two branches, but the result is the same: trying to boot from a floppy causes a seg fault. I'm compiling against Wx 3.1.5

[patrick]

I had the same issues trying to run LisaEm on a 2009 Mac Mini.
https://lisalist2.com/index.php/topic,52.msg2417.html#msg2417

[kgraaf]

I tried to compile RC2, the earliest version that has the fix for "the bug" and it had the same issue. Can anyone here confirm that they've managed to successfully compile from source into a working version? It seems like the emulator has been broken for a very long time

[jamesdenton]

So, the last correspondence I had about this particular error was on 13-Aug 2020. At the time I was using MacOS Catalina (10.15.6) on x86_64. The circumstances were the same: attempting to boot from a floppy resulted in the error.

From Ray:

"IPC is Instruction Parameter Cache. It shouldn't be null. It means something didn't work right when it allocated memory. So, generator transforms opcodes from the 68000 into basic blocks and then decodes the operands passed to the opcodes because it will like re-execute the same code repeatedly so it doesn't want to do the work twice. So, if it's null it means it wasn't allocated. A workaround can be that if I see it go to null I reallocate it. It can only be null if it's hitting some other bug somewhere else since the MMU translation table it's glued to is there already, or it got clobbered by something wiping that data structure with zeros. Either way, it's bad."

It looks like I was usign wxWidgets 3.1.2 at the time, which I don't think is compatible with my current kit (M1 ARM64/Ventura). I did manage to compile my own at the time, but no idea if I successfully tested boot from floppy at that point.

[gilles]

Compiled yesterday under ubuntu 22.04 (from github main branch)

compile is ok (both wx-widget and lisaem)
romless does not seem to work but runs fine with a real rom H image => this is a workaround to the "ipc null" error message
macworks 3.0 runs fine and installs on profile (boot floppy still needed for next start)
LOS 2.0 starts to install but fails on a profile/widget disk error (cannot write).

[jamesdenton]

Thanks, Gilles! I was beginning to suspect romless, too, but my attempts to use a ROM (on Mac, anyway) were greeted with the same IPC error.

EDIT: Apparently, this is a two-step process. If the ROM is split into hi and lo, and you select bootrom.hi in the Preferences menu and attempt to boot, ROMless mode will still be used *until* LisaEm has joined the hi/lo ROM files into a single boot.ROM. I seem to recall this happening automatically upon closing Prefs, but it's inconsistent now. You can attempt to boot and get presented with the "ROMless Boot" window, then hit cancel. The ROM files will be joined, then subsequent boot successful

[gilles]

for this test I used an already joined/interleaved boot.rom (16Kb)

I suppose this should not be too complex to fix or at least document. If romless you must have a bootable floppy inserted before pressing power on (otherwise there is no 68000 code to run at all)

[edit] something seems to be missing (even if a floppy image is inserted). we may consider romless boot to floppy non functionnal now (fixable but needs work).

[jamesdenton]

[edit] something seems to be missing (even if a floppy image is inserted). we may consider romless boot to floppy non functionnal now (fixable but needs work).

That was my experience, too.

I forked LisaEm over here (https://github.com/arcanebyte/lisaem) and documented the issue in BUGS.txt. Also added support for wxWidgets 3.2.1 and fixed the MacOS X build script. Need to look into the split ROM issue, or at least document how to combine them into a single file.

Ray's dev branch has some changes in ProFile code for LOS 2.0, but I'm not sure how functional it is.