VAXstation 4000 VLC emulator is very broken

[pmetzger]

I started down my odyssey to fix SIMH because I noticed a while ago that I could not boot NetBSD on the VAXstation 4000 VLC emulator the way I could on an actual machine, which made it hard to use the emulator to debug NetBSD problems.

As of yesterday, I discovered:

1. The dz emulator for the VAXstation (which is distinct from the usual one) is broken in several key respects.
2. The sim_scsi support is insufficient to get NetBSD to boot without a lot of warnings.
3. The NCR SCSI controller emulation is very sketchy.

I've fixed all this in my ZIMH fork (https://github.com/pmetzger/zimh). The patches are fairly clean but they are also against a greatly diverged version of the codebase. However, given fixes for those things, NetBSD 10.1 boots cleanly on the VAXstation 4000 VLC now. I will happily assist anyone who wants to do the work and pull my changes up to the main SIMH in doing so; it should be very little work.

I've dramatically improved the sim_scsi support in general; you guys might want to take back a lot of that.

(And Mark? This isn't your codebase so don't chime in.)

[LegalizeAdulthood]

@pmetzger I'll see what I can do about getting your changes into a PR here. Are the fixes in isolated commits?

[pmetzger]

Yes, and they are reasonably clean, though they may depend on other commits.

[LegalizeAdulthood]

OK, I'll see what I can work out.

[awik32]

I started down my odyssey to fix SIMH because I noticed a while ago that I could not boot NetBSD on the VAXstation 4000 VLC emulator the way I could on an actual machine, which made it hard to use the emulator to debug NetBSD problems.

I have the same problem with the VAXstation 4000 model 60 simulator -- I have an installation that boots on the actual hardware, but not on the emulator.

[pmetzger]

@awik32 Try it on my zimh fork and let me know if it works. It might very well be the same bugs.

[awik32]

@pmetzger It does not build. It complains about undefined references to strlcpy.

[pkoning2]

On Apr 29, 2026, at 9:28 AM, Albert ***@***.***> wrote: awik32 left a comment (open-simh/simh#546) <#546 (comment)> @pmetzger <https://github.com/pmetzger> It does not build. It complains about undefined references to strlcpy.

strlcpy is very nice, but it is a BSD-ism that isn't in Linux, at least not the one I have at home. paul

[pmetzger]

@awik32 I have a compat local copy of strlcpy in src/compat; the problem is it isn't being included correctly, probably because I don't have the right test for whether it is needed. Can you go onto the zimh repo and file an issue telling me the OS etc. you are on and I will fix it?

[pmetzger]

@pkoning2 strlcpy is in all recent glibc variants, has been for a while. However, I have a compatibility version of strlcpy/strlcat in my codebase that is supposed to be included if it's absent during build. Obviously I didn't do that right, I still don't have a CI covering most common OS variants. I'll have that fixed pretty soon.

[LegalizeAdulthood]

@pmetzger strlcpy isn't standard C, just use strncpy which is.

Even Linus Torvalds hates strlcpy.

[pmetzger]

strncpy is dangerous because it doesn't guarantee nul termination so you can end up hanging yourself just as you did in the old days with straight strcpy. Yes, it isn't standard C. However, I've provided an implementation in a compatibility directory that gets invoked if the platform doesn't have it.

[awik32]

@awik32 I have a compat local copy of strlcpy in src/compat; the problem is it isn't being included correctly, probably because I don't have the right test for whether it is needed. Can you go onto the zimh repo and file an issue telling me the OS etc. you are on and I will fix it?

Your repository does not have an "Issues" option. To put it briefly: The error messages refer to strlcpy() and strlcat(), and they are printed by the linker. I'm running Linux Mint 21.3, kernel 6.8.0-59-generic.

[pmetzger]

Your repository does not have an "Issues" option.

It should have had that! I'm sorry about that! I just fixed it.

To put it briefly: The error messages refer to strlcpy() and strlcat(), and they are printed by the linker. I'm running Linux Mint 21.3, kernel 6.8.0-59-generic.

So I just checked, and it turns out strlcat/strlcpy were only put into glibc quite recently. Your version of Linux Mint has glibc 2.35, and strlcpy() and strlcat() were added to glibc 2.38.

It turns out I made some bad assumptions based on using an Arch machine as a test host; I'm going to fix up things with a proper test inside CMake for the existence of strlcat/strlcopy later this morning and I'll post here when I have this and some similar issues fixed. Feel free to start an issue on the ZIMH repository now that it has that available.

[pmetzger]

@awik32 So it turns out that all my Linux test machines have strlcat/strlcpy present, but I've updated the ZIMH configuration stuff so that it should now be pulling in the right thing for you. I'll try testing it on a VM running your version of Mint later, but you can try testing it yourself if you like.

[awik32]

@pmetzger

... you can try testing it yourself if you like.

I did, and it now complains about missing Freetype. I opened an Issue at your repository.