Failures while testing SIMH on a Mac.

[pmetzger]

Howdy!

Using the head of the git repository, I have had the following happen:

The following tests FAILED:
	 52 - simh-infoserver100 (Failed)                       simh-VAX
	 53 - simh-infoserver1000 (Failed)                      simh-VAX
	 54 - simh-infoserver150vxt (Failed)                    simh-VAX
	 55 - simh-microvax1 (Failed)                           simh-VAX
	 58 - simh-microvax3100 (Failed)                        simh-VAX
	 59 - simh-microvax3100e (Failed)                       simh-VAX
	 62 - simh-vax (Failed)                                 simh-VAX
	 63 - simh-vax730 (Failed)                              simh-VAX
	 64 - simh-vax750 (Failed)                              simh-VAX
	 65 - simh-vax780 (Failed)                              simh-VAX
	 66 - simh-vax8200 (Failed)                             simh-VAX
	 67 - simh-vax8600 (Failed)                             simh-VAX
	 68 - simh-vaxstation3100m30 (Failed)                   simh-VAX
	 69 - simh-vaxstation3100m38 (Failed)                   simh-VAX
Errors while running CTest

Looking at the log, typical output seems to be of this form:

55/78 Testing: simh-microvax1
55/78 Test: simh-microvax1
Command: "/opt/local/var/macports/build/simh-09b044a3/work/simh-c5d1f8c5ef589de877a1934f2f264105d64d0ee5/BIN/microvax1" "RegisterSanityCheck" "/opt/local/var/macports/build/simh-09b044a3/work/simh-c5d1f8c5ef589de877a1934f2f264105d64d0ee5/VAX/tests/vax-diag_test.ini" "-v"
Directory: /opt/local/var/macports/build/simh-09b044a3/work/build/VAX
"simh-microvax1" start time: Aug 08 18:13 EDT
Output:
----------------------------------------------------------
 Running internal register sanity checks on MicroVAX I (KA610) simulator.
*** Good Registers in MicroVAX I (KA610) simulator.

MicroVAX I (KA610) simulator Open SIMH V4.1-0 Current        simh git commit id: c5d1f8c5
Running Hardware Core Test (EHKAA)

EHKAA v1.13 CPU Test

���??? error - test 5, subtest 19, INCL failed
Incorrect condition code bits at pc = 8000327D

expected - 0000000A
received - 0000000
*** FAILED - MicroVAX I (KA610) Hardware Core Instruction test EHKAA

<end of output>
Test time =   1.84 sec
----------------------------------------------------------
Test Failed.
"simh-microvax1" end time: Aug 08 18:13 EDT
"simh-microvax1" time elapsed: 00:00:01
----------------------------------------------------------

Any help I could get debugging this would be greatly appreciated.

[markpizz]

For the sake of comparison, try building the microvax1 using the https:/github.com/simh/simh repo.

$ git clone https://github.com/simh/simh simh-simh
$ cd simh-simh
$ make microvax1

This will run both the build and the same related test.

If you notice a different behavior, what you are seeing is an enhancement provided in the open-simh codebase.

Feel free to run which ever version you prefer.

[pmetzger]

Yah, I'm not going to run that code, it's not open source. I'm going to work on the actual open SIMH and not the one that got its license broken.

[markpizz]

You are absolutely free to do that. HOWEVER the changes I've made to https://github.com/simh/simh since the open-simh fork ARE NOT AVAILABLE for anyone to bring over to open-simh. Maybe you or someone else will be motivated to dig deeply into the problems you may uncover now or in the future and actually address them. Good Luck.

[pmetzger]

We are all aware of what you did to SIMH, and the fact that you decided to wreck an open source project that had existed for decades is precisely the reason why I do not want to work on your code or use it.

Please stop posting on this repository.

[pmetzger]

Just to be very clear here: when people are trying to fix a problem in a piece of software, saying "I have a closed source piece of software that you're not allowed to use to fix this" is not helpful or useful behavior.

[mike632t]

Could this be a duplicate of or similar to #177.

[pmetzger]

It's possible, but I am using clang on a Mac and the provided cmake infrastructure.

[pmetzger]

It occurs to me that using the compiler flags to detect undefined behavior might be a way to debug this.

[pmetzger]

Been playing with it a bunch; neither -fno-strict-overflow nor -fwrapv seem to do the trick, and although I've now found tons of UB in the code (and have fixed a bunch of it) that hasn't fixed things either.

[pmetzger]

Some of the discussion in https://www.mail-archive.com/search?l=simh@trailing-edge.com&q=subject:%22Re%5C%3A+%5C%5BSimh%5C%5D+Looking+for+EVKAA+Listing%22&o=newest&f=1 may be relevant to this.

[bscottm]

@pmetzger: My repo's mainpath branch passed all GH CI/CD builds yesterday. mainpath has the code from my submitted PRs and other things I've worked on (mainly the Ethernet code), so it'd be hard to bisect which commit fixes the problem if the tests start working again.

Unfortunately, I don't have a Mac to test on other than the GH CI/CD pipeline.

[pmetzger]

• We can probably diff the Vax files; it's almost certainly one of them.
• The GH CI/CD version of the Mac should be fine btw; I literally just ran ctest
• Is there a good reason that your mainpath fixes shouldn't be committed to this repo?

[bscottm]

@pmetzger: The reason for asking whether my mainpath branch works for you is to narrow down the possibility that what you encountered is a compiler issue or a VAX simulator code issue. Considering that the macOS macos-13 and macos-14 builds on GH pass ctest, it might be a compiler issue.

I'm not sure if GH CI/CD is building against macOS 15 ATM, but I can look and try it out if it's not.

[pkoning2]

Mark, discussions of closed forks of Open SIMH are OFF TOPIC for this forum.

[pkoning2]

Perry, I just tried it with the master branch as updated today (I merged some pending PRs, finally). I don't see any failures (I tried infoserver100, microvax1, and microvax3100; all report success.

This is on Mac OS 15.6, Apple M1 (arm64) silicon.