Crash to Desktop if starting lua script

[Calvin212]

Hi!

I've set up a new flight sim computer yesterday, always trying to use the latest versions of P3D, PMDG, etc. I use simple lua scripts to utilize my goflight devices. There are three scripts for my PMDG 737, 777 and 747. The scripts for the  PMDG 737 and 777 work both fine on my new installation. But P3D crashes to desktop as soon as I start the script for the PMDG 747 by a keyboard shortcut.

I replaced the installed version of FSUIPC 5.124 with an older version (5.103b). The script can be started without crash and the goflight devices work as expected.

If further information is required to find the cause for this issue I would be happy to help.

regards,
Oliver

FSUIPC5_crash.log

FSUIPC5_ok.log

[Pete Dowson]

22 minutes ago, Calvin212 said:

If further information is required to find the cause for this issue I would be happy to help.

More information most certainly is needed! The most relevant of course is the Lua file concerned, and possibly one of the other (similar?) ones for comparison.

Then of course there are the crash details. Check in Windows' Event Viewer. Find the crash and show me the details. It'll be in Windows Logs -> Applications.

It would also be good to download 5.124a and try with that, before sending me those files really. That is the current interim version -- see the Download Links sub-forum above. It is just a DLL replacement. That way we are both using the same.

Oh, one other thing. Does it still crash if you don't load the PMDG 747 beforehand?

Pete

 

[Calvin212]

Here we go...

- Version 5.124a crashes too, log attached

- two lua scripts attached. Both scripts do NOT crash with the PMDG 777 (though the 747 script is obviously useless), both DO crash with the PMDG 747

- eventlog: fsuipc_crash_eventlog.txt

Oh, and it does NOT crash if I start the simulation first (with whatever aircraft), call the lua script and then change to PMDG 747. Goflight devices are working fine then. This brings me to the starting point of my investigation: usually the scripts are called by the lua autoload mechanism, which crashed as well. 

 

good luck ;)

Oliver

 

 

B747.lua

B777.lua

FSUIPC5_crash124a.log

[Pete Dowson]

1 hour ago, Calvin212 said:

Both scripts do NOT crash with the PMDG 777 (though the 747 script is obviously useless), both DO crash with the PMDG 747

Aha! That is quite telling!

Is your 747 fully updated? There was a PMDG problem with that particular update which, in the early days of FSUIPC5, I worked around pending a proper PMDG fix by disabling a feature it was calling with the wrong call type (a 32-bit call only executed in the PMDG code if it detected the relevant FSUIPC facility present).

Your previous "working" version of FSUIPC, 5.103b, certainly qualifies as early enough.

Unfortunately the Windows crash data shows an error in "KernelBase.DLL", a part of Windows, so it doesn't help very much except showing it's rather more complicated than a simple code execution problem in FSUIPC.

I can't really do much with your Lua files.  I thought of comparing the two but as they both crash with the 747 but neithe with the 77 there's no value in that. And that result alone also points directly at the PMDG 747.

In the Lua(s) there is this, executed early on before any of the Events can occur. Does that display appear before the crash?

ipc.display("Preparing Cockpit Boeing 747\nFILE: B747.lua", 7)

Assuming that you cannot update 747 because it is already up to date then, though given what youve now said i think it wholly unlikely,  maybe a Lua trace/debug log leading to the point of the crash will help -- enable that in the FSUIPC Logging Tab before starting the 777 Lua (not the 747 one, so I'll know it isn't something specifically in the 747 values). In those logging options, opt for a different Lua log file. Run till the crash then show me that log which will be named after your Lua file..

But the very first thing to do is make sure your 747 is up to date.

1 hour ago, Calvin212 said:

- Version 5.124a crashes too, log attached

I wan't expecting any different. It was just that if the crash was occurring in FSUIPC code them I needed to match your code to mine so I could see exactly where it occurred.

Pete

 

[Calvin212]

Hi Pete!

- PMDG Operation Center doesn't offer me an update. Current Version of 747 is: 3.00.8605.

- The message ("ipc.display") is never shown in the case of a crash, regardless which combination I tested.

- Maybe also worth to be mentioned: It seems, that there passes a second or two from the time of calling the script and the crash.

- Log file for the B777.lua call (PMDG 747(!) loaded) is here: B777_crash_1.log

Oliver

[Pete Dowson]

11 hours ago, Calvin212 said:

- The message ("ipc.display") is never shown in the case of a crash, regardless which combination I tested.

That's strange, because the log shows it is definitely called (line 581). 

11 hours ago, Calvin212 said:

- Maybe also worth to be mentioned: It seems, that there passes a second or two from the time of calling the script and the crash.

It has got to do with one of two things:

1. One (or more) of the Events. Many of them are L:Var events, which needs FSUIPC to continually read those L:Vars to see when they change. There's an example Lua plug-in which monitors L:Vars by reading them all regularly. Colud you test by loading the 747 then starting that plugin? Leave it running for a few sconds. (It is called "log lvars".

2.  The ipc.display call, which should have done its job within the first second or two. could you try with that commented out (put -- in front)?

Failing these actions we might need to try narrowing down what in the Lua could be setting up for the crash. That will be by taking half of it at a time, and keep halving till we find the culprit. (This won't work if there are more than one of them, of course).

One other thing which might help is to run that same 777 Lua, with the 747 loaded, and the Lua logging on, using FSUIPC 5.103b, which you say doesn't crash. Don't run it too long as the log will get huge, just enough to get past where the failing one crashes (2-4 secs?).  Then I can see what will likely be the first thing it does after the point where it crashes in the later version.

Also I can then look to see if I can enable or add more logging. I think it must be related to the events being awaited as the log shows that the Lua execution comes to an end with it waiting for one, but none ever occurs.

I'm running short of time at present -- off on holiday for 11 days next Wednesday.

Pete

 

[Calvin212]

Hi!

I wasn't able to test all you asked me for, but may be this information already helps:

Commenting out the ipc.display call makes the script work. I just started it and killed it (luakillall) several times and the script works always as expected.

One more information: The style of the message  box appears different from the fsuipc version 5.103b (green box with red text) to the latest version (blue box).

The title of the blue box is "simconnect message box". Is this problem related to the versions of simconnect?

Oliver

[Pete Dowson]

43 minutes ago, Calvin212 said:

Commenting out the ipc.display call makes the script work. I just started it and killed it (luakillall) several times and the script works always as expected.

Hmm. That's interesting. Thanks.  

but it is still very odd that it is okay with other aircraft, just the PMDG 747 with this problem. I wonder if it is some sort of clash, that the 747 is trying to use the same facility at the same time. If so it may actually be a P3D bug.

43 minutes ago, Calvin212 said:

One more information: The style of the message  box appears different from the fsuipc version 5.103b (green box with red text) to the latest version (blue box).

I'll check. Single line texts should be using the single line at the top of the window. No box. There never was a box as such.

Multple line displays and Lua displays (as yours is) now use the SimConnect Message Window (not 'box' -- it shouldn't say 'box').

I've got to go out at present, but tomorrow I'll look to see if I can add a little logging around the SimConnect request being mad, to see if it is crashing in SimConnect. If so it will need a report to L-M. 

[Pete Dowson]

This was the change to how Lua displays were shown, listed for 5.110 last August (see the "History" documents:

 

7. The Lua and FSUIPC-offset controlled message Window on screen is improved by making use of
new SimConnect window facilities. It is now similar in appearance to the one used in FSUIPC4.
However, only one such Window is available at a time, and the title is fixed by SimConnect, not
configurable.

I'd like to know if that is the only change which is causing this problem, so I might find and send you version 5.110 to try. I doubt if adding more logging will help if 5.11 has the problem and 5.103b (only just preseding it) doesn't. It would certinly mean a SimConnect bug, so it might then be worth trying an earlier version of P3D. Do you have any?

Pete

 

 

 

 

[Calvin212]

2 hours ago, Pete Dowson said:

It would certinly mean a SimConnect bug, so it might then be worth trying an earlier version of P3D. Do you have any?

Unfortunately no, I uninstalled everything prior setting up my new computer to free the activations.

Oliver

[Pete Dowson]

37 minutes ago, Calvin212 said:

Unfortunately no, I uninstalled everything prior setting up my new computer to free the activations.

Okay.  Try FSUIPC 5.110 please: Install_FSUIPC511.zip
This will only install if you delete your later DLL first. And please do get back to the latest supported version after a test. I'm only wanting to narrow the problem down to that one SimConnect call.

If it is a P3D bug it might be a problem getting it fixed by L-M, as it is very very specific to that one aircraft.

There is one other thing you could do, please. With 5.124 or later, try just a Lua plugin with only that display line, nothing else.  If that fails I can ask PMDG look at it first as they could probably identify the exact nature of the problem more specifically, for reporting to L-M.

If it doesn't fail that way, the puzzle gets rather more difficult.

Pete

 

[Calvin212]

it crashed with 5.11 / PMDG 747 / the lua scripted is reduced to one line (ipc.display...)

it did NOT crash with 5.11 / PMDG 777 / the lua scripted is reduced to one line (ipc.display...)

just to be on the safe side, I double checked:

it did NOT crash with 5.103b / PMDG 747 / the lua scripted is reduced to one line (ipc.display...) (green message box at the top)

 

Oliver

 

[Pete Dowson]

1 hour ago, Calvin212 said:

it crashed with 5.11 / PMDG 747 / the lua scripted is reduced to one line (ipc.display...)

it did NOT crash with 5.11 / PMDG 777 / the lua scripted is reduced to one line (ipc.display...)

just to be on the safe side, I double checked:

it did NOT crash with 5.103b / PMDG 747 / the lua scripted is reduced to one line (ipc.display...) (green message box at the top)

Okay. That's pretty conclusive then. I'll have to ask PMDG to investigate. Maybe they can also test with an earlier version of P3D4 in case it's a new problem.

Thanks. 

For the time being I think you'll just need to run without that display line enabled. If L-M or PMDG come up with anything I'll let you know.

Pete

 

 

[Pete Dowson]

I got a reply back from PMDG straightaway. They think this is the same P3D bug they've been getting other users reporting, and which L-M know about and have in hand. It does only apply to P3D4.2.

For now you'll need to do without that display. Add it back and re-test when the next P3D update is available..

Pete

 

 

[Calvin212]

2 hours ago, Pete Dowson said:

For now you'll need to do without that display. Add it back and re-test when the next P3D update is available..

I already kicked that line. Let's see ...

Oliver