Crash Loading P3Dvhttp://forum.simflight.com/topic/82599-crash-loading-p3dv34-reporte-fsuipc4dll/3.4 reporte FSUIPC4.dll

[Pete Dowson]

1 hour ago, gbinoz said:

Tried six or so P3D shut downs and restarts with 4958 and all seems well.

Can you please show me a log (FSUIPC4.LOG) with a successful session with 4.958? I still don't understand why the minor differences will cause a problem. 

4.958 went through many increments before becoming 4.959. Is your 4.958 just "4.958" or is there a letter at the end? There were interim releases a through to j. over a 4 week period from 12/11/2016.

Do you get a crash on exit when using 4.959?

I see you are another Win10 users. That's 7 on Win10, 4 unknowns with the problem. 

1 hour ago, gbinoz said:

I usually use LINDA but I have removed it for the above tests.

Yes, others have the issue without LINDA too. I think LINDA may contribute to a separate matter, the one of crash on exit.

Pete

 

[Pete Dowson]

11 minutes ago, Joerg Alvermann said:

At the end I found 'ipc.display()' at least suspicious. If I eliminate (i.e. comment out)  all 'ipc.display()' function calls in the LUA scripts provided by LINDA, P3D closes normal, and i have no issues on following P3D starts.

As soon as ipc.display() is called only one time,  and I close P3D afterwards, I get the "prepar3d.exe stopped working" message, i.e. P3D crash on exit.

Yes, I thought it might be related to the display functions used by this in P3D3.4 HF2. But there have been no changes in the ipc.display function in FSUIPC for some time, so I'm at a loss to understand how, say, 4.958 works whilst 4.959 gives a problem -- and without LINDA too, as the previous poster has also confirmed.

Thomas and I have been testing Lus display quite extensively in the last day or two. I've had 4 Lua plug-ins using Lua display running at the same time, fighting to show their texts, but with no crashes and always successful exits. I'm on Win7. Thomas, with two simultaneous display users, can get a (trapped) access violation on exit under Win10, but P3D still reloads fine. However, it does possibly indicate a difference here between Win7 and Win10.

I've not heard from anyone have the reload problem with either Win7 or with earlier versions of P3D. Of course, those using earlier versions of P3D might not have yet decided to update to the currently supported version of FSUIPC. But I don't believe I'm the only user still using Win7!

Thanks for the feedback nonetheless. It all helps, but the situation is evidently still confused.

Pete

 

[Pete Dowson]

1 hour ago, Joerg Alvermann said:

At the end I found 'ipc.display()' at least suspicious. If I eliminate (i.e. comment out)  all 'ipc.display()' function calls in the LUA scripts provided by LINDA, P3D closes normal, and i have no issues on following P3D starts.

Another test which you could do. If you set

P3dhacks=x80

into the [General] section of the INI file, it will stop two of FSUIPC's "hacks" in P3D:

Displays like Lua Display and Radar Contact, and
AI deletion (for traffic limiters / management

(The latter was an error with this setting -- next version will separate these two).

Regards

Pete

 

[B773ER]

I'm on W10 when I was having issues with 4.959. Also had LINDA installed. I have since rolled back to 3.3.5. Too many issues with 3.4hf2

[Pete Dowson]

51 minutes ago, B773ER said:

I'm on W10 when I was having issues with 4.959. Also had LINDA installed. I have since rolled back to 3.3.5. Too many issues with 3.4hf2

And 3.3.5 + 4.959 + Win10 is okay?

Pete

 

[Joerg Alvermann]

1 hour ago, Pete Dowson said:

Another test which you could do. If you set

P3dhacks=x80

Yes, this also does the trick, i.e. preventing P3D crash on exit (with the advantage over disabling pic.display() that LINDA messages are still displayed, in the P3D default text window this time)

However, but this is probably expected, P3dhacks=x80 with 4.958 gives me P3D crash on exit

And i can confirm that - even if P3D crashes on exit - P3D 3.4 HF2 starts again without problems when 4.958 is installed.

Thinking about changes between 4.958 and 4.959, while there are not necessarily obvious changes, maybe some subtile ones, where WIn7 more tolerant than Win10?

I don't have a clue about digital signatures and so forth, but could by any chance this 

cause Win10 signature probing to fail and crash P3D? Maybe worth checking with this little nuance in the dll corrected? Just - and probably a very stupid - idea.

[Pete Dowson]

11 minutes ago, Joerg Alvermann said:

Yes, this also does the trick, i.e. preventing P3D crash on exit (with the advantage over disabling pic.display() that LINDA messages are still displayed, in the P3D default text window this time)

Okay. I have now one report saying it made no difference, and yours saying it did something "useful".

More to the point, was P3D alwats restartable without crash afterwards?

12 minutes ago, Joerg Alvermann said:

However, but this is probably expected, P3dhacks=x80 with 4.958 gives me P3D crash on exit

No, not expected at all. It does exactly the same in both. Though some things in FSUIPC just won't work in P3D3.4 HF2.

I've just been comparing the actual code between 4.958 and 4.959. There's nothing which changes normal actions. The changes are all to do with new facilities, and minor diagnostic changes (correcting or adding logging entries). And the same goes back as far as versions made in July. The only things changing affecting P3D specifically are updated links for each subsequent P3D version, and those definitely all work for the P3D versions supported by each FSUIPC release.

I'm quite prepared to believe that there's a problem with the Display facilities in P3D3.4 HF2. There are other reports which look related to this, around the DXGI module (reported on the L-M forum when using an FSL addon aircraft). The other possibility is that the hacked data I'm using to drive the Display facility through P3D's "window.DLL" has been subject to a change in format in HF2. However, I'm using the facility all the time with HF2 on Win7 with no problems.

I'm reaching the point where I'm going to say it's a problem between Win10 and P3D3.4 HF2. But it is very timing critical, because some folks can do several reboots without the startup crash.

When I give up, and I'm fast being forced to, I would  then like to be able to define a minimum setup where it (the restart crash, not necessarily the crash on exit)  so I can persuade L-M to try it themselves. I would hope to be able to do this without LINDA or GSX if I can. GSX isn't so bad, but I think LINDA would be asking quite a lot -- I'm not even wanting to try anything myself with LiINDA. (Or can someone suggest a simple LINDA install without much configuring, or with a ready-made configuration, which will do the trick?).

Pete

 

 

 

[B773ER]

1 hour ago, Pete Dowson said:

And 3.3.5 + 4.959 + Win10 is okay?

Pete

 

Nope, even on 3.2 with W10 and LINDA, latest fsuipc goes right to task manager. Just happened to me.

EDIT: replaced with 949f version and did the same thing.

[Pete Dowson]

1 hour ago, B773ER said:

Nope, even on 3.2 with W10 and LINDA, latest fsuipc goes right to task manager.

The exact same problem? No FSUIPC4.LOG produced?

1 hour ago, B773ER said:

EDIT: replaced with 949f version and did the same thing.

Well, that's looking much more like a specific LINDA problem, or just LINDA + Win10. There have been no previous such reports, and 4.949f dates back to early 2016!

I am only trying to deal with this new problem, which is more and more looking specific to P3D 3.4HF2 and Win10.

Pete

 

[zoro67]

Hi, To info

 

After several hours, and 20x reboot/restart of P3D.. it seems my config now with P3D v3.4.18.19475
+ 4.958 + Win10 + Linda 2.87(408) is working fine.

- Not had any P3D startup/shutdown issues.

 

Brgds Søren

[Joerg Alvermann]

2 hours ago, Pete Dowson said:

More to the point, was P3D alwats restartable without crash afterwards?

affirmed

2 hours ago, Pete Dowson said:

Or can someone suggest a simple LINDA install without much configuring, or with a ready-made configuration, which will do the trick?

Apart from my VRInsight MCP Combo II I have not much programmed/assigned/configured in LINDA, so a very simple LINDA install might be enough to force the error.
Simple LINDA install from my perspective would mean:

1. Unzip the LINDA Package in the respective folders
2. Run LINDA.exe once to do and save the basic configuration
3. Run P3D

--

While i'm a happy camper with the P3dhacks=x80 now, I've done some further testing

P3dhacks=x80 disabled
All LINDA lua scripts deactivated
had ipcReady.lua run a simple lua script which basically only calls ipc.display() 

This 'simple' ipc.display() call was not able to reproduce the error. Reactivating the LINDA lua package also reactivates the error.
This seems to support the theory of timing / display facility issues.

Joerg

[Pete Dowson]

5 hours ago, Joerg Alvermann said:

Simple LINDA install from my perspective would mean:

1. Unzip the LINDA Package in the respective folders
2. Run LINDA.exe once to do and save the basic configuration
3. Run P3D

Okay. Thanks  Is that with no configuring, nothing set, just the LINDA lua running?

I can try that here (on Win7).

Pete

 

[guenseli]

Quote

just the LINDA lua running?

Just the LINDA.exe running

 

Sorry, that I have not reported back the last days. Have had several BSODs, RAM was defect. Have to install everything from scratch now (WIN10 64 Pro)

 

But that is the possibility to test on a vanilla system without anything.

Will report back later

[Pete Dowson]

45 minutes ago, guenseli said:

RAM was defect. Have to install everything from scratch now (WIN10 64 Pro)

You shouldn't need to reinstall anything after replacing RAM, surel?. Or are you suspecting the integrity of everything which has been accessed through defective memory?

46 minutes ago, guenseli said:

But that is the possibility to test on a vanilla system without anything.

AND the opportunity to go back to good old Win7! ;-)

BTW we built a version of FSUIPC with the latest Microsoft compiler (VS15 -- I think a new one is in Beta, but the 2015 version must be good for P3D and Win10), and with Win10 set as the target. All this was in the hope that the problem with (seemingly) Win10 only might be eradicated by a compatible FSUIPC.

No such luck I'm afraid -- my very first volunteer tester got the same failure on the third try.

So I don't think I can solve it.  I'll need L-M to step up.

Pete

 

[guenseli]

Quote

You shouldn't need to reinstall anything after replacing RAM, surel?. Or are you suspecting the integrity of everything which has been accessed through defective memory?

 

I got a lot of BSODs and wanted to trace them down and was finally convinced it is a driver problem. So I reinstalled anything and guess what: the BSODs came back one after another till finally even my BIOS freezed.

Yes, I should have been testing RAM some steps further, but the BSODs always hinted me to some driver errors. Ok, lesson learned.Luckily I have such a good backup system, that it doen't take long

 

Win7??? Never ;)

 

So, first startup with 

UpdatedByVersion=4959

FSVersionUsed="Lockheed Martin® Prepar3D® v3",3.4.18.19475
SimConnectUsed=3.4.0.0

is working

 

I'm curious at which point the error comes up again ... if so

[Pete Dowson]

38 minutes ago, guenseli said:

I'm curious at which point the error comes up again ... if so

Yes, so am I! ;-)

I'm giving up for now. I'm pretty sure I need L-M's help with this, and with them I imagine being so busy with the next development, I doubt I'd get anything sorted for a while. So I'm going to release 4.959e as an interim update with some unrelated changes (the main being that is is completely rebuilt with the latest compilers and libraries and with Win10 as prime target). It doesn't fix this problem though.

Pete

 

[Joerg Alvermann]

50 minutes ago, guenseli said:

So, first startup with 

UpdatedByVersion=4959

FSVersionUsed="Lockheed Martin® Prepar3D® v3",3.4.18.19475
SimConnectUsed=3.4.0.0

is working

Guenther, do you use by any chance multiple displays in an NVIDIA Surround config? 

I was finally able to eliminate the P3D crash on startup with 4.959 and 4.959e by resetting my NVIDIA Surround configuration.
A new system setup is basically nothing different. Would explain why it is now working for you.

Pete, I just send you a mail with my findings.

It looks like the installation of HF2 requires also some sort of 'reset' of the NVIDIA graphics driver.
The following procudere is probably good for most users to solve the P3D crash on startup issue.  

ALL, PLEASE Report back if it helps or  not!

This is based on my system with Nvidia Driver 376.33 WHQL Win10 64 bit

1. If you use NVidia Surround, disable Nvidia Surround

2. If you use NVIDIA Profile Inspector, reset the Prepar3D profile to NVIDIA Defaults

3. Locate your Prepar3d.cfg 

4. Delete prepar3d.cfg and continue with 6.
   OR  
5. modify prepar3d.cfg as follows

5.a. in section [MAIN] delete the 2 lines 

     LocationFullScreen=0,0,5760,1200,\\.\DISPLAY1
     Location=87,0,5760,1200,\\.\DISPLAY1

5.b  delete WHOLE sections (i.e. both lines) of ALL occurences (you'll find 1 entry for each active physical screen if not on NVIDIA Surround)

     [DISPLAY.Device.NVIDIA GeForce GTX 980 Ti.0.0]     <= Exact line differs based on your GPU
     Mode=5760x1200x32                        <= Exact line differs based on your Screen resolution 

     [DISPLAY.Device.NVIDIA GeForce GTX 980 Ti.0.1]
     Mode=1920x1200x32

     [DISPLAY.Device.NVIDIA GeForce GTX 980 Ti.0.2]
     Mode=1920x1200x32

5.c  save & close prepar3d.cfg

6. Reboot 

7. It's a good idea to run FSUIPC4 Install.exe to clean remaining issues of earlier P3D crashes (those we want to fix)

8. Run P3D, let the the default flight load, and wait for all addons to complete their startup / initialization

9. close P3D and wait till the process prepar3d.exe has been terminated completely 
   use task manager or tool of your choice to check there is no running proess of prepar3d.exe remianing

10. start P3D again and check whether it's still crashing (it shouldn't though there might be other reasons causing P3D to crash) 

11. close P3D

done

12. now you can configure NVIDIA surround, NVIDIA Inspector, resolution etc. as you like

[Pete Dowson]

22 hours ago, Joerg Alvermann said:

ALL, PLEASE Report back if it helps or  not!

I asked gypsyczar to do this test. He's come back with a negative I'm afraid. As with everything about this problem, it is just too variable. What applies to one doesn't necessarily apply to another.

Shame. I'm getting L-M involved with all the details I can supply. I doubt if much will change in P3D 3.4 though, but we can but hope for a solution. Maybe they'll be able to tell me what I'm doing wrong, if anything. So far it certainly looks related to other problems folks have reported to L-M, not concerning FSUIPC at all.

Pete

 

[guenseli]

Quote

Guenther, do you use by any chance multiple displays in an NVIDIA Surround config? 

I use 2 screens, but not for the sim (just on one screen) and I do not use NVIDIA surround.

So far I haven't had an issue till now.

 

But there are some addons left I have to install

[guenseli]

ok, now for the interesting part:

 

I have now the last days installed:

P3D with HF2

LINDA

ORBX Vector, Germany North, Trees, Global

Aivlasoft EFB, PFPX, TOPCAT, AS16 and ASCA

and FSUIPC 4959 of course

 

the sim was running the last days without a problem.

 

Now today I have installed Digital Design LOWS (including SODE) and PMDG 737 and 777.

And I have the issue with vanishing P3D to a background task again.

I have then run the installer FSUIPC 4958 and it has said that there's a newer FSUIPC version and is has not been overwritten. But after that I was able to run the sim.

FSUIPC.log says also version 4959 is still installed.

 

 

[Pete Dowson]

2 hours ago, guenseli said:

I have then run the installer FSUIPC 4958 and it has said that there's a newer FSUIPC version and is has not been overwritten. But after that I was able to run the sim.

This report seems to be another repeat of the majority of those posted by folks get the reload problem.  I already have you recorded as one of the Win10 users with the same problem. I am in contact with L-M and hoping for some feedback there, but I also have two volunteer testers who have been tirelessly testing all sorts of things to isolate the causes. Perhaps you might become one of them? It is a non-flying task which I think is why most here have not volunteered. It's a matter of running with your normal add-ons, then closing, then starting again ... for any number of times, until you are happy that the problem is gone ... OR not!

On the diagnostic side with your current result, did P3D suffer a crash on exit from the last run before you got the crash on load? Please check the Windows Event viewer and let me know.

Then try setting these parameters in the [General] section of the FSUIPC4.INI file:

Debug=Please
LogExtras=5
TimeForLuaClosing=0

(If the last is already there, change it from =1 or =2 to =0).

I'd like to see the log -- from a good run if all good. From a run which crashes on exit if one does.

Thanks,
Pete

 

 

 

[Sabrefly]

For what it matters I'd like to add to this thread that my problems with FSUIPCv4.959 start even BEFORE I install P3Dv3.4.18.19475 client! 

My WideClient(s) v6.999z2, that work normal with FSUIPCv4.958, stop seeing Server (P3Dv3.4.918400 Client, not a full install) soon as I install FSUIPCv4.959. While simconnected AS16 works just fine.

 

Thanks,

Igor.

PS: and yes, FSUIPC4.959e crashes everything (P3Dv3.4.18.19475 and  P3Dv3.4.918400) all the same ))

 

[Pete Dowson]

58 minutes ago, Sabrefly said:

For what it matters I'd like to add to this thread that my problems with FSUIPCv4.959 start even BEFORE I install P3Dv3.4.18.19475 client! 

AS requested above, please try setting these parameters in the [General] section of the FSUIPC4.INI file:

Debug=Please
LogExtras=5
TimeForLuaClosing=0

(If the last is already there, change it from =1 or =2 to =0).

I'd like to see the log -- from a good run if all good. From a run which crashes on exit if one does.

Also please state version of Windows and whether you use LINDA.

Oh, you could also try assigning a key press combination to "LuaKillAll" and using it before closing P3D.

Thanks,
Pete

[Sabrefly]

8 minutes ago, Pete Dowson said:

AS requested above, please try setting these parameters in the [General] section of the FSUIPC4.INI file:

Debug=Please
LogExtras=5
TimeForLuaClosing=0

(If the last is already there, change it from =1 or =2 to =0).

I'd like to see the log -- from a good run if all good. From a run which crashes on exit if one does.

Also please state version of Windows and whether you use LINDA.

Oh, you could also try assigning a key press combination to "LuaKillAll" and using it before closing P3D.

Thanks,
Pete

Wilco, Pete.

[guenseli]

Pete, 

 

Im in holidays currently. But will come back to you.