P3D CTD interaction with FSUIPC

[gypsyczar]

Hi all,

There have been several posts regarding an issue where P3D will CTD upon the first or second start attempt following an FSUIPC install. Once that occurs he only way to get P3D to start is to do another reinstall of FSUIPC. Once FSUIPC is reinstalled P3D will load successfully for one or two time before another CTD.

Pete is back from vacation and has bee provided test data. He has requested a post here so that the LM devs can get involved

[Martyson]

Just for Info .. I posted in FSUIPC forum also:
5/Jan/2017 13:15
All ok here with FSUIPC 4.959 / Win7-64 / P3D 3.4.18.19.475

**
FSUIPC loaded near / first (top/ I always move it/put it there) in DLL.xml:
C:\Users\Vaughan\AppData\Roaming\Lockheed Martin\Prepar3D v3

example:
<?xml version="1.0" encoding="Windows-1252"?>
<SimBase.Document Type="Launch" version="1,0">
<Descr>Launch</Descr>
<Filename>dll.xml</Filename>
<Disabled>False</Disabled>
<Launch.ManualLoad>False</Launch.ManualLoad>
<Launch.Addon>
<Name>Traffic Toolbox</Name>
<Disabled>False</Disabled>
<ManualLoad>False</ManualLoad>
<Path>C:\Program Files (x86)\Lockheed Martin\Prepar3D v3 SDK 3.4.18.19475\Environment SDK\Traffic Toolbox SDK\traffictoolbox.dll</Path>
</Launch.Addon>
<Launch.Addon>
<Name>FSUIPC 4</Name>
<Disabled>False</Disabled>
<Path>Modules\FSUIPC4.dll</Path>
</Launch.Addon>
<Launch.Addon>
<Name>PMDG HUD interface</Name>
<Disabled>False</Disabled>
<Path>PMDG\DLLs\PMDG_HUD_interface.dll</Path>
<DllStartName>module_init</DllStartName>
<DllStopName>module_deinit</DllStopName>
</Launch.Addon>
<Launch.Addon>

**
FSUIPC not present:
C:\ProgramData\Lockheed Martin\Prepar3D v3

**

I do not use Linda

[Pete Dowson]

Pete is back from vacation and has bee provided test data. He has requested a post here so that the LM devs can get involved

Yes. I asked this because of the reports I've received, most seem to be cases of the DLLstart function, called by SimConnect to start it, is being called without the module actually being loaded. Where a crash occurs (as opposed to a hang) the address provided by Windows is the entry point ot that function. FSUIPC is never actually getting to run and seems not to be loaded. I can only assume that the otherwise identical problems where a hang ensues instead are because there's some rubbish at that address which doesn't case a regular crash (access protection).

One of the odd things which a couple of folks have reported is that if they re-install FSUIPC (that is simply re-run the installer), P3D runs ok -- just the once. This is puzzling because the installer will only be copying the same identical DLL module into the Modules folder and checking that the entry in the AppData\Roaming DLL.XML has the loading details in the correct position (last excepting the Active Sky AS_BTSTRP.DLL). Comparison of the DLL.XML before and after shows that no change is occurring there either. Could, perhaps, the area being called for the starting of FSUIPC be now occupied by the previous one loaded late? (But if that were the case, why would the re-install be needed?). It makes no sense.

Unfortunately neither myself of my supporting assistant, Thomas, can reproduce any sort of problems between FSUIPC and P3D, so I think we need some advice from L-M. Some way of diagnosing exactly what SimConnect is doing in this loading process, perhaps, and why things seemed to have changed with HotFix2.

I'm trying to get those reporting this problem to carry out a thorough process of elimination with other Add-ons being loaded by both DLL.XML files. So far a slightly older version of RAASPRO might be implicated, but also maybe a PMDG HUD DLL being loaded before FSUIPC, where some find it okay loading afterwards.

Pete

[Adam Breed]

Hi Pete,

For v3.4 Hotfix 2, there was only one SimConnect change and it would be unlikely to cause this issue:

◦ Fixed a hard crash related to key events via SimConnect being called before the SimConnect dialog was opened

If all other add-ons are disabled/removed, does the issue continue to occur?

Does the FSUIPC installer modify any core configs/files that would then be changed by Prepar3D/FSUIPC after running?

If you restart your machine, without reinstalling, are you able to run again?

Regards,
Adam

[Pete Dowson]

If all other add-ons are disabled/removed, does the issue continue to occur?

Still checking. I think it is definitely some sort of multi-DLL conflict. One theory at present is that P3D is actually crashing on exit, though it looks like a proper exit, and some one or other of the DLLs is leaving something in an odd state. Apparently an FSUIPC-using add-on plug-in called LINDA was, before HotFix2, causing a crash on exit. Now it doesn't appear to do so, but maybe this is just being hidden somehow.

Does the FSUIPC installer modify any core configs/files that would then be changed by Prepar3D/FSUIPC after running?

No. That's what is so puzzling. The Installer copies in the DLL itself, plus its documentation (into a subfoolder in the Modules folder), and reads the DLL.XML from the Appdata\Roaming folder, checks that FSUIPC4 is being loaded in the desired position (if not, makes it so), and writes it back. Testing so far shows the before and after state identical, character for character -- only the file timestamp gets changed.

If the problem is some critical timing thing, could the fact that FSUIPC4.DLL has recently been written make it load faster (from cache) and thus change the timing? Seems unlikely considering how much is being loaded of P3D beforehand.

If you restart your machine, without reinstalling, are you able to run again?

That's something else I've asked on my website (among other things).

I'll be back with more information as I get it. I can't make it fail on two very different PCs, and nor can my support assistant. So it's a problem for us.

Regards
Pete

[gypsyczar]

Pete, Adam

I just did a reinstall of GSX using the 1/4/17 version (download from the FSDT website). I have now restarted P3D multiple times without a crash both with a reboot before restart or restarting without a reboot. The reinstall of GSX was the only change made from the configuration that resulted in the repetitive P3D CTDs. Anyone else having this problem may want to try a GSX reinstall

[Pete Dowson]

Thanks Parker,

I'll publish this on my website too. Odd that I'm using GSX without that update (I have one from mid-December I think). The AddonManager install puts the COUATL.EXE into the EXE.XML, and the BGLMANX.DLL into the DLL.XML files, both on the ProgramData folder I think.

I expect this would be the same, then, for any DreamFleet install.

Pete

[gypsyczar]

Just to bring all up to date. The crash came back after several restarts even with the updated GSX reinstall.

I set up a bare minimum DLL.XML with just FSUIPC, Using this I did not get the CTD during 8 P3d restarts without a reboot between the P3D starts.

<?xml version="1.0" encoding="Windows-1252"?>

<SimBase.Document Type="Launch" version="1,0">
<Descr>Launch</Descr>
<Filename>dll.xml</Filename>
<Disabled>False</Disabled>
<Launch.ManualLoad>False</Launch.ManualLoad>
<Launch.Addon>
<Name>FSUIPC 4</Name>
<Disabled>False</Disabled>
<Path>Modules\FSUIPC4.dll</Path>
</Launch.Addon>
</SimBase.Document>

I know from past testing that adding back the various addons one by one will eventually lead to a repeat of the crash behavior. However, once the crash happens removing the last added "offending" addon in the DLL.XML file and resinstalling FSUIPC will not guarantee the the crash behavior will not reoccur. I have not yet been able to isolate the bad actor.

It appears that there is some persistent interaction between addons over restarts going on that I have not been able to isolate. Once again, the only constant is that a reinstall of FSUIPC will guarantee at least one good start of P3D

[Martyson]

"It appears that there is some persistent interaction between addons over restarts going on that I have not been able to isolate."

**

Just a thought:

P3D restarts do load quicker if you do not re-boot the PC between P3D (or FSX) flights ..... because the OS keeps a bunch in cache .. ready to go.

Maybe the non flushed cache is helping the problem.

[B777ER]

I'm getting this as well. Using latest version of P3D. P3D splash screen comes up and then disappears from the screen but is still running in the task manager. It does say 32 bit in parentheses though when it does this which it normally does not do. Take fsuipc out of the DLL.xml and P3D starts just fine, even with all my other add-ons still in the DLL.xml. A reinstall of fsuipc allows P3D to start normally with LINDA in the modules folder as well. At times I can get 5 to 6 P3D starts even with shutting the computer down in between some of those startups with fsuipc in the DLL.xml before the issue crops up again. Very bizarre.

[Martyson]

Pete, Adam

I just did a reinstall of GSX using the 1/4/17 version (download from the FSDT website). I have now restarted P3D multiple times without a crash both with a reboot before restart or restarting without a reboot. The reinstall of GSX was the only change made from the configuration that resulted in the repetitive P3D CTDs. Anyone else having this problem may want to try a GSX reinstall

****

In my first post in this thread I mentioned :

All ok here with FSUIPC 4.959 / Win7-64 / P3D 3.4.18.19.475
FSUIPC loaded near / first (top/ I always move it/put it there) in DLL.xml:
C:\Users\Vaughan\AppData\Roaming\Lockheed Martin\Prepar3D v3

**
Now I will add:
GSX is not in use (I uninstalled GSX months ago).

[Pete Dowson]

One user has found that an older version of FSUIPC4 (one not intended for P3D 3.4 HF2) doesn't cause the crash. It doesn't support friction tables or controls time, but else runs okay.

So, I'm now analysing the differences. I still don't understand -- as said already, the symptoms are bizarre.

Pete

[Pete Dowson]

Okay, after many tests by one user who can consistently (or eventually) reproduce the problem, we thought we'd narrowed it down. But unfortunately not.

I think it’s very timing critical, and the differences in the timings of events is making the differences. Even recompiling a module with small changes can change such things. I’ve no idea how to go about “fixing” it, or whether I have anything to actually fix or it’s completely down to L-M.

FSUIPC itself really doesn’t change very much – I may add new facilities, and of course they are changes – but 99% is unchanged from release to release. Even the differences between FSX and P3D are very minor, just different addresses mostly. And they change from each version of P3D to the next, even between Beta releases sometimes. And now, users have proven that even stopping ALL of the special changes in FSUIPC peculiar to P3D doesn't help. There's nothing else to change.

It certainly seems to be some sort of interaction between add-ons, whether only DLLs or EXE's or add-on aircraft (which use DLLs too, either by name or as GAU files). But it is so variable, so time critical, I think, that narrowing it down is going to be well nigh impossible. And when it "crashes" it is before FSUIPC is actually running, only when SimConnect loads it and is going to enter it -- at least, that is so according to the only bit of real evidence we have. It certainly doesn't run long enough to open a log file which is one of the first things it does.

I'm wondering if this is specific to P3D3.4 HF2, so I'd welcome feedback from earlier P3D version users to tell us their results with FSUIPC 4.959, which after all is my only supported version. I can't go backwards.

Pete

[Martyson]

Maybe pilots with this problem have some software or a procedure in common?
Do you exit P3D and reboot the PC before starting Active Sky and then starting P3D for a second flight?

Takes longer to get everything loaded but it clears the OS's internal cache.

Do you use Linda?
Do you use GSX?
Are you flying a complex aircraft and then restarting the flight using the P3D top menu?
Are you loading a complex aircraft's saved flight and during the flight restarting the flight using the P3D top menu?


My notes:
All ok so far here with FSUIPC 4.959 (installed 12-21-2016)

P3D 3.4.18.19.475 Hot Fix 2 (installed 12-21-2016)
ASN (build 6197) SP5
EZdok v2
Win7-64
FSGRW ( Flight Sim Global Real Weather)
Orbx global base, PNW, Fjords, NA and a couple airports
PMDG737
VRS FA18
Aerosoft DHC6
Real Air T Duke
Carenado CJ2
I reboot the PC between all my flights except when doing a few quick tests.
So far, in all my full flights and quick tests I have not seen this problem.


I do not use Linda
I do not use GSX (uninstalled it months ago).

[Pete Dowson]

Maybe pilots with this problem have some software or a procedure in common?
Do you exit P3D and reboot the PC before starting Active Sky and then starting P3D for a second flight?

Yes, these are relevant questions. I normally do that -- when I'm flying. But for these tests I just closed down P3D (which also closes all the add-on bits too, programmed that way in FSUIPC's [Programs] section).

And I've done this on two seaparate systems -- both running Win7. I'm wondering now if it's a Win10 thing.

Do you use Linda?
Do you use GSX?

No and Yes, for me.

Are you flying a complex aircraft and then restarting the flight using the P3D top menu?
Are you loading a complex aircraft's saved flight and during the flight restarting the flight using the P3D top menu

No and no for me.

I reboot the PC between all my flights except when doing a few quick tests.
So far, in all my full flights and quick tests I have not seen this problem.

Good. I think there's quite a few of us about. Really, though the thread on my forum has grown and grown, there aren't that many separate reports. There are 8 separate reporters of the exact same symptoms, of which 4 are definitely on Win8 or Win10, and the others haven't said. So I need to find out ...

Pete