CTD with latest xPilot

[Frank Herter]

Several CTD related on Xpilot.

Files attached

https://drive.google.com/file/d/1WQeQyEfyKmkcRUheoHtV3mwaMt0_Wz31/view?usp=sharing

[Robert Shearman Jr]

Might want to set view permissions to "Anyone with the link" 

[Frank Herter]

2 hours ago, Robert Shearman Jr said:

Might want to set view permissions to "Anyone with the link" 

Sorry done

https://drive.google.com/file/d/1WQeQyEfyKmkcRUheoHtV3mwaMt0_Wz31/view?usp=sharing

[Justin Shannon]

Hi Frank,

This is a crash I've been trying to track down. The crash is happening when a call is made to the X-Plane SDK function that moves the multiplayer aircraft.

I have a test build available to the users in Discord to test out which includes additional logging to try to capture more details about the crash. Join us in the Discord if you're interested: https://vats.im/xpilot-discord

Here's a link to the test build of xPilot: https://cdn.xpilot-project.org/Setup/xPilot-Setup-1.3.33.exe

 

This build will validate the size and values of the dataref vectors that are sent to the XPLM Instance API for processing. Detailed information on this can be found here: https://github.com/TwinFan/XPMP2/issues/23.

It is very important that you inspect and save a copy of the X-Plane log.txt after a flight. If the validation is triggered, it will be logged in the Log.txt file, but may not actually crash X-Plane (instead, the plane that triggered the validation checks will be destructed and no more calls to the XPLM Instance API will be called for that plane).

Additionally, the dataref values that are sent to the instancing API are also validated to make sure they are actual float values. It is possible that the validation here could cause a CTD and terminate X-Plane. If this happens, please upload your full X-Plane Log.txt and the crash dump file (../output/crash_reports/*.dmp or *.rpt).

If you find any FATAL log entries, please upload the entire Log.txt. Below are sample log entries if the validation checks are triggered:

0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:528/CreateInstances: ASSERT FAILED: numDataRefsDuringCreateInstance == 0 || numDataRefsDuringCreateInstance == DR_NAMES.size() 
0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:410/FlightLoopCB: EXCEPTION CAUGHT: 
0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:528/CreateInstances: ASSERT FAILED: numDataRefsDuringCreateInstance == 0 || numDataRefsDuringCreateInstance == DR_NAMES.size()
... 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:444/DoMove: XPMP2 Issue 23: Array sizes mismatch! v: 43, DR_NAMES: 43, numDataRefsDuringCreateInstance: 2 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:445/DoMove: while processing A/C 0x06A2B7, object instance 0000029E361CA520 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:447/DoMove: ASSERT FAILED: (v.size() == DR_NAMES.size()) && (DR_NAMES.size() == numDataRefsDuringCreateInstance) 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:410/FlightLoopCB: EXCEPTION CAUGHT: 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:447/DoMove: ASSERT FAILED: (v.size() == DR_NAMES.size()) && (DR_NAMES.size() == numDataRefsDuringCreateInstance)

 

[Frank Herter]

On 11/8/2020 at 5:26 PM, Justin Shannon said:

Hi Frank,

This is a crash I've been trying to track down. The crash is happening when a call is made to the X-Plane SDK function that moves the multiplayer aircraft.

I have a test build available to the users in Discord to test out which includes additional logging to try to capture more details about the crash. Join us in the Discord if you're interested: https://vats.im/xpilot-discord

Here's a link to the test build of xPilot: https://cdn.xpilot-project.org/Setup/xPilot-Setup-1.3.33.exe

 

This build will validate the size and values of the dataref vectors that are sent to the XPLM Instance API for processing. Detailed information on this can be found here: https://github.com/TwinFan/XPMP2/issues/23.

It is very important that you inspect and save a copy of the X-Plane log.txt after a flight. If the validation is triggered, it will be logged in the Log.txt file, but may not actually crash X-Plane (instead, the plane that triggered the validation checks will be destructed and no more calls to the XPLM Instance API will be called for that plane).

Additionally, the dataref values that are sent to the instancing API are also validated to make sure they are actual float values. It is possible that the validation here could cause a CTD and terminate X-Plane. If this happens, please upload your full X-Plane Log.txt and the crash dump file (../output/crash_reports/*.dmp or *.rpt).

If you find any FATAL log entries, please upload the entire Log.txt. Below are sample log entries if the validation checks are triggered:

0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:528/CreateInstances: ASSERT FAILED: numDataRefsDuringCreateInstance == 0 || numDataRefsDuringCreateInstance == DR_NAMES.size() 
0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:410/FlightLoopCB: EXCEPTION CAUGHT: 
0:04:57.978 xPilot/XPMP2 FATAL Aircraft.cpp:528/CreateInstances: ASSERT FAILED: numDataRefsDuringCreateInstance == 0 || numDataRefsDuringCreateInstance == DR_NAMES.size()
... 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:444/DoMove: XPMP2 Issue 23: Array sizes mismatch! v: 43, DR_NAMES: 43, numDataRefsDuringCreateInstance: 2 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:445/DoMove: while processing A/C 0x06A2B7, object instance 0000029E361CA520 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:447/DoMove: ASSERT FAILED: (v.size() == DR_NAMES.size()) && (DR_NAMES.size() == numDataRefsDuringCreateInstance) 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:410/FlightLoopCB: EXCEPTION CAUGHT: 
0:05:46.590 xPilot/XPMP2 FATAL Aircraft.cpp:447/DoMove: ASSERT FAILED: (v.size() == DR_NAMES.size()) && (DR_NAMES.size() == numDataRefsDuringCreateInstance)

 

Thx I'll try it out.

[John McPherson]

Hello guys,

I've just had my first weird XP11 'freeze' (not a CTD) since we left the 1150 beta test phase.  I've flown over 200 hrs on the release version of XP11.50r3 roughly on an equal mix of IVAO, VATSIM and offline and this is the first issue I've had.  The freeze happened on a busy night over Germany with plenty of aircraft in the vicinity.  

My log.txt is rammed with xPilot model-matching activity, right up to the moment XP11 fails.  there are no error messages and it was not one of the known XP11 issues (vulkan device loss error etc).

The one change I had made before this flight was to update xPilot to the latest version (1.3.32), and to point to a CSL model folder that had NOT been converted using the python script, as the download readme said this was no longer needed.  Is the model-matching on the fly in the latest version tipping XP11 over the edge? (Most of us already tax it as heavily as we can get away with?) 

Edit: I don't use LittleNavMap

Edited November 13, 2020 at 08:51 PM by John McPherson
afterthought

[Justin Shannon]

1 hour ago, John McPherson said:

Hello guys,

I've just had my first weird XP11 'freeze' (not a CTD) since we left the 1150 beta test phase.  I've flown over 200 hrs on the release version of XP11.50r3 roughly on an equal mix of IVAO, VATSIM and offline and this is the first issue I've had.  The freeze happened on a busy night over Germany with plenty of aircraft in the vicinity.  

My log.txt is rammed with xPilot model-matching activity, right up to the moment XP11 fails.  there are no error messages and it was not one of the known XP11 issues (vulkan device loss error etc).

The one change I had made before this flight was to update xPilot to the latest version (1.3.32), and to point to a CSL model folder that had NOT been converted using the python script, as the download readme said this was no longer needed.  Is the model-matching on the fly in the latest version tipping XP11 over the edge? (Most of us already tax it as heavily as we can get away with?) 

Edit: I don't use LittleNavMap

Does X-Plane just freeze and hang? Or does it freeze and eventually shut itself down?

There has been a handful of users in the xPilot discord that have had crashes but no indications in the log. The Windows Even Viewer indicates a graphics driver crash. If you have a NVIDIA card, graphics version 456.xxx+ has been pretty unstable for many users. Try downgrading to 452.xxx. 

[John McPherson]

17 hours ago, Justin Shannon said:

Does X-Plane just freeze and hang? Or does it freeze and eventually shut itself down?

There has been a handful of users in the xPilot discord that have had crashes but no indications in the log. The Windows Even Viewer indicates a graphics driver crash. If you have a NVIDIA card, graphics version 456.xxx+ has been pretty unstable for many users. Try downgrading to 452.xxx. 

Thanks Justin. I'm on NV drivers 457.xxx and not having any apparent issues with that.

I have explore the event viewer and its says that XP11 'hung' and was eventually closed by Windows after not responding.  The defaulting  module was ASXPconnect!  I've checked and there is a new Beta release of ASXP so I might look at that.

Meanwhile, do you have any advice about CSL choice? I use X-CSL for IVAO and up until your last xPilot release I have been pointing xPilot to a copy of that CSL folder which has been converted using the Python script. Can I now use the CSL folder without doing the conversion?  

 

[Justin Shannon]

25 minutes ago, John McPherson said:

Meanwhile, do you have any advice about CSL choice? I use X-CSL for IVAO and up until your last xPilot release I have been pointing xPilot to a copy of that CSL folder which has been converted using the Python script. Can I now use the CSL folder without doing the conversion?  

With xPilot 1.3.32+, running the Python converter script is no longer necessary as xPilot will convert the CSL models on the fly.

[Grant Williams]

On 11/14/2020 at 10:36 AM, John McPherson said:

Thanks Justin. I'm on NV drivers 457.xxx and not having any apparent issues with that.

I have explore the event viewer and its says that XP11 'hung' and was eventually closed by Windows after not responding.  The defaulting  module was ASXPconnect!  I've checked and there is a new Beta release of ASXP so I might look at that.

Meanwhile, do you have any advice about CSL choice? I use X-CSL for IVAO and up until your last xPilot release I have been pointing xPilot to a copy of that CSL folder which has been converted using the Python script. Can I now use the CSL folder without doing the conversion?  

 

Hey John, 

having a similar issue with CSLs. Did you ever come to a solution? 
thanks!

[John McPherson]

On 11/28/2020 at 6:57 AM, Grant Williams said:

Hey John, 

having a similar issue with CSLs. Did you ever come to a solution? 
thanks!

Hello Grant,

Not a solution I'm convinced by!  On the advice of Jennifer at LR I've gone back to using OGL. This has had no issues flying online with either VATSIM/xPilot or IVAO/Altitude.  I have dialled back my graphics settings a little  and also rolled back my Nvidia drivers from 457.xx  to 452.06.

I'd love to get back to using Vulkan (with it's perfomance boost) but I reckon there is still a conflict with the how the CSL models are handled (I use X-CSL, not BB) so until the combined devs at LR and on the pilot clients have tracked it down I'm taking the safe option and not selecting Vulkan