FSUIPC possible memory leak at startup

[Andre92]

Hello John,

I'm suddenly having issues with FSUIPC with started on version 7.5.1 and continued on 7.5.2. When starting FSUIPC after the MSFS2024 has loaded into the 3d world (and entered the cockpit), the sim freezes and looking at Task manager FSuipc is increasing its memory usage continuously. This does not happen every time i start fsuipc and it also does not happen when i load fsuipc before starting the sim.
Attached are my ini and log file and a picture of the task manager showing the abnormal memory usage of fsuipc.

Kind Regards,
Andre

fsuipc.zip

[John Dowson]

The changes in 7.5.1 and 7.5.2 were minimal and I don't see how any of those changes can cause such issues.

At what point (i.e. how long after FSUIPC was started) did you take that task manager screenshot?

Your log shows a strange issue after around 2.5 minutes when the lua autos were killed and input events stopped and then requested:

Quote

  ...
   154125 Killing aircraft lua Autos....
   154125 Lua threads being terminated:
   154125       1 = "E:\Sim\FSUIPC7\HidSwitch.lua"
   154125 Waiting for lua threads to process termination event....
   156328 LUA: "E:\Sim\FSUIPC7\HidSwitch.lua": killed
   156593 **** DevCom read/write threads still running - will exit anyway but could cause issues...
   156906 **** DevCom Read thread terminated
   156968       2 = "E:\Sim\FSUIPC7\bravoleds.lua"
   156984 Waiting for lua threads to process termination event....
   159187 LUA: "E:\Sim\FSUIPC7\bravoleds.lua": killed
   159672 Unsubscribed to aircraft Input Events
   159672 Requesting Input Events....
   161000 Additional 117 InputEvents received - now have 656
...

Do you know what caused this?

Input Events are also stopped and re-requested at around 6 minutes:

Quote

...
   480437 Unsubscribed to aircraft Input Events
   497797 Sim stopped: average frame rate for last 488 secs = 4.1 fps
   497797 -------------------------------------------------------------------
   511234 FS Axis Control Sent: Ctrl=67466, Param=-1324 AXIS_STEERING_SET
   512812 f:\msfs2024\packages\community\inibuilds-aircraft-a300\simobjects\airplanes\a300-600\presets\inibuilds\a300-600 freighter (ge)\config\aircraft.cfg
   550187 Requesting Input Events....
   550265 Additional 117 InputEvents received - now have 656
...

and the lua autos restarted after 9 minutes:

Quote

...
   559500 Lvars received: 3901 L:vars & 0 H:vars now available
   559547 Lvars/Hvars received - checking aircraft autos....
   559609 LUA.3: beginning "E:\Sim\FSUIPC7\HidSwitch.lua"
   559672 LUA.4: beginning "E:\Sim\FSUIPC7\bravoleds.lua"
...

All that seems strange...

And later, after FSUIPC had been running for over and hour and a half, there are a lot of device scanning entries, starting here:

Quote

...
  5680297 ---------------------- Joystick Device Scan -----------------------
  5680297 Product= WINWING JOYSTICK BASE1 + JGRIP 320
  5680297    Manufacturer= Winwing
  5680297    Serial Number= 17E7A940260D032B5125373E
  5680297    Vendor=4098, Product=BE17 (Version 1.85)
  5680297    GUIDs returned for product: VID_4098&PID_BE17:
  5680297       GUID= {52AC98A0-43B4-11EC-8001-444553540000}
  5680297       Details: Btns=7, POVs=(0, 0, 0, 0), Cal=x00000000, Max=R0,U4095,V4095,X65535,Y65535,Z0
  5680297 Product= T-Pendular-Rudder
  5680297    Manufacturer= Thrustmaster
...

which go on for 10+ minutes, and the log ends after 2 hours 7 minutes. So you kept FSUIPC running for the entire session even though you experienced these memory/performance issues at the beginning?

When you get this issue, what happens if you exit and restart FSUIPC7?

It does look like there are some issues with FSUIPC starting/stopping things when it shouldn't, and this is probably due to changes in either events. order of events or camera state changes in MSFS, and is most likely due to MSFS2024 updates and not FSUIPC updates (although you can go back to 7.5.0 if you want to confirm this).

I will look into this, but I will probably require more logs with additional logging set. I will get back to you once I have investigated.

John

[John Dowson]

Could you add offset logging for offset 026D as U8, set WAPI->Debug level logging and Extras logging and show me another log file from when you experience this issue.

Thanks,

John

Later: also please add offset logging for 0258 as U32

Edited March 6 by John Dowson
Later added

[John Dowson]

I've been testing here and cannot reproduce - memory use never goes above 35MB and CPU usage is always between 1.5 - 3%.
When are you starting FSUIPC when this occurs? i.e. what is the exact state of MSFS - in start menu, in main or other menu, plane loaded and ready to fly, plane loading, ....?

It would be ideal if I could reproduce this here (so more information on when this occurs is helpful) but failing that the logs from when this happens, with the additional logging activated,  would be useful.

Thanks,

John

[Andre92]

Sorry,  Family life caught up. I will test some more tomorrow morning, with the offsets you mentioned and will also disable the two lua files you mentioned.
And then report here.

[John Dowson]

No - please dont disable anything. They are not the cause, it is a camera state or event order change that is causing these lua scripts to be stopped and restarted. And this is most probably not related to the memory/performance issue.

Only make the recommended logging changes for now.

Edited March 6 by John Dowson
further info added

[Andre92]

Good morning,

Here is a new log file with the extra options enabled, when a memory leak occured.
This is now with SU1 installed and i had to restart fsuipc several times for it to happen again.

The memory leak does not occur every time i start fsuipc. I first noticed it when i started up fsuipc during the walk around (when you start in 3d, outside of the plane). So i now always enter the cockpit first and then start fsuipc, but it still happens.

Edit: Just tried starting fsuipc when on the main screen in msfs and also got a memory leak after starting several times. I've attached the log file.

Cheers,
Andre
 

fsuipc.zip

FSUIPC7_mainscreen_log.zip

[John Dowson]

Both of those log files and after 4 and a half seconds and were attached when FSUIPC7 was still running (or had crashed?), and don't reveal much....
Can you please always exit FSUIPC7 before attaching files (or let me know if it crashes or doesn't exit cleanly).

Maybe you can also try with the logging console open (Log->Open Console) and tell me what the last message logged is when the memory/performance issues start.

Also please use the attached beta version (7.5.3b). I have modified to not try and read the aircraft.cfg file if it cannot be opened (which is unfortunately most if not all of the time in MSFS2024 due to streaming), and there are also the following changes in this version:
  - allow helicopter-specific axes to be suppressed/disconnected via offset 0x310A
  - allow lua ipc.control function to take two parameters for the control
  - delay auto-start of WebSocketServer until offset 0x3308 populated (i.e. once connected to MSFS)
  - allow use of Input Events in macros
 

John

FSUIPC7.exe

 

[Andre92]

I can't exit fsuipc because it becomes unresponsive and i have to use the Taskmanager (see screenshots).
I've attached another zip with a log of this happening with 7.53b. It would startup normal when in 3d, but had the memory issue in the main screen.
Also attached are two screenshots with the console showing. One with 7.52 running in the 3d and the other with 7.53b in the Main screen.

Btw, either fsuip starts normal, or the memory issue starts, right from the beginning.

fsuipc.zip

[John Dowson]

17 minutes ago, Andre92 said:

I can't exit fsuipc because it becomes unresponsive and i have to use the Taskmanager (see screenshots).

Ah, ok - that is useful to know. No point attaching screenshots though really - the quality is too low to see anything and files and description of the issue are always preferred.

19 minutes ago, Andre92 said:

It would startup normal when in 3d

What does this mean (3d)? 

The log shows a close-down message:

Quote

    64562  5508 Console: Program being closed!

So it looks like the FSUIPC main thread is hitting a race condition. Very strange, especially as you are the only one reporting this.
Could you show me / attach your FSUIPC7.ini file please (although it may be in one of the zip files already attached...).

I am going to need to see more logs with additional logging activated, but not sure what yet...I will get back to you tomorrow (or maybe Sunday)...

John

 

[Andre92]

The fsuipc.ini is in the first zip file.

With 3D i mean when you've loaded into your flight. Sorry, that's an old habit from Falcon BMS days. We used 2D for when in UI and 3D when in game.

Have a nice weekend!

André

[John Dowson]

18 hours ago, Andre92 said:

I can't exit fsuipc because it becomes unresponsive and i have to use the Taskmanager (see screenshots).

Was that also the case with the first log file you attached? That was from a session over 2 hours long, shows continual logging from the main thread, and also shows that FSUIPC exited cleanly after detecting that MSFS had exited. So looks very different from those last two logs.

I will get back to you later with additional logging to try.
 

[John Dowson]

Could you please add the following to the [General] section of your FSUIPC7.ini file:
    TestOptions=x800
    LogCustom=x16

You also have a very high value set for MaxNumberOfCustomEvents (12288). The maximum value for this parameter is 8192.
Shouldn't be a problem (I will test here), but you should reduce that - or is there a need for it to be so high?

 

[Andre92]

Sorry for the late reply, i'm away from home flying. I'll be back home on Thursday and will add the requested lines to the ini file.

The first supplied ini was from a session where fsuipc worked fine. Sorry, i had the memory leak initially, restarted without problems and then saved the ini.

I increased "maxnumberofcustomevents" because inibuilds used to create an enormous amount of events when they initially released their A310 and later A300. I'll set it back to default.

Kind Regards,
André

[John Dowson]

9 hours ago, Andre92 said:

The first supplied ini was from a session where fsuipc worked fine. Sorry, i had the memory leak initially, restarted without problems and then saved the ini.

I think you mean log file, not ini...

9 hours ago, Andre92 said:

I increased "maxnumberofcustomevents" because inibuilds used to create an enormous amount of events when they initially released their A310 and later A300. I'll set it back to default.

This is not a problem - but the current max allowed value is 8192, so if you need the additional higher event numbers then set it to that.
But be aware that this parameter defines the maximum allowed number of custom events which are loaded from event files (*.evt). There is no real need to use these anymore but this is still provided for backwards compatibility. What you may need to increase is the MaxCustomControlNumber ini parameter, which defines the custom control number range allowed by FSUIPC. This does need to be increased for some aircraft.

Cheers,

John

[John Dowson]

@Andre92 Any update on this? Are you still experiencing this memory issue?

[Andre92]

Hi John,

sorry for the late reply.

I think i found the issue:
I've had buttons assigned for time acceleration.

301=PG,13,Cx02000C1A,x0100     -{offset word set, offset 0C1A}-
302=UG,13,Cx02000C1A,x0400     -{offset word set, offset 0C1A}-
303=PG,12,Cx02000C1A,x9600     -{offset word set, offset 0C1A}-
304=UG,12,Cx02000C1A,x0400     -{offset word set, offset 0C1A}-

When starting up FSUIPC, this would (sometimes) trigger fsuipc to continuously trigger the event: SIM_RATE_INCR continuously.
Since i assigned a press and not repeat function, i don't quite understand what i did wrong here?
You can see the event happening in the attached log file.

After removing the above entries, the problem stopped.

Kind Regards,
Andre

fsuipc.zip

[John Dowson]

Ok, thanks for the update.
That is strange though, especially as those controls are being sent so early....
I will look at the Sim Rate offset 0x0C1A when time permits, as this is still documented (and implemented) for FSX (same as in FSUIPC4) and will almost certainly have changed...

Anyway, glad that is now solved!

Regards,

John