Antimicro crashes when XBox 360 controller turns off automatically

[maxigaz]

Whenever my wireless XBox 360 controller switches off automatically after 15 minutes of being idle, Antimicro crashes.

From time to time, it also happens when I manually disconnect my DualShock 4 via Bluetooth, although I don't know how to reproduce it reliably.

• Operating System: Manjaro Linux KDE
• Antimicro Version: 2.24*
• Qt Version: 5.12.1
• Kernel Version: 4.19.30-1-MANJARO x86_64

*According to antimicro -v, the version number is 2.24. However, the AUR package info says it's 2.24.1.

[juliagoda]

Thanks, I will take a look at it

[maxigaz]

After further testing, I think it has more to do with autoprofiles rather than the automatic switching off of the gamepad.

I've just tried the following with both the Xbox 360 controller and DualShock 4 multiple times, and the results are the same:

1. Launch Antimicro and connect the gamepad.
2. Open an application associated with an autoprofile (in my case, mpv) and close it.
3. Disconnect the gamepad.
4. Antimicro crashes.

[juliagoda]

Thank you, any help I can get is useful. I'll check it all out after defending my thesis.

[juliagoda]

Could you check if this is still present in the last commit from the test branch?

git clone -b testing https://github.com/juliagoda/antimicro.git
cd antimicro
mkdir compile
cd compile
cmake ..
make
sudo make install

[maxigaz]

After some testing, I couldn't reproduce the crash with neither gamepad with the steps I wrote in the reply above.

Although I'm already very pleased, I'll see if the originally supposed issue (with the automatic turn off of the XBox controller) is still present tomorrow for good measure.

[juliagoda]

ok thanks

[maxigaz]

It's still working fine. Thank you so much!

[maxigaz]

Sorry, I'm afraid I'll have to take back what I said in my last reply. Antimicro still crashes when the XBox 360 controller turns off automatically. I'm not sure why it didn't happen earlier today.

However, when I switch it off manually (by removing the batteries), there are no crashes anymore at least.

[maxigaz]

OK, I think I know what was different last time. Antimicro crashed when ~/.config/antimicro was completely new. However, with the old config file, I couldn't make it crash.

I investigated further and found the following in the new config file:

[General]
AutoOpenLastProfile=1

I noticed that in the old one, the value of AutoOpenLastProfile is 0. So I changed it to 0 in the new one as well. It hasn't crashed since then.

[juliagoda]

Thanks for seeing it. You're very perceptive. Besides, I looked on the Internet and noticed that these controllers often turn off after about 15 minutes and in many situations and in some places it causes troublesome problems. So I can still think of a function that would detect a longer period of inactivity and cause motion one degree on any axis, e.g. after 5 minutes or 10 minutes. Perhaps it might even be a good idea to create such an additional option in the settings, where the user would set the frequency himself.

[juliagoda]

run in terminal:
ulimit -c unlimited
an then
antimicro

then try to cause this problem again, then copy the last few lines of the result and paste here. Save the whole result temporarily to a file just in case

[maxigaz]

Save the whole result temporarily to a file just in case

Do you mean save the output to a log file? How can I do that?

[juliagoda]

try:
antimicro &> log.txt

It should write what is happening in the program to the log.txt file. Of course you can name the "log.txt" file differently. Don't forget to call "ulimit -c unlimited" first in the same terminal session.

[maxigaz]

Well, I copied the last ~600 lines of the output manually and posted them here. (I can save the whole input using antimicro &> log.txt next time if you want.)

Out of all that, here are the last 28 lines:

Debug: Raw value is less than  32767  and greather than  -32767 ((null):0, (null))
Debug: Raw value for axis is:  0 ((null):0, (null))
Debug: Throtted value at start of function is:  0 ((null):0, (null))
Debug: Calculated value of throttle is:  0 ((null):0, (null))
Debug: Throtted value at start of function is:  0 ((null):0, (null))
Debug: Calculated value of throttle is:  0 ((null):0, (null))
Debug: Raw value is less than  32767  and greather than  -32767 ((null):0, (null))
Debug: Raw value for axis is:  0 ((null):0, (null))
Debug: Throtted value at start of function is:  0 ((null):0, (null))
Debug: Calculated value of throttle is:  0 ((null):0, (null))
Debug: Throtted value at start of function is:  0 ((null):0, (null))
Debug: Calculated value of throttle is:  0 ((null):0, (null))
Debug: Raw value is less than  32767  and greather than  -32767 ((null):0, (null))
Debug: Raw value for axis is:  0 ((null):0, (null))
Debug: Throtted value at start of function is:  0 ((null):0, (null))
Debug: Calculated value of throttle is:  0 ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 6 and type 'Read', disabling... ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 11 and type 'Read', disabling... ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 10 and type 'Read', disabling... ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 8 and type 'Read', disabling... ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 18 and type 'Read', disabling... ((null):0, (null))
Warning: QSocketNotifier: Invalid socket 9 and type 'Read', disabling... ((null):0, (null))
Debug: "/usr/bin/antimicro" ((null):0, (null))
Warning: The X11 connection broke: I/O error (code 1) ((null):0, (null))
XIO:  fatal IO error 9 (Hibás fájlleíró) on X server ":0"
      after 101251 requests (101251 known processed) with 6 events remaining.
Warning: QObject::killTimer: Timers cannot be stopped from another thread ((null):0, (null))
Warning: QObject::~QObject: Timers cannot be stopped from another thread ((null):0, (null))

[juliagoda]

Could you build it again from test branch and copy lines of code again? This time last 60 lines? It's hard to easily find the problem without such controller

[maxigaz]

I've recently recompiled the latest and left it running with the gamepad being idle for over 15 minutes (and AutoOpenLastProfile set to 1). When I came back, although the gamepad switched off, antimicro didn't crash. I also couldn't find any warnings or error messages in the terminal output this time.

So, I guess you managed to fix it. :) I'll let you know if I can reproduce the issue again and paste the output. In the meantime, thanks a lot! I appreciate your work.

[juliagoda]

Since when has this error occurred? After solving #69? Does the problem appear in other applications too (Is it possible to check this out?)?

[maxigaz]

Since when has this error occurred? After solving #69?

Yes, when I saw your question there, I cloned the master branch, compiled and tested it. Before that, I’d been using Antimicro v2.24.2 (installed from the AUR).

Does the problem appear in other applications too (Is it possible to check this out?)?

Do you mean whether other applications crash when I disconnect a gamepad? No, I haven’t noticed anything like that outside Antimicro.

[juliagoda]

I'll create a condition for wireless gamepads in #74 and we will see if it works.

[maxigaz]

@juliagoda I'm just letting you know that I can't reliably reproduce the issue anymore.

To recap the steps:

1. Make sure AutoOpenLastProfile has the value 1 in ~/.config/antimicro.
2. Launch Antimicro and connect the gamepad.
3. Open an application associated with an autoprofile (in my case, mpv) and close it.
4. Disconnect the gamepad.

By doing this, I could make Antimicro crash again three days ago more than one time in a row. Yesterday it crashed once with the Xbox gamepad. Now (still on the same version), it doesn’t crash anymore with neither gamepads.

Edit: The issue is back. I don't know what happened. Sorry about the confusion.

[juliagoda]

Because the problem may be somewhere else in the code of some library, maybe X11. SDL_JoystickCurrentPowerLevel once shows the result -1, which is SDL_JOYSTICK_POWER_UNKNOWN, and once SDL_JOYSTICK_POWER_WIRED, which is how it should be in my case. It happened in the same place. I changed the code, as I said, but so far it works the way the SDL library wants it to. SDL_GetError does not show any text when -1 occurs, as if there was no problem. There is currently no alternative to SDL_JoystickCurrentPowerLevel, so we have to wait until someone improves this feature. This is quite likely, as there have been various problems with this feature. Maybe I'll come up with another idea.

[juliagoda]

And I came up with another idea. I noticed that when the gamecontrollers were disconnected while such windows like calibration were open, the application crashed. I fixed this kind of bug, so maybe now it will work and maybe that was the reason why it worked sometimes and sometimes not.

[maxigaz]

Was it recently that you pushed the fix? I’ve had AntimicroX from the AUR package antimicrox-git installed since 18 Apr 2020. The application still crashes when I leave the XBox controller untouched for over 15 minutes.

Or if you pushed it after 18 Apr 2020, let me know and I’ll recompile it and see if it works.

[juliagoda]

Yes, I pushed it after the date. I've found #107 #108 and #109 bugs

[maxigaz]

I’ve just reinstalled and tested it. It still crashes. (Unlike the issues you linked, I have no windows of AntimicroX open when that happens, by the way.)

[juliagoda]

Ok then