The control centre fails to start, giving a Trace/breakpoint trap error

[the-paraw]

As per the title, the control centre fails to start. When I try to start it from the command line, I get Trace/breakpoint trap

This happens regardless of using the rpm or the source code. Checking it up, it looks like the problem is in electron itself, since even just electron refuses to start after installing from source. Other electron applications I have work just fine, which leads me to suspect that the specific version of electron that is installed via npm is too old.

Any help will be greatly appreciated. (Also, would it be possible to package the control centre with a more current version of electron?)

[tuxedoder]

would it be possible to package the control centre with a more current version of electron?

Updating all dependencies is currently work in progress. Electron related update issue with a few updates would be https://github.com/tuxedocomputers/tuxedo-control-center/issues/148. Currently I work on updating Angular. Due to big code changes it will take a while to complete.

When I try to start it from the command line, I get Trace/breakpoint trap

Without more information like device name, complete stack trace, operating system and version, journalctl --boot | grep "tccd\|tuxedo-control-center" or systemctl status tccd I can't really help, since it works on my side.

[the-paraw]

Hi there, thanks for the answer. So, I'm on Gentoo, currently kernel version 6.6.47.

As for the tccd service, it is started. I don't use systemd, but nothing related to the service is in any system log.

Incidentally, the control centre used to start and work fine, but this stopped after an update of the system. Since nothing else stopped working, I'm finding it difficult to pinpoint the cause.

Anyway, I will be very happy to provide a full stack trace. However, how do I go about doing it? If I run it through gdb, I still can't get any info, because the electron that gets pulled by npm install is compiled without debug symbols (or so it seems).

[tuxedoder]

If tccd or tcc don't print something useful in logs or terminal, then my guess would be that it is electron, node, wayland or kernel related. Do you use wayland and have you tried a different kernel version?

Debugging it more indepth is probably not worth it for now, because I think gentoo related problems most likely should be resolved with newer dependency versions. If you can rule out a faulty kernel and have no tuxedo logs to show, then I can only recommend to wait until I manage to update dependencies for now.

[the-paraw]

Hi, so, the problem started on 8 June. On that date, I did not change kernel version. Also, I use (roughly) the same configuration for any new kernel I install, so I tend to exclude that.

As for display server, I have been using X until last week, when I switched to wayland. Neither works.

Finally, I am pretty much 100% sure that it is an electron problem. The reason is that if I install from source and, after npm install, I just try to start electron (not the control centre) locally, with electron -h, I get the very same error, i.e., "Trace/breakpoint trap". To me, this indicates that the first and biggest problem is with electron itself, and this is why I was asking about switching to a more recent version. However, I can't imagine what it is due to.

Also note that everything else works fine. Even the RGB keyboard, which is one of the ite ones, still works without problems and is controllable.

[tuxedoder]

Also note that everything else works fine. Even the RGB keyboard, which is one of the ite ones, still works without problems and is controllable.

tuxedo-drivers and tuxedo-control-center are separate. The driver should work fine unless you encounter dkms errors.

I am pretty much 100% sure that it is an electron problem. The reason is that if I install from source

Since you are comfortable compiling I tried to give instructions on how to only update electron but that seemingly doesn't work due to bigger required code changes. Even when I fix the electron V8 segfault which was introduced in electron 21, the current code uses the outdated dependency ngx-electron which breaks the build with newer electron. That dependency is already removed in my test code.

ERROR:node_bindings.cc(153)] Fatal error in V8: v8_ArrayBuffer_NewBackingStore When the V8 Sandbox is enabled, ArrayBuffer backing stores must be allocated inside the sandbox address space. Please use an appropriate ArrayBuffer::Allocator to allocate these buffers.
libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null)
Segmentation fault (core dumped)

ERROR in node_modules/ngx-electron/lib/electron.service.d.ts:17:31 - error TS2694: Namespace 'Electron.CrossProcessExports' has no exported member 'Remote'.

17     readonly remote: Electron.Remote;

You will have to wait until I manage to fix my dependency update branch to test newer electron. Angular switched to mdc and removed vertical sliders for example, so I am not sure when it will be finished. It will take a while.