Crash on startup on virtualbox/rocky9

[luisbrandao]

Thanks in advance for your bug report!

• [X] Have you reproduced issue in safe mode?
• [X] Have you used the debugging guide to try to resolve the issue?
• [X] Have you checked our FAQs to make sure your question isn't answered there?
• [X] Does your issue already exist?
• [X] Have you checked you are on the latest release of Pulsar?

What happened?

Crash on startup on virtualbox rocky 9

Version used: pulsar-1.63.2022111616.x86_64.rpm

Which OS does this happen on?

🐧 Red Hat based (Fedora, Alma, RockyLinux, CentOS Stream, etc.)

OS details

rocky9

Which CPU architecture are you running this on?

64-bit(x86_64)

What steps are needed to reproduce this?

Just run the program in a virtualbox with rocky9

Additional Information:

techmago@rocky9 ~/Área de trabalho $ /usr/bin/pulsar, linha 194: 3051 Trace/breakpoint trap (imagem do núcleo gravada)nohup "$PULSAR_PATH" --executed-from="$(pwd)" --pid=$$ "$@" > "$ATOM_HOME/nohup.out" 2>&1 libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null) libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null) libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null) libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null) [3051:1116/193018.541178:FATAL:gpu_data_manager_impl_private.cc(415)] GPU process isn't usable. Goodbye.

[luisbrandao]

The same thing does not happen in the host machine (physical machine with rocky8)

[Daeraxa]

On the VM what happens if you start it with --no-sandbox? I've seen the same thing on other VMs where electron/chromium just doesn't play nice with the (lack of) real GPU.

[DeeDeeG]

It's also an issue with upstream Atom. --> https://github.com/atom/atom/issues/23036

I have to run it with one of the workaround flags, I've been using --in-process-gpu, but I think --no-sandbox may be more stable...? I have heard that either works.

If I understand correctly, it will be an issue on all recent Linux distros (specifically Linux with recent glibc) until the Electron version is updated to Electron 13, which is hard to do without breaking things.

[Daeraxa]

In which case we just need to keep this in mind for the Electron 19(?) fork and work. I definitely got around it with --no-sandbox in an OpenSUSE VM.

[Spiker985]

I was forced to use --no-sandbox for both Atom and Pulsar. Looks like it's a glibc error from the comments on that atom issue.

Weird part for me was that it wasn't always that way

[FibroJedi]

I had this error on a clean install of Linux Mint (no VM), but I don't get that on whichever release I have on my old computer. I'll try with the above flags. Out of curiosity I also reinstalled Atom onto the new machine (the last stable version) and got the same error.

I'll see if the workarounds mentioned give me a temporary "fix".

[FibroJedi]

--no-sandbox does run it without it crashing, so that's good. I do get this when I run it from the command line (Linux Mint): cp: cannot stat 'resources/linux/desktopenviroment/cinnamon/pulsar.nemo_action': No such file or directory

Did I miss something in my install?

[Daeraxa]

I'm using Mint but I've definitely not seen that...

[Spiker985]

--no-sandbox does run it without it crashing, so that's good. I do get this when I run it from the command line (Linux Mint): cp: cannot stat 'resources/linux/desktopenviroment/cinnamon/pulsar.nemo_action': No such file or directory

Did I miss something in my install?

There is a file that exists for that within the repo. It sounds like it isn't getting put where it's supposed to and/or, there's a reference to it in a script, which is using a folder which "technically" doesn't exist on an install (not live build)

[luisbrandao]

@Daeraxa --no-sandbox is a viable workaround. thx

[FibroJedi]

--no-sandbox does run it without it crashing, so that's good. I do get this when I run it from the command line (Linux Mint): cp: cannot stat 'resources/linux/desktopenviroment/cinnamon/pulsar.nemo_action': No such file or directory Did I miss something in my install?

There is a file that exists for that within the repo. It sounds like it isn't getting put where it's supposed to and/or, there's a reference to it in a script, which is using a folder which "technically" doesn't exist on an install (not live build)

So, I tried creating that file in the requested location (relative to pulsar.sh) to attempt to avoid the need for --no-sandbox. I had to create all the folders in that path after 'resources'.

If I run it from the command line without sudo, I still get the error (the folder is assigned to root not me). If I do run it with sudo I get

expr: syntax error: missing argument after ‘8’

and the app doesn't load. If I then add --no-sandbox, I get the missing argument error, but the app still loads.

Don't know if any of that is helpful.

[Spiker985]

My assumption is that the file is supposed to be exported from the asar during install, to a computer-wide accessible folder for nemo actions

However, it's lack of existence shouldn't be causing errors

We may need to have you make your own issue @FibroJedi

Placing that file won't fix the --no-sandbox requirement, that's a problem with some native libraries not being able to access the gpu

[FibroJedi]

We may need to have you make your own issue @FibroJedi

Placing that file won't fix the --no-sandbox requirement, that's a problem with some native libraries not being able to access the gpu

Okay, I will do. Thanks.

[confused-Techie]

Resolved in #262 (I realize there is discussion of another issue, but that was requested to be it's own separate issue. As for the original one that this issue refers to has been fixed)