Segmentation fault in sound and temperature

[GladOSkar]

After updating to current master (i3status-rs 0.30.2 (commit 76e19f29 2023-02-22)) and fixing up my config, i got:

$ i3status-rs
{"version": 1, "click_events": true}
[
fish: Job 1, 'i3status-rs' terminated by signal SIGSEGV (Address boundary error)

sometimes also:

$ i3status-rs
{"version": 1, "click_events": true}
[
[],
fish: Job 1, 'i3status-rs' terminated by signal SIGSEGV (Address boundary error)

Minimal configs to reproduce:

[[block]]
block = "sound"

and

[[block]]
block = "temperature"

both create the segfault, most other blocks don't.

This happens both when compiling in debug and in release mode.

I attached the LLVM debugger and got this stack trace (sound block config):

https://gist.github.com/GladOSkar/3352afe39d26f052d4ba6610a369a5a9#file-i3status-rs_segfault-L139
(Segfault in Thread 3, Line 139 ff.)

Looks like a read from 0x0. Going further is a bit beyond me, especially with all that threading and stuff, help would be appreciated.

My system:

• Distribution: Void
• Kernel: Linux [...] 6.1.12_1 #1 SMP PREEMPT_DYNAMIC [...] x86_64 GNU/Linux
• libc: musl
• DE: sway
• rustc: 1.67.1 (d5a82bbd2 2023-02-07)
• cargo: 1.67.1 (8ecd4f20a 2023-01-10)

[MaxVerevkin]

Can you try #401 (comment)?

[ammgws]

libc: musl

worked before with this?>

[GladOSkar]

@MaxVerevkin that works!

@ammgws yes never was a problem

[GladOSkar]

Ohh i just remembered another thing i changed. I switched from void's rust package to rustup (because void's package is stuck on 1.64.0 so i can't compile i3-status-rust with it anymore).

And they apply patches to rustc to make it work better with musl (e.g. 0002 and 0010). So that's probably what was missing.

And sound and temperature might be the blocks (of those that i use) that link to libc somewhere eventually (via libpulse/libsensors)? I thought this wasn't an issue with rust, but if they link into C code it might of course be.

[GladOSkar]

Guess I'll put a note on this into the install docs