Guard against thread creation failure in PULSEAUDIO_DetectDevices about sdl HOT 6 CLOSED

Fixed in b7dc30c for SDL3 and 58f2586 for SDL2.

from sdl.

@smcv, this is related to your other audio hardening work

from sdl.

@smcv, this is related to your other audio hardening work

Only vaguely, I think - I was looking at scenarios where an audio library can be unloaded while a background thread is still trying to run it. This is sort of the opposite: we were hoping a background thread would run audio library code, but actually it never started.

from sdl.

I think we removed the obscenely low stack size in SDL3; I'll take this one.

from sdl.

I think we removed the obscenely low stack size in SDL3; I'll take this one.

Nope, I lied, it's still there. I'll just let it choose the default.

from sdl.

(Also I just want to say that I'm not sure why Thread Local Storage would cause this, as Linux TLS is not stored on the stack, afaik, but I don't know the technical details of any of this.)

EDIT: yes, it is, nevermind.

from sdl.