Linux CPU Implementations Crashes Sporatically on all Tests

[cwfitzgerald]

Description
llvmpipe, the software GL driver we use for running our tests on CI, crashes on test-multithreaded-compute with errors that look like stack/heap smashing.

https://github.com/gfx-rs/wgpu/actions/runs/3588896082/jobs/6040764636#step:11:220

     SIGABRT [   0.185s]       wgpu::example/hello-compute tests::test_multithreaded_compute

--- STDOUT:                    wgpu::example/hello-compute tests::test_multithreaded_compute ---

running 1 test
test tests::test_multithreaded_compute ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 3 filtered out; finished in 0.06s


--- STDERR:                    wgpu::example/hello-compute tests::test_multithreaded_compute ---
[2022-12-01T02:34:54Z ERROR wgpu_hal::gles::egl] EGL 'eglInitialize' code 0x3001: DRI2: failed to load driver
corrupted double-linked list

We need to repro this locally and report this to mesa.

[cwfitzgerald]

This now happens sporadically across all tests. It happens 0-2 times every run on my own VM, a retry always fixes it.

[cwfitzgerald]

I think this is fixed