segfault from child process during Pkg.instantiate

[IanButterworth]

Bisect points to first bad commit 310f590 #49865

With a default environment with this dev-ed JuliaComputing/OmniPackage.jl@322e0e8

$ ./julia --startup-file=no -e "import Pkg; Pkg.instantiate()"

[241258] signal (11.1): Segmentation fault
in expression starting at none:0
Precompiling project...
  Progress [===============>                         ]  206/568

The segfault does not happen under rr

But does with JULIA_CPU_THREADS=1

julia> versioninfo()
Julia Version 1.10.0-DEV.1365
Commit 310f590198 (2023-05-24 12:30 UTC)
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 32 × AMD Ryzen 9 5950X 16-Core Processor
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, znver3)
  Threads: 23 on 32 virtual cores
Environment:
  JULIA_NUM_THREADS = 16

[IanButterworth]

Actually just a status --manifest call is enough

(@v1.10) pkg> st
Status `~/.julia/environments/v1.10/Project.toml`
  [992a9bcd] OmniPackage v0.1.4 `~/.julia/dev/OmniPackage`
  [295af30f] Revise v3.5.2

(@v1.10) pkg> st --manifest
Status `~/.julia/environments/v1.10/Manifest.toml`

[254117] signal (11.1): Segmentation fault
in expression starting at none:0
  [da404889] ARFFFiles v1.4.1
⌅ [c3fe647b] AbstractAlgebra v0.27.10

I suspected the CUDA artifact selection Pkg hook stuff but it doesn't happen with just

(@v1.10) pkg> st
Status `~/.julia/environments/v1.10/Project.toml`
  [052768ef] CUDA v4.3.0
  [1af6417a] CUDA_Runtime_Discovery v0.2.2

[vtjnash]

It looks like it is the CUDA hook that is crashing, according to where that crash happened. Setting JULIA_NUM_THREADS is probably not a good idea quite yet.