Multiple sync variable declarations cause runtime hang for GPU locale model

[daviditen]

With the GPU locale model and CHPL_GPU=nvidia, the example codes release/examples/primers.chpl and release/examples/benchmarks/shootout/threadring.chpl produce all of their expected output and then hang. I reduced the error down to the tiny code below, but we should verify that these two larger tests are fixed before declaring this issue fixed. These pass with CHPL_GPU=cpu.

Steps to Reproduce

Source Code:

var si: sync int;
var sb: sync bool;

Compile command: chpl --no-checks syncs.chpl

Execution command: ./syncs

Configuration Information

$ printchplenv --anonymize
CHPL_TARGET_PLATFORM: linux64
CHPL_TARGET_COMPILER: llvm *
CHPL_TARGET_ARCH: x86_64
CHPL_TARGET_CPU: native
CHPL_LOCALE_MODEL: gpu *
  CHPL_GPU: nvidia *
CHPL_COMM: none *
CHPL_TASKS: qthreads
CHPL_LAUNCHER: none
CHPL_TIMERS: generic
CHPL_UNWIND: bundled *
CHPL_MEM: jemalloc
CHPL_ATOMICS: cstdlib
CHPL_GMP: bundled *
CHPL_HWLOC: bundled
CHPL_RE2: bundled
CHPL_LLVM: bundled *
CHPL_AUX_FILESYS: none
$ chpl --version --no-checks
chpl version 1.33.0 pre-release (30ff6893f3)
  built with LLVM version 15.0.7
  available LLVM targets: amdgcn, r600, nvptx64, nvptx, aarch64_32, aarch64_be, aarch64, arm64_32, arm64, x86-64, x86
Copyright 2020-2023 Hewlett Packard Enterprise Development LP
Copyright 2004-2019 Cray Inc.
(See LICENSE file for more details)

[e-kayrakli]

On 2.2, the code in the OP works without hanging. threadring.chpl also ran successfully in 100 trials for each execopts. Whatever was causing the issue might have been resolved, but I am keeping the issue open for the time being, in case the issue occurs on some other system.