run_test.py

#!/usr/bin/env python3

import argparse
import contextlib
import copy
import glob
import json
import os
import platform
import re
import shutil
import signal
import subprocess
import sys
import sysconfig
import tempfile
import time
from collections import defaultdict
from collections.abc import Sequence
from contextlib import ExitStack
from datetime import datetime
from importlib.metadata import PackageNotFoundError, version
from pathlib import Path
from typing import Any, cast, NamedTuple

import torch
import torch.distributed as dist
from torch.multiprocessing import current_process, get_context
from torch.testing._internal.common_utils import (
    get_report_path,
    IS_CI,
    IS_MACOS,
    IS_WINDOWS,
    isRocmArchAnyOf,
    retry_shell,
    set_cwd,
    shell,
    TEST_CUDA,
    TEST_SAVE_XML,
    TEST_WITH_ASAN,
    TEST_WITH_ROCM,
    TEST_WITH_SLOW_GRADCHECK,
    TEST_XPU,
)


# using tools/ to optimize test run.
REPO_ROOT = Path(__file__).resolve().parent.parent
sys.path.insert(0, str(REPO_ROOT))

from tools.stats.import_test_stats import (
    ADDITIONAL_CI_FILES_FOLDER,
    TEST_CLASS_TIMES_FILE,
    TEST_TIMES_FILE,
)
from tools.stats.upload_metrics import add_global_metric, emit_metric
from tools.testing.discover_tests import (
    CPP_TEST_PATH,
    CPP_TEST_PREFIX,
    CPP_TESTS_DIR,
    parse_test_module,
    TESTS,
)
from tools.testing.do_target_determination_for_s3 import import_results
from tools.testing.target_determination.gen_artifact import gen_ci_artifact
from tools.testing.target_determination.heuristics.previously_failed_in_pr import (
    gen_additional_test_failures_file,
)
from tools.testing.target_determination.heuristics.utils import get_pr_number
from tools.testing.test_run import TestRun
from tools.testing.test_selections import (
    calculate_shards,
    get_job_base_name,
    get_test_case_configs,
    NUM_PROCS,
    ShardedTest,
    THRESHOLD,
)


try:
    from tools.testing.upload_artifacts import (
        parse_xml_and_upload_json,
        upload_adhoc_failure_json,
        zip_and_upload_artifacts,
    )
except ImportError:
    # some imports in those files might fail, e.g., boto3 not installed. These
    # functions are only needed under specific circumstances (CI) so we can
    # define dummy functions here.
    def parse_xml_and_upload_json():
        pass

    def zip_and_upload_artifacts(*args, **kwargs):
        pass

    def upload_adhoc_failure_json(*args, **kwargs):
        pass


from torch.testing._internal.common_utils import HardwareClassification


_HC_CHOICES = [e.name for e in HardwareClassification]


# Make sure to remove REPO_ROOT after import is done
sys.path.remove(str(REPO_ROOT))


HAVE_TEST_SELECTION_TOOLS = True
TEST_CONFIG = os.getenv("TEST_CONFIG", "")
BUILD_ENVIRONMENT = os.getenv("BUILD_ENVIRONMENT", "")
RERUN_DISABLED_TESTS = os.getenv("PYTORCH_TEST_RERUN_DISABLED_TESTS", "0") == "1"
NUM_PYTEST_RERUNS = int(os.getenv("PYTORCH_NUM_PYTEST_RERUNS", "2"))
DISTRIBUTED_TEST_PREFIX = "distributed"
INDUCTOR_TEST_PREFIX = "inductor"
IS_SLOW = "slow" in TEST_CONFIG or "slow" in BUILD_ENVIRONMENT
IS_S390X = platform.machine() == "s390x"


# Note [ROCm parallel CI testing]
# https://github.com/pytorch/pytorch/pull/85770 added file-granularity parallel testing.
# In .ci/pytorch/test.sh, TEST_CONFIG == "default", CUDA and HIP_VISIBLE_DEVICES is set to 0.
# This results in multiple test files sharing the same GPU.
# This should be a supported use case for ROCm, but it exposed issues in the kernel driver resulting in hangs.
# See https://github.com/pytorch/pytorch/issues/90940.
#
# Further, ROCm self-hosted runners have up to 4 GPUs.
# Device visibility was set to 0 to match CUDA test behavior, but this was wasting available GPU resources.
# Assigning each Pool worker their own dedicated GPU avoids the ROCm oversubscription issues.
# This should also result in better overall wall clock time since all GPUs can be utilized.
def maybe_set_hip_visible_devies():
    # Special handling of ROCm GHA runners for parallel (file granularity) tests.
    if torch.version.hip:
        p = current_process()
        if p.name != "MainProcess":
            # this is a Process from a parallel Pool, not the MainProcess
            os.environ["HIP_VISIBLE_DEVICES"] = str(p._identity[0] % NUM_PROCS)


def strtobool(s):
    return s.lower() not in {"", "0", "false", "off"}


class TestChoices(list):
    def __init__(self, *args, **kwargs):
        super().__init__(args[0])

    def __contains__(self, item):
        return list.__contains__(self, parse_test_module(item))


FSDP_TEST = [test for test in TESTS if test.startswith("distributed/fsdp")]

WINDOWS_BLOCKLIST = [
    "distributed/nn/jit/test_instantiator",
    "distributed/rpc/test_faulty_agent",
    "distributed/rpc/test_tensorpipe_agent",
    "distributed/rpc/test_share_memory",
    "distributed/rpc/cuda/test_tensorpipe_agent",
    "distributed/pipeline/sync/skip/test_api",
    "distributed/pipeline/sync/skip/test_gpipe",
    "distributed/pipeline/sync/skip/test_inspect_skip_layout",
    "distributed/pipeline/sync/skip/test_leak",
    "distributed/pipeline/sync/skip/test_portal",
    "distributed/pipeline/sync/skip/test_stash_pop",
    "distributed/pipeline/sync/skip/test_tracker",
    "distributed/pipeline/sync/skip/test_verify_skippables",
    "distributed/pipeline/sync/test_balance",
    "distributed/pipeline/sync/test_bugs",
    "distributed/pipeline/sync/test_checkpoint",
    "distributed/pipeline/sync/test_copy",
    "distributed/pipeline/sync/test_deferred_batch_norm",
    "distributed/pipeline/sync/test_dependency",
    "distributed/pipeline/sync/test_inplace",
    "distributed/pipeline/sync/test_microbatch",
    "distributed/pipeline/sync/test_phony",
    "distributed/pipeline/sync/test_pipe",
    "distributed/pipeline/sync/test_pipeline",
    "distributed/pipeline/sync/test_stream",
    "distributed/pipeline/sync/test_transparency",
    "distributed/pipeline/sync/test_worker",
    "distributed/elastic/agent/server/test/api_test",
    "distributed/elastic/multiprocessing/api_test",
    "distributed/_shard/checkpoint/test_checkpoint"
    "distributed/_shard/checkpoint/test_file_system_checkpoint"
    "distributed/_shard/sharding_spec/test_sharding_spec",
    "distributed/_shard/sharding_plan/test_sharding_plan",
    "distributed/_shard/sharded_tensor/test_sharded_tensor",
    "distributed/_shard/sharded_tensor/test_sharded_tensor_reshard",
    "distributed/_shard/sharded_tensor/ops/test_embedding",
    "distributed/_shard/sharded_tensor/ops/test_embedding_bag",
    "distributed/_shard/sharded_tensor/ops/test_binary_cmp",
    "distributed/_shard/sharded_tensor/ops/test_init",
    "distributed/_shard/sharded_optim/test_sharded_optim",
] + FSDP_TEST

ROCM_BLOCKLIST = [
    "distributed/rpc/test_faulty_agent",
    "distributed/rpc/test_tensorpipe_agent",
    "distributed/rpc/test_share_memory",
    "distributed/rpc/cuda/test_tensorpipe_agent",
    "test_determination",
    "test_jit_legacy",
    "test_cuda_nvml_based_avail",
    "test_jit_cuda_fuser",
    "distributed/pipelining/test_dtensor_pp_integration",
    "inductor/test_cpu_repro",  # excessive runtimes compared to CUDA
]

# Add architecture-specific blocklist entries
if TEST_WITH_ROCM and isRocmArchAnyOf(("gfx1100",)):
    # Some autotune tests on gfx1100 are hanging, disable for now
    ROCM_BLOCKLIST.append("inductor/test_max_autotune")
    # ROCm 7.2 gfx1100 started timing out due to these
    ROCM_BLOCKLIST.append("inductor/test_torchinductor_dynamic_shapes")
    ROCM_BLOCKLIST.append("inductor/test_torchinductor_opinfo")
    ROCM_BLOCKLIST.append("inductor/test_ck_backend")

S390X_BLOCKLIST = [
    # these tests fail due to various reasons
    "dynamo/test_misc",
    "inductor/test_cpu_repro",
    "inductor/test_cpu_select_algorithm",
    "inductor/test_torchinductor_codegen_dynamic_shapes",
    "lazy/test_meta_kernel",
    "onnx/test_utility_funs",
    "profiler/test_profiler",
    "test_jit",
    "dynamo/test_utils",
    "test_nn",
    # these tests run long and fail in addition to that
    "dynamo/test_dynamic_shapes",
    "test_quantization",
    "inductor/test_torchinductor",
    "inductor/test_torchinductor_dynamic_shapes",
    "inductor/test_torchinductor_opinfo",
    # these tests fail when cuda is not available
    "inductor/test_aot_inductor",
    "inductor/test_best_config",
    "inductor/test_cudacodecache",
    "inductor/test_inductor_utils",
    "inductor/test_inplacing_pass",
    "inductor/test_kernel_benchmark",
    "inductor/test_max_autotune",
    "inductor/test_move_constructors_to_gpu",
    "inductor/test_multi_kernel",
    "inductor/test_pattern_matcher",
    "inductor/test_perf",
    "inductor/test_select_algorithm",
    "inductor/test_snode_runtime",
    "inductor/test_triton_wrapper",
    # these tests fail when mkldnn is not available
    "inductor/test_custom_post_grad_passes",
    "inductor/test_mkldnn_pattern_matcher",
    "test_metal",
    # lacks quantization support
    "onnx/test_models_quantized_onnxruntime",
    "onnx/test_pytorch_onnx_onnxruntime",
    # sysctl -n hw.memsize is not available
    "test_mps",
    # https://github.com/pytorch/pytorch/issues/102078
    "test_decomp",
    # https://github.com/pytorch/pytorch/issues/146698
    "test_model_exports_to_core_aten",
    # runs very long, skip for now
    "inductor/test_layout_optim",
    "test_fx",
    # some false errors
    "doctests",
    # new failures to investigate and fix
    "test_tensorboard",
    # onnx + protobuf failure, see
    # https://github.com/protocolbuffers/protobuf/issues/22104
    "dynamo/test_backends",
    "dynamo/test_modules",
    "inductor/test_config",
    "test_public_bindings",
    "test_testing",
    # depend on z3-solver
    "fx/test_z3_gradual_types",
    "test_proxy_tensor",
]

XPU_BLOCKLIST = [
    "test_autograd",
    "profiler/test_memory_profiler",
]

XPU_TEST = [
    "test_xpu",
]

# The tests inside these files should never be run in parallel with each other
RUN_PARALLEL_BLOCKLIST = [
    "test_extension_utils",
    "test_cpp_extensions_jit",
    "test_cpp_extensions_stream_and_event",
    "test_cpp_extensions_mtia_backend",
    "test_jit_disabled",
    "test_mobile_optimizer",
    "test_multiprocessing",
    "test_multiprocessing_spawn",
    "test_namedtuple_return_api",
    "test_overrides",
    "test_show_pickle",
    "test_tensorexpr",
    "test_cuda_primary_ctx",
    "test_cuda_trace",
    "inductor/test_benchmark_fusion",
    "test_cuda_nvml_based_avail",
    # temporarily sets a global config
    "test_autograd_fallback",
    "inductor/test_compiler_bisector",
    "test_privateuseone_python_backend",
    "functorch/test_control_flow_cuda_initialization",
] + FSDP_TEST

# Test files that should always be run serially with other test files,
# but it's okay if the tests inside them are run in parallel with each other.
CI_SERIAL_LIST = [
    "test_nn",
    "test_fake_tensor",
    "test_cpp_api_parity",
    "test_reductions",
    "test_fx_backends",
    "test_cpp_extensions_jit",
    "test_torch",
    "test_tensor_creation_ops",
    "test_dispatch",
    "test_python_dispatch",  # torch.library creation and deletion must be serialized
    "test_spectral_ops",  # Cause CUDA illegal memory access https://github.com/pytorch/pytorch/issues/88916
    "nn/test_pooling",
    "nn/test_convolution",  # Doesn't respect set_per_process_memory_fraction, results in OOM for other tests in slow gradcheck
    "distributions/test_distributions",
    "test_fx",  # gets SIGKILL
    "functorch/test_memory_efficient_fusion",  # Cause CUDA OOM on ROCm
    "test_utils",  # OOM
    "test_sort_and_select",  # OOM
    "test_backward_compatible_arguments",  # OOM
    "test_autocast",  # OOM
    "test_native_mha",  # OOM
    "test_module_hooks",  # OOM
    "inductor/test_max_autotune",
    "inductor/test_cutlass_backend",  # slow due to many nvcc compilation steps,
    "inductor/test_flex_attention",  # OOM
]
# A subset of onnx tests that cannot run in parallel due to high memory usage.
ONNX_SERIAL_LIST = [
    "onnx/test_models",
    "onnx/test_models_quantized_onnxruntime",
    "onnx/test_models_onnxruntime",
    "onnx/test_custom_ops",
    "onnx/test_utility_funs",
]

# A subset of our TEST list that validates PyTorch's ops, modules, and autograd function as expected
CORE_TEST_LIST = [
    "test_autograd",
    "test_autograd_fallback",
    "test_modules",
    "test_nn",
    "test_ops",
    "test_ops_gradients",
    "test_ops_fwd_gradients",
    "test_ops_jit",
    "test_torch",
]


# if a test file takes longer than 5 min, we add it to TARGET_DET_LIST
SLOW_TEST_THRESHOLD = 300

DYNAMO_WRAPPED_TIMEOUT_MULTIPLIER_OVERRIDE: dict[str, int] = {
    "test_nn": 6,
}

DISTRIBUTED_TESTS_CONFIG = {}


if dist.is_available():
    num_gpus = torch.cuda.device_count()
    DISTRIBUTED_TESTS_CONFIG["test"] = {"WORLD_SIZE": "1"}
    if not TEST_WITH_ROCM and dist.is_mpi_available():
        DISTRIBUTED_TESTS_CONFIG["mpi"] = {
            "WORLD_SIZE": "3",
        }
    if dist.is_nccl_available() and num_gpus > 0:
        DISTRIBUTED_TESTS_CONFIG["nccl"] = {
            "WORLD_SIZE": f"{num_gpus}",
        }
    if dist.is_gloo_available():
        DISTRIBUTED_TESTS_CONFIG["gloo"] = {
            # TODO: retire testing gloo with CUDA
            "WORLD_SIZE": f"{num_gpus if num_gpus > 0 else 3}",
        }
    del num_gpus
    # Test with UCC backend is deprecated.
    # See https://github.com/pytorch/pytorch/pull/137161
    # if dist.is_ucc_available():
    #     DISTRIBUTED_TESTS_CONFIG["ucc"] = {
    #         "WORLD_SIZE": f"{torch.cuda.device_count()}",
    #         "UCX_TLS": "tcp,cuda",
    #         "UCC_TLS": "nccl,ucp,cuda",
    #         "UCC_TL_UCP_TUNE": "cuda:0",  # don't use UCP TL on CUDA as it is not well supported
    #         "UCC_EC_CUDA_USE_COOPERATIVE_LAUNCH": "n",  # CI nodes (M60) fail if it is on
    #     }

# https://stackoverflow.com/questions/2549939/get-signal-names-from-numbers-in-python
SIGNALS_TO_NAMES_DICT = {
    getattr(signal, n): n for n in dir(signal) if n.startswith("SIG") and "_" not in n
}

CPP_EXTENSIONS_ERROR = """
Ninja (https://ninja-build.org) is required for some of the C++ extensions
tests, but it could not be found. Install ninja with `pip install ninja`
or `conda install ninja`. Alternatively, disable said tests with
`run_test.py --exclude test_cpp_extensions_aot_ninja test_cpp_extensions_jit`.
"""

PYTORCH_COLLECT_COVERAGE = bool(os.environ.get("PYTORCH_COLLECT_COVERAGE"))

JIT_EXECUTOR_TESTS = [
    "test_jit_profiling",
    "test_jit_legacy",
    "test_jit_fuser_legacy",
]

INDUCTOR_TESTS = [test for test in TESTS if test.startswith(INDUCTOR_TEST_PREFIX)]
DISTRIBUTED_TESTS = [test for test in TESTS if test.startswith(DISTRIBUTED_TEST_PREFIX)]
TORCH_EXPORT_TESTS = [test for test in TESTS if test.startswith("export")]
AOT_DISPATCH_TESTS = [
    test for test in TESTS if test.startswith("functorch/test_aotdispatch")
]
FUNCTORCH_TESTS = [test for test in TESTS if test.startswith("functorch")]
DYNAMO_CORE_TESTS = [test for test in TESTS if test.startswith("dynamo")]
CPYTHON_TESTS = [test for test in TESTS if "cpython" in test]
ONNX_TESTS = [test for test in TESTS if test.startswith("onnx")]
QUANTIZATION_TESTS = [test for test in TESTS if test.startswith("test_quantization")]


def _is_cpp_test(test):
    # Note: tests underneath cpp_extensions are different from other cpp tests
    # in that they utilize the usual python test infrastructure.
    return test.startswith(CPP_TEST_PREFIX) and not test.startswith("cpp_extensions")


CPP_TESTS = [test for test in TESTS if _is_cpp_test(test)]

TESTS_REQUIRING_LAPACK = [
    "distributions/test_constraints",
    "distributions/test_distributions",
]

# Allowlist of the test files that actually exercise gradcheck -- either via the
# OpInfo/ModuleInfo gradient sweeps or the internal common_utils.gradcheck
# wrapper (which is what flips fast_mode off under slow gradcheck). In slow
# gradcheck mode we run ONLY these: every other file gains nothing from
# fast_mode=False and just burns the per-shard time budget, and the full suite
# is ~2x too large to fit the shards' timeout otherwise.
#
# This is an allowlist, so a NEW file that adds gradcheck coverage must be added
# here or it will silently not run in slow gradcheck. Files that call
# torch.autograd.gradcheck directly (rather than the common_utils wrapper) are
# intentionally omitted: they run identically in normal CI and slow mode adds
# nothing. Use skipIfSlowGradcheckEnv to opt out individual tests within a file.
TESTS_USING_GRADCHECK = [
    # OpInfo / ModuleInfo gradient sweeps -- the core of slow gradcheck
    "test_ops_gradients",
    "test_ops_fwd_gradients",
    "test_modules",
    # Core autograd + nn
    "test_autograd",
    "autograd/test_functional",
    "autograd/test_complex",
    "test_nn",
    "nn/test_convolution",
    "nn/test_pooling",
    "nn/test_parametrization",
    # Domain-specific autograd coverage
    "test_linalg",
    "test_sparse",
    "test_sparse_csr",
    "test_nestedtensor",
    "test_foreach",
    "test_view_ops",
    "test_segment_reductions",
    "test_transformers",
    "test_mkldnn",
    "distributions/test_distributions",
    "optim/test_optim",
]


def print_to_stderr(message):
    print(message, file=sys.stderr)


def get_executable_command(options, disable_coverage=False, is_cpp_test=False):
    if options.coverage and not disable_coverage:
        if not is_cpp_test:
            executable = ["coverage", "run", "--parallel-mode", "--source=torch"]
        else:
            # TODO: C++ with coverage is not yet supported
            executable = []
    else:
        if not is_cpp_test:
            executable = [sys.executable, "-bb"]
        else:
            executable = ["pytest"]

    return executable


def run_test(
    test_module: ShardedTest,
    test_directory,
    options,
    launcher_cmd=None,
    extra_unittest_args=None,
    env=None,
    print_log=True,
) -> int:
    scribe_token = os.getenv("SCRIBE_GRAPHQL_ACCESS_TOKEN", "")
    if scribe_token:
        print_to_stderr("SCRIBE_GRAPHQL_ACCESS_TOKEN is set")
    else:
        print_to_stderr("SCRIBE_GRAPHQL_ACCESS_TOKEN is NOT set")

    env = env or os.environ.copy()
    maybe_set_hip_visible_devies()
    unittest_args = options.additional_args.copy()
    test_file = test_module.name
    # Stable identifier for CI logs, e.g. "test_ops 3/8". Deliberately built
    # from .name rather than str(test_module), which also embeds the pytest
    # filter for partial runs ("test_ops, not (TestA or TestB) 3/8") and would
    # fragment the log classifier's grouping key.
    test_label = f"{test_file} {test_module.shard}/{test_module.num_shards}"
    stepcurrent_key = test_file

    is_distributed_test = test_file.startswith(DISTRIBUTED_TEST_PREFIX)
    is_cpp_test = _is_cpp_test(test_file)
    # NB: Rerun disabled tests depends on pytest-flakefinder and it doesn't work with
    # pytest-cpp atm. We also don't have support to disable C++ test yet, so it's ok
    # to just return successfully here
    if is_cpp_test and RERUN_DISABLED_TESTS:
        print_to_stderr(
            "Skipping C++ tests when running under RERUN_DISABLED_TESTS mode"
        )
        return 0

    if is_cpp_test:
        stepcurrent_key = f"{test_file}_{os.urandom(8).hex()}"
    else:
        unittest_args.extend(
            [
                f"--shard-id={test_module.shard}",
                f"--num-shards={test_module.num_shards}",
            ]
        )
        stepcurrent_key = f"{test_file}_{test_module.shard}_{os.urandom(8).hex()}"

    if options.verbose:
        unittest_args.append(f"-{'v' * options.verbose}")  # in case of pytest

    if test_file in RUN_PARALLEL_BLOCKLIST:
        unittest_args = [
            arg for arg in unittest_args if not arg.startswith("--run-parallel")
        ]

    if extra_unittest_args:
        if not isinstance(extra_unittest_args, list):
            raise AssertionError(
                f"extra_unittest_args must be a list, got {type(extra_unittest_args)}"
            )
        unittest_args.extend(extra_unittest_args)

    # If using pytest, replace -f with equivalent -x
    if options.pytest:
        unittest_args.extend(
            get_pytest_args(
                options,
                is_cpp_test=is_cpp_test,
                is_distributed_test=is_distributed_test,
            )
        )
        unittest_args.extend(test_module.get_pytest_args())
        replacement = {"-f": "-x", "-dist=loadfile": "--dist=loadfile"}
        unittest_args = [replacement.get(arg, arg) for arg in unittest_args]

    if options.hw_classification:
        # forward hw classification filter to test subprocess
        unittest_args += ["--hw-classification"] + options.hw_classification

    if options.showlocals:
        if options.pytest:
            unittest_args.extend(["--showlocals", "--tb=long", "--color=yes"])
        else:
            unittest_args.append("--locals")

    # NB: These features are not available for C++ tests, but there is little incentive
    # to implement it because we have never seen a flaky C++ test before.
    if IS_CI and not is_cpp_test:
        ci_args = ["--import-slow-tests", "--import-disabled-tests"]
        if RERUN_DISABLED_TESTS:
            ci_args.append("--rerun-disabled-tests")
        # use the downloaded test cases configuration, not supported in pytest
        unittest_args.extend(ci_args)

    if test_file in PYTEST_SKIP_RETRIES:
        if not options.pytest:
            raise RuntimeError(
                "A test running without pytest cannot skip retries using "
                "the PYTEST_SKIP_RETRIES set."
            )
        unittest_args = [arg for arg in unittest_args if "--reruns" not in arg]

    # Extra arguments are not supported with pytest
    executable = get_executable_command(options, is_cpp_test=is_cpp_test)
    if not executable:
        # If there is no eligible executable returning here, it means an unsupported
        # case such as coverage for C++ test. So just returning ok makes sense
        return 0

    if is_cpp_test:
        # C++ tests are not the regular test directory
        if CPP_TESTS_DIR:
            cpp_test = os.path.join(
                CPP_TESTS_DIR,
                test_file.replace(f"{CPP_TEST_PREFIX}/", ""),
            )
        else:
            cpp_test = os.path.join(
                Path(test_directory).parent,
                CPP_TEST_PATH,
                test_file.replace(f"{CPP_TEST_PREFIX}/", ""),
            )

        argv = [
            cpp_test if sys.platform != "win32" else cpp_test + ".exe"
        ] + unittest_args
    else:
        # Can't call `python -m unittest test_*` here because it doesn't run code
        # in `if __name__ == '__main__': `. So call `python test_*.py` instead.
        argv = [test_file + ".py"] + unittest_args

    os.makedirs(REPO_ROOT / "test" / "test-reports", exist_ok=True)
    if options.pipe_logs:
        log_fd, log_path = tempfile.mkstemp(
            dir=REPO_ROOT / "test" / "test-reports",
            prefix=f"{sanitize_file_name(str(test_module))}_",
            suffix="_toprint.log",
        )
        os.close(log_fd)

    command = (launcher_cmd or []) + executable + argv
    should_retry = (
        "--subprocess" not in command
        and not RERUN_DISABLED_TESTS
        and not is_cpp_test
        and "-n" not in command
    )
    timeout_multiplier = (
        DYNAMO_WRAPPED_TIMEOUT_MULTIPLIER_OVERRIDE.get(test_file, 3)
        if options.dynamo
        else 3
    )
    timeout = (
        None
        if not options.enable_timeout
        else THRESHOLD * 6
        if IS_SLOW
        else THRESHOLD * timeout_multiplier
        if should_retry
        and isinstance(test_module, ShardedTest)
        and test_module.time is not None
        else THRESHOLD * 3
        if is_cpp_test
        else None
    )
    print_to_stderr(f"Executing {command} ... [{datetime.now()}]")

    with ExitStack() as stack:
        output = None
        if options.pipe_logs:
            output = stack.enter_context(open(log_path, "w"))

        if should_retry:
            ret_code, was_rerun = run_test_retries(
                command,
                test_directory,
                env,
                timeout,
                stepcurrent_key,
                output,
                options.continue_through_error,
                test_file,
                options,
                test_label,
            )
        else:
            command.extend([f"--sc={stepcurrent_key}", "--print-items"])
            ret_code, was_rerun = retry_shell(
                command,
                test_directory,
                stdout=output,
                stderr=output,
                env=env,
                timeout=timeout,
                retries=0,
                label=test_label,
            )

            # Pytest return code 5 means no test is collected. Exit code 4 is
            # returned when the binary is not a C++ test executable, but 4 can
            # also be returned if the file fails before running any tests. All
            # binary files under build/bin that are not C++ test at the time of
            # this writing have been excluded and new ones should be added to
            # the list of exclusions in tools/testing/discover_tests.py
            ret_code = 0 if ret_code == 5 else ret_code

    if options.pipe_logs and print_log:
        handle_log_file(
            test_module, log_path, failed=(ret_code != 0), was_rerun=was_rerun
        )
    return ret_code


def install_cpp_extensions(extensions_dir, env=os.environ):
    # Wipe the build folder, if it exists already
    build_dir = os.path.join(extensions_dir, "build")
    if os.path.exists(build_dir):
        shutil.rmtree(build_dir)

    # Build the test cpp extensions modules
    cmd = [
        sys.executable,
        "-m",
        "pip",
        "install",
        "--no-build-isolation",
        ".",
        "--root",
        "./install",
    ]
    return_code = shell(cmd, cwd=extensions_dir, env=env)
    if return_code != 0:
        return None, return_code

    # Get the site-packages directory prepared for PYTHONPATH
    platlib_path = sysconfig.get_paths()["platlib"]
    platlib_rel = os.path.relpath(
        platlib_path, os.path.splitdrive(platlib_path)[0] + os.sep
    )
    install_directory = os.path.join(extensions_dir, "install", platlib_rel)

    if not install_directory:
        raise AssertionError("install_directory must not be empty")
    return install_directory, 0


@contextlib.contextmanager
def extend_python_path(install_directories):
    python_path = os.environ.get("PYTHONPATH", "")
    try:
        os.environ["PYTHONPATH"] = os.pathsep.join(install_directories + [python_path])
        yield
    finally:
        os.environ["PYTHONPATH"] = python_path


def try_set_cpp_stack_traces(env, command, set=True):
    # Print full c++ stack traces during retries
    env = env or {}
    env["TORCH_SHOW_CPP_STACKTRACES"] = "1" if set else "0"
    return env


def run_test_retries(
    command,
    test_directory,
    env,
    timeout,
    stepcurrent_key,
    output,
    continue_through_error,
    test_file,
    options,
    test_label="",
):
    # Run the test with -x to stop at first failure.  Rerun the test by itself.
    # If it succeeds, move on to the rest of the tests in a new process.  If it
    # still fails, see below
    #
    # If continue through error is not set, then we fail fast.
    #
    # If continue through error is set, then we skip that test, and keep going.
    # Basically if the same test fails 3 times in a row, skip the test on the
    # next run, but still fail in the end. I take advantage of the value saved
    # in stepcurrent to keep track of the most recently run test (which is the
    # one that failed if there was a failure).

    def print_to_file(s):
        print(s, file=output, flush=True)

    num_failures = defaultdict(int)

    def read_pytest_cache(key: str) -> Any:
        cache_file = (
            REPO_ROOT / ".pytest_cache/v/cache/stepcurrent" / stepcurrent_key / key
        )
        try:
            with open(cache_file) as f:
                return f.read()
        except FileNotFoundError:
            return None

    print_items = ["--print-items"]
    sc_command = f"--sc={stepcurrent_key}"
    while True:
        ret_code, _ = retry_shell(
            command + [sc_command] + print_items,
            test_directory,
            stdout=output,
            stderr=output,
            env=env,
            timeout=timeout,
            retries=0,  # no retries here, we do it ourselves, this is because it handles timeout exceptions well
            label=test_label,
        )
        ret_code = 0 if ret_code == 5 else ret_code
        if ret_code == 0 and not sc_command.startswith("--rs="):
            break  # Got to the end of the test suite successfully
        signal_name = f" ({SIGNALS_TO_NAMES_DICT[-ret_code]})" if ret_code < 0 else ""
        print_to_file(f"Got exit code {ret_code}{signal_name}")

        # Read what just failed/ran
        try:
            current_failure = read_pytest_cache("lastrun")
            if current_failure is None:
                raise FileNotFoundError
            if current_failure == "null":
                current_failure = f"'{test_file}'"
        except FileNotFoundError:
            print_to_file(
                "No stepcurrent file found. Either pytest didn't get to run (e.g. import error)"
                + " or file got deleted (contact dev infra)"
            )
            break

        env = try_set_cpp_stack_traces(env, command, set=False)
        if ret_code != 0:
            num_failures[current_failure] += 1

        if ret_code == 124:
            # A timeout where we know which test was in flight. This is for the
            # log classifier, same as FAILED CONSISTENTLY below: without it the
            # only evidence of a timeout is retry_shell's "Command took >Nmin"
            # line, which cannot name the test, so every hang in the fleet
            # collapses into one group and a single hanging test is invisible.
            # [1:-1] to remove quotes. NB when the hang happens during import
            # or collection no test has started, there is no stepcurrent entry,
            # and we never get here -- that case is covered by the file-level
            # label on the "Command took" line instead.
            print_to_file(f"TIMED OUT: {current_failure[1:-1]}")

        if ret_code == 0:
            # Rerunning the previously failing test succeeded, so now we can
            # skip it and move on
            sc_command = f"--scs={stepcurrent_key}"
            print_to_file(
                "Test succeeded in new process, continuing with the rest of the tests"
            )
        elif num_failures[current_failure] >= 3:
            # This is for log classifier so it can prioritize consistently
            # failing tests instead of reruns. [1:-1] to remove quotes
            print_to_file(f"FAILED CONSISTENTLY: {current_failure[1:-1]}")
            if (
                read_pytest_cache("made_failing_xml") == "false"
                and IS_CI
                and options.upload_artifacts_while_running
            ):
                upload_adhoc_failure_json(test_file, current_failure[1:-1])

            if not continue_through_error:
                print_to_file("Stopping at first consistent failure")
                break
            sc_command = f"--scs={stepcurrent_key}"
            print_to_file(
                "Test failed consistently, "
                "continuing with the rest of the tests due to continue-through-error being set"
            )
        else:
            env = try_set_cpp_stack_traces(env, command, set=True)
            sc_command = f"--rs={stepcurrent_key}"
            print_to_file("Retrying single test...")
        print_items = []  # do not continue printing them, massive waste of space

    consistent_failures = [x[1:-1] for x in num_failures if num_failures[x] >= 3]
    flaky_failures = [x[1:-1] for x in num_failures if 0 < num_failures[x] < 3]
    if len(flaky_failures) > 0:
        print_to_file(
            "The following tests failed and then succeeded when run in a new process"
            + f"{flaky_failures}",
        )
    if len(consistent_failures) > 0:
        print_to_file(f"The following tests failed consistently: {consistent_failures}")
        return 1, True
    return ret_code, any(x > 0 for x in num_failures.values())


def run_test_with_subprocess(test_module, test_directory, options):
    return run_test(
        test_module, test_directory, options, extra_unittest_args=["--subprocess"]
    )


def _test_cpp_extensions_aot(test_directory, options, use_ninja):
    if use_ninja:
        try:
            from torch.utils import cpp_extension

            cpp_extension.verify_ninja_availability()
        except RuntimeError:
            print_to_stderr(CPP_EXTENSIONS_ERROR)
            return 1

    # Wipe the build folder, if it exists already
    cpp_extensions_test_dir = os.path.join(test_directory, "cpp_extensions")
    cpp_extensions_test_build_dir = os.path.join(cpp_extensions_test_dir, "build")
    if os.path.exists(cpp_extensions_test_build_dir):
        shutil.rmtree(cpp_extensions_test_build_dir)

    # Build the test cpp extensions modules
    shell_env = os.environ.copy()
    shell_env["USE_NINJA"] = str(1 if use_ninja else 0)
    install_cmd = [
        sys.executable,
        "-m",
        "pip",
        "install",
        "--no-build-isolation",
        ".",
        "--root",
        "./install",
    ]
    wheel_cmd = [sys.executable, "-m", "build", "--wheel", "--no-isolation"]
    return_code = shell(install_cmd, cwd=cpp_extensions_test_dir, env=shell_env)
    if return_code != 0:
        return return_code
    if sys.platform != "win32":
        exts_to_build = [
            (install_cmd, "no_python_abi_suffix_test"),
        ]
        if TEST_CUDA or TEST_XPU:
            exts_to_build.append((wheel_cmd, "python_agnostic_extension"))
        if TEST_CUDA:
            exts_to_build.append((install_cmd, "libtorch_agn_2_9_extension"))
            exts_to_build.append((install_cmd, "libtorch_agn_2_10_extension"))
        for cmd, extension_dir in exts_to_build:
            return_code = shell(
                cmd,
                cwd=os.path.join(cpp_extensions_test_dir, extension_dir),
                env=shell_env,
            )
            if return_code != 0:
                return return_code

    from shutil import copyfile

    os.environ["USE_NINJA"] = shell_env["USE_NINJA"]
    test_module = "test_cpp_extensions_aot" + ("_ninja" if use_ninja else "_no_ninja")
    copyfile(
        test_directory + "/test_cpp_extensions_aot.py",
        test_directory + "/" + test_module + ".py",
    )

    try:
        cpp_extensions = os.path.join(test_directory, "cpp_extensions")
        install_directories = []
        # install directory is the one that is named site-packages
        for root, directories, _ in os.walk(os.path.join(cpp_extensions, "install")):
            for directory in directories:
                if "-packages" in directory:
                    install_directories.append(os.path.join(root, directory))

        for extension_name in [
            "libtorch_agn_2_9_extension",
            "libtorch_agn_2_10_extension",
        ]:
            for root, directories, _ in os.walk(
                os.path.join(cpp_extensions, extension_name, "install")
            ):
                for directory in directories:
                    if "-packages" in directory:
                        install_directories.append(os.path.join(root, directory))

        with extend_python_path(install_directories):
            return run_test(ShardedTest(test_module, 1, 1), test_directory, options)
    finally:
        if os.path.exists(test_directory + "/" + test_module + ".py"):
            os.remove(test_directory + "/" + test_module + ".py")
        os.environ.pop("USE_NINJA")


def test_cpp_extensions_aot_ninja(test_module, test_directory, options):
    return _test_cpp_extensions_aot(test_directory, options, use_ninja=True)


def test_cpp_extensions_aot_no_ninja(test_module, test_directory, options):
    return _test_cpp_extensions_aot(test_directory, options, use_ninja=False)


def test_autoload_enable(test_module, test_directory, options):
    return _test_autoload(test_directory, options, enable=True)


def test_autoload_disable(test_module, test_directory, options):
    return _test_autoload(test_directory, options, enable=False)


def _test_autoload(test_directory, options, enable=True):
    cpp_extensions_test_dir = os.path.join(test_directory, "cpp_extensions")
    install_directory, return_code = install_cpp_extensions(cpp_extensions_test_dir)
    if return_code != 0:
        return return_code

    try:
        os.environ["TORCH_DEVICE_BACKEND_AUTOLOAD"] = str(int(enable))
        with extend_python_path([install_directory]):
            cmd = [sys.executable, "test_autoload.py"]
            return_code = shell(cmd, cwd=test_directory, env=os.environ)
            return return_code
    finally:
        os.environ.pop("TORCH_DEVICE_BACKEND_AUTOLOAD")


# test_openreg is designed to run all tests under torch_openreg, which
# is an torch backend similar to CUDA or MPS and implemented by using
# third-party accelerator integration mechanism. Therefore, if all the
# tests under torch_openreg are passing, it can means that the mechanism
# mentioned above is working as expected.
def test_openreg(test_module, test_directory, options):
    openreg_dir = os.path.join(
        test_directory, "cpp_extensions", "open_registration_extension", "torch_openreg"
    )
    install_dir, return_code = install_cpp_extensions(openreg_dir)
    if return_code != 0:
        return return_code

    # Run the openreg C++ unit tests (gtest) built by cmake.
    ortests_bin = os.path.join(
        openreg_dir, "build", "third_party", "openreg", "ortests"
    )
    if os.path.isfile(ortests_bin):
        return_code = shell([ortests_bin], cwd=openreg_dir)
        if return_code != 0:
            return return_code

    with extend_python_path([install_dir]):
        cmd = [
            sys.executable,
            "-m",
            "unittest",
            "discover",
            "-s",
            os.path.join(openreg_dir, "tests"),
            "-v",
        ]
        return shell(cmd, cwd=test_directory, env=os.environ)


def test_distributed(test_module, test_directory, options):
    mpi_available = shutil.which("mpiexec")
    if options.verbose and not mpi_available:
        print_to_stderr("MPI not available -- MPI backend tests will be skipped")

    config = DISTRIBUTED_TESTS_CONFIG
    for backend, env_vars in config.items():
        if sys.platform == "win32" and backend != "gloo":
            continue
        if backend == "mpi" and not mpi_available:
            continue
        for with_init_file in {True, False}:
            if sys.platform == "win32" and not with_init_file:
                continue
            tmp_dir = tempfile.mkdtemp()
            init_method = "file" if with_init_file else "env"
            if options.verbose:
                with_init = f"with {init_method} init_method"
                print_to_stderr(
                    f"Running distributed tests for the {backend} backend {with_init}"
                )
            old_environ = dict(os.environ)
            os.environ["TEMP_DIR"] = tmp_dir
            os.environ["BACKEND"] = backend
            os.environ.update(env_vars)
            report_tag = f"dist-{backend}" if backend != "test" else ""
            report_tag += f"-init-{init_method}"
            os.environ["TEST_REPORT_SOURCE_OVERRIDE"] = report_tag
            try:
                os.mkdir(os.path.join(tmp_dir, "barrier"))
                os.mkdir(os.path.join(tmp_dir, "test_dir"))
                if backend == "mpi":
                    # test mpiexec for --noprefix option
                    with open(os.devnull, "w") as devnull:
                        allowrunasroot_opt = (
                            "--allow-run-as-root"
                            if subprocess.call(
                                'mpiexec --allow-run-as-root -n 1 bash -c ""',
                                shell=True,
                                stdout=devnull,
                                stderr=subprocess.STDOUT,
                            )
                            == 0
                            else ""
                        )
                        noprefix_opt = (
                            "--noprefix"
                            if subprocess.call(
                                f'mpiexec {allowrunasroot_opt} -n 1 --noprefix bash -c ""',
                                shell=True,
                                stdout=devnull,
                                stderr=subprocess.STDOUT,
                            )
                            == 0
                            else ""
                        )

                    mpiexec = ["mpiexec", "-n", "3", noprefix_opt, allowrunasroot_opt]

                    return_code = run_test(
                        test_module, test_directory, options, launcher_cmd=mpiexec
                    )
                else:
                    return_code = run_test(
                        test_module,
                        test_directory,
                        options,
                        extra_unittest_args=["--subprocess"],
                    )
                if return_code != 0:
                    return return_code
            finally:
                shutil.rmtree(tmp_dir)
                os.environ.clear()
                os.environ.update(old_environ)
    return 0


def run_doctests(test_module, test_directory, options):
    """
    Assumes the incoming test module is called doctest, and simply executes the
    xdoctest runner on the torch library itself.
    """
    import xdoctest

    pkgpath = Path(torch.__file__).parent

    exclude_module_list = ["torch._vendor.*"]
    enabled = {
        # TODO: expose these options to the user
        # For now disable all feature-conditional tests
        # 'lapack': 'auto',
        # 'cuda': 'auto',
        # 'cuda1': 'auto',
        # 'qengine': 'auto',
        "lapack": 0,
        "cuda": 0,
        "cuda1": 0,
        "qengine": 0,
        "autograd_profiler": 0,
        "cpp_ext": 0,
        "monitor": 0,
        "onnx": "auto",
    }

    # Resolve "auto" based on a test to determine if the feature is available.
    if enabled["cuda"] == "auto" and torch.cuda.is_available():
        enabled["cuda"] = True

    if (
        enabled["cuda1"] == "auto"
        and torch.cuda.is_available()
        and torch.cuda.device_count() > 1
    ):
        enabled["cuda1"] = True

    if enabled["lapack"] == "auto" and torch._C.has_lapack:
        enabled["lapack"] = True

    if enabled["qengine"] == "auto":
        try:
            # Is there a better check if quantization is enabled?
            import torch.ao.nn.quantized as nnq  # NOQA: F401

            torch.backends.quantized.engine = "qnnpack"
            torch.backends.quantized.engine = "fbgemm"
        except (ImportError, RuntimeError):
            ...
        else:
            enabled["qengine"] = True

    if enabled["onnx"] == "auto":
        try:
            import onnx  # NOQA: F401
            import onnxruntime  # NOQA: F401
            import onnxscript  # NOQA: F401
        except ImportError:
            exclude_module_list.append("torch.onnx.*")
            enabled["onnx"] = False
        else:
            enabled["onnx"] = True

    # Set doctest environment variables
    if enabled["cuda"]:
        os.environ["TORCH_DOCTEST_CUDA"] = "1"

    if enabled["cuda1"]:
        os.environ["TORCH_DOCTEST_CUDA1"] = "1"

    if enabled["lapack"]:
        os.environ["TORCH_DOCTEST_LAPACK"] = "1"

    if enabled["qengine"]:
        os.environ["TORCH_DOCTEST_QENGINE"] = "1"

    if enabled["autograd_profiler"]:
        os.environ["TORCH_DOCTEST_AUTOGRAD_PROFILER"] = "1"

    if enabled["cpp_ext"]:
        os.environ["TORCH_DOCTEST_CPP_EXT"] = "1"

    if enabled["monitor"]:
        os.environ["TORCH_DOCTEST_MONITOR"] = "1"

    if enabled["onnx"]:
        os.environ["TORCH_DOCTEST_ONNX"] = "1"

    if torch.mps.is_available():
        os.environ["TORCH_DOCTEST_MPS"] = "1"

    if torch.distributed.is_available():
        os.environ["TORCH_DOCTEST_DISTRIBUTED"] = "1"

    if 0:
        # TODO: could try to enable some of these
        os.environ["TORCH_DOCTEST_QUANTIZED_DYNAMIC"] = "1"
        os.environ["TORCH_DOCTEST_ANOMALY"] = "1"
        os.environ["TORCH_DOCTEST_AUTOGRAD"] = "1"
        os.environ["TORCH_DOCTEST_HUB"] = "1"
        os.environ["TORCH_DOCTEST_DATALOADER"] = "1"
        os.environ["TORCH_DOCTEST_FUTURES"] = "1"

    pkgpath = os.path.dirname(torch.__file__)

    xdoctest_config = {
        "global_exec": r"\n".join(
            [
                "from torch import nn",
                "import torch.nn.functional as F",
                "import torch",
                # So doctests can suppress a warning from an intentionally
                # deprecated/prototype example via a `# docs: hide`-marked
                # `warnings.filterwarnings(...)` line without a visible import.
                "import warnings",
            ]
        ),
        "analysis": "static",  # set to "auto" to test doctests in compiled modules
        "style": "google",
        "options": "+IGNORE_WHITESPACE",
    }
    xdoctest_verbose = max(1, options.verbose)
    run_summary = xdoctest.runner.doctest_module(
        os.fspath(pkgpath),
        config=xdoctest_config,
        verbose=xdoctest_verbose,
        command=options.xdoctest_command,
        argv=[],
        exclude=exclude_module_list,
    )
    n_failed = run_summary.get("n_failed", 0)
    n_warned = run_summary.get("n_warned", 0)
    if n_warned:
        print_to_stderr(
            f"ERROR: {n_warned} doctest(s) emitted run-time warnings. Doctests "
            "must be warning-free: either fix the example to use the recommended "
            "API, or if the warning is intrinsic to a deprecated/prototype API "
            "being documented, suppress it with a `# docs: hide`-marked "
            '`>>> warnings.filterwarnings("ignore", message=".*<substr>")` line '
            "(executed by xdoctest, stripped from the rendered docs)."
        )
    return 1 if (n_failed or n_warned) else 0


def sanitize_file_name(file: str):
    return file.replace("\\", ".").replace("/", ".").replace(" ", "_")


def handle_log_file(
    test: ShardedTest, file_path: str, failed: bool, was_rerun: bool
) -> None:
    test = str(test)
    with open(file_path, errors="ignore") as f:
        full_text = f.read()

    new_file = "test/test-reports/" + sanitize_file_name(
        f"{test}_{os.urandom(8).hex()}_.log"
    )
    os.rename(file_path, REPO_ROOT / new_file)

    if not failed and not was_rerun and "=== RERUNS ===" not in full_text:
        # If success + no retries (idk how else to check for test level retries
        # other than reparse xml), print only what tests ran
        print_to_stderr(
            f"\n{test} was successful, full logs can be found in artifacts with path {new_file}"
        )
        for line in full_text.splitlines():
            if re.search("Running .* items in this shard:", line):
                print_to_stderr(line.rstrip())
        print_to_stderr("")
        return

    # otherwise: print entire file
    print_to_stderr(f"\nPRINTING LOG FILE of {test} ({new_file})")
    print_to_stderr(full_text)
    print_to_stderr(f"FINISHED PRINTING LOG FILE of {test} ({new_file})\n")


def get_pytest_args(options, is_cpp_test=False, is_distributed_test=False):
    if is_distributed_test:
        # Distributed tests do not support rerun, see https://github.com/pytorch/pytorch/issues/162978
        rerun_options = ["-x", "--reruns=0"]
    elif RERUN_DISABLED_TESTS:
        # ASAN tests are too slow, so running them x50 will cause the jobs to timeout after
        # 3+ hours. So, let's opt for less number of reruns. We need at least 150 instances of the
        # test every 2 weeks to satisfy the SQL query (15 x 14 = 210).
        count = 15 if TEST_WITH_ASAN else 50
        # When under rerun-disabled-tests mode, run the same tests multiple times to determine their
        # flakiness status. Default to 50 re-runs
        rerun_options = ["--flake-finder", f"--flake-runs={count}"]
    else:
        # When under the normal mode, retry a failed test NUM_PYTEST_RERUNS more times.
        # -x means stop at the first failure. Set PYTORCH_NUM_PYTEST_RERUNS=0 to disable.
        rerun_options = ["-x", f"--reruns={NUM_PYTEST_RERUNS}"]

    pytest_args = [
        "-vv",
        "-rfEX",
    ]
    if not is_cpp_test:
        # C++ tests need to be run with pytest directly, not via python
        # We have a custom pytest shard that conflicts with the normal plugin
        pytest_args.extend(["-p", "no:xdist", "--use-pytest"])
    else:
        # Use pytext-dist to run C++ tests in parallel as running them sequentially using run_test
        # is much slower than running them directly
        pytest_args.extend(["-n", str(NUM_PROCS)])

        if TEST_SAVE_XML:
            # Add the option to generate XML test report here as C++ tests
            # won't go into common_utils
            test_report_path = get_report_path(pytest=True)
            pytest_args.extend(["--junit-xml-reruns", test_report_path])

    if options.pytest_k_expr:
        pytest_args.extend(["-k", options.pytest_k_expr])

    pytest_args.extend(rerun_options)
    return pytest_args


def run_ci_sanity_check(test: ShardedTest, test_directory, options):
    if test.name != "test_ci_sanity_check_fail":
        raise AssertionError(
            f"This handler only works for test_ci_sanity_check_fail, got {test.name}"
        )
    ret_code = run_test(test, test_directory, options, print_log=False)
    # This test should fail
    if ret_code != 1:
        return 1
    test_reports_dir = str(REPO_ROOT / "test/test-reports")
    # Delete the log files and xmls generated by the test
    for file in glob.glob(f"{test_reports_dir}/{test.name}*.log"):
        os.remove(file)
    for dirname in glob.glob(f"{test_reports_dir}/**/{test.name}"):
        shutil.rmtree(dirname)
    return 0


CUSTOM_HANDLERS = {
    "test_cuda_primary_ctx": run_test_with_subprocess,
    "test_cuda_nvml_based_avail": run_test_with_subprocess,
    "test_cuda_trace": run_test_with_subprocess,
    "test_cpp_extensions_aot_no_ninja": test_cpp_extensions_aot_no_ninja,
    "test_cpp_extensions_aot_ninja": test_cpp_extensions_aot_ninja,
    "distributed/test_distributed_spawn": test_distributed,
    "distributed/algorithms/quantization/test_quantization": test_distributed,
    "distributed/test_c10d_nccl": run_test_with_subprocess,
    "distributed/test_c10d_gloo": run_test_with_subprocess,
    "distributed/test_c10d_ucc": run_test_with_subprocess,
    "distributed/test_c10d_common": run_test_with_subprocess,
    "distributed/test_c10d_spawn_gloo": run_test_with_subprocess,
    "distributed/test_c10d_spawn_nccl": run_test_with_subprocess,
    "distributed/test_c10d_spawn_ucc": run_test_with_subprocess,
    "distributed/test_store": run_test_with_subprocess,
    "distributed/test_pg_wrapper": run_test_with_subprocess,
    "distributed/rpc/test_faulty_agent": run_test_with_subprocess,
    "distributed/rpc/test_tensorpipe_agent": run_test_with_subprocess,
    "distributed/rpc/test_share_memory": run_test_with_subprocess,
    "distributed/rpc/cuda/test_tensorpipe_agent": run_test_with_subprocess,
    "functorch/test_control_flow_cuda_initialization": run_test_with_subprocess,
    "doctests": run_doctests,
    "test_ci_sanity_check_fail": run_ci_sanity_check,
    "test_autoload_enable": test_autoload_enable,
    "test_autoload_disable": test_autoload_disable,
    "test_openreg": test_openreg,
}


PYTEST_SKIP_RETRIES = {"test_public_bindings"}


def parse_args():
    parser = argparse.ArgumentParser(
        description="Run the PyTorch unit test suite",
        epilog="where TESTS is any of: {}".format(", ".join(TESTS)),
        formatter_class=argparse.RawTextHelpFormatter,
    )
    parser.add_argument(
        "-v",
        "--verbose",
        action="count",
        default=0,
        help="Print verbose information and test-by-test results",
    )
    parser.add_argument(
        "--showlocals",
        action=argparse.BooleanOptionalAction,
        default=strtobool(os.environ.get("TEST_SHOWLOCALS", "False")),
        help="Show local variables in tracebacks (default: True)",
    )
    parser.add_argument("--jit", "--jit", action="store_true", help="run all jit tests")
    parser.add_argument(
        "--distributed-tests",
        "--distributed-tests",
        action="store_true",
        help="Run all distributed tests",
    )
    parser.add_argument(
        "--multigpu-filter",
        choices=["multigpu", "not-multigpu"],
        default=None,
        help="Restrict distributed tests by the auto-applied `multigpu` marker "
        "(see test/conftest.py). `multigpu` runs only tests that need multiple "
        "GPUs; `not-multigpu` runs only single-GPU "
        "tests, which can run on a single-GPU runner. Combined (AND) with the "
        "existing serial/not-serial split.",
    )
    parser.add_argument(
        "--include-cpython-tests",
        "--include-cpython-tests",
        action="store_true",
        help="If this flag is present, we will only run cpython tests.",
    )
    parser.add_argument(
        "--include-dynamo-core-tests",
        "--include-dynamo-core-tests",
        action="store_true",
        help=(
            "If this flag is present, we will only run dynamo tests. "
            "If this flag is not present, we will run all tests "
            "(including dynamo tests)."
        ),
    )
    parser.add_argument(
        "--include-inductor-core-tests",
        "--include-inductor-core-tests",
        action="store_true",
        help=(
            "If this flag is present, we will only run inductor tests. "
            "If this flag is not present, we will run all tests "
            "(including inductor tests)."
        ),
    )
    parser.add_argument(
        "--functorch",
        "--functorch",
        action="store_true",
        help=(
            "If this flag is present, we will only run functorch tests. "
            "If this flag is not present, we will run all tests "
            "(including functorch tests)."
        ),
    )
    parser.add_argument(
        "--einops",
        "--einops",
        action="store_true",
        help=(
            "If this flag is present, we will only run einops tests. "
            "If this flag is not present, we will run all tests "
            "(including einops tests)."
        ),
    )
    parser.add_argument(
        "--mps",
        "--mps",
        action="store_true",
        help=(
            "If this flag is present, we will only run subset of tests, such as test_mps, test_nn, ..."
        ),
    )
    parser.add_argument(
        "--xpu",
        "--xpu",
        action="store_true",
        help=("If this flag is present, we will run xpu tests except XPU_BLOCK_LIST"),
    )
    parser.add_argument(
        "--openreg",
        "--openreg",
        action="store_true",
        help=("If this flag is present, we will only run test_openreg"),
    )
    parser.add_argument(
        "--cpp",
        "--cpp",
        action="store_true",
        help=("If this flag is present, we will only run C++ tests"),
    )
    parser.add_argument(
        "-core",
        "--core",
        action="store_true",
        help="Only run core tests, or tests that validate PyTorch's ops, modules,"
        "and autograd. They are defined by CORE_TEST_LIST.",
    )
    parser.add_argument(
        "--onnx",
        "--onnx",
        action="store_true",
        help=(
            "Only run ONNX tests, or tests that validate PyTorch's ONNX export. "
            "If this flag is not present, we will exclude ONNX tests."
        ),
    )
    parser.add_argument(
        "-k",
        "--pytest-k-expr",
        default="",
        help="Pass to pytest as its -k expr argument",
    )
    parser.add_argument(
        "-c",
        "--coverage",
        action="store_true",
        help="enable coverage",
        default=PYTORCH_COLLECT_COVERAGE,
    )
    parser.add_argument(
        "-i",
        "--include",
        nargs="+",
        choices=TestChoices(TESTS),
        default=TESTS,
        metavar="TESTS",
        help="select a set of tests to include (defaults to ALL tests)."
        " tests must be a part of the TESTS list defined in run_test.py",
    )
    parser.add_argument(
        "--hw-classification",
        nargs="+",
        choices=_HC_CHOICES,
        type=str.upper,
        default=None,
        metavar="SCOPE",
        help="filter tests by hardware classification categories (e.g., GENERIC ACCELERATOR CPU CUDA MPS XPU)",
    )
    parser.add_argument(
        "-x",
        "--exclude",
        nargs="+",
        choices=TESTS,
        metavar="TESTS",
        default=[],
        help="select a set of tests to exclude",
    )
    parser.add_argument(
        "--ignore-win-blocklist",
        action="store_true",
        help="always run blocklisted windows tests",
    )
    # NS: Disable target determination until it can be made more reliable
    # parser.add_argument(
    #     "--determine-from",
    #     help="File of affected source filenames to determine which tests to run.",
    # )
    parser.add_argument(
        "--continue-through-error",
        "--keep-going",
        action="store_true",
        help="Runs the full test suite despite one of the tests failing",
        default=strtobool(os.environ.get("CONTINUE_THROUGH_ERROR", "False")),
    )
    parser.add_argument(
        "--pipe-logs",
        action="store_true",
        help="Print logs to output file while running tests.  True if in CI and env var is not set",
        default=IS_CI and not strtobool(os.environ.get("VERBOSE_TEST_LOGS", "False")),
    )
    parser.add_argument(
        "--enable-timeout",
        action="store_true",
        help="Set a timeout based on the test times json file.  Only works if there are test times available",
        default=IS_CI and not strtobool(os.environ.get("NO_TEST_TIMEOUT", "False")),
    )
    GITHUB_WORKFLOW = os.environ.get("GITHUB_WORKFLOW", "slow")
    parser.add_argument(
        "--enable-td",
        action="store_true",
        help="Enables removing tests based on TD",
        default=IS_CI
        and get_pr_number() is not None
        and not strtobool(os.environ.get("NO_TD", "False"))
        and not IS_MACOS
        and "xpu" not in BUILD_ENVIRONMENT
        and "onnx" not in BUILD_ENVIRONMENT
        and (
            GITHUB_WORKFLOW in ("trunk", "pull")
            or GITHUB_WORKFLOW.startswith(("rocm-", "periodic-rocm-"))
        ),
    )
    parser.add_argument(
        "--shard",
        nargs=2,
        type=int,
        help="runs a shard of the tests (taking into account other selections), e.g., "
        "--shard 2 3 will break up the selected tests into 3 shards and run the tests "
        "in the 2nd shard (the first number should not exceed the second)",
    )
    parser.add_argument(
        "--exclude-jit-executor",
        action="store_true",
        help="exclude tests that are run for a specific jit config",
    )
    parser.add_argument(
        "--exclude-torch-export-tests",
        action="store_true",
        help="exclude torch export tests",
    )
    parser.add_argument(
        "--exclude-aot-dispatch-tests",
        action="store_true",
        help="exclude aot dispatch tests",
    )
    parser.add_argument(
        "--exclude-distributed-tests",
        action="store_true",
        help="exclude distributed tests",
    )
    parser.add_argument(
        "--exclude-inductor-tests",
        action="store_true",
        help="exclude inductor tests",
    )
    parser.add_argument(
        "--exclude-quantization-tests",
        action="store_true",
        help="exclude quantization tests",
    )
    parser.add_argument(
        "--dry-run",
        action="store_true",
        help="Only list the test that will run.",
    )
    parser.add_argument(
        "--xdoctest-command",
        default="all",
        help=(
            "Control the specific doctest action. "
            "Use 'list' to simply parse doctests and check syntax. "
            "Use 'all' to execute all doctests or specify a specific "
            "doctest to run"
        ),
    )
    parser.add_argument(
        "--no-translation-validation",
        action="store_false",
        help="Run tests without translation validation.",
    )
    parser.add_argument(
        "--upload-artifacts-while-running",
        action="store_true",
        default=IS_CI,
    )

    group = parser.add_mutually_exclusive_group()
    group.add_argument(
        "--dynamo",
        action="store_true",
        help="Run tests with TorchDynamo+EagerBackend turned on",
    )
    group.add_argument(
        "--inductor",
        action="store_true",
        help="Run tests with TorchInductor turned on",
    )

    args, extra = parser.parse_known_args()
    if "--" in extra:
        extra.remove("--")
    args.additional_args = extra
    return args


def exclude_tests(
    exclude_list, selected_tests, exclude_message=None, exact_match=False
):
    for exclude_test in exclude_list:
        tests_copy = selected_tests[:]
        for test in tests_copy:
            if (
                not exact_match and test.startswith(exclude_test)
            ) or test == exclude_test:
                if exclude_message is not None:
                    print_to_stderr(f"Excluding {test} {exclude_message}")
                selected_tests.remove(test)
    return selected_tests


def must_serial(file: str | ShardedTest) -> bool:
    if isinstance(file, ShardedTest):
        file = file.name
    return (
        os.getenv("PYTORCH_TEST_RUN_EVERYTHING_IN_SERIAL", "0") == "1"
        or DISTRIBUTED_TEST_PREFIX in os.getenv("TEST_CONFIG", "")
        or DISTRIBUTED_TEST_PREFIX in file
        or file in CUSTOM_HANDLERS
        or file in RUN_PARALLEL_BLOCKLIST
        or file in CI_SERIAL_LIST
        or file in JIT_EXECUTOR_TESTS
        or file in ONNX_SERIAL_LIST
        or NUM_PROCS == 1
    )


def can_run_in_pytest(test):
    return os.getenv("PYTORCH_TEST_DO_NOT_USE_PYTEST", "0") == "0"


def get_selected_tests(options) -> list[str]:
    selected_tests = options.include

    # filter if there's JIT only and distributed only test options
    if options.jit:
        selected_tests = list(
            filter(lambda test_name: "jit" in test_name, selected_tests)
        )

    if options.distributed_tests:
        selected_tests = list(
            filter(lambda test_name: test_name in DISTRIBUTED_TESTS, selected_tests)
        )

    # Filter to only run core tests when --core option is specified
    if options.core:
        selected_tests = list(
            filter(lambda test_name: test_name in CORE_TEST_LIST, selected_tests)
        )

    if options.include_cpython_tests:
        selected_tests = list(
            filter(lambda test_name: test_name in CPYTHON_TESTS, selected_tests)
        )

    # Filter to only run dynamo tests when --include-dynamo-core-tests option is specified
    if options.include_dynamo_core_tests:
        selected_tests = list(
            filter(lambda test_name: test_name in DYNAMO_CORE_TESTS, selected_tests)
        )

    # Filter to only run dynamo tests when --include-inductor-core-tests option is specified
    if options.include_inductor_core_tests:
        selected_tests = list(
            filter(lambda test_name: test_name in INDUCTOR_TESTS, selected_tests)
        )

    # Filter to only run functorch tests when --functorch option is specified
    if options.functorch:
        selected_tests = list(
            filter(lambda test_name: test_name in FUNCTORCH_TESTS, selected_tests)
        )

    # Filter to only run einops tests when --einops option is specified
    if options.einops:
        selected_tests = list(
            filter(
                lambda test_name: test_name.startswith("dynamo/test_einops"),
                selected_tests,
            )
        )

    if options.cpp:
        selected_tests = list(
            filter(lambda test_name: test_name in CPP_TESTS, selected_tests)
        )
    else:
        # Exclude all C++ tests otherwise as they are still handled differently
        # than Python test at the moment
        options.exclude.extend(CPP_TESTS)

    if options.mps:
        os.environ["PYTORCH_TESTING_DEVICE_ONLY_FOR"] = "mps"
        selected_tests = [
            "test_ops",
            "test_mps",
            "test_metal",
            "test_modules",
            "nn/test_convolution",
            "nn/test_dropout",
            "nn/test_pooling",
            "test_view_ops",
            "test_nn",
            "distributions/test_distributions",
            "inductor/test_mps_basic",
            "inductor/test_torchinductor",
            "inductor/test_aot_inductor",
            "inductor/test_torchinductor_dynamic_shapes",
        ]
    else:
        # Exclude mps-only tests otherwise
        options.exclude.extend(["test_mps", "test_metal"])

    if options.xpu:
        selected_tests = exclude_tests(XPU_BLOCKLIST, selected_tests, "on XPU")
    else:
        # Exclude all xpu specific tests otherwise
        options.exclude.extend(XPU_TEST)

    if options.openreg:
        selected_tests = ["test_openreg"]
    else:
        options.exclude.append("test_openreg")

    # Filter to only run onnx tests when --onnx option is specified
    onnx_tests = [tname for tname in selected_tests if tname in ONNX_TESTS]
    if options.onnx:
        selected_tests = onnx_tests
    else:
        # Exclude all onnx tests otherwise
        options.exclude.extend(onnx_tests)

    # process exclusion
    if options.exclude_jit_executor:
        options.exclude.extend(JIT_EXECUTOR_TESTS)

    if options.exclude_distributed_tests:
        options.exclude.extend(DISTRIBUTED_TESTS)

    if options.exclude_inductor_tests:
        options.exclude.extend(INDUCTOR_TESTS)

    if options.exclude_torch_export_tests:
        options.exclude.extend(TORCH_EXPORT_TESTS)

    if options.exclude_aot_dispatch_tests:
        options.exclude.extend(AOT_DISPATCH_TESTS)

    if options.exclude_quantization_tests:
        options.exclude.extend(QUANTIZATION_TESTS)

    # these tests failing in CUDA 11.6 temporary disabling. issue https://github.com/pytorch/pytorch/issues/75375
    if torch.version.cuda is not None:
        options.exclude.extend(["distributions/test_constraints"])

    # these tests failing in Python 3.12 temporarily disabling
    if sys.version_info >= (3, 12):
        options.exclude.extend(
            [
                "functorch/test_dims",
                "functorch/test_rearrange",
                "functorch/test_parsing",
                "functorch/test_memory_efficient_fusion",
                "torch_np/numpy_tests/core/test_multiarray",
            ]
        )

    # Only include cpython tests which match the current python version
    current_cpython_prefix = (
        f"cpython/v{sys.version_info.major}_{sys.version_info.minor}/"
    )
    options.exclude.extend(
        [
            test
            for test in selected_tests
            if test.startswith("cpython/")
            and not test.startswith(current_cpython_prefix)
        ]
    )

    selected_tests = exclude_tests(options.exclude, selected_tests)

    if IS_WINDOWS and not options.ignore_win_blocklist:
        from torch.testing._internal.common_cuda import SM120OrLater, SM89OrLater

        # Disable tests on Windows for SM89 and later - tests failing in ci
        # Enable tests after fixing the failures
        if SM89OrLater:
            WINDOWS_BLOCKLIST.extend(
                [
                    # Windows fatal exception / access violation
                    "functorch/test_aotdispatch",
                    "functorch/test_control_flow",
                    "nn/test_convolution",
                    "profiler/test_profiler",
                    "test_modules",
                    "test_expanded_weights",
                    "test_jit",
                    "test_nested_tensor",
                    "test_nestedtensor",
                    "test_nn",
                    # DLL load failed errors, missing dependencies
                    "test_custom_ops",
                    "test_testing",
                    # Features not supported on Windows ( e.g. rowwise scaling)
                    "test_decomp",
                    "test_transformers",
                    "test_ops",
                    # Output mismatch errors and long running tests
                    "test_linalg",
                    "test_matmul_cuda",
                    "functorch/test_ops",
                    "test_scaled_matmul_cuda",
                ]
            )

        # Disable tests on Windows for SM120 and later - tests failing in ci
        # Enable tests after fixing the failures
        if SM120OrLater:
            WINDOWS_BLOCKLIST.extend(
                [
                    # test_api fails on Windows SM120+. Triage pending.
                    "cpp/test_api",
                ]
            )

        target_arch = os.environ.get("VSCMD_ARG_TGT_ARCH")
        if target_arch != "x64":
            WINDOWS_BLOCKLIST.append("cpp_extensions_aot_no_ninja")
            WINDOWS_BLOCKLIST.append("cpp_extensions_aot_ninja")
            WINDOWS_BLOCKLIST.append("cpp_extensions_jit")
            WINDOWS_BLOCKLIST.append("jit")
            WINDOWS_BLOCKLIST.append("jit_fuser")

        selected_tests = exclude_tests(WINDOWS_BLOCKLIST, selected_tests, "on Windows")

    elif TEST_WITH_ROCM:
        selected_tests = exclude_tests(ROCM_BLOCKLIST, selected_tests, "on ROCm")

    elif IS_S390X:
        selected_tests = exclude_tests(S390X_BLOCKLIST, selected_tests, "on s390x")
        selected_tests = exclude_tests(
            DISTRIBUTED_TESTS,
            selected_tests,
            "Skip distributed tests on s390x",
        )

    # skip all distributed tests if distributed package is not available.
    if not dist.is_available():
        selected_tests = exclude_tests(
            DISTRIBUTED_TESTS,
            selected_tests,
            "PyTorch is built without distributed support.",
        )

    # skip tests that require LAPACK when it's not available
    if not torch._C.has_lapack:
        selected_tests = exclude_tests(
            TESTS_REQUIRING_LAPACK,
            selected_tests,
            "PyTorch is built without LAPACK support.",
        )

    if TEST_WITH_SLOW_GRADCHECK:
        # Avoid running files that don't use gradcheck. See TESTS_USING_GRADCHECK.
        selected_tests = [
            test for test in selected_tests if test in TESTS_USING_GRADCHECK
        ]

    selected_tests = [parse_test_module(x) for x in selected_tests]
    return selected_tests


def load_test_times_from_file(file: str) -> dict[str, Any]:
    # Load previous test times to make sharding decisions
    path = os.path.join(str(REPO_ROOT), file)
    if not os.path.exists(path):
        print_to_stderr(
            f"::warning:: Failed to find test times file `{path}`. Using round robin sharding."
        )
        return {}

    with open(path) as f:
        test_times_file = cast(dict[str, Any], json.load(f))
    raw_job_name = os.environ.get("JOB_NAME")
    build_env = os.environ.get("BUILD_ENVIRONMENT")
    job_name = raw_job_name
    if job_name is None or job_name == "":
        # If job name isn't available, use build environment as a backup
        job_name = build_env
    else:
        job_name = get_job_base_name(job_name)
    test_config = os.environ.get("TEST_CONFIG")
    print_to_stderr(f"JOB_NAME={raw_job_name}")
    print_to_stderr(f"BUILD_ENVIRONMENT={build_env}")
    print_to_stderr(f"test-times lookup key={job_name}, test_config={test_config}")
    if test_config in test_times_file.get(job_name, {}):
        print_to_stderr("Found test times from artifacts")
        return test_times_file[job_name][test_config]
    elif test_config in test_times_file["default"]:
        print_to_stderr(
            f"::warning:: Gathered no stats from artifacts for {job_name} build env"
            f" and {test_config} test config. Using default job name and {test_config} test config instead."
        )
        return test_times_file["default"][test_config]
    else:
        print_to_stderr(
            f"::warning:: Gathered no stats from artifacts for job name {job_name} build env"
            f" and {test_config} test config. Using default job name and default test config instead."
        )
        return test_times_file["default"]["default"]


def load_test_file_times(
    file: str = ADDITIONAL_CI_FILES_FOLDER / TEST_TIMES_FILE,
) -> dict[str, float]:
    return cast(dict[str, float], load_test_times_from_file(file))


def load_test_class_times(
    file: str = ADDITIONAL_CI_FILES_FOLDER / TEST_CLASS_TIMES_FILE,
) -> dict[str, dict[str, float]]:
    return cast(dict[str, dict[str, float]], load_test_times_from_file(file))


def get_sharding_opts(options) -> tuple[int, int]:
    which_shard, num_shards = 1, 1
    if options.shard:
        if len(options.shard) != 2:
            raise AssertionError("Unexpected shard format")
        if min(options.shard) <= 0:
            raise AssertionError("Shards must be positive numbers")
        which_shard, num_shards = options.shard
        if which_shard > num_shards:
            raise AssertionError(
                "Selected shard must be less than or equal to total number of shards"
            )

    return (which_shard, num_shards)


def do_sharding(
    options,
    selected_tests: Sequence[TestRun],
    test_file_times: dict[str, float],
    test_class_times: dict[str, dict[str, float]],
    sort_by_time: bool = True,
) -> tuple[float, list[ShardedTest]]:
    which_shard, num_shards = get_sharding_opts(options)

    # Do sharding
    shards = calculate_shards(
        num_shards,
        selected_tests,
        test_file_times,
        test_class_times=test_class_times,
        must_serial=must_serial,
        sort_by_time=sort_by_time,
    )
    return shards[which_shard - 1]


class TestFailure(NamedTuple):
    test: TestRun
    message: str


def run_test_module(
    test: ShardedTest, test_directory: str, options
) -> TestFailure | None:
    try:
        maybe_set_hip_visible_devies()

        test_name = test.name

        # Printing the date here can help diagnose which tests are slow
        start = time.perf_counter()
        print_to_stderr(f"Running {str(test)} ... [{datetime.now()}][{start}]")
        handler = CUSTOM_HANDLERS.get(test_name, run_test)
        return_code = handler(test, test_directory, options)
        end = time.perf_counter()
        print_to_stderr(
            f"Finished {str(test)} ... [{datetime.now()}][{end}], took {(end - start) / 60:.2f}min"
        )
        if not isinstance(return_code, int) or isinstance(return_code, bool):
            raise AssertionError(
                f"While running {str(test)} got non integer return code {return_code}"
            )
        if return_code == 0:
            return None

        message = f"{str(test)} failed!"
        if return_code < 0:
            # subprocess.Popen returns the child process' exit signal as
            # return code -N, where N is the signal number.
            signal_name = SIGNALS_TO_NAMES_DICT[-return_code]
            message += f" Received signal: {signal_name}"
        return TestFailure(test.test, message)
    except Exception as e:
        return TestFailure(test.test, f"{str(test)} failed! {e}")


def run_tests(
    selected_tests: list[ShardedTest],
    test_directory: str,
    options,
    failures: list[TestFailure],
) -> None:
    if len(selected_tests) == 0:
        return

    # parallel = in parallel with other files
    # serial = this file on it's own.  The file might still be run in parallel with itself (ex test_ops)
    selected_tests_parallel = [x for x in selected_tests if not must_serial(x)]
    selected_tests_serial = [
        x for x in selected_tests if x not in selected_tests_parallel
    ]

    # The multigpu marker (see test/conftest.py) is orthogonal to serial: it
    # partitions distributed tests by whether they spawn multiple processes /
    # need multiple GPUs. AND it into whatever serial expression a pass uses so
    # a single-GPU config can select `not multigpu` without dropping the
    # serial/not-serial split (a bare second `-m` would clobber the first).
    multigpu_marker = {
        "multigpu": "multigpu",
        "not-multigpu": "not multigpu",
    }.get(getattr(options, "multigpu_filter", None))

    def marker_args(serial_expr: str | None) -> list[str]:
        exprs = [e for e in (serial_expr, multigpu_marker) if e]
        if not exprs:
            return []
        return ["-m", " and ".join(f"({e})" for e in exprs)]

    # NB: This is a hack to make conftest.py and files it depends on available
    # on CPP_TESTS_DIR. We should see if the file could be turned into a
    # full-fledge ptest plugin instead
    conftest_files = [
        "conftest.py",
        "pytest_shard_custom.py",
    ]
    for conftest_file in conftest_files:
        cpp_file = os.path.join(CPP_TESTS_DIR, conftest_file)
        if (
            options.cpp
            and os.path.exists(CPP_TESTS_DIR)
            and os.path.isdir(CPP_TESTS_DIR)
            and not os.path.exists(cpp_file)
        ):
            shutil.copy(os.path.join(test_directory, conftest_file), cpp_file)

    def handle_complete(failure: TestFailure | None):
        failed = failure is not None
        if IS_CI and options.upload_artifacts_while_running:
            parse_xml_and_upload_json()
            zip_and_upload_artifacts(failed)
        if not failed:
            return False
        failures.append(failure)
        print_to_stderr(failure.message)
        return True

    keep_going_message = (
        "\n\nTip: You can keep running tests even on failure by passing --keep-going to run_test.py.\n"
        "If running on CI, add the 'keep-going' label to your PR and rerun your jobs."
    )

    pool = None
    try:
        for test in selected_tests_serial:
            options_clone = copy.deepcopy(options)
            if can_run_in_pytest(test):
                options_clone.pytest = True
            options_clone.additional_args.extend(marker_args(None))
            failure = run_test_module(test, test_directory, options_clone)
            test_failed = handle_complete(failure)
            if (
                test_failed
                and not options.continue_through_error
                and not RERUN_DISABLED_TESTS
            ):
                raise RuntimeError(failure.message + keep_going_message)

        # Run tests marked as serial first
        for test in selected_tests_parallel:
            options_clone = copy.deepcopy(options)
            if can_run_in_pytest(test):
                options_clone.pytest = True
            options_clone.additional_args.extend(marker_args("serial"))
            failure = run_test_module(test, test_directory, options_clone)
            test_failed = handle_complete(failure)
            if (
                test_failed
                and not options.continue_through_error
                and not RERUN_DISABLED_TESTS
            ):
                raise RuntimeError(failure.message + keep_going_message)

        # This is used later to constrain memory per proc on the GPU. On ROCm
        # the number of procs is the number of GPUs, so we don't need to do this
        os.environ["NUM_PARALLEL_PROCS"] = str(1 if torch.version.hip else NUM_PROCS)

        # See Note [ROCm parallel CI testing]
        pool = get_context("spawn").Pool(
            NUM_PROCS, maxtasksperchild=None if torch.version.hip else 1
        )

        def parallel_test_completion_callback(failure):
            test_failed = handle_complete(failure)
            if (
                test_failed
                and not options.continue_through_error
                and not RERUN_DISABLED_TESTS
            ):
                pool.terminate()

        for test in selected_tests_parallel:
            options_clone = copy.deepcopy(options)
            if can_run_in_pytest(test):
                options_clone.pytest = True
            options_clone.additional_args.extend(marker_args("not serial"))
            pool.apply_async(
                run_test_module,
                args=(test, test_directory, options_clone),
                callback=parallel_test_completion_callback,
            )
        pool.close()
        pool.join()
        del os.environ["NUM_PARALLEL_PROCS"]

    finally:
        if pool:
            pool.terminate()
            pool.join()

    return


def check_pip_packages() -> None:
    packages = [
        "pytest-rerunfailures",
        "pytest-flakefinder",
        "pytest-xdist",
    ]
    try:
        for pkg in packages:
            version(pkg)
    except PackageNotFoundError:
        print_to_stderr(
            f"Missing pip dependency: {pkg}, please run `pip install -r .ci/docker/requirements-ci.txt`"
        )
        sys.exit(1)


def main():
    check_pip_packages()

    options = parse_args()
    tests_to_include_env = os.environ.get("TESTS_TO_INCLUDE", "").strip()
    if tests_to_include_env:
        # Parse env var tests to module names (strips .py suffix and ::method)
        env_tests = {parse_test_module(t) for t in tests_to_include_env.split()}

        if options.include != TESTS:
            # --include was explicitly provided, intersect with env var
            cli_tests = {parse_test_module(t) for t in options.include}
            options.include = list(env_tests & cli_tests)
        else:
            # No explicit --include, use env var tests
            options.include = list(env_tests)

        options.enable_td = False

    # Include sharding info in all metrics
    which_shard, num_shards = get_sharding_opts(options)
    add_global_metric("shard", which_shard)
    add_global_metric("num_shards", num_shards)

    test_directory = str(REPO_ROOT / "test")
    selected_tests = get_selected_tests(options)

    test_prioritizations = import_results()
    if len(test_prioritizations.get_all_tests()) == 0:
        options.enable_td = False
    test_prioritizations.amend_tests(selected_tests)

    os.makedirs(REPO_ROOT / "test" / "test-reports", exist_ok=True)

    if options.coverage and not PYTORCH_COLLECT_COVERAGE:
        shell(["coverage", "erase"])

    if IS_CI:
        # downloading test cases configuration to local environment
        get_test_case_configs(dirpath=test_directory)

    test_file_times_dict = load_test_file_times()
    test_class_times_dict = load_test_class_times()

    class TestBatch:
        """Defines a set of tests with similar priority that should be run together on the current shard"""

        name: str
        sharded_tests: list[ShardedTest]
        failures: list[TestFailure]

        def __init__(
            self, name: str, raw_tests: Sequence[TestRun], should_sort_shard: bool
        ):
            self.name = name
            self.failures = []
            self.time, self.sharded_tests = do_sharding(
                options,
                raw_tests,
                test_file_times_dict,
                test_class_times_dict,
                sort_by_time=should_sort_shard,
            )

        def __str__(self):
            s = f"Name: {self.name} (est. time: {round(self.time / 60, 2)}min)\n"
            serial = [test for test in self.sharded_tests if must_serial(test)]
            parallel = [test for test in self.sharded_tests if not must_serial(test)]
            s += f"  Serial tests ({len(serial)}):\n"
            s += "".join(f"    {test}\n" for test in serial)
            s += f"  Parallel tests ({len(parallel)}):\n"
            s += "".join(f"    {test}\n" for test in parallel)
            return s.strip()

    percent_to_run = 25 if options.enable_td else 100
    print_to_stderr(
        f"Running {percent_to_run}% of tests based on TD"
        if options.enable_td
        else "Running all tests"
    )
    include, exclude = test_prioritizations.get_top_per_tests(percent_to_run)

    test_batch = TestBatch("tests to run", include, False)
    test_batch_exclude = TestBatch("excluded", exclude, True)
    if IS_CI:
        gen_ci_artifact([x.to_json() for x in include], [x.to_json() for x in exclude])

    print_to_stderr(f"Running parallel tests on {NUM_PROCS} processes")
    print_to_stderr(test_batch)
    print_to_stderr(test_batch_exclude)

    if options.dry_run:
        return

    if options.dynamo:
        os.environ["PYTORCH_TEST_WITH_DYNAMO"] = "1"

    elif options.inductor:
        os.environ["PYTORCH_TEST_WITH_INDUCTOR"] = "1"

    if not options.no_translation_validation:
        os.environ["PYTORCH_TEST_WITH_TV"] = "1"

    try:
        # Actually run the tests
        start_time = time.time()
        run_tests(
            test_batch.sharded_tests, test_directory, options, test_batch.failures
        )
        elapsed_time = time.time() - start_time
        print_to_stderr(
            f"Running test batch '{test_batch.name}' cost {round(elapsed_time, 2)} seconds"
        )

    finally:
        if options.coverage:
            from coverage import Coverage

            with set_cwd(test_directory):
                cov = Coverage()
                if PYTORCH_COLLECT_COVERAGE:
                    cov.load()
                cov.combine(strict=False)
                cov.save()
                if not PYTORCH_COLLECT_COVERAGE:
                    cov.html_report()

        all_failures = test_batch.failures

        if IS_CI:
            for test, _ in all_failures:
                test_stats = test_prioritizations.get_test_stats(test)
                print_to_stderr("Emitting td_test_failure_stats_v2")
                emit_metric(
                    "td_test_failure_stats_v2",
                    {
                        "selected_tests": selected_tests,
                        "failure": str(test),
                        **test_stats,
                    },
                )
            gen_additional_test_failures_file(
                [test.test_file for test, _ in all_failures]
            )

    if len(all_failures):
        for _, err in all_failures:
            print_to_stderr(err)

        # A disabled test is expected to fail, so there is no need to report a failure here
        if not RERUN_DISABLED_TESTS:
            sys.exit(1)


if __name__ == "__main__":
    main()