[GR-64574] Bytecode DSL: add guest VM config for mx benchmark.

steve-s · steve-s · commit f264067e8b00 · 2025-04-28T08:00:47.000Z
PullRequest: graalpython/3769
diff --git a/ci.jsonnet b/ci.jsonnet
@@ -1 +1 @@
-{ "overlay": "a242e9622be463c91c2cc034a1f2c1811316bc18" }
+{ "overlay": "ae41f079f9ce46ca5f0ce5a2443c27cab856e403" }
diff --git a/graalpython/com.oracle.graal.python.test/src/tests/unittest_tags_bytecode_dsl/test_zipimport_support.txt b/graalpython/com.oracle.graal.python.test/src/tests/unittest_tags_bytecode_dsl/test_zipimport_support.txt
@@ -1,4 +1,3 @@
-test.test_zipimport_support.ZipSupportTests.test_doctest_issue4197 @ linux-x86_64
 test.test_zipimport_support.ZipSupportTests.test_doctest_main_issue4197 @ linux-x86_64
 test.test_zipimport_support.ZipSupportTests.test_inspect_getsource_issue4223 @ linux-x86_64
 test.test_zipimport_support.ZipSupportTests.test_pdb_issue4201 @ linux-x86_64
diff --git a/graalpython/com.oracle.graal.python/src/com/oracle/graal/python/runtime/PythonContext.java b/graalpython/com.oracle.graal.python/src/com/oracle/graal/python/runtime/PythonContext.java
@@ -1627,6 +1627,7 @@ public void initialize() {
             }
             releaseGil();
         }
+        PythonOptions.checkBytecodeDSLEnv();
     }
 
     public void resetPerfCounter() {
@@ -1648,6 +1649,7 @@ public void patch(Env newEnv) {
         } finally {
             releaseGil();
         }
+        PythonOptions.checkBytecodeDSLEnv();
     }
 
     private void importSiteIfForced() {
diff --git a/graalpython/com.oracle.graal.python/src/com/oracle/graal/python/runtime/PythonOptions.java b/graalpython/com.oracle.graal.python/src/com/oracle/graal/python/runtime/PythonOptions.java
@@ -93,6 +93,12 @@ private PythonOptions() {
         // no instances
     }
 
+    public static void checkBytecodeDSLEnv() {
+        if (!ENABLE_BYTECODE_DSL_INTERPRETER && "true".equalsIgnoreCase(System.getenv("BYTECODE_DSL_INTERPRETER"))) {
+            System.err.println("WARNING: found environment variable BYTECODE_DSL_INTERPRETER=true, but the python.EnableBytecodeDSLInterpreter Java property was not set.");
+        }
+    }
+
     @Option(category = OptionCategory.EXPERT, help = "Set the home of Python. Equivalent of GRAAL_PYTHONHOME env variable. " +
                     "Determines default values for the CoreHome, StdLibHome, SysBasePrefix, SysPrefix.", usageSyntax = "<path>", stability = OptionStability.STABLE) //
     public static final OptionKey<String> PythonHome = new OptionKey<>("");
diff --git a/mx.graalpython/mx_graalpython.py b/mx.graalpython/mx_graalpython.py
@@ -41,6 +41,7 @@
 from pathlib import Path
 from textwrap import dedent
 
+import mx_graalpython_benchmark
 import mx_graalpython_gradleproject
 import mx_urlrewrites
 
@@ -55,7 +56,6 @@
 
 import mx
 import mx_util
-import mx_benchmark
 import mx_gate
 import mx_native
 import mx_unittest
@@ -71,20 +71,8 @@
 from mx_graalpython_gradleproject import GradlePluginProject #pylint: disable=unused-import
 
 from mx_gate import Task
-from mx_graalpython_bench_param import PATH_MESO, BENCHMARKS, WARMUP_BENCHMARKS, JAVA_DRIVER_BENCHMARKS, \
-    HEAP_BENCHMARKS
-from mx_graalpython_benchmark import PythonBenchmarkSuite, python_vm_registry, CPythonVm, PyPyVm, JythonVm, \
-    GraalPythonVm, \
-    CONFIGURATION_DEFAULT, CONFIGURATION_SANDBOXED, CONFIGURATION_NATIVE, \
-    CONFIGURATION_DEFAULT_MULTI, CONFIGURATION_SANDBOXED_MULTI, CONFIGURATION_NATIVE_MULTI, \
-    CONFIGURATION_DEFAULT_MULTI_TIER, CONFIGURATION_NATIVE_MULTI_TIER, \
-    PythonVmWarmupBenchmarkSuite, \
-    CONFIGURATION_INTERPRETER, CONFIGURATION_INTERPRETER_MULTI, CONFIGURATION_NATIVE_INTERPRETER, \
-    CONFIGURATION_NATIVE_INTERPRETER_MULTI, PythonJavaEmbeddingBenchmarkSuite, python_java_embedding_vm_registry, \
-    GraalPythonJavaDriverVm, CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI_SHARED, \
-    CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI, CONFIGURATION_JAVA_EMBEDDING_MULTI_SHARED, \
-    CONFIGURATION_JAVA_EMBEDDING_MULTI, CONFIGURATION_PANAMA, PythonJMHDistMxBenchmarkSuite, \
-    PythonHeapBenchmarkSuite
+from mx_graalpython_bench_param import PATH_MESO
+
 
 if not sys.modules.get("__main__"):
     # workaround for pdb++
@@ -2197,76 +2185,6 @@ def bytecode_dsl_build_args():
 # post init
 #
 # ----------------------------------------------------------------------------------------------------------------------
-def _register_vms(namespace):
-    # cpython
-    python_vm_registry.add_vm(CPythonVm(config_name=CONFIGURATION_DEFAULT), SUITE)
-
-    # pypy
-    python_vm_registry.add_vm(PyPyVm(config_name=CONFIGURATION_DEFAULT), SUITE)
-
-    # jython
-    python_vm_registry.add_vm(JythonVm(config_name=CONFIGURATION_DEFAULT), SUITE)
-
-    # graalpython
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_DEFAULT), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_INTERPRETER, extra_polyglot_args=[
-        '--experimental-options', '--engine.Compilation=false'
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_DEFAULT_MULTI, extra_polyglot_args=[
-        '--experimental-options', '-multi-context',
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_INTERPRETER_MULTI, extra_polyglot_args=[
-        '--experimental-options', '-multi-context', '--engine.Compilation=false'
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_DEFAULT_MULTI_TIER, extra_polyglot_args=[
-        '--experimental-options', '--engine.MultiTier=true',
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_SANDBOXED, extra_polyglot_args=SANDBOXED_OPTIONS), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_NATIVE, extra_polyglot_args=[
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_NATIVE_INTERPRETER, extra_polyglot_args=[
-        '--experimental-options', '--engine.Compilation=false']), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_SANDBOXED_MULTI, extra_polyglot_args=[
-        '--experimental-options', '-multi-context'] + SANDBOXED_OPTIONS), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_NATIVE_MULTI, extra_polyglot_args=[
-        '--experimental-options', '-multi-context'
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_NATIVE_INTERPRETER_MULTI, extra_polyglot_args=[
-        '--experimental-options', '-multi-context', '--engine.Compilation=false'
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_NATIVE_MULTI_TIER, extra_polyglot_args=[
-        '--experimental-options', '--engine.MultiTier=true'
-    ]), SUITE, 10)
-    python_vm_registry.add_vm(GraalPythonVm(config_name=CONFIGURATION_PANAMA, extra_polyglot_args=[
-        '--experimental-options', '--python.UsePanama=true'
-    ]), SUITE, 10)
-
-    # java embedding driver
-    python_java_embedding_vm_registry.add_vm(
-        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_MULTI,
-                                extra_polyglot_args=['-multi-context']), SUITE, 10)
-    python_java_embedding_vm_registry.add_vm(
-        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_MULTI_SHARED,
-                                extra_polyglot_args=['-multi-context', '-shared-engine']), SUITE, 10)
-    python_java_embedding_vm_registry.add_vm(
-        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI,
-                                extra_polyglot_args=['-multi-context', '-interpreter']), SUITE, 10)
-    python_java_embedding_vm_registry.add_vm(
-        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI_SHARED,
-                                extra_polyglot_args=['-multi-context', '-interpreter', '-shared-engine']), SUITE, 10)
-
-
-def _register_bench_suites(namespace):
-    for py_bench_suite in PythonBenchmarkSuite.get_benchmark_suites(BENCHMARKS):
-        mx_benchmark.add_bm_suite(py_bench_suite)
-    for py_bench_suite in PythonJavaEmbeddingBenchmarkSuite.get_benchmark_suites(JAVA_DRIVER_BENCHMARKS):
-        mx_benchmark.add_bm_suite(py_bench_suite)
-    for py_bench_suite in PythonVmWarmupBenchmarkSuite.get_benchmark_suites(WARMUP_BENCHMARKS):
-        mx_benchmark.add_bm_suite(py_bench_suite)
-    mx_benchmark.add_bm_suite(PythonJMHDistMxBenchmarkSuite())
-    for py_bench_suite in PythonHeapBenchmarkSuite.get_benchmark_suites(HEAP_BENCHMARKS):
-        mx_benchmark.add_bm_suite(py_bench_suite)
-
 
 class CharsetFilteringPariticpant:
     """
@@ -2339,8 +2257,8 @@ def hardcoded_ver_is_behind_major_minor(m):
 
 def mx_post_parse_cmd_line(namespace):
     # all projects are now available at this time
-    _register_vms(namespace)
-    _register_bench_suites(namespace)
+    mx_graalpython_benchmark.register_vms(SUITE, SANDBOXED_OPTIONS)
+    mx_graalpython_benchmark.register_suites()
     mx_graalpython_python_benchmarks.register_python_benchmarks()
 
     for dist in mx.suite('graalpython').dists:
diff --git a/mx.graalpython/mx_graalpython_benchmark.py b/mx.graalpython/mx_graalpython_benchmark.py
@@ -37,7 +37,6 @@
 
 import mx
 import mx_benchmark
-import mx_graalpython
 from mx_benchmark import StdOutRule, java_vm_registry, Vm, GuestVm, VmBenchmarkSuite, AveragingBenchmarkMixin
 from mx_graalpython_bench_param import HARNESS_PATH
 
@@ -68,13 +67,17 @@
 
 PYTHON_VM_REGISTRY_NAME = "Python"
 CONFIGURATION_DEFAULT = "default"
+CONFIGURATION_DEFAULT_BC_DSL = "default-bc-dsl"
 CONFIGURATION_INTERPRETER = "interpreter"
+CONFIGURATION_INTERPRETER_BC_DSL = "interpreter-bc-dsl"
 CONFIGURATION_NATIVE_INTERPRETER = "native-interpreter"
+CONFIGURATION_NATIVE_INTERPRETER_BC_DSL = "native-interpreter-bc-dsl"
 CONFIGURATION_DEFAULT_MULTI = "default-multi"
 CONFIGURATION_INTERPRETER_MULTI = "interpreter-multi"
 CONFIGURATION_NATIVE_INTERPRETER_MULTI = "native-interpreter-multi"
 CONFIGURATION_DEFAULT_MULTI_TIER = "default-multi-tier"
 CONFIGURATION_NATIVE = "native"
+CONFIGURATION_NATIVE_BC_DSL = "native-bc-dsl"
 CONFIGURATION_NATIVE_MULTI = "native-multi"
 CONFIGURATION_NATIVE_MULTI_TIER = "native-multi-tier"
 CONFIGURATION_SANDBOXED = "sandboxed"
@@ -363,18 +366,19 @@ def _remove_vm_prefix(argument):
         else:
             return argument
 
+    @staticmethod
+    def _remove_vm_prefix_for_all(arguments):
+        return [GraalPythonVmBase._remove_vm_prefix(x) for x in arguments]
+
     def run(self, cwd, args):
         extra_polyglot_args = self.get_extra_polyglot_args()
 
-        if mx_graalpython.BYTECODE_DSL_INTERPRETER:
-            args.insert(0, "--vm.Dpython.EnableBytecodeDSLInterpreter=true")
-
         host_vm = self.host_vm()
         if hasattr(host_vm, 'run_lang'): # this is a full GraalVM build
             return self.run_in_graalvm(cwd, args, extra_polyglot_args, host_vm)
 
         # Otherwise, we're running from the source tree
-        args = [self._remove_vm_prefix(x) for x in args]
+        args = self._remove_vm_prefix_for_all(args)
         truffle_options = [
             # "-Dpolyglot.engine.CompilationExceptionsAreFatal=true"
         ]
@@ -394,7 +398,7 @@ def run(self, cwd, args):
 
         vm_args = mx.get_runtime_jvm_args(dists, cp_suffix=self._cp_suffix, cp_prefix=self._cp_prefix)
         if isinstance(self._extra_vm_args, list):
-            vm_args += self._extra_vm_args
+            vm_args += self._remove_vm_prefix_for_all(self._extra_vm_args)
         vm_args += [
             "-Dorg.graalvm.language.python.home=%s" % mx.dependency("GRAALPYTHON_GRAALVM_SUPPORT").get_output(),
             self.launcher_class(),
@@ -408,7 +412,18 @@ def run(self, cwd, args):
         if not self._env:
             self._env = dict()
         with environ(self._env):
-            return host_vm.run(cwd, cmd)
+            return self._validate_output(*host_vm.run(cwd, cmd))
+
+    def is_bytecode_dsl_config(self):
+        return self._extra_vm_args and '--vm.Dpython.EnableBytecodeDSLInterpreter=true' in self._extra_vm_args
+
+    def _validate_output(self, code, out, dims):
+        is_bytecode_dsl_config = self.is_bytecode_dsl_config()
+        if code == 0 and not f"using bytecode DSL interpreter: {is_bytecode_dsl_config}" in out:
+            print(f"ERROR: host VM config does not match what the the harness reported. "
+                  f"Expected Bytecode DSL interpreter = {is_bytecode_dsl_config}. Harness output:\n{out}", file=sys.stderr)
+            return 1, out, dims
+        return code, out, dims
 
     def name(self):
         return VM_NAME_GRAALPYTHON
@@ -435,13 +450,20 @@ def launcher_class(self):
         from mx_graalpython import GRAALPYTHON_MAIN_CLASS
         return GRAALPYTHON_MAIN_CLASS
 
+    def run(self, cwd, args):
+        if os.environ.get('BYTECODE_DSL_INTERPRETER', '').lower() == 'true' and not self.is_bytecode_dsl_config():
+            print("Found environment variable BYTECODE_DSL_INTERPRETER, but the guest vm config is not Bytecode DSL config.")
+            print("Did you want to use, e.g., `mx benchmark ... -- --host-vm-config=default-bc-dsl`?")
+            sys.exit(1)
+        return super().run(cwd, args)
+
     def run_in_graalvm(self, cwd, args, extra_polyglot_args, host_vm):
         with environ(self._env or {}):
             cp = self.get_classpath()
             if len(cp) > 0:
                 extra_polyglot_args.append("--vm.classpath=" + ":".join(cp))
             launcher_name = 'graalpy'
-            return host_vm.run_launcher(launcher_name, extra_polyglot_args + args, cwd)
+            return self._validate_output(*host_vm.run_launcher(launcher_name, extra_polyglot_args + args, cwd))
 
     def get_extra_polyglot_args(self):
         return ["--experimental-options", "-snapshot-startup", "--python.MaxNativeMemory=%s" % (2**34)] + self._extra_polyglot_args
@@ -1109,3 +1131,64 @@ def get_benchmark_suites(cls, benchmarks):
         assert isinstance(benchmarks, dict), "benchmarks must be a dict: {suite: [path, {bench: args, ... }], ...}"
         return [cls(suite_name, suite_info[0], suite_info[1])
                 for suite_name, suite_info in benchmarks.items()]
+
+
+def register_vms(suite, sandboxed_options):
+    # Other Python VMs:
+    python_vm_registry.add_vm(CPythonVm(config_name=CONFIGURATION_DEFAULT), suite)
+    python_vm_registry.add_vm(PyPyVm(config_name=CONFIGURATION_DEFAULT), suite)
+    python_vm_registry.add_vm(JythonVm(config_name=CONFIGURATION_DEFAULT), suite)
+
+    def add_graalpy_vm(name, *extra_polyglot_args, extra_vm_args=None):
+        python_vm_registry.add_vm(GraalPythonVm(config_name=name, extra_vm_args=extra_vm_args, extra_polyglot_args=extra_polyglot_args), suite, 10)
+
+    def add_graalpy_bc_dsl_vm(name, *extra_polyglot_args):
+        assert 'bc-dsl' in name
+        add_graalpy_vm(name, extra_vm_args=['--vm.Dpython.EnableBytecodeDSLInterpreter=true'], *extra_polyglot_args)
+
+    # GraalPy VMs:
+    add_graalpy_vm(CONFIGURATION_DEFAULT)
+    add_graalpy_bc_dsl_vm(CONFIGURATION_DEFAULT_BC_DSL)
+    add_graalpy_vm(CONFIGURATION_INTERPRETER, '--experimental-options', '--engine.Compilation=false')
+    add_graalpy_bc_dsl_vm(CONFIGURATION_INTERPRETER_BC_DSL, '--experimental-options', '--engine.Compilation=false')
+    add_graalpy_vm(CONFIGURATION_DEFAULT_MULTI, '--experimental-options', '-multi-context')
+    add_graalpy_vm(CONFIGURATION_INTERPRETER_MULTI, '--experimental-options', '-multi-context', '--engine.Compilation=false')
+    add_graalpy_vm(CONFIGURATION_DEFAULT_MULTI_TIER, '--experimental-options', '--engine.MultiTier=true')
+    add_graalpy_vm(CONFIGURATION_SANDBOXED, *sandboxed_options)
+    add_graalpy_vm(CONFIGURATION_NATIVE)
+    add_graalpy_bc_dsl_vm(CONFIGURATION_NATIVE_BC_DSL)
+    add_graalpy_vm(CONFIGURATION_NATIVE_INTERPRETER, '--experimental-options', '--engine.Compilation=false')
+    add_graalpy_bc_dsl_vm(CONFIGURATION_NATIVE_INTERPRETER_BC_DSL, '--experimental-options', '--engine.Compilation=false')
+    add_graalpy_vm(CONFIGURATION_SANDBOXED_MULTI, '--experimental-options', '-multi-context', *sandboxed_options)
+    add_graalpy_vm(CONFIGURATION_NATIVE_MULTI, '--experimental-options', '-multi-context')
+    add_graalpy_vm(CONFIGURATION_NATIVE_INTERPRETER_MULTI, '--experimental-options', '-multi-context', '--engine.Compilation=false')
+    add_graalpy_vm(CONFIGURATION_NATIVE_MULTI_TIER, '--experimental-options', '--engine.MultiTier=true')
+    add_graalpy_vm(CONFIGURATION_PANAMA, '--experimental-options', '--python.UsePanama=true')
+
+    # java embedding driver
+    python_java_embedding_vm_registry.add_vm(
+        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_MULTI,
+                                extra_polyglot_args=['-multi-context']), suite, 10)
+    python_java_embedding_vm_registry.add_vm(
+        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_MULTI_SHARED,
+                                extra_polyglot_args=['-multi-context', '-shared-engine']), suite, 10)
+    python_java_embedding_vm_registry.add_vm(
+        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI,
+                                extra_polyglot_args=['-multi-context', '-interpreter']), suite, 10)
+    python_java_embedding_vm_registry.add_vm(
+        GraalPythonJavaDriverVm(config_name=CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI_SHARED,
+                                extra_polyglot_args=['-multi-context', '-interpreter', '-shared-engine']), suite, 10)
+
+
+def register_suites():
+    from mx_graalpython_bench_param import BENCHMARKS, JAVA_DRIVER_BENCHMARKS, WARMUP_BENCHMARKS, HEAP_BENCHMARKS
+
+    for py_bench_suite in PythonBenchmarkSuite.get_benchmark_suites(BENCHMARKS):
+        mx_benchmark.add_bm_suite(py_bench_suite)
+    for py_bench_suite in PythonJavaEmbeddingBenchmarkSuite.get_benchmark_suites(JAVA_DRIVER_BENCHMARKS):
+        mx_benchmark.add_bm_suite(py_bench_suite)
+    for py_bench_suite in PythonVmWarmupBenchmarkSuite.get_benchmark_suites(WARMUP_BENCHMARKS):
+        mx_benchmark.add_bm_suite(py_bench_suite)
+    mx_benchmark.add_bm_suite(PythonJMHDistMxBenchmarkSuite())
+    for py_bench_suite in PythonHeapBenchmarkSuite.get_benchmark_suites(HEAP_BENCHMARKS):
+        mx_benchmark.add_bm_suite(py_bench_suite)

Original file line number	Diff line number	Diff line change
`@@ -1 +1 @@`
`1`		`-{ "overlay": "a242e9622be463c91c2cc034a1f2c1811316bc18" }`
	`1`	`+{ "overlay": "ae41f079f9ce46ca5f0ce5a2443c27cab856e403" }`
Original file line number	Diff line number	Diff line change
`@@ -1627,6 +1627,7 @@ public void initialize() {`
`1627`	`1627`	`}`
`1628`	`1628`	`releaseGil();`
`1629`	`1629`	`}`
	`1630`	`+ PythonOptions.checkBytecodeDSLEnv();`
`1630`	`1631`	`}`
`1631`	`1632`
`1632`	`1633`	`public void resetPerfCounter() {`
`@@ -1648,6 +1649,7 @@ public void patch(Env newEnv) {`
`1648`	`1649`	`} finally {`
`1649`	`1650`	`releaseGil();`
`1650`	`1651`	`}`
	`1652`	`+ PythonOptions.checkBytecodeDSLEnv();`
`1651`	`1653`	`}`
`1652`	`1654`
`1653`	`1655`	`private void importSiteIfForced() {`