Add a basic TOSA E2E backend.

We lower through linalg-on-tensors and use RefBackend to run it.
This adds enough support for a "tanh" op. Adding more ops should be
fairly mechanical now that things are wired up. Run with:
```
./tools/torchscript_e2e_test.sh -c tosa
```

The backend structure is very similar to linalg-on-tensors based E2E
backends and is a nice parallel (see `tosa_backend.py`). Actually, this
forced a nice refactoring to the layering here. We removed
`torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead
require separately running
```
torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline
```
This highlights the step that lowers to the "torch backend contract"
of cleaned up `torch` dialect ops is a critical step in the lowering.
Going forward, that is the key load-bearing contract of the torch-mlir
project, not the linalg-on-tensors backend contract.

Recommended review order:
- `TorchToTosa.cpp` / `TorchToTosa/basic.mlir`
- `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and
  the new `utils.py` file there.
- `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and
  `abc.py` in that directory for the TOSA backend e2e interface.
- other misc mechanical changes

.github/workflows/buildAndTest.yml

@@ -61,6 +61,7 @@ jobs:
         cd $GITHUB_WORKSPACE
         export PYTHONPATH="$GITHUB_WORKSPACE/build/tools/torch-mlir/python_packages/torch_mlir"
         python -m e2e_testing.torchscript.main --config=refbackend -v
+        python -m e2e_testing.torchscript.main --config=tosa -v
 
     # TODO: Only build packages in full Release mode.
     # On the other hand, having assertions on isn't too bad of an idea at this

e2e_testing/torchscript/main.py

@@ -15,12 +15,13 @@ from torch_mlir_e2e_test.torchscript.registry import GLOBAL_TEST_REGISTRY
 
 # Available test configs.
 from torch_mlir_e2e_test.torchscript.configs import (
-    LinalgOnTensorsBackendTestConfig, NativeTorchTestConfig, TorchScriptTestConfig
+    LinalgOnTensorsBackendTestConfig, NativeTorchTestConfig, TorchScriptTestConfig, TosaBackendTestConfig
 )
 
 from torch_mlir_e2e_test.linalg_on_tensors_backends.refbackend import RefBackendLinalgOnTensorsBackend
+from torch_mlir_e2e_test.tosa_backends.linalg_on_tensors import LinalgOnTensorsTosaBackend
 
-from .xfail_sets import XFAIL_SETS, COMMON_TORCH_MLIR_LOWERING_XFAILS
+from .xfail_sets import REFBACKEND_XFAIL_SET, TOSA_PASS_SET, COMMON_TORCH_MLIR_LOWERING_XFAILS
 
 # Import tests to register them in the global registry.
 # Make sure to use `tools/torchscript_e2e_test.sh` wrapper for invoking
@@ -35,7 +36,7 @@ from . import elementwise
 from . import reduction
 
 def _get_argparse():
-    config_choices = ['native_torch', 'torchscript', 'refbackend', 'external']
+    config_choices = ['native_torch', 'torchscript', 'refbackend', 'tosa', 'external']
     parser = argparse.ArgumentParser(description='Run torchscript e2e tests.')
     parser.add_argument('-c', '--config',
         choices=config_choices,
@@ -43,6 +44,7 @@ def _get_argparse():
         help=f'''
 Meaning of options:
 "refbackend": run through torch-mlir's RefBackend.
+"tosa": run through torch-mlir's default TOSA backend.
 "native_torch": run the torch.nn.Module as-is without compiling (useful for verifying model is deterministic; ALL tests should pass in this configuration).
 "torchscript": compile the model to a torch.jit.ScriptModule, and then run that as-is (useful for verifying TorchScript is modeling the program correctly).
 "external": use an external backend, specified by the `--external-backend` option.
@@ -77,16 +79,27 @@ for more information on building these artifacts.
 def main():
     args = _get_argparse().parse_args()
 
+    all_tests = list(GLOBAL_TEST_REGISTRY)
+    if args.serialized_test_dir:
+        for root, dirs, files in os.walk(args.serialized_test_dir):
+            for filename in files:
+                with open(os.path.join(root, filename), 'rb') as f:
+                    all_tests.append(pickle.load(f).as_test())
+    all_test_unique_names = set(test.unique_name for test in all_tests)
+
     # Find the selected config.
     if args.config == 'refbackend':
         config = LinalgOnTensorsBackendTestConfig(RefBackendLinalgOnTensorsBackend())
-        xfail_set = XFAIL_SETS['refbackend']
+        xfail_set = REFBACKEND_XFAIL_SET
+    if args.config == 'tosa':
+        config = TosaBackendTestConfig(LinalgOnTensorsTosaBackend())
+        xfail_set = all_test_unique_names - TOSA_PASS_SET
     elif args.config == 'native_torch':
         config = NativeTorchTestConfig()
-        xfail_set = XFAIL_SETS['native_torch']
+        xfail_set = {}
     elif args.config == 'torchscript':
         config = TorchScriptTestConfig()
-        xfail_set = XFAIL_SETS['torchscript']
+        xfail_set = {}
     elif args.config == 'external':
         with open(args.external_config, 'r') as f:
             code = compile(f.read(), args.external_config, 'exec')
@@ -106,13 +119,6 @@ def main():
             )
             sys.exit(1)
 
-    all_tests = list(GLOBAL_TEST_REGISTRY)
-    if args.serialized_test_dir:
-        for root, dirs, files in os.walk(args.serialized_test_dir):
-            for filename in files:
-                with open(os.path.join(root, filename), 'rb') as f:
-                    all_tests.append(pickle.load(f).as_test())
-
     # Find the selected tests, and emit a diagnostic if none are found.
     tests = [
         test for test in all_tests

e2e_testing/torchscript/xfail_sets.py

@@ -10,17 +10,15 @@
 # (this includes down into lower parts of the stack, where a side table
 # might be used to keep more elaborate sets of testing configurations).
 
-XFAIL_SETS = {}
-
 # Lists of tests that fail to even reach the backends.
 # These represent further work needed in torch-mlir to lower them properly
 # to the backend contract.
 COMMON_TORCH_MLIR_LOWERING_XFAILS = {
-    'QuantizedMLP_basic',
+    "QuantizedMLP_basic",
 }
 
-XFAIL_SETS['refbackend'] = COMMON_TORCH_MLIR_LOWERING_XFAILS
+REFBACKEND_XFAIL_SET = COMMON_TORCH_MLIR_LOWERING_XFAILS
 
-XFAIL_SETS['torchscript'] = {}
-
-XFAIL_SETS['native_torch'] = {}
+# Write the TOSA set as a "passing" set as it is very early in development
+# and very few tests work yet.
+TOSA_PASS_SET = {"ElementwiseUnaryModule_basic"}

examples/lazytensor_tanh.py

@@ -65,7 +65,7 @@ mlir_module.dump()
 
 # Compile the torch MLIR and execute the compiled program
 with mlir_module.context:
-    pm = PassManager.parse('torchscript-function-to-linalg-on-tensors-backend-pipeline')
+    pm = PassManager.parse('torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline')
 pm.run(mlir_module)
 
 print("BEFORE LINALG-ON-TENSORS BACKEND PIPELINE")

examples/resnet_inference.ipynb

@@ -153,7 +153,7 @@
     "    mb.import_module(scripted._c, class_annotator)\n",
     "\n",
     "    ## Lower the MLIR from TorchScript to RefBackend, passing through linalg-on-tensors.\n",
-    "    pm = PassManager.parse('torchscript-module-to-linalg-on-tensors-backend-pipeline', mb.module.context)\n",
+    "    pm = PassManager.parse('torchscript-module-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline', mb.module.context)\n",
     "    pm.run(mb.module)\n",
     "\n",
     "    ## Invoke RefBackend to compile to compiled artifact form.\n",

examples/torchfx_add_tanh_sigmoid.py

@@ -51,7 +51,7 @@ mlir_module.dump()
 print(mlir_module.operation.verify())
 
 with mlir_module.context:
-    pm = PassManager.parse('torchscript-module-to-linalg-on-tensors-backend-pipeline')
+    pm = PassManager.parse('torchscript-module-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline')
 pm.run(mlir_module)
 
 print("\n\nLOWERED MLIR")

examples/torchscript_resnet18_e2e.py

@@ -109,7 +109,7 @@ mb.import_module(recursivescriptmodule._c, class_annotator)
 
 backend = refbackend.RefBackendLinalgOnTensorsBackend()
 with mb.module.context:
-    pm = PassManager.parse('torchscript-module-to-linalg-on-tensors-backend-pipeline')
+    pm = PassManager.parse('torchscript-module-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline')
     pm.run(mb.module)
 
 compiled = backend.compile(mb.module)

include/torch-mlir/Conversion/Passes.td

@@ -105,4 +105,13 @@ def ConvertTorchToLinalg : Pass<"convert-torch-to-linalg", "FuncOp"> {
   let constructor = "mlir::torch::createConvertTorchToLinalgPass()";
 }
 
+def ConvertTorchToTosa : Pass<"convert-torch-to-tosa", "FuncOp"> {
+  let summary = "Convert Torch ops to TOSA ops";
+  let description = [{
+    This pass assumes that TOSA ops are responsible for emitting error
+    guards in case of shape mismatches.
+  }];
+  let constructor = "mlir::torch::createConvertTorchToTosaPass()";
+}
+
 #endif // TORCHMLIR_CONVERSION_PASSES

include/torch-mlir/Conversion/TorchToTosa/TorchToTosa.h

@@ -0,0 +1,22 @@
+//===------------------------------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// Also available under a BSD-style license. See LICENSE.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef TORCHMLIR_CONVERSION_TORCHTOTOSA_TORCHTOTOSA_H
+#define TORCHMLIR_CONVERSION_TORCHTOTOSA_TORCHTOTOSA_H
+
+#include "mlir/Pass/Pass.h"
+#include <memory>
+
+namespace mlir {
+namespace torch {
+std::unique_ptr<OperationPass<FuncOp>> createConvertTorchToTosaPass();
+}
+} // namespace mlir
+
+#endif // TORCHMLIR_CONVERSION_TORCHTOTOSA_TORCHTOTOSA_H

include/torch-mlir/Dialect/Torch/Transforms/Passes.h

@@ -33,13 +33,13 @@ struct TorchLoweringPipelineOptions
 
 /// Creates a pipeline that lowers the object graph IR that is produced by
 /// TorchScript import into the form expected by torch-verify-backend-contract.
-void createTorchScriptToTorchBackendPipeline(
+void createTorchScriptModuleToTorchBackendPipeline(
     OpPassManager &pm, const TorchLoweringPipelineOptions &options);
 
 /// Creates a pipeline that lowers a flat list of funcs and global slots
 /// with the torch and aten dialects and mutable arrays and converts it to
 /// the form required by torch-verify-backend-contract.
-void createGlobalizedModuleToTorchBackendPipeline(
+void createTorchFunctionToTorchBackendPipeline(
     OpPassManager &pm, const TorchLoweringPipelineOptions &options);
 
 std::unique_ptr<OperationPass<ModuleOp>> createAdjustCallingConventionsPass();

include/torch-mlir/Dialect/TorchConversion/Transforms/Passes.h

@@ -19,17 +19,15 @@ namespace mlir {
 namespace torch {
 namespace TorchConversion {
 
-/// Creates a pipeline that lowers the object graph IR that is given by a
-/// TorchScript jit.ScriptModule into the form expected by
-/// torch-verify-linalg-on-tensors-verify-backend-contract.
-void createTorchScriptModuleToLinalgOnTensorsBackendPipeline(
+/// Creates a pipeline that lowers from the torch backend contract to the
+/// linalg-on-tensors backend contract.
+void createTorchBackendToLinalgOnTensorsBackendPipeline(
     OpPassManager &pm,
     const torch::Torch::TorchLoweringPipelineOptions &options);
 
-/// Creates a pipeline that lowers the object graph IR that is given by a
-/// TorchScript jit.ScriptFunction into the form expected by
-/// torch-verify-linalg-on-tensors-verify-backend-contract.
-void createTorchScriptFunctionToLinalgOnTensorsBackendPipeline(
+/// Creates a pipeline that lowers from the torch backend contract to the
+/// TOSA backend contract.
+void createTorchBackendToTosaBackendPipeline(
     OpPassManager &pm,
     const torch::Torch::TorchLoweringPipelineOptions &options);
 
@@ -44,6 +42,8 @@ createFinalizingBackendTypeConversionPass();
 std::unique_ptr<OperationPass<ModuleOp>>
 createVerifyLinalgOnTensorsBackendContractPass();
 
+std::unique_ptr<OperationPass<ModuleOp>> createVerifyTosaBackendContractPass();
+
 } // namespace TorchConversion
 
 /// Registers all Torch transformation passes.

include/torch-mlir/Dialect/TorchConversion/Transforms/Passes.td

@@ -58,4 +58,9 @@ def VerifyLinalgOnTensorsBackendContract : Pass<"torch-verify-linalg-on-tensors-
   let constructor = "mlir::torch::TorchConversion::createVerifyLinalgOnTensorsBackendContractPass()";
 }
 
+def VerifyTosaBackendContract : Pass<"torch-verify-tosa-backend-contract", "ModuleOp"> {
+  let summary = "Verifies conformity to the linalg-on-tensors backend contract";
+  let constructor = "mlir::torch::TorchConversion::createVerifyTosaBackendContractPass()";
+}
+
 #endif // TORCHMLIR_TORCHCONVERSION_PASSES

lib/Conversion/CMakeLists.txt

@@ -1,6 +1,7 @@
 add_subdirectory(TorchToLinalg)
 add_subdirectory(TorchToSCF)
 add_subdirectory(TorchToStd)
+add_subdirectory(TorchToTosa)
 
 # TODO: Automate this with add_torch_mlir_conversion_library.
 #get_property(torch_mlir_conversion_libs GLOBAL PROPERTY TORCH_MLIR_CONVERSION_LIBS)
@@ -18,5 +19,6 @@ add_mlir_library(TorchMLIRConversionPasses
   TorchMLIRTorchToLinalg
   TorchMLIRTorchToSCF
   TorchMLIRTorchToStd
+  TorchMLIRTorchToTosa
   #${torch_mlir_conversion_libs}
 )

lib/Conversion/Passes.cpp

@@ -12,6 +12,7 @@
 #include "torch-mlir/Conversion/TorchToLinalg/TorchToLinalg.h"
 #include "torch-mlir/Conversion/TorchToSCF/TorchToSCF.h"
 #include "torch-mlir/Conversion/TorchToStd/TorchToStd.h"
+#include "torch-mlir/Conversion/TorchToTosa/TorchToTosa.h"
 
 //===----------------------------------------------------------------------===//
 // Pass registration

lib/Conversion/TorchToLinalg/TorchToLinalg.cpp

@@ -22,7 +22,6 @@
 
 using namespace mlir;
 using namespace mlir::torch;
-using namespace mlir::torch;
 using namespace mlir::torch::Torch;
 
 // -----------------------------------------------------------------------------

lib/Conversion/TorchToSCF/TorchToSCF.cpp

@@ -19,7 +19,6 @@
 
 using namespace mlir;
 using namespace mlir::torch;
-using namespace mlir::torch;
 using namespace mlir::torch::Torch;
 
 namespace {

lib/Conversion/TorchToStd/TorchToStd.cpp

@@ -20,7 +20,6 @@
 
 using namespace mlir;
 using namespace mlir::torch;
-using namespace mlir::torch;
 using namespace mlir::torch::Torch;
 
 // -----------------------------------------------------------------------------

lib/Conversion/TorchToTosa/CMakeLists.txt

@@ -0,0 +1,20 @@
+add_mlir_conversion_library(TorchMLIRTorchToTosa
+  TorchToTosa.cpp
+
+  ADDITIONAL_HEADER_DIRS
+  ${PROJECT_SOURCE_DIR}/include/torch-mlir/Conversion/TorchToTosa
+
+  DEPENDS
+  TorchMLIRConversionPassIncGen
+
+  LINK_COMPONENTS
+  Core
+
+  LINK_LIBS PUBLIC
+  MLIRIR
+  MLIRPass
+  MLIRTosa
+  TorchMLIRTorchDialect
+)
+
+torch_mlir_target_includes(TorchMLIRTorchToTosa)

lib/Conversion/TorchToTosa/TorchToTosa.cpp

@@ -0,0 +1,76 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// Also available under a BSD-style license. See LICENSE.
+//
+//===----------------------------------------------------------------------===//
+
+#include "torch-mlir/Conversion/TorchToTosa/TorchToTosa.h"
+
+#include "../PassDetail.h"
+#include "mlir/Dialect/Tosa/IR/TosaOps.h"
+#include "mlir/Dialect/Traits.h"
+#include "mlir/IR/Matchers.h"
+#include "mlir/Transforms/DialectConversion.h"
+#include "torch-mlir/Dialect/Torch/IR/TorchOps.h"
+#include "torch-mlir/Dialect/TorchConversion/IR/TorchConversionDialect.h"
+#include "torch-mlir/Dialect/TorchConversion/Transforms/BackendTypeConversion.h"
+
+using namespace mlir;
+using namespace mlir::torch;
+using namespace mlir::torch::Torch;
+
+namespace {
+class ConvertAtenTanhOp : public OpConversionPattern<AtenTanhOp> {
+public:
+  using OpConversionPattern::OpConversionPattern;
+  LogicalResult
+  matchAndRewrite(AtenTanhOp op, ArrayRef<Value> operands,
+                  ConversionPatternRewriter &rewriter) const override {
+    AtenTanhOp::Adaptor adaptor(operands);
+    rewriter.replaceOpWithNewOp<tosa::TanhOp>(
+        op, getTypeConverter()->convertType(op.getType()), adaptor.self());
+    return success();
+  }
+};
+} // namespace
+
+// -----------------------------------------------------------------------------
+// The pass
+// -----------------------------------------------------------------------------
+
+namespace {
+class ConvertTorchToTosa
+    : public ConvertTorchToTosaBase<ConvertTorchToTosa> {
+public:
+  void getDependentDialects(DialectRegistry &registry) const override {
+    registry.insert<tosa::TosaDialect>();
+    TorchConversion::getBackendTypeConversionDependentDialects(registry);
+  }
+
+  void runOnOperation() override {
+    MLIRContext *context = &getContext();
+    ConversionTarget target(*context);
+    target.addLegalDialect<tosa::TosaDialect>();
+
+    TypeConverter typeConverter;
+    typeConverter.addConversion([](Type type) { return type; });
+    TorchConversion::setupBackendTypeConversion(target, typeConverter);
+
+    RewritePatternSet patterns(context);
+    target.addIllegalOp<AtenTanhOp>();
+    patterns.add<ConvertAtenTanhOp>(typeConverter, context);
+
+    if (failed(applyPartialConversion(getOperation(), target,
+                                      std::move(patterns))))
+      return signalPassFailure();
+  }
+};
+} // namespace
+
+std::unique_ptr<OperationPass<FuncOp>>
+mlir::torch::createConvertTorchToTosaPass() {
+  return std::make_unique<ConvertTorchToTosa>();
+}

lib/Dialect/Torch/Transforms/Passes.cpp

@@ -23,16 +23,16 @@ namespace {
 void mlir::torch::registerTorchPasses() {
   ::registerPasses();
   mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
-      "torchscript-to-torch-backend-pipeline",
+      "torchscript-module-to-torch-backend-pipeline",
       "Pipeline lowering TorchScript object graph IR to Torch backend form.",
-      mlir::torch::Torch::createTorchScriptToTorchBackendPipeline);
+      mlir::torch::Torch::createTorchScriptModuleToTorchBackendPipeline);
   mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
-      "torch-globalized-module-to-torch-backend-pipeline",
-      "Pipeline lowering a globalized Torch program to Torch backend form.",
-      mlir::torch::Torch::createGlobalizedModuleToTorchBackendPipeline);
+      "torch-function-to-torch-backend-pipeline",
+      "Pipeline lowering a Torch function to Torch backend form.",
+      mlir::torch::Torch::createTorchFunctionToTorchBackendPipeline);
 }
 
-void mlir::torch::Torch::createTorchScriptToTorchBackendPipeline(
+void mlir::torch::Torch::createTorchScriptModuleToTorchBackendPipeline(
     OpPassManager &pm, const TorchLoweringPipelineOptions &options) {
   // When we import TorchScript IR, we import their entire "compilation unit",
   // which can contain numerous functions unrelated to the current program,
@@ -58,10 +58,10 @@ void mlir::torch::Torch::createTorchScriptToTorchBackendPipeline(
   // TODO: Improve shape inference.
   pm.addPass(createInlinerPass());
 
-  createGlobalizedModuleToTorchBackendPipeline(pm, options);
+  createTorchFunctionToTorchBackendPipeline(pm, options);
 }
 
-void mlir::torch::Torch::createGlobalizedModuleToTorchBackendPipeline(
+void mlir::torch::Torch::createTorchFunctionToTorchBackendPipeline(
     OpPassManager &pm, const TorchLoweringPipelineOptions &options) {
   // General considerations: As a matter of bring-up, we are simultaneously
   // building out the frontend pipeline and also co-developing the backend

lib/Dialect/TorchConversion/Transforms/BackendTypeConversion.cpp

@@ -21,7 +21,6 @@
 
 using namespace mlir;
 using namespace mlir::torch;
-using namespace mlir::torch;
 using namespace mlir::torch::TorchConversion;
 
 void mlir::torch::TorchConversion::getBackendTypeConversionDependentDialects(

lib/Dialect/TorchConversion/Transforms/CMakeLists.txt

@@ -3,6 +3,8 @@ add_mlir_library(TorchMLIRTorchConversionPasses
   Passes.cpp
   VerifyInvariantsBeforeBackendLowering.cpp
   VerifyLinalgOnTensorsBackendContract.cpp
+  VerifyTosaBackendContract.cpp
+
 
   ADDITIONAL_HEADER_DIRS
   ${PROJECT_SOURCE_DIR}/include/torch-mlir/Dialect/TorchConversion/Transforms

lib/Dialect/TorchConversion/Transforms/Passes.cpp

@@ -16,11 +16,11 @@
 #include "torch-mlir/Conversion/TorchToLinalg/TorchToLinalg.h"
 #include "torch-mlir/Conversion/TorchToSCF/TorchToSCF.h"
 #include "torch-mlir/Conversion/TorchToStd/TorchToStd.h"
+#include "torch-mlir/Conversion/TorchToTosa/TorchToTosa.h"
 #include "torch-mlir/Dialect/Torch/Transforms/Passes.h"
 
 using namespace mlir;
 using namespace mlir::torch;
-using namespace mlir::torch;
 
 //===----------------------------------------------------------------------===//
 // Pass registration
@@ -34,16 +34,18 @@ namespace {
 void mlir::torch::registerTorchConversionPasses() {
   ::registerPasses();
   mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
-      "torchscript-module-to-linalg-on-tensors-backend-pipeline",
-      "Pipeline lowering torch object graph representing a torch.jit.ScriptModule to linalg-on-tensors backend format.",
-      TorchConversion::createTorchScriptModuleToLinalgOnTensorsBackendPipeline);
+      "torch-backend-to-linalg-on-tensors-backend-pipeline",
+      "Pipeline lowering torch backend contract to linalg-on-tensors backend "
+      "contract.",
+      TorchConversion::createTorchBackendToLinalgOnTensorsBackendPipeline);
   mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
-      "torchscript-function-to-linalg-on-tensors-backend-pipeline",
-      "Pipeline lowering a flat list of functions representing a torch.jit.ScriptFunction to linalg-on-tensors backend format.",
-      TorchConversion::createTorchScriptFunctionToLinalgOnTensorsBackendPipeline);
+      "torch-backend-to-tosa-backend-pipeline",
+      "Pipeline lowering torch backend contract to TOSA backend "
+      "contract.",
+      TorchConversion::createTorchBackendToTosaBackendPipeline);
 }
 
-static void createTorchBackendToLinalgOnTensorsBackendPipeline(
+void TorchConversion::createTorchBackendToLinalgOnTensorsBackendPipeline(
     OpPassManager &pm, const Torch::TorchLoweringPipelineOptions &options) {
   // Check some invariants to catch errors in a clear way.
   pm.addPass(
@@ -80,20 +82,29 @@ static void createTorchBackendToLinalgOnTensorsBackendPipeline(
   pm.addPass(TorchConversion::createVerifyLinalgOnTensorsBackendContractPass());
 }
 
-void TorchConversion::createTorchScriptModuleToLinalgOnTensorsBackendPipeline(
+void TorchConversion::createTorchBackendToTosaBackendPipeline(
     OpPassManager &pm, const Torch::TorchLoweringPipelineOptions &options) {
+  // Check some invariants to catch errors in a clear way.
+  pm.addPass(
+      TorchConversion::createVerifyInvariantsBeforeBackendLoweringPass());
 
-  // Conversion to the linalg-on-tensors backend contract starts from the Torch
-  // backend contract.
-  Torch::createTorchScriptToTorchBackendPipeline(pm, options);
-  createTorchBackendToLinalgOnTensorsBackendPipeline(pm, options);
-}
-
-void TorchConversion::createTorchScriptFunctionToLinalgOnTensorsBackendPipeline(
-    OpPassManager &pm, const Torch::TorchLoweringPipelineOptions &options) {
-
-  // Conversion to the linalg-on-tensors backend contract starts from the Torch
-  // backend contract.
-  Torch::createGlobalizedModuleToTorchBackendPipeline(pm, options);
-  createTorchBackendToLinalgOnTensorsBackendPipeline(pm, options);
+  pm.addNestedPass<FuncOp>(createConvertTorchToTosaPass());
+
+  if (options.optimize) {
+    // Clean up any non-canonical code introduced above..
+    pm.addNestedPass<FuncOp>(createCanonicalizerPass());
+    // The resolution of `dim` ops tends to create identical ops. CSE them.
+    pm.addNestedPass<FuncOp>(createCSEPass());
+  }
+
+  // Finish the type conversion from `torch` types to the types of the
+  // TOSA backend contract.
+  pm.addPass(TorchConversion::createFuncBackendTypeConversionPass());
+  pm.addNestedPass<FuncOp>(
+      TorchConversion::createFinalizingBackendTypeConversionPass());
+
+  // Verify that we have lowered to the form that TOSA backends
+  // expect. This fails compilation (signalPassFailure) if the IR is not in the
+  // correct form.
+  pm.addPass(TorchConversion::createVerifyTosaBackendContractPass());
 }

lib/Dialect/TorchConversion/Transforms/VerifyTosaBackendContract.cpp

@@ -0,0 +1,62 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// Also available under a BSD-style license. See LICENSE.
+//
+//===----------------------------------------------------------------------===//
+
+#include "PassDetail.h"
+
+#include "mlir/Dialect/StandardOps/IR/Ops.h"
+#include "mlir/Dialect/Tensor/IR/Tensor.h"
+#include "mlir/Dialect/Tosa/IR/TosaOps.h"
+#include "mlir/IR/BuiltinOps.h"
+#include "mlir/IR/OpDefinition.h"
+#include "mlir/Transforms/DialectConversion.h"
+#include "torch-mlir/Dialect/TorchConversion/Transforms/Passes.h"
+
+using namespace mlir;
+using namespace mlir::torch;
+using namespace mlir::torch::TorchConversion;
+
+namespace {
+class VerifyTosaBackendContractPass
+    : public VerifyTosaBackendContractBase<VerifyTosaBackendContractPass> {
+  void runOnOperation() override {
+    MLIRContext *context = &getContext();
+    auto module = getOperation();
+    TypeConverter converter;
+    converter.addConversion([](TensorType type) -> Type {
+      if (BaseMemRefType::isValidElementType(type.getElementType()))
+        return type;
+      return nullptr;
+    });
+
+    auto opHasLegalTypes = [&](Operation *op) { return converter.isLegal(op); };
+
+    ConversionTarget target(*context);
+
+    // Structural operations.
+    target.addDynamicallyLegalOp<ModuleOp, FuncOp, ReturnOp>(opHasLegalTypes);
+    // Basic scalar operations.
+    target.addLegalDialect<tosa::TosaDialect>();
+
+    RewritePatternSet patterns(context);
+    if (failed(applyFullConversion(module, target, std::move(patterns)))) {
+      // We avoid `module.emitError()` so that mlir-print-op-on-diagnostics
+      // doesn't unnecessarily spew out the entire module.
+      emitError(module.getLoc())
+          << "Module does not conform to the TOSA backend contract. "
+             "See dialect conversion legality information above.";
+      return signalPassFailure();
+    }
+  }
+};
+} // namespace
+
+std::unique_ptr<OperationPass<ModuleOp>>
+mlir::torch::TorchConversion::createVerifyTosaBackendContractPass() {
+  return std::make_unique<VerifyTosaBackendContractPass>();
+}

python/torch_mlir_e2e_test/torchscript/configs/__init__.py

@@ -6,3 +6,4 @@
 from .linalg_on_tensors_backend import LinalgOnTensorsBackendTestConfig
 from .native_torch import NativeTorchTestConfig
 from .torchscript import TorchScriptTestConfig
+from .tosa_backend import TosaBackendTestConfig

python/torch_mlir_e2e_test/torchscript/configs/linalg_on_tensors_backend.py

@@ -12,47 +12,15 @@ import tempfile
 import numpy as np
 import torch
 
-from torch_mlir.dialects.torch.importer.jit_ir import ClassAnnotator, ModuleBuilder
-from torch_mlir.dialects.torch.importer.jit_ir.torchscript_annotations import extract_annotations
-from torch_mlir.passmanager import PassManager
 from torch_mlir_e2e_test.linalg_on_tensors_backends.abc import LinalgOnTensorsBackend
 from torch_mlir_e2e_test.torchscript.framework import TestConfig, Trace, TraceItem
+from .utils import (
+    recursively_convert_to_numpy,
+    recursively_convert_from_numpy,
+    convert_torchscript_module_to_torch_backend_contract_mlir,
+    run_pipeline_with_repro_report
+)
 
-def _recursively_convert_to_numpy(o: Any):
-    if isinstance(o, torch.Tensor):
-        return o.numpy()
-    if isinstance(o, tuple):
-        return tuple(_recursively_convert_to_numpy(x) for x in o)
-    if isinstance(o, list):
-        return [_recursively_convert_to_numpy(x) for x in o]
-    if isinstance(o, dict):
-        return {k: _recursively_convert_to_numpy(v) for k, v in o.items()}
-    # No-op cases. Explicitly enumerated to avoid things sneaking through.
-    if isinstance(o, str):
-        return o
-    if isinstance(o, float):
-        return o
-    if isinstance(o, int):
-        return o
-    raise Exception(f"Unexpected Python function input: {o}")
-
-def _recursively_convert_from_numpy(o: Any):
-    if isinstance(o, np.ndarray):
-        return torch.from_numpy(o)
-    if isinstance(o, tuple):
-        return tuple(_recursively_convert_from_numpy(x) for x in o)
-    if isinstance(o, list):
-        return [_recursively_convert_from_numpy(x) for x in o]
-    if isinstance(o, dict):
-        return {k: _recursively_convert_from_numpy(v) for k, v in o.items()}
-    # No-op cases. Explicitly enumerated to avoid things sneaking through.
-    if isinstance(o, str):
-        return o
-    if isinstance(o, float):
-        return o
-    if isinstance(o, int):
-        return o
-    raise Exception(f"Unexpected Python function output: {o}")
 
 class LinalgOnTensorsBackendTestConfig(TestConfig):
     """Base class for TestConfig's that are implemented with linalg-on-tensors.
@@ -65,73 +33,28 @@ class LinalgOnTensorsBackendTestConfig(TestConfig):
         self.backend = backend
 
     def compile(self, program: torch.nn.Module) -> Any:
-        mb = ModuleBuilder()
-        scripted = torch.jit.script(program)
-        class_annotator = ClassAnnotator()
 
-        extract_annotations(program, scripted, class_annotator)
+        module = convert_torchscript_module_to_torch_backend_contract_mlir(
+            program)
 
-        # TODO: Find a way to make each of these calls own its own
-        # "debuggable error report" situation.
-        try:
-            sys.stderr = StringIO()
-            # Import the TorchScript module to MLIR
-            mb.import_module(scripted._c, class_annotator)
-        except Exception as e:
-            raise Exception(f"""
-PyTorch TorchScript module -> torch-mlir Object Graph IR import failed with:
-Exception:
-{e}
-Diagnostics:
-{sys.stderr.getvalue()}
-""") from None
-        finally:
-            sys.stderr = sys.__stderr__
+        run_pipeline_with_repro_report(
+            module,
+            "torch-backend-to-linalg-on-tensors-backend-pipeline",
+            "Lower Torch Backend IR -> Linalg-on-Tensors Backend IR",
+            program.__class__.__name__)
 
         try:
             sys.stderr = StringIO()
-            asm_for_error_report = mb.module.operation.get_asm(
+            asm_for_error_report = module.operation.get_asm(
                 large_elements_limit=10, enable_debug_info=True)
-            pipeline_str = "torchscript-module-to-linalg-on-tensors-backend-pipeline"
-            # Lower module in place to make it ready for compiler backends.
-            with mb.module.context:
-                pm = PassManager.parse(pipeline_str)
-                pm.run(mb.module)
+            return self.backend.compile(module)
         except Exception as e:
-            # TODO: More robust.
-            # - don't arbitrarily clutter up /tmp. When a test suite has many
-            #   tests, this can be a big disk cost (also, /tmp/ is frequently a
-            #   RAM fs, which increases worries about capacity).
-            # - don't have colliding filenames (hard to do without cluttering
-            #   up /tmp)
-            # - if we do have have colliding filenames, writes should at least
-            #   avoid being racy.
             filename = os.path.join(tempfile.gettempdir(),
-                                    scripted.original_name + '.mlir')
+                                    program.__class__.__name__ + ".mlir")
             with open(filename, 'w') as f:
                 f.write(asm_for_error_report)
             raise Exception(f"""
-torch-mlir TorchScript Object Graph IR -> linalg-on-tensors backend IR lowering failed with the following diagnostics:
-{sys.stderr.getvalue()}
-
-Error can be reproduced with:
-$ torch-mlir-opt -{pipeline_str} {filename}
-""") from None
-        finally:
-            sys.stderr = sys.__stderr__
-
-        try:
-            sys.stderr = StringIO()
-            asm_for_error_report = mb.module.operation.get_asm(
-                large_elements_limit=10, enable_debug_info=True)
-            return self.backend.compile(mb.module)
-        except Exception as e:
-            filename = os.path.join(tempfile.gettempdir(),
-                                    scripted.original_name + '.mlir')
-            with open(filename, 'w') as f:
-                f.write(asm_for_error_report)
-            raise Exception(f"""
-torch-mlir linalg-on-tensors Backend lowering for {self.backend.__class__.__name__} failed with the following diagnostics:
+Linalg-on-Tensors Backend lowering for {self.backend.__class__.__name__} failed with the following diagnostics:
 ## Exception:
 {e}
 
@@ -148,9 +71,9 @@ torch-mlir linalg-on-tensors Backend lowering for {self.backend.__class__.__name
         backend_module = self.backend.load(artifact)
         result: Trace = []
         for item in trace:
-            numpy_inputs = _recursively_convert_to_numpy(item.inputs)
+            numpy_inputs = recursively_convert_to_numpy(item.inputs)
             outputs = getattr(backend_module, item.symbol)(*numpy_inputs)
-            output = _recursively_convert_from_numpy(outputs)
+            output = recursively_convert_from_numpy(outputs)
             result.append(
                 TraceItem(symbol=item.symbol,
                           inputs=item.inputs,

python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py

@@ -0,0 +1,81 @@
+# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# Also available under a BSD-style license. See LICENSE.
+
+import sys
+from typing import Any
+from io import StringIO
+import os
+import tempfile
+
+import numpy as np
+import torch
+
+from torch_mlir_e2e_test.tosa_backends.abc import TosaBackend
+from torch_mlir_e2e_test.torchscript.framework import TestConfig, Trace, TraceItem
+from .utils import (
+    recursively_convert_to_numpy,
+    recursively_convert_from_numpy,
+    convert_torchscript_module_to_torch_backend_contract_mlir,
+    run_pipeline_with_repro_report
+)
+
+
+class TosaBackendTestConfig(TestConfig):
+    """Base class for TestConfig's that are implemented with linalg-on-tensors.
+
+    This class handles all the common lowering that torch-mlir does before
+    reaching the linalg-on-tensors abstraction level.
+    """
+    def __init__(self, backend: TosaBackend):
+        super().__init__()
+        self.backend = backend
+
+    def compile(self, program: torch.nn.Module) -> Any:
+
+        module = convert_torchscript_module_to_torch_backend_contract_mlir(
+            program)
+
+        run_pipeline_with_repro_report(
+            module,
+            "torch-backend-to-tosa-backend-pipeline",
+            "Lower Torch Backend IR -> TOSA Backend IR",
+            program.__class__.__name__)
+
+        try:
+            sys.stderr = StringIO()
+            asm_for_error_report = module.operation.get_asm(
+                large_elements_limit=10, enable_debug_info=True)
+            return self.backend.compile(module)
+        except Exception as e:
+            filename = os.path.join(tempfile.gettempdir(),
+                                    program.__class__.__name__ + ".mlir")
+            with open(filename, 'w') as f:
+                f.write(asm_for_error_report)
+            raise Exception(f"""
+TOSA Backend lowering for {self.backend.__class__.__name__} failed with the following diagnostics:
+## Exception:
+{e}
+
+## Stderr:
+{sys.stderr.getvalue()}
+
+## Input IR has been saved in {filename}
+""") from None
+        finally:
+            sys.stderr = sys.__stderr__
+
+
+    def run(self, artifact: Any, trace: Trace) -> Trace:
+        backend_module = self.backend.load(artifact)
+        result: Trace = []
+        for item in trace:
+            numpy_inputs = recursively_convert_to_numpy(item.inputs)
+            outputs = getattr(backend_module, item.symbol)(*numpy_inputs)
+            output = recursively_convert_from_numpy(outputs)
+            result.append(
+                TraceItem(symbol=item.symbol,
+                          inputs=item.inputs,
+                          output=output))
+        return result

python/torch_mlir_e2e_test/torchscript/configs/utils.py

@@ -0,0 +1,129 @@
+# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# Also available under a BSD-style license. See LICENSE.
+
+import sys
+from typing import Any
+from io import StringIO
+import os
+import tempfile
+
+import numpy as np
+import torch
+
+from torch_mlir.dialects.torch.importer.jit_ir import ClassAnnotator, ModuleBuilder
+from torch_mlir.dialects.torch.importer.jit_ir.torchscript_annotations import extract_annotations
+from torch_mlir.passmanager import PassManager
+from torch_mlir_e2e_test.linalg_on_tensors_backends.abc import LinalgOnTensorsBackend
+from torch_mlir_e2e_test.torchscript.framework import TestConfig, Trace, TraceItem
+
+def recursively_convert_to_numpy(o: Any):
+    if isinstance(o, torch.Tensor):
+        return o.numpy()
+    if isinstance(o, tuple):
+        return tuple(recursively_convert_to_numpy(x) for x in o)
+    if isinstance(o, list):
+        return [recursively_convert_to_numpy(x) for x in o]
+    if isinstance(o, dict):
+        return {k: recursively_convert_to_numpy(v) for k, v in o.items()}
+    # No-op cases. Explicitly enumerated to avoid things sneaking through.
+    if isinstance(o, str):
+        return o
+    if isinstance(o, float):
+        return o
+    if isinstance(o, int):
+        return o
+    raise Exception(f"Unexpected Python function input: {o}")
+
+def recursively_convert_from_numpy(o: Any):
+    if isinstance(o, np.ndarray):
+        return torch.from_numpy(o)
+    if isinstance(o, tuple):
+        return tuple(recursively_convert_from_numpy(x) for x in o)
+    if isinstance(o, list):
+        return [recursively_convert_from_numpy(x) for x in o]
+    if isinstance(o, dict):
+        return {k: recursively_convert_from_numpy(v) for k, v in o.items()}
+    # No-op cases. Explicitly enumerated to avoid things sneaking through.
+    if isinstance(o, str):
+        return o
+    if isinstance(o, float):
+        return o
+    if isinstance(o, int):
+        return o
+    raise Exception(f"Unexpected Python function output: {o}")
+
+
+def run_pipeline_with_repro_report(module,
+                                   pipeline: str,
+                                   description: str,
+                                   module_name: str):
+    """Runs `pipeline` on `module`, with a nice repro report if it fails."""
+    try:
+        sys.stderr = StringIO()
+        asm_for_error_report = module.operation.get_asm(
+            large_elements_limit=10, enable_debug_info=True)
+        # Lower module in place to make it ready for compiler backends.
+        with module.context:
+            pm = PassManager.parse(pipeline)
+            pm.run(module)
+    except Exception as e:
+        # TODO: More robust.
+        # - don't arbitrarily clutter up /tmp. When a test suite has many
+        #   tests, this can be a big disk cost (also, /tmp/ is frequently a
+        #   RAM fs, which increases worries about capacity).
+        # - don't have colliding filenames (hard to do without cluttering
+        #   up /tmp)
+        # - if we do have have colliding filenames, writes should at least
+        #   avoid being racy.
+        filename = os.path.join(tempfile.gettempdir(), module_name + ".mlir")
+        with open(filename, 'w') as f:
+            f.write(asm_for_error_report)
+        raise Exception(f"""
+{description} failed with the following diagnostics:
+{sys.stderr.getvalue()}
+
+Error can be reproduced with:
+$ torch-mlir-opt -pass-pipeline='{pipeline}' {filename}
+""") from None
+    finally:
+        sys.stderr = sys.__stderr__
+
+
+def convert_torchscript_module_to_torch_backend_contract_mlir(program: torch.nn.Module):
+    """Perform common lowering from TorchScript to Torch MLIR
+
+    Returns an MLIR module that satisfies the Torch backend contract.
+    """
+    mb = ModuleBuilder()
+    scripted = torch.jit.script(program)
+    class_annotator = ClassAnnotator()
+
+    extract_annotations(program, scripted, class_annotator)
+
+
+    # TODO: Find a way to make each of these calls own its own
+    # "debuggable error report" situation.
+    try:
+        sys.stderr = StringIO()
+        # Import the TorchScript module to MLIR
+        mb.import_module(scripted._c, class_annotator)
+    except Exception as e:
+        raise Exception(f"""
+PyTorch TorchScript module -> torch-mlir Object Graph IR import failed with:
+Exception:
+{e}
+Diagnostics:
+{sys.stderr.getvalue()}
+""") from None
+    finally:
+        sys.stderr = sys.__stderr__
+
+    run_pipeline_with_repro_report(
+        mb.module,
+        "torchscript-module-to-torch-backend-pipeline",
+        "Lowering TorchScript Object Graph IR -> Torch Backend IR",
+        program.__class__.__name__)
+
+    return mb.module

python/torch_mlir_e2e_test/tosa_backends/abc.py

@@ -0,0 +1,46 @@
+# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# Also available under a BSD-style license. See LICENSE.
+
+import abc
+from typing import TypeVar
+
+import torch
+
+from torch_mlir.ir import Module
+
+# A type shared between the result of `TosaBackend.compile` and the
+# input to `TosaBackend.load`. Each backend will likely have a
+# different definition of this type.
+CompiledArtifact = TypeVar('CompiledArtifact')
+
+# A wrapper around a backend-specific loaded program representation
+# that uniformly translates the `x.method(...)` interface expected of
+# Torch modules into appropriate lower-level operations.
+Invoker = TypeVar('Invoker')
+
+
+class TosaBackend(abc.ABC):
+    """The interface to an TOSA backend.
+    """
+    @abc.abstractmethod
+    def compile(self, module: Module) -> CompiledArtifact:
+        """Compile the provided MLIR module into a compiled artifact.
+
+        The module adheres to the TOSA backend contract
+        (see the VerifyTosaBackendContract pass).
+
+        The compiled artifact can be any type, but must be correctly
+        interpreted by the `load` method.
+        """
+
+    @abc.abstractmethod
+    def load(self, artifact: CompiledArtifact) -> Invoker:
+        """Load the compiled artifact into a uniformly invokable form.
+
+        The compiled artifact is the result of a previous call to `compile`.
+
+        See the description of `Invoker` for the requirements on the returned
+        type.
+        """

python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py

@@ -0,0 +1,48 @@
+# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# Also available under a BSD-style license. See LICENSE.
+
+from torch_mlir.ir import *
+from torch_mlir.passmanager import *
+# Imported for side effects.
+import torch_mlir.all_passes_registration
+
+from torch_mlir_e2e_test.linalg_on_tensors_backends.refbackend import RefBackendLinalgOnTensorsBackend
+
+from .abc import TosaBackend
+
+__all__ = [
+    "LinalgOnTensorsTosaBackend",
+]
+
+class LinalgOnTensorsTosaBackend(TosaBackend):
+    """Main entry-point for the linalg-on-tensors based TOSA backend.
+
+    This currently uses the linalg-on-tensors RefBackend for actual execution.
+    """
+    def __init__(self):
+        super().__init__()
+        self.refbackend = RefBackendLinalgOnTensorsBackend()
+
+    def compile(self, imported_module: Module):
+        """Compiles an imported module that satisfied the TOSA backend contract.
+
+        Args:
+          imported_module: The MLIR module consisting of funcs in the TOSA
+            dialect.
+        Returns:
+          An opaque, backend specific compiled artifact object that can be
+          passed to `load`.
+        """
+        # TODO: Error/repro reporting.
+        # We should store the program name as the symbol name of the MLIR
+        # module so we don't have to have access to the original program for it.
+        with imported_module.context:
+            pm = PassManager.parse("builtin.func(tosa-to-linalg-on-tensors)")
+            pm.run(imported_module)
+        return self.refbackend.compile(imported_module)
+
+    def load(self, module):
+        """Loads a compiled artifact into the runtime."""
+        return self.refbackend.load(module)

test/Conversion/TorchToTosa/basic.mlir

@@ -0,0 +1,12 @@
+// RUN: torch-mlir-opt <%s -convert-torch-to-tosa -split-input-file -verify-diagnostics | FileCheck %s
+
+// CHECK-LABEL:   func @torch.aten.tanh$basic(
+// CHECK-SAME:                                %[[ARG:.*]]: !torch.vtensor<[?,?],f32>) -> !torch.vtensor<[?,?],f32> {
+// CHECK:           %[[ARG_BUILTIN:.*]] = torch_c.to_builtin_tensor %[[ARG]] : !torch.vtensor<[?,?],f32> -> tensor<?x?xf32>
+// CHECK:           %[[RESULT_BUILTIN:.*]] = "tosa.tanh"(%[[ARG_BUILTIN]]) : (tensor<?x?xf32>) -> tensor<?x?xf32>
+// CHECK:           %[[RESULT:.*]] = torch_c.from_builtin_tensor %[[RESULT_BUILTIN]] : tensor<?x?xf32> -> !torch.vtensor<[?,?],f32>
+// CHECK:           return %[[RESULT]] : !torch.vtensor<[?,?],f32>
+func @torch.aten.tanh$basic(%arg0: !torch.vtensor<[?,?],f32>) -> !torch.vtensor<[?,?],f32> {
+  %0 = torch.aten.tanh %arg0 : !torch.vtensor<[?,?],f32> -> !torch.vtensor<[?,?],f32>
+  return %0 : !torch.vtensor<[?,?],f32>
+}

test/Dialect/TorchConversion/verify-linalg-on-tensors-backend-contract.mlir

@@ -36,7 +36,7 @@ module {
 //
 // The reporting we inherit from dialect conversion is not precise.
 // For example, here we want it to explicitly call out that
-// `tensor<?x!numpy.any_dtype>` is the problem here, which suggests
+// `!torch.tensor` is the problem here, which suggests
 // that type inference didn't succeed, or insufficient type information
 // was available.
 //
@@ -46,8 +46,8 @@ module {
 
 // expected-error@+1 {{Module does not conform to the linalg-on-tensors backend contract.}}
 module {
-  func @disallowed(%arg0: tensor<?x!numpy.any_dtype>) -> tensor<?x!numpy.any_dtype> {
+  func @disallowed(%arg0: !torch.tensor) -> !torch.tensor {
     // expected-error@+1 {{failed to legalize operation 'std.return'}}
-    return %arg0 : tensor<?x!numpy.any_dtype>
+    return %arg0 : !torch.tensor
   }
 }

test/Dialect/TorchConversion/verify-tosa-backend-contract.mlir

@@ -0,0 +1,42 @@
+// RUN: torch-mlir-opt -torch-verify-tosa-backend-contract -split-input-file -verify-diagnostics -allow-unregistered-dialect %s | FileCheck %s
+
+// CHECK: func @tanh
+func @tanh(%arg0: tensor<?x?xf32>) -> tensor<?x?xf32> {
+  %0 = "tosa.tanh"(%arg0) : (tensor<?x?xf32>) -> tensor<?x?xf32>
+  return %0 : tensor<?x?xf32>
+}
+
+// -----
+
+// Basic check of error reporting.
+
+// expected-error@+1 {{Module does not conform to the TOSA backend contract.}}
+module {
+  func @disallowed() {
+    // expected-error@+1 {{failed to legalize operation 'unknown_dialect.unknown_op'}}
+    "unknown_dialect.unknown_op"() : () -> ()
+    return
+  }
+}
+
+// -----
+
+// TODO: Improve these errors to give more exact reporting.
+//
+// The reporting we inherit from dialect conversion is not precise.
+// For example, here we want it to explicitly call out that
+// `!torch.tensor` is the problem here, which suggests
+// that type inference didn't succeed, or insufficient type information
+// was available.
+//
+// Ultimately, the output of this pass needs to be conveyed to the user
+// in an understandable way, such as suggesting a particular place where
+// a shape annotation is needed.
+
+// expected-error@+1 {{Module does not conform to the TOSA backend contract.}}
+module {
+  func @disallowed(%arg0: !torch.tensor) -> !torch.tensor {
+    // expected-error@+1 {{failed to legalize operation 'std.return'}}
+    return %arg0 : !torch.tensor
+  }
+}