[mlir][bufferization] Remove finalizing-bufferize pass
#114154
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
matthias-springer
requested review from
aartbik,
PeimingLiu and
yinying-lisa-li
as code owners
llvmbot
added
mlir:sparse
|
@llvm/pr-subscribers-mlir @llvm/pr-subscribers-mlir-bufferization Author: Matthias Springer (matthias-springer) ChangesThe dialect conversion-based bufferization passes have been migrated to One-Shot Bufferize about two years ago. To clean up the code base, this commit removes the Note for LLVM integration: If you depend on this pass, migrate to One-Shot Bufferize or copy the pass to your codebase. Depends on #114152. Full diff: https://github.com/llvm/llvm-project/pull/114154.diff 5 Files Affected:
diff --git a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
index 72abb5b3f1f94e..8b957577292796 100644
--- a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
+++ b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
@@ -200,10 +200,6 @@ std::unique_ptr<Pass> createEmptyTensorToAllocTensorPass();
/// Drop all memref function results that are equivalent to a function argument.
LogicalResult dropEquivalentBufferResults(ModuleOp module);
-/// Creates a pass that finalizes a partial bufferization by removing remaining
-/// bufferization.to_tensor and bufferization.to_memref operations.
-std::unique_ptr<OperationPass<func::FuncOp>> createFinalizingBufferizePass();
-
/// Create a pass that bufferizes all ops that implement BufferizableOpInterface
/// with One-Shot Bufferize.
std::unique_ptr<Pass> createOneShotBufferizePass();
diff --git a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
index cc5463ea968fc3..3e1b55cf99767f 100644
--- a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
+++ b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
@@ -343,22 +343,6 @@ def BufferResultsToOutParams : Pass<"buffer-results-to-out-params", "ModuleOp">
let dependentDialects = ["memref::MemRefDialect"];
}
-def FinalizingBufferize : Pass<"finalizing-bufferize", "func::FuncOp"> {
- let summary = "Finalize a partial bufferization";
- let description = [{
- A bufferize pass that finalizes a partial bufferization by removing
- remaining `bufferization.to_tensor` and `bufferization.to_buffer` operations.
-
- The removal of those operations is only possible if the operations only
- exist in pairs, i.e., all uses of `bufferization.to_tensor` operations are
- `bufferization.to_buffer` operations.
-
- This pass will fail if not all operations can be removed or if any operation
- with tensor typed operands remains.
- }];
- let constructor = "mlir::bufferization::createFinalizingBufferizePass()";
-}
-
def DropEquivalentBufferResults : Pass<"drop-equivalent-buffer-results", "ModuleOp"> {
let summary = "Remove MemRef return values that are equivalent to a bbArg";
let description = [{
diff --git a/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp b/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
index 1d009b03754c52..62ce2583f4fa1d 100644
--- a/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
+++ b/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
@@ -26,7 +26,6 @@
namespace mlir {
namespace bufferization {
-#define GEN_PASS_DEF_FINALIZINGBUFFERIZE
#define GEN_PASS_DEF_BUFFERIZATIONBUFFERIZE
#define GEN_PASS_DEF_ONESHOTBUFFERIZE
#include "mlir/Dialect/Bufferization/Transforms/Passes.h.inc"
@@ -98,75 +97,6 @@ void mlir::bufferization::populateBufferizeMaterializationLegality(
}
namespace {
-// In a finalizing bufferize conversion, we know that all tensors have been
-// converted to memrefs, thus, this op becomes an identity.
-class BufferizeToTensorOp
- : public OpConversionPattern<bufferization::ToTensorOp> {
-public:
- using OpConversionPattern::OpConversionPattern;
- LogicalResult
- matchAndRewrite(bufferization::ToTensorOp op, OpAdaptor adaptor,
- ConversionPatternRewriter &rewriter) const override {
- rewriter.replaceOp(op, adaptor.getMemref());
- return success();
- }
-};
-} // namespace
-
-namespace {
-// In a finalizing bufferize conversion, we know that all tensors have been
-// converted to memrefs, thus, this op becomes an identity.
-class BufferizeToMemrefOp
- : public OpConversionPattern<bufferization::ToMemrefOp> {
-public:
- using OpConversionPattern::OpConversionPattern;
- LogicalResult
- matchAndRewrite(bufferization::ToMemrefOp op, OpAdaptor adaptor,
- ConversionPatternRewriter &rewriter) const override {
- rewriter.replaceOp(op, adaptor.getTensor());
- return success();
- }
-};
-} // namespace
-
-void mlir::bufferization::populateEliminateBufferizeMaterializationsPatterns(
- const BufferizeTypeConverter &typeConverter, RewritePatternSet &patterns) {
- patterns.add<BufferizeToTensorOp, BufferizeToMemrefOp>(typeConverter,
- patterns.getContext());
-}
-
-namespace {
-struct FinalizingBufferizePass
- : public bufferization::impl::FinalizingBufferizeBase<
- FinalizingBufferizePass> {
- using FinalizingBufferizeBase<
- FinalizingBufferizePass>::FinalizingBufferizeBase;
-
- void runOnOperation() override {
- auto func = getOperation();
- auto *context = &getContext();
-
- BufferizeTypeConverter typeConverter;
- RewritePatternSet patterns(context);
- ConversionTarget target(*context);
-
- populateEliminateBufferizeMaterializationsPatterns(typeConverter, patterns);
-
- // If all result types are legal, and all block arguments are legal (ensured
- // by func conversion above), then all types in the program are legal.
- //
- // We also check that the operand types are legal to avoid creating invalid
- // IR. For example, this prevents
- // populateEliminateBufferizeMaterializationsPatterns from updating the
- // types of the operands to a return op without updating the enclosing
- // function.
- target.markUnknownOpDynamicallyLegal(
- [&](Operation *op) { return typeConverter.isLegal(op); });
-
- if (failed(applyFullConversion(func, target, std::move(patterns))))
- signalPassFailure();
- }
-};
static LayoutMapOption parseLayoutMapOption(const std::string &s) {
if (s == "fully-dynamic-layout-map")
@@ -331,11 +261,6 @@ std::unique_ptr<Pass> mlir::bufferization::createOneShotBufferizePass(
return std::make_unique<OneShotBufferizePass>(options);
}
-std::unique_ptr<OperationPass<func::FuncOp>>
-mlir::bufferization::createFinalizingBufferizePass() {
- return std::make_unique<FinalizingBufferizePass>();
-}
-
//===----------------------------------------------------------------------===//
// BufferizableOpInterface-based Bufferization
//===----------------------------------------------------------------------===//
diff --git a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
index abc4a4c252841b..96ccf9a9a24087 100644
--- a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
@@ -55,8 +55,6 @@ void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
// Storage specifier lowering and bufferization wrap-up.
pm.addPass(createStorageSpecifierToLLVMPass());
pm.addNestedPass<func::FuncOp>(createCanonicalizerPass());
- pm.addNestedPass<func::FuncOp>(
- mlir::bufferization::createFinalizingBufferizePass());
// GPU code generation.
const bool gpuCodegen = options.gpuTriple.hasValue();
diff --git a/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir b/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir
deleted file mode 100644
index ab18ce05e355d3..00000000000000
--- a/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir
+++ /dev/null
@@ -1,95 +0,0 @@
-// RUN: mlir-opt %s -finalizing-bufferize -split-input-file -verify-diagnostics | FileCheck %s
-
-// CHECK-LABEL: func @eliminate_materializations(
-// CHECK-SAME: %[[ARG:.*]]: memref<f32>) -> memref<f32> {
-// CHECK: return %[[ARG]] : memref<f32>
-func.func @eliminate_materializations(%arg0: memref<f32>) -> memref<f32> {
- %0 = bufferization.to_tensor %arg0 : memref<f32>
- %1 = bufferization.to_memref %0 : memref<f32>
- return %1 : memref<f32>
-}
-
-// -----
-
-func.func @unable_to_convert_lone_buffer_cast() -> memref<f32> {
- // expected-error @+1 {{failed to legalize operation 'test.source'}}
- %0 = "test.source"() : () -> tensor<f32>
- %1 = bufferization.to_memref %0 : memref<f32>
- return %1 : memref<f32>
-}
-
-// -----
-
-func.func @unable_to_convert_lone_tensor_load(%arg0: memref<f32>) {
- %0 = bufferization.to_tensor %arg0 : memref<f32>
- // expected-error @+1 {{failed to legalize operation 'test.sink'}}
- "test.sink"(%0) : (tensor<f32>) -> ()
- return
-}
-
-// -----
-
-// CHECK-LABEL: func @dyn_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[1], offset: ?>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @dyn_layout_to_no_layout_cast(%m: memref<?xf32, strided<[1], offset: ?>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[1], offset: ?>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// CHECK-LABEL: func @fancy_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[100], offset: ?>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @fancy_layout_to_no_layout_cast(%m: memref<?xf32, strided<[100], offset: ?>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[100], offset: ?>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// CHECK-LABEL: func @static_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[1], offset: 25>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @static_layout_to_no_layout_cast(%m: memref<?xf32, strided<[1], offset: 25>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[1], offset: 25>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// TODO: to_memref with layout maps not supported yet. This should fold to a
-// memref.cast.
-func.func @no_layout_to_dyn_layout_cast(%m: memref<?xf32>) -> memref<?xf32, strided<[1], offset: ?>> {
- %0 = bufferization.to_tensor %m : memref<?xf32>
- // expected-error @+1 {{failed to legalize unresolved materialization from ('memref<?xf32>') to 'memref<?xf32, strided<[1], offset: ?>>' that remained live after conversion}}
- %1 = bufferization.to_memref %0 : memref<?xf32, strided<[1], offset: ?>>
- // expected-note @below{{see existing live user here}}
- return %1 : memref<?xf32, strided<[1], offset: ?>>
-}
-
-// -----
-
-func.func @illegal_unranked_to_rank(%m: memref<*xf32>) -> memref<?xf32> {
- // expected-note @+1 {{prior use here}}
- %0 = bufferization.to_tensor %m : memref<*xf32>
- // expected-error @+1 {{expects different type than prior uses: 'tensor<?xf32>' vs 'tensor<*xf32>'}}
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
|
|
@llvm/pr-subscribers-mlir-sparse Author: Matthias Springer (matthias-springer) ChangesThe dialect conversion-based bufferization passes have been migrated to One-Shot Bufferize about two years ago. To clean up the code base, this commit removes the Note for LLVM integration: If you depend on this pass, migrate to One-Shot Bufferize or copy the pass to your codebase. Depends on #114152. Full diff: https://github.com/llvm/llvm-project/pull/114154.diff 5 Files Affected:
diff --git a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
index 72abb5b3f1f94e..8b957577292796 100644
--- a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
+++ b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.h
@@ -200,10 +200,6 @@ std::unique_ptr<Pass> createEmptyTensorToAllocTensorPass();
/// Drop all memref function results that are equivalent to a function argument.
LogicalResult dropEquivalentBufferResults(ModuleOp module);
-/// Creates a pass that finalizes a partial bufferization by removing remaining
-/// bufferization.to_tensor and bufferization.to_memref operations.
-std::unique_ptr<OperationPass<func::FuncOp>> createFinalizingBufferizePass();
-
/// Create a pass that bufferizes all ops that implement BufferizableOpInterface
/// with One-Shot Bufferize.
std::unique_ptr<Pass> createOneShotBufferizePass();
diff --git a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
index cc5463ea968fc3..3e1b55cf99767f 100644
--- a/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
+++ b/mlir/include/mlir/Dialect/Bufferization/Transforms/Passes.td
@@ -343,22 +343,6 @@ def BufferResultsToOutParams : Pass<"buffer-results-to-out-params", "ModuleOp">
let dependentDialects = ["memref::MemRefDialect"];
}
-def FinalizingBufferize : Pass<"finalizing-bufferize", "func::FuncOp"> {
- let summary = "Finalize a partial bufferization";
- let description = [{
- A bufferize pass that finalizes a partial bufferization by removing
- remaining `bufferization.to_tensor` and `bufferization.to_buffer` operations.
-
- The removal of those operations is only possible if the operations only
- exist in pairs, i.e., all uses of `bufferization.to_tensor` operations are
- `bufferization.to_buffer` operations.
-
- This pass will fail if not all operations can be removed or if any operation
- with tensor typed operands remains.
- }];
- let constructor = "mlir::bufferization::createFinalizingBufferizePass()";
-}
-
def DropEquivalentBufferResults : Pass<"drop-equivalent-buffer-results", "ModuleOp"> {
let summary = "Remove MemRef return values that are equivalent to a bbArg";
let description = [{
diff --git a/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp b/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
index 1d009b03754c52..62ce2583f4fa1d 100644
--- a/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
+++ b/mlir/lib/Dialect/Bufferization/Transforms/Bufferize.cpp
@@ -26,7 +26,6 @@
namespace mlir {
namespace bufferization {
-#define GEN_PASS_DEF_FINALIZINGBUFFERIZE
#define GEN_PASS_DEF_BUFFERIZATIONBUFFERIZE
#define GEN_PASS_DEF_ONESHOTBUFFERIZE
#include "mlir/Dialect/Bufferization/Transforms/Passes.h.inc"
@@ -98,75 +97,6 @@ void mlir::bufferization::populateBufferizeMaterializationLegality(
}
namespace {
-// In a finalizing bufferize conversion, we know that all tensors have been
-// converted to memrefs, thus, this op becomes an identity.
-class BufferizeToTensorOp
- : public OpConversionPattern<bufferization::ToTensorOp> {
-public:
- using OpConversionPattern::OpConversionPattern;
- LogicalResult
- matchAndRewrite(bufferization::ToTensorOp op, OpAdaptor adaptor,
- ConversionPatternRewriter &rewriter) const override {
- rewriter.replaceOp(op, adaptor.getMemref());
- return success();
- }
-};
-} // namespace
-
-namespace {
-// In a finalizing bufferize conversion, we know that all tensors have been
-// converted to memrefs, thus, this op becomes an identity.
-class BufferizeToMemrefOp
- : public OpConversionPattern<bufferization::ToMemrefOp> {
-public:
- using OpConversionPattern::OpConversionPattern;
- LogicalResult
- matchAndRewrite(bufferization::ToMemrefOp op, OpAdaptor adaptor,
- ConversionPatternRewriter &rewriter) const override {
- rewriter.replaceOp(op, adaptor.getTensor());
- return success();
- }
-};
-} // namespace
-
-void mlir::bufferization::populateEliminateBufferizeMaterializationsPatterns(
- const BufferizeTypeConverter &typeConverter, RewritePatternSet &patterns) {
- patterns.add<BufferizeToTensorOp, BufferizeToMemrefOp>(typeConverter,
- patterns.getContext());
-}
-
-namespace {
-struct FinalizingBufferizePass
- : public bufferization::impl::FinalizingBufferizeBase<
- FinalizingBufferizePass> {
- using FinalizingBufferizeBase<
- FinalizingBufferizePass>::FinalizingBufferizeBase;
-
- void runOnOperation() override {
- auto func = getOperation();
- auto *context = &getContext();
-
- BufferizeTypeConverter typeConverter;
- RewritePatternSet patterns(context);
- ConversionTarget target(*context);
-
- populateEliminateBufferizeMaterializationsPatterns(typeConverter, patterns);
-
- // If all result types are legal, and all block arguments are legal (ensured
- // by func conversion above), then all types in the program are legal.
- //
- // We also check that the operand types are legal to avoid creating invalid
- // IR. For example, this prevents
- // populateEliminateBufferizeMaterializationsPatterns from updating the
- // types of the operands to a return op without updating the enclosing
- // function.
- target.markUnknownOpDynamicallyLegal(
- [&](Operation *op) { return typeConverter.isLegal(op); });
-
- if (failed(applyFullConversion(func, target, std::move(patterns))))
- signalPassFailure();
- }
-};
static LayoutMapOption parseLayoutMapOption(const std::string &s) {
if (s == "fully-dynamic-layout-map")
@@ -331,11 +261,6 @@ std::unique_ptr<Pass> mlir::bufferization::createOneShotBufferizePass(
return std::make_unique<OneShotBufferizePass>(options);
}
-std::unique_ptr<OperationPass<func::FuncOp>>
-mlir::bufferization::createFinalizingBufferizePass() {
- return std::make_unique<FinalizingBufferizePass>();
-}
-
//===----------------------------------------------------------------------===//
// BufferizableOpInterface-based Bufferization
//===----------------------------------------------------------------------===//
diff --git a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
index abc4a4c252841b..96ccf9a9a24087 100644
--- a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
@@ -55,8 +55,6 @@ void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
// Storage specifier lowering and bufferization wrap-up.
pm.addPass(createStorageSpecifierToLLVMPass());
pm.addNestedPass<func::FuncOp>(createCanonicalizerPass());
- pm.addNestedPass<func::FuncOp>(
- mlir::bufferization::createFinalizingBufferizePass());
// GPU code generation.
const bool gpuCodegen = options.gpuTriple.hasValue();
diff --git a/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir b/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir
deleted file mode 100644
index ab18ce05e355d3..00000000000000
--- a/mlir/test/Dialect/Bufferization/Transforms/finalizing-bufferize.mlir
+++ /dev/null
@@ -1,95 +0,0 @@
-// RUN: mlir-opt %s -finalizing-bufferize -split-input-file -verify-diagnostics | FileCheck %s
-
-// CHECK-LABEL: func @eliminate_materializations(
-// CHECK-SAME: %[[ARG:.*]]: memref<f32>) -> memref<f32> {
-// CHECK: return %[[ARG]] : memref<f32>
-func.func @eliminate_materializations(%arg0: memref<f32>) -> memref<f32> {
- %0 = bufferization.to_tensor %arg0 : memref<f32>
- %1 = bufferization.to_memref %0 : memref<f32>
- return %1 : memref<f32>
-}
-
-// -----
-
-func.func @unable_to_convert_lone_buffer_cast() -> memref<f32> {
- // expected-error @+1 {{failed to legalize operation 'test.source'}}
- %0 = "test.source"() : () -> tensor<f32>
- %1 = bufferization.to_memref %0 : memref<f32>
- return %1 : memref<f32>
-}
-
-// -----
-
-func.func @unable_to_convert_lone_tensor_load(%arg0: memref<f32>) {
- %0 = bufferization.to_tensor %arg0 : memref<f32>
- // expected-error @+1 {{failed to legalize operation 'test.sink'}}
- "test.sink"(%0) : (tensor<f32>) -> ()
- return
-}
-
-// -----
-
-// CHECK-LABEL: func @dyn_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[1], offset: ?>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @dyn_layout_to_no_layout_cast(%m: memref<?xf32, strided<[1], offset: ?>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[1], offset: ?>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// CHECK-LABEL: func @fancy_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[100], offset: ?>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @fancy_layout_to_no_layout_cast(%m: memref<?xf32, strided<[100], offset: ?>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[100], offset: ?>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// CHECK-LABEL: func @static_layout_to_no_layout_cast(
-// CHECK-SAME: %[[arg:.*]]: memref<?xf32, strided<[1], offset: 25>>)
-// CHECK: %[[c0:.*]] = arith.constant 0 : index
-// CHECK: %[[dim:.*]] = memref.dim %[[arg]], %[[c0]]
-// CHECK: %[[alloc:.*]] = memref.alloc(%[[dim]]) : memref<?xf32>
-// CHECK: memref.copy %[[arg]], %[[alloc]]
-// CHECK: return %[[alloc]]
-func.func @static_layout_to_no_layout_cast(%m: memref<?xf32, strided<[1], offset: 25>>) -> memref<?xf32> {
- %0 = bufferization.to_tensor %m : memref<?xf32, strided<[1], offset: 25>>
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
-
-// -----
-
-// TODO: to_memref with layout maps not supported yet. This should fold to a
-// memref.cast.
-func.func @no_layout_to_dyn_layout_cast(%m: memref<?xf32>) -> memref<?xf32, strided<[1], offset: ?>> {
- %0 = bufferization.to_tensor %m : memref<?xf32>
- // expected-error @+1 {{failed to legalize unresolved materialization from ('memref<?xf32>') to 'memref<?xf32, strided<[1], offset: ?>>' that remained live after conversion}}
- %1 = bufferization.to_memref %0 : memref<?xf32, strided<[1], offset: ?>>
- // expected-note @below{{see existing live user here}}
- return %1 : memref<?xf32, strided<[1], offset: ?>>
-}
-
-// -----
-
-func.func @illegal_unranked_to_rank(%m: memref<*xf32>) -> memref<?xf32> {
- // expected-note @+1 {{prior use here}}
- %0 = bufferization.to_tensor %m : memref<*xf32>
- // expected-error @+1 {{expects different type than prior uses: 'tensor<?xf32>' vs 'tensor<*xf32>'}}
- %1 = bufferization.to_memref %0 : memref<?xf32>
- return %1 : memref<?xf32>
-}
|
matthias-springer
force-pushed
the
users/matthias-springer/remove-finalizing-bufferize
branch
from
e353503 to
7c2f6b6
Compare
matthias-springer
added a commit
that referenced
this pull request
Oct 30, 2024
… parts This commit removes the last remaining components of the dialect conversion-based bufferization passes. Note for LLVM integration: If you depend on these components, migrate to One-Shot Bufferize or copy them to your codebase. Depends on #114154.
javedabsar1
reviewed
Oct 30, 2024
There was a problem hiding this comment.
Hi Matthias.
Will you delete references in docs in a different diff ? https://github.com/llvm/llvm-project/blob/main/mlir/docs/Bufferization.md?plain=1#L561
matthias-springer
force-pushed
the
users/matthias-springer/remove-func-bufferize-pass
branch
from
be96e32 to
d491c19
Compare
matthias-springer
force-pushed
the
users/matthias-springer/remove-finalizing-bufferize
branch
from
7c2f6b6 to
e0b61b5
Compare
The previous PR #114152 already deletes the entire bufferization documentation for the partial bufferization passes. #114152 deletes the last partial bufferization pass. The other two PR (including this one) remove some infrastructure that is no longer needed. The reason why this PR is not part of #114152 is to give downstream users more time to adapt to the changes. (I'm going to leave a few days between merging these.) |
javedabsar1
approved these changes
Oct 30, 2024
matthias-springer
force-pushed
the
users/matthias-springer/remove-func-bufferize-pass
branch
from
d491c19 to
34218ee
Compare
matthias-springer
force-pushed
the
users/matthias-springer/remove-finalizing-bufferize
branch
from
e0b61b5 to
1e194b3
Compare
matthias-springer
added a commit
that referenced
this pull request
Nov 14, 2024
… parts This commit removes the last remaining components of the dialect conversion-based bufferization passes. Note for LLVM integration: If you depend on these components, migrate to One-Shot Bufferize or copy them to your codebase. Depends on #114154.
Base automatically changed from
users/matthias-springer/remove-func-bufferize-pass
to
main
November 20, 2024 00:29
matthias-springer
force-pushed
the
users/matthias-springer/remove-finalizing-bufferize
branch
2 times, most recently
from
bf089ca to
1beeac3
Compare
The dialect conversion-based bufferization passes have been migrated to One-Shot Bufferize about two years ago. To clean up the code base, this commit removes the `finalizing-bufferize` pass, one of the few remaining parts of the old infrastructure. Most bufferization passes have already been removed. Note for LLVM integration: If you depend on this pass, migrate to One-Shot Bufferize or copy the pass to your codebase. Depends on #114152.
matthias-springer
force-pushed
the
users/matthias-springer/remove-finalizing-bufferize
branch
from
1beeac3 to
5353a10
Compare
matthias-springer
deleted the
users/matthias-springer/remove-finalizing-bufferize
branch
matthias-springer
added a commit
that referenced
this pull request
Nov 22, 2024
… parts This commit removes the last remaining components of the dialect conversion-based bufferization passes. Note for LLVM integration: If you depend on these components, migrate to One-Shot Bufferize or copy them to your codebase. Depends on #114154.
matthias-springer
added a commit
that referenced
this pull request
Nov 27, 2024
… parts This commit removes the last remaining components of the dialect conversion-based bufferization passes. Note for LLVM integration: If you depend on these components, migrate to One-Shot Bufferize or copy them to your codebase. Depends on #114154.
zezhang
added a commit
to llvm/mlir-tcp
that referenced
this pull request
Feb 1, 2025
[LLVM](https://github.com/llvm/llvm-project/tree/5d6d982df61d16b6d498e6d59dd91c059679d3d8) (Jan 27) [Torch-MLIR](https://github.com/llvm/torch-mlir/tree/169032010793ee7fe3e305ab920e4119fdfc3b11) (Jan 30) [StableHlo](https://github.com/openxla/stablehlo/tree/b62dc66da9946b4c400c0d99c9d5bb8e04edaee6) (Jan 29) Remove the createFinalizingBufferizePass since it's no longer exist in [upstream](llvm/llvm-project#114154) Add [createConvertControlFlowToLLVMPass](https://github.com/llvm/llvm-project/blob/bd51fa2ad2982811fe76a9d2c0a94073c65e895d/mlir/test/lib/Dialect/LLVM/TestLowerToLLVM.cpp#L81) to lower cf.br and cf.cond_br
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The dialect conversion-based bufferization passes have been migrated to One-Shot Bufferize about two years ago. To clean up the code base, this commit removes the
finalizing-bufferizepass, one of the few remaining parts of the old infrastructure. Most bufferization passes have already been removed.Note for LLVM integration: If you depend on this pass, migrate to One-Shot Bufferize or copy the pass to your codebase.
Depends on #114152.