[Mlir-commits] [mlir] 3d51010 - Revert "[mlir][bufferization] Switch tests to new deallocation pass pipeline (#66471)"
Martin Erhart
llvmlistbot at llvm.org
Fri Sep 15 02:20:15 PDT 2023
Author: Martin Erhart
Date: 2023-09-15T09:19:54Z
New Revision: 3d51010a3350660160981c6b8e624dcc87c208a3
URL: https://github.com/llvm/llvm-project/commit/3d51010a3350660160981c6b8e624dcc87c208a3
DIFF: https://github.com/llvm/llvm-project/commit/3d51010a3350660160981c6b8e624dcc87c208a3.diff
LOG: Revert "[mlir][bufferization] Switch tests to new deallocation pass pipeline (#66471)"
This reverts commit ea42b49f1069ec647cbc67e6360026bc4e5d9a60.
Some GPU integration tests are failing that I didn't observe locally.
Reverting until I have a fix.
Added:
Modified:
mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-allow-return-allocs.mlir
mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-compat.mlir
mlir/test/Dialect/Bufferization/Transforms/one-shot-module-bufferize-out-params.mlir
mlir/test/Dialect/SCF/one-shot-bufferize.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-collapse-tensor.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-elementwise.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-expand-tensor.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-padtensor.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert-multiple-uses.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-e2e.mlir
mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir
Removed:
################################################################################
diff --git a/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-allow-return-allocs.mlir b/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-allow-return-allocs.mlir
index bccff8ef8d65aaa..ab9360e45c65ff3 100644
--- a/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-allow-return-allocs.mlir
+++ b/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-allow-return-allocs.mlir
@@ -1,4 +1,4 @@
-// RUN: mlir-opt %s -one-shot-bufferize="allow-return-allocs allow-unknown-ops" -canonicalize -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -one-shot-bufferize="allow-return-allocs allow-unknown-ops" -buffer-deallocation -canonicalize -split-input-file | FileCheck %s
// Run fuzzer with
diff erent seeds.
// RUN: mlir-opt %s -one-shot-bufferize="allow-return-allocs test-analysis-only analysis-fuzzer-seed=23" -split-input-file -o /dev/null
@@ -14,17 +14,20 @@ func.func @buffer_not_deallocated(%t : tensor<?xf32>, %c : i1) -> tensor<?xf32>
// CHECK: %[[some_op:.*]] = "test.some_op"
// CHECK: %[[alloc:.*]] = memref.alloc(%[[some_op]])
// CHECK: %[[casted:.*]] = memref.cast %[[alloc]]
+ // CHECK-NOT: dealloc
// CHECK: scf.yield %[[casted]]
%sz = "test.some_op"() : () -> (index)
%0 = bufferization.alloc_tensor(%sz) : tensor<?xf32>
scf.yield %0 : tensor<?xf32>
} else {
// CHECK: } else {
- // CHECK: scf.yield %[[m]]
+ // CHECK: %[[cloned:.*]] = bufferization.clone %[[m]]
+ // CHECK: scf.yield %[[cloned]]
scf.yield %t : tensor<?xf32>
}
// CHECK: }
// CHECK: %[[r_tensor:.*]] = bufferization.to_tensor %[[r]]
+ // CHECK: memref.dealloc %[[r]]
// CHECK: return %[[r_tensor]]
return %r : tensor<?xf32>
}
@@ -39,7 +42,8 @@ func.func @write_to_alloc_tensor_or_readonly_tensor(%arg0: tensor<i32>,
{
// CHECK: %[[arg0_m:.*]] = bufferization.to_memref %[[arg0]]
// CHECK: %[[r:.*]] = scf.if {{.*}} {
- // CHECK: scf.yield %[[arg0_m]]
+ // CHECK: %[[clone:.*]] = bufferization.clone %[[arg0_m]]
+ // CHECK: scf.yield %[[clone]]
// CHECK: } else {
// CHECK: %[[alloc:.*]] = memref.alloc
// CHECK: memref.store %{{.*}}, %[[alloc]]
@@ -47,6 +51,7 @@ func.func @write_to_alloc_tensor_or_readonly_tensor(%arg0: tensor<i32>,
// CHECK: scf.yield %[[casted]]
// CHECK: }
// CHECK: %[[r_t:.*]] = bufferization.to_tensor %[[r]]
+ // CHECK: memref.dealloc %[[r]]
// CHECK: return %[[r_t]]
%3 = scf.if %cond -> (tensor<i32>) {
scf.yield %arg0 : tensor<i32>
diff --git a/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-compat.mlir b/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-compat.mlir
index 9ebd60f50328e14..06c79d450cea705 100644
--- a/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-compat.mlir
+++ b/mlir/test/Dialect/Bufferization/Transforms/one-shot-bufferize-compat.mlir
@@ -5,7 +5,7 @@
// RUN: mlir-opt %s \
// RUN: -one-shot-bufferize="allow-unknown-ops create-deallocs=0" \
-// RUN: -buffer-deallocation-pipeline | \
+// RUN: -buffer-deallocation | \
// RUN: FileCheck %s --check-prefix=CHECK-BUFFERDEALLOC
// CHECK-NODEALLOC-LABEL: func @out_of_place_bufferization
diff --git a/mlir/test/Dialect/Bufferization/Transforms/one-shot-module-bufferize-out-params.mlir b/mlir/test/Dialect/Bufferization/Transforms/one-shot-module-bufferize-out-params.mlir
index 4e4340c9db8acea..5d876e935df1af6 100644
--- a/mlir/test/Dialect/Bufferization/Transforms/one-shot-module-bufferize-out-params.mlir
+++ b/mlir/test/Dialect/Bufferization/Transforms/one-shot-module-bufferize-out-params.mlir
@@ -1,6 +1,6 @@
-// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=fully-dynamic-layout-map" -drop-equivalent-buffer-results -buffer-results-to-out-params -split-input-file | FileCheck %s
-// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=identity-layout-map" -drop-equivalent-buffer-results -buffer-results-to-out-params -split-input-file | FileCheck %s --check-prefix=CHECK-NO-LAYOUT
-// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=infer-layout-map" -drop-equivalent-buffer-results -split-input-file | FileCheck %s --check-prefix=CHECK-BASELINE
+// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=fully-dynamic-layout-map" -drop-equivalent-buffer-results -buffer-results-to-out-params -buffer-deallocation -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=identity-layout-map" -drop-equivalent-buffer-results -buffer-results-to-out-params -buffer-deallocation -split-input-file | FileCheck %s --check-prefix=CHECK-NO-LAYOUT
+// RUN: mlir-opt %s -one-shot-bufferize="bufferize-function-boundaries allow-return-allocs function-boundary-type-conversion=infer-layout-map" -drop-equivalent-buffer-results -buffer-deallocation -split-input-file | FileCheck %s --check-prefix=CHECK-BASELINE
// Note: function-boundary-type-conversion=infer-layout-map with
// promote-buffer-results-to-out-params is an unsupported combination.
@@ -18,6 +18,7 @@
// CHECK: memref.store %{{.*}}, %[[alloc]]
// CHECK: %[[casted:.*]] = memref.cast %[[alloc]]
// CHECK: memref.copy %[[casted]], %[[arg1]]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK: return
// CHECK: }
@@ -28,6 +29,7 @@
// CHECK-NO-LAYOUT: memref.copy %[[arg0]], %[[alloc]]
// CHECK-NO-LAYOUT: memref.store {{.*}}, %[[alloc]]
// CHECK-NO-LAYOUT: memref.copy %[[alloc]], %[[arg1]]
+// CHECK-NO-LAYOUT: memref.dealloc %[[alloc]]
// CHECK-BASELINE-LABEL: func @callee(
// CHECK-BASELINE-SAME: %[[arg0:.*]]: memref<5xf32, strided<[?], offset: ?>>) -> memref<5xf32> {
@@ -51,6 +53,7 @@ func.func @callee(%t: tensor<5xf32>) -> (tensor<5xf32>, tensor<5xf32>) {
// CHECK: call @callee(%[[arg0]], %[[casted]])
// CHECK: %[[l1:.*]] = memref.load %[[arg0]]
// CHECK: %[[l2:.*]] = memref.load %[[casted]]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK: return %[[l1]], %[[l2]]
// CHECK: }
@@ -75,6 +78,7 @@ func.func @main(%t: tensor<5xf32>) -> (f32, f32) {
// CHECK: %[[subview:.*]] = memref.subview %[[alloc]]{{.*}} : memref<10x20xf32> to memref<2x5xf32, strided<[20, 1], offset: ?>>
// CHECK: %[[casted:.*]] = memref.cast %[[subview]]
// CHECK: memref.copy %[[casted]], %[[r]]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK-NO-LAYOUT-LABEL: func @callee(
// CHECK-NO-LAYOUT-SAME: %{{.*}}: index,
@@ -86,7 +90,9 @@ func.func @main(%t: tensor<5xf32>) -> (f32, f32) {
// value and function signature.
// CHECK-NO-LAYOUT: %[[alloc2:.*]] = memref.alloc() : memref<2x5xf32>
// CHECK-NO-LAYOUT: memref.copy %[[subview]], %[[alloc2]]
+// CHECK-NO-LAYOUT: memref.dealloc %[[alloc]]
// CHECK-NO-LAYOUT: memref.copy %[[alloc2]], %[[r]]
+// CHECK-NO-LAYOUT: memref.dealloc %[[alloc2]]
// CHECK-BASELINE-LABEL: func @callee(
// CHECK-BASELINE-SAME: %{{.*}}: index) -> memref<2x5xf32, strided<[20, 1], offset: ?>> {
@@ -104,11 +110,13 @@ func.func @callee(%idx: index) -> tensor<2x5xf32> {
// CHECK: %[[casted:.*]] = memref.cast %[[alloc]] : memref<2x5xf32> to memref<2x5xf32, strided<[?, ?], offset: ?>>
// CHECK: call @callee(%{{.*}}, %[[casted]])
// CHECK: memref.load %[[casted]]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK-NO-LAYOUT: func @main(
// CHECK-NO-LAYOUT: %[[alloc:.*]] = memref.alloc() : memref<2x5xf32>
// CHECK-NO-LAYOUT: call @callee(%{{.*}}, %[[alloc]])
// CHECK-NO-LAYOUT: memref.load %[[alloc]]
+// CHECK-NO-LAYOUT: memref.dealloc
// CHECK-BASELINE: func @main(
// CHECK-BASELINE: %[[call:.*]] = call @callee
diff --git a/mlir/test/Dialect/SCF/one-shot-bufferize.mlir b/mlir/test/Dialect/SCF/one-shot-bufferize.mlir
index a8c488461c74edd..b01592dcad74488 100644
--- a/mlir/test/Dialect/SCF/one-shot-bufferize.mlir
+++ b/mlir/test/Dialect/SCF/one-shot-bufferize.mlir
@@ -1,4 +1,5 @@
-// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs bufferize-function-boundaries" -cse -canonicalize -drop-equivalent-buffer-results -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs bufferize-function-boundaries" -cse -canonicalize -drop-equivalent-buffer-results -buffer-deallocation -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs bufferize-function-boundaries" -drop-equivalent-buffer-results -split-input-file | FileCheck %s --check-prefix=CHECK-NO-DEALLOC-PASS
// Run fuzzer with
diff erent seeds.
// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs test-analysis-only analysis-fuzzer-seed=23 bufferize-function-boundaries" -split-input-file -o /dev/null
@@ -6,7 +7,7 @@
// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs test-analysis-only analysis-fuzzer-seed=91 bufferize-function-boundaries" -split-input-file -o /dev/null
// Test bufferization using memref types that have no layout map.
-// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs unknown-type-conversion=identity-layout-map function-boundary-type-conversion=identity-layout-map bufferize-function-boundaries" -split-input-file -o /dev/null
+// RUN: mlir-opt %s -allow-unregistered-dialect -one-shot-bufferize="allow-return-allocs unknown-type-conversion=identity-layout-map function-boundary-type-conversion=identity-layout-map bufferize-function-boundaries" -buffer-deallocation -split-input-file -o /dev/null
// CHECK-LABEL: func @scf_for_yield_only(
// CHECK-SAME: %[[A:[a-zA-Z0-9]*]]: memref<?xf32, strided<[?], offset: ?>>,
@@ -51,7 +52,8 @@ func.func @scf_for_is_reading(%A : tensor<?xf32>, %B : tensor<?xf32>,
// CHECK: %[[alloc:.*]] = memref.alloc
// CHECK: memref.copy %[[A]], %[[alloc]]
- // CHECK: scf.for {{.*}} iter_args(%{{.*}} = %[[alloc]])
+ // CHECK: %[[clone:.*]] = bufferization.clone %[[alloc]]
+ // CHECK: scf.for {{.*}} iter_args(%{{.*}} = %[[clone]])
%0 = scf.for %iv = %lb to %ub step %c1 iter_args(%1 = %A) -> tensor<?xf32> {
%r = linalg.fill ins(%cst : f32) outs(%1 : tensor<?xf32>) -> tensor<?xf32>
scf.yield %B : tensor<?xf32>
@@ -233,6 +235,7 @@ func.func @scf_if_non_equiv_yields(
// CHECK-LABEL: func @scf_execute_region_yield_non_equivalent(
// CHECK: %[[alloc:.*]] = memref.alloc(%{{.*}})
// CHECK: %[[r:.*]] = memref.load %[[alloc]][%{{.*}}]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK: return %[[r]]
func.func @scf_execute_region_yield_non_equivalent(%i: index, %j: index) -> f32 {
%r = scf.execute_region -> (tensor<?xf32>) {
@@ -253,11 +256,16 @@ func.func @scf_execute_region_yield_non_equivalent(%i: index, %j: index) -> f32
// CHECK-SAME: %[[t:.*]]: memref<?xf32
// CHECK: %[[alloc:.*]] = memref.alloc(%{{.*}})
// CHECK: memref.copy %[[t]], %[[alloc]]
-// CHECK: %[[for:.*]] = scf.for {{.*}} iter_args(%[[iter:.*]] = %[[t]])
+// CHECK: %[[cloned:.*]] = bufferization.clone %[[t]]
+// CHECK: %[[for:.*]] = scf.for {{.*}} iter_args(%[[iter:.*]] = %[[cloned]])
+// CHECK-DAG: memref.dealloc %[[iter]]
// CHECK-DAG: %[[alloc2:.*]] = memref.alloc(%{{.*}})
// CHECK: memref.copy %[[alloc]], %[[alloc2]]
// CHECK: %[[alloc2_casted:.*]] = memref.cast %[[alloc2]]
-// CHECK: scf.yield %[[alloc2_casted]]
+// CHECK: %[[cloned2:.*]] = bufferization.clone %[[alloc2_casted]]
+// CHECK: memref.dealloc %[[alloc2]]
+// CHECK: scf.yield %[[cloned2]]
+// CHECK: memref.dealloc %[[alloc]]
// CHECK: return %[[for]]
func.func @scf_for_yield_non_equivalent(
%t: tensor<?xf32>, %lb : index, %ub : index, %step : index) -> tensor<?xf32> {
@@ -276,14 +284,19 @@ func.func @scf_for_yield_non_equivalent(
// CHECK-LABEL: func @scf_for_yield_allocation(
// CHECK-SAME: %[[t:.*]]: memref<?xf32
-// CHECK: %[[for:.*]] = scf.for {{.*}} iter_args(%[[iter:.*]] = %[[t]])
+// CHECK: %[[cloned:.*]] = bufferization.clone %[[t]]
+// CHECK: %[[for:.*]] = scf.for {{.*}} iter_args(%[[iter:.*]] = %[[cloned]])
// This alloc is for the bufferization.alloc_tensor.
// CHECK-DAG: %[[alloc2:.*]] = memref.alloc(%{{.*}})
+// CHECK-DAG: memref.dealloc %[[iter]]
// This alloc is for the scf.yield.
// CHECK: %[[alloc3:.*]] = memref.alloc(%{{.*}})
// CHECK: memref.copy %[[alloc2]], %[[alloc3]]
+// CHECK: memref.dealloc %[[alloc2]]
// CHECK: %[[casted3:.*]] = memref.cast %[[alloc3]]
-// CHECK: scf.yield %[[casted3]]
+// CHECK: %[[cloned3:.*]] = bufferization.clone %[[casted3]]
+// CHECK: memref.dealloc %[[alloc3]]
+// CHECK: scf.yield %[[cloned3]]
// CHECK: return %[[for]]
func.func @scf_for_yield_allocation(%t: tensor<?xf32>, %lb : index, %ub : index,
%step : index) -> tensor<?xf32> {
@@ -307,7 +320,9 @@ func.func @scf_for_swapping_yields(
%C : tensor<4xf32>, %lb : index, %ub : index, %step : index)
-> (f32, f32)
{
-// CHECK: %[[for:.*]]:2 = scf.for {{.*}} iter_args(%[[iter1:.*]] = %[[A]], %[[iter2:.*]] = %[[B]])
+// CHECK-DAG: %[[clone1:.*]] = bufferization.clone %[[A]]
+// CHECK-DAG: %[[clone2:.*]] = bufferization.clone %[[B]]
+// CHECK: %[[for:.*]]:2 = scf.for {{.*}} iter_args(%[[iter1:.*]] = %[[clone1]], %[[iter2:.*]] = %[[clone2]])
%r0:2 = scf.for %i = %lb to %ub step %step iter_args(%tA = %A, %tB = %B)
-> (tensor<?xf32>, tensor<?xf32>)
{
@@ -320,17 +335,25 @@ func.func @scf_for_swapping_yields(
// CHECK: %[[alloc2:.*]] = memref.alloc(%{{.*}})
// CHECK: memref.copy %[[iter2]], %[[alloc2]]
+// CHECK: memref.dealloc %[[iter2]]
// CHECK: %[[alloc1:.*]] = memref.alloc(%{{.*}})
// CHECK: memref.copy %[[iter1]], %[[alloc1]]
+// CHECK: memref.dealloc %[[iter1]]
// CHECK: %[[casted2:.*]] = memref.cast %[[alloc2]]
// CHECK: %[[casted1:.*]] = memref.cast %[[alloc1]]
-// CHECK: scf.yield %[[casted2]], %[[casted1]]
+// CHECK: %[[cloned1:.*]] = bufferization.clone %[[casted1]]
+// CHECK: memref.dealloc %[[alloc1]]
+// CHECK: %[[cloned2:.*]] = bufferization.clone %[[casted2]]
+// CHECK: memref.dealloc %[[alloc2]]
+// CHECK: scf.yield %[[cloned2]], %[[cloned1]]
// Yield tensors in
diff erent order.
scf.yield %ttB, %ttA : tensor<?xf32>, tensor<?xf32>
}
// CHECK: %[[r0:.*]] = memref.load %[[for]]#0
+// CHECK: memref.dealloc %[[for]]#0
// CHECK: %[[r1:.*]] = memref.load %[[for]]#1
+// CHECK: memref.dealloc %[[for]]#1
%f0 = tensor.extract %r0#0[%step] : tensor<?xf32>
%f1 = tensor.extract %r0#1[%step] : tensor<?xf32>
// CHECK: return %[[r0]], %[[r1]]
@@ -376,15 +399,23 @@ func.func @scf_while_non_equiv_condition(%arg0: tensor<5xi1>,
%idx: index)
-> (tensor<5xi1>, tensor<5xi1>)
{
- // CHECK: %[[loop:.*]]:2 = scf.while (%[[w0:.*]] = %[[arg0]], %[[w1:.*]] = %[[arg1]]) {{.*}} {
+ // CHECK: %[[clone1:.*]] = bufferization.clone %[[arg1]]
+ // CHECK: %[[clone0:.*]] = bufferization.clone %[[arg0]]
+ // CHECK: %[[loop:.*]]:2 = scf.while (%[[w0:.*]] = %[[clone0]], %[[w1:.*]] = %[[clone1]]) {{.*}} {
%r0, %r1 = scf.while (%w0 = %arg0, %w1 = %arg1)
: (tensor<5xi1>, tensor<5xi1>) -> (tensor<5xi1>, tensor<5xi1>) {
// CHECK: %[[condition:.*]] = memref.load %[[w0]]
// CHECK: %[[a1:.*]] = memref.alloc() {{.*}} : memref<5xi1>
// CHECK: memref.copy %[[w1]], %[[a1]]
+ // CHECK: memref.dealloc %[[w1]]
// CHECK: %[[a0:.*]] = memref.alloc() {{.*}} : memref<5xi1>
// CHECK: memref.copy %[[w0]], %[[a0]]
- // CHECK: scf.condition(%[[condition]]) %[[a1]], %[[a0]]
+ // CHECK: memref.dealloc %[[w0]]
+ // CHECK: %[[cloned1:.*]] = bufferization.clone %[[a1]]
+ // CHECK: memref.dealloc %[[a1]]
+ // CHECK: %[[cloned0:.*]] = bufferization.clone %[[a0]]
+ // CHECK: memref.dealloc %[[a0]]
+ // CHECK: scf.condition(%[[condition]]) %[[cloned1]], %[[cloned0]]
%condition = tensor.extract %w0[%idx] : tensor<5xi1>
scf.condition(%condition) %w1, %w0 : tensor<5xi1>, tensor<5xi1>
} do {
@@ -394,7 +425,11 @@ func.func @scf_while_non_equiv_condition(%arg0: tensor<5xi1>,
// CHECK: memref.store %{{.*}}, %[[b0]]
// CHECK: %[[casted0:.*]] = memref.cast %[[b0]] : memref<5xi1> to memref<5xi1, strided{{.*}}>
// CHECK: %[[casted1:.*]] = memref.cast %[[b1]] : memref<5xi1> to memref<5xi1, strided{{.*}}>
- // CHECK: scf.yield %[[casted0]], %[[casted1]]
+ // CHECK: %[[cloned2:.*]] = bufferization.clone %[[casted1]]
+ // CHECK: memref.dealloc %[[b1]]
+ // CHECK: %[[cloned3:.*]] = bufferization.clone %[[casted0]]
+ // CHECK: memref.dealloc %[[b0]]
+ // CHECK: scf.yield %[[cloned3]], %[[cloned2]]
// CHECK: }
%pos = "dummy.some_op"() : () -> (index)
%val = "dummy.another_op"() : () -> (i1)
@@ -417,15 +452,23 @@ func.func @scf_while_non_equiv_condition_and_body(%arg0: tensor<5xi1>,
%idx: index)
-> (tensor<5xi1>, tensor<5xi1>)
{
- // CHECK: %[[loop:.*]]:2 = scf.while (%[[w0:.*]] = %[[arg0]], %[[w1:.*]] = %[[arg1]]) {{.*}} {
+ // CHECK-DAG: %[[clone1:.*]] = bufferization.clone %[[arg1]]
+ // CHECK-DAG: %[[clone0:.*]] = bufferization.clone %[[arg0]]
+ // CHECK: %[[loop:.*]]:2 = scf.while (%[[w0:.*]] = %[[clone0]], %[[w1:.*]] = %[[clone1]]) {{.*}} {
%r0, %r1 = scf.while (%w0 = %arg0, %w1 = %arg1)
: (tensor<5xi1>, tensor<5xi1>) -> (tensor<5xi1>, tensor<5xi1>) {
// CHECK: %[[condition:.*]] = memref.load %[[w0]]
// CHECK: %[[a1:.*]] = memref.alloc() {{.*}} : memref<5xi1>
// CHECK: memref.copy %[[w1]], %[[a1]]
+ // CHECK: memref.dealloc %[[w1]]
// CHECK: %[[a0:.*]] = memref.alloc() {{.*}} : memref<5xi1>
// CHECK: memref.copy %[[w0]], %[[a0]]
- // CHECK: scf.condition(%[[condition]]) %[[a1]], %[[a0]]
+ // CHECK: memref.dealloc %[[w0]]
+ // CHECK: %[[cloned1:.*]] = bufferization.clone %[[a1]]
+ // CHECK: memref.dealloc %[[a1]]
+ // CHECK: %[[cloned0:.*]] = bufferization.clone %[[a0]]
+ // CHECK: memref.dealloc %[[a0]]
+ // CHECK: scf.condition(%[[condition]]) %[[cloned1]], %[[cloned0]]
%condition = tensor.extract %w0[%idx] : tensor<5xi1>
scf.condition(%condition) %w1, %w0 : tensor<5xi1>, tensor<5xi1>
} do {
@@ -435,7 +478,11 @@ func.func @scf_while_non_equiv_condition_and_body(%arg0: tensor<5xi1>,
// CHECK: memref.store %{{.*}}, %[[b0]]
// CHECK: %[[casted1:.*]] = memref.cast %[[b1]]
// CHECK: %[[casted0:.*]] = memref.cast %[[b0]]
- // CHECK: scf.yield %[[casted1]], %[[casted0]]
+ // CHECK: %[[cloned1:.*]] = bufferization.clone %[[casted1]]
+ // CHECK: memref.dealloc %[[b1]]
+ // CHECK: %[[cloned0:.*]] = bufferization.clone %[[casted0]]
+ // CHECK: memref.dealloc %[[b0]]
+ // CHECK: scf.yield %[[cloned1]], %[[cloned0]]
// CHECK: }
%pos = "dummy.some_op"() : () -> (index)
%val = "dummy.another_op"() : () -> (i1)
@@ -451,7 +498,9 @@ func.func @scf_while_non_equiv_condition_and_body(%arg0: tensor<5xi1>,
// CHECK-LABEL: func @scf_while_iter_arg_result_mismatch(
// CHECK-SAME: %[[arg0:.*]]: memref<5xi1, strided{{.*}}>, %[[arg1:.*]]: memref<5xi1, strided{{.*}}>
-// CHECK: scf.while (%[[arg3:.*]] = %[[arg1]]) : (memref<5xi1, strided{{.*}}) -> () {
+// CHECK: %[[clone:.*]] = bufferization.clone %[[arg1]]
+// CHECK: scf.while (%[[arg3:.*]] = %[[clone]]) : (memref<5xi1, strided{{.*}}) -> () {
+// CHECK-DAG: memref.dealloc %[[arg3]]
// CHECK-DAG: %[[load:.*]] = memref.load %[[arg0]]
// CHECK: scf.condition(%[[load]])
// CHECK: } do {
@@ -459,7 +508,9 @@ func.func @scf_while_non_equiv_condition_and_body(%arg0: tensor<5xi1>,
// CHECK: memref.copy %[[arg0]], %[[alloc2]]
// CHECK: memref.store %{{.*}}, %[[alloc2]]
// CHECK: %[[casted:.*]] = memref.cast %[[alloc2]] : memref<5xi1> to memref<5xi1, strided{{.*}}>
-// CHECK: scf.yield %[[casted]]
+// CHECK: %[[cloned:.*]] = bufferization.clone %[[casted]]
+// CHECK: memref.dealloc %[[alloc2]]
+// CHECK: scf.yield %[[cloned]]
// CHECK: }
func.func @scf_while_iter_arg_result_mismatch(%arg0: tensor<5xi1>,
%arg1: tensor<5xi1>,
@@ -558,6 +609,7 @@ func.func @parallel_insert_slice_with_conflict(
// CHECK: %[[load:.*]] = memref.load %[[arg2]]
// CHECK: %[[load2:.*]] = memref.load %[[alloc1]]
+ // CHECK: memref.dealloc %[[alloc1]]
%f = tensor.extract %arg2[%c0] : tensor<?xf32>
%f2 = tensor.extract %2[%c0] : tensor<?xf32>
@@ -669,12 +721,15 @@ func.func @scf_if_memory_space(%c: i1, %f: f32, %cst: f32) -> (f32, f32)
%filled = linalg.fill ins(%cst : f32) outs(%alloc : tensor<5xf32>) -> tensor<5xf32>
// CHECK: scf.if %{{.*}} -> (memref<5xf32, 1>) {
%1 = scf.if %c -> tensor<5xf32> {
- // CHECK: scf.yield %[[alloc]]
+ // CHECK: %[[cloned:.*]] = bufferization.clone %[[alloc]]
+ // CHECK: scf.yield %[[cloned]]
scf.yield %filled : tensor<5xf32>
} else {
// CHECK: %[[alloc2:.*]] = memref.alloc() {{.*}} : memref<5xf32, 1>
// CHECK: memref.store %{{.*}}, %[[alloc2]]
- // CHECK: scf.yield %[[alloc2]]
+ // CHECK: %[[cloned2:.*]] = bufferization.clone %[[alloc2]]
+ // CHECK: memref.dealloc %[[alloc2]]
+ // CHECK: scf.yield %[[cloned2]]
%2 = tensor.insert %f into %filled[%c0] : tensor<5xf32>
scf.yield %2 : tensor<5xf32>
}
@@ -689,6 +744,7 @@ func.func @scf_if_memory_space(%c: i1, %f: f32, %cst: f32) -> (f32, f32)
// CHECK: memref.alloc() {{.*}} : memref<5xf32, 1>
// CHECK: memref.store
// CHECK: memref.load
+// CHECK: memref.dealloc
func.func @scf_execute_region_memory_space(%f: f32) -> f32 {
%c0 = arith.constant 0 : index
%0 = scf.execute_region -> tensor<5xf32> {
@@ -899,24 +955,24 @@ func.func @regression_cast_in_loop() -> tensor<2xindex> {
// This test does not compute anything meaningful but it tests that
// bufferizesToMemoryWrite is correctly propagated through regions.
-// CHECK-LABEL: func @elide_copy_of_non_writing_scf_if(
+// CHECK-NO-DEALLOC-PASS-LABEL: func @elide_copy_of_non_writing_scf_if(
func.func @elide_copy_of_non_writing_scf_if(%c: i1, %p1: index, %p2: index, %f: f32)
-> (tensor<10xf32>, f32)
{
%r = scf.if %c -> tensor<10xf32> {
- // CHECK: memref.alloc
+ // CHECK-NO-DEALLOC-PASS: memref.alloc
%t1 = bufferization.alloc_tensor() : tensor<10xf32>
scf.yield %t1 : tensor<10xf32>
} else {
- // CHECK: memref.alloc
+ // CHECK-NO-DEALLOC-PASS: memref.alloc
%t2 = bufferization.alloc_tensor() : tensor<10xf32>
scf.yield %t2 : tensor<10xf32>
}
// No copy should be inserted because %r does not bufferize to a memory write.
// I.e., %r does not have defined contents and the copy can be elided.
- // CHECK-NOT: memref.alloc
- // CHECK-NOT: memref.copy
+ // CHECK-NO-DEALLOC-PASS-NOT: memref.alloc
+ // CHECK-NO-DEALLOC-PASS-NOT: memref.copy
%r2 = tensor.insert %f into %r[%p1] : tensor<10xf32>
%r3 = tensor.extract %r[%p2] : tensor<10xf32>
return %r2, %r3 : tensor<10xf32>, f32
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-collapse-tensor.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-collapse-tensor.mlir
index 53cdcc57ec1b185..1580778167321ed 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-collapse-tensor.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-collapse-tensor.mlir
@@ -1,7 +1,7 @@
// RUN: mlir-opt %s -linalg-bufferize \
// RUN: -arith-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -finalizing-bufferize -buffer-deallocation-pipeline \
-// RUN: -convert-scf-to-cf -expand-strided-metadata -lower-affine -convert-cf-to-llvm -convert-arith-to-llvm \
+// RUN: -finalizing-bufferize -buffer-deallocation \
+// RUN: -expand-strided-metadata -lower-affine \
// RUN: -finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_runner_utils \
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-elementwise.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-elementwise.mlir
index 57b504222c241ab..8331757000ba487 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-elementwise.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-elementwise.mlir
@@ -1,8 +1,8 @@
// RUN: mlir-opt %s -convert-elementwise-to-linalg \
-// RUN: -arith-bufferize -linalg-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -canonicalize -buffer-deallocation-pipeline -convert-linalg-to-loops \
-// RUN: -convert-scf-to-cf -convert-arith-to-llvm -convert-cf-to-llvm --finalize-memref-to-llvm \
-// RUN: -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -arith-bufferize -linalg-bufferize -tensor-bufferize \
+// RUN: -func-bufferize -buffer-deallocation -convert-linalg-to-loops \
+// RUN: --finalize-memref-to-llvm -convert-func-to-llvm \
+// RUN: -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_runner_utils \
// RUN: | FileCheck %s
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-expand-tensor.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-expand-tensor.mlir
index 7dddeb145262173..b7084a30654ab62 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-expand-tensor.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-expand-tensor.mlir
@@ -1,7 +1,7 @@
// RUN: mlir-opt %s -linalg-bufferize \
// RUN: -arith-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -finalizing-bufferize -buffer-deallocation-pipeline \
-// RUN: -convert-scf-to-cf -expand-strided-metadata -lower-affine -convert-cf-to-llvm -convert-arith-to-llvm \
+// RUN: -finalizing-bufferize -buffer-deallocation \
+// RUN: -expand-strided-metadata -lower-affine \
// RUN: -finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_runner_utils \
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-padtensor.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-padtensor.mlir
index c9bc964a124ff45..967c4d58e71b4f2 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-padtensor.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-padtensor.mlir
@@ -1,9 +1,8 @@
// RUN: mlir-opt %s -test-linalg-transform-patterns=test-linalg-to-vector-patterns \
// RUN: -empty-tensor-to-alloc-tensor -linalg-bufferize -arith-bufferize \
// RUN: -bufferization-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -finalizing-bufferize -buffer-deallocation-pipeline \
-// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata \
-// RUN: -lower-affine -convert-arith-to-llvm -finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -finalizing-bufferize -buffer-deallocation \
+// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata -lower-affine -convert-arith-to-llvm -finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_c_runner_utils,%mlir_runner_utils \
// RUN: | FileCheck %s
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert-multiple-uses.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert-multiple-uses.mlir
index d749fca36a5ba52..18dd8bf183cb90a 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert-multiple-uses.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert-multiple-uses.mlir
@@ -1,9 +1,7 @@
// RUN: mlir-opt %s -linalg-bufferize \
// RUN: -arith-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -finalizing-bufferize -buffer-deallocation-pipeline \
-// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata \
-// RUN: -lower-affine -convert-arith-to-llvm --finalize-memref-to-llvm \
-// RUN: -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -finalizing-bufferize -buffer-deallocation \
+// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata -lower-affine -convert-arith-to-llvm --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_c_runner_utils,%mlir_runner_utils \
// RUN: | FileCheck %s
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert.mlir
index f3d416d4809794c..8002db794a40172 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-subtensor-insert.mlir
@@ -1,9 +1,7 @@
// RUN: mlir-opt %s -linalg-bufferize \
// RUN: -arith-bufferize -tensor-bufferize -func-bufferize \
-// RUN: -finalizing-bufferize -buffer-deallocation-pipeline \
-// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata \
-// RUN: -lower-affine -convert-arith-to-llvm --finalize-memref-to-llvm \
-// RUN: -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -finalizing-bufferize -buffer-deallocation \
+// RUN: -convert-linalg-to-loops -convert-scf-to-cf -expand-strided-metadata -lower-affine -convert-arith-to-llvm --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_runner_utils \
// RUN: | FileCheck %s
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-e2e.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-e2e.mlir
index 6be2c18060f3777..715d62137f342c6 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-e2e.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-e2e.mlir
@@ -1,6 +1,6 @@
// RUN: mlir-opt %s -arith-bufferize -linalg-bufferize \
-// RUN: -tensor-bufferize -func-bufferize -finalizing-bufferize -buffer-deallocation-pipeline -convert-linalg-to-loops \
-// RUN: -convert-arith-to-llvm -convert-scf-to-cf -convert-cf-to-llvm --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -tensor-bufferize -func-bufferize -finalizing-bufferize -buffer-deallocation -convert-linalg-to-loops \
+// RUN: --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_runner_utils \
// RUN: | FileCheck %s
diff --git a/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir b/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir
index 0d5db35079aceee..e1b2e247d0619dd 100644
--- a/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir
+++ b/mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir
@@ -1,7 +1,7 @@
// UNSUPPORTED: asan
// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -linalg-bufferize -arith-bufferize \
-// RUN: -tensor-bufferize -func-bufferize -finalizing-bufferize -buffer-deallocation-pipeline -convert-linalg-to-loops -convert-scf-to-cf \
-// RUN: -expand-strided-metadata -lower-affine -convert-arith-to-llvm -convert-scf-to-cf --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: -tensor-bufferize -func-bufferize -finalizing-bufferize -buffer-deallocation -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: -expand-strided-metadata -lower-affine -convert-arith-to-llvm -convert-scf-to-cf --finalize-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
// RUN: mlir-cpu-runner -e main -entry-point-result=void \
// RUN: -shared-libs=%mlir_c_runner_utils,%mlir_runner_utils \
// RUN: | FileCheck %s
More information about the Mlir-commits
mailing list