1 /*
2  * Copyright (C) 2018 The Android Open Source Project
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  *      http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 
17 #define LOG_TAG "Operations"
18 
19 #include "HalInterfaces.h"
20 #include "IndexedShapeWrapper.h"
21 #include "OperationResolver.h"
22 #include "OperationsUtils.h"
23 
24 namespace android {
25 namespace nn {
26 namespace logical {
27 
28 constexpr uint32_t kNumInputs = 2;
29 constexpr uint32_t kInputTensor1 = 0;
30 constexpr uint32_t kInputTensor2 = 1;
31 
32 constexpr uint32_t kNumOutputs = 1;
33 constexpr uint32_t kOutputTensor = 0;
34 
35 namespace {
36 
compute(const std::function<bool (bool,bool)> & func,const bool8 * aData,const Shape & aShape,const bool8 * bData,const Shape & bShape,bool8 * outputData,const Shape & outputShape)37 bool compute(const std::function<bool(bool, bool)>& func, const bool8* aData, const Shape& aShape,
38              const bool8* bData, const Shape& bShape, bool8* outputData, const Shape& outputShape) {
39     IndexedShapeWrapper aShapeIndexed(aShape);
40     IndexedShapeWrapper bShapeIndexed(bShape);
41     IndexedShapeWrapper outputShapeIndexed(outputShape);
42     std::vector<uint32_t> curIndex(outputShape.dimensions.size(), 0);
43     bool lastIndex = false;
44     do {
45         uint32_t outputFlatIndex;
46         NN_RET_CHECK(outputShapeIndexed.indexToFlatIndex(curIndex, &outputFlatIndex));
47         uint32_t aFlatIndex;
48         NN_RET_CHECK(aShapeIndexed.broadcastedIndexToFlatIndex(curIndex, &aFlatIndex));
49         uint32_t bFlatIndex;
50         NN_RET_CHECK(bShapeIndexed.broadcastedIndexToFlatIndex(curIndex, &bFlatIndex));
51 
52         outputData[outputFlatIndex] = func(aData[aFlatIndex], bData[bFlatIndex]);
53 
54         NN_RET_CHECK(outputShapeIndexed.nextIndexInplace(&curIndex, &lastIndex));
55     } while (!lastIndex);
56     return true;
57 }
58 
59 }  // namespace
60 
validate(const IOperationValidationContext * context)61 bool validate(const IOperationValidationContext* context) {
62     NN_RET_CHECK_EQ(context->getNumInputs(), kNumInputs);
63     NN_RET_CHECK_EQ(context->getNumOutputs(), kNumOutputs);
64     OperandType inputType = context->getInputType(kInputTensor1);
65     NN_RET_CHECK(inputType == OperandType::TENSOR_BOOL8)
66             << "Unsupported tensor type for a logical operation";
67     NN_RET_CHECK(validateInputTypes(context, {inputType, inputType}));
68     NN_RET_CHECK(validateOutputTypes(context, {inputType}));
69     return validateHalVersion(context, HalVersion::V1_2);
70 }
71 
prepare(IOperationExecutionContext * context)72 bool prepare(IOperationExecutionContext* context) {
73     Shape input1 = context->getInputShape(kInputTensor1);
74     Shape input2 = context->getInputShape(kInputTensor2);
75     Shape output = context->getOutputShape(kOutputTensor);
76     NN_RET_CHECK(calculateBroadcastedShape(input1, input2, &output));
77     return context->setOutputShape(kOutputTensor, output);
78 }
79 
executeAnd(IOperationExecutionContext * context)80 bool executeAnd(IOperationExecutionContext* context) {
81     return compute(
82             std::logical_and<bool>(), context->getInputBuffer<bool8>(kInputTensor1),
83             context->getInputShape(kInputTensor1), context->getInputBuffer<bool8>(kInputTensor2),
84             context->getInputShape(kInputTensor2), context->getOutputBuffer<bool8>(kOutputTensor),
85             context->getOutputShape(kOutputTensor));
86 }
87 
executeOr(IOperationExecutionContext * context)88 bool executeOr(IOperationExecutionContext* context) {
89     return compute(
90             std::logical_or<bool>(), context->getInputBuffer<bool8>(kInputTensor1),
91             context->getInputShape(kInputTensor1), context->getInputBuffer<bool8>(kInputTensor2),
92             context->getInputShape(kInputTensor2), context->getOutputBuffer<bool8>(kOutputTensor),
93             context->getOutputShape(kOutputTensor));
94 }
95 
96 }  // namespace logical
97 
98 NN_REGISTER_OPERATION(LOGICAL_AND, "LOGICAL_AND", logical::validate, logical::prepare,
99                       logical::executeAnd);
100 NN_REGISTER_OPERATION(LOGICAL_OR, "LOGICAL_OR", logical::validate, logical::prepare,
101                       logical::executeOr);
102 
103 }  // namespace nn
104 }  // namespace android
105