/*
* Copyright (C) 2018 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <gmock/gmock.h>
#include <gtest/gtest.h>
#include <limits>
#include <utility>
#include <vector>
#include "HalInterfaces.h"
#include "MemoryUtils.h"
#include "OperationsExecutionUtils.h"
#include "QuantUtils.h"
#include "Utils.h"
#include "ValidateHal.h"
#include "nnapi/TypeUtils.h"
#include "nnapi/Types.h"
namespace android {
namespace nn {
namespace wrapper {
namespace {
using ::testing::ElementsAreArray;
} // namespace
TEST(CalculateBroadcastedShapeTest, Basic) {
Shape shape1;
Shape shape2;
shape1.dimensions = {4, 3, 2, 1};
shape2.dimensions = {3, 1, 5};
Shape expectedOutputShape;
expectedOutputShape.dimensions = {4, 3, 2, 5};
Shape actualOutputShape;
EXPECT_TRUE(calculateBroadcastedShape(shape1, shape2, &actualOutputShape));
EXPECT_THAT(actualOutputShape.dimensions, ElementsAreArray(expectedOutputShape.dimensions));
EXPECT_TRUE(calculateBroadcastedShape(shape2, shape1, &actualOutputShape));
EXPECT_THAT(actualOutputShape.dimensions, ElementsAreArray(expectedOutputShape.dimensions));
}
TEST(CalculateBroadcastedShapeTest, FailsOnIncompatible) {
Shape shape1;
Shape shape2;
shape1.dimensions = {5};
shape2.dimensions = {3};
Shape actualOutputShape;
EXPECT_FALSE(calculateBroadcastedShape(shape1, shape2, &actualOutputShape));
EXPECT_FALSE(calculateBroadcastedShape(shape2, shape1, &actualOutputShape));
}
static int32_t getExtensionType(uint16_t extensionPrefix, uint16_t typeWithinExtension) {
int32_t type = (extensionPrefix << kExtensionTypeBits) | typeWithinExtension;
EXPECT_TRUE(isExtensionOperandType(static_cast<V1_3::OperandType>(type)));
return type;
}
TEST(TensorHasUnspecifiedDimensionsTest, ExtensionTensorWithUnspecifiedRank) {
// Regression test for b/124285861.
EXPECT_TRUE(tensorHasUnspecifiedDimensions(getExtensionType(1, 0), /*dim=*/nullptr,
/*dimCount=*/0));
}
TEST(ValidateOperandTypeTest, ExtensionTensorWithUnspecifiedRank) {
// Regression test for b/124104123.
constexpr uint16_t kExtensionPrefix = 1;
constexpr uint16_t kTypeWithinExtension = 0;
int32_t extensionType = getExtensionType(kExtensionPrefix, kTypeWithinExtension);
ANeuralNetworksOperandType type = {
.type = extensionType,
.dimensionCount = 0,
.dimensions = nullptr,
};
Extension::OperandTypeInformation info = {
.type = kTypeWithinExtension,
.isTensor = true,
.byteSize = 4,
};
EXPECT_EQ(validateOperandType(type, &info, /*tag=*/"test", /*allowPartial=*/true),
ANEURALNETWORKS_NO_ERROR);
EXPECT_EQ(validateOperandType(type, &info, /*tag=*/"test", /*allowPartial=*/false),
ANEURALNETWORKS_BAD_DATA);
}
TEST(ValidateOperandTypeTest, ExtensionTypeDimensionProductOverflow) {
// Regression test for b/146044137.
constexpr uint16_t kExtensionPrefix = 1;
constexpr uint16_t kTypeWithinExtension = 0;
int32_t extensionType = getExtensionType(kExtensionPrefix, kTypeWithinExtension);
uint32_t dimensions[] = {5, 4, 4, 786433, 5, 3, 16777216, 4, 5};
ANeuralNetworksOperandType type = {
.type = extensionType,
.dimensionCount = std::size(dimensions),
.dimensions = dimensions,
};
Extension::OperandTypeInformation info = {
.type = kTypeWithinExtension,
.isTensor = true,
.byteSize = 1,
};
EXPECT_EQ(validateOperandType(type, &info, /*tag=*/"test", /*allowPartial=*/true),
ANEURALNETWORKS_BAD_DATA);
}
TEST(ValidateOperandTypeTest, TensorSizeDimensionProductOverflow) {
// Regression test for b/146044137.
uint32_t dimensions[] = {256, 256, 256, 256};
ANeuralNetworksOperandType type = {
.type = ANEURALNETWORKS_TENSOR_FLOAT32,
.dimensionCount = std::size(dimensions),
.dimensions = dimensions,
};
EXPECT_EQ(validateOperandType(type, nullptr, /*tag=*/"test", /*allowPartial=*/true),
ANEURALNETWORKS_BAD_DATA);
}
TEST(ValidateRequestTest, UnknownOutputRank) {
V1_3::Request::MemoryPool pool;
pool.hidlMemory(allocateSharedMemory(2 * sizeof(float)));
ASSERT_TRUE(pool.hidlMemory().valid());
const V1_3::Model model = {
.main =
{
.operands = {{
.type = V1_3::OperandType::TENSOR_FLOAT32,
.dimensions = {1},
.numberOfConsumers = 1,
.lifetime = V1_3::OperandLifeTime::SUBGRAPH_INPUT,
},
{
.type = V1_3::OperandType::TENSOR_FLOAT32,
.dimensions = {}, // unknown output rank
.numberOfConsumers = 0,
.lifetime = V1_3::OperandLifeTime::SUBGRAPH_OUTPUT,
}},
.operations = {{
.type = V1_3::OperationType::ABS,
.inputs = {0},
.outputs = {1},
}},
.inputIndexes = {0},
.outputIndexes = {1},
},
};
const V1_3::Request request = {
.inputs = {{
.location = {.poolIndex = 0, .offset = 0, .length = sizeof(float)},
.dimensions = {},
}},
.outputs = {{
.location = {.poolIndex = 0, .offset = sizeof(float), .length = sizeof(float)},
.dimensions = {},
}},
.pools = {std::move(pool)},
};
EXPECT_FALSE(validateRequest(request, model, /*allowUnspecifiedOutput=*/false));
}
TEST(ValidateRequestTest, ScalarOutput) {
V1_3::Request::MemoryPool pool;
pool.hidlMemory(allocateSharedMemory(sizeof(float) + sizeof(int32_t)));
ASSERT_TRUE(pool.hidlMemory().valid());
const V1_3::Model model = {
.main =
{
.operands = {{
.type = V1_3::OperandType::TENSOR_FLOAT32,
.dimensions = {1},
.numberOfConsumers = 1,
.lifetime = V1_3::OperandLifeTime::SUBGRAPH_INPUT,
},
{
.type = V1_3::OperandType::INT32,
.dimensions = {},
.numberOfConsumers = 0,
.lifetime = V1_3::OperandLifeTime::SUBGRAPH_OUTPUT,
}},
.operations = {{
.type = V1_3::OperationType::RANK,
.inputs = {0},
.outputs = {1},
}},
.inputIndexes = {0},
.outputIndexes = {1},
},
};
const V1_3::Request request = {
.inputs = {{
.location = {.poolIndex = 0, .offset = 0, .length = sizeof(float)},
.dimensions = {},
}},
.outputs = {{
.location = {.poolIndex = 0,
.offset = sizeof(float),
.length = sizeof(int32_t)},
.dimensions = {},
}},
.pools = {std::move(pool)},
};
EXPECT_TRUE(validateRequest(request, model, /*allowUnspecifiedOutput=*/false));
}
class CombineDimensionsTest : public ::testing::Test {
protected:
void testCompatible(const std::vector<uint32_t>& lhs, const std::vector<uint32_t>& rhs,
const std::vector<uint32_t>& expected) {
SCOPED_TRACE("lhs = " + toString(lhs) + ", rhs = " + toString(rhs));
const auto res = combineDimensions(lhs, rhs);
ASSERT_TRUE(res.has_value());
EXPECT_EQ(res.value(), expected);
}
void testIncompatible(const std::vector<uint32_t>& lhs, const std::vector<uint32_t>& rhs) {
SCOPED_TRACE("lhs = " + toString(lhs) + ", rhs = " + toString(rhs));
const auto res = combineDimensions(lhs, rhs);
EXPECT_FALSE(res.has_value());
}
};
TEST_F(CombineDimensionsTest, Rank) {
testCompatible({}, {1, 2, 3, 4}, {1, 2, 3, 4});
testCompatible({1, 2, 3, 4}, {}, {1, 2, 3, 4});
testCompatible({}, {}, {});
testIncompatible({1, 2, 3}, {1, 2, 3, 4});
testIncompatible({1, 2, 3, 4}, {1, 2, 3});
}
TEST_F(CombineDimensionsTest, Dimensions) {
testCompatible({0, 0, 0, 0}, {1, 2, 3, 4}, {1, 2, 3, 4});
testCompatible({1, 2, 3, 4}, {0, 0, 0, 0}, {1, 2, 3, 4});
testCompatible({0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0});
testIncompatible({1, 2, 3, 4}, {2, 2, 3, 4});
testIncompatible({1, 2, 3, 4}, {1, 2, 3, 3});
}
TEST(QuantizationUtilsTest, QuantizeMultiplierSmallerThanOneExp) {
auto checkInvalidQuantization = [](double value) {
int32_t q;
int s;
EXPECT_FALSE(QuantizeMultiplierSmallerThanOneExp(value, &q, &s));
};
checkInvalidQuantization(-0.1);
checkInvalidQuantization(0.0);
// If we get close enough to 1.0 it crashes and dies in one of two ways:
// Either the shift becomes negative or we trigger the 'less-than-one' CHECK.
checkInvalidQuantization(1 - 1e-15);
checkInvalidQuantization(1 - 1e-17);
checkInvalidQuantization(1.0);
auto checkQuantization = [](double value, int32_t goldenQuantized, int goldenShift) {
int32_t q;
int s;
EXPECT_TRUE(QuantizeMultiplierSmallerThanOneExp(value, &q, &s));
EXPECT_EQ(q, goldenQuantized);
EXPECT_EQ(s, goldenShift);
};
checkQuantization(0.25, 1073741824, -1);
checkQuantization(0.50 - 5e-9, 2147483627, -1);
checkQuantization(0.50 - 1e-10, 1073741824, 0);
checkQuantization(0.50, 1073741824, 0);
checkQuantization(0.75, 1610612736, 0);
checkQuantization(1 - 1e-9, 2147483646, 0);
}
TEST(QuantizationUtilsTest, QuantizeMultiplierGreaterThanOne) {
auto checkInvalidQuantization = [](double value) {
int32_t q;
int s;
EXPECT_FALSE(QuantizeMultiplierGreaterThanOne(value, &q, &s));
};
checkInvalidQuantization(1 + 1e-16);
auto checkQuantization = [](double value, int32_t goldenQuantized, int goldenShift) {
int32_t q;
int s;
EXPECT_TRUE(QuantizeMultiplierGreaterThanOne(value, &q, &s));
EXPECT_EQ(q, goldenQuantized);
EXPECT_EQ(s, goldenShift);
};
checkQuantization(1 + 1e-11, 1073741824, 1);
checkQuantization(1.25, 1342177280, 1);
checkQuantization(1.50, 1610612736, 1);
checkQuantization(1.50, 1610612736, 1);
checkQuantization(1.75, 1879048192, 1);
checkQuantization(2 - 1e-9, 2147483647, 1);
checkQuantization(2 - 1e-11, 1073741824, 2);
checkQuantization(2, 1073741824, 2);
}
TEST(QuantizationUtilTest, QuantizeMultiplier) {
auto checkQuantization = [](double value, int32_t goldenQuantized, int goldenShift) {
int32_t q;
int s;
EXPECT_TRUE(QuantizeMultiplier(value, &q, &s));
EXPECT_EQ(q, goldenQuantized);
EXPECT_EQ(s, goldenShift);
};
checkQuantization(-4, -1073741824, 3);
checkQuantization(-2, -1073741824, 2);
checkQuantization(-1, -1073741824, 1);
checkQuantization(-0.5, -1073741824, 0);
checkQuantization(-0.25, -1073741824, -1);
checkQuantization(-0.125, -1073741824, -2);
checkQuantization(0, 0, 0);
checkQuantization(0.125, 1073741824, -2);
checkQuantization(0.25, 1073741824, -1);
checkQuantization(0.5, 1073741824, 0);
checkQuantization(1, 1073741824, 1);
checkQuantization(2, 1073741824, 2);
checkQuantization(4, 1073741824, 3);
}
TEST(QuantizationUtilTest, QuantizeMultiplierUnderflow) {
auto checkQuantization = [](double value, int32_t goldenQuantized, int goldenShift) {
int32_t q;
int s;
EXPECT_TRUE(QuantizeMultiplier(value, &q, &s));
EXPECT_EQ(q, goldenQuantized);
EXPECT_EQ(s, goldenShift);
};
checkQuantization(std::ldexp(1.0f, -31), 1073741824, -30);
checkQuantization(std::ldexp(1.0f, -32), 1073741824, -31);
checkQuantization(std::ldexp(0.99f, -32), 0, 0);
checkQuantization(std::ldexp(1.0f, -33), 0, 0);
}
TEST(QuantizationUtilTest, GetInvSqrtQuantizedMultiplierExp) {
auto checkInvSqrtQuantization = [](int32_t input, int32_t goldenInvSqrt, int goldenShift) {
int32_t q;
int s;
EXPECT_TRUE(GetInvSqrtQuantizedMultiplierExp(input, 1, &q, &s));
EXPECT_EQ(q, goldenInvSqrt);
EXPECT_EQ(s, goldenShift);
};
const auto kInt32Max = std::numeric_limits<std::int32_t>::max();
checkInvSqrtQuantization(0, kInt32Max, 0);
checkInvSqrtQuantization(1, kInt32Max, 0);
checkInvSqrtQuantization(2, 1518498372, 0);
checkInvSqrtQuantization(3, 1239850284, 0);
checkInvSqrtQuantization(4, 1073741828, 0);
checkInvSqrtQuantization(100, 214748363, 0);
checkInvSqrtQuantization(10000, 343597361, 4);
checkInvSqrtQuantization(1000000, 274877901, 7);
checkInvSqrtQuantization(100000000, 219902323, 10);
checkInvSqrtQuantization((1 << 30), 268435457, 12);
checkInvSqrtQuantization(kInt32Max, 189812531, 12);
}
} // namespace wrapper
} // namespace nn
} // namespace android