Introduce reusable execution to canonical interface -- HAL.

This CL modifies the canonical interface for reusable executions:
- Add new interface: IExecution with compute and computeFenced methods
- Add new method IPreparedModel::createExecution

In NNAPI runtime, the new interface IExecution is used to
memoize request-specific execution resources (e.g. converted HAL
request). The expected usage is that, IPreparedModel::createExecution
will be invoked in the first computation of a reusable NDK ANNExecution
object, and IExecution::compute* will be invoked repeatedly.

The IPreparedModel::execute* methods are preserved to avoid redundant
object creation and memoization overhead for a single-time
(non-reusable) execution.

For a vendor implementing the canonical interfaces, only the
IPreparedModel::execute* methods will be called because there is
currently no reusable execution at HAL interface. A DefaultExecution
implementation is provided to reduce the work needed on the vendor side.

Bug: 184073769
Test: NNT_static
Test: neuralnetworks_utils_hal_1_0_test
Test: neuralnetworks_utils_hal_1_1_test
Test: neuralnetworks_utils_hal_1_2_test
Test: neuralnetworks_utils_hal_1_3_test
Test: neuralnetworks_utils_hal_common_test
Test: neuralnetworks_utils_hal_aidl_test
Change-Id: I91790bb5ccf5ae648687fe603f88ffda2c9fd2b2

1 files changed, 31 insertions, 0 deletions

diff --git a/neuralnetworks/utils/common/test/ResilientPreparedModelTest.cpp b/neuralnetworks/utils/common/test/ResilientPreparedModelTest.cpp
index 6d86e10df2..d396ca88df 100644
--- a/neuralnetworks/utils/common/test/ResilientPreparedModelTest.cpp
+++ b/neuralnetworks/utils/common/test/ResilientPreparedModelTest.cpp
@@ -55,6 +55,7 @@ constexpr auto makeError = [](nn::ErrorStatus status) {
 const auto kReturnGeneralFailure = makeError(nn::ErrorStatus::GENERAL_FAILURE);
 const auto kReturnDeadObject = makeError(nn::ErrorStatus::DEAD_OBJECT);
 
+const auto kNoCreateReusableExecutionError = nn::GeneralResult<nn::SharedExecution>{};
 const auto kNoExecutionError =
         nn::ExecutionResult<std::pair<std::vector<nn::OutputShape>, nn::Timing>>{};
 const auto kNoFencedExecutionError =
@@ -231,6 +232,36 @@ TEST(ResilientPreparedModelTest, executeFencedDeadObjectSuccessfulRecovery) {
             << "Failed with " << result.error().code << ": " << result.error().message;
 }
 
+TEST(ResilientPreparedModelTest, createReusableExecution) {
+    // setup call
+    const auto [mockPreparedModel, mockPreparedModelFactory, preparedModel] = setup();
+    EXPECT_CALL(*mockPreparedModel, createReusableExecution(_, _, _))
+            .Times(1)
+            .WillOnce(Return(kNoCreateReusableExecutionError));
+
+    // run test
+    const auto result = preparedModel->createReusableExecution({}, {}, {});
+
+    // verify result
+    ASSERT_TRUE(result.has_value())
+            << "Failed with " << result.error().code << ": " << result.error().message;
+}
+
+TEST(ResilientPreparedModelTest, createReusableExecutionError) {
+    // setup call
+    const auto [mockPreparedModel, mockPreparedModelFactory, preparedModel] = setup();
+    EXPECT_CALL(*mockPreparedModel, createReusableExecution(_, _, _))
+            .Times(1)
+            .WillOnce(kReturnGeneralFailure);
+
+    // run test
+    const auto result = preparedModel->createReusableExecution({}, {}, {});
+
+    // verify result
+    ASSERT_FALSE(result.has_value());
+    EXPECT_EQ(result.error().code, nn::ErrorStatus::GENERAL_FAILURE);
+}
+
 TEST(ResilientPreparedModelTest, getUnderlyingResource) {
     // setup call
     const auto [mockPreparedModel, mockPreparedModelFactory, preparedModel] = setup();