You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

311 lines
18 KiB

/*
* Copyright (C) 2019 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.
*/
package android.hardware.neuralnetworks@1.3;
import @1.2::IPreparedModel;
import @1.2::MeasureTiming;
import @1.2::OutputShape;
import @1.2::Timing;
import ErrorStatus;
import OptionalTimeoutDuration;
import OptionalTimePoint;
import Request;
import IExecutionCallback;
import IFencedExecutionCallback;
/**
* IPreparedModel describes a model that has been prepared for execution and
* is used to launch executions.
*/
interface IPreparedModel extends @1.2::IPreparedModel {
/**
* Launches an asynchronous execution on a prepared model.
*
* The execution is performed asynchronously with respect to the caller.
* execute_1_3 must verify the inputs to the function are correct, and the usages
* of memory pools allocated by IDevice::allocate are valid. If there is
* an error, execute_1_3 must immediately invoke the callback with the
* appropriate ErrorStatus value, then return with the same ErrorStatus. If
* the inputs to the function are valid and there is no error, execute_1_3 must
* launch an asynchronous task to perform the execution in the background,
* and immediately return with ErrorStatus::NONE. If the asynchronous task
* fails to launch, execute_1_3 must immediately invoke the callback with
* ErrorStatus::GENERAL_FAILURE, then return with
* ErrorStatus::GENERAL_FAILURE.
*
* When the asynchronous task has finished its execution, it must
* immediately invoke the callback object provided as an input to the
* execute_1_3 function. This callback must be provided with the ErrorStatus of
* the execution.
*
* If the launch is successful, the caller must not change the content of
* any data object referenced by 'request' (described by the
* {@link @1.0::DataLocation} of a {@link @1.0::RequestArgument}) until the
* asynchronous task has invoked the callback object. The asynchronous task
* must not change the content of any of the data objects corresponding to
* 'request' inputs.
*
* If the prepared model was prepared from a model wherein all tensor
* operands have fully specified dimensions, and the inputs to the function
* are valid, then:
* - the execution should launch successfully (ErrorStatus::NONE): There
* must be no failure unless the device itself is in a bad state.
* - if at execution time every operation's input operands have legal
* values, the execution should complete successfully (ErrorStatus::NONE):
* There must be no failure unless the device itself is in a bad state.
*
* execute_1_3 can be called with an optional deadline. If the execution
* is not able to be completed before the provided deadline, the execution
* may be aborted, and either {@link
* ErrorStatus::MISSED_DEADLINE_TRANSIENT} or {@link
* ErrorStatus::MISSED_DEADLINE_PERSISTENT} may be returned. The error due
* to an abort must be sent the same way as other errors, described above.
* The deadline is represented as nanoseconds since the epoch of the steady
* clock (as if from std::chrono::steady_clock::time_point), but the service
* may convert it to the nanoseconds since boot time (as if from
* clock_gettime(CLOCK_BOOTTIME, &ts) or
* android::base::boot_clock::time_point) to account for time when the
* system is suspended. This conversion can by done by finding the timeout
* duration remaining compared to the steady_clock and adding it to the
* current boot_clock time.
*
* Any number of calls to the execute* and executeSynchronously* functions,
* in any combination, may be made concurrently, even on the same
* IPreparedModel object.
*
* @param request The input and output information on which the prepared
* model is to be executed.
* @param measure Specifies whether or not to measure duration of the execution.
* The duration runs from the time the driver sees the call
* to the execute_1_3 function to the time the driver invokes
* the callback.
* @param deadline The time by which the execution is expected to complete.
* If the execution cannot be completed by the deadline, the
* execution may be aborted.
* @param loopTimeoutDuration The maximum amount of time that should be spent
* executing a {@link OperationType::WHILE}
* operation. If a loop condition model does not
* output false within this duration, the
* execution must be aborted. If no loop timeout
* duration is provided, the maximum amount of
* time is {@link LoopTimeoutDurationNs::DEFAULT}.
* When provided, the duration must not exceed
* {@link LoopTimeoutDurationNs::MAXIMUM}.
* @param callback A callback object used to return the error status of
* the execution, shape information of model output operands, and
* duration of execution. The callback object's notify function must
* be called exactly once, even if the execution was
* unsuccessful.
* @return status Error status of the call, must be:
* - NONE if task is successfully launched
* - DEVICE_UNAVAILABLE if driver is offline or busy
* - GENERAL_FAILURE if there is an unspecified error
* - OUTPUT_INSUFFICIENT_SIZE if provided output buffer is
* not large enough to store the resultant values
* - INVALID_ARGUMENT if one of the input arguments is
* invalid
* - MISSED_DEADLINE_* if the execution is aborted because it
* cannot be completed by the deadline
* - RESOURCE_EXHAUSTED_* if the task was aborted by the
* driver
*/
execute_1_3(Request request, MeasureTiming measure, OptionalTimePoint deadline,
OptionalTimeoutDuration loopTimeoutDuration, IExecutionCallback callback)
generates (ErrorStatus status);
/**
* Performs a synchronous execution on a prepared model.
*
* The execution is performed synchronously with respect to the caller.
* executeSynchronously_1_3 must verify the inputs to the function are
* correct, and the usages of memory pools allocated by IDevice::allocate
* are valid. If there is an error, executeSynchronously_1_3 must immediately
* return with the appropriate ErrorStatus value. If the inputs to the
* function are valid and there is no error, executeSynchronously_1_3 must
* perform the execution, and must not return until the execution is
* complete.
*
* The caller must not change the content of any data object referenced by
* 'request' (described by the {@link @1.0::DataLocation} of a
* {@link @1.0::RequestArgument}) until executeSynchronously_1_3
* returns. executeSynchronously_1_3 must not change the content of any of the
* data objects corresponding to 'request' inputs.
*
* If the prepared model was prepared from a model wherein all tensor
* operands have fully specified dimensions, and the inputs to the function
* are valid, and at execution time every operation's input operands have
* legal values, then the execution should complete successfully
* (ErrorStatus::NONE): There must be no failure unless the device itself is
* in a bad state.
*
* executeSynchronously_1_3 may be called with an optional deadline. If the
* execution is not able to be completed before the provided deadline, the
* execution may be aborted, and either {@link
* ErrorStatus::MISSED_DEADLINE_TRANSIENT} or {@link
* ErrorStatus::MISSED_DEADLINE_PERSISTENT} may be returned. The error due
* to an abort must be sent the same way as other errors, described above.
* The deadline is represented as nanoseconds since the epoch of the steady
* clock (as if from std::chrono::steady_clock::time_point), but the service
* may convert it to the nanoseconds since boot time (as if from
* clock_gettime(CLOCK_BOOTTIME, &ts) or
* android::base::boot_clock::time_point) to account for time when the
* system is suspended. This conversion can by done by finding the timeout
* duration remaining compared to the steady_clock and adding it to the
* current boot_clock time.
*
* Any number of calls to the execute* and executeSynchronously* functions,
* in any combination, may be made concurrently, even on the same
* IPreparedModel object.
*
* @param request The input and output information on which the prepared
* model is to be executed.
* @param measure Specifies whether or not to measure duration of the execution.
* The duration runs from the time the driver sees the call
* to the executeSynchronously_1_3 function to the time the driver
* returns from the function.
* @param deadline The time by which the execution is expected to complete.
* If the execution cannot be finished by the deadline, the
* execution may be aborted.
* @param loopTimeoutDuration The maximum amount of time that should be spent
* executing a {@link OperationType::WHILE}
* operation. If a loop condition model does not
* output false within this duration, the
* execution must be aborted. If no loop timeout
* duration is provided, the maximum amount of
* time is {@link LoopTimeoutDurationNs::DEFAULT}.
* When provided, the duration must not exceed
* {@link LoopTimeoutDurationNs::MAXIMUM}.
* @return status Error status of the execution, must be:
* - NONE if execution is performed successfully
* - DEVICE_UNAVAILABLE if driver is offline or busy
* - GENERAL_FAILURE if there is an unspecified error
* - OUTPUT_INSUFFICIENT_SIZE if at least one output
* operand buffer is not large enough to store the
* corresponding output
* - INVALID_ARGUMENT if one of the input arguments is
* invalid
* - MISSED_DEADLINE_* if the execution is aborted because it
* cannot be completed by the deadline
* - RESOURCE_EXHAUSTED_* if the task was aborted by the
* driver
* @return outputShapes A list of shape information of model output operands.
* The index into "outputShapes" corresponds to the index
* of the output operand in the Request outputs vector.
* outputShapes must be empty unless the status is either
* NONE or OUTPUT_INSUFFICIENT_SIZE.
* @return timing Duration of execution. Unless measure is YES and status is
* NONE, all times must be reported as UINT64_MAX. A driver may
* choose to report any time as UINT64_MAX, indicating that
* measurement is not available.
*/
executeSynchronously_1_3(Request request, MeasureTiming measure,
OptionalTimePoint deadline,
OptionalTimeoutDuration loopTimeoutDuration)
generates (ErrorStatus status, vec<OutputShape> outputShapes,
Timing timing);
/**
* Launch a fenced asynchronous execution on a prepared model.
*
* The execution is performed asynchronously with respect to the caller.
* executeFenced must verify the inputs to the function are correct, and the usages
* of memory pools allocated by IDevice::allocate are valid. If there is an error,
* executeFenced must immediately return with the corresponding ErrorStatus, an empty
* handle for syncFence, and nullptr for callback. If the inputs to the function
* are valid and there is no error, executeFenced must dispatch an asynchronous task
* to perform the execution in the background, and immediately return with
* ErrorStatus::NONE, a sync fence that will be signaled once the execution is completed,
* and a callback that can be used by the client to query the duration and runtime error
* status. If the task has finished before the call returns, an empty handle may be returned
* for syncFence. The execution must wait for all the sync fences (if any) in waitFor
* to be signaled before starting the actual execution.
*
* When the asynchronous task has finished its execution, it must
* immediately signal the syncFence returned from the executeFenced call. After
* the syncFence is signaled, the task must not modify the content of
* any data object referenced by 'request' (described by the
* {@link @1.0::DataLocation} of a {@link @1.0::RequestArgument}).
*
* executeFenced may be called with an optional deadline and an optional duration.
* If the execution is not able to be completed before the provided deadline or
* within the timeout duration (measured from when all sync fences in waitFor are
* signaled), whichever comes earlier, the execution may be aborted, and either
* {@link ErrorStatus::MISSED_DEADLINE_TRANSIENT} or {@link
* ErrorStatus::MISSED_DEADLINE_PERSISTENT} may be returned. The error due
* to an abort must be sent the same way as other errors, described above.
* The deadline is represented as nanoseconds since the epoch of the steady
* clock (as if from std::chrono::steady_clock::time_point), but the service
* may convert it to the nanoseconds since boot time (as if from
* clock_gettime(CLOCK_BOOTTIME, &ts) or
* android::base::boot_clock::time_point) to account for time when the
* system is suspended. This conversion can by done by finding the timeout
* duration remaining compared to the steady_clock and adding it to the
* current boot_clock time.
*
* If any of the sync fences in waitFor changes to error status after the executeFenced
* call succeeds, or the execution is aborted because it cannot finish before the deadline
* has been reached or the duration has elapsed, the driver must immediately set the returned
* syncFence to error status.
*
* Any number of calls to the executeFenced, execute* and executeSynchronously*
* functions, in any combination, may be made concurrently, even on the same
* IPreparedModel object.
*
* @param request The input and output information on which the prepared
* model is to be executed. The outputs in the request must have
* fully specified dimensions.
* @param waitFor A vector of sync fence file descriptors.
* Execution must not start until all sync fences have been signaled.
* @param measure Specifies whether or not to measure duration of the execution.
* @param deadline The time by which the execution is expected to complete.
* If the execution cannot be finished by the deadline, the
* execution may be aborted.
* @param loopTimeoutDuration The maximum amount of time that should be spent
* executing a {@link OperationType::WHILE}
* operation. If a loop condition model does not
* output false within this duration, the
* execution must be aborted. If no loop timeout
* duration is provided, the maximum amount of
* time is {@link LoopTimeoutDurationNs::DEFAULT}.
* When provided, the duration must not exceed
* {@link LoopTimeoutDurationNs::MAXIMUM}.
* @param duration The length of time within which the execution is expected
* to complete after all sync fences in waitFor are signaled.
* If the execution cannot be finished within the duration,
* the execution may be aborted.
* @return status Error status of the call, must be:
* - NONE if task is successfully launched
* - DEVICE_UNAVAILABLE if driver is offline or busy
* - GENERAL_FAILURE if there is an unspecified error
* - INVALID_ARGUMENT if one of the input arguments is invalid, including
* fences in error states.
* - MISSED_DEADLINE_* if the execution is aborted because it
* cannot be completed by the deadline
* - RESOURCE_EXHAUSTED_* if the task was aborted by the
* driver
* @return syncFence The sync fence that will be signaled when the task is completed.
* The sync fence will be set to error if a critical error,
* e.g. hardware failure or kernel panic, occurs when doing execution.
* @return callback The IFencedExecutionCallback can be used to query information like duration
* and error status when the execution is completed.
*/
executeFenced(Request request, vec<handle> waitFor, MeasureTiming measure,
OptionalTimePoint deadline, OptionalTimeoutDuration loopTimeoutDuration,
OptionalTimeoutDuration duration)
generates (ErrorStatus status, handle syncFence, IFencedExecutionCallback callback);
};