processingUnit.hpp

Go to the documentation of this file.

/*
 *   Copyright 2025 Huawei Technologies Co., Ltd.
 *
 * 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.
 */
 
#pragma once
 
#include <csignal>
#include <set>
#include <sched.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <pthread.h>
#include <hicr/core/definitions.hpp>
#include <hicr/core/exceptions.hpp>
#include <hicr/core/processingUnit.hpp>
#include <hicr/backends/boost/executionState.hpp>
#include <hicr/backends/boost/executionUnit.hpp>
#include <hicr/backends/hwloc/computeResource.hpp>
 
#ifndef _GNU_SOURCE
  #define _GNU_SOURCE
#endif
 
namespace HiCR::backend::pthreads
{
class ComputeManager;
}
 
namespace HiCR::backend::pthreads
{
 
#define HICR_SUSPEND_SIGNAL SIGUSR1
 
#define HICR_RESUME_SIGNAL SIGUSR2
 
class ProcessingUnit;
 
class ProcessingUnit final : public HiCR::ProcessingUnit
{
  friend class HiCR::backend::pthreads::ComputeManager;
 
  public:
 
  [[nodiscard]] __INLINE__ std::string getType() override { return "POSIX Thread"; }
 
  __INLINE__ static void updateAffinity(const std::set<hwloc::ComputeResource::logicalProcessorId_t> &affinity)
  {
    cpu_set_t cpuset;
    CPU_ZERO(&cpuset);
    for (const auto c : affinity) CPU_SET_S(c, sizeof(cpu_set_t), &cpuset);
 
    // Attempting to use the pthread interface first
    int status = pthread_setaffinity_np(pthread_self(), sizeof(cpu_set_t), &cpuset);
 
    // If failed, attempt to use the sched interface
    if (status != 0) status = sched_getaffinity(getpid(), sizeof(cpu_set_t), &cpuset);
 
    // Throw exception if none of them worked
    if (status != 0) HICR_THROW_RUNTIME("Problem assigning affinity.");
  }
 
  __INLINE__ static std::set<HiCR::backend::hwloc::ComputeResource::logicalProcessorId_t> getAffinity()
  {
    auto      affinity = std::set<HiCR::backend::hwloc::ComputeResource::logicalProcessorId_t>();
    cpu_set_t cpuset;
 
    // Attempting to use the pthread interface first
    int status = pthread_getaffinity_np(pthread_self(), sizeof(cpu_set_t), &cpuset);
 
    // If failed, attempt to use the sched interface
    if (status != 0) status = sched_getaffinity(getpid(), sizeof(cpu_set_t), &cpuset);
 
    // Throw exception if none of them worked
    if (status != 0) HICR_THROW_RUNTIME("Problem obtaining affinity.");
    for (int i = 0; i < CPU_SETSIZE; i++)
      if (CPU_ISSET(i, &cpuset)) affinity.insert(i);
 
    return affinity;
  }
 
  __INLINE__ ProcessingUnit(const std::shared_ptr<HiCR::ComputeResource> &computeResource)
    : HiCR::ProcessingUnit(computeResource)
  {
    // Getting up-casted pointer for the processing unit
    auto c = dynamic_pointer_cast<HiCR::backend::hwloc::ComputeResource>(computeResource);
 
    // Checking whether the execution unit passed is compatible with this backend
    if (c == nullptr) HICR_THROW_LOGIC("The passed compute resource is not supported by this processing unit type\n");
 
    // Creating initialization barrier
    _initializationBarrier = std::make_unique<pthread_barrier_t>();
  };
 
  private:
 
  pthread_t _pthreadId = 0;
 
  std::unique_ptr<HiCR::ExecutionState> _executionState;
 
  std::unique_ptr<pthread_barrier_t> _initializationBarrier;
 
  __INLINE__ static void *launchWrapper(void *p)
  {
    // Gathering thread object
    auto thread = static_cast<pthreads::ProcessingUnit *>(p);
 
    // Getting associated compute unit reference
    auto computeResource = dynamic_pointer_cast<HiCR::backend::hwloc::ComputeResource>(thread->getComputeResource());
 
    // Setting signal to hear for suspend/resume
    signal(HICR_SUSPEND_SIGNAL, ProcessingUnit::catchSuspendSignal);
    signal(HICR_RESUME_SIGNAL, ProcessingUnit::catchResumeSignal);
 
    // Setting initial thread affinity
    thread->updateAffinity(std::set<hwloc::ComputeResource::logicalProcessorId_t>({computeResource->getProcessorId()}));
 
    // Yielding execution to allow affinity to refresh
    sched_yield();
 
    // The thread has now been properly initialized
    pthread_barrier_wait(thread->_initializationBarrier.get());
 
    // Calling main loop
    thread->_executionState->resume();
 
    // No returns
    return nullptr;
  }
 
  __INLINE__ static void catchSuspendSignal(int sig)
  {
    int      status    = 0;
    int      signalSet = 0;
    sigset_t suspendSet;
 
    // Waiting for that signal to arrive
    status = sigaddset(&suspendSet, HICR_RESUME_SIGNAL);
    if (status != 0) HICR_THROW_RUNTIME("Could not set resume signal thread\n");
 
    status = sigwait(&suspendSet, &signalSet);
    if (status != 0) HICR_THROW_RUNTIME("Could not suspend thread\n");
  }
 
  __INLINE__ static void catchResumeSignal(int sig) {}
 
  __INLINE__ void initialize()
  {
    // Nothing to do for the initialization
  }
 
  __INLINE__ void suspend()
  {
    auto status = pthread_kill(_pthreadId, HICR_SUSPEND_SIGNAL);
    if (status != 0) HICR_THROW_RUNTIME("Could not suspend thread %lu\n", _pthreadId);
  }
 
  __INLINE__ void resume()
  {
    auto status = pthread_kill(_pthreadId, HICR_RESUME_SIGNAL);
    if (status != 0) HICR_THROW_RUNTIME("Could not resume thread %lu\n", _pthreadId);
  }
 
  __INLINE__ void start(std::unique_ptr<HiCR::ExecutionState> &executionState)
  {
    // Initializing barrier
    pthread_barrier_init(_initializationBarrier.get(), nullptr, 2);
 
    // Obtaining execution state
    _executionState = std::move(executionState);
 
    // Launching thread function wrapper
    auto status = pthread_create(&_pthreadId, nullptr, launchWrapper, this);
    if (status != 0) HICR_THROW_RUNTIME("Could not create thread %lu\n", _pthreadId);
 
    // Waiting for proper initialization of the thread
    pthread_barrier_wait(_initializationBarrier.get());
 
    // Destroying barrier
    pthread_barrier_destroy(_initializationBarrier.get());
  }
 
  __INLINE__ void terminate()
  {
    // Nothing to do actively, just wait for the thread to finalize in its own accord
  }
 
  __INLINE__ void await()
  {
    // Waiting for thread after execution
    pthread_join(_pthreadId, nullptr);
  }
};
 
} // namespace HiCR::backend::pthreads