lib/iomgr/combiner.cc

/*
 *
 * Copyright 2016 gRPC authors.
 *
 * 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 <grpc/support/port_platform.h>

#include "src/core/lib/iomgr/combiner.h"

#include <assert.h>
#include <inttypes.h>
#include <string.h>

#include <grpc/support/alloc.h>
#include <grpc/support/log.h>

#include "src/core/lib/debug/stats.h"
#include "src/core/lib/iomgr/executor.h"
#include "src/core/lib/profiling/timers.h"

grpc_core::DebugOnlyTraceFlag grpc_combiner_trace(false, "combiner");

#define GRPC_COMBINER_TRACE(fn)          \
  do {                                   \
    if (grpc_combiner_trace.enabled()) { \
      fn;                                \
    }                                    \
  } while (0)

#define STATE_UNORPHANED 1
#define STATE_ELEM_COUNT_LOW_BIT 2

struct grpc_combiner {
  grpc_combiner* next_combiner_on_this_exec_ctx;
  grpc_closure_scheduler scheduler;
  grpc_closure_scheduler finally_scheduler;
  gpr_mpscq queue;
  // either:
  // a pointer to the initiating exec ctx if that is the only exec_ctx that has
  // ever queued to this combiner, or NULL. If this is non-null, it's not
  // dereferencable (since the initiating exec_ctx may have gone out of scope)
  gpr_atm initiating_exec_ctx_or_null;
  // state is:
  // lower bit - zero if orphaned (STATE_UNORPHANED)
  // other bits - number of items queued on the lock (STATE_ELEM_COUNT_LOW_BIT)
  gpr_atm state;
  bool time_to_execute_final_list;
  grpc_closure_list final_list;
  grpc_closure offload;
  gpr_refcount refs;
};

static void combiner_run(grpc_closure* closure, grpc_error* error);
static void combiner_exec(grpc_closure* closure, grpc_error* error);
static void combiner_finally_exec(grpc_closure* closure, grpc_error* error);

static const grpc_closure_scheduler_vtable scheduler = {
    combiner_run, combiner_exec, "combiner:immediately"};
static const grpc_closure_scheduler_vtable finally_scheduler = {
    combiner_finally_exec, combiner_finally_exec, "combiner:finally"};

static void offload(void* arg, grpc_error* error);

grpc_combiner* grpc_combiner_create(void) {
  grpc_combiner* lock = static_cast<grpc_combiner*>(gpr_zalloc(sizeof(*lock)));
  gpr_ref_init(&lock->refs, 1);
  lock->scheduler.vtable = &scheduler;
  lock->finally_scheduler.vtable = &finally_scheduler;
  gpr_atm_no_barrier_store(&lock->state, STATE_UNORPHANED);
  gpr_mpscq_init(&lock->queue);
  grpc_closure_list_init(&lock->final_list);
  GRPC_CLOSURE_INIT(&lock->offload, offload, lock,
                    grpc_executor_scheduler(GRPC_EXECUTOR_SHORT));
  GRPC_COMBINER_TRACE(gpr_log(GPR_INFO, "C:%p create", lock));
  return lock;
}

static void really_destroy(grpc_combiner* lock) {
  GRPC_COMBINER_TRACE(gpr_log(GPR_INFO, "C:%p really_destroy", lock));
  GPR_ASSERT(gpr_atm_no_barrier_load(&lock->state) == 0);
  gpr_mpscq_destroy(&lock->queue);
  gpr_free(lock);
}

static void start_destroy(grpc_combiner* lock) {
  gpr_atm old_state = gpr_atm_full_fetch_add(&lock->state, -STATE_UNORPHANED);
  GRPC_COMBINER_TRACE(gpr_log(
      GPR_INFO, "C:%p really_destroy old_state=%" PRIdPTR, lock, old_state));
  if (old_state == 1) {
    really_destroy(lock);
  }
}

#ifndef NDEBUG
#define GRPC_COMBINER_DEBUG_SPAM(op, delta)                                \
  if (grpc_combiner_trace.enabled()) {                                     \
    gpr_log(file, line, GPR_LOG_SEVERITY_DEBUG,                            \
            "C:%p %s %" PRIdPTR " --> %" PRIdPTR " %s", lock, (op),        \
            gpr_atm_no_barrier_load(&lock->refs.count),                    \
            gpr_atm_no_barrier_load(&lock->refs.count) + (delta), reason); \
  }
#else
#define GRPC_COMBINER_DEBUG_SPAM(op, delta)
#endif

void grpc_combiner_unref(grpc_combiner* lock GRPC_COMBINER_DEBUG_ARGS) {
  GRPC_COMBINER_DEBUG_SPAM("UNREF", -1);
  if (gpr_unref(&lock->refs)) {
    start_destroy(lock);
  }
}

grpc_combiner* grpc_combiner_ref(grpc_combiner* lock GRPC_COMBINER_DEBUG_ARGS) {
  GRPC_COMBINER_DEBUG_SPAM("  REF", 1);
  gpr_ref(&lock->refs);
  return lock;
}

static void push_last_on_exec_ctx(grpc_combiner* lock) {
  lock->next_combiner_on_this_exec_ctx = nullptr;
  if (grpc_core::ExecCtx::Get()->combiner_data()->active_combiner == nullptr) {
    grpc_core::ExecCtx::Get()->combiner_data()->active_combiner =
        grpc_core::ExecCtx::Get()->combiner_data()->last_combiner = lock;
  } else {
    grpc_core::ExecCtx::Get()
        ->combiner_data()
        ->last_combiner->next_combiner_on_this_exec_ctx = lock;
    grpc_core::ExecCtx::Get()->combiner_data()->last_combiner = lock;
  }
}

static void push_first_on_exec_ctx(grpc_combiner* lock) {
  lock->next_combiner_on_this_exec_ctx =
      grpc_core::ExecCtx::Get()->combiner_data()->active_combiner;
  grpc_core::ExecCtx::Get()->combiner_data()->active_combiner = lock;
  if (lock->next_combiner_on_this_exec_ctx == nullptr) {
    grpc_core::ExecCtx::Get()->combiner_data()->last_combiner = lock;
  }
}

#define COMBINER_FROM_CLOSURE_SCHEDULER(closure, scheduler_name) \
  ((grpc_combiner*)(((char*)((closure)->scheduler)) -            \
                    offsetof(grpc_combiner, scheduler_name)))

static void combiner_exec(grpc_closure* cl, grpc_error* error) {
  GPR_TIMER_SCOPE("combiner.execute", 0);
  GRPC_STATS_INC_COMBINER_LOCKS_SCHEDULED_ITEMS();
  grpc_combiner* lock = COMBINER_FROM_CLOSURE_SCHEDULER(cl, scheduler);
  gpr_atm last = gpr_atm_full_fetch_add(&lock->state, STATE_ELEM_COUNT_LOW_BIT);
  GRPC_COMBINER_TRACE(gpr_log(GPR_INFO,
                              "C:%p grpc_combiner_execute c=%p last=%" PRIdPTR,
                              lock, cl, last));
  if (last == 1) {
    GRPC_STATS_INC_COMBINER_LOCKS_INITIATED();
    GPR_TIMER_MARK("combiner.initiated", 0);
    gpr_atm_no_barrier_store(&lock->initiating_exec_ctx_or_null,
                             (gpr_atm)grpc_core::ExecCtx::Get());
    // first element on this list: add it to the list of combiner locks
    // executing within this exec_ctx
    push_last_on_exec_ctx(lock);
  } else {
    // there may be a race with setting here: if that happens, we may delay
    // offload for one or two actions, and that's fine
    gpr_atm initiator =
        gpr_atm_no_barrier_load(&lock->initiating_exec_ctx_or_null);
    if (initiator != 0 && initiator != (gpr_atm)grpc_core::ExecCtx::Get()) {
      gpr_atm_no_barrier_store(&lock->initiating_exec_ctx_or_null, 0);
    }
  }
  GPR_ASSERT(last & STATE_UNORPHANED);  // ensure lock has not been destroyed
  assert(cl->cb);
  cl->error_data.error = error;
  gpr_mpscq_push(&lock->queue, &cl->next_data.atm_next);
}

static void move_next() {
  grpc_core::ExecCtx::Get()->combiner_data()->active_combiner =
      grpc_core::ExecCtx::Get()
          ->combiner_data()
          ->active_combiner->next_combiner_on_this_exec_ctx;
  if (grpc_core::ExecCtx::Get()->combiner_data()->active_combiner == nullptr) {
    grpc_core::ExecCtx::Get()->combiner_data()->last_combiner = nullptr;
  }
}

static void offload(void* arg, grpc_error* error) {
  grpc_combiner* lock = static_cast<grpc_combiner*>(arg);
  push_last_on_exec_ctx(lock);
}

static void queue_offload(grpc_combiner* lock) {
  GRPC_STATS_INC_COMBINER_LOCKS_OFFLOADED();
  move_next();
  GRPC_COMBINER_TRACE(gpr_log(GPR_INFO, "C:%p queue_offload", lock));
  GRPC_CLOSURE_SCHED(&lock->offload, GRPC_ERROR_NONE);
}

bool grpc_combiner_continue_exec_ctx() {
  GPR_TIMER_SCOPE("combiner.continue_exec_ctx", 0);
  grpc_combiner* lock =
      grpc_core::ExecCtx::Get()->combiner_data()->active_combiner;
  if (lock == nullptr) {
    return false;
  }

  bool contended =
      gpr_atm_no_barrier_load(&lock->initiating_exec_ctx_or_null) == 0;

  GRPC_COMBINER_TRACE(gpr_log(GPR_INFO,
                              "C:%p grpc_combiner_continue_exec_ctx "
                              "contended=%d "
                              "exec_ctx_ready_to_finish=%d "
                              "time_to_execute_final_list=%d",
                              lock, contended,
                              grpc_core::ExecCtx::Get()->IsReadyToFinish(),
                              lock->time_to_execute_final_list));

  if (contended && grpc_core::ExecCtx::Get()->IsReadyToFinish() &&
      grpc_executor_is_threaded()) {
    GPR_TIMER_MARK("offload_from_finished_exec_ctx", 0);
    // this execution context wants to move on: schedule remaining work to be
    // picked up on the executor
    queue_offload(lock);
    return true;
  }

  if (!lock->time_to_execute_final_list ||
      // peek to see if something new has shown up, and execute that with
      // priority
      (gpr_atm_acq_load(&lock->state) >> 1) > 1) {
    gpr_mpscq_node* n = gpr_mpscq_pop(&lock->queue);
    GRPC_COMBINER_TRACE(
        gpr_log(GPR_INFO, "C:%p maybe_finish_one n=%p", lock, n));
    if (n == nullptr) {
      // queue is in an inconsistent state: use this as a cue that we should
      // go off and do something else for a while (and come back later)
      GPR_TIMER_MARK("delay_busy", 0);
      queue_offload(lock);
      return true;
    }
    GPR_TIMER_SCOPE("combiner.exec1", 0);
    grpc_closure* cl = reinterpret_cast<grpc_closure*>(n);
    grpc_error* cl_err = cl->error_data.error;
#ifndef NDEBUG
    cl->scheduled = false;
#endif
    cl->cb(cl->cb_arg, cl_err);
    GRPC_ERROR_UNREF(cl_err);
  } else {
    grpc_closure* c = lock->final_list.head;
    GPR_ASSERT(c != nullptr);
    grpc_closure_list_init(&lock->final_list);
    int loops = 0;
    while (c != nullptr) {
      GPR_TIMER_SCOPE("combiner.exec_1final", 0);
      GRPC_COMBINER_TRACE(
          gpr_log(GPR_INFO, "C:%p execute_final[%d] c=%p", lock, loops, c));
      grpc_closure* next = c->next_data.next;
      grpc_error* error = c->error_data.error;
#ifndef NDEBUG
      c->scheduled = false;
#endif
      c->cb(c->cb_arg, error);
      GRPC_ERROR_UNREF(error);
      c = next;
    }
  }

  GPR_TIMER_MARK("unref", 0);
  move_next();
  lock->time_to_execute_final_list = false;
  gpr_atm old_state =
      gpr_atm_full_fetch_add(&lock->state, -STATE_ELEM_COUNT_LOW_BIT);
  GRPC_COMBINER_TRACE(
      gpr_log(GPR_INFO, "C:%p finish old_state=%" PRIdPTR, lock, old_state));
// Define a macro to ease readability of the following switch statement.
#define OLD_STATE_WAS(orphaned, elem_count) \
  (((orphaned) ? 0 : STATE_UNORPHANED) |    \
   ((elem_count)*STATE_ELEM_COUNT_LOW_BIT))
  // Depending on what the previous state was, we need to perform different
  // actions.
  switch (old_state) {
    default:
      // we have multiple queued work items: just continue executing them
      break;
    case OLD_STATE_WAS(false, 2):
    case OLD_STATE_WAS(true, 2):
      // we're down to one queued item: if it's the final list we should do that
      if (!grpc_closure_list_empty(lock->final_list)) {
        lock->time_to_execute_final_list = true;
      }
      break;
    case OLD_STATE_WAS(false, 1):
      // had one count, one unorphaned --> unlocked unorphaned
      return true;
    case OLD_STATE_WAS(true, 1):
      // and one count, one orphaned --> unlocked and orphaned
      really_destroy(lock);
      return true;
    case OLD_STATE_WAS(false, 0):
    case OLD_STATE_WAS(true, 0):
      // these values are illegal - representing an already unlocked or
      // deleted lock
      GPR_UNREACHABLE_CODE(return true);
  }
  push_first_on_exec_ctx(lock);
  return true;
}

static void enqueue_finally(void* closure, grpc_error* error);

static void combiner_finally_exec(grpc_closure* closure, grpc_error* error) {
  GPR_TIMER_SCOPE("combiner.execute_finally", 0);
  GRPC_STATS_INC_COMBINER_LOCKS_SCHEDULED_FINAL_ITEMS();
  grpc_combiner* lock =
      COMBINER_FROM_CLOSURE_SCHEDULER(closure, finally_scheduler);
  GRPC_COMBINER_TRACE(gpr_log(
      GPR_INFO, "C:%p grpc_combiner_execute_finally c=%p; ac=%p", lock, closure,
      grpc_core::ExecCtx::Get()->combiner_data()->active_combiner));
  if (grpc_core::ExecCtx::Get()->combiner_data()->active_combiner != lock) {
    GPR_TIMER_MARK("slowpath", 0);
    GRPC_CLOSURE_SCHED(GRPC_CLOSURE_CREATE(enqueue_finally, closure,
                                           grpc_combiner_scheduler(lock)),
                       error);
    return;
  }

  if (grpc_closure_list_empty(lock->final_list)) {
    gpr_atm_full_fetch_add(&lock->state, STATE_ELEM_COUNT_LOW_BIT);
  }
  grpc_closure_list_append(&lock->final_list, closure, error);
}

static void combiner_run(grpc_closure* closure, grpc_error* error) {
  grpc_combiner* lock = COMBINER_FROM_CLOSURE_SCHEDULER(closure, scheduler);
#ifndef NDEBUG
  closure->scheduled = false;
  GRPC_COMBINER_TRACE(gpr_log(
      GPR_DEBUG,
      "Combiner:%p grpc_combiner_run closure:%p created [%s:%d] run [%s:%d]",
      lock, closure, closure->file_created, closure->line_created,
      closure->file_initiated, closure->line_initiated));
#endif
  GPR_ASSERT(grpc_core::ExecCtx::Get()->combiner_data()->active_combiner ==
             lock);
  closure->cb(closure->cb_arg, error);
  GRPC_ERROR_UNREF(error);
}

static void enqueue_finally(void* closure, grpc_error* error) {
  combiner_finally_exec(static_cast<grpc_closure*>(closure),
                        GRPC_ERROR_REF(error));
}

grpc_closure_scheduler* grpc_combiner_scheduler(grpc_combiner* combiner) {
  return &combiner->scheduler;
}

grpc_closure_scheduler* grpc_combiner_finally_scheduler(
    grpc_combiner* combiner) {
  return &combiner->finally_scheduler;
}