src/core/ext/filters/client_channel/lb_policy/grpclb/grpclb.c

/*
 *
 * 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.
 *
 */

/** Implementation of the gRPC LB policy.
 *
 * This policy takes as input a set of resolved addresses {a1..an} for which the
 * LB set was set (it's the resolver's responsibility to ensure this). That is
 * to say, {a1..an} represent a collection of LB servers.
 *
 * An internal channel (\a glb_lb_policy.lb_channel) is created over {a1..an}.
 * This channel behaves just like a regular channel. In particular, the
 * constructed URI over the addresses a1..an will use the default pick first
 * policy to select from this list of LB server backends.
 *
 * The first time the policy gets a request for a pick, a ping, or to exit the
 * idle state, \a query_for_backends_locked() is called. This function sets up
 * and initiates the internal communication with the LB server. In particular,
 * it's responsible for instantiating the internal *streaming* call to the LB
 * server (whichever address from {a1..an} pick-first chose). This call is
 * serviced by two callbacks, \a lb_on_server_status_received and \a
 * lb_on_response_received. The former will be called when the call to the LB
 * server completes. This can happen if the LB server closes the connection or
 * if this policy itself cancels the call (for example because it's shutting
 * down). If the internal call times out, the usual behavior of pick-first
 * applies, continuing to pick from the list {a1..an}.
 *
 * Upon sucesss, the incoming \a LoadBalancingResponse is processed by \a
 * res_recv. An invalid one results in the termination of the streaming call. A
 * new streaming call should be created if possible, failing the original call
 * otherwise. For a valid \a LoadBalancingResponse, the server list of actual
 * backends is extracted. A Round Robin policy will be created from this list.
 * There are two possible scenarios:
 *
 * 1. This is the first server list received. There was no previous instance of
 *    the Round Robin policy. \a rr_handover_locked() will instantiate the RR
 *    policy and perform all the pending operations over it.
 * 2. There's already a RR policy instance active. We need to introduce the new
 *    one build from the new serverlist, but taking care not to disrupt the
 *    operations in progress over the old RR instance. This is done by
 *    decreasing the reference count on the old policy. The moment no more
 *    references are held on the old RR policy, it'll be destroyed and \a
 *    glb_rr_connectivity_changed notified with a \a GRPC_CHANNEL_SHUTDOWN
 *    state. At this point we can transition to a new RR instance safely, which
 *    is done once again via \a rr_handover_locked().
 *
 *
 * Once a RR policy instance is in place (and getting updated as described),
 * calls to for a pick, a ping or a cancellation will be serviced right away by
 * forwarding them to the RR instance. Any time there's no RR policy available
 * (ie, right after the creation of the gRPCLB policy, if an empty serverlist is
 * received, etc), pick/ping requests are added to a list of pending picks/pings
 * to be flushed and serviced as part of \a rr_handover_locked() the moment the
 * RR policy instance becomes available.
 *
 * \see https://github.com/grpc/grpc/blob/master/doc/load-balancing.md for the
 * high level design and details. */

/* TODO(dgq):
 * - Implement LB service forwarding (point 2c. in the doc's diagram).
 */

/* With the addition of a libuv endpoint, sockaddr.h now includes uv.h when
   using that endpoint. Because of various transitive includes in uv.h,
   including windows.h on Windows, uv.h must be included before other system
   headers. Therefore, sockaddr.h must always be included first */
#include "src/core/lib/iomgr/sockaddr.h"

#include <limits.h>
#include <string.h>

#include <grpc/byte_buffer_reader.h>
#include <grpc/grpc.h>
#include <grpc/support/alloc.h>
#include <grpc/support/host_port.h>
#include <grpc/support/string_util.h>
#include <grpc/support/time.h>

#include "src/core/ext/filters/client_channel/client_channel.h"
#include "src/core/ext/filters/client_channel/client_channel_factory.h"
#include "src/core/ext/filters/client_channel/lb_policy/grpclb/client_load_reporting_filter.h"
#include "src/core/ext/filters/client_channel/lb_policy/grpclb/grpclb.h"
#include "src/core/ext/filters/client_channel/lb_policy/grpclb/grpclb_channel.h"
#include "src/core/ext/filters/client_channel/lb_policy/grpclb/grpclb_client_stats.h"
#include "src/core/ext/filters/client_channel/lb_policy/grpclb/load_balancer_api.h"
#include "src/core/ext/filters/client_channel/lb_policy_factory.h"
#include "src/core/ext/filters/client_channel/lb_policy_registry.h"
#include "src/core/ext/filters/client_channel/parse_address.h"
#include "src/core/ext/filters/client_channel/resolver/fake/fake_resolver.h"
#include "src/core/ext/filters/client_channel/subchannel_index.h"
#include "src/core/lib/channel/channel_args.h"
#include "src/core/lib/channel/channel_stack.h"
#include "src/core/lib/iomgr/combiner.h"
#include "src/core/lib/iomgr/sockaddr.h"
#include "src/core/lib/iomgr/sockaddr_utils.h"
#include "src/core/lib/iomgr/timer.h"
#include "src/core/lib/slice/slice_hash_table.h"
#include "src/core/lib/slice/slice_internal.h"
#include "src/core/lib/slice/slice_string_helpers.h"
#include "src/core/lib/support/backoff.h"
#include "src/core/lib/support/string.h"
#include "src/core/lib/surface/call.h"
#include "src/core/lib/surface/channel.h"
#include "src/core/lib/surface/channel_init.h"
#include "src/core/lib/transport/static_metadata.h"

#define GRPC_GRPCLB_MIN_CONNECT_TIMEOUT_SECONDS 20
#define GRPC_GRPCLB_INITIAL_CONNECT_BACKOFF_SECONDS 1
#define GRPC_GRPCLB_RECONNECT_BACKOFF_MULTIPLIER 1.6
#define GRPC_GRPCLB_RECONNECT_MAX_BACKOFF_SECONDS 120
#define GRPC_GRPCLB_RECONNECT_JITTER 0.2

grpc_tracer_flag grpc_lb_glb_trace = GRPC_TRACER_INITIALIZER(false, "glb");

/* add lb_token of selected subchannel (address) to the call's initial
 * metadata */
static grpc_error *initial_metadata_add_lb_token(
    grpc_exec_ctx *exec_ctx, grpc_metadata_batch *initial_metadata,
    grpc_linked_mdelem *lb_token_mdelem_storage, grpc_mdelem lb_token) {
  GPR_ASSERT(lb_token_mdelem_storage != NULL);
  GPR_ASSERT(!GRPC_MDISNULL(lb_token));
  return grpc_metadata_batch_add_tail(exec_ctx, initial_metadata,
                                      lb_token_mdelem_storage, lb_token);
}

static void destroy_client_stats(void *arg) {
  grpc_grpclb_client_stats_unref((grpc_grpclb_client_stats *)arg);
}

typedef struct wrapped_rr_closure_arg {
  /* the closure instance using this struct as argument */
  grpc_closure wrapper_closure;

  /* the original closure. Usually a on_complete/notify cb for pick() and ping()
   * calls against the internal RR instance, respectively. */
  grpc_closure *wrapped_closure;

  /* the pick's initial metadata, kept in order to append the LB token for the
   * pick */
  grpc_metadata_batch *initial_metadata;

  /* the picked target, used to determine which LB token to add to the pick's
   * initial metadata */
  grpc_connected_subchannel **target;

  /* the context to be populated for the subchannel call */
  grpc_call_context_element *context;

  /* Stats for client-side load reporting. Note that this holds a
   * reference, which must be either passed on via context or unreffed. */
  grpc_grpclb_client_stats *client_stats;

  /* the LB token associated with the pick */
  grpc_mdelem lb_token;

  /* storage for the lb token initial metadata mdelem */
  grpc_linked_mdelem *lb_token_mdelem_storage;

  /* The RR instance related to the closure */
  grpc_lb_policy *rr_policy;

  /* heap memory to be freed upon closure execution. */
  void *free_when_done;
} wrapped_rr_closure_arg;

/* The \a on_complete closure passed as part of the pick requires keeping a
 * reference to its associated round robin instance. We wrap this closure in
 * order to unref the round robin instance upon its invocation */
static void wrapped_rr_closure(grpc_exec_ctx *exec_ctx, void *arg,
                               grpc_error *error) {
  wrapped_rr_closure_arg *wc_arg = (wrapped_rr_closure_arg *)arg;

  GPR_ASSERT(wc_arg->wrapped_closure != NULL);
  GRPC_CLOSURE_SCHED(exec_ctx, wc_arg->wrapped_closure, GRPC_ERROR_REF(error));

  if (wc_arg->rr_policy != NULL) {
    /* if *target is NULL, no pick has been made by the RR policy (eg, all
     * addresses failed to connect). There won't be any user_data/token
     * available */
    if (*wc_arg->target != NULL) {
      if (!GRPC_MDISNULL(wc_arg->lb_token)) {
        initial_metadata_add_lb_token(exec_ctx, wc_arg->initial_metadata,
                                      wc_arg->lb_token_mdelem_storage,
                                      GRPC_MDELEM_REF(wc_arg->lb_token));
      } else {
        gpr_log(GPR_ERROR,
                "No LB token for connected subchannel pick %p (from RR "
                "instance %p).",
                (void *)*wc_arg->target, (void *)wc_arg->rr_policy);
        abort();
      }
      // Pass on client stats via context. Passes ownership of the reference.
      GPR_ASSERT(wc_arg->client_stats != NULL);
      wc_arg->context[GRPC_GRPCLB_CLIENT_STATS].value = wc_arg->client_stats;
      wc_arg->context[GRPC_GRPCLB_CLIENT_STATS].destroy = destroy_client_stats;
    } else {
      grpc_grpclb_client_stats_unref(wc_arg->client_stats);
    }
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Unreffing RR %p", (void *)wc_arg->rr_policy);
    }
    GRPC_LB_POLICY_UNREF(exec_ctx, wc_arg->rr_policy, "wrapped_rr_closure");
  }
  GPR_ASSERT(wc_arg->free_when_done != NULL);
  gpr_free(wc_arg->free_when_done);
}

/* Linked list of pending pick requests. It stores all information needed to
 * eventually call (Round Robin's) pick() on them. They mainly stay pending
 * waiting for the RR policy to be created/updated.
 *
 * One particularity is the wrapping of the user-provided \a on_complete closure
 * (in \a wrapped_on_complete and \a wrapped_on_complete_arg). This is needed in
 * order to correctly unref the RR policy instance upon completion of the pick.
 * See \a wrapped_rr_closure for details. */
typedef struct pending_pick {
  struct pending_pick *next;

  /* original pick()'s arguments */
  grpc_lb_policy_pick_args pick_args;

  /* output argument where to store the pick()ed connected subchannel, or NULL
   * upon error. */
  grpc_connected_subchannel **target;

  /* args for wrapped_on_complete */
  wrapped_rr_closure_arg wrapped_on_complete_arg;
} pending_pick;

static void add_pending_pick(pending_pick **root,
                             const grpc_lb_policy_pick_args *pick_args,
                             grpc_connected_subchannel **target,
                             grpc_call_context_element *context,
                             grpc_closure *on_complete) {
  pending_pick *pp = (pending_pick *)gpr_zalloc(sizeof(*pp));
  pp->next = *root;
  pp->pick_args = *pick_args;
  pp->target = target;
  pp->wrapped_on_complete_arg.wrapped_closure = on_complete;
  pp->wrapped_on_complete_arg.target = target;
  pp->wrapped_on_complete_arg.context = context;
  pp->wrapped_on_complete_arg.initial_metadata = pick_args->initial_metadata;
  pp->wrapped_on_complete_arg.lb_token_mdelem_storage =
      pick_args->lb_token_mdelem_storage;
  pp->wrapped_on_complete_arg.free_when_done = pp;
  GRPC_CLOSURE_INIT(&pp->wrapped_on_complete_arg.wrapper_closure,
                    wrapped_rr_closure, &pp->wrapped_on_complete_arg,
                    grpc_schedule_on_exec_ctx);
  *root = pp;
}

/* Same as the \a pending_pick struct but for ping operations */
typedef struct pending_ping {
  struct pending_ping *next;

  /* args for wrapped_notify */
  wrapped_rr_closure_arg wrapped_notify_arg;
} pending_ping;

static void add_pending_ping(pending_ping **root, grpc_closure *notify) {
  pending_ping *pping = (pending_ping *)gpr_zalloc(sizeof(*pping));
  pping->wrapped_notify_arg.wrapped_closure = notify;
  pping->wrapped_notify_arg.free_when_done = pping;
  pping->next = *root;
  GRPC_CLOSURE_INIT(&pping->wrapped_notify_arg.wrapper_closure,
                    wrapped_rr_closure, &pping->wrapped_notify_arg,
                    grpc_schedule_on_exec_ctx);
  *root = pping;
}

/*
 * glb_lb_policy
 */
typedef struct rr_connectivity_data rr_connectivity_data;

typedef struct glb_lb_policy {
  /** base policy: must be first */
  grpc_lb_policy base;

  /** who the client is trying to communicate with */
  const char *server_name;
  grpc_client_channel_factory *cc_factory;
  grpc_channel_args *args;

  /** timeout in milliseconds for the LB call. 0 means no deadline. */
  int lb_call_timeout_ms;

  /** for communicating with the LB server */
  grpc_channel *lb_channel;

  /** response generator to inject address updates into \a lb_channel */
  grpc_fake_resolver_response_generator *response_generator;

  /** the RR policy to use of the backend servers returned by the LB server */
  grpc_lb_policy *rr_policy;

  bool started_picking;

  /** our connectivity state tracker */
  grpc_connectivity_state_tracker state_tracker;

  /** connectivity state of the LB channel */
  grpc_connectivity_state lb_channel_connectivity;

  /** stores the deserialized response from the LB. May be NULL until one such
   * response has arrived. */
  grpc_grpclb_serverlist *serverlist;

  /** Index into serverlist for next pick.
   * If the server at this index is a drop, we return a drop.
   * Otherwise, we delegate to the RR policy. */
  size_t serverlist_index;

  /** list of picks that are waiting on RR's policy connectivity */
  pending_pick *pending_picks;

  /** list of pings that are waiting on RR's policy connectivity */
  pending_ping *pending_pings;

  bool shutting_down;

  /** are we currently updating lb_call? */
  bool updating_lb_call;

  /** are we currently updating lb_channel? */
  bool updating_lb_channel;

  /** are we already watching the LB channel's connectivity? */
  bool watching_lb_channel;

  /** is \a lb_call_retry_timer active? */
  bool retry_timer_active;

  /** called upon changes to the LB channel's connectivity. */
  grpc_closure lb_channel_on_connectivity_changed;

  /** args from the latest update received while already updating, or NULL */
  grpc_lb_policy_args *pending_update_args;

  /************************************************************/
  /*  client data associated with the LB server communication */
  /************************************************************/
  /* Finished sending initial request. */
  grpc_closure lb_on_sent_initial_request;

  /* Status from the LB server has been received. This signals the end of the LB
   * call. */
  grpc_closure lb_on_server_status_received;

  /* A response from the LB server has been received. Process it */
  grpc_closure lb_on_response_received;

  /* LB call retry timer callback. */
  grpc_closure lb_on_call_retry;

  grpc_call *lb_call; /* streaming call to the LB server, */

  grpc_metadata_array lb_initial_metadata_recv; /* initial MD from LB server */
  grpc_metadata_array
      lb_trailing_metadata_recv; /* trailing MD from LB server */

  /* what's being sent to the LB server. Note that its value may vary if the LB
   * server indicates a redirect. */
  grpc_byte_buffer *lb_request_payload;

  /* response the LB server, if any. Processed in lb_on_response_received() */
  grpc_byte_buffer *lb_response_payload;

  /* call status code and details, set in lb_on_server_status_received() */
  grpc_status_code lb_call_status;
  grpc_slice lb_call_status_details;

  /** LB call retry backoff state */
  gpr_backoff lb_call_backoff_state;

  /** LB call retry timer */
  grpc_timer lb_call_retry_timer;

  bool initial_request_sent;
  bool seen_initial_response;

  /* Stats for client-side load reporting. Should be unreffed and
   * recreated whenever lb_call is replaced. */
  grpc_grpclb_client_stats *client_stats;
  /* Interval and timer for next client load report. */
  gpr_timespec client_stats_report_interval;
  grpc_timer client_load_report_timer;
  bool client_load_report_timer_pending;
  bool last_client_load_report_counters_were_zero;
  /* Closure used for either the load report timer or the callback for
   * completion of sending the load report. */
  grpc_closure client_load_report_closure;
  /* Client load report message payload. */
  grpc_byte_buffer *client_load_report_payload;
} glb_lb_policy;

/* Keeps track and reacts to changes in connectivity of the RR instance */
struct rr_connectivity_data {
  grpc_closure on_change;
  grpc_connectivity_state state;
  glb_lb_policy *glb_policy;
};

static bool is_server_valid(const grpc_grpclb_server *server, size_t idx,
                            bool log) {
  if (server->drop) return false;
  const grpc_grpclb_ip_address *ip = &server->ip_address;
  if (server->port >> 16 != 0) {
    if (log) {
      gpr_log(GPR_ERROR,
              "Invalid port '%d' at index %lu of serverlist. Ignoring.",
              server->port, (unsigned long)idx);
    }
    return false;
  }
  if (ip->size != 4 && ip->size != 16) {
    if (log) {
      gpr_log(GPR_ERROR,
              "Expected IP to be 4 or 16 bytes, got %d at index %lu of "
              "serverlist. Ignoring",
              ip->size, (unsigned long)idx);
    }
    return false;
  }
  return true;
}

/* vtable for LB tokens in grpc_lb_addresses. */
static void *lb_token_copy(void *token) {
  return token == NULL
             ? NULL
             : (void *)GRPC_MDELEM_REF((grpc_mdelem){(uintptr_t)token}).payload;
}
static void lb_token_destroy(grpc_exec_ctx *exec_ctx, void *token) {
  if (token != NULL) {
    GRPC_MDELEM_UNREF(exec_ctx, (grpc_mdelem){(uintptr_t)token});
  }
}
static int lb_token_cmp(void *token1, void *token2) {
  if (token1 > token2) return 1;
  if (token1 < token2) return -1;
  return 0;
}
static const grpc_lb_user_data_vtable lb_token_vtable = {
    lb_token_copy, lb_token_destroy, lb_token_cmp};

static void parse_server(const grpc_grpclb_server *server,
                         grpc_resolved_address *addr) {
  memset(addr, 0, sizeof(*addr));
  if (server->drop) return;
  const uint16_t netorder_port = htons((uint16_t)server->port);
  /* the addresses are given in binary format (a in(6)_addr struct) in
   * server->ip_address.bytes. */
  const grpc_grpclb_ip_address *ip = &server->ip_address;
  if (ip->size == 4) {
    addr->len = sizeof(struct sockaddr_in);
    struct sockaddr_in *addr4 = (struct sockaddr_in *)&addr->addr;
    addr4->sin_family = AF_INET;
    memcpy(&addr4->sin_addr, ip->bytes, ip->size);
    addr4->sin_port = netorder_port;
  } else if (ip->size == 16) {
    addr->len = sizeof(struct sockaddr_in6);
    struct sockaddr_in6 *addr6 = (struct sockaddr_in6 *)&addr->addr;
    addr6->sin6_family = AF_INET6;
    memcpy(&addr6->sin6_addr, ip->bytes, ip->size);
    addr6->sin6_port = netorder_port;
  }
}

/* Returns addresses extracted from \a serverlist. */
static grpc_lb_addresses *process_serverlist_locked(
    grpc_exec_ctx *exec_ctx, const grpc_grpclb_serverlist *serverlist) {
  size_t num_valid = 0;
  /* first pass: count how many are valid in order to allocate the necessary
   * memory in a single block */
  for (size_t i = 0; i < serverlist->num_servers; ++i) {
    if (is_server_valid(serverlist->servers[i], i, true)) ++num_valid;
  }
  grpc_lb_addresses *lb_addresses =
      grpc_lb_addresses_create(num_valid, &lb_token_vtable);
  /* second pass: actually populate the addresses and LB tokens (aka user data
   * to the outside world) to be read by the RR policy during its creation.
   * Given that the validity tests are very cheap, they are performed again
   * instead of marking the valid ones during the first pass, as this would
   * incurr in an allocation due to the arbitrary number of server */
  size_t addr_idx = 0;
  for (size_t sl_idx = 0; sl_idx < serverlist->num_servers; ++sl_idx) {
    const grpc_grpclb_server *server = serverlist->servers[sl_idx];
    if (!is_server_valid(serverlist->servers[sl_idx], sl_idx, false)) continue;
    GPR_ASSERT(addr_idx < num_valid);
    /* address processing */
    grpc_resolved_address addr;
    parse_server(server, &addr);
    /* lb token processing */
    void *user_data;
    if (server->has_load_balance_token) {
      const size_t lb_token_max_length =
          GPR_ARRAY_SIZE(server->load_balance_token);
      const size_t lb_token_length =
          strnlen(server->load_balance_token, lb_token_max_length);
      grpc_slice lb_token_mdstr = grpc_slice_from_copied_buffer(
          server->load_balance_token, lb_token_length);
      user_data = (void *)grpc_mdelem_from_slices(exec_ctx, GRPC_MDSTR_LB_TOKEN,
                                                  lb_token_mdstr)
                      .payload;
    } else {
      char *uri = grpc_sockaddr_to_uri(&addr);
      gpr_log(GPR_INFO,
              "Missing LB token for backend address '%s'. The empty token will "
              "be used instead",
              uri);
      gpr_free(uri);
      user_data = (void *)GRPC_MDELEM_LB_TOKEN_EMPTY.payload;
    }

    grpc_lb_addresses_set_address(lb_addresses, addr_idx, &addr.addr, addr.len,
                                  false /* is_balancer */,
                                  NULL /* balancer_name */, user_data);
    ++addr_idx;
  }
  GPR_ASSERT(addr_idx == num_valid);
  return lb_addresses;
}

static void update_lb_connectivity_status_locked(
    grpc_exec_ctx *exec_ctx, glb_lb_policy *glb_policy,
    grpc_connectivity_state rr_state, grpc_error *rr_state_error) {
  const grpc_connectivity_state curr_glb_state =
      grpc_connectivity_state_check(&glb_policy->state_tracker);

  /* The new connectivity status is a function of the previous one and the new
   * input coming from the status of the RR policy.
   *
   *  current state (grpclb's)
   *  |
   *  v  || I  |  C  |  R  |  TF  |  SD  |  <- new state (RR's)
   *  ===++====+=====+=====+======+======+
   *   I || I  |  C  |  R  | [I]  | [I]  |
   *  ---++----+-----+-----+------+------+
   *   C || I  |  C  |  R  | [C]  | [C]  |
   *  ---++----+-----+-----+------+------+
   *   R || I  |  C  |  R  | [R]  | [R]  |
   *  ---++----+-----+-----+------+------+
   *  TF || I  |  C  |  R  | [TF] | [TF] |
   *  ---++----+-----+-----+------+------+
   *  SD || NA |  NA |  NA |  NA  |  NA  | (*)
   *  ---++----+-----+-----+------+------+
   *
   * A [STATE] indicates that the old RR policy is kept. In those cases, STATE
   * is the current state of grpclb, which is left untouched.
   *
   *  In summary, if the new state is TRANSIENT_FAILURE or SHUTDOWN, stick to
   *  the previous RR instance.
   *
   *  Note that the status is never updated to SHUTDOWN as a result of calling
   *  this function. Only glb_shutdown() has the power to set that state.
   *
   *  (*) This function mustn't be called during shutting down. */
  GPR_ASSERT(curr_glb_state != GRPC_CHANNEL_SHUTDOWN);

  switch (rr_state) {
    case GRPC_CHANNEL_TRANSIENT_FAILURE:
    case GRPC_CHANNEL_SHUTDOWN:
      GPR_ASSERT(rr_state_error != GRPC_ERROR_NONE);
      break;
    case GRPC_CHANNEL_INIT:
    case GRPC_CHANNEL_IDLE:
    case GRPC_CHANNEL_CONNECTING:
    case GRPC_CHANNEL_READY:
      GPR_ASSERT(rr_state_error == GRPC_ERROR_NONE);
  }

  if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
    gpr_log(
        GPR_INFO, "Setting grpclb's state to %s from new RR policy %p state.",
        grpc_connectivity_state_name(rr_state), (void *)glb_policy->rr_policy);
  }
  grpc_connectivity_state_set(exec_ctx, &glb_policy->state_tracker, rr_state,
                              rr_state_error,
                              "update_lb_connectivity_status_locked");
}

/* Perform a pick over \a glb_policy->rr_policy. Given that a pick can return
 * immediately (ignoring its completion callback), we need to perform the
 * cleanups this callback would otherwise be resposible for.
 * If \a force_async is true, then we will manually schedule the
 * completion callback even if the pick is available immediately. */
static bool pick_from_internal_rr_locked(
    grpc_exec_ctx *exec_ctx, glb_lb_policy *glb_policy,
    const grpc_lb_policy_pick_args *pick_args, bool force_async,
    grpc_connected_subchannel **target, wrapped_rr_closure_arg *wc_arg) {
  // Look at the index into the serverlist to see if we should drop this call.
  grpc_grpclb_server *server =
      glb_policy->serverlist->servers[glb_policy->serverlist_index++];
  if (glb_policy->serverlist_index == glb_policy->serverlist->num_servers) {
    glb_policy->serverlist_index = 0;  // Wrap-around.
  }
  if (server->drop) {
    // Not using the RR policy, so unref it.
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Unreffing RR for drop (0x%" PRIxPTR ")",
              (intptr_t)wc_arg->rr_policy);
    }
    GRPC_LB_POLICY_UNREF(exec_ctx, wc_arg->rr_policy, "glb_pick_sync");
    // Update client load reporting stats to indicate the number of
    // dropped calls.  Note that we have to do this here instead of in
    // the client_load_reporting filter, because we do not create a
    // subchannel call (and therefore no client_load_reporting filter)
    // for dropped calls.
    grpc_grpclb_client_stats_add_call_dropped_locked(server->load_balance_token,
                                                     wc_arg->client_stats);
    grpc_grpclb_client_stats_unref(wc_arg->client_stats);
    if (force_async) {
      GPR_ASSERT(wc_arg->wrapped_closure != NULL);
      GRPC_CLOSURE_SCHED(exec_ctx, wc_arg->wrapped_closure, GRPC_ERROR_NONE);
      gpr_free(wc_arg->free_when_done);
      return false;
    }
    gpr_free(wc_arg->free_when_done);
    return true;
  }
  // Pick via the RR policy.
  const bool pick_done = grpc_lb_policy_pick_locked(
      exec_ctx, wc_arg->rr_policy, pick_args, target, wc_arg->context,
      (void **)&wc_arg->lb_token, &wc_arg->wrapper_closure);
  if (pick_done) {
    /* synchronous grpc_lb_policy_pick call. Unref the RR policy. */
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Unreffing RR (0x%" PRIxPTR ")",
              (intptr_t)wc_arg->rr_policy);
    }
    GRPC_LB_POLICY_UNREF(exec_ctx, wc_arg->rr_policy, "glb_pick_sync");
    /* add the load reporting initial metadata */
    initial_metadata_add_lb_token(exec_ctx, pick_args->initial_metadata,
                                  pick_args->lb_token_mdelem_storage,
                                  GRPC_MDELEM_REF(wc_arg->lb_token));
    // Pass on client stats via context. Passes ownership of the reference.
    GPR_ASSERT(wc_arg->client_stats != NULL);
    wc_arg->context[GRPC_GRPCLB_CLIENT_STATS].value = wc_arg->client_stats;
    wc_arg->context[GRPC_GRPCLB_CLIENT_STATS].destroy = destroy_client_stats;
    if (force_async) {
      GPR_ASSERT(wc_arg->wrapped_closure != NULL);
      GRPC_CLOSURE_SCHED(exec_ctx, wc_arg->wrapped_closure, GRPC_ERROR_NONE);
      gpr_free(wc_arg->free_when_done);
      return false;
    }
    gpr_free(wc_arg->free_when_done);
  }
  /* else, the pending pick will be registered and taken care of by the
   * pending pick list inside the RR policy (glb_policy->rr_policy).
   * Eventually, wrapped_on_complete will be called, which will -among other
   * things- add the LB token to the call's initial metadata */
  return pick_done;
}

static grpc_lb_policy_args *lb_policy_args_create(grpc_exec_ctx *exec_ctx,
                                                  glb_lb_policy *glb_policy) {
  grpc_lb_addresses *addresses =
      process_serverlist_locked(exec_ctx, glb_policy->serverlist);
  GPR_ASSERT(addresses != NULL);
  grpc_lb_policy_args *args = (grpc_lb_policy_args *)gpr_zalloc(sizeof(*args));
  args->client_channel_factory = glb_policy->cc_factory;
  args->combiner = glb_policy->base.combiner;
  // Replace the LB addresses in the channel args that we pass down to
  // the subchannel.
  static const char *keys_to_remove[] = {GRPC_ARG_LB_ADDRESSES};
  const grpc_arg arg = grpc_lb_addresses_create_channel_arg(addresses);
  args->args = grpc_channel_args_copy_and_add_and_remove(
      glb_policy->args, keys_to_remove, GPR_ARRAY_SIZE(keys_to_remove), &arg,
      1);
  grpc_lb_addresses_destroy(exec_ctx, addresses);
  return args;
}

static void lb_policy_args_destroy(grpc_exec_ctx *exec_ctx,
                                   grpc_lb_policy_args *args) {
  grpc_channel_args_destroy(exec_ctx, args->args);
  gpr_free(args);
}

static void glb_rr_connectivity_changed_locked(grpc_exec_ctx *exec_ctx,
                                               void *arg, grpc_error *error);
static void create_rr_locked(grpc_exec_ctx *exec_ctx, glb_lb_policy *glb_policy,
                             grpc_lb_policy_args *args) {
  GPR_ASSERT(glb_policy->rr_policy == NULL);

  grpc_lb_policy *new_rr_policy =
      grpc_lb_policy_create(exec_ctx, "round_robin", args);
  if (new_rr_policy == NULL) {
    gpr_log(GPR_ERROR,
            "Failure creating a RoundRobin policy for serverlist update with "
            "%lu entries. The previous RR instance (%p), if any, will continue "
            "to be used. Future updates from the LB will attempt to create new "
            "instances.",
            (unsigned long)glb_policy->serverlist->num_servers,
            (void *)glb_policy->rr_policy);
    return;
  }
  glb_policy->rr_policy = new_rr_policy;
  grpc_error *rr_state_error = NULL;
  const grpc_connectivity_state rr_state =
      grpc_lb_policy_check_connectivity_locked(exec_ctx, glb_policy->rr_policy,
                                               &rr_state_error);
  /* Connectivity state is a function of the RR policy updated/created */
  update_lb_connectivity_status_locked(exec_ctx, glb_policy, rr_state,
                                       rr_state_error);
  /* Add the gRPC LB's interested_parties pollset_set to that of the newly
   * created RR policy. This will make the RR policy progress upon activity on
   * gRPC LB, which in turn is tied to the application's call */
  grpc_pollset_set_add_pollset_set(exec_ctx,
                                   glb_policy->rr_policy->interested_parties,
                                   glb_policy->base.interested_parties);

  /* Allocate the data for the tracking of the new RR policy's connectivity.
   * It'll be deallocated in glb_rr_connectivity_changed() */
  rr_connectivity_data *rr_connectivity =
      (rr_connectivity_data *)gpr_zalloc(sizeof(rr_connectivity_data));
  GRPC_CLOSURE_INIT(&rr_connectivity->on_change,
                    glb_rr_connectivity_changed_locked, rr_connectivity,
                    grpc_combiner_scheduler(glb_policy->base.combiner));
  rr_connectivity->glb_policy = glb_policy;
  rr_connectivity->state = rr_state;

  /* Subscribe to changes to the connectivity of the new RR */
  GRPC_LB_POLICY_WEAK_REF(&glb_policy->base, "glb_rr_connectivity_cb");
  grpc_lb_policy_notify_on_state_change_locked(exec_ctx, glb_policy->rr_policy,
                                               &rr_connectivity->state,
                                               &rr_connectivity->on_change);
  grpc_lb_policy_exit_idle_locked(exec_ctx, glb_policy->rr_policy);

  /* Update picks and pings in wait */
  pending_pick *pp;
  while ((pp = glb_policy->pending_picks)) {
    glb_policy->pending_picks = pp->next;
    GRPC_LB_POLICY_REF(glb_policy->rr_policy, "rr_handover_pending_pick");
    pp->wrapped_on_complete_arg.rr_policy = glb_policy->rr_policy;
    pp->wrapped_on_complete_arg.client_stats =
        grpc_grpclb_client_stats_ref(glb_policy->client_stats);
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Pending pick about to (async) PICK from %p",
              (void *)glb_policy->rr_policy);
    }
    pick_from_internal_rr_locked(exec_ctx, glb_policy, &pp->pick_args,
                                 true /* force_async */, pp->target,
                                 &pp->wrapped_on_complete_arg);
  }

  pending_ping *pping;
  while ((pping = glb_policy->pending_pings)) {
    glb_policy->pending_pings = pping->next;
    GRPC_LB_POLICY_REF(glb_policy->rr_policy, "rr_handover_pending_ping");
    pping->wrapped_notify_arg.rr_policy = glb_policy->rr_policy;
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Pending ping about to PING from 0x%" PRIxPTR "",
              (intptr_t)glb_policy->rr_policy);
    }
    grpc_lb_policy_ping_one_locked(exec_ctx, glb_policy->rr_policy,
                                   &pping->wrapped_notify_arg.wrapper_closure);
  }
}

/* glb_policy->rr_policy may be NULL (initial handover) */
static void rr_handover_locked(grpc_exec_ctx *exec_ctx,
                               glb_lb_policy *glb_policy) {
  GPR_ASSERT(glb_policy->serverlist != NULL &&
             glb_policy->serverlist->num_servers > 0);
  if (glb_policy->shutting_down) return;
  grpc_lb_policy_args *args = lb_policy_args_create(exec_ctx, glb_policy);
  GPR_ASSERT(args != NULL);
  if (glb_policy->rr_policy != NULL) {
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_DEBUG, "Updating Round Robin policy (%p)",
              (void *)glb_policy->rr_policy);
    }
    grpc_lb_policy_update_locked(exec_ctx, glb_policy->rr_policy, args);
  } else {
    create_rr_locked(exec_ctx, glb_policy, args);
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_DEBUG, "Created new Round Robin policy (%p)",
              (void *)glb_policy->rr_policy);
    }
  }
  lb_policy_args_destroy(exec_ctx, args);
}

static void glb_rr_connectivity_changed_locked(grpc_exec_ctx *exec_ctx,
                                               void *arg, grpc_error *error) {
  rr_connectivity_data *rr_connectivity = (rr_connectivity_data *)arg;
  glb_lb_policy *glb_policy = rr_connectivity->glb_policy;
  if (glb_policy->shutting_down) {
    GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                              "glb_rr_connectivity_cb");
    gpr_free(rr_connectivity);
    return;
  }
  if (rr_connectivity->state == GRPC_CHANNEL_SHUTDOWN) {
    /* An RR policy that has transitioned into the SHUTDOWN connectivity state
     * should not be considered for picks or updates: the SHUTDOWN state is a
     * sink, policies can't transition back from it. .*/
    GRPC_LB_POLICY_UNREF(exec_ctx, glb_policy->rr_policy,
                         "rr_connectivity_shutdown");
    glb_policy->rr_policy = NULL;
    GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                              "glb_rr_connectivity_cb");
    gpr_free(rr_connectivity);
    return;
  }
  /* rr state != SHUTDOWN && !glb_policy->shutting down: biz as usual */
  update_lb_connectivity_status_locked(
      exec_ctx, glb_policy, rr_connectivity->state, GRPC_ERROR_REF(error));
  /* Resubscribe. Reuse the "glb_rr_connectivity_cb" weak ref. */
  grpc_lb_policy_notify_on_state_change_locked(exec_ctx, glb_policy->rr_policy,
                                               &rr_connectivity->state,
                                               &rr_connectivity->on_change);
}

static void destroy_balancer_name(grpc_exec_ctx *exec_ctx,
                                  void *balancer_name) {
  gpr_free(balancer_name);
}

static grpc_slice_hash_table_entry targets_info_entry_create(
    const char *address, const char *balancer_name) {
  grpc_slice_hash_table_entry entry;
  entry.key = grpc_slice_from_copied_string(address);
  entry.value = gpr_strdup(balancer_name);
  return entry;
}

static int balancer_name_cmp_fn(void *a, void *b) {
  const char *a_str = (const char *)a;
  const char *b_str = (const char *)b;
  return strcmp(a_str, b_str);
}

/* Returns the channel args for the LB channel, used to create a bidirectional
 * stream for the reception of load balancing updates.
 *
 * Inputs:
 *   - \a addresses: corresponding to the balancers.
 *   - \a response_generator: in order to propagate updates from the resolver
 *   above the grpclb policy.
 *   - \a args: other args inherited from the grpclb policy. */
static grpc_channel_args *build_lb_channel_args(
    grpc_exec_ctx *exec_ctx, const grpc_lb_addresses *addresses,
    grpc_fake_resolver_response_generator *response_generator,
    const grpc_channel_args *args) {
  size_t num_grpclb_addrs = 0;
  for (size_t i = 0; i < addresses->num_addresses; ++i) {
    if (addresses->addresses[i].is_balancer) ++num_grpclb_addrs;
  }
  /* All input addresses come from a resolver that claims they are LB services.
   * It's the resolver's responsibility to make sure this policy is only
   * instantiated and used in that case. Otherwise, something has gone wrong. */
  GPR_ASSERT(num_grpclb_addrs > 0);
  grpc_lb_addresses *lb_addresses =
      grpc_lb_addresses_create(num_grpclb_addrs, NULL);
  grpc_slice_hash_table_entry *targets_info_entries =
      (grpc_slice_hash_table_entry *)gpr_zalloc(sizeof(*targets_info_entries) *
                                                num_grpclb_addrs);

  size_t lb_addresses_idx = 0;
  for (size_t i = 0; i < addresses->num_addresses; ++i) {
    if (!addresses->addresses[i].is_balancer) continue;
    if (addresses->addresses[i].user_data != NULL) {
      gpr_log(GPR_ERROR,
              "This LB policy doesn't support user data. It will be ignored");
    }
    char *addr_str;
    GPR_ASSERT(grpc_sockaddr_to_string(
                   &addr_str, &addresses->addresses[i].address, true) > 0);
    targets_info_entries[lb_addresses_idx] = targets_info_entry_create(
        addr_str, addresses->addresses[i].balancer_name);
    gpr_free(addr_str);

    grpc_lb_addresses_set_address(
        lb_addresses, lb_addresses_idx++, addresses->addresses[i].address.addr,
        addresses->addresses[i].address.len, false /* is balancer */,
        addresses->addresses[i].balancer_name, NULL /* user data */);
  }
  GPR_ASSERT(num_grpclb_addrs == lb_addresses_idx);
  grpc_slice_hash_table *targets_info =
      grpc_slice_hash_table_create(num_grpclb_addrs, targets_info_entries,
                                   destroy_balancer_name, balancer_name_cmp_fn);
  gpr_free(targets_info_entries);

  grpc_channel_args *lb_channel_args =
      grpc_lb_policy_grpclb_build_lb_channel_args(exec_ctx, targets_info,
                                                  response_generator, args);

  grpc_arg lb_channel_addresses_arg =
      grpc_lb_addresses_create_channel_arg(lb_addresses);

  grpc_channel_args *result = grpc_channel_args_copy_and_add(
      lb_channel_args, &lb_channel_addresses_arg, 1);
  grpc_slice_hash_table_unref(exec_ctx, targets_info);
  grpc_channel_args_destroy(exec_ctx, lb_channel_args);
  grpc_lb_addresses_destroy(exec_ctx, lb_addresses);
  return result;
}

static void glb_destroy(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  GPR_ASSERT(glb_policy->pending_picks == NULL);
  GPR_ASSERT(glb_policy->pending_pings == NULL);
  gpr_free((void *)glb_policy->server_name);
  grpc_channel_args_destroy(exec_ctx, glb_policy->args);
  if (glb_policy->client_stats != NULL) {
    grpc_grpclb_client_stats_unref(glb_policy->client_stats);
  }
  grpc_connectivity_state_destroy(exec_ctx, &glb_policy->state_tracker);
  if (glb_policy->serverlist != NULL) {
    grpc_grpclb_destroy_serverlist(glb_policy->serverlist);
  }
  grpc_fake_resolver_response_generator_unref(glb_policy->response_generator);
  grpc_subchannel_index_unref();
  if (glb_policy->pending_update_args != NULL) {
    grpc_channel_args_destroy(exec_ctx, glb_policy->pending_update_args->args);
    gpr_free(glb_policy->pending_update_args);
  }
  gpr_free(glb_policy);
}

static void glb_shutdown_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  glb_policy->shutting_down = true;

  /* We need a copy of the lb_call pointer because we can't cancell the call
   * while holding glb_policy->mu: lb_on_server_status_received, invoked due to
   * the cancel, needs to acquire that same lock */
  grpc_call *lb_call = glb_policy->lb_call;

  /* glb_policy->lb_call and this local lb_call must be consistent at this point
   * because glb_policy->lb_call is only assigned in lb_call_init_locked as part
   * of query_for_backends_locked, which can only be invoked while
   * glb_policy->shutting_down is false. */
  if (lb_call != NULL) {
    grpc_call_cancel(lb_call, NULL);
    /* lb_on_server_status_received will pick up the cancel and clean up */
  }
  if (glb_policy->retry_timer_active) {
    grpc_timer_cancel(exec_ctx, &glb_policy->lb_call_retry_timer);
    glb_policy->retry_timer_active = false;
  }

  pending_pick *pp = glb_policy->pending_picks;
  glb_policy->pending_picks = NULL;
  pending_ping *pping = glb_policy->pending_pings;
  glb_policy->pending_pings = NULL;
  if (glb_policy->rr_policy != NULL) {
    GRPC_LB_POLICY_UNREF(exec_ctx, glb_policy->rr_policy, "glb_shutdown");
  }
  // We destroy the LB channel here because
  // glb_lb_channel_on_connectivity_changed_cb needs a valid glb_policy
  // instance.  Destroying the lb channel in glb_destroy would likely result in
  // a callback invocation without a valid glb_policy arg.
  if (glb_policy->lb_channel != NULL) {
    grpc_channel_destroy(glb_policy->lb_channel);
    glb_policy->lb_channel = NULL;
  }
  grpc_connectivity_state_set(
      exec_ctx, &glb_policy->state_tracker, GRPC_CHANNEL_SHUTDOWN,
      GRPC_ERROR_CREATE_FROM_STATIC_STRING("Channel Shutdown"), "glb_shutdown");

  while (pp != NULL) {
    pending_pick *next = pp->next;
    *pp->target = NULL;
    GRPC_CLOSURE_SCHED(exec_ctx, &pp->wrapped_on_complete_arg.wrapper_closure,
                       GRPC_ERROR_NONE);
    pp = next;
  }

  while (pping != NULL) {
    pending_ping *next = pping->next;
    GRPC_CLOSURE_SCHED(exec_ctx, &pping->wrapped_notify_arg.wrapper_closure,
                       GRPC_ERROR_NONE);
    pping = next;
  }
}

// Cancel a specific pending pick.
//
// A grpclb pick progresses as follows:
// - If there's a Round Robin policy (glb_policy->rr_policy) available, it'll be
//   handed over to the RR policy (in create_rr_locked()). From that point
//   onwards, it'll be RR's responsibility. For cancellations, that implies the
//   pick needs also be cancelled by the RR instance.
// - Otherwise, without an RR instance, picks stay pending at this policy's
//   level (grpclb), inside the glb_policy->pending_picks list. To cancel these,
//   we invoke the completion closure and set *target to NULL right here.
static void glb_cancel_pick_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol,
                                   grpc_connected_subchannel **target,
                                   grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  pending_pick *pp = glb_policy->pending_picks;
  glb_policy->pending_picks = NULL;
  while (pp != NULL) {
    pending_pick *next = pp->next;
    if (pp->target == target) {
      *target = NULL;
      GRPC_CLOSURE_SCHED(exec_ctx, &pp->wrapped_on_complete_arg.wrapper_closure,
                         GRPC_ERROR_CREATE_REFERENCING_FROM_STATIC_STRING(
                             "Pick Cancelled", &error, 1));
    } else {
      pp->next = glb_policy->pending_picks;
      glb_policy->pending_picks = pp;
    }
    pp = next;
  }
  if (glb_policy->rr_policy != NULL) {
    grpc_lb_policy_cancel_pick_locked(exec_ctx, glb_policy->rr_policy, target,
                                      GRPC_ERROR_REF(error));
  }
  GRPC_ERROR_UNREF(error);
}

// Cancel all pending picks.
//
// A grpclb pick progresses as follows:
// - If there's a Round Robin policy (glb_policy->rr_policy) available, it'll be
//   handed over to the RR policy (in create_rr_locked()). From that point
//   onwards, it'll be RR's responsibility. For cancellations, that implies the
//   pick needs also be cancelled by the RR instance.
// - Otherwise, without an RR instance, picks stay pending at this policy's
//   level (grpclb), inside the glb_policy->pending_picks list. To cancel these,
//   we invoke the completion closure and set *target to NULL right here.
static void glb_cancel_picks_locked(grpc_exec_ctx *exec_ctx,
                                    grpc_lb_policy *pol,
                                    uint32_t initial_metadata_flags_mask,
                                    uint32_t initial_metadata_flags_eq,
                                    grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  pending_pick *pp = glb_policy->pending_picks;
  glb_policy->pending_picks = NULL;
  while (pp != NULL) {
    pending_pick *next = pp->next;
    if ((pp->pick_args.initial_metadata_flags & initial_metadata_flags_mask) ==
        initial_metadata_flags_eq) {
      GRPC_CLOSURE_SCHED(exec_ctx, &pp->wrapped_on_complete_arg.wrapper_closure,
                         GRPC_ERROR_CREATE_REFERENCING_FROM_STATIC_STRING(
                             "Pick Cancelled", &error, 1));
    } else {
      pp->next = glb_policy->pending_picks;
      glb_policy->pending_picks = pp;
    }
    pp = next;
  }
  if (glb_policy->rr_policy != NULL) {
    grpc_lb_policy_cancel_picks_locked(
        exec_ctx, glb_policy->rr_policy, initial_metadata_flags_mask,
        initial_metadata_flags_eq, GRPC_ERROR_REF(error));
  }
  GRPC_ERROR_UNREF(error);
}

static void query_for_backends_locked(grpc_exec_ctx *exec_ctx,
                                      glb_lb_policy *glb_policy);
static void start_picking_locked(grpc_exec_ctx *exec_ctx,
                                 glb_lb_policy *glb_policy) {
  glb_policy->started_picking = true;
  gpr_backoff_reset(&glb_policy->lb_call_backoff_state);
  query_for_backends_locked(exec_ctx, glb_policy);
}

static void glb_exit_idle_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  if (!glb_policy->started_picking) {
    start_picking_locked(exec_ctx, glb_policy);
  }
}

static int glb_pick_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol,
                           const grpc_lb_policy_pick_args *pick_args,
                           grpc_connected_subchannel **target,
                           grpc_call_context_element *context, void **user_data,
                           grpc_closure *on_complete) {
  if (pick_args->lb_token_mdelem_storage == NULL) {
    *target = NULL;
    GRPC_CLOSURE_SCHED(exec_ctx, on_complete,
                       GRPC_ERROR_CREATE_FROM_STATIC_STRING(
                           "No mdelem storage for the LB token. Load reporting "
                           "won't work without it. Failing"));
    return 0;
  }

  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  bool pick_done;

  if (glb_policy->rr_policy != NULL) {
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "grpclb %p about to PICK from RR %p",
              (void *)glb_policy, (void *)glb_policy->rr_policy);
    }
    GRPC_LB_POLICY_REF(glb_policy->rr_policy, "glb_pick");

    wrapped_rr_closure_arg *wc_arg =
        (wrapped_rr_closure_arg *)gpr_zalloc(sizeof(wrapped_rr_closure_arg));

    GRPC_CLOSURE_INIT(&wc_arg->wrapper_closure, wrapped_rr_closure, wc_arg,
                      grpc_schedule_on_exec_ctx);
    wc_arg->rr_policy = glb_policy->rr_policy;
    wc_arg->target = target;
    wc_arg->context = context;
    GPR_ASSERT(glb_policy->client_stats != NULL);
    wc_arg->client_stats =
        grpc_grpclb_client_stats_ref(glb_policy->client_stats);
    wc_arg->wrapped_closure = on_complete;
    wc_arg->lb_token_mdelem_storage = pick_args->lb_token_mdelem_storage;
    wc_arg->initial_metadata = pick_args->initial_metadata;
    wc_arg->free_when_done = wc_arg;
    pick_done =
        pick_from_internal_rr_locked(exec_ctx, glb_policy, pick_args,
                                     false /* force_async */, target, wc_arg);
  } else {
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_DEBUG,
              "No RR policy in grpclb instance %p. Adding to grpclb's pending "
              "picks",
              (void *)(glb_policy));
    }
    add_pending_pick(&glb_policy->pending_picks, pick_args, target, context,
                     on_complete);

    if (!glb_policy->started_picking) {
      start_picking_locked(exec_ctx, glb_policy);
    }
    pick_done = false;
  }
  return pick_done;
}

static grpc_connectivity_state glb_check_connectivity_locked(
    grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol,
    grpc_error **connectivity_error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  return grpc_connectivity_state_get(&glb_policy->state_tracker,
                                     connectivity_error);
}

static void glb_ping_one_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *pol,
                                grpc_closure *closure) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  if (glb_policy->rr_policy) {
    grpc_lb_policy_ping_one_locked(exec_ctx, glb_policy->rr_policy, closure);
  } else {
    add_pending_ping(&glb_policy->pending_pings, closure);
    if (!glb_policy->started_picking) {
      start_picking_locked(exec_ctx, glb_policy);
    }
  }
}

static void glb_notify_on_state_change_locked(grpc_exec_ctx *exec_ctx,
                                              grpc_lb_policy *pol,
                                              grpc_connectivity_state *current,
                                              grpc_closure *notify) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)pol;
  grpc_connectivity_state_notify_on_state_change(
      exec_ctx, &glb_policy->state_tracker, current, notify);
}

static void send_client_load_report_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                           grpc_error *error);

static void schedule_next_client_load_report(grpc_exec_ctx *exec_ctx,
                                             glb_lb_policy *glb_policy) {
  const gpr_timespec now = gpr_now(GPR_CLOCK_MONOTONIC);
  const gpr_timespec next_client_load_report_time =
      gpr_time_add(now, glb_policy->client_stats_report_interval);
  GRPC_CLOSURE_INIT(&glb_policy->client_load_report_closure,
                    send_client_load_report_locked, glb_policy,
                    grpc_combiner_scheduler(glb_policy->base.combiner));
  grpc_timer_init(exec_ctx, &glb_policy->client_load_report_timer,
                  next_client_load_report_time,
                  &glb_policy->client_load_report_closure, now);
}

static void client_load_report_done_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                           grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  grpc_byte_buffer_destroy(glb_policy->client_load_report_payload);
  glb_policy->client_load_report_payload = NULL;
  if (error != GRPC_ERROR_NONE || glb_policy->lb_call == NULL) {
    glb_policy->client_load_report_timer_pending = false;
    GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                              "client_load_report");
    return;
  }
  schedule_next_client_load_report(exec_ctx, glb_policy);
}

static void do_send_client_load_report_locked(grpc_exec_ctx *exec_ctx,
                                              glb_lb_policy *glb_policy) {
  grpc_op op;
  memset(&op, 0, sizeof(op));
  op.op = GRPC_OP_SEND_MESSAGE;
  op.data.send_message.send_message = glb_policy->client_load_report_payload;
  GRPC_CLOSURE_INIT(&glb_policy->client_load_report_closure,
                    client_load_report_done_locked, glb_policy,
                    grpc_combiner_scheduler(glb_policy->base.combiner));
  grpc_call_error call_error = grpc_call_start_batch_and_execute(
      exec_ctx, glb_policy->lb_call, &op, 1,
      &glb_policy->client_load_report_closure);
  GPR_ASSERT(GRPC_CALL_OK == call_error);
}

static bool load_report_counters_are_zero(grpc_grpclb_request *request) {
  grpc_grpclb_dropped_call_counts *drop_entries =
      (grpc_grpclb_dropped_call_counts *)
          request->client_stats.calls_finished_with_drop.arg;
  return request->client_stats.num_calls_started == 0 &&
         request->client_stats.num_calls_finished == 0 &&
         request->client_stats.num_calls_finished_with_client_failed_to_send ==
             0 &&
         request->client_stats.num_calls_finished_known_received == 0 &&
         (drop_entries == NULL || drop_entries->num_entries == 0);
}

static void send_client_load_report_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                           grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  if (error == GRPC_ERROR_CANCELLED || glb_policy->lb_call == NULL) {
    glb_policy->client_load_report_timer_pending = false;
    GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                              "client_load_report");
    return;
  }
  // Construct message payload.
  GPR_ASSERT(glb_policy->client_load_report_payload == NULL);
  grpc_grpclb_request *request =
      grpc_grpclb_load_report_request_create_locked(glb_policy->client_stats);
  // Skip client load report if the counters were all zero in the last
  // report and they are still zero in this one.
  if (load_report_counters_are_zero(request)) {
    if (glb_policy->last_client_load_report_counters_were_zero) {
      grpc_grpclb_request_destroy(request);
      schedule_next_client_load_report(exec_ctx, glb_policy);
      return;
    }
    glb_policy->last_client_load_report_counters_were_zero = true;
  } else {
    glb_policy->last_client_load_report_counters_were_zero = false;
  }
  grpc_slice request_payload_slice = grpc_grpclb_request_encode(request);
  glb_policy->client_load_report_payload =
      grpc_raw_byte_buffer_create(&request_payload_slice, 1);
  grpc_slice_unref_internal(exec_ctx, request_payload_slice);
  grpc_grpclb_request_destroy(request);
  // If we've already sent the initial request, then we can go ahead and
  // sent the load report.  Otherwise, we need to wait until the initial
  // request has been sent to send this
  // (see lb_on_sent_initial_request_locked() below).
  if (glb_policy->initial_request_sent) {
    do_send_client_load_report_locked(exec_ctx, glb_policy);
  }
}

static void lb_on_sent_initial_request_locked(grpc_exec_ctx *exec_ctx,
                                              void *arg, grpc_error *error);
static void lb_on_server_status_received_locked(grpc_exec_ctx *exec_ctx,
                                                void *arg, grpc_error *error);
static void lb_on_response_received_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                           grpc_error *error);
static void lb_call_init_locked(grpc_exec_ctx *exec_ctx,
                                glb_lb_policy *glb_policy) {
  GPR_ASSERT(glb_policy->server_name != NULL);
  GPR_ASSERT(glb_policy->server_name[0] != '\0');
  GPR_ASSERT(glb_policy->lb_call == NULL);
  GPR_ASSERT(!glb_policy->shutting_down);

  /* Note the following LB call progresses every time there's activity in \a
   * glb_policy->base.interested_parties, which is comprised of the polling
   * entities from \a client_channel. */
  grpc_slice host = grpc_slice_from_copied_string(glb_policy->server_name);
  gpr_timespec deadline =
      glb_policy->lb_call_timeout_ms == 0
          ? gpr_inf_future(GPR_CLOCK_MONOTONIC)
          : gpr_time_add(gpr_now(GPR_CLOCK_MONOTONIC),
                         gpr_time_from_millis(glb_policy->lb_call_timeout_ms,
                                              GPR_TIMESPAN));
  glb_policy->lb_call = grpc_channel_create_pollset_set_call(
      exec_ctx, glb_policy->lb_channel, NULL, GRPC_PROPAGATE_DEFAULTS,
      glb_policy->base.interested_parties,
      GRPC_MDSTR_SLASH_GRPC_DOT_LB_DOT_V1_DOT_LOADBALANCER_SLASH_BALANCELOAD,
      &host, deadline, NULL);
  grpc_slice_unref_internal(exec_ctx, host);

  if (glb_policy->client_stats != NULL) {
    grpc_grpclb_client_stats_unref(glb_policy->client_stats);
  }
  glb_policy->client_stats = grpc_grpclb_client_stats_create();

  grpc_metadata_array_init(&glb_policy->lb_initial_metadata_recv);
  grpc_metadata_array_init(&glb_policy->lb_trailing_metadata_recv);

  grpc_grpclb_request *request =
      grpc_grpclb_request_create(glb_policy->server_name);
  grpc_slice request_payload_slice = grpc_grpclb_request_encode(request);
  glb_policy->lb_request_payload =
      grpc_raw_byte_buffer_create(&request_payload_slice, 1);
  grpc_slice_unref_internal(exec_ctx, request_payload_slice);
  grpc_grpclb_request_destroy(request);

  GRPC_CLOSURE_INIT(&glb_policy->lb_on_sent_initial_request,
                    lb_on_sent_initial_request_locked, glb_policy,
                    grpc_combiner_scheduler(glb_policy->base.combiner));
  GRPC_CLOSURE_INIT(&glb_policy->lb_on_server_status_received,
                    lb_on_server_status_received_locked, glb_policy,
                    grpc_combiner_scheduler(glb_policy->base.combiner));
  GRPC_CLOSURE_INIT(&glb_policy->lb_on_response_received,
                    lb_on_response_received_locked, glb_policy,
                    grpc_combiner_scheduler(glb_policy->base.combiner));

  gpr_backoff_init(&glb_policy->lb_call_backoff_state,
                   GRPC_GRPCLB_INITIAL_CONNECT_BACKOFF_SECONDS,
                   GRPC_GRPCLB_RECONNECT_BACKOFF_MULTIPLIER,
                   GRPC_GRPCLB_RECONNECT_JITTER,
                   GRPC_GRPCLB_MIN_CONNECT_TIMEOUT_SECONDS * 1000,
                   GRPC_GRPCLB_RECONNECT_MAX_BACKOFF_SECONDS * 1000);

  glb_policy->initial_request_sent = false;
  glb_policy->seen_initial_response = false;
  glb_policy->last_client_load_report_counters_were_zero = false;
}

static void lb_call_destroy_locked(grpc_exec_ctx *exec_ctx,
                                   glb_lb_policy *glb_policy) {
  GPR_ASSERT(glb_policy->lb_call != NULL);
  grpc_call_unref(glb_policy->lb_call);
  glb_policy->lb_call = NULL;

  grpc_metadata_array_destroy(&glb_policy->lb_initial_metadata_recv);
  grpc_metadata_array_destroy(&glb_policy->lb_trailing_metadata_recv);

  grpc_byte_buffer_destroy(glb_policy->lb_request_payload);
  grpc_slice_unref_internal(exec_ctx, glb_policy->lb_call_status_details);

  if (!glb_policy->client_load_report_timer_pending) {
    grpc_timer_cancel(exec_ctx, &glb_policy->client_load_report_timer);
  }
}

/*
 * Auxiliary functions and LB client callbacks.
 */
static void query_for_backends_locked(grpc_exec_ctx *exec_ctx,
                                      glb_lb_policy *glb_policy) {
  GPR_ASSERT(glb_policy->lb_channel != NULL);
  if (glb_policy->shutting_down) return;

  lb_call_init_locked(exec_ctx, glb_policy);

  if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
    gpr_log(GPR_INFO,
            "Query for backends (grpclb: %p, lb_channel: %p, lb_call: %p)",
            (void *)glb_policy, (void *)glb_policy->lb_channel,
            (void *)glb_policy->lb_call);
  }
  GPR_ASSERT(glb_policy->lb_call != NULL);

  grpc_call_error call_error;
  grpc_op ops[4];
  memset(ops, 0, sizeof(ops));

  grpc_op *op = ops;
  op->op = GRPC_OP_SEND_INITIAL_METADATA;
  op->data.send_initial_metadata.count = 0;
  op->flags = 0;
  op->reserved = NULL;
  op++;
  op->op = GRPC_OP_RECV_INITIAL_METADATA;
  op->data.recv_initial_metadata.recv_initial_metadata =
      &glb_policy->lb_initial_metadata_recv;
  op->flags = 0;
  op->reserved = NULL;
  op++;
  GPR_ASSERT(glb_policy->lb_request_payload != NULL);
  op->op = GRPC_OP_SEND_MESSAGE;
  op->data.send_message.send_message = glb_policy->lb_request_payload;
  op->flags = 0;
  op->reserved = NULL;
  op++;
  /* take a weak ref (won't prevent calling of \a glb_shutdown if the strong ref
   * count goes to zero) to be unref'd in lb_on_sent_initial_request_locked() */
  GRPC_LB_POLICY_WEAK_REF(&glb_policy->base,
                          "lb_on_sent_initial_request_locked");
  call_error = grpc_call_start_batch_and_execute(
      exec_ctx, glb_policy->lb_call, ops, (size_t)(op - ops),
      &glb_policy->lb_on_sent_initial_request);
  GPR_ASSERT(GRPC_CALL_OK == call_error);

  op = ops;
  op->op = GRPC_OP_RECV_STATUS_ON_CLIENT;
  op->data.recv_status_on_client.trailing_metadata =
      &glb_policy->lb_trailing_metadata_recv;
  op->data.recv_status_on_client.status = &glb_policy->lb_call_status;
  op->data.recv_status_on_client.status_details =
      &glb_policy->lb_call_status_details;
  op->flags = 0;
  op->reserved = NULL;
  op++;
  /* take a weak ref (won't prevent calling of \a glb_shutdown if the strong ref
   * count goes to zero) to be unref'd in lb_on_server_status_received_locked */
  GRPC_LB_POLICY_WEAK_REF(&glb_policy->base,
                          "lb_on_server_status_received_locked");
  call_error = grpc_call_start_batch_and_execute(
      exec_ctx, glb_policy->lb_call, ops, (size_t)(op - ops),
      &glb_policy->lb_on_server_status_received);
  GPR_ASSERT(GRPC_CALL_OK == call_error);

  op = ops;
  op->op = GRPC_OP_RECV_MESSAGE;
  op->data.recv_message.recv_message = &glb_policy->lb_response_payload;
  op->flags = 0;
  op->reserved = NULL;
  op++;
  /* take another weak ref to be unref'd/reused in
   * lb_on_response_received_locked */
  GRPC_LB_POLICY_WEAK_REF(&glb_policy->base, "lb_on_response_received_locked");
  call_error = grpc_call_start_batch_and_execute(
      exec_ctx, glb_policy->lb_call, ops, (size_t)(op - ops),
      &glb_policy->lb_on_response_received);
  GPR_ASSERT(GRPC_CALL_OK == call_error);
}

static void lb_on_sent_initial_request_locked(grpc_exec_ctx *exec_ctx,
                                              void *arg, grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  glb_policy->initial_request_sent = true;
  // If we attempted to send a client load report before the initial
  // request was sent, send the load report now.
  if (glb_policy->client_load_report_payload != NULL) {
    do_send_client_load_report_locked(exec_ctx, glb_policy);
  }
  GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                            "lb_on_sent_initial_request_locked");
}

static void lb_on_response_received_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                           grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  grpc_op ops[2];
  memset(ops, 0, sizeof(ops));
  grpc_op *op = ops;
  if (glb_policy->lb_response_payload != NULL) {
    gpr_backoff_reset(&glb_policy->lb_call_backoff_state);
    /* Received data from the LB server. Look inside
     * glb_policy->lb_response_payload, for a serverlist. */
    grpc_byte_buffer_reader bbr;
    grpc_byte_buffer_reader_init(&bbr, glb_policy->lb_response_payload);
    grpc_slice response_slice = grpc_byte_buffer_reader_readall(&bbr);
    grpc_byte_buffer_reader_destroy(&bbr);
    grpc_byte_buffer_destroy(glb_policy->lb_response_payload);

    grpc_grpclb_initial_response *response = NULL;
    if (!glb_policy->seen_initial_response &&
        (response = grpc_grpclb_initial_response_parse(response_slice)) !=
            NULL) {
      if (response->has_client_stats_report_interval) {
        glb_policy->client_stats_report_interval =
            gpr_time_max(gpr_time_from_seconds(1, GPR_TIMESPAN),
                         grpc_grpclb_duration_to_timespec(
                             &response->client_stats_report_interval));
        if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
          gpr_log(GPR_INFO,
                  "received initial LB response message; "
                  "client load reporting interval = %" PRId64 ".%09d sec",
                  glb_policy->client_stats_report_interval.tv_sec,
                  glb_policy->client_stats_report_interval.tv_nsec);
        }
        /* take a weak ref (won't prevent calling of \a glb_shutdown() if the
         * strong ref count goes to zero) to be unref'd in
         * send_client_load_report_locked() */
        glb_policy->client_load_report_timer_pending = true;
        GRPC_LB_POLICY_WEAK_REF(&glb_policy->base, "client_load_report");
        schedule_next_client_load_report(exec_ctx, glb_policy);
      } else if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
        gpr_log(GPR_INFO,
                "received initial LB response message; "
                "client load reporting NOT enabled");
      }
      grpc_grpclb_initial_response_destroy(response);
      glb_policy->seen_initial_response = true;
    } else {
      grpc_grpclb_serverlist *serverlist =
          grpc_grpclb_response_parse_serverlist(response_slice);
      if (serverlist != NULL) {
        GPR_ASSERT(glb_policy->lb_call != NULL);
        if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
          gpr_log(GPR_INFO, "Serverlist with %lu servers received",
                  (unsigned long)serverlist->num_servers);
          for (size_t i = 0; i < serverlist->num_servers; ++i) {
            grpc_resolved_address addr;
            parse_server(serverlist->servers[i], &addr);
            char *ipport;
            grpc_sockaddr_to_string(&ipport, &addr, false);
            gpr_log(GPR_INFO, "Serverlist[%lu]: %s", (unsigned long)i, ipport);
            gpr_free(ipport);
          }
        }
        /* update serverlist */
        if (serverlist->num_servers > 0) {
          if (grpc_grpclb_serverlist_equals(glb_policy->serverlist,
                                            serverlist)) {
            if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
              gpr_log(GPR_INFO,
                      "Incoming server list identical to current, ignoring.");
            }
            grpc_grpclb_destroy_serverlist(serverlist);
          } else { /* new serverlist */
            if (glb_policy->serverlist != NULL) {
              /* dispose of the old serverlist */
              grpc_grpclb_destroy_serverlist(glb_policy->serverlist);
            }
            /* and update the copy in the glb_lb_policy instance. This
             * serverlist instance will be destroyed either upon the next
             * update or in glb_destroy() */
            glb_policy->serverlist = serverlist;
            glb_policy->serverlist_index = 0;
            rr_handover_locked(exec_ctx, glb_policy);
          }
        } else {
          if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
            gpr_log(GPR_INFO,
                    "Received empty server list. Picks will stay pending until "
                    "a response with > 0 servers is received");
          }
          grpc_grpclb_destroy_serverlist(serverlist);
        }
      } else { /* serverlist == NULL */
        gpr_log(GPR_ERROR, "Invalid LB response received: '%s'. Ignoring.",
                grpc_dump_slice(response_slice, GPR_DUMP_ASCII | GPR_DUMP_HEX));
      }
    }
    grpc_slice_unref_internal(exec_ctx, response_slice);
    if (!glb_policy->shutting_down) {
      /* keep listening for serverlist updates */
      op->op = GRPC_OP_RECV_MESSAGE;
      op->data.recv_message.recv_message = &glb_policy->lb_response_payload;
      op->flags = 0;
      op->reserved = NULL;
      op++;
      /* reuse the "lb_on_response_received_locked" weak ref taken in
       * query_for_backends_locked() */
      const grpc_call_error call_error = grpc_call_start_batch_and_execute(
          exec_ctx, glb_policy->lb_call, ops, (size_t)(op - ops),
          &glb_policy->lb_on_response_received); /* loop */
      GPR_ASSERT(GRPC_CALL_OK == call_error);
    }
  } else { /* empty payload: call cancelled. */
           /* dispose of the "lb_on_response_received_locked" weak ref taken in
            * query_for_backends_locked() and reused in every reception loop */
    GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                              "lb_on_response_received_locked_empty_payload");
  }
}

static void lb_call_on_retry_timer_locked(grpc_exec_ctx *exec_ctx, void *arg,
                                          grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  glb_policy->retry_timer_active = false;
  if (!glb_policy->shutting_down && error == GRPC_ERROR_NONE) {
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO, "Restaring call to LB server (grpclb %p)",
              (void *)glb_policy);
    }
    GPR_ASSERT(glb_policy->lb_call == NULL);
    query_for_backends_locked(exec_ctx, glb_policy);
  }
  GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base, "grpclb_retry_timer");
}

static void lb_on_server_status_received_locked(grpc_exec_ctx *exec_ctx,
                                                void *arg, grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  GPR_ASSERT(glb_policy->lb_call != NULL);
  if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
    char *status_details =
        grpc_slice_to_c_string(glb_policy->lb_call_status_details);
    gpr_log(GPR_INFO,
            "Status from LB server received. Status = %d, Details = '%s', "
            "(call: %p), error %p",
            glb_policy->lb_call_status, status_details,
            (void *)glb_policy->lb_call, (void *)error);
    gpr_free(status_details);
  }
  /* We need to perform cleanups no matter what. */
  lb_call_destroy_locked(exec_ctx, glb_policy);
  if (glb_policy->started_picking && glb_policy->updating_lb_call) {
    if (glb_policy->retry_timer_active) {
      grpc_timer_cancel(exec_ctx, &glb_policy->lb_call_retry_timer);
    }
    if (!glb_policy->shutting_down) start_picking_locked(exec_ctx, glb_policy);
    glb_policy->updating_lb_call = false;
  } else if (!glb_policy->shutting_down) {
    /* if we aren't shutting down, restart the LB client call after some time */
    gpr_timespec now = gpr_now(GPR_CLOCK_MONOTONIC);
    gpr_timespec next_try =
        gpr_backoff_step(&glb_policy->lb_call_backoff_state, now);
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_DEBUG, "Connection to LB server lost (grpclb: %p)...",
              (void *)glb_policy);
      gpr_timespec timeout = gpr_time_sub(next_try, now);
      if (gpr_time_cmp(timeout, gpr_time_0(timeout.clock_type)) > 0) {
        gpr_log(GPR_DEBUG,
                "... retry_timer_active in %" PRId64 ".%09d seconds.",
                timeout.tv_sec, timeout.tv_nsec);
      } else {
        gpr_log(GPR_DEBUG, "... retry_timer_active immediately.");
      }
    }
    GRPC_LB_POLICY_WEAK_REF(&glb_policy->base, "grpclb_retry_timer");
    GRPC_CLOSURE_INIT(&glb_policy->lb_on_call_retry,
                      lb_call_on_retry_timer_locked, glb_policy,
                      grpc_combiner_scheduler(glb_policy->base.combiner));
    glb_policy->retry_timer_active = true;
    grpc_timer_init(exec_ctx, &glb_policy->lb_call_retry_timer, next_try,
                    &glb_policy->lb_on_call_retry, now);
  }
  GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                            "lb_on_server_status_received_locked");
}

static void glb_update_locked(grpc_exec_ctx *exec_ctx, grpc_lb_policy *policy,
                              const grpc_lb_policy_args *args) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)policy;
  if (glb_policy->updating_lb_channel) {
    if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
      gpr_log(GPR_INFO,
              "Update already in progress for grpclb %p. Deferring update.",
              (void *)glb_policy);
    }
    if (glb_policy->pending_update_args != NULL) {
      grpc_channel_args_destroy(exec_ctx,
                                glb_policy->pending_update_args->args);
      gpr_free(glb_policy->pending_update_args);
    }
    glb_policy->pending_update_args = (grpc_lb_policy_args *)gpr_zalloc(
        sizeof(*glb_policy->pending_update_args));
    glb_policy->pending_update_args->client_channel_factory =
        args->client_channel_factory;
    glb_policy->pending_update_args->args = grpc_channel_args_copy(args->args);
    glb_policy->pending_update_args->combiner = args->combiner;
    return;
  }

  glb_policy->updating_lb_channel = true;
  // Propagate update to lb_channel (pick first).
  const grpc_arg *arg =
      grpc_channel_args_find(args->args, GRPC_ARG_LB_ADDRESSES);
  if (arg == NULL || arg->type != GRPC_ARG_POINTER) {
    if (glb_policy->lb_channel == NULL) {
      // If we don't have a current channel to the LB, go into TRANSIENT
      // FAILURE.
      grpc_connectivity_state_set(
          exec_ctx, &glb_policy->state_tracker, GRPC_CHANNEL_TRANSIENT_FAILURE,
          GRPC_ERROR_CREATE_FROM_STATIC_STRING("Missing update in args"),
          "glb_update_missing");
    } else {
      // otherwise, keep using the current LB channel (ignore this update).
      gpr_log(GPR_ERROR,
              "No valid LB addresses channel arg for grpclb %p update, "
              "ignoring.",
              (void *)glb_policy);
    }
  }
  const grpc_lb_addresses *addresses =
      (const grpc_lb_addresses *)arg->value.pointer.p;
  GPR_ASSERT(glb_policy->lb_channel != NULL);
  grpc_channel_args *lb_channel_args = build_lb_channel_args(
      exec_ctx, addresses, glb_policy->response_generator, args->args);
  /* Propagate updates to the LB channel through the fake resolver */
  grpc_fake_resolver_response_generator_set_response(
      exec_ctx, glb_policy->response_generator, lb_channel_args);
  grpc_channel_args_destroy(exec_ctx, lb_channel_args);

  if (!glb_policy->watching_lb_channel) {
    // Watch the LB channel connectivity for connection.
    glb_policy->lb_channel_connectivity = grpc_channel_check_connectivity_state(
        glb_policy->lb_channel, true /* try to connect */);
    grpc_channel_element *client_channel_elem = grpc_channel_stack_last_element(
        grpc_channel_get_channel_stack(glb_policy->lb_channel));
    GPR_ASSERT(client_channel_elem->filter == &grpc_client_channel_filter);
    glb_policy->watching_lb_channel = true;
    GRPC_LB_POLICY_WEAK_REF(&glb_policy->base, "watch_lb_channel_connectivity");
    grpc_client_channel_watch_connectivity_state(
        exec_ctx, client_channel_elem,
        grpc_polling_entity_create_from_pollset_set(
            glb_policy->base.interested_parties),
        &glb_policy->lb_channel_connectivity,
        &glb_policy->lb_channel_on_connectivity_changed, NULL);
  }
}

// Invoked as part of the update process. It continues watching the LB channel
// until it shuts down or becomes READY. It's invoked even if the LB channel
// stayed READY throughout the update (for example if the update is identical).
static void glb_lb_channel_on_connectivity_changed_cb(grpc_exec_ctx *exec_ctx,
                                                      void *arg,
                                                      grpc_error *error) {
  glb_lb_policy *glb_policy = (glb_lb_policy *)arg;
  if (glb_policy->shutting_down) goto done;
  // Re-initialize the lb_call. This should also take care of updating the
  // embedded RR policy. Note that the current RR policy, if any, will stay in
  // effect until an update from the new lb_call is received.
  switch (glb_policy->lb_channel_connectivity) {
    case GRPC_CHANNEL_INIT:
    case GRPC_CHANNEL_CONNECTING:
    case GRPC_CHANNEL_TRANSIENT_FAILURE: {
      /* resub. */
      grpc_channel_element *client_channel_elem =
          grpc_channel_stack_last_element(
              grpc_channel_get_channel_stack(glb_policy->lb_channel));
      GPR_ASSERT(client_channel_elem->filter == &grpc_client_channel_filter);
      grpc_client_channel_watch_connectivity_state(
          exec_ctx, client_channel_elem,
          grpc_polling_entity_create_from_pollset_set(
              glb_policy->base.interested_parties),
          &glb_policy->lb_channel_connectivity,
          &glb_policy->lb_channel_on_connectivity_changed, NULL);
      break;
    }
    case GRPC_CHANNEL_IDLE:
      // lb channel inactive (probably shutdown prior to update). Restart lb
      // call to kick the lb channel into gear.
      GPR_ASSERT(glb_policy->lb_call == NULL);
    /* fallthrough */
    case GRPC_CHANNEL_READY:
      if (glb_policy->lb_call != NULL) {
        glb_policy->updating_lb_channel = false;
        glb_policy->updating_lb_call = true;
        grpc_call_cancel(glb_policy->lb_call, NULL);
        // lb_on_server_status_received will pick up the cancel and reinit
        // lb_call.
        if (glb_policy->pending_update_args != NULL) {
          grpc_lb_policy_args *args = glb_policy->pending_update_args;
          glb_policy->pending_update_args = NULL;
          glb_update_locked(exec_ctx, &glb_policy->base, args);
          grpc_channel_args_destroy(exec_ctx, args->args);
          gpr_free(args);
        }
      } else if (glb_policy->started_picking && !glb_policy->shutting_down) {
        if (glb_policy->retry_timer_active) {
          grpc_timer_cancel(exec_ctx, &glb_policy->lb_call_retry_timer);
          glb_policy->retry_timer_active = false;
        }
        start_picking_locked(exec_ctx, glb_policy);
      }
    /* fallthrough */
    case GRPC_CHANNEL_SHUTDOWN:
    done:
      glb_policy->watching_lb_channel = false;
      GRPC_LB_POLICY_WEAK_UNREF(exec_ctx, &glb_policy->base,
                                "watch_lb_channel_connectivity_cb_shutdown");
      break;
  }
}

/* Code wiring the policy with the rest of the core */
static const grpc_lb_policy_vtable glb_lb_policy_vtable = {
    glb_destroy,
    glb_shutdown_locked,
    glb_pick_locked,
    glb_cancel_pick_locked,
    glb_cancel_picks_locked,
    glb_ping_one_locked,
    glb_exit_idle_locked,
    glb_check_connectivity_locked,
    glb_notify_on_state_change_locked,
    glb_update_locked};

static grpc_lb_policy *glb_create(grpc_exec_ctx *exec_ctx,
                                  grpc_lb_policy_factory *factory,
                                  grpc_lb_policy_args *args) {
  /* Count the number of gRPC-LB addresses. There must be at least one.
   * TODO(roth): For now, we ignore non-balancer addresses, but in the
   * future, we may change the behavior such that we fall back to using
   * the non-balancer addresses if we cannot reach any balancers. In the
   * fallback case, we should use the LB policy indicated by
   * GRPC_ARG_LB_POLICY_NAME (although if that specifies grpclb or is
   * unset, we should default to pick_first). */
  const grpc_arg *arg =
      grpc_channel_args_find(args->args, GRPC_ARG_LB_ADDRESSES);
  if (arg == NULL || arg->type != GRPC_ARG_POINTER) {
    return NULL;
  }
  grpc_lb_addresses *addresses = (grpc_lb_addresses *)arg->value.pointer.p;
  size_t num_grpclb_addrs = 0;
  for (size_t i = 0; i < addresses->num_addresses; ++i) {
    if (addresses->addresses[i].is_balancer) ++num_grpclb_addrs;
  }
  if (num_grpclb_addrs == 0) return NULL;

  glb_lb_policy *glb_policy = (glb_lb_policy *)gpr_zalloc(sizeof(*glb_policy));

  /* Get server name. */
  arg = grpc_channel_args_find(args->args, GRPC_ARG_SERVER_URI);
  GPR_ASSERT(arg != NULL);
  GPR_ASSERT(arg->type == GRPC_ARG_STRING);
  grpc_uri *uri = grpc_uri_parse(exec_ctx, arg->value.string, true);
  GPR_ASSERT(uri->path[0] != '\0');
  glb_policy->server_name =
      gpr_strdup(uri->path[0] == '/' ? uri->path + 1 : uri->path);
  if (GRPC_TRACER_ON(grpc_lb_glb_trace)) {
    gpr_log(GPR_INFO, "Will use '%s' as the server name for LB request.",
            glb_policy->server_name);
  }
  grpc_uri_destroy(uri);

  glb_policy->cc_factory = args->client_channel_factory;
  GPR_ASSERT(glb_policy->cc_factory != NULL);

  arg = grpc_channel_args_find(args->args, GRPC_ARG_GRPCLB_CALL_TIMEOUT_MS);
  glb_policy->lb_call_timeout_ms =
      grpc_channel_arg_get_integer(arg, (grpc_integer_options){0, 0, INT_MAX});

  // Make sure that GRPC_ARG_LB_POLICY_NAME is set in channel args,
  // since we use this to trigger the client_load_reporting filter.
  grpc_arg new_arg =
      grpc_channel_arg_string_create(GRPC_ARG_LB_POLICY_NAME, "grpclb");
  static const char *args_to_remove[] = {GRPC_ARG_LB_POLICY_NAME};
  glb_policy->args = grpc_channel_args_copy_and_add_and_remove(
      args->args, args_to_remove, GPR_ARRAY_SIZE(args_to_remove), &new_arg, 1);

  /* Create a client channel over them to communicate with a LB service */
  glb_policy->response_generator =
      grpc_fake_resolver_response_generator_create();
  grpc_channel_args *lb_channel_args = build_lb_channel_args(
      exec_ctx, addresses, glb_policy->response_generator, args->args);
  char *uri_str;
  gpr_asprintf(&uri_str, "fake:///%s", glb_policy->server_name);
  glb_policy->lb_channel = grpc_lb_policy_grpclb_create_lb_channel(
      exec_ctx, uri_str, args->client_channel_factory, lb_channel_args);

  /* Propagate initial resolution */
  grpc_fake_resolver_response_generator_set_response(
      exec_ctx, glb_policy->response_generator, lb_channel_args);
  grpc_channel_args_destroy(exec_ctx, lb_channel_args);
  gpr_free(uri_str);
  if (glb_policy->lb_channel == NULL) {
    gpr_free((void *)glb_policy->server_name);
    grpc_channel_args_destroy(exec_ctx, glb_policy->args);
    gpr_free(glb_policy);
    return NULL;
  }
  grpc_subchannel_index_ref();
  GRPC_CLOSURE_INIT(&glb_policy->lb_channel_on_connectivity_changed,
                    glb_lb_channel_on_connectivity_changed_cb, glb_policy,
                    grpc_combiner_scheduler(args->combiner));
  grpc_lb_policy_init(&glb_policy->base, &glb_lb_policy_vtable, args->combiner);
  grpc_connectivity_state_init(&glb_policy->state_tracker, GRPC_CHANNEL_IDLE,
                               "grpclb");
  return &glb_policy->base;
}

static void glb_factory_ref(grpc_lb_policy_factory *factory) {}

static void glb_factory_unref(grpc_lb_policy_factory *factory) {}

static const grpc_lb_policy_factory_vtable glb_factory_vtable = {
    glb_factory_ref, glb_factory_unref, glb_create, "grpclb"};

static grpc_lb_policy_factory glb_lb_policy_factory = {&glb_factory_vtable};

grpc_lb_policy_factory *grpc_glb_lb_factory_create() {
  return &glb_lb_policy_factory;
}

/* Plugin registration */

// Only add client_load_reporting filter if the grpclb LB policy is used.
static bool maybe_add_client_load_reporting_filter(
    grpc_exec_ctx *exec_ctx, grpc_channel_stack_builder *builder, void *arg) {
  const grpc_channel_args *args =
      grpc_channel_stack_builder_get_channel_arguments(builder);
  const grpc_arg *channel_arg =
      grpc_channel_args_find(args, GRPC_ARG_LB_POLICY_NAME);
  if (channel_arg != NULL && channel_arg->type == GRPC_ARG_STRING &&
      strcmp(channel_arg->value.string, "grpclb") == 0) {
    return grpc_channel_stack_builder_append_filter(
        builder, (const grpc_channel_filter *)arg, NULL, NULL);
  }
  return true;
}

void grpc_lb_policy_grpclb_init() {
  grpc_register_lb_policy(grpc_glb_lb_factory_create());
  grpc_register_tracer(&grpc_lb_glb_trace);
#ifndef NDEBUG
  grpc_register_tracer(&grpc_trace_lb_policy_refcount);
#endif
  grpc_channel_init_register_stage(GRPC_CLIENT_SUBCHANNEL,
                                   GRPC_CHANNEL_INIT_BUILTIN_PRIORITY,
                                   maybe_add_client_load_reporting_filter,
                                   (void *)&grpc_client_load_reporting_filter);
}

void grpc_lb_policy_grpclb_shutdown() {}