Merge pull request #6721 from vjpai/cleanup

Remove some unused enums/structs
diff --git a/BUILD b/BUILD
index fd03d52..7601a4a 100644
--- a/BUILD
+++ b/BUILD
@@ -178,6 +178,7 @@
     "src/core/lib/iomgr/closure.h",
     "src/core/lib/iomgr/endpoint.h",
     "src/core/lib/iomgr/endpoint_pair.h",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.h",
     "src/core/lib/iomgr/ev_poll_posix.h",
     "src/core/lib/iomgr/ev_posix.h",
     "src/core/lib/iomgr/exec_ctx.h",
@@ -321,6 +322,7 @@
     "src/core/lib/iomgr/endpoint.c",
     "src/core/lib/iomgr/endpoint_pair_posix.c",
     "src/core/lib/iomgr/endpoint_pair_windows.c",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.c",
     "src/core/lib/iomgr/ev_poll_posix.c",
     "src/core/lib/iomgr/ev_posix.c",
     "src/core/lib/iomgr/exec_ctx.c",
@@ -546,6 +548,7 @@
     "src/core/lib/iomgr/closure.h",
     "src/core/lib/iomgr/endpoint.h",
     "src/core/lib/iomgr/endpoint_pair.h",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.h",
     "src/core/lib/iomgr/ev_poll_posix.h",
     "src/core/lib/iomgr/ev_posix.h",
     "src/core/lib/iomgr/exec_ctx.h",
@@ -666,6 +669,7 @@
     "src/core/lib/iomgr/endpoint.c",
     "src/core/lib/iomgr/endpoint_pair_posix.c",
     "src/core/lib/iomgr/endpoint_pair_windows.c",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.c",
     "src/core/lib/iomgr/ev_poll_posix.c",
     "src/core/lib/iomgr/ev_posix.c",
     "src/core/lib/iomgr/exec_ctx.c",
@@ -1358,6 +1362,7 @@
     "src/core/lib/iomgr/endpoint.c",
     "src/core/lib/iomgr/endpoint_pair_posix.c",
     "src/core/lib/iomgr/endpoint_pair_windows.c",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.c",
     "src/core/lib/iomgr/ev_poll_posix.c",
     "src/core/lib/iomgr/ev_posix.c",
     "src/core/lib/iomgr/exec_ctx.c",
@@ -1562,6 +1567,7 @@
     "src/core/lib/iomgr/closure.h",
     "src/core/lib/iomgr/endpoint.h",
     "src/core/lib/iomgr/endpoint_pair.h",
+    "src/core/lib/iomgr/ev_poll_and_epoll_posix.h",
     "src/core/lib/iomgr/ev_poll_posix.h",
     "src/core/lib/iomgr/ev_posix.h",
     "src/core/lib/iomgr/exec_ctx.h",
diff --git a/Makefile b/Makefile
index 817fcd0..710f626 100644
--- a/Makefile
+++ b/Makefile
@@ -2486,6 +2486,7 @@
     src/core/lib/iomgr/endpoint.c \
     src/core/lib/iomgr/endpoint_pair_posix.c \
     src/core/lib/iomgr/endpoint_pair_windows.c \
+    src/core/lib/iomgr/ev_poll_and_epoll_posix.c \
     src/core/lib/iomgr/ev_poll_posix.c \
     src/core/lib/iomgr/ev_posix.c \
     src/core/lib/iomgr/exec_ctx.c \
@@ -2840,6 +2841,7 @@
     src/core/lib/iomgr/endpoint.c \
     src/core/lib/iomgr/endpoint_pair_posix.c \
     src/core/lib/iomgr/endpoint_pair_windows.c \
+    src/core/lib/iomgr/ev_poll_and_epoll_posix.c \
     src/core/lib/iomgr/ev_poll_posix.c \
     src/core/lib/iomgr/ev_posix.c \
     src/core/lib/iomgr/exec_ctx.c \
diff --git a/binding.gyp b/binding.gyp
index 7b18707..f2271cb 100644
--- a/binding.gyp
+++ b/binding.gyp
@@ -581,6 +581,7 @@
         'src/core/lib/iomgr/endpoint.c',
         'src/core/lib/iomgr/endpoint_pair_posix.c',
         'src/core/lib/iomgr/endpoint_pair_windows.c',
+        'src/core/lib/iomgr/ev_poll_and_epoll_posix.c',
         'src/core/lib/iomgr/ev_poll_posix.c',
         'src/core/lib/iomgr/ev_posix.c',
         'src/core/lib/iomgr/exec_ctx.c',
diff --git a/build.yaml b/build.yaml
index 429dbb3..b76bd12 100644
--- a/build.yaml
+++ b/build.yaml
@@ -165,6 +165,7 @@
   - src/core/lib/iomgr/closure.h
   - src/core/lib/iomgr/endpoint.h
   - src/core/lib/iomgr/endpoint_pair.h
+  - src/core/lib/iomgr/ev_poll_and_epoll_posix.h
   - src/core/lib/iomgr/ev_poll_posix.h
   - src/core/lib/iomgr/ev_posix.h
   - src/core/lib/iomgr/exec_ctx.h
@@ -239,6 +240,7 @@
   - src/core/lib/iomgr/endpoint.c
   - src/core/lib/iomgr/endpoint_pair_posix.c
   - src/core/lib/iomgr/endpoint_pair_windows.c
+  - src/core/lib/iomgr/ev_poll_and_epoll_posix.c
   - src/core/lib/iomgr/ev_poll_posix.c
   - src/core/lib/iomgr/ev_posix.c
   - src/core/lib/iomgr/exec_ctx.c
diff --git a/config.m4 b/config.m4
index 29df77c..67c9bef 100644
--- a/config.m4
+++ b/config.m4
@@ -100,6 +100,7 @@
     src/core/lib/iomgr/endpoint.c \
     src/core/lib/iomgr/endpoint_pair_posix.c \
     src/core/lib/iomgr/endpoint_pair_windows.c \
+    src/core/lib/iomgr/ev_poll_and_epoll_posix.c \
     src/core/lib/iomgr/ev_poll_posix.c \
     src/core/lib/iomgr/ev_posix.c \
     src/core/lib/iomgr/exec_ctx.c \
diff --git a/doc/connectivity-semantics-and-api.md b/doc/connectivity-semantics-and-api.md
index 5427900..cc007ea 100644
--- a/doc/connectivity-semantics-and-api.md
+++ b/doc/connectivity-semantics-and-api.md
@@ -101,7 +101,7 @@
     <td>Shutdown triggered by application.</td>
   </tr>
   <tr>
-    <th>FATAL_FAILURE</th>
+    <th>SHUTDOWN</th>
     <td></td>
     <td></td>
     <td></td>
diff --git a/gRPC.podspec b/gRPC.podspec
index cc8ba3d..436cec4 100644
--- a/gRPC.podspec
+++ b/gRPC.podspec
@@ -181,6 +181,7 @@
                       'src/core/lib/iomgr/closure.h',
                       'src/core/lib/iomgr/endpoint.h',
                       'src/core/lib/iomgr/endpoint_pair.h',
+                      'src/core/lib/iomgr/ev_poll_and_epoll_posix.h',
                       'src/core/lib/iomgr/ev_poll_posix.h',
                       'src/core/lib/iomgr/ev_posix.h',
                       'src/core/lib/iomgr/exec_ctx.h',
@@ -358,6 +359,7 @@
                       'src/core/lib/iomgr/endpoint.c',
                       'src/core/lib/iomgr/endpoint_pair_posix.c',
                       'src/core/lib/iomgr/endpoint_pair_windows.c',
+                      'src/core/lib/iomgr/ev_poll_and_epoll_posix.c',
                       'src/core/lib/iomgr/ev_poll_posix.c',
                       'src/core/lib/iomgr/ev_posix.c',
                       'src/core/lib/iomgr/exec_ctx.c',
@@ -546,6 +548,7 @@
                               'src/core/lib/iomgr/closure.h',
                               'src/core/lib/iomgr/endpoint.h',
                               'src/core/lib/iomgr/endpoint_pair.h',
+                              'src/core/lib/iomgr/ev_poll_and_epoll_posix.h',
                               'src/core/lib/iomgr/ev_poll_posix.h',
                               'src/core/lib/iomgr/ev_posix.h',
                               'src/core/lib/iomgr/exec_ctx.h',
diff --git a/grpc.gemspec b/grpc.gemspec
index ae7f9b7..55ec9a9 100755
--- a/grpc.gemspec
+++ b/grpc.gemspec
@@ -190,6 +190,7 @@
   s.files += %w( src/core/lib/iomgr/closure.h )
   s.files += %w( src/core/lib/iomgr/endpoint.h )
   s.files += %w( src/core/lib/iomgr/endpoint_pair.h )
+  s.files += %w( src/core/lib/iomgr/ev_poll_and_epoll_posix.h )
   s.files += %w( src/core/lib/iomgr/ev_poll_posix.h )
   s.files += %w( src/core/lib/iomgr/ev_posix.h )
   s.files += %w( src/core/lib/iomgr/exec_ctx.h )
@@ -337,6 +338,7 @@
   s.files += %w( src/core/lib/iomgr/endpoint.c )
   s.files += %w( src/core/lib/iomgr/endpoint_pair_posix.c )
   s.files += %w( src/core/lib/iomgr/endpoint_pair_windows.c )
+  s.files += %w( src/core/lib/iomgr/ev_poll_and_epoll_posix.c )
   s.files += %w( src/core/lib/iomgr/ev_poll_posix.c )
   s.files += %w( src/core/lib/iomgr/ev_posix.c )
   s.files += %w( src/core/lib/iomgr/exec_ctx.c )
diff --git a/package.xml b/package.xml
index 507a2a7..c7fbaf5 100644
--- a/package.xml
+++ b/package.xml
@@ -197,6 +197,7 @@
     <file baseinstalldir="/" name="src/core/lib/iomgr/closure.h" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/endpoint.h" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/endpoint_pair.h" role="src" />
+    <file baseinstalldir="/" name="src/core/lib/iomgr/ev_poll_and_epoll_posix.h" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/ev_poll_posix.h" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/ev_posix.h" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/exec_ctx.h" role="src" />
@@ -344,6 +345,7 @@
     <file baseinstalldir="/" name="src/core/lib/iomgr/endpoint.c" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/endpoint_pair_posix.c" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/endpoint_pair_windows.c" role="src" />
+    <file baseinstalldir="/" name="src/core/lib/iomgr/ev_poll_and_epoll_posix.c" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/ev_poll_posix.c" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/ev_posix.c" role="src" />
     <file baseinstalldir="/" name="src/core/lib/iomgr/exec_ctx.c" role="src" />
diff --git a/src/core/ext/transport/cronet/transport/cronet_transport.c b/src/core/ext/transport/cronet/transport/cronet_transport.c
index 5bb0851..df160aa 100644
--- a/src/core/ext/transport/cronet/transport/cronet_transport.c
+++ b/src/core/ext/transport/cronet/transport/cronet_transport.c
@@ -218,8 +218,11 @@
 static void process_recv_message(stream_obj *s, const uint8_t *recv_data) {
   gpr_slice read_data_slice = gpr_slice_malloc((uint32_t)s->total_read_bytes);
   uint8_t *dst_p = GPR_SLICE_START_PTR(read_data_slice);
-  memcpy(dst_p, recv_data, (size_t)s->total_read_bytes);
-  gpr_slice_buffer_add(&s->read_slice_buffer, read_data_slice);
+  if (s->total_read_bytes > 0) {
+    // Only copy if there is non-zero number of bytes
+    memcpy(dst_p, recv_data, (size_t)s->total_read_bytes);
+    gpr_slice_buffer_add(&s->read_slice_buffer, read_data_slice);
+  }
   grpc_slice_buffer_stream_init(&s->sbs, &s->read_slice_buffer, 0);
   *s->recv_message = (grpc_byte_buffer *)&s->sbs;
 }
@@ -347,8 +350,17 @@
           if (grpc_cronet_trace) {
             gpr_log(GPR_DEBUG, "R: cronet_bidirectional_stream_read()");
           }
-          cronet_bidirectional_stream_read(s->cbs, (char *)s->read_buffer,
-                                           s->remaining_read_bytes);
+          if (s->remaining_read_bytes > 0) {
+            cronet_bidirectional_stream_read(s->cbs, (char *)s->read_buffer,
+                                             s->remaining_read_bytes);
+          } else {
+            // Calling the closing callback directly since this is a 0 byte read
+            // for an empty message.
+            process_recv_message(s, NULL);
+            enqueue_callbacks(s->callback_list[CB_RECV_MESSAGE]);
+            invoke_closing_callback(s);
+            set_recv_state(s, CRONET_RECV_CLOSED);
+          }
         }
       }
       break;
diff --git a/src/core/lib/iomgr/ev_poll_and_epoll_posix.c b/src/core/lib/iomgr/ev_poll_and_epoll_posix.c
new file mode 100644
index 0000000..943c404
--- /dev/null
+++ b/src/core/lib/iomgr/ev_poll_and_epoll_posix.c
@@ -0,0 +1,1978 @@
+/*
+ *
+ * Copyright 2015, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+/* This file will be removed shortly: it's here to keep refactoring
+ * steps simple and auditable.
+ * It's the combination of the old files:
+ *  - fd_posix.{h,c}
+ *  - pollset_posix.{h,c}
+ *  - pullset_multipoller_with_{poll,epoll}.{h,c}
+ * The new version will be split into:
+ *  - ev_poll_posix.{h,c}
+ *  - ev_epoll_posix.{h,c}
+ */
+
+#include <grpc/support/port_platform.h>
+
+#ifdef GPR_POSIX_SOCKET
+
+#include "src/core/lib/iomgr/ev_poll_and_epoll_posix.h"
+
+#include <assert.h>
+#include <errno.h>
+#include <poll.h>
+#include <string.h>
+#include <sys/socket.h>
+#include <unistd.h>
+
+#include <grpc/support/alloc.h>
+#include <grpc/support/log.h>
+#include <grpc/support/string_util.h>
+#include <grpc/support/tls.h>
+#include <grpc/support/useful.h>
+
+#include "src/core/lib/iomgr/iomgr_internal.h"
+#include "src/core/lib/iomgr/wakeup_fd_posix.h"
+#include "src/core/lib/profiling/timers.h"
+#include "src/core/lib/support/block_annotate.h"
+
+/*******************************************************************************
+ * FD declarations
+ */
+
+typedef struct grpc_fd_watcher {
+  struct grpc_fd_watcher *next;
+  struct grpc_fd_watcher *prev;
+  grpc_pollset *pollset;
+  grpc_pollset_worker *worker;
+  grpc_fd *fd;
+} grpc_fd_watcher;
+
+struct grpc_fd {
+  int fd;
+  /* refst format:
+     bit0:   1=active/0=orphaned
+     bit1-n: refcount
+     meaning that mostly we ref by two to avoid altering the orphaned bit,
+     and just unref by 1 when we're ready to flag the object as orphaned */
+  gpr_atm refst;
+
+  gpr_mu mu;
+  int shutdown;
+  int closed;
+  int released;
+
+  /* The watcher list.
+
+     The following watcher related fields are protected by watcher_mu.
+
+     An fd_watcher is an ephemeral object created when an fd wants to
+     begin polling, and destroyed after the poll.
+
+     It denotes the fd's interest in whether to read poll or write poll
+     or both or neither on this fd.
+
+     If a watcher is asked to poll for reads or writes, the read_watcher
+     or write_watcher fields are set respectively. A watcher may be asked
+     to poll for both, in which case both fields will be set.
+
+     read_watcher and write_watcher may be NULL if no watcher has been
+     asked to poll for reads or writes.
+
+     If an fd_watcher is not asked to poll for reads or writes, it's added
+     to a linked list of inactive watchers, rooted at inactive_watcher_root.
+     If at a later time there becomes need of a poller to poll, one of
+     the inactive pollers may be kicked out of their poll loops to take
+     that responsibility. */
+  grpc_fd_watcher inactive_watcher_root;
+  grpc_fd_watcher *read_watcher;
+  grpc_fd_watcher *write_watcher;
+
+  grpc_closure *read_closure;
+  grpc_closure *write_closure;
+
+  struct grpc_fd *freelist_next;
+
+  grpc_closure *on_done_closure;
+
+  grpc_iomgr_object iomgr_object;
+
+  /* The pollset that last noticed and notified that the fd is readable */
+  grpc_pollset *read_notifier_pollset;
+};
+
+/* Begin polling on an fd.
+   Registers that the given pollset is interested in this fd - so that if read
+   or writability interest changes, the pollset can be kicked to pick up that
+   new interest.
+   Return value is:
+     (fd_needs_read? read_mask : 0) | (fd_needs_write? write_mask : 0)
+   i.e. a combination of read_mask and write_mask determined by the fd's current
+   interest in said events.
+   Polling strategies that do not need to alter their behavior depending on the
+   fd's current interest (such as epoll) do not need to call this function.
+   MUST NOT be called with a pollset lock taken */
+static uint32_t fd_begin_poll(grpc_fd *fd, grpc_pollset *pollset,
+                              grpc_pollset_worker *worker, uint32_t read_mask,
+                              uint32_t write_mask, grpc_fd_watcher *rec);
+/* Complete polling previously started with fd_begin_poll
+   MUST NOT be called with a pollset lock taken
+   if got_read or got_write are 1, also does the become_{readable,writable} as
+   appropriate. */
+static void fd_end_poll(grpc_exec_ctx *exec_ctx, grpc_fd_watcher *rec,
+                        int got_read, int got_write,
+                        grpc_pollset *read_notifier_pollset);
+
+/* Return 1 if this fd is orphaned, 0 otherwise */
+static bool fd_is_orphaned(grpc_fd *fd);
+
+/* Reference counting for fds */
+/*#define GRPC_FD_REF_COUNT_DEBUG*/
+#ifdef GRPC_FD_REF_COUNT_DEBUG
+static void fd_ref(grpc_fd *fd, const char *reason, const char *file, int line);
+static void fd_unref(grpc_fd *fd, const char *reason, const char *file,
+                     int line);
+#define GRPC_FD_REF(fd, reason) fd_ref(fd, reason, __FILE__, __LINE__)
+#define GRPC_FD_UNREF(fd, reason) fd_unref(fd, reason, __FILE__, __LINE__)
+#else
+static void fd_ref(grpc_fd *fd);
+static void fd_unref(grpc_fd *fd);
+#define GRPC_FD_REF(fd, reason) fd_ref(fd)
+#define GRPC_FD_UNREF(fd, reason) fd_unref(fd)
+#endif
+
+static void fd_global_init(void);
+static void fd_global_shutdown(void);
+
+#define CLOSURE_NOT_READY ((grpc_closure *)0)
+#define CLOSURE_READY ((grpc_closure *)1)
+
+/*******************************************************************************
+ * pollset declarations
+ */
+
+typedef struct grpc_pollset_vtable grpc_pollset_vtable;
+
+typedef struct grpc_cached_wakeup_fd {
+  grpc_wakeup_fd fd;
+  struct grpc_cached_wakeup_fd *next;
+} grpc_cached_wakeup_fd;
+
+struct grpc_pollset_worker {
+  grpc_cached_wakeup_fd *wakeup_fd;
+  int reevaluate_polling_on_wakeup;
+  int kicked_specifically;
+  struct grpc_pollset_worker *next;
+  struct grpc_pollset_worker *prev;
+};
+
+struct grpc_pollset {
+  /* pollsets under posix can mutate representation as fds are added and
+     removed.
+     For example, we may choose a poll() based implementation on linux for
+     few fds, and an epoll() based implementation for many fds */
+  const grpc_pollset_vtable *vtable;
+  gpr_mu mu;
+  grpc_pollset_worker root_worker;
+  int in_flight_cbs;
+  int shutting_down;
+  int called_shutdown;
+  int kicked_without_pollers;
+  grpc_closure *shutdown_done;
+  grpc_closure_list idle_jobs;
+  union {
+    int fd;
+    void *ptr;
+  } data;
+  /* Local cache of eventfds for workers */
+  grpc_cached_wakeup_fd *local_wakeup_cache;
+};
+
+struct grpc_pollset_vtable {
+  void (*add_fd)(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                 struct grpc_fd *fd, int and_unlock_pollset);
+  void (*maybe_work_and_unlock)(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                                grpc_pollset_worker *worker,
+                                gpr_timespec deadline, gpr_timespec now);
+  void (*finish_shutdown)(grpc_pollset *pollset);
+  void (*destroy)(grpc_pollset *pollset);
+};
+
+/* Add an fd to a pollset */
+static void pollset_add_fd(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                           struct grpc_fd *fd);
+
+static void pollset_set_add_fd(grpc_exec_ctx *exec_ctx,
+                               grpc_pollset_set *pollset_set, grpc_fd *fd);
+
+/* Convert a timespec to milliseconds:
+   - very small or negative poll times are clamped to zero to do a
+     non-blocking poll (which becomes spin polling)
+   - other small values are rounded up to one millisecond
+   - longer than a millisecond polls are rounded up to the next nearest
+     millisecond to avoid spinning
+   - infinite timeouts are converted to -1 */
+static int poll_deadline_to_millis_timeout(gpr_timespec deadline,
+                                           gpr_timespec now);
+
+/* Allow kick to wakeup the currently polling worker */
+#define GRPC_POLLSET_CAN_KICK_SELF 1
+/* Force the wakee to repoll when awoken */
+#define GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP 2
+/* As per pollset_kick, with an extended set of flags (defined above)
+   -- mostly for fd_posix's use. */
+static void pollset_kick_ext(grpc_pollset *p,
+                             grpc_pollset_worker *specific_worker,
+                             uint32_t flags);
+
+/* turn a pollset into a multipoller: platform specific */
+typedef void (*platform_become_multipoller_type)(grpc_exec_ctx *exec_ctx,
+                                                 grpc_pollset *pollset,
+                                                 struct grpc_fd **fds,
+                                                 size_t fd_count);
+static platform_become_multipoller_type platform_become_multipoller;
+
+/* Return 1 if the pollset has active threads in pollset_work (pollset must
+ * be locked) */
+static int pollset_has_workers(grpc_pollset *pollset);
+
+static void remove_fd_from_all_epoll_sets(int fd);
+
+/*******************************************************************************
+ * pollset_set definitions
+ */
+
+struct grpc_pollset_set {
+  gpr_mu mu;
+
+  size_t pollset_count;
+  size_t pollset_capacity;
+  grpc_pollset **pollsets;
+
+  size_t pollset_set_count;
+  size_t pollset_set_capacity;
+  struct grpc_pollset_set **pollset_sets;
+
+  size_t fd_count;
+  size_t fd_capacity;
+  grpc_fd **fds;
+};
+
+/*******************************************************************************
+ * fd_posix.c
+ */
+
+/* We need to keep a freelist not because of any concerns of malloc performance
+ * but instead so that implementations with multiple threads in (for example)
+ * epoll_wait deal with the race between pollset removal and incoming poll
+ * notifications.
+ *
+ * The problem is that the poller ultimately holds a reference to this
+ * object, so it is very difficult to know when is safe to free it, at least
+ * without some expensive synchronization.
+ *
+ * If we keep the object freelisted, in the worst case losing this race just
+ * becomes a spurious read notification on a reused fd.
+ */
+/* TODO(klempner): We could use some form of polling generation count to know
+ * when these are safe to free. */
+/* TODO(klempner): Consider disabling freelisting if we don't have multiple
+ * threads in poll on the same fd */
+/* TODO(klempner): Batch these allocations to reduce fragmentation */
+static grpc_fd *fd_freelist = NULL;
+static gpr_mu fd_freelist_mu;
+
+static void freelist_fd(grpc_fd *fd) {
+  gpr_mu_lock(&fd_freelist_mu);
+  fd->freelist_next = fd_freelist;
+  fd_freelist = fd;
+  grpc_iomgr_unregister_object(&fd->iomgr_object);
+  gpr_mu_unlock(&fd_freelist_mu);
+}
+
+static grpc_fd *alloc_fd(int fd) {
+  grpc_fd *r = NULL;
+  gpr_mu_lock(&fd_freelist_mu);
+  if (fd_freelist != NULL) {
+    r = fd_freelist;
+    fd_freelist = fd_freelist->freelist_next;
+  }
+  gpr_mu_unlock(&fd_freelist_mu);
+  if (r == NULL) {
+    r = gpr_malloc(sizeof(grpc_fd));
+    gpr_mu_init(&r->mu);
+  }
+
+  gpr_mu_lock(&r->mu);
+  gpr_atm_rel_store(&r->refst, 1);
+  r->shutdown = 0;
+  r->read_closure = CLOSURE_NOT_READY;
+  r->write_closure = CLOSURE_NOT_READY;
+  r->fd = fd;
+  r->inactive_watcher_root.next = r->inactive_watcher_root.prev =
+      &r->inactive_watcher_root;
+  r->freelist_next = NULL;
+  r->read_watcher = r->write_watcher = NULL;
+  r->on_done_closure = NULL;
+  r->closed = 0;
+  r->released = 0;
+  r->read_notifier_pollset = NULL;
+  gpr_mu_unlock(&r->mu);
+  return r;
+}
+
+static void destroy(grpc_fd *fd) {
+  gpr_mu_destroy(&fd->mu);
+  gpr_free(fd);
+}
+
+#ifdef GRPC_FD_REF_COUNT_DEBUG
+#define REF_BY(fd, n, reason) ref_by(fd, n, reason, __FILE__, __LINE__)
+#define UNREF_BY(fd, n, reason) unref_by(fd, n, reason, __FILE__, __LINE__)
+static void ref_by(grpc_fd *fd, int n, const char *reason, const char *file,
+                   int line) {
+  gpr_log(GPR_DEBUG, "FD %d %p   ref %d %d -> %d [%s; %s:%d]", fd->fd, fd, n,
+          gpr_atm_no_barrier_load(&fd->refst),
+          gpr_atm_no_barrier_load(&fd->refst) + n, reason, file, line);
+#else
+#define REF_BY(fd, n, reason) ref_by(fd, n)
+#define UNREF_BY(fd, n, reason) unref_by(fd, n)
+static void ref_by(grpc_fd *fd, int n) {
+#endif
+  GPR_ASSERT(gpr_atm_no_barrier_fetch_add(&fd->refst, n) > 0);
+}
+
+#ifdef GRPC_FD_REF_COUNT_DEBUG
+static void unref_by(grpc_fd *fd, int n, const char *reason, const char *file,
+                     int line) {
+  gpr_atm old;
+  gpr_log(GPR_DEBUG, "FD %d %p unref %d %d -> %d [%s; %s:%d]", fd->fd, fd, n,
+          gpr_atm_no_barrier_load(&fd->refst),
+          gpr_atm_no_barrier_load(&fd->refst) - n, reason, file, line);
+#else
+static void unref_by(grpc_fd *fd, int n) {
+  gpr_atm old;
+#endif
+  old = gpr_atm_full_fetch_add(&fd->refst, -n);
+  if (old == n) {
+    freelist_fd(fd);
+  } else {
+    GPR_ASSERT(old > n);
+  }
+}
+
+static void fd_global_init(void) { gpr_mu_init(&fd_freelist_mu); }
+
+static void fd_global_shutdown(void) {
+  gpr_mu_lock(&fd_freelist_mu);
+  gpr_mu_unlock(&fd_freelist_mu);
+  while (fd_freelist != NULL) {
+    grpc_fd *fd = fd_freelist;
+    fd_freelist = fd_freelist->freelist_next;
+    destroy(fd);
+  }
+  gpr_mu_destroy(&fd_freelist_mu);
+}
+
+static grpc_fd *fd_create(int fd, const char *name) {
+  grpc_fd *r = alloc_fd(fd);
+  char *name2;
+  gpr_asprintf(&name2, "%s fd=%d", name, fd);
+  grpc_iomgr_register_object(&r->iomgr_object, name2);
+  gpr_free(name2);
+#ifdef GRPC_FD_REF_COUNT_DEBUG
+  gpr_log(GPR_DEBUG, "FD %d %p create %s", fd, r, name);
+#endif
+  return r;
+}
+
+static bool fd_is_orphaned(grpc_fd *fd) {
+  return (gpr_atm_acq_load(&fd->refst) & 1) == 0;
+}
+
+static void pollset_kick_locked(grpc_fd_watcher *watcher) {
+  gpr_mu_lock(&watcher->pollset->mu);
+  GPR_ASSERT(watcher->worker);
+  pollset_kick_ext(watcher->pollset, watcher->worker,
+                   GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP);
+  gpr_mu_unlock(&watcher->pollset->mu);
+}
+
+static void maybe_wake_one_watcher_locked(grpc_fd *fd) {
+  if (fd->inactive_watcher_root.next != &fd->inactive_watcher_root) {
+    pollset_kick_locked(fd->inactive_watcher_root.next);
+  } else if (fd->read_watcher) {
+    pollset_kick_locked(fd->read_watcher);
+  } else if (fd->write_watcher) {
+    pollset_kick_locked(fd->write_watcher);
+  }
+}
+
+static void wake_all_watchers_locked(grpc_fd *fd) {
+  grpc_fd_watcher *watcher;
+  for (watcher = fd->inactive_watcher_root.next;
+       watcher != &fd->inactive_watcher_root; watcher = watcher->next) {
+    pollset_kick_locked(watcher);
+  }
+  if (fd->read_watcher) {
+    pollset_kick_locked(fd->read_watcher);
+  }
+  if (fd->write_watcher && fd->write_watcher != fd->read_watcher) {
+    pollset_kick_locked(fd->write_watcher);
+  }
+}
+
+static int has_watchers(grpc_fd *fd) {
+  return fd->read_watcher != NULL || fd->write_watcher != NULL ||
+         fd->inactive_watcher_root.next != &fd->inactive_watcher_root;
+}
+
+static void close_fd_locked(grpc_exec_ctx *exec_ctx, grpc_fd *fd) {
+  fd->closed = 1;
+  if (!fd->released) {
+    close(fd->fd);
+  } else {
+    remove_fd_from_all_epoll_sets(fd->fd);
+  }
+  grpc_exec_ctx_enqueue(exec_ctx, fd->on_done_closure, true, NULL);
+}
+
+static int fd_wrapped_fd(grpc_fd *fd) {
+  if (fd->released || fd->closed) {
+    return -1;
+  } else {
+    return fd->fd;
+  }
+}
+
+static void fd_orphan(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                      grpc_closure *on_done, int *release_fd,
+                      const char *reason) {
+  fd->on_done_closure = on_done;
+  fd->released = release_fd != NULL;
+  if (!fd->released) {
+    shutdown(fd->fd, SHUT_RDWR);
+  } else {
+    *release_fd = fd->fd;
+  }
+  gpr_mu_lock(&fd->mu);
+  REF_BY(fd, 1, reason); /* remove active status, but keep referenced */
+  if (!has_watchers(fd)) {
+    close_fd_locked(exec_ctx, fd);
+  } else {
+    wake_all_watchers_locked(fd);
+  }
+  gpr_mu_unlock(&fd->mu);
+  UNREF_BY(fd, 2, reason); /* drop the reference */
+}
+
+/* increment refcount by two to avoid changing the orphan bit */
+#ifdef GRPC_FD_REF_COUNT_DEBUG
+static void fd_ref(grpc_fd *fd, const char *reason, const char *file,
+                   int line) {
+  ref_by(fd, 2, reason, file, line);
+}
+
+static void fd_unref(grpc_fd *fd, const char *reason, const char *file,
+                     int line) {
+  unref_by(fd, 2, reason, file, line);
+}
+#else
+static void fd_ref(grpc_fd *fd) { ref_by(fd, 2); }
+
+static void fd_unref(grpc_fd *fd) { unref_by(fd, 2); }
+#endif
+
+static void notify_on_locked(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                             grpc_closure **st, grpc_closure *closure) {
+  if (*st == CLOSURE_NOT_READY) {
+    /* not ready ==> switch to a waiting state by setting the closure */
+    *st = closure;
+  } else if (*st == CLOSURE_READY) {
+    /* already ready ==> queue the closure to run immediately */
+    *st = CLOSURE_NOT_READY;
+    grpc_exec_ctx_enqueue(exec_ctx, closure, !fd->shutdown, NULL);
+    maybe_wake_one_watcher_locked(fd);
+  } else {
+    /* upcallptr was set to a different closure.  This is an error! */
+    gpr_log(GPR_ERROR,
+            "User called a notify_on function with a previous callback still "
+            "pending");
+    abort();
+  }
+}
+
+/* returns 1 if state becomes not ready */
+static int set_ready_locked(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                            grpc_closure **st) {
+  if (*st == CLOSURE_READY) {
+    /* duplicate ready ==> ignore */
+    return 0;
+  } else if (*st == CLOSURE_NOT_READY) {
+    /* not ready, and not waiting ==> flag ready */
+    *st = CLOSURE_READY;
+    return 0;
+  } else {
+    /* waiting ==> queue closure */
+    grpc_exec_ctx_enqueue(exec_ctx, *st, !fd->shutdown, NULL);
+    *st = CLOSURE_NOT_READY;
+    return 1;
+  }
+}
+
+static void set_read_notifier_pollset_locked(
+    grpc_exec_ctx *exec_ctx, grpc_fd *fd, grpc_pollset *read_notifier_pollset) {
+  fd->read_notifier_pollset = read_notifier_pollset;
+}
+
+static void fd_shutdown(grpc_exec_ctx *exec_ctx, grpc_fd *fd) {
+  gpr_mu_lock(&fd->mu);
+  GPR_ASSERT(!fd->shutdown);
+  fd->shutdown = 1;
+  set_ready_locked(exec_ctx, fd, &fd->read_closure);
+  set_ready_locked(exec_ctx, fd, &fd->write_closure);
+  gpr_mu_unlock(&fd->mu);
+}
+
+static void fd_notify_on_read(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                              grpc_closure *closure) {
+  gpr_mu_lock(&fd->mu);
+  notify_on_locked(exec_ctx, fd, &fd->read_closure, closure);
+  gpr_mu_unlock(&fd->mu);
+}
+
+static void fd_notify_on_write(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                               grpc_closure *closure) {
+  gpr_mu_lock(&fd->mu);
+  notify_on_locked(exec_ctx, fd, &fd->write_closure, closure);
+  gpr_mu_unlock(&fd->mu);
+}
+
+/* Return the read-notifier pollset */
+static grpc_pollset *fd_get_read_notifier_pollset(grpc_exec_ctx *exec_ctx,
+                                                  grpc_fd *fd) {
+  grpc_pollset *notifier = NULL;
+
+  gpr_mu_lock(&fd->mu);
+  notifier = fd->read_notifier_pollset;
+  gpr_mu_unlock(&fd->mu);
+
+  return notifier;
+}
+
+static uint32_t fd_begin_poll(grpc_fd *fd, grpc_pollset *pollset,
+                              grpc_pollset_worker *worker, uint32_t read_mask,
+                              uint32_t write_mask, grpc_fd_watcher *watcher) {
+  uint32_t mask = 0;
+  grpc_closure *cur;
+  int requested;
+  /* keep track of pollers that have requested our events, in case they change
+   */
+  GRPC_FD_REF(fd, "poll");
+
+  gpr_mu_lock(&fd->mu);
+
+  /* if we are shutdown, then don't add to the watcher set */
+  if (fd->shutdown) {
+    watcher->fd = NULL;
+    watcher->pollset = NULL;
+    watcher->worker = NULL;
+    gpr_mu_unlock(&fd->mu);
+    GRPC_FD_UNREF(fd, "poll");
+    return 0;
+  }
+
+  /* if there is nobody polling for read, but we need to, then start doing so */
+  cur = fd->read_closure;
+  requested = cur != CLOSURE_READY;
+  if (read_mask && fd->read_watcher == NULL && requested) {
+    fd->read_watcher = watcher;
+    mask |= read_mask;
+  }
+  /* if there is nobody polling for write, but we need to, then start doing so
+   */
+  cur = fd->write_closure;
+  requested = cur != CLOSURE_READY;
+  if (write_mask && fd->write_watcher == NULL && requested) {
+    fd->write_watcher = watcher;
+    mask |= write_mask;
+  }
+  /* if not polling, remember this watcher in case we need someone to later */
+  if (mask == 0 && worker != NULL) {
+    watcher->next = &fd->inactive_watcher_root;
+    watcher->prev = watcher->next->prev;
+    watcher->next->prev = watcher->prev->next = watcher;
+  }
+  watcher->pollset = pollset;
+  watcher->worker = worker;
+  watcher->fd = fd;
+  gpr_mu_unlock(&fd->mu);
+
+  return mask;
+}
+
+static void fd_end_poll(grpc_exec_ctx *exec_ctx, grpc_fd_watcher *watcher,
+                        int got_read, int got_write,
+                        grpc_pollset *read_notifier_pollset) {
+  int was_polling = 0;
+  int kick = 0;
+  grpc_fd *fd = watcher->fd;
+
+  if (fd == NULL) {
+    return;
+  }
+
+  gpr_mu_lock(&fd->mu);
+
+  if (watcher == fd->read_watcher) {
+    /* remove read watcher, kick if we still need a read */
+    was_polling = 1;
+    if (!got_read) {
+      kick = 1;
+    }
+    fd->read_watcher = NULL;
+  }
+  if (watcher == fd->write_watcher) {
+    /* remove write watcher, kick if we still need a write */
+    was_polling = 1;
+    if (!got_write) {
+      kick = 1;
+    }
+    fd->write_watcher = NULL;
+  }
+  if (!was_polling && watcher->worker != NULL) {
+    /* remove from inactive list */
+    watcher->next->prev = watcher->prev;
+    watcher->prev->next = watcher->next;
+  }
+  if (got_read) {
+    if (set_ready_locked(exec_ctx, fd, &fd->read_closure)) {
+      kick = 1;
+    }
+
+    if (read_notifier_pollset != NULL) {
+      set_read_notifier_pollset_locked(exec_ctx, fd, read_notifier_pollset);
+    }
+  }
+  if (got_write) {
+    if (set_ready_locked(exec_ctx, fd, &fd->write_closure)) {
+      kick = 1;
+    }
+  }
+  if (kick) {
+    maybe_wake_one_watcher_locked(fd);
+  }
+  if (fd_is_orphaned(fd) && !has_watchers(fd) && !fd->closed) {
+    close_fd_locked(exec_ctx, fd);
+  }
+  gpr_mu_unlock(&fd->mu);
+
+  GRPC_FD_UNREF(fd, "poll");
+}
+
+/*******************************************************************************
+ * pollset_posix.c
+ */
+
+GPR_TLS_DECL(g_current_thread_poller);
+GPR_TLS_DECL(g_current_thread_worker);
+
+/** The alarm system needs to be able to wakeup 'some poller' sometimes
+ *  (specifically when a new alarm needs to be triggered earlier than the next
+ *  alarm 'epoch').
+ *  This wakeup_fd gives us something to alert on when such a case occurs. */
+grpc_wakeup_fd grpc_global_wakeup_fd;
+
+static void remove_worker(grpc_pollset *p, grpc_pollset_worker *worker) {
+  worker->prev->next = worker->next;
+  worker->next->prev = worker->prev;
+}
+
+static int pollset_has_workers(grpc_pollset *p) {
+  return p->root_worker.next != &p->root_worker;
+}
+
+static grpc_pollset_worker *pop_front_worker(grpc_pollset *p) {
+  if (pollset_has_workers(p)) {
+    grpc_pollset_worker *w = p->root_worker.next;
+    remove_worker(p, w);
+    return w;
+  } else {
+    return NULL;
+  }
+}
+
+static void push_back_worker(grpc_pollset *p, grpc_pollset_worker *worker) {
+  worker->next = &p->root_worker;
+  worker->prev = worker->next->prev;
+  worker->prev->next = worker->next->prev = worker;
+}
+
+static void push_front_worker(grpc_pollset *p, grpc_pollset_worker *worker) {
+  worker->prev = &p->root_worker;
+  worker->next = worker->prev->next;
+  worker->prev->next = worker->next->prev = worker;
+}
+
+static void pollset_kick_ext(grpc_pollset *p,
+                             grpc_pollset_worker *specific_worker,
+                             uint32_t flags) {
+  GPR_TIMER_BEGIN("pollset_kick_ext", 0);
+
+  /* pollset->mu already held */
+  if (specific_worker != NULL) {
+    if (specific_worker == GRPC_POLLSET_KICK_BROADCAST) {
+      GPR_TIMER_BEGIN("pollset_kick_ext.broadcast", 0);
+      GPR_ASSERT((flags & GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP) == 0);
+      for (specific_worker = p->root_worker.next;
+           specific_worker != &p->root_worker;
+           specific_worker = specific_worker->next) {
+        grpc_wakeup_fd_wakeup(&specific_worker->wakeup_fd->fd);
+      }
+      p->kicked_without_pollers = 1;
+      GPR_TIMER_END("pollset_kick_ext.broadcast", 0);
+    } else if (gpr_tls_get(&g_current_thread_worker) !=
+               (intptr_t)specific_worker) {
+      GPR_TIMER_MARK("different_thread_worker", 0);
+      if ((flags & GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP) != 0) {
+        specific_worker->reevaluate_polling_on_wakeup = 1;
+      }
+      specific_worker->kicked_specifically = 1;
+      grpc_wakeup_fd_wakeup(&specific_worker->wakeup_fd->fd);
+    } else if ((flags & GRPC_POLLSET_CAN_KICK_SELF) != 0) {
+      GPR_TIMER_MARK("kick_yoself", 0);
+      if ((flags & GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP) != 0) {
+        specific_worker->reevaluate_polling_on_wakeup = 1;
+      }
+      specific_worker->kicked_specifically = 1;
+      grpc_wakeup_fd_wakeup(&specific_worker->wakeup_fd->fd);
+    }
+  } else if (gpr_tls_get(&g_current_thread_poller) != (intptr_t)p) {
+    GPR_ASSERT((flags & GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP) == 0);
+    GPR_TIMER_MARK("kick_anonymous", 0);
+    specific_worker = pop_front_worker(p);
+    if (specific_worker != NULL) {
+      if (gpr_tls_get(&g_current_thread_worker) == (intptr_t)specific_worker) {
+        /* Prefer not to kick self. Push the worker to the end of the list and
+         * pop the one from front */
+        GPR_TIMER_MARK("kick_anonymous_not_self", 0);
+        push_back_worker(p, specific_worker);
+        specific_worker = pop_front_worker(p);
+        /* If there was only one worker on the pollset, we would get the same
+         * worker we pushed (the one set on current thread local) back. If so,
+         * kick it only if GRPC_POLLSET_CAN_KICK_SELF flag is set */
+        if ((flags & GRPC_POLLSET_CAN_KICK_SELF) == 0 &&
+            gpr_tls_get(&g_current_thread_worker) ==
+                (intptr_t)specific_worker) {
+          push_back_worker(p, specific_worker);
+          specific_worker = NULL;
+        }
+      }
+      if (specific_worker != NULL) {
+        GPR_TIMER_MARK("finally_kick", 0);
+        push_back_worker(p, specific_worker);
+        grpc_wakeup_fd_wakeup(&specific_worker->wakeup_fd->fd);
+      }
+    } else {
+      GPR_TIMER_MARK("kicked_no_pollers", 0);
+      p->kicked_without_pollers = 1;
+    }
+  }
+
+  GPR_TIMER_END("pollset_kick_ext", 0);
+}
+
+static void pollset_kick(grpc_pollset *p,
+                         grpc_pollset_worker *specific_worker) {
+  pollset_kick_ext(p, specific_worker, 0);
+}
+
+/* global state management */
+
+static void pollset_global_init(void) {
+  gpr_tls_init(&g_current_thread_poller);
+  gpr_tls_init(&g_current_thread_worker);
+  grpc_wakeup_fd_init(&grpc_global_wakeup_fd);
+}
+
+static void pollset_global_shutdown(void) {
+  grpc_wakeup_fd_destroy(&grpc_global_wakeup_fd);
+  gpr_tls_destroy(&g_current_thread_poller);
+  gpr_tls_destroy(&g_current_thread_worker);
+}
+
+static void kick_poller(void) { grpc_wakeup_fd_wakeup(&grpc_global_wakeup_fd); }
+
+/* main interface */
+
+static void become_basic_pollset(grpc_pollset *pollset, grpc_fd *fd_or_null);
+
+static void pollset_init(grpc_pollset *pollset, gpr_mu **mu) {
+  gpr_mu_init(&pollset->mu);
+  *mu = &pollset->mu;
+  pollset->root_worker.next = pollset->root_worker.prev = &pollset->root_worker;
+  pollset->in_flight_cbs = 0;
+  pollset->shutting_down = 0;
+  pollset->called_shutdown = 0;
+  pollset->kicked_without_pollers = 0;
+  pollset->idle_jobs.head = pollset->idle_jobs.tail = NULL;
+  pollset->local_wakeup_cache = NULL;
+  pollset->kicked_without_pollers = 0;
+  become_basic_pollset(pollset, NULL);
+}
+
+static void pollset_destroy(grpc_pollset *pollset) {
+  GPR_ASSERT(pollset->in_flight_cbs == 0);
+  GPR_ASSERT(!pollset_has_workers(pollset));
+  GPR_ASSERT(pollset->idle_jobs.head == pollset->idle_jobs.tail);
+  pollset->vtable->destroy(pollset);
+  while (pollset->local_wakeup_cache) {
+    grpc_cached_wakeup_fd *next = pollset->local_wakeup_cache->next;
+    grpc_wakeup_fd_destroy(&pollset->local_wakeup_cache->fd);
+    gpr_free(pollset->local_wakeup_cache);
+    pollset->local_wakeup_cache = next;
+  }
+  gpr_mu_destroy(&pollset->mu);
+}
+
+static void pollset_reset(grpc_pollset *pollset) {
+  GPR_ASSERT(pollset->shutting_down);
+  GPR_ASSERT(pollset->in_flight_cbs == 0);
+  GPR_ASSERT(!pollset_has_workers(pollset));
+  GPR_ASSERT(pollset->idle_jobs.head == pollset->idle_jobs.tail);
+  pollset->vtable->destroy(pollset);
+  pollset->shutting_down = 0;
+  pollset->called_shutdown = 0;
+  pollset->kicked_without_pollers = 0;
+  become_basic_pollset(pollset, NULL);
+}
+
+static void pollset_add_fd(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                           grpc_fd *fd) {
+  gpr_mu_lock(&pollset->mu);
+  pollset->vtable->add_fd(exec_ctx, pollset, fd, 1);
+/* the following (enabled only in debug) will reacquire and then release
+   our lock - meaning that if the unlocking flag passed to add_fd above is
+   not respected, the code will deadlock (in a way that we have a chance of
+   debugging) */
+#ifndef NDEBUG
+  gpr_mu_lock(&pollset->mu);
+  gpr_mu_unlock(&pollset->mu);
+#endif
+}
+
+static void finish_shutdown(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset) {
+  GPR_ASSERT(grpc_closure_list_empty(pollset->idle_jobs));
+  pollset->vtable->finish_shutdown(pollset);
+  grpc_exec_ctx_enqueue(exec_ctx, pollset->shutdown_done, true, NULL);
+}
+
+static void pollset_work(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                         grpc_pollset_worker **worker_hdl, gpr_timespec now,
+                         gpr_timespec deadline) {
+  grpc_pollset_worker worker;
+  *worker_hdl = &worker;
+
+  /* pollset->mu already held */
+  int added_worker = 0;
+  int locked = 1;
+  int queued_work = 0;
+  int keep_polling = 0;
+  GPR_TIMER_BEGIN("pollset_work", 0);
+  /* this must happen before we (potentially) drop pollset->mu */
+  worker.next = worker.prev = NULL;
+  worker.reevaluate_polling_on_wakeup = 0;
+  if (pollset->local_wakeup_cache != NULL) {
+    worker.wakeup_fd = pollset->local_wakeup_cache;
+    pollset->local_wakeup_cache = worker.wakeup_fd->next;
+  } else {
+    worker.wakeup_fd = gpr_malloc(sizeof(*worker.wakeup_fd));
+    grpc_wakeup_fd_init(&worker.wakeup_fd->fd);
+  }
+  worker.kicked_specifically = 0;
+  /* If there's work waiting for the pollset to be idle, and the
+     pollset is idle, then do that work */
+  if (!pollset_has_workers(pollset) &&
+      !grpc_closure_list_empty(pollset->idle_jobs)) {
+    GPR_TIMER_MARK("pollset_work.idle_jobs", 0);
+    grpc_exec_ctx_enqueue_list(exec_ctx, &pollset->idle_jobs, NULL);
+    goto done;
+  }
+  /* If we're shutting down then we don't execute any extended work */
+  if (pollset->shutting_down) {
+    GPR_TIMER_MARK("pollset_work.shutting_down", 0);
+    goto done;
+  }
+  /* Give do_promote priority so we don't starve it out */
+  if (pollset->in_flight_cbs) {
+    GPR_TIMER_MARK("pollset_work.in_flight_cbs", 0);
+    gpr_mu_unlock(&pollset->mu);
+    locked = 0;
+    goto done;
+  }
+  /* Start polling, and keep doing so while we're being asked to
+     re-evaluate our pollers (this allows poll() based pollers to
+     ensure they don't miss wakeups) */
+  keep_polling = 1;
+  while (keep_polling) {
+    keep_polling = 0;
+    if (!pollset->kicked_without_pollers) {
+      if (!added_worker) {
+        push_front_worker(pollset, &worker);
+        added_worker = 1;
+        gpr_tls_set(&g_current_thread_worker, (intptr_t)&worker);
+      }
+      gpr_tls_set(&g_current_thread_poller, (intptr_t)pollset);
+      GPR_TIMER_BEGIN("maybe_work_and_unlock", 0);
+      pollset->vtable->maybe_work_and_unlock(exec_ctx, pollset, &worker,
+                                             deadline, now);
+      GPR_TIMER_END("maybe_work_and_unlock", 0);
+      locked = 0;
+      gpr_tls_set(&g_current_thread_poller, 0);
+    } else {
+      GPR_TIMER_MARK("pollset_work.kicked_without_pollers", 0);
+      pollset->kicked_without_pollers = 0;
+    }
+  /* Finished execution - start cleaning up.
+     Note that we may arrive here from outside the enclosing while() loop.
+     In that case we won't loop though as we haven't added worker to the
+     worker list, which means nobody could ask us to re-evaluate polling). */
+  done:
+    if (!locked) {
+      queued_work |= grpc_exec_ctx_flush(exec_ctx);
+      gpr_mu_lock(&pollset->mu);
+      locked = 1;
+    }
+    /* If we're forced to re-evaluate polling (via pollset_kick with
+       GRPC_POLLSET_REEVALUATE_POLLING_ON_WAKEUP) then we land here and force
+       a loop */
+    if (worker.reevaluate_polling_on_wakeup) {
+      worker.reevaluate_polling_on_wakeup = 0;
+      pollset->kicked_without_pollers = 0;
+      if (queued_work || worker.kicked_specifically) {
+        /* If there's queued work on the list, then set the deadline to be
+           immediate so we get back out of the polling loop quickly */
+        deadline = gpr_inf_past(GPR_CLOCK_MONOTONIC);
+      }
+      keep_polling = 1;
+    }
+  }
+  if (added_worker) {
+    remove_worker(pollset, &worker);
+    gpr_tls_set(&g_current_thread_worker, 0);
+  }
+  /* release wakeup fd to the local pool */
+  worker.wakeup_fd->next = pollset->local_wakeup_cache;
+  pollset->local_wakeup_cache = worker.wakeup_fd;
+  /* check shutdown conditions */
+  if (pollset->shutting_down) {
+    if (pollset_has_workers(pollset)) {
+      pollset_kick(pollset, NULL);
+    } else if (!pollset->called_shutdown && pollset->in_flight_cbs == 0) {
+      pollset->called_shutdown = 1;
+      gpr_mu_unlock(&pollset->mu);
+      finish_shutdown(exec_ctx, pollset);
+      grpc_exec_ctx_flush(exec_ctx);
+      /* Continuing to access pollset here is safe -- it is the caller's
+       * responsibility to not destroy when it has outstanding calls to
+       * pollset_work.
+       * TODO(dklempner): Can we refactor the shutdown logic to avoid this? */
+      gpr_mu_lock(&pollset->mu);
+    } else if (!grpc_closure_list_empty(pollset->idle_jobs)) {
+      grpc_exec_ctx_enqueue_list(exec_ctx, &pollset->idle_jobs, NULL);
+      gpr_mu_unlock(&pollset->mu);
+      grpc_exec_ctx_flush(exec_ctx);
+      gpr_mu_lock(&pollset->mu);
+    }
+  }
+  *worker_hdl = NULL;
+  GPR_TIMER_END("pollset_work", 0);
+}
+
+static void pollset_shutdown(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                             grpc_closure *closure) {
+  GPR_ASSERT(!pollset->shutting_down);
+  pollset->shutting_down = 1;
+  pollset->shutdown_done = closure;
+  pollset_kick(pollset, GRPC_POLLSET_KICK_BROADCAST);
+  if (!pollset_has_workers(pollset)) {
+    grpc_exec_ctx_enqueue_list(exec_ctx, &pollset->idle_jobs, NULL);
+  }
+  if (!pollset->called_shutdown && pollset->in_flight_cbs == 0 &&
+      !pollset_has_workers(pollset)) {
+    pollset->called_shutdown = 1;
+    finish_shutdown(exec_ctx, pollset);
+  }
+}
+
+static int poll_deadline_to_millis_timeout(gpr_timespec deadline,
+                                           gpr_timespec now) {
+  gpr_timespec timeout;
+  static const int64_t max_spin_polling_us = 10;
+  if (gpr_time_cmp(deadline, gpr_inf_future(deadline.clock_type)) == 0) {
+    return -1;
+  }
+  if (gpr_time_cmp(deadline, gpr_time_add(now, gpr_time_from_micros(
+                                                   max_spin_polling_us,
+                                                   GPR_TIMESPAN))) <= 0) {
+    return 0;
+  }
+  timeout = gpr_time_sub(deadline, now);
+  return gpr_time_to_millis(gpr_time_add(
+      timeout, gpr_time_from_nanos(GPR_NS_PER_MS - 1, GPR_TIMESPAN)));
+}
+
+/*
+ * basic_pollset - a vtable that provides polling for zero or one file
+ *                 descriptor via poll()
+ */
+
+typedef struct grpc_unary_promote_args {
+  const grpc_pollset_vtable *original_vtable;
+  grpc_pollset *pollset;
+  grpc_fd *fd;
+  grpc_closure promotion_closure;
+} grpc_unary_promote_args;
+
+static void basic_do_promote(grpc_exec_ctx *exec_ctx, void *args,
+                             bool success) {
+  grpc_unary_promote_args *up_args = args;
+  const grpc_pollset_vtable *original_vtable = up_args->original_vtable;
+  grpc_pollset *pollset = up_args->pollset;
+  grpc_fd *fd = up_args->fd;
+
+  /*
+   * This is quite tricky. There are a number of cases to keep in mind here:
+   * 1. fd may have been orphaned
+   * 2. The pollset may no longer be a unary poller (and we can't let case #1
+   * leak to other pollset types!)
+   * 3. pollset's fd (which may have changed) may have been orphaned
+   * 4. The pollset may be shutting down.
+   */
+
+  gpr_mu_lock(&pollset->mu);
+  /* First we need to ensure that nobody is polling concurrently */
+  GPR_ASSERT(!pollset_has_workers(pollset));
+
+  gpr_free(up_args);
+  /* At this point the pollset may no longer be a unary poller. In that case
+   * we should just call the right add function and be done. */
+  /* TODO(klempner): If we're not careful this could cause infinite recursion.
+   * That's not a problem for now because empty_pollset has a trivial poller
+   * and we don't have any mechanism to unbecome multipoller. */
+  pollset->in_flight_cbs--;
+  if (pollset->shutting_down) {
+    /* We don't care about this pollset anymore. */
+    if (pollset->in_flight_cbs == 0 && !pollset->called_shutdown) {
+      pollset->called_shutdown = 1;
+      finish_shutdown(exec_ctx, pollset);
+    }
+  } else if (fd_is_orphaned(fd)) {
+    /* Don't try to add it to anything, we'll drop our ref on it below */
+  } else if (pollset->vtable != original_vtable) {
+    pollset->vtable->add_fd(exec_ctx, pollset, fd, 0);
+  } else if (fd != pollset->data.ptr) {
+    grpc_fd *fds[2];
+    fds[0] = pollset->data.ptr;
+    fds[1] = fd;
+
+    if (fds[0] && !fd_is_orphaned(fds[0])) {
+      platform_become_multipoller(exec_ctx, pollset, fds, GPR_ARRAY_SIZE(fds));
+      GRPC_FD_UNREF(fds[0], "basicpoll");
+    } else {
+      /* old fd is orphaned and we haven't cleaned it up until now, so remain a
+       * unary poller */
+      /* Note that it is possible that fds[1] is also orphaned at this point.
+       * That's okay, we'll correct it at the next add or poll. */
+      if (fds[0]) GRPC_FD_UNREF(fds[0], "basicpoll");
+      pollset->data.ptr = fd;
+      GRPC_FD_REF(fd, "basicpoll");
+    }
+  }
+
+  gpr_mu_unlock(&pollset->mu);
+
+  /* Matching ref in basic_pollset_add_fd */
+  GRPC_FD_UNREF(fd, "basicpoll_add");
+}
+
+static void basic_pollset_add_fd(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                                 grpc_fd *fd, int and_unlock_pollset) {
+  grpc_unary_promote_args *up_args;
+  GPR_ASSERT(fd);
+  if (fd == pollset->data.ptr) goto exit;
+
+  if (!pollset_has_workers(pollset)) {
+    /* Fast path -- no in flight cbs */
+    /* TODO(klempner): Comment this out and fix any test failures or establish
+     * they are due to timing issues */
+    grpc_fd *fds[2];
+    fds[0] = pollset->data.ptr;
+    fds[1] = fd;
+
+    if (fds[0] == NULL) {
+      pollset->data.ptr = fd;
+      GRPC_FD_REF(fd, "basicpoll");
+    } else if (!fd_is_orphaned(fds[0])) {
+      platform_become_multipoller(exec_ctx, pollset, fds, GPR_ARRAY_SIZE(fds));
+      GRPC_FD_UNREF(fds[0], "basicpoll");
+    } else {
+      /* old fd is orphaned and we haven't cleaned it up until now, so remain a
+       * unary poller */
+      GRPC_FD_UNREF(fds[0], "basicpoll");
+      pollset->data.ptr = fd;
+      GRPC_FD_REF(fd, "basicpoll");
+    }
+    goto exit;
+  }
+
+  /* Now we need to promote. This needs to happen when we're not polling. Since
+   * this may be called from poll, the wait needs to happen asynchronously. */
+  GRPC_FD_REF(fd, "basicpoll_add");
+  pollset->in_flight_cbs++;
+  up_args = gpr_malloc(sizeof(*up_args));
+  up_args->fd = fd;
+  up_args->original_vtable = pollset->vtable;
+  up_args->pollset = pollset;
+  up_args->promotion_closure.cb = basic_do_promote;
+  up_args->promotion_closure.cb_arg = up_args;
+
+  grpc_closure_list_add(&pollset->idle_jobs, &up_args->promotion_closure, 1);
+  pollset_kick(pollset, GRPC_POLLSET_KICK_BROADCAST);
+
+exit:
+  if (and_unlock_pollset) {
+    gpr_mu_unlock(&pollset->mu);
+  }
+}
+
+static void basic_pollset_maybe_work_and_unlock(grpc_exec_ctx *exec_ctx,
+                                                grpc_pollset *pollset,
+                                                grpc_pollset_worker *worker,
+                                                gpr_timespec deadline,
+                                                gpr_timespec now) {
+#define POLLOUT_CHECK (POLLOUT | POLLHUP | POLLERR)
+#define POLLIN_CHECK (POLLIN | POLLHUP | POLLERR)
+
+  struct pollfd pfd[3];
+  grpc_fd *fd;
+  grpc_fd_watcher fd_watcher;
+  int timeout;
+  int r;
+  nfds_t nfds;
+
+  fd = pollset->data.ptr;
+  if (fd && fd_is_orphaned(fd)) {
+    GRPC_FD_UNREF(fd, "basicpoll");
+    fd = pollset->data.ptr = NULL;
+  }
+  timeout = poll_deadline_to_millis_timeout(deadline, now);
+  pfd[0].fd = GRPC_WAKEUP_FD_GET_READ_FD(&grpc_global_wakeup_fd);
+  pfd[0].events = POLLIN;
+  pfd[0].revents = 0;
+  pfd[1].fd = GRPC_WAKEUP_FD_GET_READ_FD(&worker->wakeup_fd->fd);
+  pfd[1].events = POLLIN;
+  pfd[1].revents = 0;
+  nfds = 2;
+  if (fd) {
+    pfd[2].fd = fd->fd;
+    pfd[2].revents = 0;
+    GRPC_FD_REF(fd, "basicpoll_begin");
+    gpr_mu_unlock(&pollset->mu);
+    pfd[2].events =
+        (short)fd_begin_poll(fd, pollset, worker, POLLIN, POLLOUT, &fd_watcher);
+    if (pfd[2].events != 0) {
+      nfds++;
+    }
+  } else {
+    gpr_mu_unlock(&pollset->mu);
+  }
+
+  /* TODO(vpai): Consider first doing a 0 timeout poll here to avoid
+     even going into the blocking annotation if possible */
+  /* poll fd count (argument 2) is shortened by one if we have no events
+     to poll on - such that it only includes the kicker */
+  GPR_TIMER_BEGIN("poll", 0);
+  GRPC_SCHEDULING_START_BLOCKING_REGION;
+  r = grpc_poll_function(pfd, nfds, timeout);
+  GRPC_SCHEDULING_END_BLOCKING_REGION;
+  GPR_TIMER_END("poll", 0);
+
+  if (r < 0) {
+    if (errno != EINTR) {
+      gpr_log(GPR_ERROR, "poll() failed: %s", strerror(errno));
+    }
+    if (fd) {
+      fd_end_poll(exec_ctx, &fd_watcher, 0, 0, NULL);
+    }
+  } else if (r == 0) {
+    if (fd) {
+      fd_end_poll(exec_ctx, &fd_watcher, 0, 0, NULL);
+    }
+  } else {
+    if (pfd[0].revents & POLLIN_CHECK) {
+      grpc_wakeup_fd_consume_wakeup(&grpc_global_wakeup_fd);
+    }
+    if (pfd[1].revents & POLLIN_CHECK) {
+      grpc_wakeup_fd_consume_wakeup(&worker->wakeup_fd->fd);
+    }
+    if (nfds > 2) {
+      fd_end_poll(exec_ctx, &fd_watcher, pfd[2].revents & POLLIN_CHECK,
+                  pfd[2].revents & POLLOUT_CHECK, pollset);
+    } else if (fd) {
+      fd_end_poll(exec_ctx, &fd_watcher, 0, 0, NULL);
+    }
+  }
+
+  if (fd) {
+    GRPC_FD_UNREF(fd, "basicpoll_begin");
+  }
+}
+
+static void basic_pollset_destroy(grpc_pollset *pollset) {
+  if (pollset->data.ptr != NULL) {
+    GRPC_FD_UNREF(pollset->data.ptr, "basicpoll");
+    pollset->data.ptr = NULL;
+  }
+}
+
+static const grpc_pollset_vtable basic_pollset = {
+    basic_pollset_add_fd, basic_pollset_maybe_work_and_unlock,
+    basic_pollset_destroy, basic_pollset_destroy};
+
+static void become_basic_pollset(grpc_pollset *pollset, grpc_fd *fd_or_null) {
+  pollset->vtable = &basic_pollset;
+  pollset->data.ptr = fd_or_null;
+  if (fd_or_null != NULL) {
+    GRPC_FD_REF(fd_or_null, "basicpoll");
+  }
+}
+
+/*******************************************************************************
+ * pollset_multipoller_with_poll_posix.c
+ */
+
+#ifndef GPR_LINUX_MULTIPOLL_WITH_EPOLL
+
+typedef struct {
+  /* all polled fds */
+  size_t fd_count;
+  size_t fd_capacity;
+  grpc_fd **fds;
+  /* fds that have been removed from the pollset explicitly */
+  size_t del_count;
+  size_t del_capacity;
+  grpc_fd **dels;
+} poll_hdr;
+
+static void multipoll_with_poll_pollset_add_fd(grpc_exec_ctx *exec_ctx,
+                                               grpc_pollset *pollset,
+                                               grpc_fd *fd,
+                                               int and_unlock_pollset) {
+  size_t i;
+  poll_hdr *h = pollset->data.ptr;
+  /* TODO(ctiller): this is O(num_fds^2); maybe switch to a hash set here */
+  for (i = 0; i < h->fd_count; i++) {
+    if (h->fds[i] == fd) goto exit;
+  }
+  if (h->fd_count == h->fd_capacity) {
+    h->fd_capacity = GPR_MAX(h->fd_capacity + 8, h->fd_count * 3 / 2);
+    h->fds = gpr_realloc(h->fds, sizeof(grpc_fd *) * h->fd_capacity);
+  }
+  h->fds[h->fd_count++] = fd;
+  GRPC_FD_REF(fd, "multipoller");
+exit:
+  if (and_unlock_pollset) {
+    gpr_mu_unlock(&pollset->mu);
+  }
+}
+
+static void multipoll_with_poll_pollset_maybe_work_and_unlock(
+    grpc_exec_ctx *exec_ctx, grpc_pollset *pollset, grpc_pollset_worker *worker,
+    gpr_timespec deadline, gpr_timespec now) {
+#define POLLOUT_CHECK (POLLOUT | POLLHUP | POLLERR)
+#define POLLIN_CHECK (POLLIN | POLLHUP | POLLERR)
+
+  int timeout;
+  int r;
+  size_t i, j, fd_count;
+  nfds_t pfd_count;
+  poll_hdr *h;
+  /* TODO(ctiller): inline some elements to avoid an allocation */
+  grpc_fd_watcher *watchers;
+  struct pollfd *pfds;
+
+  h = pollset->data.ptr;
+  timeout = poll_deadline_to_millis_timeout(deadline, now);
+  /* TODO(ctiller): perform just one malloc here if we exceed the inline case */
+  pfds = gpr_malloc(sizeof(*pfds) * (h->fd_count + 2));
+  watchers = gpr_malloc(sizeof(*watchers) * (h->fd_count + 2));
+  fd_count = 0;
+  pfd_count = 2;
+  pfds[0].fd = GRPC_WAKEUP_FD_GET_READ_FD(&grpc_global_wakeup_fd);
+  pfds[0].events = POLLIN;
+  pfds[0].revents = 0;
+  pfds[1].fd = GRPC_WAKEUP_FD_GET_READ_FD(&worker->wakeup_fd->fd);
+  pfds[1].events = POLLIN;
+  pfds[1].revents = 0;
+  for (i = 0; i < h->fd_count; i++) {
+    int remove = fd_is_orphaned(h->fds[i]);
+    for (j = 0; !remove && j < h->del_count; j++) {
+      if (h->fds[i] == h->dels[j]) remove = 1;
+    }
+    if (remove) {
+      GRPC_FD_UNREF(h->fds[i], "multipoller");
+    } else {
+      h->fds[fd_count++] = h->fds[i];
+      watchers[pfd_count].fd = h->fds[i];
+      GRPC_FD_REF(watchers[pfd_count].fd, "multipoller_start");
+      pfds[pfd_count].fd = h->fds[i]->fd;
+      pfds[pfd_count].revents = 0;
+      pfd_count++;
+    }
+  }
+  for (j = 0; j < h->del_count; j++) {
+    GRPC_FD_UNREF(h->dels[j], "multipoller_del");
+  }
+  h->del_count = 0;
+  h->fd_count = fd_count;
+  gpr_mu_unlock(&pollset->mu);
+
+  for (i = 2; i < pfd_count; i++) {
+    grpc_fd *fd = watchers[i].fd;
+    pfds[i].events = (short)fd_begin_poll(fd, pollset, worker, POLLIN, POLLOUT,
+                                          &watchers[i]);
+    GRPC_FD_UNREF(fd, "multipoller_start");
+  }
+
+  /* TODO(vpai): Consider first doing a 0 timeout poll here to avoid
+     even going into the blocking annotation if possible */
+  GRPC_SCHEDULING_START_BLOCKING_REGION;
+  r = grpc_poll_function(pfds, pfd_count, timeout);
+  GRPC_SCHEDULING_END_BLOCKING_REGION;
+
+  if (r < 0) {
+    if (errno != EINTR) {
+      gpr_log(GPR_ERROR, "poll() failed: %s", strerror(errno));
+    }
+    for (i = 2; i < pfd_count; i++) {
+      fd_end_poll(exec_ctx, &watchers[i], 0, 0, NULL);
+    }
+  } else if (r == 0) {
+    for (i = 2; i < pfd_count; i++) {
+      fd_end_poll(exec_ctx, &watchers[i], 0, 0, NULL);
+    }
+  } else {
+    if (pfds[0].revents & POLLIN_CHECK) {
+      grpc_wakeup_fd_consume_wakeup(&grpc_global_wakeup_fd);
+    }
+    if (pfds[1].revents & POLLIN_CHECK) {
+      grpc_wakeup_fd_consume_wakeup(&worker->wakeup_fd->fd);
+    }
+    for (i = 2; i < pfd_count; i++) {
+      if (watchers[i].fd == NULL) {
+        fd_end_poll(exec_ctx, &watchers[i], 0, 0, NULL);
+        continue;
+      }
+      fd_end_poll(exec_ctx, &watchers[i], pfds[i].revents & POLLIN_CHECK,
+                  pfds[i].revents & POLLOUT_CHECK, pollset);
+    }
+  }
+
+  gpr_free(pfds);
+  gpr_free(watchers);
+}
+
+static void multipoll_with_poll_pollset_finish_shutdown(grpc_pollset *pollset) {
+  size_t i;
+  poll_hdr *h = pollset->data.ptr;
+  for (i = 0; i < h->fd_count; i++) {
+    GRPC_FD_UNREF(h->fds[i], "multipoller");
+  }
+  for (i = 0; i < h->del_count; i++) {
+    GRPC_FD_UNREF(h->dels[i], "multipoller_del");
+  }
+  h->fd_count = 0;
+  h->del_count = 0;
+}
+
+static void multipoll_with_poll_pollset_destroy(grpc_pollset *pollset) {
+  poll_hdr *h = pollset->data.ptr;
+  multipoll_with_poll_pollset_finish_shutdown(pollset);
+  gpr_free(h->fds);
+  gpr_free(h->dels);
+  gpr_free(h);
+}
+
+static const grpc_pollset_vtable multipoll_with_poll_pollset = {
+    multipoll_with_poll_pollset_add_fd,
+    multipoll_with_poll_pollset_maybe_work_and_unlock,
+    multipoll_with_poll_pollset_finish_shutdown,
+    multipoll_with_poll_pollset_destroy};
+
+static void poll_become_multipoller(grpc_exec_ctx *exec_ctx,
+                                    grpc_pollset *pollset, grpc_fd **fds,
+                                    size_t nfds) {
+  size_t i;
+  poll_hdr *h = gpr_malloc(sizeof(poll_hdr));
+  pollset->vtable = &multipoll_with_poll_pollset;
+  pollset->data.ptr = h;
+  h->fd_count = nfds;
+  h->fd_capacity = nfds;
+  h->fds = gpr_malloc(nfds * sizeof(grpc_fd *));
+  h->del_count = 0;
+  h->del_capacity = 0;
+  h->dels = NULL;
+  for (i = 0; i < nfds; i++) {
+    h->fds[i] = fds[i];
+    GRPC_FD_REF(fds[i], "multipoller");
+  }
+}
+
+#endif /* !GPR_LINUX_MULTIPOLL_WITH_EPOLL */
+
+/*******************************************************************************
+ * pollset_multipoller_with_epoll_posix.c
+ */
+
+#ifdef GPR_LINUX_MULTIPOLL_WITH_EPOLL
+
+#include <errno.h>
+#include <poll.h>
+#include <string.h>
+#include <sys/epoll.h>
+#include <unistd.h>
+
+#include <grpc/support/alloc.h>
+#include <grpc/support/log.h>
+#include <grpc/support/useful.h>
+
+#include "src/core/lib/iomgr/ev_posix.h"
+#include "src/core/lib/profiling/timers.h"
+#include "src/core/lib/support/block_annotate.h"
+
+static void set_ready(grpc_exec_ctx *exec_ctx, grpc_fd *fd, grpc_closure **st,
+                      grpc_pollset *read_notifier_pollset) {
+  /* only one set_ready can be active at once (but there may be a racing
+     notify_on) */
+  gpr_mu_lock(&fd->mu);
+  set_ready_locked(exec_ctx, fd, st);
+
+  /* A non-NULL read_notifier_pollset means that the fd is readable. */
+  if (read_notifier_pollset != NULL) {
+    /* Note: Since the fd might be a part of multiple pollsets, this might be
+     * called multiple times (for each time the fd becomes readable) and it is
+     * okay to set the fd's read-notifier pollset to anyone of these pollsets */
+    set_read_notifier_pollset_locked(exec_ctx, fd, read_notifier_pollset);
+  }
+
+  gpr_mu_unlock(&fd->mu);
+}
+
+static void fd_become_readable(grpc_exec_ctx *exec_ctx, grpc_fd *fd,
+                               grpc_pollset *notifier_pollset) {
+  set_ready(exec_ctx, fd, &fd->read_closure, notifier_pollset);
+}
+
+static void fd_become_writable(grpc_exec_ctx *exec_ctx, grpc_fd *fd) {
+  set_ready(exec_ctx, fd, &fd->write_closure, NULL);
+}
+
+struct epoll_fd_list {
+  int *epoll_fds;
+  size_t count;
+  size_t capacity;
+};
+
+static struct epoll_fd_list epoll_fd_global_list;
+static gpr_once init_epoll_fd_list_mu = GPR_ONCE_INIT;
+static gpr_mu epoll_fd_list_mu;
+
+static void init_mu(void) { gpr_mu_init(&epoll_fd_list_mu); }
+
+static void add_epoll_fd_to_global_list(int epoll_fd) {
+  gpr_once_init(&init_epoll_fd_list_mu, init_mu);
+
+  gpr_mu_lock(&epoll_fd_list_mu);
+  if (epoll_fd_global_list.count == epoll_fd_global_list.capacity) {
+    epoll_fd_global_list.capacity =
+        GPR_MAX((size_t)8, epoll_fd_global_list.capacity * 2);
+    epoll_fd_global_list.epoll_fds =
+        gpr_realloc(epoll_fd_global_list.epoll_fds,
+                    epoll_fd_global_list.capacity * sizeof(int));
+  }
+  epoll_fd_global_list.epoll_fds[epoll_fd_global_list.count++] = epoll_fd;
+  gpr_mu_unlock(&epoll_fd_list_mu);
+}
+
+static void remove_epoll_fd_from_global_list(int epoll_fd) {
+  gpr_mu_lock(&epoll_fd_list_mu);
+  GPR_ASSERT(epoll_fd_global_list.count > 0);
+  for (size_t i = 0; i < epoll_fd_global_list.count; i++) {
+    if (epoll_fd == epoll_fd_global_list.epoll_fds[i]) {
+      epoll_fd_global_list.epoll_fds[i] =
+          epoll_fd_global_list.epoll_fds[--(epoll_fd_global_list.count)];
+      break;
+    }
+  }
+  gpr_mu_unlock(&epoll_fd_list_mu);
+}
+
+static void remove_fd_from_all_epoll_sets(int fd) {
+  int err;
+  gpr_once_init(&init_epoll_fd_list_mu, init_mu);
+  gpr_mu_lock(&epoll_fd_list_mu);
+  if (epoll_fd_global_list.count == 0) {
+    gpr_mu_unlock(&epoll_fd_list_mu);
+    return;
+  }
+  for (size_t i = 0; i < epoll_fd_global_list.count; i++) {
+    err = epoll_ctl(epoll_fd_global_list.epoll_fds[i], EPOLL_CTL_DEL, fd, NULL);
+    if (err < 0 && errno != ENOENT) {
+      gpr_log(GPR_ERROR, "epoll_ctl del for %d failed: %s", fd,
+              strerror(errno));
+    }
+  }
+  gpr_mu_unlock(&epoll_fd_list_mu);
+}
+
+typedef struct {
+  grpc_pollset *pollset;
+  grpc_fd *fd;
+  grpc_closure closure;
+} delayed_add;
+
+typedef struct { int epoll_fd; } epoll_hdr;
+
+static void finally_add_fd(grpc_exec_ctx *exec_ctx, grpc_pollset *pollset,
+                           grpc_fd *fd) {
+  epoll_hdr *h = pollset->data.ptr;
+  struct epoll_event ev;
+  int err;
+  grpc_fd_watcher watcher;
+
+  /* We pretend to be polling whilst adding an fd to keep the fd from being
+     closed during the add. This may result in a spurious wakeup being assigned
+     to this pollset whilst adding, but that should be benign. */
+  GPR_ASSERT(fd_begin_poll(fd, pollset, NULL, 0, 0, &watcher) == 0);
+  if (watcher.fd != NULL) {
+    ev.events = (uint32_t)(EPOLLIN | EPOLLOUT | EPOLLET);
+    ev.data.ptr = fd;
+    err = epoll_ctl(h->epoll_fd, EPOLL_CTL_ADD, fd->fd, &ev);
+    if (err < 0) {
+      /* FDs may be added to a pollset multiple times, so EEXIST is normal. */
+      if (errno != EEXIST) {
+        gpr_log(GPR_ERROR, "epoll_ctl add for %d failed: %s", fd->fd,
+                strerror(errno));
+      }
+    }
+  }
+  fd_end_poll(exec_ctx, &watcher, 0, 0, NULL);
+}
+
+static void perform_delayed_add(grpc_exec_ctx *exec_ctx, void *arg,
+                                bool iomgr_status) {
+  delayed_add *da = arg;
+
+  if (!fd_is_orphaned(da->fd)) {
+    finally_add_fd(exec_ctx, da->pollset, da->fd);
+  }
+
+  gpr_mu_lock(&da->pollset->mu);
+  da->pollset->in_flight_cbs--;
+  if (da->pollset->shutting_down) {
+    /* We don't care about this pollset anymore. */
+    if (da->pollset->in_flight_cbs == 0 && !da->pollset->called_shutdown) {
+      da->pollset->called_shutdown = 1;
+      grpc_exec_ctx_enqueue(exec_ctx, da->pollset->shutdown_done, true, NULL);
+    }
+  }
+  gpr_mu_unlock(&da->pollset->mu);
+
+  GRPC_FD_UNREF(da->fd, "delayed_add");
+
+  gpr_free(da);
+}
+
+static void multipoll_with_epoll_pollset_add_fd(grpc_exec_ctx *exec_ctx,
+                                                grpc_pollset *pollset,
+                                                grpc_fd *fd,
+                                                int and_unlock_pollset) {
+  if (and_unlock_pollset) {
+    gpr_mu_unlock(&pollset->mu);
+    finally_add_fd(exec_ctx, pollset, fd);
+  } else {
+    delayed_add *da = gpr_malloc(sizeof(*da));
+    da->pollset = pollset;
+    da->fd = fd;
+    GRPC_FD_REF(fd, "delayed_add");
+    grpc_closure_init(&da->closure, perform_delayed_add, da);
+    pollset->in_flight_cbs++;
+    grpc_exec_ctx_enqueue(exec_ctx, &da->closure, true, NULL);
+  }
+}
+
+/* TODO(klempner): We probably want to turn this down a bit */
+#define GRPC_EPOLL_MAX_EVENTS 1000
+
+static void multipoll_with_epoll_pollset_maybe_work_and_unlock(
+    grpc_exec_ctx *exec_ctx, grpc_pollset *pollset, grpc_pollset_worker *worker,
+    gpr_timespec deadline, gpr_timespec now) {
+  struct epoll_event ep_ev[GRPC_EPOLL_MAX_EVENTS];
+  int ep_rv;
+  int poll_rv;
+  epoll_hdr *h = pollset->data.ptr;
+  int timeout_ms;
+  struct pollfd pfds[2];
+
+  /* If you want to ignore epoll's ability to sanely handle parallel pollers,
+   * for a more apples-to-apples performance comparison with poll, add a
+   * if (pollset->counter != 0) { return 0; }
+   * here.
+   */
+
+  gpr_mu_unlock(&pollset->mu);
+
+  timeout_ms = poll_deadline_to_millis_timeout(deadline, now);
+
+  pfds[0].fd = GRPC_WAKEUP_FD_GET_READ_FD(&worker->wakeup_fd->fd);
+  pfds[0].events = POLLIN;
+  pfds[0].revents = 0;
+  pfds[1].fd = h->epoll_fd;
+  pfds[1].events = POLLIN;
+  pfds[1].revents = 0;
+
+  /* TODO(vpai): Consider first doing a 0 timeout poll here to avoid
+     even going into the blocking annotation if possible */
+  GPR_TIMER_BEGIN("poll", 0);
+  GRPC_SCHEDULING_START_BLOCKING_REGION;
+  poll_rv = grpc_poll_function(pfds, 2, timeout_ms);
+  GRPC_SCHEDULING_END_BLOCKING_REGION;
+  GPR_TIMER_END("poll", 0);
+
+  if (poll_rv < 0) {
+    if (errno != EINTR) {
+      gpr_log(GPR_ERROR, "poll() failed: %s", strerror(errno));
+    }
+  } else if (poll_rv == 0) {
+    /* do nothing */
+  } else {
+    if (pfds[0].revents) {
+      grpc_wakeup_fd_consume_wakeup(&worker->wakeup_fd->fd);
+    }
+    if (pfds[1].revents) {
+      do {
+        /* The following epoll_wait never blocks; it has a timeout of 0 */
+        ep_rv = epoll_wait(h->epoll_fd, ep_ev, GRPC_EPOLL_MAX_EVENTS, 0);
+        if (ep_rv < 0) {
+          if (errno != EINTR) {
+            gpr_log(GPR_ERROR, "epoll_wait() failed: %s", strerror(errno));
+          }
+        } else {
+          int i;
+          for (i = 0; i < ep_rv; ++i) {
+            grpc_fd *fd = ep_ev[i].data.ptr;
+            /* TODO(klempner): We might want to consider making err and pri
+             * separate events */
+            int cancel = ep_ev[i].events & (EPOLLERR | EPOLLHUP);
+            int read_ev = ep_ev[i].events & (EPOLLIN | EPOLLPRI);
+            int write_ev = ep_ev[i].events & EPOLLOUT;
+            if (fd == NULL) {
+              grpc_wakeup_fd_consume_wakeup(&grpc_global_wakeup_fd);
+            } else {
+              if (read_ev || cancel) {
+                fd_become_readable(exec_ctx, fd, pollset);
+              }
+              if (write_ev || cancel) {
+                fd_become_writable(exec_ctx, fd);
+              }
+            }
+          }
+        }
+      } while (ep_rv == GRPC_EPOLL_MAX_EVENTS);
+    }
+  }
+}
+
+static void multipoll_with_epoll_pollset_finish_shutdown(
+    grpc_pollset *pollset) {}
+
+static void multipoll_with_epoll_pollset_destroy(grpc_pollset *pollset) {
+  epoll_hdr *h = pollset->data.ptr;
+  close(h->epoll_fd);
+  remove_epoll_fd_from_global_list(h->epoll_fd);
+  gpr_free(h);
+}
+
+static const grpc_pollset_vtable multipoll_with_epoll_pollset = {
+    multipoll_with_epoll_pollset_add_fd,
+    multipoll_with_epoll_pollset_maybe_work_and_unlock,
+    multipoll_with_epoll_pollset_finish_shutdown,
+    multipoll_with_epoll_pollset_destroy};
+
+static void epoll_become_multipoller(grpc_exec_ctx *exec_ctx,
+                                     grpc_pollset *pollset, grpc_fd **fds,
+                                     size_t nfds) {
+  size_t i;
+  epoll_hdr *h = gpr_malloc(sizeof(epoll_hdr));
+  struct epoll_event ev;
+  int err;
+
+  pollset->vtable = &multipoll_with_epoll_pollset;
+  pollset->data.ptr = h;
+  h->epoll_fd = epoll_create1(EPOLL_CLOEXEC);
+  if (h->epoll_fd < 0) {
+    /* TODO(klempner): Fall back to poll here, especially on ENOSYS */
+    gpr_log(GPR_ERROR, "epoll_create1 failed: %s", strerror(errno));
+    abort();
+  }
+  add_epoll_fd_to_global_list(h->epoll_fd);
+
+  ev.events = (uint32_t)(EPOLLIN | EPOLLET);
+  ev.data.ptr = NULL;
+  err = epoll_ctl(h->epoll_fd, EPOLL_CTL_ADD,
+                  GRPC_WAKEUP_FD_GET_READ_FD(&grpc_global_wakeup_fd), &ev);
+  if (err < 0) {
+    gpr_log(GPR_ERROR, "epoll_ctl add for %d failed: %s",
+            GRPC_WAKEUP_FD_GET_READ_FD(&grpc_global_wakeup_fd),
+            strerror(errno));
+  }
+
+  for (i = 0; i < nfds; i++) {
+    multipoll_with_epoll_pollset_add_fd(exec_ctx, pollset, fds[i], 0);
+  }
+}
+
+#else /* GPR_LINUX_MULTIPOLL_WITH_EPOLL */
+
+static void remove_fd_from_all_epoll_sets(int fd) {}
+
+#endif /* GPR_LINUX_MULTIPOLL_WITH_EPOLL */
+
+/*******************************************************************************
+ * pollset_set_posix.c
+ */
+
+static grpc_pollset_set *pollset_set_create(void) {
+  grpc_pollset_set *pollset_set = gpr_malloc(sizeof(*pollset_set));
+  memset(pollset_set, 0, sizeof(*pollset_set));
+  gpr_mu_init(&pollset_set->mu);
+  return pollset_set;
+}
+
+static void pollset_set_destroy(grpc_pollset_set *pollset_set) {
+  size_t i;
+  gpr_mu_destroy(&pollset_set->mu);
+  for (i = 0; i < pollset_set->fd_count; i++) {
+    GRPC_FD_UNREF(pollset_set->fds[i], "pollset_set");
+  }
+  gpr_free(pollset_set->pollsets);
+  gpr_free(pollset_set->pollset_sets);
+  gpr_free(pollset_set->fds);
+  gpr_free(pollset_set);
+}
+
+static void pollset_set_add_pollset(grpc_exec_ctx *exec_ctx,
+                                    grpc_pollset_set *pollset_set,
+                                    grpc_pollset *pollset) {
+  size_t i, j;
+  gpr_mu_lock(&pollset_set->mu);
+  if (pollset_set->pollset_count == pollset_set->pollset_capacity) {
+    pollset_set->pollset_capacity =
+        GPR_MAX(8, 2 * pollset_set->pollset_capacity);
+    pollset_set->pollsets =
+        gpr_realloc(pollset_set->pollsets, pollset_set->pollset_capacity *
+                                               sizeof(*pollset_set->pollsets));
+  }
+  pollset_set->pollsets[pollset_set->pollset_count++] = pollset;
+  for (i = 0, j = 0; i < pollset_set->fd_count; i++) {
+    if (fd_is_orphaned(pollset_set->fds[i])) {
+      GRPC_FD_UNREF(pollset_set->fds[i], "pollset_set");
+    } else {
+      pollset_add_fd(exec_ctx, pollset, pollset_set->fds[i]);
+      pollset_set->fds[j++] = pollset_set->fds[i];
+    }
+  }
+  pollset_set->fd_count = j;
+  gpr_mu_unlock(&pollset_set->mu);
+}
+
+static void pollset_set_del_pollset(grpc_exec_ctx *exec_ctx,
+                                    grpc_pollset_set *pollset_set,
+                                    grpc_pollset *pollset) {
+  size_t i;
+  gpr_mu_lock(&pollset_set->mu);
+  for (i = 0; i < pollset_set->pollset_count; i++) {
+    if (pollset_set->pollsets[i] == pollset) {
+      pollset_set->pollset_count--;
+      GPR_SWAP(grpc_pollset *, pollset_set->pollsets[i],
+               pollset_set->pollsets[pollset_set->pollset_count]);
+      break;
+    }
+  }
+  gpr_mu_unlock(&pollset_set->mu);
+}
+
+static void pollset_set_add_pollset_set(grpc_exec_ctx *exec_ctx,
+                                        grpc_pollset_set *bag,
+                                        grpc_pollset_set *item) {
+  size_t i, j;
+  gpr_mu_lock(&bag->mu);
+  if (bag->pollset_set_count == bag->pollset_set_capacity) {
+    bag->pollset_set_capacity = GPR_MAX(8, 2 * bag->pollset_set_capacity);
+    bag->pollset_sets =
+        gpr_realloc(bag->pollset_sets,
+                    bag->pollset_set_capacity * sizeof(*bag->pollset_sets));
+  }
+  bag->pollset_sets[bag->pollset_set_count++] = item;
+  for (i = 0, j = 0; i < bag->fd_count; i++) {
+    if (fd_is_orphaned(bag->fds[i])) {
+      GRPC_FD_UNREF(bag->fds[i], "pollset_set");
+    } else {
+      pollset_set_add_fd(exec_ctx, item, bag->fds[i]);
+      bag->fds[j++] = bag->fds[i];
+    }
+  }
+  bag->fd_count = j;
+  gpr_mu_unlock(&bag->mu);
+}
+
+static void pollset_set_del_pollset_set(grpc_exec_ctx *exec_ctx,
+                                        grpc_pollset_set *bag,
+                                        grpc_pollset_set *item) {
+  size_t i;
+  gpr_mu_lock(&bag->mu);
+  for (i = 0; i < bag->pollset_set_count; i++) {
+    if (bag->pollset_sets[i] == item) {
+      bag->pollset_set_count--;
+      GPR_SWAP(grpc_pollset_set *, bag->pollset_sets[i],
+               bag->pollset_sets[bag->pollset_set_count]);
+      break;
+    }
+  }
+  gpr_mu_unlock(&bag->mu);
+}
+
+static void pollset_set_add_fd(grpc_exec_ctx *exec_ctx,
+                               grpc_pollset_set *pollset_set, grpc_fd *fd) {
+  size_t i;
+  gpr_mu_lock(&pollset_set->mu);
+  if (pollset_set->fd_count == pollset_set->fd_capacity) {
+    pollset_set->fd_capacity = GPR_MAX(8, 2 * pollset_set->fd_capacity);
+    pollset_set->fds = gpr_realloc(
+        pollset_set->fds, pollset_set->fd_capacity * sizeof(*pollset_set->fds));
+  }
+  GRPC_FD_REF(fd, "pollset_set");
+  pollset_set->fds[pollset_set->fd_count++] = fd;
+  for (i = 0; i < pollset_set->pollset_count; i++) {
+    pollset_add_fd(exec_ctx, pollset_set->pollsets[i], fd);
+  }
+  for (i = 0; i < pollset_set->pollset_set_count; i++) {
+    pollset_set_add_fd(exec_ctx, pollset_set->pollset_sets[i], fd);
+  }
+  gpr_mu_unlock(&pollset_set->mu);
+}
+
+static void pollset_set_del_fd(grpc_exec_ctx *exec_ctx,
+                               grpc_pollset_set *pollset_set, grpc_fd *fd) {
+  size_t i;
+  gpr_mu_lock(&pollset_set->mu);
+  for (i = 0; i < pollset_set->fd_count; i++) {
+    if (pollset_set->fds[i] == fd) {
+      pollset_set->fd_count--;
+      GPR_SWAP(grpc_fd *, pollset_set->fds[i],
+               pollset_set->fds[pollset_set->fd_count]);
+      GRPC_FD_UNREF(fd, "pollset_set");
+      break;
+    }
+  }
+  for (i = 0; i < pollset_set->pollset_set_count; i++) {
+    pollset_set_del_fd(exec_ctx, pollset_set->pollset_sets[i], fd);
+  }
+  gpr_mu_unlock(&pollset_set->mu);
+}
+
+/*******************************************************************************
+ * event engine binding
+ */
+
+static void shutdown_engine(void) {
+  fd_global_shutdown();
+  pollset_global_shutdown();
+}
+
+static const grpc_event_engine_vtable vtable = {
+    .pollset_size = sizeof(grpc_pollset),
+
+    .fd_create = fd_create,
+    .fd_wrapped_fd = fd_wrapped_fd,
+    .fd_orphan = fd_orphan,
+    .fd_shutdown = fd_shutdown,
+    .fd_notify_on_read = fd_notify_on_read,
+    .fd_notify_on_write = fd_notify_on_write,
+    .fd_get_read_notifier_pollset = fd_get_read_notifier_pollset,
+
+    .pollset_init = pollset_init,
+    .pollset_shutdown = pollset_shutdown,
+    .pollset_reset = pollset_reset,
+    .pollset_destroy = pollset_destroy,
+    .pollset_work = pollset_work,
+    .pollset_kick = pollset_kick,
+    .pollset_add_fd = pollset_add_fd,
+
+    .pollset_set_create = pollset_set_create,
+    .pollset_set_destroy = pollset_set_destroy,
+    .pollset_set_add_pollset = pollset_set_add_pollset,
+    .pollset_set_del_pollset = pollset_set_del_pollset,
+    .pollset_set_add_pollset_set = pollset_set_add_pollset_set,
+    .pollset_set_del_pollset_set = pollset_set_del_pollset_set,
+    .pollset_set_add_fd = pollset_set_add_fd,
+    .pollset_set_del_fd = pollset_set_del_fd,
+
+    .kick_poller = kick_poller,
+
+    .shutdown_engine = shutdown_engine,
+};
+
+const grpc_event_engine_vtable *grpc_init_poll_and_epoll_posix(void) {
+#ifdef GPR_LINUX_MULTIPOLL_WITH_EPOLL
+  platform_become_multipoller = epoll_become_multipoller;
+#else
+  platform_become_multipoller = poll_become_multipoller;
+#endif
+  fd_global_init();
+  pollset_global_init();
+  return &vtable;
+}
+
+#endif
diff --git a/src/core/lib/iomgr/ev_poll_and_epoll_posix.h b/src/core/lib/iomgr/ev_poll_and_epoll_posix.h
new file mode 100644
index 0000000..06d6dbf
--- /dev/null
+++ b/src/core/lib/iomgr/ev_poll_and_epoll_posix.h
@@ -0,0 +1,41 @@
+/*
+ *
+ * Copyright 2015, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#ifndef GRPC_CORE_LIB_IOMGR_EV_POLL_AND_EPOLL_POSIX_H
+#define GRPC_CORE_LIB_IOMGR_EV_POLL_AND_EPOLL_POSIX_H
+
+#include "src/core/lib/iomgr/ev_posix.h"
+
+const grpc_event_engine_vtable *grpc_init_poll_and_epoll_posix(void);
+
+#endif /* GRPC_CORE_LIB_IOMGR_EV_POLL_AND_EPOLL_POSIX_H */
diff --git a/src/core/lib/iomgr/ev_poll_posix.c b/src/core/lib/iomgr/ev_poll_posix.c
index e2a2123..0167999 100644
--- a/src/core/lib/iomgr/ev_poll_posix.c
+++ b/src/core/lib/iomgr/ev_poll_posix.c
@@ -59,8 +59,6 @@
  * FD declarations
  */
 
-grpc_wakeup_fd grpc_global_wakeup_fd;
-
 typedef struct grpc_fd_watcher {
   struct grpc_fd_watcher *next;
   struct grpc_fd_watcher *prev;
diff --git a/src/core/lib/iomgr/ev_posix.c b/src/core/lib/iomgr/ev_posix.c
index 95520b0..6477b05 100644
--- a/src/core/lib/iomgr/ev_posix.c
+++ b/src/core/lib/iomgr/ev_posix.c
@@ -44,6 +44,7 @@
 #include <grpc/support/string_util.h>
 #include <grpc/support/useful.h>
 
+#include "src/core/lib/iomgr/ev_poll_and_epoll_posix.h"
 #include "src/core/lib/iomgr/ev_poll_posix.h"
 #include "src/core/lib/support/env.h"
 
@@ -61,7 +62,7 @@
 } event_engine_factory;
 
 static const event_engine_factory g_factories[] = {
-    {"poll", grpc_init_poll_posix},
+    {"poll", grpc_init_poll_posix}, {"legacy", grpc_init_poll_and_epoll_posix},
 };
 
 static void add(const char *beg, const char *end, char ***ss, size_t *ns) {
diff --git a/src/csharp/Grpc.Core.Tests/ClientServerTest.cs b/src/csharp/Grpc.Core.Tests/ClientServerTest.cs
index d92addb..dcdddc7 100644
--- a/src/csharp/Grpc.Core.Tests/ClientServerTest.cs
+++ b/src/csharp/Grpc.Core.Tests/ClientServerTest.cs
@@ -235,8 +235,16 @@
             await barrier.Task;  // make sure the handler has started.
             cts.Cancel();
 
-            var ex = Assert.ThrowsAsync<RpcException>(async () => await call.ResponseAsync);
-            Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
+            try
+            {
+                // cannot use Assert.ThrowsAsync because it uses Task.Wait and would deadlock.
+                await call.ResponseAsync;
+                Assert.Fail();
+            }
+            catch (RpcException ex)
+            {
+                Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
+            }
         }
 
         [Test]
@@ -265,9 +273,15 @@
             await handlerStartedBarrier.Task;
             cts.Cancel();
 
-            var ex = Assert.ThrowsAsync<RpcException>(async () => await call.ResponseAsync);
-            Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
-
+            try
+            {
+                await call.ResponseAsync;
+                Assert.Fail();
+            }
+            catch (RpcException ex)
+            {
+                Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
+            }
             Assert.AreEqual("SUCCESS", await successTcs.Task);
         }
 
diff --git a/src/csharp/Grpc.Core.Tests/ContextPropagationTest.cs b/src/csharp/Grpc.Core.Tests/ContextPropagationTest.cs
index cec8c7c..6a15629 100644
--- a/src/csharp/Grpc.Core.Tests/ContextPropagationTest.cs
+++ b/src/csharp/Grpc.Core.Tests/ContextPropagationTest.cs
@@ -105,7 +105,15 @@
             var parentCall = Calls.AsyncClientStreamingCall(helper.CreateClientStreamingCall(new CallOptions(cancellationToken: cts.Token)));
             await readyToCancelTcs.Task;
             cts.Cancel();
-            Assert.ThrowsAsync(typeof(RpcException), async () => await parentCall);
+            try
+            {
+                // cannot use Assert.ThrowsAsync because it uses Task.Wait and would deadlock.
+                await parentCall;
+                Assert.Fail();
+            }
+            catch (RpcException)
+            {
+            }
             Assert.AreEqual("CHILD_CALL_CANCELLED", await successTcs.Task);
         }
 
diff --git a/src/csharp/Grpc.Core.Tests/GrpcEnvironmentTest.cs b/src/csharp/Grpc.Core.Tests/GrpcEnvironmentTest.cs
index ab12c12..6fe3827 100644
--- a/src/csharp/Grpc.Core.Tests/GrpcEnvironmentTest.cs
+++ b/src/csharp/Grpc.Core.Tests/GrpcEnvironmentTest.cs
@@ -32,7 +32,7 @@
 #endregion
 
 using System;
-using System.Threading;
+using System.Linq;
 using Grpc.Core;
 using NUnit.Framework;
 
@@ -44,7 +44,11 @@
         public void InitializeAndShutdownGrpcEnvironment()
         {
             var env = GrpcEnvironment.AddRef();
-            Assert.IsNotNull(env.CompletionQueue);
+            Assert.IsTrue(env.CompletionQueues.Count > 0);
+            for (int i = 0; i < env.CompletionQueues.Count; i++)
+            {
+                Assert.IsNotNull(env.CompletionQueues.ElementAt(i));
+            }
             GrpcEnvironment.Release();
         }
 
diff --git a/src/csharp/Grpc.Core.Tests/Internal/AsyncCallServerTest.cs b/src/csharp/Grpc.Core.Tests/Internal/AsyncCallServerTest.cs
index 0e20476..c35aaf6 100644
--- a/src/csharp/Grpc.Core.Tests/Internal/AsyncCallServerTest.cs
+++ b/src/csharp/Grpc.Core.Tests/Internal/AsyncCallServerTest.cs
@@ -53,8 +53,6 @@
         [SetUp]
         public void Init()
         {
-            var environment = GrpcEnvironment.AddRef();
-
             // Create a fake server just so we have an instance to refer to.
             // The server won't actually be used at all.
             server = new Server()
@@ -66,7 +64,6 @@
             fakeCall = new FakeNativeCall();
             asyncCallServer = new AsyncCallServer<string, string>(
                 Marshallers.StringMarshaller.Serializer, Marshallers.StringMarshaller.Deserializer,
-                environment,
                 server);
             asyncCallServer.InitializeForTesting(fakeCall);
         }
@@ -75,7 +72,6 @@
         public void Cleanup()
         {
             server.ShutdownAsync().Wait();
-            GrpcEnvironment.Release();
         }
 
         [Test]
@@ -136,7 +132,6 @@
         public void WriteAfterCancelNotificationFails()
         {
             var finishedTask = asyncCallServer.ServerSideCallAsync();
-            var requestStream = new ServerRequestStream<string, string>(asyncCallServer);
             var responseStream = new ServerResponseStream<string, string>(asyncCallServer);
 
             fakeCall.ReceivedCloseOnServerHandler(true, cancelled: true);
@@ -181,6 +176,21 @@
             AssertFinished(asyncCallServer, fakeCall, finishedTask);
         }
 
+        [Test]
+        public void WriteAfterWriteStatusThrowsInvalidOperationException()
+        {
+            var finishedTask = asyncCallServer.ServerSideCallAsync();
+            var responseStream = new ServerResponseStream<string, string>(asyncCallServer);
+
+            asyncCallServer.SendStatusFromServerAsync(Status.DefaultSuccess, new Metadata(), null);
+            Assert.ThrowsAsync(typeof(InvalidOperationException), async () => await responseStream.WriteAsync("request1"));
+
+            fakeCall.SendStatusFromServerHandler(true);
+            fakeCall.ReceivedCloseOnServerHandler(true, cancelled: true);
+
+            AssertFinished(asyncCallServer, fakeCall, finishedTask);
+        }
+
         static void AssertFinished(AsyncCallServer<string, string> asyncCallServer, FakeNativeCall fakeCall, Task finishedTask)
         {
             Assert.IsTrue(fakeCall.IsDisposed);
diff --git a/src/csharp/Grpc.Core.Tests/Internal/AsyncCallTest.cs b/src/csharp/Grpc.Core.Tests/Internal/AsyncCallTest.cs
index 777a1c8..81897f8 100644
--- a/src/csharp/Grpc.Core.Tests/Internal/AsyncCallTest.cs
+++ b/src/csharp/Grpc.Core.Tests/Internal/AsyncCallTest.cs
@@ -33,7 +33,6 @@
 
 using System;
 using System.Collections.Generic;
-using System.Runtime.InteropServices;
 using System.Threading.Tasks;
 
 using Grpc.Core.Internal;
@@ -82,7 +81,7 @@
             Assert.ThrowsAsync(typeof(InvalidOperationException),
                 async () => await asyncCall.ReadMessageAsync());
             Assert.Throws(typeof(InvalidOperationException),
-                () => asyncCall.StartSendMessage("abc", new WriteFlags(), (x,y) => {}));
+                () => asyncCall.SendMessageAsync("abc", new WriteFlags()));
         }
 
         [Test]
@@ -103,7 +102,7 @@
             var resultTask = asyncCall.UnaryCallAsync("request1");
             fakeCall.UnaryResponseClientHandler(true,
                 CreateClientSideStatus(StatusCode.InvalidArgument),
-                CreateResponsePayload(),
+                null,
                 new Metadata());
 
             AssertUnaryResponseError(asyncCall, fakeCall, resultTask, StatusCode.InvalidArgument);
@@ -148,7 +147,7 @@
             var resultTask = asyncCall.ClientStreamingCallAsync();
             fakeCall.UnaryResponseClientHandler(true,
                 CreateClientSideStatus(StatusCode.InvalidArgument),
-                CreateResponsePayload(),
+                null,
                 new Metadata());
 
             AssertUnaryResponseError(asyncCall, fakeCall, resultTask, StatusCode.InvalidArgument);
@@ -193,7 +192,7 @@
 
             fakeCall.UnaryResponseClientHandler(true,
                 CreateClientSideStatus(StatusCode.Internal),
-                CreateResponsePayload(),
+                null,
                 new Metadata());
 
             AssertUnaryResponseError(asyncCall, fakeCall, resultTask, StatusCode.Internal);
@@ -211,7 +210,9 @@
                 new Metadata());
 
             AssertUnaryResponseSuccess(asyncCall, fakeCall, resultTask);
-            var ex = Assert.Throws<RpcException>(() => requestStream.WriteAsync("request1"));
+
+            var writeTask = requestStream.WriteAsync("request1");
+            var ex = Assert.ThrowsAsync<RpcException>(async () => await writeTask);
             Assert.AreEqual(Status.DefaultSuccess, ex.Status);
         }
 
@@ -223,11 +224,13 @@
 
             fakeCall.UnaryResponseClientHandler(true,
                 new ClientSideStatus(new Status(StatusCode.OutOfRange, ""), new Metadata()),
-                CreateResponsePayload(),
+                null,
                 new Metadata());
 
             AssertUnaryResponseError(asyncCall, fakeCall, resultTask, StatusCode.OutOfRange);
-            var ex = Assert.Throws<RpcException>(() => requestStream.WriteAsync("request1"));
+
+            var writeTask = requestStream.WriteAsync("request1");
+            var ex = Assert.ThrowsAsync<RpcException>(async () => await writeTask);
             Assert.AreEqual(StatusCode.OutOfRange, ex.Status.StatusCode);
         }
 
@@ -267,7 +270,7 @@
         }
 
         [Test]
-        public void ClientStreaming_WriteAfterCancellationRequestThrowsOperationCancelledException()
+        public void ClientStreaming_WriteAfterCancellationRequestThrowsTaskCanceledException()
         {
             var resultTask = asyncCall.ClientStreamingCallAsync();
             var requestStream = new ClientRequestStream<string, string>(asyncCall);
@@ -275,11 +278,12 @@
             asyncCall.Cancel();
             Assert.IsTrue(fakeCall.IsCancelled);
 
-            Assert.Throws(typeof(OperationCanceledException), () => requestStream.WriteAsync("request1"));
+            var writeTask = requestStream.WriteAsync("request1");
+            Assert.ThrowsAsync(typeof(TaskCanceledException), async () => await writeTask);
 
             fakeCall.UnaryResponseClientHandler(true,
                 CreateClientSideStatus(StatusCode.Cancelled),
-                CreateResponsePayload(),
+                null,
                 new Metadata());
 
             AssertUnaryResponseError(asyncCall, fakeCall, resultTask, StatusCode.Cancelled);
@@ -290,7 +294,7 @@
         {
             asyncCall.StartServerStreamingCall("request1");
             Assert.Throws(typeof(InvalidOperationException),
-                () => asyncCall.StartSendMessage("abc", new WriteFlags(), (x,y) => {}));
+                () => asyncCall.SendMessageAsync("abc", new WriteFlags()));
         }
 
         [Test]
@@ -390,12 +394,13 @@
 
             AssertStreamingResponseSuccess(asyncCall, fakeCall, readTask);
 
-            var ex = Assert.ThrowsAsync<RpcException>(async () => await requestStream.WriteAsync("request1"));
+            var writeTask = requestStream.WriteAsync("request1");
+            var ex = Assert.ThrowsAsync<RpcException>(async () => await writeTask);
             Assert.AreEqual(Status.DefaultSuccess, ex.Status);
         }
 
         [Test]
-        public void DuplexStreaming_CompleteAfterReceivingStatusFails()
+        public void DuplexStreaming_CompleteAfterReceivingStatusSuceeds()
         {
             asyncCall.StartDuplexStreamingCall();
             var requestStream = new ClientRequestStream<string, string>(asyncCall);
@@ -411,7 +416,7 @@
         }
 
         [Test]
-        public void DuplexStreaming_WriteAfterCancellationRequestThrowsOperationCancelledException()
+        public void DuplexStreaming_WriteAfterCancellationRequestThrowsTaskCanceledException()
         {
             asyncCall.StartDuplexStreamingCall();
             var requestStream = new ClientRequestStream<string, string>(asyncCall);
@@ -419,7 +424,9 @@
 
             asyncCall.Cancel();
             Assert.IsTrue(fakeCall.IsCancelled);
-            Assert.Throws(typeof(OperationCanceledException), () => requestStream.WriteAsync("request1"));
+
+            var writeTask = requestStream.WriteAsync("request1");
+            Assert.ThrowsAsync(typeof(TaskCanceledException), async () => await writeTask);
 
             var readTask = responseStream.MoveNext();
             fakeCall.ReceivedMessageHandler(true, null);
diff --git a/src/csharp/Grpc.Core.Tests/MarshallingErrorsTest.cs b/src/csharp/Grpc.Core.Tests/MarshallingErrorsTest.cs
index 0663e77..d770f82 100644
--- a/src/csharp/Grpc.Core.Tests/MarshallingErrorsTest.cs
+++ b/src/csharp/Grpc.Core.Tests/MarshallingErrorsTest.cs
@@ -134,7 +134,15 @@
         {
             helper.ClientStreamingHandler = new ClientStreamingServerMethod<string, string>(async (requestStream, context) =>
             {
-                Assert.ThrowsAsync<IOException>(async () => await requestStream.MoveNext());
+                try
+                {
+                    // cannot use Assert.ThrowsAsync because it uses Task.Wait and would deadlock.
+                    await requestStream.MoveNext();
+                    Assert.Fail();
+                }
+                catch (IOException)
+                {
+                }
                 return "RESPONSE";
             });
 
diff --git a/src/csharp/Grpc.Core/Channel.cs b/src/csharp/Grpc.Core/Channel.cs
index 93a6e6a..886adfe 100644
--- a/src/csharp/Grpc.Core/Channel.cs
+++ b/src/csharp/Grpc.Core/Channel.cs
@@ -31,7 +31,6 @@
 
 using System;
 using System.Collections.Generic;
-using System.Linq;
 using System.Threading;
 using System.Threading.Tasks;
 
@@ -56,6 +55,7 @@
 
         readonly string target;
         readonly GrpcEnvironment environment;
+        readonly CompletionQueueSafeHandle completionQueue;
         readonly ChannelSafeHandle handle;
         readonly Dictionary<string, ChannelOption> options;
 
@@ -75,6 +75,7 @@
             EnsureUserAgentChannelOption(this.options);
             this.environment = GrpcEnvironment.AddRef();
 
+            this.completionQueue = this.environment.PickCompletionQueue();
             using (var nativeCredentials = credentials.ToNativeCredentials())
             using (var nativeChannelArgs = ChannelOptions.CreateChannelArgs(this.options.Values))
             {
@@ -135,7 +136,7 @@
                     tcs.SetCanceled();
                 }
             });
-            handle.WatchConnectivityState(lastObservedState, deadlineTimespec, environment.CompletionQueue, environment.CompletionRegistry, handler);
+            handle.WatchConnectivityState(lastObservedState, deadlineTimespec, completionQueue, handler);
             return tcs.Task;
         }
 
@@ -231,6 +232,14 @@
             }
         }
 
+        internal CompletionQueueSafeHandle CompletionQueue
+        {
+            get
+            {
+                return this.completionQueue;
+            }
+        }
+
         internal void AddCallReference(object call)
         {
             activeCallCounter.Increment();
diff --git a/src/csharp/Grpc.Core/Grpc.Core.csproj b/src/csharp/Grpc.Core/Grpc.Core.csproj
index 4bf30e8..a8b7b5f 100644
--- a/src/csharp/Grpc.Core/Grpc.Core.csproj
+++ b/src/csharp/Grpc.Core/Grpc.Core.csproj
@@ -86,7 +86,6 @@
     <Compile Include="Utils\BenchmarkUtil.cs" />
     <Compile Include="ChannelCredentials.cs" />
     <Compile Include="Internal\ChannelArgsSafeHandle.cs" />
-    <Compile Include="Internal\AsyncCompletion.cs" />
     <Compile Include="Internal\AsyncCallBase.cs" />
     <Compile Include="Internal\AsyncCallServer.cs" />
     <Compile Include="Internal\AsyncCall.cs" />
diff --git a/src/csharp/Grpc.Core/GrpcEnvironment.cs b/src/csharp/Grpc.Core/GrpcEnvironment.cs
index bee0ef1..18af109 100644
--- a/src/csharp/Grpc.Core/GrpcEnvironment.cs
+++ b/src/csharp/Grpc.Core/GrpcEnvironment.cs
@@ -32,8 +32,9 @@
 #endregion
 
 using System;
+using System.Collections.Generic;
+using System.Linq;
 using System.Runtime.InteropServices;
-using System.Threading.Tasks;
 using Grpc.Core.Internal;
 using Grpc.Core.Logging;
 using Grpc.Core.Utils;
@@ -51,12 +52,13 @@
         static GrpcEnvironment instance;
         static int refCount;
         static int? customThreadPoolSize;
+        static int? customCompletionQueueCount;
 
         static ILogger logger = new ConsoleLogger();
 
         readonly GrpcThreadPool threadPool;
-        readonly CompletionRegistry completionRegistry;
         readonly DebugStats debugStats = new DebugStats();
+        readonly AtomicCounter cqPickerCounter = new AtomicCounter();
         bool isClosed;
 
         /// <summary>
@@ -141,36 +143,51 @@
         }
 
         /// <summary>
+        /// Sets the number of completion queues in the  gRPC thread pool that polls for internal RPC events.
+        /// Can be only invoke before the <c>GrpcEnviroment</c> is started and cannot be changed afterwards.
+        /// Setting the number of completions queues is an advanced setting and you should only use it if you know what you are doing.
+        /// Most users should rely on the default value provided by gRPC library.
+        /// Note: this method is part of an experimental API that can change or be removed without any prior notice.
+        /// </summary>
+        public static void SetCompletionQueueCount(int completionQueueCount)
+        {
+            lock (staticLock)
+            {
+                GrpcPreconditions.CheckState(instance == null, "Can only be set before GrpcEnvironment is initialized");
+                GrpcPreconditions.CheckArgument(completionQueueCount > 0, "threadCount needs to be a positive number");
+                customCompletionQueueCount = completionQueueCount;
+            }
+        }
+
+        /// <summary>
         /// Creates gRPC environment.
         /// </summary>
         private GrpcEnvironment()
         {
             GrpcNativeInit();
-            completionRegistry = new CompletionRegistry(this);
-            threadPool = new GrpcThreadPool(this, GetThreadPoolSizeOrDefault());
+            threadPool = new GrpcThreadPool(this, GetThreadPoolSizeOrDefault(), GetCompletionQueueCountOrDefault());
             threadPool.Start();
         }
 
         /// <summary>
-        /// Gets the completion registry used by this gRPC environment.
+        /// Gets the completion queues used by this gRPC environment.
         /// </summary>
-        internal CompletionRegistry CompletionRegistry
+        internal IReadOnlyCollection<CompletionQueueSafeHandle> CompletionQueues
         {
             get
             {
-                return this.completionRegistry;
+                return this.threadPool.CompletionQueues;
             }
         }
 
         /// <summary>
-        /// Gets the completion queue used by this gRPC environment.
+        /// Picks a completion queue in a round-robin fashion.
+        /// Shouldn't be invoked on a per-call basis (used at per-channel basis).
         /// </summary>
-        internal CompletionQueueSafeHandle CompletionQueue
+        internal CompletionQueueSafeHandle PickCompletionQueue()
         {
-            get
-            {
-                return this.threadPool.CompletionQueue;
-            }
+            var cqIndex = (int) ((cqPickerCounter.Increment() - 1) % this.threadPool.CompletionQueues.Count);
+            return this.threadPool.CompletionQueues.ElementAt(cqIndex);
         }
 
         /// <summary>
@@ -230,5 +247,15 @@
             // more work, but seems to work reasonably well for a start.
             return Math.Max(MinDefaultThreadPoolSize, Environment.ProcessorCount / 2);
         }
+
+        private int GetCompletionQueueCountOrDefault()
+        {
+            if (customCompletionQueueCount.HasValue)
+            {
+                return customCompletionQueueCount.Value;
+            }
+            // by default, create a completion queue for each thread
+            return GetThreadPoolSizeOrDefault();
+        }
     }
 }
diff --git a/src/csharp/Grpc.Core/Internal/AsyncCall.cs b/src/csharp/Grpc.Core/Internal/AsyncCall.cs
index 5535186..895be69 100644
--- a/src/csharp/Grpc.Core/Internal/AsyncCall.cs
+++ b/src/csharp/Grpc.Core/Internal/AsyncCall.cs
@@ -32,12 +32,7 @@
 #endregion
 
 using System;
-using System.Diagnostics;
-using System.Runtime.CompilerServices;
-using System.Runtime.InteropServices;
-using System.Threading;
 using System.Threading.Tasks;
-using Grpc.Core.Internal;
 using Grpc.Core.Logging;
 using Grpc.Core.Profiling;
 using Grpc.Core.Utils;
@@ -57,9 +52,11 @@
         // Completion of a pending unary response if not null.
         TaskCompletionSource<TResponse> unaryResponseTcs;
 
+        // TODO(jtattermusch): this field doesn't need to be initialized for unary response calls.
         // Indicates that response streaming call has finished.
         TaskCompletionSource<object> streamingCallFinishedTcs = new TaskCompletionSource<object>();
 
+        // TODO(jtattermusch): this field could be lazy-initialized (only if someone requests the response headers).
         // Response headers set here once received.
         TaskCompletionSource<Metadata> responseHeadersTcs = new TaskCompletionSource<Metadata>();
 
@@ -67,7 +64,7 @@
         ClientSideStatus? finishedStatus;
 
         public AsyncCall(CallInvocationDetails<TRequest, TResponse> callDetails)
-            : base(callDetails.RequestMarshaller.Serializer, callDetails.ResponseMarshaller.Deserializer, callDetails.Channel.Environment)
+            : base(callDetails.RequestMarshaller.Serializer, callDetails.ResponseMarshaller.Deserializer)
         {
             this.details = callDetails.WithOptions(callDetails.Options.Normalize());
             this.initialMetadataSent = true;  // we always send metadata at the very beginning of the call.
@@ -144,7 +141,7 @@
                 GrpcPreconditions.CheckState(!started);
                 started = true;
 
-                Initialize(environment.CompletionQueue);
+                Initialize(details.Channel.CompletionQueue);
 
                 halfcloseRequested = true;
                 readingDone = true;
@@ -171,7 +168,7 @@
                 GrpcPreconditions.CheckState(!started);
                 started = true;
 
-                Initialize(environment.CompletionQueue);
+                Initialize(details.Channel.CompletionQueue);
 
                 readingDone = true;
 
@@ -195,7 +192,7 @@
                 GrpcPreconditions.CheckState(!started);
                 started = true;
 
-                Initialize(environment.CompletionQueue);
+                Initialize(details.Channel.CompletionQueue);
 
                 halfcloseRequested = true;
 
@@ -220,7 +217,7 @@
                 GrpcPreconditions.CheckState(!started);
                 started = true;
 
-                Initialize(environment.CompletionQueue);
+                Initialize(details.Channel.CompletionQueue);
 
                 using (var metadataArray = MetadataArraySafeHandle.Create(details.Options.Headers))
                 {
@@ -232,11 +229,10 @@
 
         /// <summary>
         /// Sends a streaming request. Only one pending send action is allowed at any given time.
-        /// completionDelegate is called when the operation finishes.
         /// </summary>
-        public void StartSendMessage(TRequest msg, WriteFlags writeFlags, AsyncCompletionDelegate<object> completionDelegate)
+        public Task SendMessageAsync(TRequest msg, WriteFlags writeFlags)
         {
-            StartSendMessageInternal(msg, writeFlags, completionDelegate);
+            return SendMessageInternalAsync(msg, writeFlags);
         }
 
         /// <summary>
@@ -250,29 +246,32 @@
         /// <summary>
         /// Sends halfclose, indicating client is done with streaming requests.
         /// Only one pending send action is allowed at any given time.
-        /// completionDelegate is called when the operation finishes.
         /// </summary>
-        public void StartSendCloseFromClient(AsyncCompletionDelegate<object> completionDelegate)
+        public Task SendCloseFromClientAsync()
         {
             lock (myLock)
             {
-                GrpcPreconditions.CheckNotNull(completionDelegate, "Completion delegate cannot be null");
-                CheckSendingAllowed(allowFinished: true);
+                GrpcPreconditions.CheckState(started);
 
-                if (!disposed && !finished)
+                var earlyResult = CheckSendPreconditionsClientSide();
+                if (earlyResult != null)
                 {
-                    call.StartSendCloseFromClient(HandleSendCloseFromClientFinished);
+                    return earlyResult;
                 }
-                else
+
+                if (disposed || finished)
                 {
                     // In case the call has already been finished by the serverside,
-                    // the halfclose has already been done implicitly, so we only
-                    // emit the notification for the completion delegate.
-                    Task.Run(() => HandleSendCloseFromClientFinished(true));
+                    // the halfclose has already been done implicitly, so just return
+                    // completed task here.
+                    halfcloseRequested = true;
+                    return Task.FromResult<object>(null);
                 }
+                call.StartSendCloseFromClient(HandleSendCloseFromClientFinished);
 
                 halfcloseRequested = true;
-                sendCompletionDelegate = completionDelegate;
+                streamingWriteTcs = new TaskCompletionSource<object>();
+                return streamingWriteTcs.Task;
             }
         }
 
@@ -342,6 +341,45 @@
             get { return true; }
         }
 
+        protected override Task CheckSendAllowedOrEarlyResult()
+        {
+            var earlyResult = CheckSendPreconditionsClientSide();
+            if (earlyResult != null)
+            {
+                return earlyResult;
+            }
+
+            if (finishedStatus.HasValue)
+            {
+                // throwing RpcException if we already received status on client
+                // side makes the most sense.
+                // Note that this throws even for StatusCode.OK.
+                // Writing after the call has finished is not a programming error because server can close
+                // the call anytime, so don't throw directly, but let the write task finish with an error.
+                var tcs = new TaskCompletionSource<object>();
+                tcs.SetException(new RpcException(finishedStatus.Value.Status));
+                return tcs.Task;
+            }
+
+            return null;
+        }
+
+        private Task CheckSendPreconditionsClientSide()
+        {
+            GrpcPreconditions.CheckState(!halfcloseRequested, "Request stream has already been completed.");
+            GrpcPreconditions.CheckState(streamingWriteTcs == null, "Only one write can be pending at a time.");
+
+            if (cancelRequested)
+            {
+                // Return a cancelled task.
+                var tcs = new TaskCompletionSource<object>();
+                tcs.SetCanceled();
+                return tcs.Task;
+            }
+
+            return null;
+        }
+
         private void Initialize(CompletionQueueSafeHandle cq)
         {
             using (Profilers.ForCurrentThread().NewScope("AsyncCall.Initialize"))
@@ -368,7 +406,7 @@
                 var credentials = details.Options.Credentials;
                 using (var nativeCredentials = credentials != null ? credentials.ToNativeCredentials() : null)
                 {
-                    var result = details.Channel.Handle.CreateCall(environment.CompletionRegistry,
+                    var result = details.Channel.Handle.CreateCall(
                                  parentCall, ContextPropagationToken.DefaultMask, cq,
                                  details.Method, details.Host, Timespec.FromDateTime(details.Options.Deadline.Value), nativeCredentials);
                     return result;
@@ -400,6 +438,7 @@
         /// </summary>
         private void HandleReceivedResponseHeaders(bool success, Metadata responseHeaders)
         {
+            // TODO(jtattermusch): handle success==false
             responseHeadersTcs.SetResult(responseHeaders);
         }
 
@@ -443,19 +482,6 @@
             }
         }
 
-        protected override void CheckSendingAllowed(bool allowFinished)
-        {
-            base.CheckSendingAllowed(true);
-
-            // throwing RpcException if we already received status on client
-            // side makes the most sense.
-            // Note that this throws even for StatusCode.OK.
-            if (!allowFinished && finishedStatus.HasValue)
-            {
-                throw new RpcException(finishedStatus.Value.Status);
-            }
-        }
-
         /// <summary>
         /// Handles receive status completion for calls with streaming response.
         /// </summary>
diff --git a/src/csharp/Grpc.Core/Internal/AsyncCallBase.cs b/src/csharp/Grpc.Core/Internal/AsyncCallBase.cs
index 4de2370..cb8366c 100644
--- a/src/csharp/Grpc.Core/Internal/AsyncCallBase.cs
+++ b/src/csharp/Grpc.Core/Internal/AsyncCallBase.cs
@@ -58,7 +58,6 @@
         readonly Func<TWrite, byte[]> serializer;
         readonly Func<byte[], TRead> deserializer;
 
-        protected readonly GrpcEnvironment environment;
         protected readonly object myLock = new object();
 
         protected INativeCall call;
@@ -67,8 +66,8 @@
         protected bool started;
         protected bool cancelRequested;
 
-        protected AsyncCompletionDelegate<object> sendCompletionDelegate;  // Completion of a pending send or sendclose if not null.
         protected TaskCompletionSource<TRead> streamingReadTcs;  // Completion of a pending streaming read if not null.
+        protected TaskCompletionSource<object> streamingWriteTcs;  // Completion of a pending streaming write or send close from client if not null.
         protected TaskCompletionSource<object> sendStatusFromServerTcs;
 
         protected bool readingDone;  // True if last read (i.e. read with null payload) was already received.
@@ -78,11 +77,10 @@
         protected bool initialMetadataSent;
         protected long streamingWritesCounter;  // Number of streaming send operations started so far.
 
-        public AsyncCallBase(Func<TWrite, byte[]> serializer, Func<byte[], TRead> deserializer, GrpcEnvironment environment)
+        public AsyncCallBase(Func<TWrite, byte[]> serializer, Func<byte[], TRead> deserializer)
         {
             this.serializer = GrpcPreconditions.CheckNotNull(serializer);
             this.deserializer = GrpcPreconditions.CheckNotNull(deserializer);
-            this.environment = GrpcPreconditions.CheckNotNull(environment);
         }
 
         /// <summary>
@@ -128,28 +126,31 @@
 
         /// <summary>
         /// Initiates sending a message. Only one send operation can be active at a time.
-        /// completionDelegate is invoked upon completion.
         /// </summary>
-        protected void StartSendMessageInternal(TWrite msg, WriteFlags writeFlags, AsyncCompletionDelegate<object> completionDelegate)
+        protected Task SendMessageInternalAsync(TWrite msg, WriteFlags writeFlags)
         {
             byte[] payload = UnsafeSerialize(msg);
 
             lock (myLock)
             {
-                GrpcPreconditions.CheckNotNull(completionDelegate, "Completion delegate cannot be null");
-                CheckSendingAllowed(allowFinished: false);
+                GrpcPreconditions.CheckState(started);
+                var earlyResult = CheckSendAllowedOrEarlyResult();
+                if (earlyResult != null)
+                {
+                    return earlyResult;
+                }
 
                 call.StartSendMessage(HandleSendFinished, payload, writeFlags, !initialMetadataSent);
 
-                sendCompletionDelegate = completionDelegate;
                 initialMetadataSent = true;
                 streamingWritesCounter++;
+                streamingWriteTcs = new TaskCompletionSource<object>();
+                return streamingWriteTcs.Task;
             }
         }
 
         /// <summary>
         /// Initiates reading a message. Only one read operation can be active at a time.
-        /// completionDelegate is invoked upon completion.
         /// </summary>
         protected Task<TRead> ReadMessageInternalAsync()
         {
@@ -159,7 +160,7 @@
                 if (readingDone)
                 {
                     // the last read that returns null or throws an exception is idempotent
-                    // and maintain its state.
+                    // and maintains its state.
                     GrpcPreconditions.CheckState(streamingReadTcs != null, "Call does not support streaming reads.");
                     return streamingReadTcs.Task;
                 }
@@ -183,7 +184,7 @@
             {
                 if (!disposed && call != null)
                 {
-                    bool noMoreSendCompletions = sendCompletionDelegate == null && (halfcloseRequested || cancelRequested || finished);
+                    bool noMoreSendCompletions = streamingWriteTcs == null && (halfcloseRequested || cancelRequested || finished);
                     if (noMoreSendCompletions && readingDone && finished)
                     {
                         ReleaseResources();
@@ -213,24 +214,11 @@
         {
         }
 
-        protected virtual void CheckSendingAllowed(bool allowFinished)
-        {
-            GrpcPreconditions.CheckState(started);
-            CheckNotCancelled();
-            GrpcPreconditions.CheckState(!disposed || allowFinished);
-
-            GrpcPreconditions.CheckState(!halfcloseRequested, "Already halfclosed.");
-            GrpcPreconditions.CheckState(!finished || allowFinished, "Already finished.");
-            GrpcPreconditions.CheckState(sendCompletionDelegate == null, "Only one write can be pending at a time");
-        }
-
-        protected void CheckNotCancelled()
-        {
-            if (cancelRequested)
-            {
-                throw new OperationCanceledException("Remote call has been cancelled.");
-            }
-        }
+        /// <summary>
+        /// Checks if sending is allowed and possibly returns a Task that allows short-circuiting the send
+        /// logic by directly returning the write operation result task. Normally, null is returned.
+        /// </summary>
+        protected abstract Task CheckSendAllowedOrEarlyResult();
 
         protected byte[] UnsafeSerialize(TWrite msg)
         {
@@ -259,39 +247,27 @@
             }
         }
 
-        protected void FireCompletion<T>(AsyncCompletionDelegate<T> completionDelegate, T value, Exception error)
-        {
-            try
-            {
-                completionDelegate(value, error);
-            }
-            catch (Exception e)
-            {
-                Logger.Error(e, "Exception occured while invoking completion delegate.");
-            }
-        }
-
         /// <summary>
         /// Handles send completion.
         /// </summary>
         protected void HandleSendFinished(bool success)
         {
-            AsyncCompletionDelegate<object> origCompletionDelegate = null;
+            TaskCompletionSource<object> origTcs = null;
             lock (myLock)
             {
-                origCompletionDelegate = sendCompletionDelegate;
-                sendCompletionDelegate = null;
+                origTcs = streamingWriteTcs;
+                streamingWriteTcs = null;
 
                 ReleaseResourcesIfPossible();
             }
 
             if (!success)
             {
-                FireCompletion(origCompletionDelegate, null, new InvalidOperationException("Send failed"));
+                origTcs.SetException(new InvalidOperationException("Send failed"));
             }
             else
             {
-                FireCompletion(origCompletionDelegate, null, null);
+                origTcs.SetResult(null);
             }
         }
 
@@ -300,22 +276,23 @@
         /// </summary>
         protected void HandleSendCloseFromClientFinished(bool success)
         {
-            AsyncCompletionDelegate<object> origCompletionDelegate = null;
+            TaskCompletionSource<object> origTcs = null;
             lock (myLock)
             {
-                origCompletionDelegate = sendCompletionDelegate;
-                sendCompletionDelegate = null;
+                origTcs = streamingWriteTcs;
+                streamingWriteTcs = null;
 
                 ReleaseResourcesIfPossible();
             }
 
             if (!success)
             {
-                FireCompletion(origCompletionDelegate, null, new InvalidOperationException("Sending close from client has failed."));
+                // TODO(jtattermusch): this method is same as HandleSendFinished (only the error message differs).
+                origTcs.SetException(new InvalidOperationException("Sending close from client has failed."));
             }
             else
             {
-                FireCompletion(origCompletionDelegate, null, null);
+                origTcs.SetResult(null);
             }
         }
 
diff --git a/src/csharp/Grpc.Core/Internal/AsyncCallServer.cs b/src/csharp/Grpc.Core/Internal/AsyncCallServer.cs
index b1566b4..56c23ba 100644
--- a/src/csharp/Grpc.Core/Internal/AsyncCallServer.cs
+++ b/src/csharp/Grpc.Core/Internal/AsyncCallServer.cs
@@ -51,14 +51,14 @@
         readonly CancellationTokenSource cancellationTokenSource = new CancellationTokenSource();
         readonly Server server;
 
-        public AsyncCallServer(Func<TResponse, byte[]> serializer, Func<byte[], TRequest> deserializer, GrpcEnvironment environment, Server server) : base(serializer, deserializer, environment)
+        public AsyncCallServer(Func<TResponse, byte[]> serializer, Func<byte[], TRequest> deserializer, Server server) : base(serializer, deserializer)
         {
             this.server = GrpcPreconditions.CheckNotNull(server);
         }
 
-        public void Initialize(CallSafeHandle call)
+        public void Initialize(CallSafeHandle call, CompletionQueueSafeHandle completionQueue)
         {
-            call.Initialize(environment.CompletionRegistry, environment.CompletionQueue);
+            call.Initialize(completionQueue);
 
             server.AddCallReference(this);
             InitializeInternal(call);
@@ -91,11 +91,10 @@
 
         /// <summary>
         /// Sends a streaming response. Only one pending send action is allowed at any given time.
-        /// completionDelegate is called when the operation finishes.
         /// </summary>
-        public void StartSendMessage(TResponse msg, WriteFlags writeFlags, AsyncCompletionDelegate<object> completionDelegate)
+        public Task SendMessageAsync(TResponse msg, WriteFlags writeFlags)
         {
-            StartSendMessageInternal(msg, writeFlags, completionDelegate);
+            return SendMessageInternalAsync(msg, writeFlags);
         }
 
         /// <summary>
@@ -110,20 +109,22 @@
         /// Initiates sending a initial metadata. 
         /// Even though C-core allows sending metadata in parallel to sending messages, we will treat sending metadata as a send message operation
         /// to make things simpler.
-        /// completionDelegate is invoked upon completion.
         /// </summary>
-        public void StartSendInitialMetadata(Metadata headers, AsyncCompletionDelegate<object> completionDelegate)
+        public Task SendInitialMetadataAsync(Metadata headers)
         {
             lock (myLock)
             {
                 GrpcPreconditions.CheckNotNull(headers, "metadata");
-                GrpcPreconditions.CheckNotNull(completionDelegate, "Completion delegate cannot be null");
 
+                GrpcPreconditions.CheckState(started);
                 GrpcPreconditions.CheckState(!initialMetadataSent, "Response headers can only be sent once per call.");
                 GrpcPreconditions.CheckState(streamingWritesCounter == 0, "Response headers can only be sent before the first write starts.");
-                CheckSendingAllowed(allowFinished: false);
 
-                GrpcPreconditions.CheckNotNull(completionDelegate, "Completion delegate cannot be null");
+                var earlyResult = CheckSendAllowedOrEarlyResult();
+                if (earlyResult != null)
+                {
+                    return earlyResult;
+                }
 
                 using (var metadataArray = MetadataArraySafeHandle.Create(headers))
                 {
@@ -131,7 +132,8 @@
                 }
 
                 this.initialMetadataSent = true;
-                sendCompletionDelegate = completionDelegate;
+                streamingWriteTcs = new TaskCompletionSource<object>();
+                return streamingWriteTcs.Task;
             }
         }
 
@@ -196,6 +198,16 @@
             server.RemoveCallReference(this);
         }
 
+        protected override Task CheckSendAllowedOrEarlyResult()
+        {
+            GrpcPreconditions.CheckState(!halfcloseRequested, "Response stream has already been completed.");
+            GrpcPreconditions.CheckState(!finished, "Already finished.");
+            GrpcPreconditions.CheckState(streamingWriteTcs == null, "Only one write can be pending at a time");
+            GrpcPreconditions.CheckState(!disposed);
+
+            return null;
+        }
+
         /// <summary>
         /// Handles the server side close completion.
         /// </summary>
diff --git a/src/csharp/Grpc.Core/Internal/AsyncCompletion.cs b/src/csharp/Grpc.Core/Internal/AsyncCompletion.cs
deleted file mode 100644
index 7e86fdd..0000000
--- a/src/csharp/Grpc.Core/Internal/AsyncCompletion.cs
+++ /dev/null
@@ -1,94 +0,0 @@
-#region Copyright notice and license
-
-// Copyright 2015, Google Inc.
-// All rights reserved.
-// 
-// Redistribution and use in source and binary forms, with or without
-// modification, are permitted provided that the following conditions are
-// met:
-// 
-//     * Redistributions of source code must retain the above copyright
-// notice, this list of conditions and the following disclaimer.
-//     * Redistributions in binary form must reproduce the above
-// copyright notice, this list of conditions and the following disclaimer
-// in the documentation and/or other materials provided with the
-// distribution.
-//     * Neither the name of Google Inc. nor the names of its
-// contributors may be used to endorse or promote products derived from
-// this software without specific prior written permission.
-// 
-// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
-#endregion
-
-using System;
-using System.Diagnostics;
-using System.Runtime.CompilerServices;
-using System.Runtime.InteropServices;
-using System.Threading;
-using System.Threading.Tasks;
-using Grpc.Core.Internal;
-using Grpc.Core.Utils;
-
-namespace Grpc.Core.Internal
-{
-    /// <summary>
-    /// If error != null, there's been an error or operation has been cancelled.
-    /// </summary>
-    internal delegate void AsyncCompletionDelegate<T>(T result, Exception error);
-
-    /// <summary>
-    /// Helper for transforming AsyncCompletionDelegate into full-fledged Task.
-    /// </summary>
-    internal class AsyncCompletionTaskSource<T>
-    {
-        readonly TaskCompletionSource<T> tcs = new TaskCompletionSource<T>();
-        readonly AsyncCompletionDelegate<T> completionDelegate;
-
-        public AsyncCompletionTaskSource()
-        {
-            completionDelegate = new AsyncCompletionDelegate<T>(HandleCompletion);
-        }
-
-        public Task<T> Task
-        {
-            get
-            {
-                return tcs.Task;
-            }
-        }
-
-        public AsyncCompletionDelegate<T> CompletionDelegate
-        {
-            get
-            {
-                return completionDelegate;
-            }
-        }
-
-        private void HandleCompletion(T value, Exception error)
-        {
-            if (error == null)
-            {
-                tcs.SetResult(value);
-                return;
-            }
-            if (error is OperationCanceledException)
-            {
-                tcs.SetCanceled();
-                return;
-            }
-            tcs.SetException(error);
-        }
-    }
-}
diff --git a/src/csharp/Grpc.Core/Internal/CallSafeHandle.cs b/src/csharp/Grpc.Core/Internal/CallSafeHandle.cs
index 244b97d..82361f5 100644
--- a/src/csharp/Grpc.Core/Internal/CallSafeHandle.cs
+++ b/src/csharp/Grpc.Core/Internal/CallSafeHandle.cs
@@ -47,16 +47,14 @@
         static readonly NativeMethods Native = NativeMethods.Get();
 
         const uint GRPC_WRITE_BUFFER_HINT = 1;
-        CompletionRegistry completionRegistry;
         CompletionQueueSafeHandle completionQueue;
 
         private CallSafeHandle()
         {
         }
 
-        public void Initialize(CompletionRegistry completionRegistry, CompletionQueueSafeHandle completionQueue)
+        public void Initialize(CompletionQueueSafeHandle completionQueue)
         {
-            this.completionRegistry = completionRegistry;
             this.completionQueue = completionQueue;
         }
 
@@ -70,7 +68,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient(), context.GetReceivedMessage(), context.GetReceivedInitialMetadata()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient(), context.GetReceivedMessage(), context.GetReceivedInitialMetadata()));
                 Native.grpcsharp_call_start_unary(this, ctx, payload, new UIntPtr((ulong)payload.Length), metadataArray, writeFlags)
                     .CheckOk();
             }
@@ -90,7 +88,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient(), context.GetReceivedMessage(), context.GetReceivedInitialMetadata()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient(), context.GetReceivedMessage(), context.GetReceivedInitialMetadata()));
                 Native.grpcsharp_call_start_client_streaming(this, ctx, metadataArray).CheckOk();
             }
         }
@@ -100,7 +98,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient()));
                 Native.grpcsharp_call_start_server_streaming(this, ctx, payload, new UIntPtr((ulong)payload.Length), metadataArray, writeFlags).CheckOk();
             }
         }
@@ -110,7 +108,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedStatusOnClient()));
                 Native.grpcsharp_call_start_duplex_streaming(this, ctx, metadataArray).CheckOk();
             }
         }
@@ -120,7 +118,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
                 Native.grpcsharp_call_send_message(this, ctx, payload, new UIntPtr((ulong)payload.Length), writeFlags, sendEmptyInitialMetadata).CheckOk();
             }
         }
@@ -130,7 +128,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
                 Native.grpcsharp_call_send_close_from_client(this, ctx).CheckOk();
             }
         }
@@ -142,7 +140,7 @@
             {
                 var ctx = BatchContextSafeHandle.Create();
                 var optionalPayloadLength = optionalPayload != null ? new UIntPtr((ulong)optionalPayload.Length) : UIntPtr.Zero;
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
                 Native.grpcsharp_call_send_status_from_server(this, ctx, status.StatusCode, status.Detail, metadataArray, sendEmptyInitialMetadata,
                     optionalPayload, optionalPayloadLength, writeFlags).CheckOk();
             }
@@ -153,7 +151,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedMessage()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedMessage()));
                 Native.grpcsharp_call_recv_message(this, ctx).CheckOk();
             }
         }
@@ -163,7 +161,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedInitialMetadata()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedInitialMetadata()));
                 Native.grpcsharp_call_recv_initial_metadata(this, ctx).CheckOk();
             }
         }
@@ -173,7 +171,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedCloseOnServerCancelled()));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success, context.GetReceivedCloseOnServerCancelled()));
                 Native.grpcsharp_call_start_serverside(this, ctx).CheckOk();
             }
         }
@@ -183,7 +181,7 @@
             using (completionQueue.NewScope())
             {
                 var ctx = BatchContextSafeHandle.Create();
-                completionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
+                completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, (success, context) => callback(success));
                 Native.grpcsharp_call_send_initial_metadata(this, ctx, metadataArray).CheckOk();
             }
         }
diff --git a/src/csharp/Grpc.Core/Internal/ChannelSafeHandle.cs b/src/csharp/Grpc.Core/Internal/ChannelSafeHandle.cs
index 1dbd1f4..62864df 100644
--- a/src/csharp/Grpc.Core/Internal/ChannelSafeHandle.cs
+++ b/src/csharp/Grpc.Core/Internal/ChannelSafeHandle.cs
@@ -63,7 +63,7 @@
             return Native.grpcsharp_secure_channel_create(credentials, target, channelArgs);
         }
 
-        public CallSafeHandle CreateCall(CompletionRegistry registry, CallSafeHandle parentCall, ContextPropagationFlags propagationMask, CompletionQueueSafeHandle cq, string method, string host, Timespec deadline, CallCredentialsSafeHandle credentials)
+        public CallSafeHandle CreateCall(CallSafeHandle parentCall, ContextPropagationFlags propagationMask, CompletionQueueSafeHandle cq, string method, string host, Timespec deadline, CallCredentialsSafeHandle credentials)
         {
             using (Profilers.ForCurrentThread().NewScope("ChannelSafeHandle.CreateCall"))
             {
@@ -72,7 +72,7 @@
                 {
                     result.SetCredentials(credentials);
                 }
-                result.Initialize(registry, cq);
+                result.Initialize(cq);
                 return result;
             }
         }
@@ -82,11 +82,10 @@
             return Native.grpcsharp_channel_check_connectivity_state(this, tryToConnect ? 1 : 0);
         }
 
-        public void WatchConnectivityState(ChannelState lastObservedState, Timespec deadline, CompletionQueueSafeHandle cq,
-            CompletionRegistry completionRegistry, BatchCompletionDelegate callback)
+        public void WatchConnectivityState(ChannelState lastObservedState, Timespec deadline, CompletionQueueSafeHandle cq, BatchCompletionDelegate callback)
         {
             var ctx = BatchContextSafeHandle.Create();
-            completionRegistry.RegisterBatchCompletion(ctx, callback);
+            cq.CompletionRegistry.RegisterBatchCompletion(ctx, callback);
             Native.grpcsharp_channel_watch_connectivity_state(this, lastObservedState, deadline, cq, ctx);
         }
 
diff --git a/src/csharp/Grpc.Core/Internal/ClientRequestStream.cs b/src/csharp/Grpc.Core/Internal/ClientRequestStream.cs
index 013f00f..924de02 100644
--- a/src/csharp/Grpc.Core/Internal/ClientRequestStream.cs
+++ b/src/csharp/Grpc.Core/Internal/ClientRequestStream.cs
@@ -50,16 +50,12 @@
 
         public Task WriteAsync(TRequest message)
         {
-            var taskSource = new AsyncCompletionTaskSource<object>();
-            call.StartSendMessage(message, GetWriteFlags(), taskSource.CompletionDelegate);
-            return taskSource.Task;
+            return call.SendMessageAsync(message, GetWriteFlags());
         }
 
         public Task CompleteAsync()
         {
-            var taskSource = new AsyncCompletionTaskSource<object>();
-            call.StartSendCloseFromClient(taskSource.CompletionDelegate);
-            return taskSource.Task;
+            return call.SendCloseFromClientAsync();
         }
 
         public WriteOptions WriteOptions
diff --git a/src/csharp/Grpc.Core/Internal/CompletionQueueSafeHandle.cs b/src/csharp/Grpc.Core/Internal/CompletionQueueSafeHandle.cs
index 91364cd..46f5624 100644
--- a/src/csharp/Grpc.Core/Internal/CompletionQueueSafeHandle.cs
+++ b/src/csharp/Grpc.Core/Internal/CompletionQueueSafeHandle.cs
@@ -45,6 +45,7 @@
         static readonly NativeMethods Native = NativeMethods.Get();
 
         AtomicCounter shutdownRefcount = new AtomicCounter(1);
+        CompletionRegistry completionRegistry;
 
         private CompletionQueueSafeHandle()
         {
@@ -53,7 +54,13 @@
         public static CompletionQueueSafeHandle Create()
         {
             return Native.grpcsharp_completion_queue_create();
+        }
 
+        public static CompletionQueueSafeHandle Create(CompletionRegistry completionRegistry)
+        {
+            var cq = Native.grpcsharp_completion_queue_create();
+            cq.completionRegistry = completionRegistry;
+            return cq;
         }
 
         public CompletionQueueEvent Next()
@@ -83,6 +90,15 @@
             DecrementShutdownRefcount();
         }
 
+        /// <summary>
+        /// Completion registry associated with this completion queue.
+        /// Doesn't need to be set if only using Pluck() operations.
+        /// </summary>
+        public CompletionRegistry CompletionRegistry
+        {
+            get { return completionRegistry; }
+        }
+
         protected override bool ReleaseHandle()
         {
             Native.grpcsharp_completion_queue_destroy(handle);
diff --git a/src/csharp/Grpc.Core/Internal/GrpcThreadPool.cs b/src/csharp/Grpc.Core/Internal/GrpcThreadPool.cs
index b538726..4de543b 100644
--- a/src/csharp/Grpc.Core/Internal/GrpcThreadPool.cs
+++ b/src/csharp/Grpc.Core/Internal/GrpcThreadPool.cs
@@ -33,15 +33,15 @@
 
 using System;
 using System.Collections.Generic;
-using System.Runtime.InteropServices;
+using System.Linq;
 using System.Threading;
-using System.Threading.Tasks;
 using Grpc.Core.Logging;
+using Grpc.Core.Utils;
 
 namespace Grpc.Core.Internal
 {
     /// <summary>
-    /// Pool of threads polling on the same completion queue.
+    /// Pool of threads polling on a set of completions queues.
     /// </summary>
     internal class GrpcThreadPool
     {
@@ -51,25 +51,31 @@
         readonly object myLock = new object();
         readonly List<Thread> threads = new List<Thread>();
         readonly int poolSize;
+        readonly int completionQueueCount;
 
-        CompletionQueueSafeHandle cq;
+        IReadOnlyCollection<CompletionQueueSafeHandle> completionQueues;
 
-        public GrpcThreadPool(GrpcEnvironment environment, int poolSize)
+        /// <summary>
+        /// Creates a thread pool threads polling on a set of completions queues.
+        /// </summary>
+        /// <param name="environment">Environment.</param>
+        /// <param name="poolSize">Pool size.</param>
+        /// <param name="completionQueueCount">Completion queue count.</param>
+        public GrpcThreadPool(GrpcEnvironment environment, int poolSize, int completionQueueCount)
         {
             this.environment = environment;
             this.poolSize = poolSize;
+            this.completionQueueCount = completionQueueCount;
+            GrpcPreconditions.CheckArgument(poolSize >= completionQueueCount,
+                "Thread pool size cannot be smaller than the number of completion queues used.");
         }
 
         public void Start()
         {
             lock (myLock)
             {
-                if (cq != null)
-                {
-                    throw new InvalidOperationException("Already started.");
-                }
-
-                cq = CompletionQueueSafeHandle.Create();
+                GrpcPreconditions.CheckState(completionQueues == null, "Already started.");
+                completionQueues = CreateCompletionQueueList(environment, completionQueueCount);
 
                 for (int i = 0; i < poolSize; i++)
                 {
@@ -82,37 +88,48 @@
         {
             lock (myLock)
             {
-                cq.Shutdown();
+                foreach (var cq in completionQueues)
+                {
+                    cq.Shutdown();
+                }
+
                 foreach (var thread in threads)
                 {
                     thread.Join();
                 }
 
-                cq.Dispose();
+                foreach (var cq in completionQueues)
+                {
+                    cq.Dispose();
+                }
             }
         }
 
-        internal CompletionQueueSafeHandle CompletionQueue
+        internal IReadOnlyCollection<CompletionQueueSafeHandle> CompletionQueues
         {
             get
             {
-                return cq;
+                return completionQueues;
             }
         }
 
-        private Thread CreateAndStartThread(int i)
+        private Thread CreateAndStartThread(int threadIndex)
         {
-            var thread = new Thread(new ThreadStart(RunHandlerLoop));
+            var cqIndex = threadIndex % completionQueues.Count;
+            var cq = completionQueues.ElementAt(cqIndex);
+
+            var thread = new Thread(new ThreadStart(() => RunHandlerLoop(cq)));
             thread.IsBackground = false;
+            thread.Name = string.Format("grpc {0} (cq {1})", threadIndex, cqIndex);
             thread.Start();
-            thread.Name = "grpc " + i;
+
             return thread;
         }
 
         /// <summary>
         /// Body of the polling thread.
         /// </summary>
-        private void RunHandlerLoop()
+        private void RunHandlerLoop(CompletionQueueSafeHandle cq)
         {
             CompletionQueueEvent ev;
             do
@@ -124,7 +141,7 @@
                     IntPtr tag = ev.tag;
                     try
                     {
-                        var callback = environment.CompletionRegistry.Extract(tag);
+                        var callback = cq.CompletionRegistry.Extract(tag);
                         callback(success);
                     }
                     catch (Exception e)
@@ -135,5 +152,16 @@
             }
             while (ev.type != CompletionQueueEvent.CompletionType.Shutdown);
         }
+
+        private static IReadOnlyCollection<CompletionQueueSafeHandle> CreateCompletionQueueList(GrpcEnvironment environment, int completionQueueCount)
+        {
+            var list = new List<CompletionQueueSafeHandle>();
+            for (int i = 0; i < completionQueueCount; i++)
+            {
+                var completionRegistry = new CompletionRegistry(environment);
+                list.Add(CompletionQueueSafeHandle.Create(completionRegistry));
+            }
+            return list.AsReadOnly();
+        }
     }
 }
diff --git a/src/csharp/Grpc.Core/Internal/NativeMethods.cs b/src/csharp/Grpc.Core/Internal/NativeMethods.cs
index 42fd4d4..65607ed 100644
--- a/src/csharp/Grpc.Core/Internal/NativeMethods.cs
+++ b/src/csharp/Grpc.Core/Internal/NativeMethods.cs
@@ -137,6 +137,7 @@
         public readonly Delegates.grpcsharp_server_credentials_release_delegate grpcsharp_server_credentials_release;
 
         public readonly Delegates.grpcsharp_server_create_delegate grpcsharp_server_create;
+        public readonly Delegates.grpcsharp_server_register_completion_queue_delegate grpcsharp_server_register_completion_queue;
         public readonly Delegates.grpcsharp_server_add_insecure_http2_port_delegate grpcsharp_server_add_insecure_http2_port;
         public readonly Delegates.grpcsharp_server_add_secure_http2_port_delegate grpcsharp_server_add_secure_http2_port;
         public readonly Delegates.grpcsharp_server_start_delegate grpcsharp_server_start;
@@ -244,6 +245,7 @@
                 this.grpcsharp_server_credentials_release = GetMethodDelegate<Delegates.grpcsharp_server_credentials_release_delegate>(library);
 
                 this.grpcsharp_server_create = GetMethodDelegate<Delegates.grpcsharp_server_create_delegate>(library);
+                this.grpcsharp_server_register_completion_queue = GetMethodDelegate<Delegates.grpcsharp_server_register_completion_queue_delegate>(library);
                 this.grpcsharp_server_add_insecure_http2_port = GetMethodDelegate<Delegates.grpcsharp_server_add_insecure_http2_port_delegate>(library);
                 this.grpcsharp_server_add_secure_http2_port = GetMethodDelegate<Delegates.grpcsharp_server_add_secure_http2_port_delegate>(library);
                 this.grpcsharp_server_start = GetMethodDelegate<Delegates.grpcsharp_server_start_delegate>(library);
@@ -348,6 +350,7 @@
                 this.grpcsharp_server_credentials_release = PInvokeMethods.grpcsharp_server_credentials_release;
 
                 this.grpcsharp_server_create = PInvokeMethods.grpcsharp_server_create;
+                this.grpcsharp_server_register_completion_queue = PInvokeMethods.grpcsharp_server_register_completion_queue;
                 this.grpcsharp_server_add_insecure_http2_port = PInvokeMethods.grpcsharp_server_add_insecure_http2_port;
                 this.grpcsharp_server_add_secure_http2_port = PInvokeMethods.grpcsharp_server_add_secure_http2_port;
                 this.grpcsharp_server_start = PInvokeMethods.grpcsharp_server_start;
@@ -493,7 +496,8 @@
             public delegate ServerCredentialsSafeHandle grpcsharp_ssl_server_credentials_create_delegate(string pemRootCerts, string[] keyCertPairCertChainArray, string[] keyCertPairPrivateKeyArray, UIntPtr numKeyCertPairs, bool forceClientAuth);
             public delegate void grpcsharp_server_credentials_release_delegate(IntPtr credentials);
 
-            public delegate ServerSafeHandle grpcsharp_server_create_delegate(CompletionQueueSafeHandle cq, ChannelArgsSafeHandle args);
+            public delegate ServerSafeHandle grpcsharp_server_create_delegate(ChannelArgsSafeHandle args);
+            public delegate void grpcsharp_server_register_completion_queue_delegate(ServerSafeHandle server, CompletionQueueSafeHandle cq);
             public delegate int grpcsharp_server_add_insecure_http2_port_delegate(ServerSafeHandle server, string addr);
             public delegate int grpcsharp_server_add_secure_http2_port_delegate(ServerSafeHandle server, string addr, ServerCredentialsSafeHandle creds);
             public delegate void grpcsharp_server_start_delegate(ServerSafeHandle server);
@@ -773,7 +777,10 @@
             // ServerSafeHandle
 
             [DllImport("grpc_csharp_ext.dll")]
-            public static extern ServerSafeHandle grpcsharp_server_create(CompletionQueueSafeHandle cq, ChannelArgsSafeHandle args);
+            public static extern ServerSafeHandle grpcsharp_server_create(ChannelArgsSafeHandle args);
+
+            [DllImport("grpc_csharp_ext.dll")]
+            public static extern void grpcsharp_server_register_completion_queue(ServerSafeHandle server, CompletionQueueSafeHandle cq);
 
             [DllImport("grpc_csharp_ext.dll")]
             public static extern int grpcsharp_server_add_insecure_http2_port(ServerSafeHandle server, string addr);
diff --git a/src/csharp/Grpc.Core/Internal/ServerCallHandler.cs b/src/csharp/Grpc.Core/Internal/ServerCallHandler.cs
index febebba..6a2f520 100644
--- a/src/csharp/Grpc.Core/Internal/ServerCallHandler.cs
+++ b/src/csharp/Grpc.Core/Internal/ServerCallHandler.cs
@@ -44,7 +44,7 @@
 {
     internal interface IServerCallHandler
     {
-        Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment);
+        Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq);
     }
 
     internal class UnaryServerCallHandler<TRequest, TResponse> : IServerCallHandler
@@ -62,14 +62,14 @@
             this.handler = handler;
         }
 
-        public async Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment)
+        public async Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             var asyncCall = new AsyncCallServer<TRequest, TResponse>(
                 method.ResponseMarshaller.Serializer,
                 method.RequestMarshaller.Deserializer,
-                environment, newRpc.Server);
+                newRpc.Server);
 
-            asyncCall.Initialize(newRpc.Call);
+            asyncCall.Initialize(newRpc.Call, cq);
             var finishedTask = asyncCall.ServerSideCallAsync();
             var requestStream = new ServerRequestStream<TRequest, TResponse>(asyncCall);
             var responseStream = new ServerResponseStream<TRequest, TResponse>(asyncCall);
@@ -121,14 +121,14 @@
             this.handler = handler;
         }
 
-        public async Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment)
+        public async Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             var asyncCall = new AsyncCallServer<TRequest, TResponse>(
                 method.ResponseMarshaller.Serializer,
                 method.RequestMarshaller.Deserializer,
-                environment, newRpc.Server);
+                newRpc.Server);
 
-            asyncCall.Initialize(newRpc.Call);
+            asyncCall.Initialize(newRpc.Call, cq);
             var finishedTask = asyncCall.ServerSideCallAsync();
             var requestStream = new ServerRequestStream<TRequest, TResponse>(asyncCall);
             var responseStream = new ServerResponseStream<TRequest, TResponse>(asyncCall);
@@ -179,14 +179,14 @@
             this.handler = handler;
         }
 
-        public async Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment)
+        public async Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             var asyncCall = new AsyncCallServer<TRequest, TResponse>(
                 method.ResponseMarshaller.Serializer,
                 method.RequestMarshaller.Deserializer,
-                environment, newRpc.Server);
+                newRpc.Server);
 
-            asyncCall.Initialize(newRpc.Call);
+            asyncCall.Initialize(newRpc.Call, cq);
             var finishedTask = asyncCall.ServerSideCallAsync();
             var requestStream = new ServerRequestStream<TRequest, TResponse>(asyncCall);
             var responseStream = new ServerResponseStream<TRequest, TResponse>(asyncCall);
@@ -237,14 +237,14 @@
             this.handler = handler;
         }
 
-        public async Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment)
+        public async Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             var asyncCall = new AsyncCallServer<TRequest, TResponse>(
                 method.ResponseMarshaller.Serializer,
                 method.RequestMarshaller.Deserializer,
-                environment, newRpc.Server);
+                newRpc.Server);
 
-            asyncCall.Initialize(newRpc.Call);
+            asyncCall.Initialize(newRpc.Call, cq);
             var finishedTask = asyncCall.ServerSideCallAsync();
             var requestStream = new ServerRequestStream<TRequest, TResponse>(asyncCall);
             var responseStream = new ServerResponseStream<TRequest, TResponse>(asyncCall);
@@ -281,13 +281,13 @@
     {
         public static readonly NoSuchMethodCallHandler Instance = new NoSuchMethodCallHandler();
 
-        public async Task HandleCall(ServerRpcNew newRpc, GrpcEnvironment environment)
+        public async Task HandleCall(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             // We don't care about the payload type here.
             var asyncCall = new AsyncCallServer<byte[], byte[]>(
-                (payload) => payload, (payload) => payload, environment, newRpc.Server);
+                (payload) => payload, (payload) => payload, newRpc.Server);
             
-            asyncCall.Initialize(newRpc.Call);
+            asyncCall.Initialize(newRpc.Call, cq);
             var finishedTask = asyncCall.ServerSideCallAsync();
             await asyncCall.SendStatusFromServerAsync(new Status(StatusCode.Unimplemented, ""), Metadata.Empty, null).ConfigureAwait(false);
             await finishedTask.ConfigureAwait(false);
diff --git a/src/csharp/Grpc.Core/Internal/ServerResponseStream.cs b/src/csharp/Grpc.Core/Internal/ServerResponseStream.cs
index ecfee0b..25b79b4 100644
--- a/src/csharp/Grpc.Core/Internal/ServerResponseStream.cs
+++ b/src/csharp/Grpc.Core/Internal/ServerResponseStream.cs
@@ -52,16 +52,12 @@
 
         public Task WriteAsync(TResponse message)
         {
-            var taskSource = new AsyncCompletionTaskSource<object>();
-            call.StartSendMessage(message, GetWriteFlags(), taskSource.CompletionDelegate);
-            return taskSource.Task;
+            return call.SendMessageAsync(message, GetWriteFlags());
         }
 
         public Task WriteResponseHeadersAsync(Metadata responseHeaders)
         {
-            var taskSource = new AsyncCompletionTaskSource<object>();
-            call.StartSendInitialMetadata(responseHeaders, taskSource.CompletionDelegate);
-            return taskSource.Task;
+            return call.SendInitialMetadataAsync(responseHeaders);
         }
 
         public WriteOptions WriteOptions
diff --git a/src/csharp/Grpc.Core/Internal/ServerSafeHandle.cs b/src/csharp/Grpc.Core/Internal/ServerSafeHandle.cs
index 6b5f70e..8581302 100644
--- a/src/csharp/Grpc.Core/Internal/ServerSafeHandle.cs
+++ b/src/csharp/Grpc.Core/Internal/ServerSafeHandle.cs
@@ -31,12 +31,6 @@
 
 #endregion
 
-using System;
-using System.Collections.Concurrent;
-using System.Diagnostics;
-using System.Runtime.InteropServices;
-using Grpc.Core.Utils;
-
 namespace Grpc.Core.Internal
 {
     /// <summary>
@@ -50,12 +44,17 @@
         {
         }
 
-        public static ServerSafeHandle NewServer(CompletionQueueSafeHandle cq, ChannelArgsSafeHandle args)
+        public static ServerSafeHandle NewServer(ChannelArgsSafeHandle args)
         {
             // Increment reference count for the native gRPC environment to make sure we don't do grpc_shutdown() before destroying the server handle.
             // Doing so would make object finalizer crash if we end up abandoning the handle.
             GrpcEnvironment.GrpcNativeInit();
-            return Native.grpcsharp_server_create(cq, args);
+            return Native.grpcsharp_server_create(args);
+        }
+
+        public void RegisterCompletionQueue(CompletionQueueSafeHandle cq)
+        {
+            Native.grpcsharp_server_register_completion_queue(this, cq);
         }
 
         public int AddInsecurePort(string addr)
@@ -73,18 +72,18 @@
             Native.grpcsharp_server_start(this);
         }
     
-        public void ShutdownAndNotify(BatchCompletionDelegate callback, GrpcEnvironment environment)
+        public void ShutdownAndNotify(BatchCompletionDelegate callback, CompletionQueueSafeHandle completionQueue)
         {
             var ctx = BatchContextSafeHandle.Create();
-            environment.CompletionRegistry.RegisterBatchCompletion(ctx, callback);
-            Native.grpcsharp_server_shutdown_and_notify_callback(this, environment.CompletionQueue, ctx);
+            completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, callback);
+            Native.grpcsharp_server_shutdown_and_notify_callback(this, completionQueue, ctx);
         }
 
-        public void RequestCall(BatchCompletionDelegate callback, GrpcEnvironment environment)
+        public void RequestCall(BatchCompletionDelegate callback, CompletionQueueSafeHandle completionQueue)
         {
             var ctx = BatchContextSafeHandle.Create();
-            environment.CompletionRegistry.RegisterBatchCompletion(ctx, callback);
-            Native.grpcsharp_server_request_call(this, environment.CompletionQueue, ctx).CheckOk();
+            completionQueue.CompletionRegistry.RegisterBatchCompletion(ctx, callback);
+            Native.grpcsharp_server_request_call(this, completionQueue, ctx).CheckOk();
         }
 
         protected override bool ReleaseHandle()
diff --git a/src/csharp/Grpc.Core/Server.cs b/src/csharp/Grpc.Core/Server.cs
index d538a46..069185e 100644
--- a/src/csharp/Grpc.Core/Server.cs
+++ b/src/csharp/Grpc.Core/Server.cs
@@ -34,8 +34,7 @@
 using System;
 using System.Collections;
 using System.Collections.Generic;
-using System.Diagnostics;
-using System.Runtime.InteropServices;
+using System.Linq;
 using System.Threading.Tasks;
 using Grpc.Core.Internal;
 using Grpc.Core.Logging;
@@ -48,7 +47,7 @@
     /// </summary>
     public class Server
     {
-        const int InitialAllowRpcTokenCount = 10;
+        const int InitialAllowRpcTokenCountPerCq = 10;
         static readonly ILogger Logger = GrpcEnvironment.Logger.ForType<Server>();
 
         readonly AtomicCounter activeCallCounter = new AtomicCounter();
@@ -80,7 +79,12 @@
             this.options = options != null ? new List<ChannelOption>(options) : new List<ChannelOption>();
             using (var channelArgs = ChannelOptions.CreateChannelArgs(this.options))
             {
-                this.handle = ServerSafeHandle.NewServer(environment.CompletionQueue, channelArgs);
+                this.handle = ServerSafeHandle.NewServer(channelArgs);
+            }
+
+            foreach (var cq in environment.CompletionQueues)
+            {
+                this.handle.RegisterCompletionQueue(cq);
             }
         }
 
@@ -133,9 +137,12 @@
 
                 // Starting with more than one AllowOneRpc tokens can significantly increase
                 // unary RPC throughput.
-                for (int i = 0; i < InitialAllowRpcTokenCount; i++)
+                for (int i = 0; i < InitialAllowRpcTokenCountPerCq; i++)
                 {
-                    AllowOneRpc();
+                    foreach (var cq in environment.CompletionQueues)
+                    {
+                        AllowOneRpc(cq);
+                    }
                 }
             }
         }
@@ -154,7 +161,8 @@
                 shutdownRequested = true;
             }
 
-            handle.ShutdownAndNotify(HandleServerShutdown, environment);
+            var cq = environment.CompletionQueues.First();  // any cq will do
+            handle.ShutdownAndNotify(HandleServerShutdown, cq);
             await shutdownTcs.Task.ConfigureAwait(false);
             DisposeHandle();
 
@@ -174,7 +182,8 @@
                 shutdownRequested = true;
             }
 
-            handle.ShutdownAndNotify(HandleServerShutdown, environment);
+            var cq = environment.CompletionQueues.First();  // any cq will do
+            handle.ShutdownAndNotify(HandleServerShutdown, cq);
             handle.CancelAllCalls();
             await shutdownTcs.Task.ConfigureAwait(false);
             DisposeHandle();
@@ -244,11 +253,11 @@
         /// <summary>
         /// Allows one new RPC call to be received by server.
         /// </summary>
-        private void AllowOneRpc()
+        private void AllowOneRpc(CompletionQueueSafeHandle cq)
         {
             if (!shutdownRequested)
             {
-                handle.RequestCall(HandleNewServerRpc, environment);
+                handle.RequestCall((success, ctx) => HandleNewServerRpc(success, ctx, cq), cq);
             }
         }
 
@@ -265,7 +274,7 @@
         /// <summary>
         /// Selects corresponding handler for given call and handles the call.
         /// </summary>
-        private async Task HandleCallAsync(ServerRpcNew newRpc)
+        private async Task HandleCallAsync(ServerRpcNew newRpc, CompletionQueueSafeHandle cq)
         {
             try
             {
@@ -274,7 +283,7 @@
                 {
                     callHandler = NoSuchMethodCallHandler.Instance;
                 }
-                await callHandler.HandleCall(newRpc, environment).ConfigureAwait(false);
+                await callHandler.HandleCall(newRpc, cq).ConfigureAwait(false);
             }
             catch (Exception e)
             {
@@ -285,9 +294,9 @@
         /// <summary>
         /// Handles the native callback.
         /// </summary>
-        private void HandleNewServerRpc(bool success, BatchContextSafeHandle ctx)
+        private void HandleNewServerRpc(bool success, BatchContextSafeHandle ctx, CompletionQueueSafeHandle cq)
         {
-			Task.Run(() => AllowOneRpc());
+			Task.Run(() => AllowOneRpc(cq));
 
             if (success)
             {
@@ -296,7 +305,7 @@
                 // after server shutdown, the callback returns with null call
                 if (!newRpc.Call.IsInvalid)
                 {
-                    HandleCallAsync(newRpc);  // we don't need to await.
+                    HandleCallAsync(newRpc, cq);  // we don't need to await.
                 }
             }
         }
diff --git a/src/csharp/Grpc.Core/WriteOptions.cs b/src/csharp/Grpc.Core/WriteOptions.cs
index 7523ada..4c9706d 100644
--- a/src/csharp/Grpc.Core/WriteOptions.cs
+++ b/src/csharp/Grpc.Core/WriteOptions.cs
@@ -1,6 +1,6 @@
 #region Copyright notice and license
 
-// Copyright 2015, Google Inc.
+// Copyright 2015-2016, Google Inc.
 // All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
@@ -64,7 +64,7 @@
         /// </summary>
         public static readonly WriteOptions Default = new WriteOptions();
             
-        private WriteFlags flags;
+        private readonly WriteFlags flags;
 
         /// <summary>
         /// Initializes a new instance of <c>WriteOptions</c> class.
diff --git a/src/csharp/Grpc.IntegrationTesting/InteropClient.cs b/src/csharp/Grpc.IntegrationTesting/InteropClient.cs
index 1541cfd..aea40af 100644
--- a/src/csharp/Grpc.IntegrationTesting/InteropClient.cs
+++ b/src/csharp/Grpc.IntegrationTesting/InteropClient.cs
@@ -471,8 +471,16 @@
 
                 cts.Cancel();
 
-                var ex = Assert.ThrowsAsync<RpcException>(async () => await call.ResponseStream.MoveNext());
-                Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
+                try
+                {
+                    // cannot use Assert.ThrowsAsync because it uses Task.Wait and would deadlock.
+                    await call.ResponseStream.MoveNext();
+                    Assert.Fail();
+                }
+                catch (RpcException ex)
+                {
+                    Assert.AreEqual(StatusCode.Cancelled, ex.Status.StatusCode);
+                }
             }
             Console.WriteLine("Passed!");
         }
@@ -497,9 +505,16 @@
                     // Deadline was reached before write has started. Eat the exception and continue.
                 }
 
-                var ex = Assert.ThrowsAsync<RpcException>(async () => await call.ResponseStream.MoveNext());
-                // We can't guarantee the status code always DeadlineExceeded. See issue #2685.
-                Assert.Contains(ex.Status.StatusCode, new[] { StatusCode.DeadlineExceeded, StatusCode.Internal });
+                try
+                {
+                    await call.ResponseStream.MoveNext();
+                    Assert.Fail();
+                }
+                catch (RpcException ex)
+                {
+                    // We can't guarantee the status code always DeadlineExceeded. See issue #2685.
+                    Assert.Contains(ex.Status.StatusCode, new[] { StatusCode.DeadlineExceeded, StatusCode.Internal });
+                }
             }
             Console.WriteLine("Passed!");
         }
@@ -577,9 +592,17 @@
                 await call.RequestStream.WriteAsync(request);
                 await call.RequestStream.CompleteAsync();
 
-                var e = Assert.ThrowsAsync<RpcException>(async () => await call.ResponseStream.ToListAsync());
-                Assert.AreEqual(StatusCode.Unknown, e.Status.StatusCode);
-                Assert.AreEqual(echoStatus.Message, e.Status.Detail);
+                try
+                {
+                    // cannot use Assert.ThrowsAsync because it uses Task.Wait and would deadlock.
+                    await call.ResponseStream.ToListAsync();
+                    Assert.Fail();
+                }
+                catch (RpcException e)
+                {
+                    Assert.AreEqual(StatusCode.Unknown, e.Status.StatusCode);
+                    Assert.AreEqual(echoStatus.Message, e.Status.Detail);
+                }
             }
 
             Console.WriteLine("Passed!");
diff --git a/src/csharp/ext/grpc_csharp_ext.c b/src/csharp/ext/grpc_csharp_ext.c
index 5b8ff9b..4beef9d 100644
--- a/src/csharp/ext/grpc_csharp_ext.c
+++ b/src/csharp/ext/grpc_csharp_ext.c
@@ -806,11 +806,14 @@
 /* Server */
 
 GPR_EXPORT grpc_server *GPR_CALLTYPE
-grpcsharp_server_create(grpc_completion_queue *cq,
-                        const grpc_channel_args *args) {
-  grpc_server *server = grpc_server_create(args, NULL);
+grpcsharp_server_create(const grpc_channel_args *args) {
+  return grpc_server_create(args, NULL);
+}
+
+GPR_EXPORT void GPR_CALLTYPE
+grpcsharp_server_register_completion_queue(grpc_server *server,
+                                           grpc_completion_queue *cq) {
   grpc_server_register_completion_queue(server, cq, NULL);
-  return server;
 }
 
 GPR_EXPORT int32_t GPR_CALLTYPE
diff --git a/src/node/ext/byte_buffer.cc b/src/node/ext/byte_buffer.cc
index 8e0b691..3479a67 100644
--- a/src/node/ext/byte_buffer.cc
+++ b/src/node/ext/byte_buffer.cc
@@ -72,17 +72,13 @@
   if (buffer == NULL) {
     return scope.Escape(Nan::Null());
   }
-  size_t length = grpc_byte_buffer_length(buffer);
-  char *result = new char[length];
-  size_t offset = 0;
   grpc_byte_buffer_reader reader;
   grpc_byte_buffer_reader_init(&reader, buffer);
-  gpr_slice next;
-  while (grpc_byte_buffer_reader_next(&reader, &next) != 0) {
-    memcpy(result + offset, GPR_SLICE_START_PTR(next), GPR_SLICE_LENGTH(next));
-    offset += GPR_SLICE_LENGTH(next);
-    gpr_slice_unref(next);
-  }
+  gpr_slice slice = grpc_byte_buffer_reader_readall(&reader);
+  size_t length = GPR_SLICE_LENGTH(slice);
+  char *result = new char[length];
+  memcpy(result, GPR_SLICE_START_PTR(slice), length);
+  gpr_slice_unref(slice);
   return scope.Escape(MakeFastBuffer(
       Nan::NewBuffer(result, length, delete_buffer, NULL).ToLocalChecked()));
 }
diff --git a/src/objective-c/CronetFramework.podspec b/src/objective-c/CronetFramework.podspec
new file mode 100644
index 0000000..20af764
--- /dev/null
+++ b/src/objective-c/CronetFramework.podspec
@@ -0,0 +1,43 @@
+# Copyright 2016, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+#     * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+#     * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+#     * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+Pod::Spec.new do |s|
+  s.name         = "CronetFramework"
+  s.version      = "0.0.2"
+  s.summary      = "Cronet, precompiled and used as a framework."
+  s.homepage     = "http://chromium.org"
+  s.license      = { :type => 'BSD' }
+  s.vendored_framework = "Cronet.framework"
+  s.author             = "The Chromium Authors"
+  s.ios.deployment_target = "8.0"
+  s.source       = { :http => 'https://storage.googleapis.com/grpc-precompiled-binaries/cronet/Cronet.framework.zip' }
+  s.preserve_paths = "Cronet.framework"
+  s.public_header_files = "Cronet.framework/Headers/**/*{.h}"
+end
diff --git a/src/objective-c/GRPCClient/GRPCCall+Cronet.h b/src/objective-c/GRPCClient/GRPCCall+Cronet.h
new file mode 100644
index 0000000..7f7fc6c
--- /dev/null
+++ b/src/objective-c/GRPCClient/GRPCCall+Cronet.h
@@ -0,0 +1,55 @@
+/*
+ *
+ * Copyright 2016, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+#import <Cronet/Cronet.h>
+
+#import "GRPCCall.h"
+
+/**
+ * Methods for using cronet transport.
+ */
+@interface GRPCCall (Cronet)
+
+/**
+ * This method should be called before issuing the first RPC. It should be
+ * called only once. Create an instance of Cronet engine in your app elsewhere
+ * and pass the instance pointer in the cronet_engine parameter. Once set,
+ * all subsequent RPCs will use Cronet transport. The method is not thread
+ * safe.
+ */
++(void)useCronetWithEngine:(cronet_engine *)engine;
+
++(cronet_engine *)cronetEngine;
+
++(BOOL)isUsingCronet;
+
+@end
diff --git a/src/objective-c/GRPCClient/GRPCCall+Cronet.m b/src/objective-c/GRPCClient/GRPCCall+Cronet.m
new file mode 100644
index 0000000..69a410e
--- /dev/null
+++ b/src/objective-c/GRPCClient/GRPCCall+Cronet.m
@@ -0,0 +1,54 @@
+/*
+ *
+ * Copyright 2016, Google Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *     * Neither the name of Google Inc. nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#import "GRPCCall+Cronet.h"
+
+static BOOL useCronet = NO;
+static cronet_engine *globalCronetEngine;
+
+@implementation GRPCCall (Cronet)
+
++ (void)useCronetWithEngine:(cronet_engine *)engine {
+  useCronet = YES;
+  globalCronetEngine = engine;
+}
+
++ (cronet_engine *)cronetEngine {
+  return globalCronetEngine;
+}
+
++ (BOOL)isUsingCronet {
+  return useCronet;
+}
+
+@end
diff --git a/src/objective-c/GRPCClient/private/GRPCChannel.h b/src/objective-c/GRPCClient/private/GRPCChannel.h
index 70d1a9b..3219835 100644
--- a/src/objective-c/GRPCClient/private/GRPCChannel.h
+++ b/src/objective-c/GRPCClient/private/GRPCChannel.h
@@ -56,6 +56,12 @@
 + (nullable GRPCChannel *)secureChannelWithHost:(nonnull NSString *)host;
 
 /**
+ * Creates a secure channel to the specified @c host using Cronet as a transport mechanism.
+ */
++ (nullable GRPCChannel *)secureCronetChannelWithHost:(NSString *)host
+                                          channelArgs:(NSDictionary *)channelArgs;
+
+/**
  * Creates a secure channel to the specified @c host using the specified @c credentials and
  * @c channelArgs. Only in tests should @c GRPC_SSL_TARGET_NAME_OVERRIDE_ARG channel arg be set.
  */
diff --git a/src/objective-c/GRPCClient/private/GRPCChannel.m b/src/objective-c/GRPCClient/private/GRPCChannel.m
index 203ef58..e4e0dbe 100644
--- a/src/objective-c/GRPCClient/private/GRPCChannel.m
+++ b/src/objective-c/GRPCClient/private/GRPCChannel.m
@@ -34,10 +34,13 @@
 #import "GRPCChannel.h"
 
 #include <grpc/grpc_security.h>
+#include <grpc/grpc_cronet.h>
 #include <grpc/support/alloc.h>
 #include <grpc/support/log.h>
 #include <grpc/support/string_util.h>
 
+#import <Cronet/Cronet.h>
+#import <GRPCClient/GRPCCall+Cronet.h>
 #import "GRPCCompletionQueue.h"
 
 void freeChannelArgs(grpc_channel_args *channel_args) {
@@ -99,6 +102,22 @@
   grpc_channel_args *_channelArgs;
 }
 
+- (instancetype)initWithHost:(NSString *)host
+                cronetEngine:(cronet_engine *)cronetEngine
+                 channelArgs:(NSDictionary *)channelArgs {
+  if (!host) {
+    [NSException raise:NSInvalidArgumentException format:@"host argument missing"];
+  }
+
+  if (self = [super init]) {
+    _channelArgs = buildChannelArgs(channelArgs);
+    _host = [host copy];
+    _unmanagedChannel = grpc_cronet_secure_channel_create(cronetEngine, _host.UTF8String, _channelArgs,
+                                                     NULL);
+  }
+
+  return self;
+}
 
 - (instancetype)initWithHost:(NSString *)host
                       secure:(BOOL)secure
@@ -133,6 +152,17 @@
   freeChannelArgs(_channelArgs);
 }
 
++ (GRPCChannel *)secureCronetChannelWithHost:(NSString *)host
+                                 channelArgs:(NSDictionary *)channelArgs {
+  cronet_engine *engine = [GRPCCall cronetEngine];
+  if (!engine) {
+    [NSException raise:NSInvalidArgumentException
+                format:@"cronet_engine is NULL. Set it first."];
+    return nil;
+  }
+  return [[GRPCChannel alloc] initWithHost:host cronetEngine:engine channelArgs:channelArgs];
+}
+
 + (GRPCChannel *)secureChannelWithHost:(NSString *)host {
   return [[GRPCChannel alloc] initWithHost:host secure:YES credentials:NULL channelArgs:NULL];
 }
diff --git a/src/objective-c/GRPCClient/private/GRPCHost.m b/src/objective-c/GRPCClient/private/GRPCHost.m
index 43166cb..7da5088 100644
--- a/src/objective-c/GRPCClient/private/GRPCHost.m
+++ b/src/objective-c/GRPCClient/private/GRPCHost.m
@@ -37,6 +37,7 @@
 #include <grpc/grpc_security.h>
 #import <GRPCClient/GRPCCall.h>
 #import <GRPCClient/GRPCCall+ChannelArg.h>
+#import <GRPCClient/GRPCCall+Cronet.h>
 
 #import "GRPCChannel.h"
 #import "GRPCCompletionQueue.h"
@@ -200,15 +201,21 @@
 
 - (GRPCChannel *)newChannel {
   NSDictionary *args = [self channelArgs];
+  BOOL useCronet = [GRPCCall isUsingCronet];
   if (_secure) {
       GRPCChannel *channel;
       @synchronized(self) {
         if (_channelCreds == nil) {
           [self setTLSPEMRootCerts:nil withPrivateKey:nil withCertChain:nil error:nil];
         }
-        channel = [GRPCChannel secureChannelWithHost:_address
-                                          credentials:_channelCreds
-                                          channelArgs:args];
+        if (useCronet) {
+          channel = [GRPCChannel secureCronetChannelWithHost:_address
+                                                 channelArgs:args];
+        } else {
+          channel = [GRPCChannel secureChannelWithHost:_address
+                                            credentials:_channelCreds
+                                            channelArgs:args];
+        }
       }
       return channel;
   } else {
diff --git a/src/objective-c/tests/InteropTests.m b/src/objective-c/tests/InteropTests.m
index 4f096b9..781c500 100644
--- a/src/objective-c/tests/InteropTests.m
+++ b/src/objective-c/tests/InteropTests.m
@@ -35,7 +35,9 @@
 
 #include <grpc/status.h>
 
+#import <Cronet/Cronet.h>
 #import <GRPCClient/GRPCCall+Tests.h>
+#import <GRPCClient/GRPCCall+Cronet.h>
 #import <ProtoRPC/ProtoRPC.h>
 #import <RemoteTest/Empty.pbobjc.h>
 #import <RemoteTest/Messages.pbobjc.h>
@@ -78,6 +80,8 @@
 
 #pragma mark Tests
 
+static cronet_engine *cronetEngine = NULL;
+
 @implementation InteropTests {
   RMTTestService *_service;
 }
@@ -88,6 +92,15 @@
 
 - (void)setUp {
   _service = self.class.host ? [RMTTestService serviceWithHost:self.class.host] : nil;
+#ifdef GRPC_COMPILE_WITH_CRONET
+  if (cronetEngine == NULL) {
+    // Cronet setup
+    [Cronet setHttp2Enabled:YES];
+    [Cronet start];
+    cronetEngine = [Cronet getGlobalEngine];
+    [GRPCCall useCronetWithEngine:cronetEngine];
+  }
+#endif
 }
 
 - (void)testEmptyUnaryRPC {
@@ -245,6 +258,8 @@
   [self waitForExpectationsWithTimeout:4 handler:nil];
 }
 
+#ifndef GRPC_COMPILE_WITH_CRONET
+// TODO(makdharma@): Fix this test
 - (void)testEmptyStreamRPC {
   XCTAssertNotNil(self.class.host);
   __weak XCTestExpectation *expectation = [self expectationWithDescription:@"EmptyStream"];
@@ -258,6 +273,7 @@
   }];
   [self waitForExpectationsWithTimeout:2 handler:nil];
 }
+#endif
 
 - (void)testCancelAfterBeginRPC {
   XCTAssertNotNil(self.class.host);
diff --git a/src/objective-c/tests/Podfile b/src/objective-c/tests/Podfile
index 7ec7a25..508641d 100644
--- a/src/objective-c/tests/Podfile
+++ b/src/objective-c/tests/Podfile
@@ -3,6 +3,7 @@
 
 pod 'Protobuf', :path => "../../../third_party/protobuf"
 pod 'BoringSSL', :podspec => ".."
+pod 'CronetFramework', :podspec => ".."
 pod 'gRPC', :path => "../../.."
 pod 'RemoteTest', :path => "RemoteTestClient"
 
diff --git a/src/python/grpcio/grpc/__init__.py b/src/python/grpcio/grpc/__init__.py
index b844a14..8644731 100644
--- a/src/python/grpcio/grpc/__init__.py
+++ b/src/python/grpcio/grpc/__init__.py
@@ -27,5 +27,763 @@
 # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
+"""gRPC's Python API."""
+
 __import__('pkg_resources').declare_namespace(__name__)
 
+import abc
+import enum
+
+import six
+
+from grpc._cython import cygrpc as _cygrpc
+
+
+############################## Future Interface  ###############################
+
+
+class FutureTimeoutError(Exception):
+  """Indicates that a method call on a Future timed out."""
+
+
+class FutureCancelledError(Exception):
+  """Indicates that the computation underlying a Future was cancelled."""
+
+
+class Future(six.with_metaclass(abc.ABCMeta)):
+  """A representation of a computation in another control flow.
+
+  Computations represented by a Future may be yet to be begun, may be ongoing,
+  or may have already completed.
+  """
+
+  @abc.abstractmethod
+  def cancel(self):
+    """Attempts to cancel the computation.
+
+    This method does not block.
+
+    Returns:
+      True if the computation has not yet begun, will not be allowed to take
+        place, and determination of both was possible without blocking. False
+        under all other circumstances including but not limited to the
+        computation's already having begun, the computation's already having
+        finished, and the computation's having been scheduled for execution on a
+        remote system for which a determination of whether or not it commenced
+        before being cancelled cannot be made without blocking.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def cancelled(self):
+    """Describes whether the computation was cancelled.
+
+    This method does not block.
+
+    Returns:
+      True if the computation was cancelled any time before its result became
+        immediately available. False under all other circumstances including but
+        not limited to this object's cancel method not having been called and
+        the computation's result having become immediately available.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def running(self):
+    """Describes whether the computation is taking place.
+
+    This method does not block.
+
+    Returns:
+      True if the computation is scheduled to take place in the future or is
+        taking place now, or False if the computation took place in the past or
+        was cancelled.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def done(self):
+    """Describes whether the computation has taken place.
+
+    This method does not block.
+
+    Returns:
+      True if the computation is known to have either completed or have been
+        unscheduled or interrupted. False if the computation may possibly be
+        executing or scheduled to execute later.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def result(self, timeout=None):
+    """Accesses the outcome of the computation or raises its exception.
+
+    This method may return immediately or may block.
+
+    Args:
+      timeout: The length of time in seconds to wait for the computation to
+        finish or be cancelled, or None if this method should block until the
+        computation has finished or is cancelled no matter how long that takes.
+
+    Returns:
+      The return value of the computation.
+
+    Raises:
+      FutureTimeoutError: If a timeout value is passed and the computation does
+        not terminate within the allotted time.
+      FutureCancelledError: If the computation was cancelled.
+      Exception: If the computation raised an exception, this call will raise
+        the same exception.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def exception(self, timeout=None):
+    """Return the exception raised by the computation.
+
+    This method may return immediately or may block.
+
+    Args:
+      timeout: The length of time in seconds to wait for the computation to
+        terminate or be cancelled, or None if this method should block until
+        the computation is terminated or is cancelled no matter how long that
+        takes.
+
+    Returns:
+      The exception raised by the computation, or None if the computation did
+        not raise an exception.
+
+    Raises:
+      FutureTimeoutError: If a timeout value is passed and the computation does
+        not terminate within the allotted time.
+      FutureCancelledError: If the computation was cancelled.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def traceback(self, timeout=None):
+    """Access the traceback of the exception raised by the computation.
+
+    This method may return immediately or may block.
+
+    Args:
+      timeout: The length of time in seconds to wait for the computation to
+        terminate or be cancelled, or None if this method should block until
+        the computation is terminated or is cancelled no matter how long that
+        takes.
+
+    Returns:
+      The traceback of the exception raised by the computation, or None if the
+        computation did not raise an exception.
+
+    Raises:
+      FutureTimeoutError: If a timeout value is passed and the computation does
+        not terminate within the allotted time.
+      FutureCancelledError: If the computation was cancelled.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def add_done_callback(self, fn):
+    """Adds a function to be called at completion of the computation.
+
+    The callback will be passed this Future object describing the outcome of
+    the computation.
+
+    If the computation has already completed, the callback will be called
+    immediately.
+
+    Args:
+      fn: A callable taking this Future object as its single parameter.
+    """
+    raise NotImplementedError()
+
+
+################################  gRPC Enums  ##################################
+
+
+@enum.unique
+class ChannelConnectivity(enum.Enum):
+  """Mirrors grpc_connectivity_state in the gRPC Core.
+
+  Attributes:
+    IDLE: The channel is idle.
+    CONNECTING: The channel is connecting.
+    READY: The channel is ready to conduct RPCs.
+    TRANSIENT_FAILURE: The channel has seen a failure from which it expects to
+      recover.
+    FATAL_FAILURE: The channel has seen a failure from which it cannot recover.
+  """
+  IDLE              = (_cygrpc.ConnectivityState.idle, 'idle')
+  CONNECTING        = (_cygrpc.ConnectivityState.connecting, 'connecting')
+  READY             = (_cygrpc.ConnectivityState.ready, 'ready')
+  TRANSIENT_FAILURE = (
+      _cygrpc.ConnectivityState.transient_failure, 'transient failure')
+  FATAL_FAILURE     = (_cygrpc.ConnectivityState.fatal_failure, 'fatal failure')
+
+
+@enum.unique
+class StatusCode(enum.Enum):
+  """Mirrors grpc_status_code in the gRPC Core."""
+  OK                  = (_cygrpc.StatusCode.ok, 'ok')
+  CANCELLED           = (_cygrpc.StatusCode.cancelled, 'cancelled')
+  UNKNOWN             = (_cygrpc.StatusCode.unknown, 'unknown')
+  INVALID_ARGUMENT    = (
+      _cygrpc.StatusCode.invalid_argument, 'invalid argument')
+  DEADLINE_EXCEEDED   = (
+      _cygrpc.StatusCode.deadline_exceeded, 'deadline exceeded')
+  NOT_FOUND           = (_cygrpc.StatusCode.not_found, 'not found')
+  ALREADY_EXISTS      = (_cygrpc.StatusCode.already_exists, 'already exists')
+  PERMISSION_DENIED   = (
+      _cygrpc.StatusCode.permission_denied, 'permission denied')
+  RESOURCE_EXHAUSTED  = (
+      _cygrpc.StatusCode.resource_exhausted, 'resource exhausted')
+  FAILED_PRECONDITION = (
+      _cygrpc.StatusCode.failed_precondition, 'failed precondition')
+  ABORTED             = (_cygrpc.StatusCode.aborted, 'aborted')
+  OUT_OF_RANGE        = (_cygrpc.StatusCode.out_of_range, 'out of range')
+  UNIMPLEMENTED       = (_cygrpc.StatusCode.unimplemented, 'unimplemented')
+  INTERNAL            = (_cygrpc.StatusCode.internal, 'internal')
+  UNAVAILABLE         = (_cygrpc.StatusCode.unavailable, 'unavailable')
+  DATA_LOSS           = (_cygrpc.StatusCode.data_loss, 'data loss')
+  UNAUTHENTICATED     = (_cygrpc.StatusCode.unauthenticated, 'unauthenticated')
+
+
+#############################  gRPC Exceptions  ################################
+
+
+class RpcError(Exception):
+  """Raised by the gRPC library to indicate non-OK-status RPC termination."""
+
+
+##############################  Shared Context  ################################
+
+
+class RpcContext(six.with_metaclass(abc.ABCMeta)):
+  """Provides RPC-related information and action."""
+
+  @abc.abstractmethod
+  def is_active(self):
+    """Describes whether the RPC is active or has terminated."""
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def time_remaining(self):
+    """Describes the length of allowed time remaining for the RPC.
+
+    Returns:
+      A nonnegative float indicating the length of allowed time in seconds
+      remaining for the RPC to complete before it is considered to have timed
+      out, or None if no deadline was specified for the RPC.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def cancel(self):
+    """Cancels the RPC.
+
+    Idempotent and has no effect if the RPC has already terminated.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def add_callback(self, callback):
+    """Registers a callback to be called on RPC termination.
+
+    Args:
+      callback: A no-parameter callable to be called on RPC termination.
+
+    Returns:
+      True if the callback was added and will be called later; False if the
+        callback was not added and will not later be called (because the RPC
+        already terminated or some other reason).
+    """
+    raise NotImplementedError()
+
+
+#########################  Invocation-Side Context  ############################
+
+
+class Call(six.with_metaclass(abc.ABCMeta, RpcContext)):
+  """Invocation-side utility object for an RPC."""
+
+  @abc.abstractmethod
+  def initial_metadata(self):
+    """Accesses the initial metadata from the service-side of the RPC.
+
+    This method blocks until the value is available.
+
+    Returns:
+      The initial metadata as a sequence of pairs of bytes.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def trailing_metadata(self):
+    """Accesses the trailing metadata from the service-side of the RPC.
+
+    This method blocks until the value is available.
+
+    Returns:
+      The trailing metadata as a sequence of pairs of bytes.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def code(self):
+    """Accesses the status code emitted by the service-side of the RPC.
+
+    This method blocks until the value is available.
+
+    Returns:
+      The StatusCode value for the RPC.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def details(self):
+    """Accesses the details value emitted by the service-side of the RPC.
+
+    This method blocks until the value is available.
+
+    Returns:
+      The bytes of the details of the RPC.
+    """
+    raise NotImplementedError()
+
+
+########################  Multi-Callable Interfaces  ###########################
+
+
+class UnaryUnaryMultiCallable(six.with_metaclass(abc.ABCMeta)):
+  """Affords invoking a unary-unary RPC."""
+
+  @abc.abstractmethod
+  def __call__(self, request, timeout=None, metadata=None, with_call=False):
+    """Synchronously invokes the underlying RPC.
+
+    Args:
+      request: The request value for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+      with_call: Whether or not to include return a Call for the RPC in addition
+        to the response.
+
+    Returns:
+      The response value for the RPC, and a Call for the RPC if with_call was
+        set to True at invocation.
+
+    Raises:
+      RpcError: Indicating that the RPC terminated with non-OK status. The
+        raised RpcError will also be a Call for the RPC affording the RPC's
+        metadata, status code, and details.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def future(self, request, timeout=None, metadata=None):
+    """Asynchronously invokes the underlying RPC.
+
+    Args:
+      request: The request value for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+
+    Returns:
+      An object that is both a Call for the RPC and a Future. In the event of
+        RPC completion, the return Future's result value will be the response
+        message of the RPC. Should the event terminate with non-OK status, the
+        returned Future's exception value will be an RpcError.
+    """
+    raise NotImplementedError()
+
+
+class UnaryStreamMultiCallable(six.with_metaclass(abc.ABCMeta)):
+  """Affords invoking a unary-stream RPC."""
+
+  @abc.abstractmethod
+  def __call__(self, request, timeout=None, metadata=None):
+    """Invokes the underlying RPC.
+
+    Args:
+      request: The request value for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+
+    Returns:
+      An object that is both a Call for the RPC and an iterator of response
+        values. Drawing response values from the returned iterator may raise
+        RpcError indicating termination of the RPC with non-OK status.
+    """
+    raise NotImplementedError()
+
+
+class StreamUnaryMultiCallable(six.with_metaclass(abc.ABCMeta)):
+  """Affords invoking a stream-unary RPC in any call style."""
+
+  @abc.abstractmethod
+  def __call__(
+      self, request_iterator, timeout=None, metadata=None, with_call=False):
+    """Synchronously invokes the underlying RPC.
+
+    Args:
+      request_iterator: An iterator that yields request values for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+      with_call: Whether or not to include return a Call for the RPC in addition
+        to the response.
+
+    Returns:
+      The response value for the RPC, and a Call for the RPC if with_call was
+        set to True at invocation.
+
+    Raises:
+      RpcError: Indicating that the RPC terminated with non-OK status. The
+        raised RpcError will also be a Call for the RPC affording the RPC's
+        metadata, status code, and details.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def future(self, request_iterator, timeout=None, metadata=None):
+    """Asynchronously invokes the underlying RPC.
+
+    Args:
+      request_iterator: An iterator that yields request values for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+
+    Returns:
+      An object that is both a Call for the RPC and a Future. In the event of
+        RPC completion, the return Future's result value will be the response
+        message of the RPC. Should the event terminate with non-OK status, the
+        returned Future's exception value will be an RpcError.
+    """
+    raise NotImplementedError()
+
+
+class StreamStreamMultiCallable(six.with_metaclass(abc.ABCMeta)):
+  """Affords invoking a stream-stream RPC in any call style."""
+
+  @abc.abstractmethod
+  def __call__(self, request_iterator, timeout=None, metadata=None):
+    """Invokes the underlying RPC.
+
+    Args:
+      request_iterator: An iterator that yields request values for the RPC.
+      timeout: An optional duration of time in seconds to allow for the RPC.
+      metadata: An optional sequence of pairs of bytes to be transmitted to the
+        service-side of the RPC.
+
+    Returns:
+      An object that is both a Call for the RPC and an iterator of response
+        values. Drawing response values from the returned iterator may raise
+        RpcError indicating termination of the RPC with non-OK status.
+    """
+    raise NotImplementedError()
+
+
+#############################  Channel Interface  ##############################
+
+
+class Channel(six.with_metaclass(abc.ABCMeta)):
+  """Affords RPC invocation via generic methods."""
+
+  @abc.abstractmethod
+  def subscribe(self, callback, try_to_connect=False):
+    """Subscribes to this Channel's connectivity.
+
+    Args:
+      callback: A callable to be invoked and passed a ChannelConnectivity value
+        describing this Channel's connectivity. The callable will be invoked
+        immediately upon subscription and again for every change to this
+        Channel's connectivity thereafter until it is unsubscribed or this
+        Channel object goes out of scope.
+      try_to_connect: A boolean indicating whether or not this Channel should
+        attempt to connect if it is not already connected and ready to conduct
+        RPCs.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def unsubscribe(self, callback):
+    """Unsubscribes a callback from this Channel's connectivity.
+
+    Args:
+      callback: A callable previously registered with this Channel from having
+        been passed to its "subscribe" method.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def unary_unary(
+      self, method, request_serializer=None, response_deserializer=None):
+    """Creates a UnaryUnaryMultiCallable for a unary-unary method.
+
+    Args:
+      method: The name of the RPC method.
+
+    Returns:
+      A UnaryUnaryMultiCallable value for the named unary-unary method.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def unary_stream(
+      self, method, request_serializer=None, response_deserializer=None):
+    """Creates a UnaryStreamMultiCallable for a unary-stream method.
+
+    Args:
+      method: The name of the RPC method.
+
+    Returns:
+      A UnaryStreamMultiCallable value for the name unary-stream method.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def stream_unary(
+      self, method, request_serializer=None, response_deserializer=None):
+    """Creates a StreamUnaryMultiCallable for a stream-unary method.
+
+    Args:
+      method: The name of the RPC method.
+
+    Returns:
+      A StreamUnaryMultiCallable value for the named stream-unary method.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def stream_stream(
+      self, method, request_serializer=None, response_deserializer=None):
+    """Creates a StreamStreamMultiCallable for a stream-stream method.
+
+    Args:
+      method: The name of the RPC method.
+
+    Returns:
+      A StreamStreamMultiCallable value for the named stream-stream method.
+    """
+    raise NotImplementedError()
+
+
+##########################  Service-Side Context  ##############################
+
+
+class ServicerContext(six.with_metaclass(abc.ABCMeta, RpcContext)):
+  """A context object passed to method implementations."""
+
+  @abc.abstractmethod
+  def invocation_metadata(self):
+    """Accesses the metadata from the invocation-side of the RPC.
+
+    Returns:
+      The invocation metadata object as a sequence of pairs of bytes.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def peer(self):
+    """Identifies the peer that invoked the RPC being serviced.
+
+    Returns:
+      A string identifying the peer that invoked the RPC being serviced.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def send_initial_metadata(self, initial_metadata):
+    """Sends the initial metadata value to the invocation-side of the RPC.
+
+    This method need not be called by method implementations if they have no
+    service-side initial metadata to transmit.
+
+    Args:
+      initial_metadata: The initial metadata of the RPC as a sequence of pairs
+        of bytes.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def set_trailing_metadata(self, trailing_metadata):
+    """Accepts the trailing metadata value of the RPC.
+
+    This method need not be called by method implementations if they have no
+    service-side trailing metadata to transmit.
+
+    Args:
+      trailing_metadata: The trailing metadata of the RPC as a sequence of pairs
+        of bytes.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def set_code(self, code):
+    """Accepts the status code of the RPC.
+
+    This method need not be called by method implementations if they wish the
+    gRPC runtime to determine the status code of the RPC.
+
+    Args:
+      code: The integer status code of the RPC to be transmitted to the
+        invocation side of the RPC.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def set_details(self, details):
+    """Accepts the service-side details of the RPC.
+
+    This method need not be called by method implementations if they have no
+    details to transmit.
+
+    Args:
+      details: The details bytes of the RPC to be transmitted to
+        the invocation side of the RPC.
+    """
+    raise NotImplementedError()
+
+
+#####################  Service-Side Handler Interfaces  ########################
+
+
+class RpcMethodHandler(six.with_metaclass(abc.ABCMeta)):
+  """An implementation of a single RPC method.
+
+  Attributes:
+    request_streaming: Whether the RPC supports exactly one request message or
+      any arbitrary number of request messages.
+    response_streaming: Whether the RPC supports exactly one response message or
+      any arbitrary number of response messages.
+    request_deserializer: A callable behavior that accepts a byte string and
+      returns an object suitable to be passed to this object's business logic,
+      or None to indicate that this object's business logic should be passed the
+      raw request bytes.
+    response_serializer: A callable behavior that accepts an object produced by
+      this object's business logic and returns a byte string, or None to
+      indicate that the byte strings produced by this object's business logic
+      should be transmitted on the wire as they are.
+    unary_unary: This object's application-specific business logic as a callable
+      value that takes a request value and a ServicerContext object and returns
+      a response value. Only non-None if both request_streaming and
+      response_streaming are False.
+    unary_stream: This object's application-specific business logic as a
+      callable value that takes a request value and a ServicerContext object and
+      returns an iterator of response values. Only non-None if request_streaming
+      is False and response_streaming is True.
+    stream_unary: This object's application-specific business logic as a
+      callable value that takes an iterator of request values and a
+      ServicerContext object and returns a response value. Only non-None if
+      request_streaming is True and response_streaming is False.
+    stream_stream: This object's application-specific business logic as a
+      callable value that takes an iterator of request values and a
+      ServicerContext object and returns an iterator of response values. Only
+      non-None if request_streaming and response_streaming are both True.
+  """
+
+
+class HandlerCallDetails(six.with_metaclass(abc.ABCMeta)):
+  """Describes an RPC that has just arrived for service.
+
+  Attributes:
+    method: The method name of the RPC.
+    invocation_metadata: The metadata from the invocation side of the RPC.
+  """
+
+
+class GenericRpcHandler(six.with_metaclass(abc.ABCMeta)):
+  """An implementation of arbitrarily many RPC methods."""
+
+  @abc.abstractmethod
+  def service(self, handler_call_details):
+    """Services an RPC (or not).
+
+    Args:
+      handler_call_details: A HandlerCallDetails describing the RPC.
+
+    Returns:
+      An RpcMethodHandler with which the RPC may be serviced, or None to
+        indicate that this object will not be servicing the RPC.
+    """
+    raise NotImplementedError()
+
+
+#############################  Server Interface  ###############################
+
+
+class Server(six.with_metaclass(abc.ABCMeta)):
+  """Services RPCs."""
+
+  @abc.abstractmethod
+  def add_generic_rpc_handlers(self, generic_rpc_handlers):
+    """Registers GenericRpcHandlers with this Server.
+
+    This method is only safe to call before the server is started.
+
+    Args:
+      generic_rpc_handlers: An iterable of GenericRpcHandlers that will be used
+        to service RPCs after this Server is started.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def add_insecure_port(self, address):
+    """Reserves a port for insecure RPC service once this Server becomes active.
+
+    This method may only be called before calling this Server's start method is
+    called.
+
+    Args:
+      address: The address for which to open a port.
+
+    Returns:
+      An integer port on which RPCs will be serviced after this link has been
+        started. This is typically the same number as the port number contained
+        in the passed address, but will likely be different if the port number
+        contained in the passed address was zero.
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def start(self):
+    """Starts this Server's service of RPCs.
+
+    This method may only be called while the server is not serving RPCs (i.e. it
+    is not idempotent).
+    """
+    raise NotImplementedError()
+
+  @abc.abstractmethod
+  def stop(self, grace):
+    """Stops this Server's service of RPCs.
+
+    All calls to this method immediately stop service of new RPCs. When existing
+    RPCs are aborted is controlled by the grace period parameter passed to this
+    method.
+
+    This method may be called at any time and is idempotent. Passing a smaller
+    grace value than has been passed in a previous call will have the effect of
+    stopping the Server sooner. Passing a larger grace value than has been
+    passed in a previous call will not have the effect of stopping the server
+    later.
+
+    Args:
+      grace: A duration of time in seconds to allow existing RPCs to complete
+        before being aborted by this Server's stopping. If None, this method
+        will block until the server is completely stopped.
+
+    Returns:
+      A threading.Event that will be set when this Server has completely
+      stopped. The returned event may not be set until after the full grace
+      period (if some ongoing RPC continues for the full length of the period)
+      of it may be set much sooner (such as if this Server had no RPCs underway
+      at the time it was stopped or if all RPCs that it had underway completed
+      very early in the grace period).
+    """
+    raise NotImplementedError()
diff --git a/src/python/grpcio/grpc_core_dependencies.py b/src/python/grpcio/grpc_core_dependencies.py
index be24dc7..9c067ad 100644
--- a/src/python/grpcio/grpc_core_dependencies.py
+++ b/src/python/grpcio/grpc_core_dependencies.py
@@ -94,6 +94,7 @@
   'src/core/lib/iomgr/endpoint.c',
   'src/core/lib/iomgr/endpoint_pair_posix.c',
   'src/core/lib/iomgr/endpoint_pair_windows.c',
+  'src/core/lib/iomgr/ev_poll_and_epoll_posix.c',
   'src/core/lib/iomgr/ev_poll_posix.c',
   'src/core/lib/iomgr/ev_posix.c',
   'src/core/lib/iomgr/exec_ctx.c',
diff --git a/src/python/grpcio/tests/unit/framework/common/test_control.py b/src/python/grpcio/tests/unit/framework/common/test_control.py
index ca5ba3a..088e2f8 100644
--- a/src/python/grpcio/tests/unit/framework/common/test_control.py
+++ b/src/python/grpcio/tests/unit/framework/common/test_control.py
@@ -60,10 +60,16 @@
 
 
 class PauseFailControl(Control):
-  """A Control that can be used to pause or fail code under control."""
+  """A Control that can be used to pause or fail code under control.
+
+  This object is only safe for use from two threads: one of the system under
+  test calling control and the other from the test system calling pause,
+  block_until_paused, and fail.
+  """
 
   def __init__(self):
     self._condition = threading.Condition()
+    self._pause = False
     self._paused = False
     self._fail = False
 
@@ -72,19 +78,31 @@
       if self._fail:
         raise Defect()
 
-      while self._paused:
+      while self._pause:
+        self._paused = True
+        self._condition.notify_all()
         self._condition.wait()
+      self._paused = False
 
   @contextlib.contextmanager
   def pause(self):
     """Pauses code under control while controlling code is in context."""
     with self._condition:
-      self._paused = True
+      self._pause = True
     yield
     with self._condition:
-      self._paused = False
+      self._pause = False
       self._condition.notify_all()
 
+  def block_until_paused(self):
+    """Blocks controlling code until code under control is paused.
+
+    May only be called within the context of a pause call.
+    """
+    with self._condition:
+      while not self._paused:
+        self._condition.wait()
+
   @contextlib.contextmanager
   def fail(self):
     """Fails code under control while controlling code is in context."""
diff --git a/tools/doxygen/Doxyfile.core.internal b/tools/doxygen/Doxyfile.core.internal
index c3fd59b..386ef7d 100644
--- a/tools/doxygen/Doxyfile.core.internal
+++ b/tools/doxygen/Doxyfile.core.internal
@@ -807,6 +807,7 @@
 src/core/lib/iomgr/closure.h \
 src/core/lib/iomgr/endpoint.h \
 src/core/lib/iomgr/endpoint_pair.h \
+src/core/lib/iomgr/ev_poll_and_epoll_posix.h \
 src/core/lib/iomgr/ev_poll_posix.h \
 src/core/lib/iomgr/ev_posix.h \
 src/core/lib/iomgr/exec_ctx.h \
@@ -954,6 +955,7 @@
 src/core/lib/iomgr/endpoint.c \
 src/core/lib/iomgr/endpoint_pair_posix.c \
 src/core/lib/iomgr/endpoint_pair_windows.c \
+src/core/lib/iomgr/ev_poll_and_epoll_posix.c \
 src/core/lib/iomgr/ev_poll_posix.c \
 src/core/lib/iomgr/ev_posix.c \
 src/core/lib/iomgr/exec_ctx.c \
diff --git a/tools/jenkins/run_full_performance.sh b/tools/jenkins/run_full_performance.sh
index f1e09a8..3feda86 100755
--- a/tools/jenkins/run_full_performance.sh
+++ b/tools/jenkins/run_full_performance.sh
@@ -40,7 +40,7 @@
     --netperf \
     --category all \
     --bq_result_table performance_test.performance_experiment \
-    --remote_worker_host grpc-performance-server-8core grpc-performance-client-8core \
+    --remote_worker_host grpc-performance-server-8core grpc-performance-client-8core grpc-performance-client2-8core \
     || EXIT_CODE=1
 
 # scalability with 32cores (and upload to a different BQ table)
diff --git a/tools/run_tests/run_tests.py b/tools/run_tests/run_tests.py
index 0538dce..0a5625c 100755
--- a/tools/run_tests/run_tests.py
+++ b/tools/run_tests/run_tests.py
@@ -157,7 +157,7 @@
       'windows': ['all'],
       'mac': ['all'],
       'posix': ['all'],
-      'linux': ['poll'],
+      'linux': ['poll', 'legacy']
     }
     for target in binaries:
       polling_strategies = (POLLING_STRATEGIES[self.platform]
diff --git a/tools/run_tests/sources_and_headers.json b/tools/run_tests/sources_and_headers.json
index 24d23fe..fe67426 100644
--- a/tools/run_tests/sources_and_headers.json
+++ b/tools/run_tests/sources_and_headers.json
@@ -5530,6 +5530,7 @@
       "src/core/lib/iomgr/closure.h", 
       "src/core/lib/iomgr/endpoint.h", 
       "src/core/lib/iomgr/endpoint_pair.h", 
+      "src/core/lib/iomgr/ev_poll_and_epoll_posix.h", 
       "src/core/lib/iomgr/ev_poll_posix.h", 
       "src/core/lib/iomgr/ev_posix.h", 
       "src/core/lib/iomgr/exec_ctx.h", 
@@ -5629,6 +5630,8 @@
       "src/core/lib/iomgr/endpoint_pair.h", 
       "src/core/lib/iomgr/endpoint_pair_posix.c", 
       "src/core/lib/iomgr/endpoint_pair_windows.c", 
+      "src/core/lib/iomgr/ev_poll_and_epoll_posix.c", 
+      "src/core/lib/iomgr/ev_poll_and_epoll_posix.h", 
       "src/core/lib/iomgr/ev_poll_posix.c", 
       "src/core/lib/iomgr/ev_poll_posix.h", 
       "src/core/lib/iomgr/ev_posix.c", 
diff --git a/vsprojects/vcxproj/grpc/grpc.vcxproj b/vsprojects/vcxproj/grpc/grpc.vcxproj
index e5379dc..7a39070 100644
--- a/vsprojects/vcxproj/grpc/grpc.vcxproj
+++ b/vsprojects/vcxproj/grpc/grpc.vcxproj
@@ -316,6 +316,7 @@
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\closure.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair.h" />
+    <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\exec_ctx.h" />
@@ -483,6 +484,8 @@
     </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair_windows.c">
     </ClCompile>
+    <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.c">
+    </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.c">
     </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_posix.c">
diff --git a/vsprojects/vcxproj/grpc/grpc.vcxproj.filters b/vsprojects/vcxproj/grpc/grpc.vcxproj.filters
index 95a5a73..ae70e0a 100644
--- a/vsprojects/vcxproj/grpc/grpc.vcxproj.filters
+++ b/vsprojects/vcxproj/grpc/grpc.vcxproj.filters
@@ -55,6 +55,9 @@
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair_windows.c">
       <Filter>src\core\lib\iomgr</Filter>
     </ClCompile>
+    <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.c">
+      <Filter>src\core\lib\iomgr</Filter>
+    </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.c">
       <Filter>src\core\lib\iomgr</Filter>
     </ClCompile>
@@ -674,6 +677,9 @@
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair.h">
       <Filter>src\core\lib\iomgr</Filter>
     </ClInclude>
+    <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.h">
+      <Filter>src\core\lib\iomgr</Filter>
+    </ClInclude>
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.h">
       <Filter>src\core\lib\iomgr</Filter>
     </ClInclude>
diff --git a/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj b/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj
index 90ad80f..09748f0 100644
--- a/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj
+++ b/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj
@@ -304,6 +304,7 @@
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\closure.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair.h" />
+    <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_posix.h" />
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\exec_ctx.h" />
@@ -449,6 +450,8 @@
     </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair_windows.c">
     </ClCompile>
+    <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.c">
+    </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.c">
     </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_posix.c">
diff --git a/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj.filters b/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj.filters
index 2b19c0f..a85bfee 100644
--- a/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj.filters
+++ b/vsprojects/vcxproj/grpc_unsecure/grpc_unsecure.vcxproj.filters
@@ -58,6 +58,9 @@
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair_windows.c">
       <Filter>src\core\lib\iomgr</Filter>
     </ClCompile>
+    <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.c">
+      <Filter>src\core\lib\iomgr</Filter>
+    </ClCompile>
     <ClCompile Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.c">
       <Filter>src\core\lib\iomgr</Filter>
     </ClCompile>
@@ -572,6 +575,9 @@
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\endpoint_pair.h">
       <Filter>src\core\lib\iomgr</Filter>
     </ClInclude>
+    <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_and_epoll_posix.h">
+      <Filter>src\core\lib\iomgr</Filter>
+    </ClInclude>
     <ClInclude Include="$(SolutionDir)\..\src\core\lib\iomgr\ev_poll_posix.h">
       <Filter>src\core\lib\iomgr</Filter>
     </ClInclude>