Allow specific pollers to be woken

Currently, if two threads call grpc_completion_queue_pluck on the same
completion queue for different tags, there is a 50% chance that we
deliver the completion wakeup to the wrong poller - forcing the correct
poller to wait until its polling times out before it can return an event
up to the application.

This change tweaks our polling interfaces so that we can indeed wake a
specific poller.

Nothing has been performance tuned yet. It's definitely sub-optimal in a
number of places. Wakeup file-descriptors should be recycled. We should
have a path that avoids calling poll() followed by epoll(). We can
probably live without it right at the second though.

This code will fail on Windows at least (I'll do that port when I'm in the office and have a Windows
machine).

1 files changed, 3 insertions, 2 deletions

diff --git a/test/core/iomgr/tcp_server_posix_test.c b/test/core/iomgr/tcp_server_posix_test.c
index f8d0fe8217..b82d7c08b1 100644
--- a/test/core/iomgr/tcp_server_posix_test.c
+++ b/test/core/iomgr/tcp_server_posix_test.c
@@ -54,7 +54,7 @@ static void on_connect(void *arg, grpc_endpoint *tcp) {
 
   gpr_mu_lock(GRPC_POLLSET_MU(&g_pollset));
   g_nconnects++;
-  grpc_pollset_kick(&g_pollset);
+  grpc_pollset_kick(&g_pollset, NULL);
   gpr_mu_unlock(GRPC_POLLSET_MU(&g_pollset));
 }
 
@@ -136,7 +136,8 @@ static void test_connect(int n) {
     gpr_log(GPR_DEBUG, "wait");
     while (g_nconnects == nconnects_before &&
            gpr_time_cmp(deadline, gpr_now(deadline.clock_type)) > 0) {
-      grpc_pollset_work(&g_pollset, deadline);
+      grpc_pollset_worker worker;
+      grpc_pollset_work(&g_pollset, &worker, deadline);
     }
     gpr_log(GPR_DEBUG, "wait done");