Expand documentation

2 files changed, 23 insertions, 0 deletions

diff --git a/src/core/lib/iomgr/ev_epoll_linux.c b/src/core/lib/iomgr/ev_epoll_linux.c
index 98fe2defea..8de42bb7a9 100644
--- a/src/core/lib/iomgr/ev_epoll_linux.c
+++ b/src/core/lib/iomgr/ev_epoll_linux.c
@@ -165,6 +165,7 @@ static void fd_global_shutdown(void);
 
 #endif /* !defined(GPRC_PI_REF_COUNT_DEBUG) */
 
+/* This is also used as grpc_workqueue (by directly casing it) */
 typedef struct polling_island {
   gpr_mu mu;
   /* Ref count. Use PI_ADD_REF() and PI_UNREF() macros to increment/decrement
@@ -184,10 +185,16 @@ typedef struct polling_island {
    * (except mu and ref_count) are invalid and must be ignored. */
   gpr_atm merged_to;
 
+  /* Number of threads currently polling on this island */
   gpr_atm poller_count;
+  /* Mutex guarding the read end of the workqueue (must be held to pop from
+   * workqueue_items) */
   gpr_mu workqueue_read_mu;
+  /* Queue of closures to be executed */
   gpr_mpscq workqueue_items;
+  /* Count of items in workqueue_items */
   gpr_atm workqueue_item_count;
+  /* Wakeup fd used to wake pollers to check the contents of workqueue_items */
   grpc_wakeup_fd workqueue_wakeup_fd;
 
   /* The fd of the underlying epoll set */
@@ -1396,6 +1403,9 @@ static bool maybe_do_workqueue_work(grpc_exec_ctx *exec_ctx,
       grpc_closure_run(exec_ctx, c, c->error_data.error);
       return true;
     } else if (gpr_atm_no_barrier_load(&pi->workqueue_item_count) > 0) {
+      /* n == NULL might mean there's work but it's not available to be popped
+       * yet - try to ensure another workqueue wakes up to check shortly if so
+       */
       workqueue_maybe_wakeup(pi);
     }
   }
@@ -1457,6 +1467,9 @@ static void pollset_work_and_unlock(grpc_exec_ctx *exec_ctx,
   PI_ADD_REF(pi, "ps_work");
   gpr_mu_unlock(&pollset->mu);
 
+  /* If we get some workqueue work to do, it might end up completing an item on
+     the completion queue, so there's no need to poll... so we skip that and
+     redo the complete loop to verify */
   if (!maybe_do_workqueue_work(exec_ctx, pi)) {
     gpr_atm_no_barrier_fetch_add(&pi->poller_count, 1);
     g_current_thread_polling_island = pi;
diff --git a/src/core/lib/iomgr/exec_ctx.h b/src/core/lib/iomgr/exec_ctx.h
index 4744e21c5e..7e50cb9825 100644
--- a/src/core/lib/iomgr/exec_ctx.h
+++ b/src/core/lib/iomgr/exec_ctx.h
@@ -66,10 +66,20 @@ typedef struct grpc_combiner grpc_combiner;
 #ifndef GRPC_EXECUTION_CONTEXT_SANITIZER
 struct grpc_exec_ctx {
   grpc_closure_list closure_list;
+  /** The workqueue we're stealing work from.
+      As items are queued to the execution context, we try to steal one
+      workqueue item and execute it inline (assuming the exec_ctx is not
+      finished) - doing so does not invalidate the workqueue's contract, and
+      provides a small latency win in cases where we get a hit */
   grpc_workqueue *stealing_from_workqueue;
+  /** The workqueue item that was stolen from the workqueue above. When new
+      items are scheduled to be offloaded to that workqueue, we need to update
+      this like a 1-deep fifo to maintain the invariant that workqueue items
+      queued by one thread are started in order */
   grpc_closure *stolen_closure;
   /** currently active combiner: updated only via combiner.c */
   grpc_combiner *active_combiner;
+  /** last active combiner in the active combiner list */
   grpc_combiner *last_combiner;
   bool cached_ready_to_finish;
   void *check_ready_to_finish_arg;