--- libev/ev_iouring.c	2019/12/27 22:16:10	1.12
+++ libev/ev_iouring.c	2019/12/28 07:39:18	1.17
@@ -230,20 +230,62 @@
 #define EV_SQES         ((struct io_uring_sqe *)         iouring_sqes)
 #define EV_CQES         ((struct io_uring_cqe *)((char *)iouring_cq_ring + iouring_cq_cqes))
 
+inline_speed
+int
+iouring_enter (EV_P_ ev_tstamp timeout)
+{
+  int res;
+
+  EV_RELEASE_CB;
+
+  res = evsys_io_uring_enter (iouring_fd, iouring_to_submit, 1,
+                              timeout > EV_TS_CONST (0.) ? IORING_ENTER_GETEVENTS : 0, 0, 0);
+
+  assert (("libev: io_uring_enter did not consume all sqes", (res < 0 || res == iouring_to_submit)));
+
+  iouring_to_submit = 0;
+
+  EV_ACQUIRE_CB;
+
+  return res;
+}
+
+/* TODO: can we move things around so we don't need this forward-reference? */
+static void
+iouring_poll (EV_P_ ev_tstamp timeout);
+
 static
 struct io_uring_sqe *
 iouring_sqe_get (EV_P)
 {
-  unsigned tail = EV_SQ_VAR (tail);
-
-  if (tail + 1 - EV_SQ_VAR (head) > EV_SQ_VAR (ring_entries))
+  unsigned tail;
+  
+  for (;;)
     {
-      /* queue full, flush */
-      evsys_io_uring_enter (iouring_fd, iouring_to_submit, 0, 0, 0, 0);
-      iouring_to_submit = 0;
+      tail = EV_SQ_VAR (tail);
+
+      if (ecb_expect_true (tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)))
+        break; /* whats the problem, we have free sqes */
+
+      /* queue full, need to flush and possibly handle some events */
+
+#if EV_FEATURE_CODE
+      /* first we ask the kernel nicely, most often this frees up some sqes */
+      int res = iouring_enter (EV_A_ EV_TS_CONST (0.));
+
+      ECB_MEMORY_FENCE_ACQUIRE; /* better safe than sorry */
+
+      if (res >= 0)
+        continue; /* yes, it worked, try again */
+#endif
+
+      /* some problem, possibly EBUSY - do the full poll and let it handle any issues */
+
+      iouring_poll (EV_A_ EV_TS_CONST (0.));
+      /* iouring_poll should have done ECB_MEMORY_FENCE_ACQUIRE for us */
     }
 
-  assert (("libev: io_uring queue full after flush", tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)));
+  /*assert (("libev: io_uring queue full after flush", tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)));*/
 
   return EV_SQES + (tail & EV_SQ_VAR (ring_mask));
 }
@@ -319,7 +361,7 @@
         return -1; /* we failed */
 
 #if TODO
-      if ((~params.features) & (IORING_FEAT_NODROP | IORING_FEATURE_SINGLE_MMAP))
+      if ((~params.features) & (IORING_FEAT_NODROP | IORING_FEATURE_SINGLE_MMAP | IORING_FEAT_SUBMIT_STABLE))
         return -1; /* we require the above features */
 #endif
 
@@ -402,7 +444,13 @@
       struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
       sqe->opcode    = IORING_OP_POLL_REMOVE;
       sqe->fd        = fd;
-      sqe->user_data = -1;
+      /* Jens Axboe notified me that user_data is not what is documented, but is
+       * some kind of unique ID that has to match, otherwise the request cannot
+       * be removed. Since we don't *really* have that, we pass in the old
+       * generation counter - if that fails, too bad, it will hopefully be removed
+       * at close time and then be ignored. */
+      sqe->addr      = (uint32_t)fd | ((__u64)(uint32_t)anfds [fd].egen << 32);
+      sqe->user_data = (uint64_t)-1;
       iouring_sqe_submit (EV_A_ sqe);
 
       /* increment generation counter to avoid handling old events */
@@ -453,9 +501,8 @@
   uint32_t gen = cqe->user_data >> 32;
   int      res = cqe->res;
 
-  /* ignore fd removal events, if there are any. TODO: verify */
-  /* TODO: yes, this triggers */
-  if (cqe->user_data == (__u64)-1)
+  /* user_data -1 is a remove that we are not atm. interested in */
+  if (cqe->user_data == (uint64_t)-1)
     return;
 
   assert (("libev: io_uring fd must be in-bounds", fd >= 0 && fd < anfdmax));
@@ -467,6 +514,7 @@
    */
 
   /* ignore event if generation doesn't match */
+  /* other than skipping removal events, */
   /* this should actually be very rare */
   if (ecb_expect_false (gen != (uint32_t)anfds [fd].egen))
     return;
@@ -590,7 +638,11 @@
 iouring_poll (EV_P_ ev_tstamp timeout)
 {
   /* if we have events, no need for extra syscalls, but we might have to queue events */
-  if (iouring_handle_cq (EV_A))
+  /* we also clar the timeout if there are outstanding fdchanges */
+  /* the latter should only happen if both the sq and cq are full, most likely */
+  /* because we have a lot of event sources that immediately complete */
+  /* TODO: fdchacngecnt is always 0 because fd_reify does not have two buffers yet */
+  if (iouring_handle_cq (EV_A) || fdchangecnt)
     timeout = EV_TS_CONST (0.);
   else
     /* no events, so maybe wait for some */
@@ -599,19 +651,13 @@
   /* only enter the kernel if we have something to submit, or we need to wait */
   if (timeout || iouring_to_submit)
     {
-      int res;
-
-      EV_RELEASE_CB;
-
-      res = evsys_io_uring_enter (iouring_fd, iouring_to_submit, 1,
-                                  timeout > EV_TS_CONST (0.) ? IORING_ENTER_GETEVENTS : 0, 0, 0);
-      iouring_to_submit = 0;
-
-      EV_ACQUIRE_CB;
+      int res = iouring_enter (EV_A_ timeout);
 
       if (ecb_expect_false (res < 0))
         if (errno == EINTR)
           /* ignore */;
+        else if (errno == EBUSY)
+          /* cq full, cannot submit - should be rare because we flush the cq first, so simply ignore */;
         else
           ev_syserr ("(libev) iouring setup");
       else