erts: Remove ets traversal yielding if heap fragment

Many heap fragments do no longer make the GC slow.

Even worse, we are not guaranteed that a yield will provoke a GC
removing the fragments, which might lead to a one-yield-per-bucket
scenario if the heap fragment(s) still remains after each yield.

1 files changed, 2 insertions, 10 deletions

diff --git a/erts/emulator/beam/erl_db_hash.c b/erts/emulator/beam/erl_db_hash.c
index cf928a9035..47d304f860 100644
--- a/erts/emulator/beam/erl_db_hash.c
+++ b/erts/emulator/beam/erl_db_hash.c
@@ -1328,11 +1328,7 @@ static int match_traverse(Process* p, DbTableHash* tb,
                 unlock_hash_function(lck);
                 break;
             }
-            if (iterations_left <= 0 || MBUF(p)) {
-                /*
-                 * We have either reached our limit, or just created some heap fragments.
-                 * Since many heap fragments will make the GC slower, trap and GC now.
-                 */
+            if (iterations_left <= 0) {
                 unlock_hash_function(lck);
                 ret_value = on_trap(context_ptr, slot_ix, got, &mpi.mp, ret);
                 goto done;
@@ -1451,11 +1447,7 @@ static int match_traverse_continue(Process* p, DbTableHash* tb,
                 unlock_hash_function(lck);
                 break;
             }
-            if (iterations_left <= 0 || MBUF(p)) {
-                /*
-                 * We have either reached our limit, or just created some heap fragments.
-                 * Since many heap fragments will make the GC slower, trap and GC now.
-                 */
+            if (iterations_left <= 0) {
                 unlock_hash_function(lck);
                 ret_value = on_trap(context_ptr, slot_ix, got, mpp, ret);
                 goto done;