further enhancements for dirty schedulers

Add support for setting the number of dirty CPU schedulers online via
erlang:system_flag/2. Assuming the emulator is built with dirty schedulers
enabled, the number of dirty CPU schedulers online may not be less than 1,
nor greater than the number of dirty CPU schedulers available, nor greater
than the number of normal schedulers online. Dirty CPU scheduler threads
that are taken offline via system_flag/2 are suspended. The number of dirty
CPU schedulers online may be adjusted independently of the number of normal
schedulers online, but if system_flag/2 is used to set the number of normal
schedulers online to a value less than the current number of normal
schedulers online, the number of dirty CPU schedulers online is decreased
proportionally. Likewise, if the number of normal schedulers online is
increased, the number of dirty CPU schedulers online is increased
proportionally. For example, if 8 normal schedulers and 4 dirty CPU
schedulers are online, and system_flag/2 is called to set the number of
normal schedulers online to 4, the number of dirty CPU schedulers online is
also decreased by half, to 2. Subsequently setting the number of normal
schedulers online back to 8 also sets the number of dirty CPU schedulers
online back to 4. Augment the system_flag/2 documentation in the erlang man
page to explain this relationship between schedulers_online and
dirty_cpu_schedulers_online.

Also ensure that all dirty CPU and I/O schedulers are suspended when
multi-scheduling is blocked via system_flag/2, and brought back online when
multi-scheduling is unblocked.

Add Rickard Green's rewritten check_enqueue_in_prio_queue() function that
inspects process state more thoroughly to determine if to enqueue it and if
so on what queue, including dirty queues when appropriate.

Make sure dirty NIF jobs do not trigger erlang:system_monitor long_schedule
messages.

Add more dirty scheduler testing to the scheduler test suite.

Remove the erts_no_dirty_cpu_schedulers_online global variable, since it's
no longer needed.

Execute dirty NIFs on a normal scheduler thread while multi-scheduling
blocking is in effect. Evacuate any dirty jobs residing in the dirty run
queues over to a normal run queue when multi-scheduling is blocked.

Allow dirty schedulers to execute aux work.

Set the dirty run queues halt_in_progress flag when halting the normal
schedulers.

Change dirty scheduler numbers to a structure including both scheduler
number and type, either dirty CPU or dirty I/O. Add some assertions to
ensure that dirty CPU schedulers operate only on dirty CPU scheduler
process flags, and the same for dirty I/O schedulers.

1 files changed, 92 insertions, 3 deletions

diff --git a/erts/emulator/test/scheduler_SUITE.erl b/erts/emulator/test/scheduler_SUITE.erl
index 6a43e2b0e7..3906471f87 100644
--- a/erts/emulator/test/scheduler_SUITE.erl
+++ b/erts/emulator/test/scheduler_SUITE.erl
@@ -54,6 +54,7 @@
 	 sbt_cmd/1,
 	 scheduler_threads/1,
 	 scheduler_suspend/1,
+	 dirty_scheduler_threads/1,
 	 reader_groups/1]).
 
 -define(DEFAULT_TIMEOUT, ?t:minutes(15)).
@@ -68,6 +69,7 @@ all() ->
      equal_and_high_with_part_time_max, equal_with_high,
      equal_with_high_max, bound_process,
      {group, scheduler_bind}, scheduler_threads, scheduler_suspend,
+     dirty_scheduler_threads,
      reader_groups].
 
 groups() -> 
@@ -1092,6 +1094,55 @@ scheduler_threads(Config) when is_list(Config) ->
     end,
     ok.
 
+dirty_scheduler_threads(Config) when is_list(Config) ->
+    SmpSupport = erlang:system_info(smp_support),
+    try
+	erlang:system_info(dirty_cpu_schedulers),
+	dirty_scheduler_threads_test(Config, SmpSupport)
+    catch
+	error:badarg ->
+	    {skipped, "No dirty scheduler support"}
+    end.
+
+dirty_scheduler_threads_test(Config, SmpSupport) ->
+    {Sched, SchedOnln, _} = get_dsstate(Config, ""),
+    {HalfSched, HalfSchedOnln} = case SmpSupport of
+                                     false -> {1,1};
+                                     true ->
+                                         {Sched div 2,
+                                          SchedOnln div 2}
+                                 end,
+    Cmd1 = "+SDcpu "++integer_to_list(HalfSched)++":"++
+	integer_to_list(HalfSchedOnln),
+    {HalfSched, HalfSchedOnln, _} = get_dsstate(Config, Cmd1),
+    {HalfSched, HalfSchedOnln, _} = get_dsstate(Config, "+SDPcpu 50:50"),
+    IOSched = 20,
+    {_, _, IOSched} = get_dsstate(Config, "+SDio "++integer_to_list(IOSched)),
+    {ok, Node} = start_node(Config, ""),
+    [ok] = mcall(Node, [fun() -> dirty_schedulers_online_test() end]),
+    ok.
+
+dirty_schedulers_online_test() ->
+    dirty_schedulers_online_test(erlang:system_info(smp_support)).
+dirty_schedulers_online_test(false) -> ok;
+dirty_schedulers_online_test(true) ->
+    dirty_schedulers_online_smp_test(erlang:system_info(schedulers_online)).
+dirty_schedulers_online_smp_test(SchedOnln) when SchedOnln < 4 -> ok;
+dirty_schedulers_online_smp_test(SchedOnln) ->
+    DirtyCPUSchedOnln = erlang:system_info(dirty_cpu_schedulers_online),
+    SchedOnln = DirtyCPUSchedOnln,
+    HalfSchedOnln = SchedOnln div 2,
+    SchedOnln = erlang:system_flag(schedulers_online, HalfSchedOnln),
+    HalfDirtyCPUSchedOnln = DirtyCPUSchedOnln div 2,
+    HalfDirtyCPUSchedOnln = erlang:system_flag(schedulers_online, SchedOnln),
+    DirtyCPUSchedOnln = erlang:system_flag(dirty_cpu_schedulers_online,
+					   HalfDirtyCPUSchedOnln),
+    HalfDirtyCPUSchedOnln = erlang:system_info(dirty_cpu_schedulers_online),
+    QrtrDirtyCPUSchedOnln = HalfDirtyCPUSchedOnln div 2,
+    SchedOnln = erlang:system_flag(schedulers_online, HalfSchedOnln),
+    QrtrDirtyCPUSchedOnln = erlang:system_info(dirty_cpu_schedulers_online),
+    ok.
+
 get_sstate(Config, Cmd) ->
     {ok, Node} = start_node(Config, Cmd),
     [SState] = mcall(Node, [fun () ->
@@ -1100,6 +1151,20 @@ get_sstate(Config, Cmd) ->
     stop_node(Node),
     SState.
 
+get_dsstate(Config, Cmd) ->
+    {ok, Node} = start_node(Config, Cmd),
+    [DSCPU] = mcall(Node, [fun () ->
+				   erlang:system_info(dirty_cpu_schedulers)
+			   end]),
+    [DSCPUOnln] = mcall(Node, [fun () ->
+				       erlang:system_info(dirty_cpu_schedulers_online)
+			       end]),
+    [DSIO] = mcall(Node, [fun () ->
+				  erlang:system_info(dirty_io_schedulers)
+			  end]),
+    stop_node(Node),
+    {DSCPU, DSCPUOnln, DSIO}.
+
 scheduler_suspend(Config) when is_list(Config) ->
     ?line Dog = ?t:timetrap(?t:minutes(5)),
     ?line lists:foreach(fun (S) -> scheduler_suspend_test(Config, S) end,
@@ -1171,16 +1236,40 @@ sst2_loop(N) ->
     erlang:system_flag(multi_scheduling, unblock),
     sst2_loop(N-1).
 
-sst3_loop(_S, 0) ->
-    ok;
 sst3_loop(S, N) ->
+    try erlang:system_info(dirty_cpu_schedulers) of
+	DS ->
+	    sst3_loop_with_dirty_schedulers(S, DS, N)
+    catch
+	error:badarg ->
+	    sst3_loop_normal_schedulers_only(S, N)
+    end.
+
+sst3_loop_normal_schedulers_only(_S, 0) ->
+    ok;
+sst3_loop_normal_schedulers_only(S, N) ->
+    erlang:system_flag(schedulers_online, (S div 2)+1),
+    erlang:system_flag(schedulers_online, 1),
+    erlang:system_flag(schedulers_online, (S div 2)+1),
+    erlang:system_flag(schedulers_online, S),
+    erlang:system_flag(schedulers_online, 1),
+    erlang:system_flag(schedulers_online, S),
+    sst3_loop_normal_schedulers_only(S, N-1).
+
+sst3_loop_with_dirty_schedulers(_S, _DS, 0) ->
+    ok;
+sst3_loop_with_dirty_schedulers(S, DS, N) ->
     erlang:system_flag(schedulers_online, (S div 2)+1),
+    erlang:system_flag(dirty_cpu_schedulers_online, (DS div 2)+1),
     erlang:system_flag(schedulers_online, 1),
     erlang:system_flag(schedulers_online, (S div 2)+1),
+    erlang:system_flag(dirty_cpu_schedulers_online, 1),
     erlang:system_flag(schedulers_online, S),
+    erlang:system_flag(dirty_cpu_schedulers_online, DS),
     erlang:system_flag(schedulers_online, 1),
     erlang:system_flag(schedulers_online, S),
-    sst3_loop(S, N-1).
+    erlang:system_flag(dirty_cpu_schedulers_online, DS),
+    sst3_loop_with_dirty_schedulers(S, DS, N-1).
 
 reader_groups(Config) when is_list(Config) ->
     %% White box testing. These results are correct, but other results