Allow caching HiPE-compilation results

That way HiPE compilation can be performed during package installation
and will not waste time during every startup.

rabbit_hipe is refactored to support both modes of compilation - during
every server startup or separately with caching in the filesystem.

6 files changed, 121 insertions, 25 deletions

diff --git a/docs/rabbitmqctl.1.xml b/docs/rabbitmqctl.1.xml
index 1ecf8d4d3a..ec864af6cf 100644
--- a/docs/rabbitmqctl.1.xml
+++ b/docs/rabbitmqctl.1.xml
@@ -290,6 +290,33 @@
             </para>
           </listitem>
         </varlistentry>
+
+        <varlistentry>
+          <term><cmdsynopsis><command>hipe_compile</command> <arg choice="req"><replaceable>directory</replaceable></arg></cmdsynopsis></term>
+          <listitem>
+            <para>
+              Performs HiPE-compilation and caches resulting
+              .beam-files in the given directory.
+            </para>
+            <para>
+              Parent directories are created if necessary. Any
+              existing <command>.beam</command> files from the
+              directory are automatically deleted prior to
+              compilation.
+            </para>
+            <para>
+              To use this precompiled files, you should set
+              <command>RABBITMQ_SERVER_CODE_PATH</command> environment
+              variable to directory specified in
+              <command>hipe_compile</command> invokation.
+            </para>
+            <para role="example-prefix">For example:</para>
+            <screen role="example">rabbitmqctl hipe_compile /tmp/rabbit-hipe/ebin</screen>
+            <para role="example">
+              HiPE-compiles modules and stores them to /tmp/rabbit-hipe/ebin directory.
+            </para>
+          </listitem>
+        </varlistentry>
       </variablelist>
     </refsect2>
 
diff --git a/scripts/rabbitmq-env b/scripts/rabbitmq-env
index 35239620ca..def47ebd27 100644..100755
--- a/scripts/rabbitmq-env
+++ b/scripts/rabbitmq-env
@@ -177,6 +177,7 @@ DEFAULT_NODE_PORT=5672
 [ "x" = "x$RABBITMQ_MNESIA_BASE" ] && RABBITMQ_MNESIA_BASE=${MNESIA_BASE}
 [ "x" = "x$RABBITMQ_SERVER_START_ARGS" ] && RABBITMQ_SERVER_START_ARGS=${SERVER_START_ARGS}
 [ "x" = "x$RABBITMQ_SERVER_ADDITIONAL_ERL_ARGS" ] && RABBITMQ_SERVER_ADDITIONAL_ERL_ARGS=${SERVER_ADDITIONAL_ERL_ARGS}
+[ "x" = "x$RABBITMQ_SERVER_CODE_PATH" ] && RABBITMQ_SERVER_CODE_PATH=${SERVER_CODE_PATH}
 [ "x" = "x$RABBITMQ_MNESIA_DIR" ] && RABBITMQ_MNESIA_DIR=${MNESIA_DIR}
 [ "x" = "x$RABBITMQ_MNESIA_DIR" ] && RABBITMQ_MNESIA_DIR=${RABBITMQ_MNESIA_BASE}/${RABBITMQ_NODENAME}
 
diff --git a/scripts/rabbitmq-server b/scripts/rabbitmq-server
index ab2975feb1..74337311cd 100755
--- a/scripts/rabbitmq-server
+++ b/scripts/rabbitmq-server
@@ -133,11 +133,15 @@ ensure_thread_pool_size() {
 }
 
 start_rabbitmq_server() {
+    # "-pa ${RABBITMQ_SERVER_CODE_PATH}" should be the very first
+    # command-line argument. In case of using cached HiPE-compilation,
+    # this will allow for compiled versions of erlang built-in modules
+    # (e.g. lists) to be loaded.
     ensure_thread_pool_size
     check_start_params &&
     RABBITMQ_CONFIG_FILE=$RABBITMQ_CONFIG_FILE \
     exec ${ERL_DIR}erl \
-        -pa ${RABBITMQ_EBIN_ROOT} \
+        -pa ${RABBITMQ_SERVER_CODE_PATH} ${RABBITMQ_EBIN_ROOT} \
         ${RABBITMQ_START_RABBIT} \
         ${RABBITMQ_NAME_TYPE} ${RABBITMQ_NODENAME} \
         -boot "${SASL_BOOT_FILE}" \
diff --git a/src/rabbit_control_main.erl b/src/rabbit_control_main.erl
index f63694b657..2df4fd96c0 100644
--- a/src/rabbit_control_main.erl
+++ b/src/rabbit_control_main.erl
@@ -37,6 +37,7 @@
          reset,
          force_reset,
          rotate_logs,
+         hipe_compile,
 
          {join_cluster, [?RAM_DEF]},
          change_cluster_node_type,
@@ -113,7 +114,7 @@
         [stop, stop_app, start_app, wait, reset, force_reset, rotate_logs,
          join_cluster, change_cluster_node_type, update_cluster_nodes,
          forget_cluster_node, rename_cluster_node, cluster_status, status,
-         environment, eval, force_boot, help, node_health_check]).
+         environment, eval, force_boot, help, node_health_check, hipe_compile]).
 
 -define(COMMANDS_WITH_TIMEOUT,
         [list_user_permissions, list_policies, list_queues, list_exchanges,
@@ -383,6 +384,16 @@ action(rotate_logs, Node, Args = [Suffix], _Opts, Inform) ->
     Inform("Rotating logs to files with suffix \"~s\"", [Suffix]),
     call(Node, {rabbit, rotate_logs, Args});
 
+action(hipe_compile, _Node, [TargetDir], _Opts, _Inform) ->
+    ok = application:load(rabbit),
+    case rabbit_hipe:can_hipe_compile() of
+        true ->
+            {ok, _, _} = rabbit_hipe:compile_to_directory(TargetDir),
+            ok;
+        false ->
+            {error, "HiPE compilation is not supported"}
+    end;
+
 action(close_connection, Node, [PidStr, Explanation], _Opts, Inform) ->
     Inform("Closing connection \"~s\"", [PidStr]),
     rpc_call(Node, rabbit_networking, close_connection,
diff --git a/src/rabbit_file.erl b/src/rabbit_file.erl
index 6c4f0e5ccd..1088f2c2dd 100644
--- a/src/rabbit_file.erl
+++ b/src/rabbit_file.erl
@@ -23,6 +23,7 @@
 -export([append_file/2, ensure_parent_dirs_exist/1]).
 -export([rename/2, delete/1, recursive_delete/1, recursive_copy/2]).
 -export([lock_file/1]).
+-export([filename_as_a_directory/1]).
 
 -import(file_handle_cache, [with_handle/1, with_handle/2]).
 
@@ -58,6 +59,7 @@
         (file:filename(), file:filename())
         -> rabbit_types:ok_or_error({file:filename(), file:filename(), any()})).
 -spec(lock_file/1 :: (file:filename()) -> rabbit_types:ok_or_error('eexist')).
+-spec(filename_as_a_directory/1 :: (file:filename()) -> file:filename()).
 
 -endif.
 
@@ -305,3 +307,11 @@ lock_file(Path) ->
                              ok = prim_file:close(Lock)
                    end)
     end.
+
+filename_as_a_directory(FileName) ->
+    case lists:last(FileName) of
+        "/" ->
+            FileName;
+        _ ->
+            FileName ++ "/"
+    end.
diff --git a/src/rabbit_hipe.erl b/src/rabbit_hipe.erl
index 05b5f3719d..494f0e06b7 100644
--- a/src/rabbit_hipe.erl
+++ b/src/rabbit_hipe.erl
@@ -5,15 +5,15 @@
 %% practice 2 processes seems just as fast as any other number > 1,
 %% and keeps the progress bar realistic-ish.
 -define(HIPE_PROCESSES, 2).
--export([maybe_hipe_compile/0, log_hipe_result/1]).
 
-%% HiPE compilation happens before we have log handlers - so we have
-%% to io:format/2, it's all we can do.
+-export([maybe_hipe_compile/0, log_hipe_result/1]).
+-export([compile_to_directory/1]).
+-export([can_hipe_compile/0]).
 
+%% Compile and load during server startup sequence
 maybe_hipe_compile() ->
     {ok, Want} = application:get_env(rabbit, hipe_compile),
-    Can = code:which(hipe) =/= non_existing,
-    case {Want, Can} of
+    case {Want, can_hipe_compile()} of
         {true,  true}  -> hipe_compile();
         {true,  false} -> false;
         {false, _}     -> {ok, disabled}
@@ -33,38 +33,49 @@ log_hipe_result(false) ->
     rabbit_log:warning(
       "Not HiPE compiling: HiPE not found in this Erlang installation.~n").
 
+hipe_compile() ->
+    hipe_compile(fun compile_and_load/1, false).
+
+compile_to_directory(Dir0) ->
+    Dir = rabbit_file:filename_as_a_directory(Dir0),
+    ok = prepare_ebin_directory(Dir),
+    hipe_compile(fun (Mod) -> compile_and_save(Mod, Dir) end, true).
+
+need_compile(Mod, Force) ->
+    Exists = code:which(Mod) =/= non_existing,
+    %% We skip modules already natively compiled. This
+    %% happens when RabbitMQ is stopped (just the
+    %% application, not the entire node) and started
+    %% again.
+    NotYetCompiled = not already_hipe_compiled(Mod),
+    NotVersioned = not compiled_with_version_support(Mod),
+    Exists andalso (Force orelse (NotYetCompiled andalso NotVersioned)).
+
 %% HiPE compilation happens before we have log handlers and can take a
 %% long time, so make an exception to our no-stdout policy and display
 %% progress via stdout.
-hipe_compile() ->
+hipe_compile(CompileFun, Force) ->
     {ok, HipeModulesAll} = application:get_env(rabbit, hipe_modules),
-    HipeModules = [HM || HM <- HipeModulesAll,
-                   code:which(HM) =/= non_existing andalso
-                   %% We skip modules already natively compiled. This
-                   %% happens when RabbitMQ is stopped (just the
-                   %% application, not the entire node) and started
-                   %% again.
-                   already_hipe_compiled(HM)
-                             andalso (not compiled_with_version_support(HM))],
+    HipeModules = lists:filter(fun(Mod) -> need_compile(Mod, Force) end, HipeModulesAll),
     case HipeModules of
         [] -> {ok, already_compiled};
-        _  -> do_hipe_compile(HipeModules)
+        _  -> do_hipe_compile(HipeModules, CompileFun)
     end.
 
 already_hipe_compiled(Mod) ->
     try
     %% OTP 18.x or later
-	Mod:module_info(native) =:= false
+        Mod:module_info(native) =:= true
     %% OTP prior to 18.x
     catch error:badarg ->
-	code:is_module_native(Mod) =:= false
+        code:is_module_native(Mod) =:= true
     end.
 
 compiled_with_version_support(Mod) ->
     proplists:get_value(erlang_version_support, Mod:module_info(attributes))
         =/= undefined.
 
-do_hipe_compile(HipeModules) ->
+do_hipe_compile(HipeModules, CompileFun) ->
     Count = length(HipeModules),
     io:format("~nHiPE compiling:  |~s|~n                 |",
               [string:copies("-", Count)]),
@@ -79,11 +90,7 @@ do_hipe_compile(HipeModules) ->
     %% advanced API does not load automatically the code, except if the
     %% 'load' option is set.
     PidMRefs = [spawn_monitor(fun () -> [begin
-                                             {M, Beam, _} =
-                                               code:get_object_code(M),
-                                             {ok, _} =
-                                               hipe:compile(M, [], Beam,
-                                                            [o3, load]),
+                                             CompileFun(M),
                                              io:format("#")
                                          end || M <- Ms]
                               end) ||
@@ -101,3 +108,39 @@ split(L, N) -> split0(L, [[] || _ <- lists:seq(1, N)]).
 
 split0([],       Ls)       -> Ls;
 split0([I | Is], [L | Ls]) -> split0(Is, Ls ++ [[I | L]]).
+
+prepare_ebin_directory(Dir) ->
+    ok = rabbit_file:ensure_dir(Dir),
+    ok = delete_beam_files(Dir),
+    ok.
+
+delete_beam_files(Dir) ->
+    {ok, Files} = file:list_dir(Dir),
+    lists:foreach(fun(File) ->
+                          case filename:extension(File) of
+                              ".beam" ->
+                                  ok = file:delete(filename:join([Dir, File]));
+                              _ ->
+                                  ok
+                          end
+                  end,
+                  Files).
+
+compile_and_load(Mod) ->
+    {Mod, Beam, _} = code:get_object_code(Mod),
+    {ok, _} = hipe:compile(Mod, [], Beam, [o3, load]).
+
+compile_and_save(Module, Dir) ->
+    {Module, BeamCode, _} = code:get_object_code(Module),
+    BeamName = filename:join([Dir, atom_to_list(Module) ++ ".beam"]),
+    {ok, {Architecture, NativeCode}} = hipe:compile(Module, [], BeamCode, [o3]),
+    {ok, _, Chunks0} = beam_lib:all_chunks(BeamCode),
+    ChunkName = hipe_unified_loader:chunk_name(Architecture),
+    Chunks1 = lists:keydelete(ChunkName, 1, Chunks0),
+    Chunks = Chunks1 ++ [{ChunkName,NativeCode}],
+    {ok, BeamPlusNative} = beam_lib:build_module(Chunks),
+    ok = file:write_file(BeamName, BeamPlusNative),
+    BeamName.
+
+can_hipe_compile() ->
+    code:which(hipe) =/= non_existing.