dialyzer: Optimize memory consumption

Some SCC info is no longer cached. It reduces the peak memory
consumption of workers during the typesig phase, and seems to cost
very little time.

5 files changed, 50 insertions, 38 deletions

diff --git a/lib/dialyzer/src/dialyzer_analysis_callgraph.erl b/lib/dialyzer/src/dialyzer_analysis_callgraph.erl
index 8b3f70aa23..c8129cfbaa 100644
--- a/lib/dialyzer/src/dialyzer_analysis_callgraph.erl
+++ b/lib/dialyzer/src/dialyzer_analysis_callgraph.erl
@@ -121,6 +121,7 @@ loop(#server_state{parent = Parent} = State,
 %% The Analysis
 %%--------------------------------------------------------------------
 
+%% Calls to erlang:garbage_collect() help to reduce the heap size.
 analysis_start(Parent, Analysis, LegalWarnings) ->
   CServer = dialyzer_codeserver:new(),
   Plt = Analysis#analysis.plt,
@@ -157,6 +158,7 @@ analysis_start(Parent, Analysis, LegalWarnings) ->
       TmpCServer1 = dialyzer_codeserver:set_temp_records(MergedRecords, TmpCServer0),
       TmpCServer2 =
         dialyzer_codeserver:finalize_exported_types(MergedExpTypes, TmpCServer1),
+      erlang:garbage_collect(),
       ?timing(State#analysis_state.timing_server, "remote",
               begin
                 TmpCServer3 =
@@ -185,8 +187,6 @@ analysis_start(Parent, Analysis, LegalWarnings) ->
       true -> dialyzer_callgraph:put_race_detection(true, Callgraph);
       false -> Callgraph
     end,
-  %% Calls to erlang:garbage_collect() help to reduce the heap size.
-  %% An alternative is to spawn more processes to do the job(s).
   erlang:garbage_collect(),
   State2 = analyze_callgraph(NewCallgraph, State1),
   #analysis_state{plt = MiniPlt2, doc_plt = DocPlt} = State2,
diff --git a/lib/dialyzer/src/dialyzer_codeserver.erl b/lib/dialyzer/src/dialyzer_codeserver.erl
index 03cd9671af..5028c4f3ba 100644
--- a/lib/dialyzer/src/dialyzer_codeserver.erl
+++ b/lib/dialyzer/src/dialyzer_codeserver.erl
@@ -48,6 +48,7 @@
 	 is_exported/2,
 	 lookup_mod_code/2,
 	 lookup_mfa_code/2,
+	 lookup_mfa_var_label/2,
 	 lookup_mod_records/2,
 	 lookup_mod_contracts/2,
 	 lookup_mfa_contract/2,
@@ -173,8 +174,8 @@ insert(Mod, ModCode, CS) ->
   As = cerl:get_ann(ModCode),
   Funs =
     [{{Mod, cerl:fname_id(Var), cerl:fname_arity(Var)},
-      Val} || Val = {Var, _Fun} <- Defs],
-  Keys = [Key || {Key, _Value} <- Funs],
+      Val, {Var, cerl_trees:get_label(Fun)}} || Val = {Var, Fun} <- Defs],
+  Keys = [Key || {Key, _Value, _Label} <- Funs],
   ModEntry = {Mod, {Name, Exports, Attrs, Keys, As}},
   true = ets:insert(CS#codeserver.code, [ModEntry|Funs]),
   CS.
@@ -237,6 +238,11 @@ lookup_mod_code(Mod, CS) when is_atom(Mod) ->
 lookup_mfa_code({_M, _F, _A} = MFA, CS) ->
   table__lookup(CS#codeserver.code, MFA).
 
+-spec lookup_mfa_var_label(mfa(), codeserver()) -> {cerl:c_var(), label()}.
+
+lookup_mfa_var_label({_M, _F, _A} = MFA, CS) ->
+  ets:lookup_element(CS#codeserver.code, MFA, 3).
+
 -spec get_next_core_label(codeserver()) -> label().
 
 get_next_core_label(#codeserver{next_core_label = NCL}) ->
diff --git a/lib/dialyzer/src/dialyzer_contracts.erl b/lib/dialyzer/src/dialyzer_contracts.erl
index 82275d5df7..2cca16d085 100644
--- a/lib/dialyzer/src/dialyzer_contracts.erl
+++ b/lib/dialyzer/src/dialyzer_contracts.erl
@@ -176,7 +176,7 @@ process_contract_remote_types(CodeServer) ->
   NewCallbackList = lists:map(ModuleFun, dict:to_list(TmpCallbackDict)),
   NewContractDict = dict:from_list(NewContractList),
   NewCallbackDict = dict:from_list(NewCallbackList),
-  %% Make sure the (huge) cache is garbage collected:
+  %% Make sure temporary data and the (huge) cache are garbage collected:
   erlang:garbage_collect(),
   dialyzer_codeserver:finalize_contracts(NewContractDict, NewCallbackDict,
                                          CodeServer).
diff --git a/lib/dialyzer/src/dialyzer_succ_typings.erl b/lib/dialyzer/src/dialyzer_succ_typings.erl
index 986def52bf..d92173875c 100644
--- a/lib/dialyzer/src/dialyzer_succ_typings.erl
+++ b/lib/dialyzer/src/dialyzer_succ_typings.erl
@@ -349,21 +349,25 @@ find_succ_typings(SCCs, #st{codeserver = Codeserver, callgraph = Callgraph,
 
 -spec find_succ_types_for_scc(scc(), typesig_init_data()) -> [mfa_or_funlbl()].
 
-find_succ_types_for_scc(SCC, {Codeserver, Callgraph, Plt, Solvers}) ->
-  SCC_Info = [{MFA, 
-	       dialyzer_codeserver:lookup_mfa_code(MFA, Codeserver),
-	       dialyzer_codeserver:lookup_mod_records(M, Codeserver)}
-	      || {M, _, _} = MFA <- SCC],
+find_succ_types_for_scc(SCC0, {Codeserver, Callgraph, Plt, Solvers}) ->
+  SCC = [MFA || {_, _, _} = MFA <- SCC0],
   Contracts1 = [{MFA, dialyzer_codeserver:lookup_mfa_contract(MFA, Codeserver)}
-		|| {_, _, _} = MFA <- SCC],
+		|| MFA <- SCC],
   Contracts2 = [{MFA, Contract} || {MFA, {ok, Contract}} <- Contracts1],
   Contracts3 = orddict:from_list(Contracts2),
   Label = dialyzer_codeserver:get_next_core_label(Codeserver),
-  AllFuns = collect_fun_info([Fun || {_MFA, {_Var, Fun}, _Rec} <- SCC_Info]),
+  AllFuns = lists:append(
+              [begin
+                 {_Var, Fun} =
+                   dialyzer_codeserver:lookup_mfa_code(MFA, Codeserver),
+                 collect_fun_info([Fun])
+               end || MFA <- SCC]),
+  erlang:garbage_collect(),
   PropTypes = get_fun_types_from_plt(AllFuns, Callgraph, Plt),
   %% Assume that the PLT contains the current propagated types
-  FunTypes = dialyzer_typesig:analyze_scc(SCC_Info, Label, Callgraph,
-                                          Plt, PropTypes, Solvers),
+  FunTypes = dialyzer_typesig:analyze_scc(SCC, Label, Callgraph,
+                                          Codeserver, Plt, PropTypes,
+                                          Solvers),
   AllFunSet = sets:from_list([X || {X, _} <- AllFuns]),
   FilteredFunTypes =
     dict:filter(fun(X, _) -> sets:is_element(X, AllFunSet) end, FunTypes),
diff --git a/lib/dialyzer/src/dialyzer_typesig.erl b/lib/dialyzer/src/dialyzer_typesig.erl
index 1787b66192..075b6ee165 100644
--- a/lib/dialyzer/src/dialyzer_typesig.erl
+++ b/lib/dialyzer/src/dialyzer_typesig.erl
@@ -29,7 +29,7 @@
 
 -module(dialyzer_typesig).
 
--export([analyze_scc/6]).
+-export([analyze_scc/7]).
 -export([get_safe_underapprox/2]).
 
 %%-import(helper, %% 'helper' could be any module doing sanity checks...
@@ -101,7 +101,6 @@
 
 -type types() :: erl_types:type_table().
 
--type typesig_scc()    :: [{mfa(), {cerl:c_var(), cerl:c_fun()}, types()}].
 -type typesig_funmap() :: #{type_var() => type_var()}.
 
 -type prop_types() :: dict:dict(label(), types()).
@@ -160,11 +159,10 @@
 %%-----------------------------------------------------------------------------
 %% Analysis of strongly connected components.
 %%
-%% analyze_scc(SCC, NextLabel, CallGraph, PLT, PropTypes, Solvers) -> FunTypes
+%% analyze_scc(SCC, NextLabel, CallGraph, CodeServer,
+%%             PLT, PropTypes, Solvers) -> FunTypes
 %%
-%% SCC       - [{MFA, Def, Records}]
-%%             where Def = {Var, Fun} as in the Core Erlang module definitions.
-%%                   Records = dict(RecName, {Arity, [{FieldName, FieldType}]})
+%% SCC       - [{MFA}]
 %% NextLabel - An integer that is higher than any label in the code.
 %% CallGraph - A callgraph as produced by dialyzer_callgraph.erl
 %%             Note: The callgraph must have been built with all the
@@ -176,28 +174,27 @@
 %% Solvers   - User specified solvers.
 %%-----------------------------------------------------------------------------
 
--spec analyze_scc(typesig_scc(), label(),
+-spec analyze_scc([mfa()], label(),
 		  dialyzer_callgraph:callgraph(),
+                  dialyzer_codeserver:codeserver(),
 		  dialyzer_plt:plt(), prop_types(), [solver()]) -> prop_types().
 
-analyze_scc(SCC, NextLabel, CallGraph, Plt, PropTypes, Solvers0) ->
+analyze_scc(SCC, NextLabel, CallGraph, CServer, Plt, PropTypes, Solvers0) ->
   Solvers = solvers(Solvers0),
-  assert_format_of_scc(SCC),
-  State1 = new_state(SCC, NextLabel, CallGraph, Plt, PropTypes, Solvers),
-  DefSet = add_def_list([Var || {_MFA, {Var, _Fun}, _Rec} <- SCC], sets:new()),
-  State2 = traverse_scc(SCC, DefSet, State1),
+  State1 = new_state(SCC, NextLabel, CallGraph, CServer, Plt, PropTypes,
+                     Solvers),
+  DefSet = add_def_list(maps:values(State1#state.name_map), sets:new()),
+  ModRecs = [{M, dialyzer_codeserver:lookup_mod_records(M, CServer)} ||
+              M <- lists:usort([M || {M, _, _} <- SCC])],
+  State2 = traverse_scc(SCC, CServer, DefSet, ModRecs, State1),
   State3 = state__finalize(State2),
+  erlang:garbage_collect(),
   Funs = state__scc(State3),
   pp_constrs_scc(Funs, State3),
   constraints_to_dot_scc(Funs, State3),
   T = solve(Funs, State3),
   dict:from_list(maps:to_list(T)).
 
-assert_format_of_scc([{_MFA, {_Var, _Fun}, _Records}|Left]) ->
-  assert_format_of_scc(Left);
-assert_format_of_scc([]) ->
-  ok.
-
 solvers([]) -> [v2];
 solvers(Solvers) -> Solvers.
 
@@ -207,12 +204,14 @@ solvers(Solvers) -> Solvers.
 %%
 %% ============================================================================
 
-traverse_scc([{_MFA, Def, Rec}|Left], DefSet, AccState) ->
+traverse_scc([{M,_,_}=MFA|Left], Codeserver, DefSet, ModRecs, AccState) ->
+  Def = dialyzer_codeserver:lookup_mfa_code(MFA, Codeserver),
+  {M, Rec} = lists:keyfind(M, 1, ModRecs),
   TmpState1 = state__set_rec_dict(AccState, Rec),
   DummyLetrec = cerl:c_letrec([Def], cerl:c_atom(foo)),
   {NewAccState, _} = traverse(DummyLetrec, DefSet, TmpState1),
-  traverse_scc(Left, DefSet, NewAccState);
-traverse_scc([], _DefSet, AccState) ->
+  traverse_scc(Left, Codeserver, DefSet, ModRecs, NewAccState);
+traverse_scc([], _Codeserver, _DefSet, _ModRecs, AccState) ->
   AccState.
 
 traverse(Tree, DefinedVars, State) ->
@@ -2702,11 +2701,14 @@ pp_map(_S, _Map) ->
 %%
 %% ============================================================================
 
-new_state(SCC0, NextLabel, CallGraph, Plt, PropTypes, Solvers) ->
-  List = [{MFA, Var} || {MFA, {Var, _Fun}, _Rec} <- SCC0],
+new_state(MFAs, NextLabel, CallGraph, CServer, Plt, PropTypes, Solvers) ->
+  List_SCC =
+    [begin
+       {Var, Label} = dialyzer_codeserver:lookup_mfa_var_label(MFA, CServer),
+       {{MFA, Var}, t_var(Label)}
+   end || MFA <- MFAs],
+  {List, SCC} = lists:unzip(List_SCC),
   NameMap = maps:from_list(List),
-  MFAs = [MFA || {MFA, _Var} <- List],
-  SCC = [mk_var(Fun) || {_MFA, {_Var, Fun}, _Rec} <- SCC0],
   SelfRec =
     case SCC of
       [OneF] ->