diff options
Diffstat (limited to 'lib/compiler/src')
40 files changed, 3529 insertions, 2128 deletions
diff --git a/lib/compiler/src/Makefile b/lib/compiler/src/Makefile index ef6db66ff6..9e96147787 100644 --- a/lib/compiler/src/Makefile +++ b/lib/compiler/src/Makefile @@ -83,6 +83,7 @@ MODULES = \ core_scan \ erl_bifs \ rec_env \ + sys_core_alias \ sys_core_bsm \ sys_core_dsetel \ sys_core_fold \ @@ -92,16 +93,14 @@ MODULES = \ v3_codegen \ v3_core \ v3_kernel \ - v3_kernel_pp \ - v3_life + v3_kernel_pp BEAM_H = $(wildcard ../priv/beam_h/*.h) HRL_FILES= \ beam_disasm.hrl \ core_parse.hrl \ - v3_kernel.hrl \ - v3_life.hrl + v3_kernel.hrl YRL_FILE = core_parse.yrl @@ -186,7 +185,7 @@ release_docs_spec: # ---------------------------------------------------- $(EBIN)/beam_disasm.beam: $(EGEN)/beam_opcodes.hrl beam_disasm.hrl -$(EBIN)/beam_listing.beam: v3_life.hrl +$(EBIN)/beam_listing.beam: core_parse.hrl v3_kernel.hrl $(EBIN)/beam_validator.beam: beam_disasm.hrl $(EBIN)/cerl.beam: core_parse.hrl $(EBIN)/compile.beam: core_parse.hrl ../../stdlib/include/erl_compile.hrl @@ -194,12 +193,12 @@ $(EBIN)/core_lib.beam: core_parse.hrl $(EBIN)/core_lint.beam: core_parse.hrl $(EBIN)/core_parse.beam: core_parse.hrl $(EGEN)/core_parse.erl $(EBIN)/core_pp.beam: core_parse.hrl +$(EBIN)/sys_core_alias.beam: core_parse.hrl $(EBIN)/sys_core_dsetel.beam: core_parse.hrl $(EBIN)/sys_core_fold.beam: core_parse.hrl $(EBIN)/sys_core_fold_lists.beam: core_parse.hrl $(EBIN)/sys_core_inline.beam: core_parse.hrl -$(EBIN)/v3_codegen.beam: v3_life.hrl +$(EBIN)/v3_codegen.beam: v3_kernel.hrl $(EBIN)/v3_core.beam: core_parse.hrl $(EBIN)/v3_kernel.beam: core_parse.hrl v3_kernel.hrl $(EBIN)/v3_kernel_pp.beam: v3_kernel.hrl -$(EBIN)/v3_life.beam: v3_kernel.hrl v3_life.hrl diff --git a/lib/compiler/src/beam_a.erl b/lib/compiler/src/beam_a.erl index cdb32d5d55..91acb19971 100644 --- a/lib/compiler/src/beam_a.erl +++ b/lib/compiler/src/beam_a.erl @@ -42,8 +42,7 @@ function({function,Name,Arity,CLabel,Is0}) -> Is = beam_jump:remove_unused_labels(Is1), {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -59,8 +58,17 @@ rename_instrs([{call_only,A,F}|Is]) -> rename_instrs([{call_ext_only,A,F}|Is]) -> [{call_ext,A,F},return|rename_instrs(Is)]; rename_instrs([{'%live',_}|Is]) -> - %% When compiling from old .S files. + %% Ignore old type of live annotation. Only happens when compiling + %% from very old .S files. rename_instrs(Is); +rename_instrs([{get_list,S,D1,D2}|Is]) -> + %% Only happens when compiling from old .S files. + if + D1 =:= S -> + [{get_tl,S,D2},{get_hd,S,D1}|rename_instrs(Is)]; + true -> + [{get_hd,S,D1},{get_tl,S,D2}|rename_instrs(Is)] + end; rename_instrs([I|Is]) -> [rename_instr(I)|rename_instrs(Is)]; rename_instrs([]) -> []. diff --git a/lib/compiler/src/beam_asm.erl b/lib/compiler/src/beam_asm.erl index f7c838e392..fa919ca862 100644 --- a/lib/compiler/src/beam_asm.erl +++ b/lib/compiler/src/beam_asm.erl @@ -21,10 +21,10 @@ -module(beam_asm). --export([module/5]). +-export([module/4]). -export([encode/2]). --export_type([fail/0,label/0,reg/0,src/0,module_code/0,function_name/0]). +-export_type([fail/0,label/0,reg/0,reg_num/0,src/0,module_code/0,function_name/0]). -import(lists, [map/2,member/2,keymember/3,duplicate/2,splitwith/2]). -include("beam_opcodes.hrl"). @@ -55,20 +55,20 @@ -type module_code() :: {module(),[_],[_],[asm_function()],pos_integer()}. --spec module(module_code(), [{binary(), binary()}], [_], [compile:option()], [compile:option()]) -> +-spec module(module_code(), [{binary(), binary()}], [{atom(),term()}], [compile:option()]) -> {'ok',binary()}. -module(Code, ExtraChunks, SourceFile, Opts, CompilerOpts) -> - {ok,assemble(Code, ExtraChunks, SourceFile, Opts, CompilerOpts)}. +module(Code, ExtraChunks, CompileInfo, CompilerOpts) -> + {ok,assemble(Code, ExtraChunks, CompileInfo, CompilerOpts)}. -assemble({Mod,Exp0,Attr0,Asm0,NumLabels}, ExtraChunks, SourceFile, Opts, CompilerOpts) -> +assemble({Mod,Exp0,Attr0,Asm0,NumLabels}, ExtraChunks, CompileInfo, CompilerOpts) -> {1,Dict0} = beam_dict:atom(Mod, beam_dict:new()), {0,Dict1} = beam_dict:fname(atom_to_list(Mod) ++ ".erl", Dict0), NumFuncs = length(Asm0), {Asm,Attr} = on_load(Asm0, Attr0), Exp = cerl_sets:from_list(Exp0), {Code,Dict2} = assemble_1(Asm, Exp, Dict1, []), - build_file(Code, Attr, Dict2, NumLabels, NumFuncs, ExtraChunks, SourceFile, Opts, CompilerOpts). + build_file(Code, Attr, Dict2, NumLabels, NumFuncs, ExtraChunks, CompileInfo, CompilerOpts). on_load(Fs0, Attr0) -> case proplists:get_value(on_load, Attr0) of @@ -111,7 +111,7 @@ assemble_function([H|T], Acc, Dict0) -> assemble_function([], Code, Dict) -> {Code, Dict}. -build_file(Code, Attr, Dict, NumLabels, NumFuncs, ExtraChunks, SourceFile, Opts, CompilerOpts) -> +build_file(Code, Attr, Dict, NumLabels, NumFuncs, ExtraChunks, CompileInfo, CompilerOpts) -> %% Create the code chunk. CodeChunk = chunk(<<"Code">>, @@ -182,8 +182,7 @@ build_file(Code, Attr, Dict, NumLabels, NumFuncs, ExtraChunks, SourceFile, Opts, Essentials1 = [iolist_to_binary(C) || C <- Essentials0], MD5 = module_md5(Essentials1), Essentials = finalize_fun_table(Essentials1, MD5), - {Attributes,Compile} = build_attributes(Opts, CompilerOpts, SourceFile, - Attr, MD5), + {Attributes,Compile} = build_attributes(Attr, CompileInfo, MD5), AttrChunk = chunk(<<"Attr">>, Attributes), CompileChunk = chunk(<<"CInf">>, Compile), @@ -193,7 +192,7 @@ build_file(Code, Attr, Dict, NumLabels, NumFuncs, ExtraChunks, SourceFile, Opts, %% Create IFF chunk. - Chunks = case member(slim, Opts) of + Chunks = case member(slim, CompilerOpts) of true -> [Essentials,AttrChunk]; false -> @@ -241,9 +240,7 @@ build_form(Id, Chunks0) when byte_size(Id) =:= 4, is_list(Chunks0) -> chunk(Id, Contents) when byte_size(Id) =:= 4, is_binary(Contents) -> Size = byte_size(Contents), - [<<Id/binary,Size:32>>,Contents|pad(Size)]; -chunk(Id, Contents) when is_list(Contents) -> - chunk(Id, list_to_binary(Contents)). + [<<Id/binary,Size:32>>,Contents|pad(Size)]. %% Build a correctly padded chunk (with a sub-header). @@ -265,22 +262,10 @@ flatten_exports(Exps) -> flatten_imports(Imps) -> list_to_binary(map(fun({M,F,A}) -> <<M:32,F:32,A:32>> end, Imps)). -build_attributes(Opts, CompilerOpts, SourceFile, Attr, MD5) -> - Misc0 = case SourceFile of - [] -> []; - [_|_] -> [{source,SourceFile}] - end, - Misc = case member(slim, CompilerOpts) of - false -> Misc0; - true -> [] - end, - Compile = case member(deterministic, CompilerOpts) of - false -> - [{options,Opts},{version,?COMPILER_VSN}|Misc]; - true -> - [{version,?COMPILER_VSN}] - end, - {term_to_binary(set_vsn_attribute(Attr, MD5)),term_to_binary(Compile)}. +build_attributes(Attr, Compile, MD5) -> + AttrBinary = term_to_binary(set_vsn_attribute(Attr, MD5)), + CompileBinary = term_to_binary([{version,?COMPILER_VSN}|Compile]), + {AttrBinary,CompileBinary}. build_line_table(Dict) -> {NumLineInstrs,NumFnames0,Fnames0,NumLines,Lines0} = @@ -422,7 +407,17 @@ encode_arg({atom, Atom}, Dict0) when is_atom(Atom) -> {Index, Dict} = beam_dict:atom(Atom, Dict0), {encode(?tag_a, Index), Dict}; encode_arg({integer, N}, Dict) -> - {encode(?tag_i, N), Dict}; + %% Conservatily assume that all integers whose absolute + %% value is greater than 1 bsl 128 will be bignums in + %% the runtime system. + if + N >= 1 bsl 128 -> + encode_arg({literal, N}, Dict); + N =< -(1 bsl 128) -> + encode_arg({literal, N}, Dict); + true -> + {encode(?tag_i, N), Dict} + end; encode_arg(nil, Dict) -> {encode(?tag_a, 0), Dict}; encode_arg({f, W}, Dict) -> diff --git a/lib/compiler/src/beam_block.erl b/lib/compiler/src/beam_block.erl index 6543e05e20..47a2be8ab5 100644 --- a/lib/compiler/src/beam_block.erl +++ b/lib/compiler/src/beam_block.erl @@ -23,30 +23,36 @@ -module(beam_block). -export([module/2]). --import(lists, [reverse/1,reverse/2,foldl/3,member/2]). +-import(lists, [reverse/1,reverse/2,member/2]). -spec module(beam_utils:module_code(), [compile:option()]) -> {'ok',beam_utils:module_code()}. -module({Mod,Exp,Attr,Fs0,Lc}, _Opt) -> - Fs = [function(F) || F <- Fs0], +module({Mod,Exp,Attr,Fs0,Lc}, Opts) -> + Blockify = not member(no_blockify, Opts), + Fs = [function(F, Blockify) || F <- Fs0], {ok,{Mod,Exp,Attr,Fs,Lc}}. -function({function,Name,Arity,CLabel,Is0}) -> +function({function,Name,Arity,CLabel,Is0}, Blockify) -> try %% Collect basic blocks and optimize them. - Is1 = blockify(Is0), - Is2 = embed_lines(Is1), - Is3 = move_allocates(Is2), - Is4 = beam_utils:live_opt(Is3), - Is5 = opt_blocks(Is4), - Is6 = beam_utils:delete_live_annos(Is5), - - %% Done. - {function,Name,Arity,CLabel,Is6} + Is1 = case Blockify of + false -> Is0; + true -> blockify(Is0) + end, + Is2 = embed_lines(Is1), + Is3 = local_cse(Is2), + Is4 = beam_utils:anno_defs(Is3), + Is5 = move_allocates(Is4), + Is6 = beam_utils:live_opt(Is5), + Is7 = opt_blocks(Is6), + Is8 = beam_utils:delete_annos(Is7), + Is = opt_allocs(Is8), + + %% Done. + {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -102,7 +108,8 @@ collect({put_tuple,A,D}) -> {set,[D],[],{put_tuple,A}}; collect({put,S}) -> {set,[],[S],put}; collect({get_tuple_element,S,I,D}) -> {set,[D],[S],{get_tuple_element,I}}; collect({set_tuple_element,S,D,I}) -> {set,[],[S,D],{set_tuple_element,I}}; -collect({get_list,S,D1,D2}) -> {set,[D1,D2],[S],get_list}; +collect({get_hd,S,D}) -> {set,[D],[S],get_hd}; +collect({get_tl,S,D}) -> {set,[D],[S],get_tl}; collect(remove_message) -> {set,[],[],remove_message}; collect({put_map,F,Op,S,D,R,{list,Puts}}) -> {set,[D],[S|Puts],{alloc,R,{put_map,Op,F}}}; @@ -130,23 +137,27 @@ embed_lines([{block,B2},{line,_}=Line,{block,B1}|T], Acc) -> embed_lines([{block,B1},{line,_}=Line|T], Acc) -> B = {block,[{set,[],[],Line}|B1]}, embed_lines([B|T], Acc); +embed_lines([{block,B2},{block,B1}|T], Acc) -> + %% This can only happen when beam_block is run for + %% the second time. + B = {block,B1++B2}, + embed_lines([B|T], Acc); embed_lines([I|Is], Acc) -> embed_lines(Is, [I|Acc]); embed_lines([], Acc) -> Acc. opt_blocks([{block,Bl0}|Is]) -> %% The live annotation at the beginning is not useful. - [{'%live',_,_}|Bl] = Bl0, + [{'%anno',_}|Bl] = Bl0, [{block,opt_block(Bl)}|opt_blocks(Is)]; opt_blocks([I|Is]) -> [I|opt_blocks(Is)]; opt_blocks([]) -> []. opt_block(Is0) -> - Is = find_fixpoint(fun(Is) -> - opt_tuple_element(opt(Is)) - end, Is0), - opt_alloc(Is). + find_fixpoint(fun(Is) -> + opt_tuple_element(opt(Is)) + end, Is0). find_fixpoint(OptFun, Is0) -> case OptFun(Is0) of @@ -173,7 +184,7 @@ find_fixpoint(OptFun, Is0) -> %% safe to assume that if x(N) is initialized, then all lower-numbered %% x registers are also initialized. %% -%% For example, in general it is not safe to transform the following +%% For example, we must be careful when transforming the following %% instructions: %% %% get_tuple_element x(0) Element => x(1) @@ -185,13 +196,9 @@ find_fixpoint(OptFun, Is0) -> %% get_tuple_element x(0) Element => x(1) %% %% The transformation is safe if and only if x(1) has been -%% initialized previously. Unfortunately, beam_reorder may have moved -%% a get_tuple_element instruction so that x(1) is not always -%% initialized when this code is reached. To find whether or not x(1) -%% is initialized, we would need to analyze all code preceding these -%% two instructions (across branches). Since we currently don't have -%% any practical mechanism for doing that, we will have to -%% conservatively assume that the transformation is unsafe. +%% initialized previously. We will use the annotations added by +%% beam_utils:anno_defs/1 to determine whether x(a) has been +%% initialized. move_allocates([{block,Bl0}|Is]) -> Bl = move_allocates_1(reverse(Bl0), []), @@ -200,40 +207,77 @@ move_allocates([I|Is]) -> [I|move_allocates(Is)]; move_allocates([]) -> []. -move_allocates_1([I|Is], [{set,[],[],{alloc,Live0,Info}}|Acc]=Acc0) -> - case {alloc_may_pass(I),alloc_live_regs(I, Live0)} of - {false,_} -> - move_allocates_1(Is, [I|Acc0]); - {true,not_possible} -> - move_allocates_1(Is, [I|Acc0]); - {true,Live} when is_integer(Live) -> - A = {set,[],[],{alloc,Live,Info}}, - move_allocates_1(Is, [A,I|Acc]) +move_allocates_1([{'%anno',_}|Is], Acc) -> + move_allocates_1(Is, Acc); +move_allocates_1([I|Is], [{set,[],[],{alloc,Live0,Info0}}|Acc]=Acc0) -> + case alloc_may_pass(I) of + false -> + move_allocates_1(Is, [I|Acc0]); + true -> + case alloc_live_regs(I, Is, Live0) of + not_possible -> + move_allocates_1(Is, [I|Acc0]); + Live when is_integer(Live) -> + Info = safe_info(Info0), + A = {set,[],[],{alloc,Live,Info}}, + move_allocates_1(Is, [A,I|Acc]) + end end; move_allocates_1([I|Is], Acc) -> move_allocates_1(Is, [I|Acc]); move_allocates_1([], Acc) -> Acc. +alloc_may_pass({set,_,[{fr,_}],fmove}) -> false; alloc_may_pass({set,_,_,{alloc,_,_}}) -> false; alloc_may_pass({set,_,_,{set_tuple_element,_}}) -> false; alloc_may_pass({set,_,_,put_list}) -> false; alloc_may_pass({set,_,_,put}) -> false; alloc_may_pass({set,_,_,_}) -> true. - + +safe_info({nozero,Stack,Heap,_}) -> + %% nozero is not safe if the allocation instruction is moved + %% upwards past an instruction that may throw an exception + %% (such as element/2). + {zero,Stack,Heap,[]}; +safe_info(Info) -> Info. + %% opt([Instruction]) -> [Instruction] %% Optimize the instruction stream inside a basic block. opt([{set,[X],[X],move}|Is]) -> opt(Is); +opt([{set,[Dst],_,move},{set,[Dst],[Src],move}=I|Is]) when Dst =/= Src -> + opt([I|Is]); +opt([{set,[{x,0}],[S1],move}=I1,{set,[D2],[{x,0}],move}|Is]) -> + opt([I1,{set,[D2],[S1],move}|Is]); +opt([{set,[{x,0}],[S1],move}=I1,{set,[D2],[S2],move}|Is0]) when S1 =/= D2 -> + %% Place move S x0 at the end of move sequences so that + %% loader can merge with the following instruction + {Ds,Is} = opt_moves([D2], Is0), + [{set,Ds,[S2],move}|opt([I1|Is])]; opt([{set,_,_,{line,_}}=Line1, {set,[D1],[{integer,Idx1},Reg],{bif,element,{f,0}}}=I1, {set,_,_,{line,_}}=Line2, {set,[D2],[{integer,Idx2},Reg],{bif,element,{f,0}}}=I2|Is]) when Idx1 < Idx2, D1 =/= D2, D1 =/= Reg, D2 =/= Reg -> opt([Line2,I2,Line1,I1|Is]); +opt([{set,[D1],[{integer,Idx1},Reg],{bif,element,{f,L}}}=I1, + {set,[D2],[{integer,Idx2},Reg],{bif,element,{f,L}}}=I2|Is]) + when Idx1 < Idx2, D1 =/= D2, D1 =/= Reg, D2 =/= Reg -> + opt([I2,I1|Is]); +opt([{set,Hd0,Cons,get_hd}=GetHd, + {set,Tl0,Cons,get_tl}=GetTl|Is0]) -> + case {opt_moves(Hd0, [GetTl|Is0]),opt_moves(Tl0, [GetHd|Is0])} of + {{Hd0,Is},{Tl0,_}} -> + [GetHd|opt(Is)]; + {{Hd,Is},{Tl0,_}} -> + [{set,Hd,Cons,get_hd}|opt(Is)]; + {{_,_},{Tl,Is}} -> + [{set,Tl,Cons,get_tl}|opt(Is)] + end; opt([{set,Ds0,Ss,Op}|Is0]) -> {Ds,Is} = opt_moves(Ds0, Is0), [{set,Ds,Ss,Op}|opt(Is)]; -opt([{'%live',_,_}=I|Is]) -> +opt([{'%anno',_}=I|Is]) -> [I|opt(Is)]; opt([]) -> []. @@ -245,17 +289,6 @@ opt_moves([D0]=Ds, Is0) -> case opt_move(D0, Is0) of not_possible -> {Ds,Is0}; {D1,Is} -> {[D1],Is} - end; -opt_moves([X0,Y0], Is0) -> - {X,Is2} = case opt_move(X0, Is0) of - not_possible -> {X0,Is0}; - {Y0,_} -> {X0,Is0}; - {_X1,_Is1} = XIs1 -> XIs1 - end, - case opt_move(Y0, Is2) of - not_possible -> {[X,Y0],Is2}; - {X,_} -> {[X,Y0],Is2}; - {Y,Is} -> {[X,Y],Is} end. %% opt_move(Dest, [Instruction]) -> {UpdatedDest,[Instruction]} | not_possible @@ -269,7 +302,7 @@ opt_move(Dest, Is) -> opt_move_1(R, [{set,[D],[R],move}|Is0], Acc) -> %% Provided that the source register is killed by instructions %% that follow, the optimization is safe. - case eliminate_use_of_from_reg(Is0, R, D, []) of + case eliminate_use_of_from_reg(Is0, R, D) of {yes,Is} -> opt_move_rev(D, Acc, Is); no -> not_possible end; @@ -327,7 +360,7 @@ opt_tuple_element_1([{set,_,_,{alloc,_,_}}|_], _, _, _) -> opt_tuple_element_1([{set,_,_,{try_catch,_,_}}|_], _, _, _) -> no; opt_tuple_element_1([{set,[D],[S],move}|Is0], I0, {_,S}, Acc) -> - case eliminate_use_of_from_reg(Is0, S, D, []) of + case eliminate_use_of_from_reg(Is0, S, D) of no -> no; {yes,Is} -> @@ -369,6 +402,14 @@ is_killed_or_used(R, {set,Ss,Ds,_}) -> %% that FromRegister is still used and that the optimization is not %% possible. +eliminate_use_of_from_reg(Is, From, To) -> + try + eliminate_use_of_from_reg(Is, From, To, []) + catch + throw:not_possible -> + no + end. + eliminate_use_of_from_reg([{set,_,_,{alloc,Live,_}}|_]=Is0, {x,X}, _, Acc) -> if X < Live -> @@ -377,21 +418,32 @@ eliminate_use_of_from_reg([{set,_,_,{alloc,Live,_}}|_]=Is0, {x,X}, _, Acc) -> {yes,reverse(Acc, Is0)} end; eliminate_use_of_from_reg([{set,Ds,Ss0,Op}=I0|Is], From, To, Acc) -> + ensure_safe_tuple(I0, To), I = case member(From, Ss0) of - true -> - Ss = [case S of - From -> To; - _ -> S - end || S <- Ss0], - {set,Ds,Ss,Op}; - false -> - I0 - end, + true -> + Ss = [case S of + From -> To; + _ -> S + end || S <- Ss0], + {set,Ds,Ss,Op}; + false -> + I0 + end, case member(From, Ds) of - true -> - {yes,reverse(Acc, [I|Is])}; - false -> - eliminate_use_of_from_reg(Is, From, To, [I|Acc]) + true -> + {yes,reverse(Acc, [I|Is])}; + false -> + case member(To, Ds) of + true -> + case beam_utils:is_killed_block(From, Is) of + true -> + {yes,reverse(Acc, [I|Is])}; + false -> + no + end; + false -> + eliminate_use_of_from_reg(Is, From, To, [I|Acc]) + end end; eliminate_use_of_from_reg([I]=Is, From, _To, Acc) -> case beam_utils:is_killed_block(From, [I]) of @@ -401,31 +453,51 @@ eliminate_use_of_from_reg([I]=Is, From, _To, Acc) -> no end. +ensure_safe_tuple({set,[To],[],{put_tuple,_}}, To) -> + throw(not_possible); +ensure_safe_tuple(_, _) -> ok. + +%% opt_allocs(Instructions) -> Instructions. Optimize allocate +%% instructions inside blocks. If safe, replace an allocate_zero +%% instruction with the slightly cheaper allocate instruction. + +opt_allocs(Is) -> + D = beam_utils:index_labels(Is), + opt_allocs_1(Is, D). + +opt_allocs_1([{block,Bl0}|Is], D) -> + Bl = opt_alloc(Bl0, {D,Is}), + [{block,Bl}|opt_allocs_1(Is, D)]; +opt_allocs_1([I|Is], D) -> + [I|opt_allocs_1(Is, D)]; +opt_allocs_1([], _) -> []. + %% opt_alloc(Instructions) -> Instructions' %% Optimises all allocate instructions. opt_alloc([{set,[],[],{alloc,Live0,Info0}}, - {set,[],[],{alloc,Live,Info}}|Is]) -> + {set,[],[],{alloc,Live,Info}}|Is], D) -> Live = Live0, %Assertion. Alloc = combine_alloc(Info0, Info), I = {set,[],[],{alloc,Live,Alloc}}, - opt_alloc([I|Is]); -opt_alloc([{set,[],[],{alloc,R,{_,Ns,Nh,[]}}}|Is]) -> - [{set,[],[],opt_alloc(Is, Ns, Nh, R)}|Is]; -opt_alloc([I|Is]) -> [I|opt_alloc(Is)]; -opt_alloc([]) -> []. + opt_alloc([I|Is], D); +opt_alloc([{set,[],[],{alloc,R,{_,Ns,Nh,[]}}}|Is], D) -> + [{set,[],[],opt_alloc(Is, D, Ns, Nh, R)}|Is]; +opt_alloc([I|Is], D) -> [I|opt_alloc(Is, D)]; +opt_alloc([], _) -> []. combine_alloc({_,Ns,Nh1,Init}, {_,nostack,Nh2,[]}) -> {zero,Ns,beam_utils:combine_heap_needs(Nh1, Nh2),Init}. - + %% opt_alloc(Instructions, FrameSize, HeapNeed, LivingRegs) -> [Instr] %% Generates the optimal sequence of instructions for %% allocating and initalizing the stack frame and needed heap. -opt_alloc(_Is, nostack, Nh, LivingRegs) -> +opt_alloc(_Is, _D, nostack, Nh, LivingRegs) -> {alloc,LivingRegs,{nozero,nostack,Nh,[]}}; -opt_alloc(Is, Ns, Nh, LivingRegs) -> - InitRegs = init_yreg(Is, 0), +opt_alloc(Bl, {D,OuterIs}, Ns, Nh, LivingRegs) -> + Is = [{block,Bl}|OuterIs], + InitRegs = init_yregs(Ns, Is, D), case count_ones(InitRegs) of N when N*2 > Ns -> {alloc,LivingRegs,{nozero,Ns,Nh,gen_init(Ns, InitRegs)}}; @@ -441,19 +513,14 @@ gen_init(Fs, Regs, Y, Acc) when Regs band 1 =:= 0 -> gen_init(Fs, Regs, Y, Acc) -> gen_init(Fs, Regs bsr 1, Y+1, Acc). -%% init_yreg(Instructions, RegSet) -> RegSetInitialized -%% Calculate the set of initialized y registers. - -init_yreg([{set,_,_,{bif,_,_}}|_], Reg) -> Reg; -init_yreg([{set,_,_,{alloc,_,{gc_bif,_,_}}}|_], Reg) -> Reg; -init_yreg([{set,_,_,{alloc,_,{put_map,_,_}}}|_], Reg) -> Reg; -init_yreg([{set,Ds,_,_}|Is], Reg) -> init_yreg(Is, add_yregs(Ds, Reg)); -init_yreg(_Is, Reg) -> Reg. - -add_yregs(Ys, Reg) -> foldl(fun(Y, R0) -> add_yreg(Y, R0) end, Reg, Ys). - -add_yreg({y,Y}, Reg) -> Reg bor (1 bsl Y); -add_yreg(_, Reg) -> Reg. +init_yregs(Y, Is, D) when Y >= 0 -> + case beam_utils:is_killed({y,Y}, Is, D) of + true -> + (1 bsl Y) bor init_yregs(Y-1, Is, D); + false -> + init_yregs(Y-1, Is, D) + end; +init_yregs(_, _, _) -> 0. count_ones(Bits) -> count_ones(Bits, 0). count_ones(0, Acc) -> Acc; @@ -463,16 +530,34 @@ count_ones(Bits, Acc) -> %% Calculate the new number of live registers when we move an allocate %% instruction upwards, passing a 'set' instruction. -alloc_live_regs({set,Ds,Ss,_}, Regs0) -> +alloc_live_regs({set,Ds,Ss,_}, Is, Regs0) -> Rset = x_live(Ss, x_dead(Ds, (1 bsl Regs0)-1)), - live_regs(0, Rset). + Live = live_regs(0, Rset), + case ensure_contiguous(Rset, Live) of + not_possible -> + %% Liveness information (looking forward in the + %% instruction stream) can't prove that moving this + %% allocation instruction is safe. Now use the annotation + %% of defined registers at the beginning of the current + %% block to see whether moving would be safe. + Def0 = defined_regs(Is, 0), + Def = Def0 band ((1 bsl Live) - 1), + ensure_contiguous(Rset bor Def, Live); + Live -> + %% Safe based on liveness information. + Live + end. live_regs(N, 0) -> N; -live_regs(N, Regs) when Regs band 1 =:= 1 -> - live_regs(N+1, Regs bsr 1); -live_regs(_, _) -> - not_possible. +live_regs(N, Regs) -> + live_regs(N+1, Regs bsr 1). + +ensure_contiguous(Regs, Live) -> + case (1 bsl Live) - 1 of + Regs -> Live; + _ -> not_possible + end. x_dead([{x,N}|Rs], Regs) -> x_dead(Rs, Regs band (bnot (1 bsl N))); x_dead([_|Rs], Regs) -> x_dead(Rs, Regs); @@ -481,3 +566,120 @@ x_dead([], Regs) -> Regs. x_live([{x,N}|Rs], Regs) -> x_live(Rs, Regs bor (1 bsl N)); x_live([_|Rs], Regs) -> x_live(Rs, Regs); x_live([], Regs) -> Regs. + +%% defined_regs(ReversedInstructions) -> RegBitmap. +%% Given a reversed instruction stream, determine the +%% the registers that are defined. + +defined_regs([{'%anno',{def,Def}}|_], Regs) -> + Def bor Regs; +defined_regs([{set,Ds,_,{alloc,Live,_}}|_], Regs) -> + x_live(Ds, Regs bor ((1 bsl Live) - 1)); +defined_regs([{set,Ds,_,_}|Is], Regs) -> + defined_regs(Is, x_live(Ds, Regs)). + +%%% +%%% Do local common sub expression elimination (CSE) in each block. +%%% + +local_cse([{block,Bl0}|Is]) -> + Bl = cse_block(Bl0, orddict:new(), []), + [{block,Bl}|local_cse(Is)]; +local_cse([I|Is]) -> + [I|local_cse(Is)]; +local_cse([]) -> []. + +cse_block([I|Is], Es0, Acc0) -> + Es1 = cse_clear(I, Es0), + case cse_expr(I) of + none -> + %% Instruction is not suitable for CSE. + cse_block(Is, Es1, [I|Acc0]); + {ok,D,Expr} -> + %% Suitable instruction. First update the dictionary of + %% suitable expressions for the next iteration. + Es = cse_add(D, Expr, Es1), + + %% Search for a previous identical expression. + case cse_find(Expr, Es0) of + error -> + %% Nothing found + cse_block(Is, Es, [I|Acc0]); + Src -> + %% Use the previously calculated result. + %% Also eliminate any line instruction. + Move = {set,[D],[Src],move}, + case Acc0 of + [{set,_,_,{line,_}}|Acc] -> + cse_block(Is, Es, [Move|Acc]); + [_|_] -> + cse_block(Is, Es, [Move|Acc0]) + end + end + end; +cse_block([], _, Acc) -> + reverse(Acc). + +%% cse_find(Expr, Expressions) -> error | Register. +%% Find a previously evaluated expression whose result can be reused, +%% or return 'error' if no such expression is found. + +cse_find(Expr, Es) -> + case orddict:find(Expr, Es) of + {ok,{Src,_}} -> Src; + error -> error + end. + +cse_expr({set,[D],Ss,{bif,N,_}}) -> + case D of + {fr,_} -> + %% There are too many things that can go wrong. + none; + _ -> + {ok,D,{{bif,N},Ss}} + end; +cse_expr({set,[D],Ss,{alloc,_,{gc_bif,N,_}}}) -> + {ok,D,{{gc_bif,N},Ss}}; +cse_expr({set,[D],Ss,put_list}) -> + {ok,D,{put_list,Ss}}; +cse_expr(_) -> none. + +%% cse_clear(Instr, Expressions0) -> Expressions. +%% Remove all previous expressions that will become +%% invalid when this instruction is executed. Basically, +%% an expression is no longer safe to reuse when the +%% register it has been stored to has been modified, killed, +%% or if any of the source operands have changed. + +cse_clear({set,Ds,_,{alloc,Live,_}}, Es) -> + cse_clear_1(Es, Live, Ds); +cse_clear({set,Ds,_,_}, Es) -> + cse_clear_1(Es, all, Ds). + +cse_clear_1(Es, Live, Ds0) -> + Ds = ordsets:from_list(Ds0), + [E || E <- Es, cse_is_safe(E, Live, Ds)]. + +cse_is_safe({_,{Dst,Interfering}}, Live, Ds) -> + ordsets:is_disjoint(Interfering, Ds) andalso + case Dst of + {x,X} -> + X < Live; + _ -> + true + end. + +%% cse_add(Dest, Expr, Expressions0) -> Expressions. +%% Provided that it is safe, add a new expression to the dictionary +%% of already evaluated expressions. + +cse_add(D, {_,Ss}=Expr, Es) -> + case member(D, Ss) of + false -> + Interfering = ordsets:from_list([D|Ss]), + orddict:store(Expr, {D,Interfering}, Es); + true -> + %% Unsafe because the instruction overwrites one of + %% source operands. + Es + end. diff --git a/lib/compiler/src/beam_bs.erl b/lib/compiler/src/beam_bs.erl index beb055b23d..14cedbb582 100644 --- a/lib/compiler/src/beam_bs.erl +++ b/lib/compiler/src/beam_bs.erl @@ -38,8 +38,7 @@ function({function,Name,Arity,CLabel,Is0}, Lc0) -> {Is,Lc} = bsm_opt(Is1, Lc0), {{function,Name,Arity,CLabel,Is},Lc} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. diff --git a/lib/compiler/src/beam_bsm.erl b/lib/compiler/src/beam_bsm.erl index 9a4e7fb133..9f3b9d788f 100644 --- a/lib/compiler/src/beam_bsm.erl +++ b/lib/compiler/src/beam_bsm.erl @@ -105,8 +105,7 @@ function({function,Name,Arity,Entry,Is}, FIndex) -> D = #btb{f=FIndex,index=Index}, {function,Name,Arity,Entry,btb_opt_1(Is, D, [])} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -125,20 +124,21 @@ btb_opt_1([{test,bs_get_binary2,F,_,[Reg,{atom,all},U,Fs],Reg}=I0|Is], D, Acc0) end, btb_opt_1(Is, D, Acc) end; -btb_opt_1([{test,bs_get_binary2,F,_,[Ctx,{atom,all},U,Fs],Dst}=I0|Is], D, Acc0) -> - case btb_reaches_match(Is, [Ctx,Dst], D) of +btb_opt_1([{test,bs_get_binary2,F,_,[Ctx,{atom,all},U,Fs],Dst}=I0|Is0], D, Acc0) -> + case btb_reaches_match(Is0, [Ctx,Dst], D) of {error,Reason} -> Comment = btb_comment_no_opt(Reason, Fs), - btb_opt_1(Is, D, [Comment,I0|Acc0]); + btb_opt_1(Is0, D, [Comment,I0|Acc0]); {ok,MustSave} when U =:= 1 -> Comment = btb_comment_opt(Fs), - Acc1 = btb_gen_save(MustSave, Ctx, [Comment|Acc0]), - Acc = [{move,Ctx,Dst}|Acc1], + Acc = btb_gen_save(MustSave, Ctx, [Comment|Acc0]), + Is = prepend_move(Ctx, Dst, Is0), btb_opt_1(Is, D, Acc); {ok,MustSave} -> Comment = btb_comment_opt(Fs), Acc1 = btb_gen_save(MustSave, Ctx, [Comment|Acc0]), - Acc = [{move,Ctx,Dst},{test,bs_test_unit,F,[Ctx,U]}|Acc1], + Acc = [{test,bs_test_unit,F,[Ctx,U]}|Acc1], + Is = prepend_move(Ctx, Dst, Is0), btb_opt_1(Is, D, Acc) end; btb_opt_1([I|Is], D, Acc) -> @@ -151,6 +151,12 @@ btb_gen_save(true, Reg, Acc) -> [{bs_save2,Reg,{atom,start}}|Acc]; btb_gen_save(false, _, Acc) -> Acc. +prepend_move(Ctx, Dst, [{block,Bl0}|Is]) -> + Bl = [{set,[Dst],[Ctx],move}|Bl0], + [{block,Bl}|Is]; +prepend_move(Ctx, Dst, Is) -> + [{move,Ctx,Dst}|Is]. + %% btb_reaches_match([Instruction], [Register], D) -> %% {ok,MustSave}|{error,Reason} %% diff --git a/lib/compiler/src/beam_clean.erl b/lib/compiler/src/beam_clean.erl index b736d39f9c..7ddf9fa2e2 100644 --- a/lib/compiler/src/beam_clean.erl +++ b/lib/compiler/src/beam_clean.erl @@ -24,7 +24,7 @@ -export([module/2]). -export([bs_clean_saves/1]). -export([clean_labels/1]). --import(lists, [map/2,foldl/3,reverse/1,filter/2]). +-import(lists, [foldl/3,reverse/1]). -spec module(beam_utils:module_code(), [compile:option()]) -> {'ok',beam_utils:module_code()}. @@ -118,7 +118,7 @@ add_to_work_list(F, {Fs,Used}=Sets) -> clean_labels(Fs0) -> St0 = #st{lmap=[],entry=1,lc=1}, {Fs1,#st{lmap=Lmap0,lc=Lc}} = function_renumber(Fs0, St0, []), - Lmap = gb_trees:from_orddict(ordsets:from_list(Lmap0)), + Lmap = maps:from_list(Lmap0), Fs = function_replace(Fs1, Lmap, []), {Fs,Lc}. @@ -187,7 +187,8 @@ is_record_tuple(_, _, _) -> no. function_replace([{function,Name,Arity,Entry,Asm0}|Fs], Dict, Acc) -> Asm = try - replace(Asm0, [], Dict) + Fb = fun(Old) -> throw({error,{undefined_label,Old}}) end, + beam_utils:replace_labels(Asm0, [], Dict, Fb) catch throw:{error,{undefined_label,Lbl}=Reason} -> io:format("Function ~s/~w refers to undefined label ~w\n", @@ -197,57 +198,6 @@ function_replace([{function,Name,Arity,Entry,Asm0}|Fs], Dict, Acc) -> function_replace(Fs, Dict, [{function,Name,Arity,Entry,Asm}|Acc]); function_replace([], _, Acc) -> Acc. -replace([{test,Test,{f,Lbl},Ops}|Is], Acc, D) -> - replace(Is, [{test,Test,{f,label(Lbl, D)},Ops}|Acc], D); -replace([{test,Test,{f,Lbl},Live,Ops,Dst}|Is], Acc, D) -> - replace(Is, [{test,Test,{f,label(Lbl, D)},Live,Ops,Dst}|Acc], D); -replace([{select,I,R,{f,Fail0},Vls0}|Is], Acc, D) -> - Vls = map(fun ({f,L}) -> {f,label(L, D)}; - (Other) -> Other - end, Vls0), - Fail = label(Fail0, D), - replace(Is, [{select,I,R,{f,Fail},Vls}|Acc], D); -replace([{'try',R,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{'try',R,{f,label(Lbl, D)}}|Acc], D); -replace([{'catch',R,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{'catch',R,{f,label(Lbl, D)}}|Acc], D); -replace([{jump,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{jump,{f,label(Lbl, D)}}|Acc], D); -replace([{loop_rec,{f,Lbl},R}|Is], Acc, D) -> - replace(Is, [{loop_rec,{f,label(Lbl, D)},R}|Acc], D); -replace([{loop_rec_end,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{loop_rec_end,{f,label(Lbl, D)}}|Acc], D); -replace([{wait,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{wait,{f,label(Lbl, D)}}|Acc], D); -replace([{wait_timeout,{f,Lbl},To}|Is], Acc, D) -> - replace(Is, [{wait_timeout,{f,label(Lbl, D)},To}|Acc], D); -replace([{bif,Name,{f,Lbl},As,R}|Is], Acc, D) when Lbl =/= 0 -> - replace(Is, [{bif,Name,{f,label(Lbl, D)},As,R}|Acc], D); -replace([{gc_bif,Name,{f,Lbl},Live,As,R}|Is], Acc, D) when Lbl =/= 0 -> - replace(Is, [{gc_bif,Name,{f,label(Lbl, D)},Live,As,R}|Acc], D); -replace([{call,Ar,{f,Lbl}}|Is], Acc, D) -> - replace(Is, [{call,Ar,{f,label(Lbl,D)}}|Acc], D); -replace([{make_fun2,{f,Lbl},U1,U2,U3}|Is], Acc, D) -> - replace(Is, [{make_fun2,{f,label(Lbl, D)},U1,U2,U3}|Acc], D); -replace([{bs_init,{f,Lbl},Info,Live,Ss,Dst}|Is], Acc, D) when Lbl =/= 0 -> - replace(Is, [{bs_init,{f,label(Lbl, D)},Info,Live,Ss,Dst}|Acc], D); -replace([{bs_put,{f,Lbl},Info,Ss}|Is], Acc, D) when Lbl =/= 0 -> - replace(Is, [{bs_put,{f,label(Lbl, D)},Info,Ss}|Acc], D); -replace([{put_map=I,{f,Lbl},Op,Src,Dst,Live,List}|Is], Acc, D) - when Lbl =/= 0 -> - replace(Is, [{I,{f,label(Lbl, D)},Op,Src,Dst,Live,List}|Acc], D); -replace([{get_map_elements=I,{f,Lbl},Src,List}|Is], Acc, D) when Lbl =/= 0 -> - replace(Is, [{I,{f,label(Lbl, D)},Src,List}|Acc], D); -replace([I|Is], Acc, D) -> - replace(Is, [I|Acc], D); -replace([], Acc, _) -> Acc. - -label(Old, D) -> - case gb_trees:lookup(Old, D) of - {value,Val} -> Val; - none -> throw({error,{undefined_label,Old}}) - end. - %%% %%% Final fixup of bs_start_match2/5,bs_save2/bs_restore2 instructions for %%% new bit syntax matching (introduced in R11B). @@ -353,8 +303,21 @@ maybe_remove_lines(Fs, Opts) -> end. remove_lines([{function,N,A,Lbl,Is0}|T]) -> - Is = filter(fun({line,_}) -> false; - (_) -> true - end, Is0), + Is = remove_lines_fun(Is0), [{function,N,A,Lbl,Is}|remove_lines(T)]; remove_lines([]) -> []. + +remove_lines_fun([{line,_}|Is]) -> + remove_lines_fun(Is); +remove_lines_fun([{block,Bl0}|Is]) -> + Bl = remove_lines_block(Bl0), + [{block,Bl}|remove_lines_fun(Is)]; +remove_lines_fun([I|Is]) -> + [I|remove_lines_fun(Is)]; +remove_lines_fun([]) -> []. + +remove_lines_block([{set,_,_,{line,_}}|Is]) -> + remove_lines_block(Is); +remove_lines_block([I|Is]) -> + [I|remove_lines_block(Is)]; +remove_lines_block([]) -> []. diff --git a/lib/compiler/src/beam_dead.erl b/lib/compiler/src/beam_dead.erl index d379fdc4eb..dbbaae05eb 100644 --- a/lib/compiler/src/beam_dead.erl +++ b/lib/compiler/src/beam_dead.erl @@ -56,8 +56,7 @@ function({function,Name,Arity,CLabel,Is0}, Lc0) -> Is = move_move_into_block(Is3, []), {{function,Name,Arity,CLabel,Is},Lc} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -272,7 +271,8 @@ backward([{jump,{f,To0}},{move,Src,Reg}=Move|Is], D, Acc) -> end; backward([{jump,{f,To}}=J|[{bif,Op,{f,BifFail},Ops,Reg}|Is]=Is0], D, Acc) -> try replace_comp_op(To, Reg, Op, Ops, D) of - I -> backward(Is, D, I++Acc) + {Test,Jump} -> + backward([Jump,Test|Is], D, Acc) catch throw:not_possible -> case To =:= BifFail of @@ -294,24 +294,25 @@ backward([{jump,{f,To}}=J|[{gc_bif,_,{f,To},_,_,_Dst}|Is]], D, Acc) -> %% register is initialized, and it is therefore no need to test %% for liveness of the destination register at label To. backward([J|Is], D, Acc); -backward([{test,bs_start_match2,F,Live,[R,_]=Args,Ctxt}|Is], D, - [{test,bs_match_string,F,[Ctxt,Bs]}, - {test,bs_test_tail2,F,[Ctxt,0]}|Acc0]=Acc) -> +backward([{test,bs_start_match2,F,Live,[Src,_]=Args,Ctxt}|Is], D, Acc0) -> {f,To0} = F, - case beam_utils:is_killed(Ctxt, Acc0, D) of - true -> - To = shortcut_bs_context_to_binary(To0, R, D), - Eq = {test,is_eq_exact,{f,To},[R,{literal,Bs}]}, - backward(Is, D, [Eq|Acc0]); - false -> - To = shortcut_bs_start_match(To0, R, D), - I = {test,bs_start_match2,{f,To},Live,Args,Ctxt}, - backward(Is, D, [I|Acc]) + case test_bs_literal(F, Ctxt, D, Acc0) of + {none,Acc} -> + %% Ctxt killed immediately after bs_start_match2. + To = shortcut_bs_context_to_binary(To0, Src, D), + I = {test,is_bitstr,{f,To},[Src]}, + backward(Is, D, [I|Acc]); + {Literal,Acc} -> + %% Ctxt killed after matching a literal. + To = shortcut_bs_context_to_binary(To0, Src, D), + Eq = {test,is_eq_exact,{f,To},[Src,{literal,Literal}]}, + backward(Is, D, [Eq|Acc]); + not_killed -> + %% Ctxt not killed. Not much to do. + To = shortcut_bs_start_match(To0, Src, D), + I = {test,bs_start_match2,{f,To},Live,Args,Ctxt}, + backward(Is, D, [I|Acc0]) end; -backward([{test,bs_start_match2,{f,To0},Live,[Src|_]=Info,Dst}|Is], D, Acc) -> - To = shortcut_bs_start_match(To0, Src, D), - I = {test,bs_start_match2,{f,To},Live,Info,Dst}, - backward(Is, D, [I|Acc]); backward([{test,Op,{f,To0},Ops0}|Is], D, Acc) -> To1 = shortcut_bs_test(To0, Is, D), To2 = shortcut_label(To1, D), @@ -446,7 +447,7 @@ prune_redundant([], _) -> []. replace_comp_op(To, Reg, Op, Ops, D) -> False = comp_op_find_shortcut(To, Reg, {atom,false}, D), True = comp_op_find_shortcut(To, Reg, {atom,true}, D), - [bif_to_test(Op, Ops, False),{jump,{f,True}}]. + {bif_to_test(Op, Ops, False),{jump,{f,True}}}. comp_op_find_shortcut(To0, Reg, Val, D) -> case shortcut_select_label(To0, Reg, Val, D) of @@ -483,15 +484,22 @@ not_possible() -> throw(not_possible). %% F1: is_eq_exact F2 Reg Lit2 F1: is_eq_exact F2 Reg Lit2 %% L2: .... L2: %% -combine_eqs(To, [Reg,{Type,_}=Lit1]=Ops, D, [{label,L1}|_]) - when Type =:= atom; Type =:= integer -> +combine_eqs(To, [Reg,{Type,_}=Lit1]=Ops, D, Acc) + when Type =:= atom; Type =:= integer -> + Next = case Acc of + [{label,Lbl}|_] -> Lbl; + [{jump,{f,Lbl}}|_] -> Lbl + end, case beam_utils:code_at(To, D) of [{test,is_eq_exact,{f,F2},[Reg,{Type,_}=Lit2]}, {label,L2}|_] when Lit1 =/= Lit2 -> - {select,select_val,Reg,{f,F2},[Lit1,{f,L1},Lit2,{f,L2}]}; + {select,select_val,Reg,{f,F2},[Lit1,{f,Next},Lit2,{f,L2}]}; + [{test,is_eq_exact,{f,F2},[Reg,{Type,_}=Lit2]}, + {jump,{f,L2}}|_] when Lit1 =/= Lit2 -> + {select,select_val,Reg,{f,F2},[Lit1,{f,Next},Lit2,{f,L2}]}; [{select,select_val,Reg,{f,F2},[{Type,_}|_]=List0}|_] -> List = remove_from_list(Lit1, List0), - {select,select_val,Reg,{f,F2},[Lit1,{f,L1}|List]}; + {select,select_val,Reg,{f,F2},[Lit1,{f,Next}|List]}; _Is -> {test,is_eq_exact,{f,To},Ops} end; @@ -504,6 +512,22 @@ remove_from_list(Lit, [Val,{f,_}=Fail|T]) -> [Val,Fail|remove_from_list(Lit, T)]; remove_from_list(_, []) -> []. + +test_bs_literal(F, Ctxt, D, + [{test,bs_match_string,F,[Ctxt,Bs]}, + {test,bs_test_tail2,F,[Ctxt,0]}|Acc]) -> + test_bs_literal_1(Ctxt, Acc, D, Bs); +test_bs_literal(F, Ctxt, D, [{test,bs_test_tail2,F,[Ctxt,0]}|Acc]) -> + test_bs_literal_1(Ctxt, Acc, D, <<>>); +test_bs_literal(_, Ctxt, D, Acc) -> + test_bs_literal_1(Ctxt, Acc, D, none). + +test_bs_literal_1(Ctxt, Is, D, Literal) -> + case beam_utils:is_killed(Ctxt, Is, D) of + true -> {Literal,Is}; + false -> not_killed + end. + %% shortcut_bs_test(TargetLabel, ReversedInstructions, D) -> TargetLabel' %% Try to shortcut the failure label for bit syntax matching. diff --git a/lib/compiler/src/beam_disasm.erl b/lib/compiler/src/beam_disasm.erl index 8fd0b36d05..a68c4b5367 100644 --- a/lib/compiler/src/beam_disasm.erl +++ b/lib/compiler/src/beam_disasm.erl @@ -163,8 +163,8 @@ pp_instr(I) -> file(File) -> try process_chunks(File) - catch error:Reason -> - {error,?MODULE,{internal,{Reason,erlang:get_stacktrace()}}} + catch error:Reason:Stack -> + {error,?MODULE,{internal,{Reason,Stack}}} end. %%----------------------------------------------------------------------- @@ -719,42 +719,6 @@ resolve_inst({wait,[Lbl]},_,_,_) -> {wait,Lbl}; resolve_inst({wait_timeout,[Lbl,Int]},_,_,_) -> {wait_timeout,Lbl,resolve_arg(Int)}; -resolve_inst({m_plus,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'+',W,[SrcR1,SrcR2],DstR}; -resolve_inst({m_minus,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'-',W,[SrcR1,SrcR2],DstR}; -resolve_inst({m_times,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'*',W,[SrcR1,SrcR2],DstR}; -resolve_inst({m_div,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'/',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_div,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'div',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_rem,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'rem',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_band,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'band',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_bor,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'bor',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_bxor,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'bxor',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_bsl,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'bsl',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_bsr,Args},_,_,_) -> - [W,SrcR1,SrcR2,DstR] = resolve_args(Args), - {arithbif,'bsr',W,[SrcR1,SrcR2],DstR}; -resolve_inst({int_bnot,Args},_,_,_) -> - [W,SrcR,DstR] = resolve_args(Args), - {arithbif,'bnot',W,[SrcR],DstR}; resolve_inst({is_lt=I,Args0},_,_,_) -> [L|Args] = resolve_args(Args0), {test,I,L,Args}; @@ -800,9 +764,6 @@ resolve_inst({is_nil=I,Args0},_,_,_) -> resolve_inst({is_binary=I,Args0},_,_,_) -> [L|Args] = resolve_args(Args0), {test,I,L,Args}; -resolve_inst({is_constant=I,Args0},_,_,_) -> - [L|Args] = resolve_args(Args0), - {test,I,L,Args}; resolve_inst({is_list=I,Args0},_,_,_) -> [L|Args] = resolve_args(Args0), {test,I,L,Args}; @@ -840,11 +801,6 @@ resolve_inst({get_tuple_element,[Src,{u,Off},Dst]},_,_,_) -> {get_tuple_element,resolve_arg(Src),Off,resolve_arg(Dst)}; resolve_inst({set_tuple_element,[Src,Dst,{u,Off}]},_,_,_) -> {set_tuple_element,resolve_arg(Src),resolve_arg(Dst),Off}; -resolve_inst({put_string,[{u,Len},{u,Off},Dst]},_,Strings,_) -> - String = if Len > 0 -> binary_to_list(Strings, Off+1, Off+Len); - true -> "" - end, - {put_string,Len,{string,String},Dst}; resolve_inst({put_list,[Src1,Src2,Dst]},_,_,_) -> {put_list,resolve_arg(Src1),resolve_arg(Src2),Dst}; resolve_inst({put_tuple,[{u,Arity},Dst]},_,_,_) -> @@ -859,9 +815,6 @@ resolve_inst({case_end,[X]},_,_,_) -> {case_end,resolve_arg(X)}; resolve_inst({call_fun,[{u,N}]},_,_,_) -> {call_fun,N}; -resolve_inst({make_fun,Args},_,_,Lbls) -> - [{f,L},Magic,FreeVars] = resolve_args(Args), - {make_fun,lookup(L,Lbls),Magic,FreeVars}; resolve_inst({is_function=I,Args0},_,_,_) -> [L|Args] = resolve_args(Args0), {test,I,L,Args}; @@ -870,30 +823,6 @@ resolve_inst({call_ext_only,[{u,N},{u,MFAix}]},Imports,_,_) -> %% %% Instructions for handling binaries added in R7A & R7B %% -resolve_inst({bs_start_match,[F,Reg]},_,_,_) -> - {bs_start_match,F,Reg}; -resolve_inst({bs_get_integer=I,[Lbl,Arg2,{u,N},{u,U},Arg5]},_,_,_) -> - [A2,A5] = resolve_args([Arg2,Arg5]), - {test,I,Lbl,[A2,N,decode_field_flags(U),A5]}; -resolve_inst({bs_get_float=I,[Lbl,Arg2,{u,N},{u,U},Arg5]},_,_,_) -> - [A2,A5] = resolve_args([Arg2,Arg5]), - {test,I,Lbl,[A2,N,decode_field_flags(U),A5]}; -resolve_inst({bs_get_binary=I,[Lbl,Arg2,{u,N},{u,U},Arg5]},_,_,_) -> - [A2,A5] = resolve_args([Arg2,Arg5]), - {test,I,Lbl,[A2,N,decode_field_flags(U),A5]}; -resolve_inst({bs_skip_bits,[Lbl,Arg2,{u,N},{u,U}]},_,_,_) -> - A2 = resolve_arg(Arg2), - {test,bs_skip_bits,Lbl,[A2,N,decode_field_flags(U)]}; -resolve_inst({bs_test_tail,[F,{u,N}]},_,_,_) -> - {test,bs_test_tail,F,[N]}; -resolve_inst({bs_save,[{u,N}]},_,_,_) -> - {bs_save,N}; -resolve_inst({bs_restore,[{u,N}]},_,_,_) -> - {bs_restore,N}; -resolve_inst({bs_init,[{u,N},{u,U}]},_,_,_) -> - {bs_init,N,decode_field_flags(U)}; -resolve_inst({bs_final,[F,X]},_,_,_) -> - {bs_final,F,X}; resolve_inst({bs_put_integer,[Lbl,Arg2,{u,N},{u,U},Arg5]},_,_,_) -> [A2,A5] = resolve_args([Arg2,Arg5]), {bs_put_integer,Lbl,A2,N,decode_field_flags(U),A5}; @@ -908,8 +837,6 @@ resolve_inst({bs_put_string,[{u,Len},{u,Off}]},_,Strings,_) -> true -> "" end, {bs_put_string,Len,{string,String}}; -resolve_inst({bs_need_buf,[{u,N}]},_,_,_) -> - {bs_need_buf,N}; %% %% Instructions for handling floating point numbers added in June 2001 (R8). @@ -961,9 +888,6 @@ resolve_inst({raise,[_Reg1,_Reg2]=Regs},_,_,_) -> resolve_inst({bs_init2,[Lbl,Arg2,{u,W},{u,R},{u,F},Arg6]},_,_,_) -> [A2,A6] = resolve_args([Arg2,Arg6]), {bs_init2,Lbl,A2,W,R,decode_field_flags(F),A6}; -resolve_inst({bs_bits_to_bytes,[Lbl,Arg2,Arg3]},_,_,_) -> - [A2,A3] = resolve_args([Arg2,Arg3]), - {bs_bits_to_bytes,Lbl,A2,A3}; resolve_inst({bs_add=I,[Lbl,Arg2,Arg3,Arg4,Arg5]},_,_,_) -> [A2,A3,A4,A5] = resolve_args([Arg2,Arg3,Arg4,Arg5]), {I,Lbl,[A2,A3,A4],A5}; @@ -1041,12 +965,6 @@ resolve_inst({gc_bif3,Args},Imports,_,_) -> {gc_bif,BifName,F,Live,[A1,A2,A3],Reg}; %% -%% New instructions for creating non-byte aligned binaries. -%% -resolve_inst({bs_final2,[X,Y]},_,_,_) -> - {bs_final2,X,Y}; - -%% %% R11B-5. %% resolve_inst({is_bitstr=I,Args0},_,_,_) -> @@ -1165,6 +1083,19 @@ resolve_inst({get_map_elements,Args0},_,_,_) -> {get_map_elements,FLbl,Src,{list,List}}; %% +%% OTP 21. +%% + +resolve_inst({build_stacktrace,[]},_,_,_) -> + build_stacktrace; +resolve_inst({raw_raise,[]},_,_,_) -> + raw_raise; +resolve_inst({get_hd,[Src,Dst]},_,_,_) -> + {get_hd,Src,Dst}; +resolve_inst({get_tl,[Src,Dst]},_,_,_) -> + {get_tl,Src,Dst}; + +%% %% Catches instructions that are not yet handled. %% resolve_inst(X,_,_,_) -> ?exit({resolve_inst,X}). diff --git a/lib/compiler/src/beam_disasm.hrl b/lib/compiler/src/beam_disasm.hrl index d968cd9587..c3326c15a0 100644 --- a/lib/compiler/src/beam_disasm.hrl +++ b/lib/compiler/src/beam_disasm.hrl @@ -26,7 +26,8 @@ %% IT SHOULD BE MOVED TO A FILE THAT DEFINES (AND EXPORTS) %% PROPER TYPES FOR THE SET OF BEAM INSTRUCTIONS. %% --type beam_instr() :: 'bs_init_writable' | 'fclearerror' | 'if_end' +-type beam_instr() :: 'bs_init_writable' | 'build_stacktrace' + | 'fclearerror' | 'if_end' | 'raw_raise' | 'remove_message' | 'return' | 'send' | 'timeout' | tuple(). %% XXX: Very underspecified - FIX THIS diff --git a/lib/compiler/src/beam_except.erl b/lib/compiler/src/beam_except.erl index 9801c68ee2..abd39c661d 100644 --- a/lib/compiler/src/beam_except.erl +++ b/lib/compiler/src/beam_except.erl @@ -45,8 +45,7 @@ function({function,Name,Arity,CLabel,Is0}) -> Is = function_1(Is0), {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. diff --git a/lib/compiler/src/beam_flatten.erl b/lib/compiler/src/beam_flatten.erl index a4d45a4ca6..c60211f516 100644 --- a/lib/compiler/src/beam_flatten.erl +++ b/lib/compiler/src/beam_flatten.erl @@ -50,6 +50,9 @@ norm_block([{set,[],[],{alloc,R,Alloc}}|Is], Acc0) -> Acc -> norm_block(Is, Acc) end; +norm_block([{set,[D1],[S],get_hd},{set,[D2],[S],get_tl}|Is], Acc) -> + I = {get_list,S,D1,D2}, + norm_block(Is, [I|Acc]); norm_block([I|Is], Acc) -> norm_block(Is, [norm(I)|Acc]); norm_block([], Acc) -> Acc. @@ -64,12 +67,14 @@ norm({set,[D],[],{put_tuple,A}}) -> {put_tuple,A,D}; norm({set,[],[S],put}) -> {put,S}; norm({set,[D],[S],{get_tuple_element,I}}) -> {get_tuple_element,S,I,D}; norm({set,[],[S,D],{set_tuple_element,I}}) -> {set_tuple_element,S,D,I}; -norm({set,[D1,D2],[S],get_list}) -> {get_list,S,D1,D2}; +norm({set,[D],[S],get_hd}) -> {get_hd,S,D}; +norm({set,[D],[S],get_tl}) -> {get_tl,S,D}; norm({set,[D],[S|Puts],{alloc,R,{put_map,Op,F}}}) -> {put_map,F,Op,S,D,R,{list,Puts}}; norm({set,[],[],remove_message}) -> remove_message; norm({set,[],[],fclearerror}) -> fclearerror; -norm({set,[],[],fcheckerror}) -> {fcheckerror,{f,0}}. +norm({set,[],[],fcheckerror}) -> {fcheckerror,{f,0}}; +norm({set,[],[],{line,_}=Line}) -> Line. norm_allocate({_Zero,nostack,Nh,[]}, Regs) -> [{test_heap,Nh,Regs}]; diff --git a/lib/compiler/src/beam_jump.erl b/lib/compiler/src/beam_jump.erl index 4365451356..c33de217bd 100644 --- a/lib/compiler/src/beam_jump.erl +++ b/lib/compiler/src/beam_jump.erl @@ -71,9 +71,9 @@ %%% %%% jump L2 %%% . . . -%%% L1: %%% L2: ... %%% +%%% and all preceding uses of L1 renamed to L2. %%% If the jump is unreachable, it will be removed according to (1). %%% %%% (5) In @@ -156,41 +156,46 @@ function({function,Name,Arity,CLabel,Asm0}) -> %%% share(Is0) -> - %% We will get more sharing if we never fall through to a label. - Is = eliminate_fallthroughs(Is0, []), - share_1(Is, #{}, [], []). + Is1 = eliminate_fallthroughs(Is0, []), + Is2 = find_fixpoint(fun(Is) -> + share_1(Is, #{}, #{}, [], []) + end, Is1), + reverse(Is2). -share_1([{label,L}=Lbl|Is], Dict0, [_|_]=Seq, Acc) -> +share_1([{label,L}=Lbl|Is], Dict0, Lbls0, [_|_]=Seq, Acc) -> case maps:find(Seq, Dict0) of error -> Dict = maps:put(Seq, L, Dict0), - share_1(Is, Dict, [], [Lbl|Seq ++ Acc]); + share_1(Is, Dict, Lbls0, [], [Lbl|Seq ++ Acc]); {ok,Label} -> - share_1(Is, Dict0, [], [Lbl,{jump,{f,Label}}|Acc]) + Lbls = maps:put(L, Label, Lbls0), + share_1(Is, Dict0, Lbls, [], [Lbl,{jump,{f,Label}}|Acc]) end; -share_1([{func_info,_,_,_}=I|Is], _, [], Acc) -> - reverse(Is, [I|Acc]); -share_1([{'catch',_,_}=I|Is], Dict0, Seq, Acc) -> - Dict = clean_non_sharable(Dict0), - share_1(Is, Dict, [I|Seq], Acc); -share_1([{'try',_,_}=I|Is], Dict0, Seq, Acc) -> - Dict = clean_non_sharable(Dict0), - share_1(Is, Dict, [I|Seq], Acc); -share_1([{try_case,_}=I|Is], Dict0, Seq, Acc) -> - Dict = clean_non_sharable(Dict0), - share_1(Is, Dict, [I|Seq], Acc); -share_1([{catch_end,_}=I|Is], Dict0, Seq, Acc) -> - Dict = clean_non_sharable(Dict0), - share_1(Is, Dict, [I|Seq], Acc); -share_1([I|Is], Dict, Seq, Acc) -> +share_1([{func_info,_,_,_}|_]=Is, _, Lbls, [], Acc) when Lbls =/= #{} -> + beam_utils:replace_labels(Acc, Is, Lbls, fun(Old) -> Old end); +share_1([{func_info,_,_,_}|_]=Is, _, Lbls, [], Acc) when Lbls =:= #{} -> + reverse(Acc, Is); +share_1([{'catch',_,_}=I|Is], Dict0, Lbls0, Seq, Acc) -> + {Dict,Lbls} = clean_non_sharable(Dict0, Lbls0), + share_1(Is, Dict, Lbls, [I|Seq], Acc); +share_1([{'try',_,_}=I|Is], Dict0, Lbls0, Seq, Acc) -> + {Dict,Lbls} = clean_non_sharable(Dict0, Lbls0), + share_1(Is, Dict, Lbls, [I|Seq], Acc); +share_1([{try_case,_}=I|Is], Dict0, Lbls0, Seq, Acc) -> + {Dict,Lbls} = clean_non_sharable(Dict0, Lbls0), + share_1(Is, Dict, Lbls, [I|Seq], Acc); +share_1([{catch_end,_}=I|Is], Dict0, Lbls0, Seq, Acc) -> + {Dict,Lbls} = clean_non_sharable(Dict0, Lbls0), + share_1(Is, Dict, Lbls, [I|Seq], Acc); +share_1([I|Is], Dict, Lbls, Seq, Acc) -> case is_unreachable_after(I) of false -> - share_1(Is, Dict, [I|Seq], Acc); + share_1(Is, Dict, Lbls, [I|Seq], Acc); true -> - share_1(Is, Dict, [I], Acc) + share_1(Is, Dict, Lbls, [I], Acc) end. -clean_non_sharable(Dict) -> +clean_non_sharable(Dict0, Lbls0) -> %% We are passing in or out of a 'catch' or 'try' block. Remove %% sequences that should not be shared over the boundaries of the %% block. Since the end of the sequence must match, the only @@ -198,7 +203,17 @@ clean_non_sharable(Dict) -> %% the 'catch'/'try' block is a sequence that ends with an %% instruction that causes an exception. Any sequence that causes %% an exception must contain a line/1 instruction. - maps:filter(fun(K, _V) -> sharable_with_try(K) end, Dict). + Dict1 = maps:to_list(Dict0), + Lbls1 = maps:to_list(Lbls0), + {Dict2,Lbls2} = foldl(fun({K, V}, {Dict,Lbls}) -> + case sharable_with_try(K) of + true -> + {[{K,V}|Dict],lists:keydelete(V, 2, Lbls)}; + false -> + {Dict,Lbls} + end + end, {[],Lbls1}, Dict1), + {maps:from_list(Dict2),maps:from_list(Lbls2)}. sharable_with_try([{line,_}|_]) -> %% This sequence may cause an exception and may potentially @@ -275,14 +290,15 @@ extract_seq_1(_, _) -> no. -record(st, { entry :: beam_asm:label(), %Entry label (must not be moved). - mlbl :: #{beam_asm:label() := [beam_asm:label()]}, %Moved labels. - labels :: cerl_sets:set() %Set of referenced labels. + replace :: #{beam_asm:label() := beam_asm:label()}, %Labels to replace. + labels :: cerl_sets:set(), %Set of referenced labels. + index :: beam_utils:code_index() | {lazy,[beam_utils:instruction()]} %Index built lazily only if needed }). opt(Is0, CLabel) -> find_fixpoint(fun(Is) -> Lbls = initial_labels(Is), - St = #st{entry=CLabel,mlbl=#{},labels=Lbls}, + St = #st{entry=CLabel,replace=#{},labels=Lbls,index={lazy,Is}}, opt(Is, [], St) end, Is0). @@ -292,7 +308,7 @@ find_fixpoint(OptFun, Is0) -> Is -> find_fixpoint(OptFun, Is) end. -opt([{test,_,{f,L}=Lbl,_}=I|[{jump,{f,L}}|_]=Is], Acc, St) -> +opt([{test,_,{f,L}=Lbl,_}=I|[{jump,{f,L}}|_]=Is], Acc0, St0) -> %% We have %% Test Label Ops %% jump Label @@ -301,10 +317,23 @@ opt([{test,_,{f,L}=Lbl,_}=I|[{jump,{f,L}}|_]=Is], Acc, St) -> case beam_utils:is_pure_test(I) of false -> %% Test is not pure; we must keep it. - opt(Is, [I|Acc], label_used(Lbl, St)); + opt(Is, [I|Acc0], label_used(Lbl, St0)); true -> %% The test is pure and its failure label is the same %% as in the jump that follows -- thus it is not needed. + %% Check if any of the previous instructions could also be eliminated. + {Acc,St} = opt_useless_loads(Acc0, L, St0), + opt(Is, Acc, St) + end; +opt([{test,_,{f,L}=Lbl,_}=I|[{label,L}|_]=Is], Acc0, St0) -> + %% Similar to the above, except we have a fall-through rather than jump + %% Test Label Ops + %% label Label + case beam_utils:is_pure_test(I) of + false -> + opt(Is, [I|Acc0], label_used(Lbl, St0)); + true -> + {Acc,St} = opt_useless_loads(Acc0, L, St0), opt(Is, Acc, St) end; opt([{test,Test0,{f,L}=Lbl,Ops}=I|[{jump,To}|Is]=Is0], Acc, St) -> @@ -326,30 +355,16 @@ opt([{test,_,{f,_}=Lbl,_,_,_}=I|Is], Acc, St) -> opt(Is, [I|Acc], label_used(Lbl, St)); opt([{select,_,_R,Fail,Vls}=I|Is], Acc, St) -> skip_unreachable(Is, [I|Acc], label_used([Fail|Vls], St)); -opt([{label,Lbl}=I|Is], Acc, #st{mlbl=Mlbl}=St0) -> - case maps:find(Lbl, Mlbl) of - {ok,Lbls} -> - %% Essential to remove the list of labels from the dictionary, - %% since we will rescan the inserted labels. We MUST rescan. - St = St0#st{mlbl=maps:remove(Lbl, Mlbl)}, - insert_labels([Lbl|Lbls], Is, Acc, St); - error -> - opt(Is, [I|Acc], St0) - end; +opt([{label,From}=I,{label,To}|Is], Acc, #st{replace=Replace}=St) -> + opt([I|Is], Acc, St#st{replace=Replace#{To => From}}); opt([{jump,{f,_}=X}|[{label,_},{jump,X}|_]=Is], Acc, St) -> opt(Is, Acc, St); opt([{jump,{f,Lbl}}|[{label,Lbl}|_]=Is], Acc, St) -> opt(Is, Acc, St); -opt([{jump,{f,L}=Lbl}=I|Is], Acc0, #st{mlbl=Mlbl0}=St0) -> - %% All labels before this jump instruction should now be - %% moved to the location of the jump's target. - {Lbls,Acc} = collect_labels(Acc0, St0), - St = case Lbls of - [] -> St0; - [_|_] -> - Mlbl = maps_append_list(L, Lbls, Mlbl0), - St0#st{mlbl=Mlbl} - end, +opt([{jump,{f,L}=Lbl}=I|Is], Acc0, St0) -> + %% Replace all labels before this jump instruction into the + %% location of the jump's target. + {Acc,St} = collect_labels(Acc0, L, St0), skip_unreachable(Is, [I|Acc], label_used(Lbl, St)); %% Optimization: quickly handle some common instructions that don't %% have any failure labels and where is_unreachable_after(I) =:= false. @@ -369,36 +384,72 @@ opt([I|Is], Acc, #st{labels=Used0}=St0) -> true -> skip_unreachable(Is, [I|Acc], St); false -> opt(Is, [I|Acc], St) end; -opt([], Acc, #st{mlbl=Mlbl}) -> - Code = reverse(Acc), - insert_fc_labels(Code, Mlbl). - -insert_fc_labels([{label,L}=I|Is0], Mlbl) -> - case maps:find(L, Mlbl) of - error -> - [I|insert_fc_labels(Is0, Mlbl)]; - {ok,Lbls} -> - Is = [{label,Lb} || Lb <- Lbls] ++ Is0, - [I|insert_fc_labels(Is, maps:remove(L, Mlbl))] +opt([], Acc, #st{replace=Replace0}) when Replace0 =/= #{} -> + Replace = normalize_replace(maps:to_list(Replace0), Replace0, []), + beam_utils:replace_labels(Acc, [], Replace, fun(Old) -> Old end); +opt([], Acc, #st{replace=Replace}) when Replace =:= #{} -> + reverse(Acc). + +normalize_replace([{From,To0}|Rest], Replace, Acc) -> + case Replace of + #{To0 := To} -> + normalize_replace([{From,To}|Rest], Replace, Acc); + _ -> + normalize_replace(Rest, Replace, [{From,To0}|Acc]) end; -insert_fc_labels([_|_]=Is, _) -> Is. - -maps_append_list(K,Vs,M) -> - case M of - #{K:=Vs0} -> M#{K:=Vs0++Vs}; % same order as dict - _ -> M#{K => Vs} - end. +normalize_replace([], _Replace, Acc) -> + maps:from_list(Acc). + +%% After eliminating a test, it might happen, that a register was only used +%% in this test. Let's check if that was the case and if it was so, we can +%% eliminate the load into the register completely. +opt_useless_loads([{block,_}|_]=Is, L, #st{index={lazy,FIs}}=St) -> + opt_useless_loads(Is, L, St#st{index=beam_utils:index_labels(FIs)}); +opt_useless_loads([{block,Block0}|Is], L, #st{index=Index}=St) -> + case opt_useless_block_loads(Block0, L, Index) of + [] -> + opt_useless_loads(Is, L, St); + [_|_]=Block -> + {[{block,Block}|Is],St} + end; +%% After eliminating the test and useless blocks, it might happen, +%% that the previous test could also be eliminated. +%% It might be that the label was already marked as used, even if ultimately, +%% it never will be - we can't do much about it at that point, though +opt_useless_loads([{test,_,{f,L},_}=I|Is], L, St) -> + case beam_utils:is_pure_test(I) of + false -> + {[I|Is],St}; + true -> + opt_useless_loads(Is, L, St) + end; +opt_useless_loads(Is, _L, St) -> + {Is,St}. + +opt_useless_block_loads([{set,[Dst],_,_}=I|Is], L, Index) -> + BlockJump = [{block,Is},{jump,{f,L}}], + case beam_utils:is_killed(Dst, BlockJump, Index) of + true -> + %% The register is killed and not used, we can remove the load + opt_useless_block_loads(Is, L, Index); + false -> + [I|opt_useless_block_loads(Is, L, Index)] + end; +opt_useless_block_loads([I|Is], L, Index) -> + [I|opt_useless_block_loads(Is, L, Index)]; +opt_useless_block_loads([], _L, _Index) -> + []. -collect_labels(Is, #st{entry=Entry}) -> - collect_labels_1(Is, Entry, []). +collect_labels(Is, Label, #st{entry=Entry,replace=Replace} = St) -> + collect_labels_1(Is, Label, Entry, Replace, St). -collect_labels_1([{label,Entry}|_]=Is, Entry, Acc) -> +collect_labels_1([{label,Entry}|_]=Is, _Label, Entry, Acc, St) -> %% Never move the entry label. - {Acc,Is}; -collect_labels_1([{label,L}|Is], Entry, Acc) -> - collect_labels_1(Is, Entry, [L|Acc]); -collect_labels_1(Is, _Entry, Acc) -> - {Acc,Is}. + {Is,St#st{replace=Acc}}; +collect_labels_1([{label,L}|Is], Label, Entry, Acc, St) -> + collect_labels_1(Is, Label, Entry, Acc#{L => Label}, St); +collect_labels_1(Is, _Label, _Entry, Acc, St) -> + {Is,St#st{replace=Acc}}. %% label_defined(Is, Label) -> true | false. %% Test whether the label Label is defined at the start of the instruction @@ -418,13 +469,6 @@ invert_test(is_eq_exact) -> is_ne_exact; invert_test(is_ne_exact) -> is_eq_exact; invert_test(_) -> not_possible. -insert_labels([L|Ls], Is, [{jump,{f,L}}|Acc], St) -> - insert_labels(Ls, [{label,L}|Is], Acc, St); -insert_labels([L|Ls], Is, Acc, St) -> - insert_labels(Ls, [{label,L}|Is], Acc, St); -insert_labels([], Is, Acc, St) -> - opt(Is, Acc, St). - %% skip_unreachable([Instruction], St). %% Remove all instructions (including definitions of labels %% that have not been referenced yet) up to the next diff --git a/lib/compiler/src/beam_listing.erl b/lib/compiler/src/beam_listing.erl index 836378727b..73c6501fe5 100644 --- a/lib/compiler/src/beam_listing.erl +++ b/lib/compiler/src/beam_listing.erl @@ -23,7 +23,6 @@ -include("core_parse.hrl"). -include("v3_kernel.hrl"). --include("v3_life.hrl"). -include("beam_disasm.hrl"). -import(lists, [foreach/2]). @@ -31,7 +30,6 @@ -type code() :: cerl:c_module() | beam_utils:module_code() | #k_mdef{} - | {module(),_,_,_} %v3_life | [_]. %form-based format -spec module(file:io_device(), code()) -> 'ok'. @@ -43,13 +41,9 @@ module(File, #k_mdef{}=Kern) -> %% This is a kernel module. io:put_chars(File, v3_kernel_pp:format(Kern)); %%io:put_chars(File, io_lib:format("~p~n", [Kern])); -module(File, {Mod,Exp,Attr,Kern}) -> - %% This is output from beam_life (v3). - io:fwrite(File, "~w.~n~p.~n~p.~n", [Mod,Exp,Attr]), - foreach(fun (F) -> function(File, F) end, Kern); module(Stream, {Mod,Exp,Attr,Code,NumLabels}) -> - %% This is output from beam_codegen. - io:format(Stream, "{module, ~p}. %% version = ~w\n", + %% This is output from v3_codegen. + io:format(Stream, "{module, ~p}. %% version = ~w\n", [Mod, beam_opcodes:format_number()]), io:format(Stream, "\n{exports, ~p}.\n", [Exp]), io:format(Stream, "\n{attributes, ~p}.\n", [Attr]), @@ -60,19 +54,6 @@ module(Stream, {Mod,Exp,Attr,Code,NumLabels}) -> [Name, Arity, Entry]), io:put_chars(Stream, format_asm(Asm)) end, Code); -module(Stream, Code) when is_binary(Code) -> - #beam_file{ module = Module, compile_info = CInfo } = beam_disasm:file(Code), - Loaded = code:is_loaded(Module), - Sticky = code:is_sticky(Module), - [code:unstick_mod(Module) || Sticky], - - {module, Module} = code:load_binary(Module, proplists:get_value(source, CInfo), Code), - ok = erts_debug:df(Stream, Module), - - %% Restore loaded module - _ = [{module, Module} = code:load_file(Module) || Loaded =/= false], - [code:stick_mod(Module) || Sticky], - ok; module(Stream, [_|_]=Fs) -> %% Form-based abstract format. foreach(fun (F) -> io:format(Stream, "~p.\n", [F]) end, Fs). @@ -82,60 +63,3 @@ format_asm([{label,L}|Is]) -> format_asm([I|Is]) -> [io_lib:format(" ~p", [I]),".\n"|format_asm(Is)]; format_asm([]) -> []. - -function(File, {function,Name,Arity,Args,Body,Vdb,_Anno}) -> - io:nl(File), - io:format(File, "function ~p/~p.\n", [Name,Arity]), - io:format(File, " ~p.\n", [Args]), - print_vdb(File, Vdb), - put(beam_listing_nl, false), - nl(File), - foreach(fun(F) -> format(File, F, []) end, Body), - nl(File), - erase(beam_listing_nl). - -format(File, #l{ke=Ke,i=I,vdb=Vdb}, Ind) -> - nl(File), - ind_format(File, Ind, "~p ", [I]), - print_vdb(File, Vdb), - nl(File), - format(File, Ke, Ind); -format(File, Tuple, Ind) when is_tuple(Tuple) -> - ind_format(File, Ind, "{", []), - format_list(File, tuple_to_list(Tuple), [$\s|Ind]), - ind_format(File, Ind, "}", []); -format(File, List, Ind) when is_list(List) -> - ind_format(File, Ind, "[", []), - format_list(File, List, [$\s|Ind]), - ind_format(File, Ind, "]", []); -format(File, F, Ind) -> - ind_format(File, Ind, "~p", [F]). - -format_list(File, [F], Ind) -> - format(File, F, Ind); -format_list(File, [F|Fs], Ind) -> - format(File, F, Ind), - ind_format(File, Ind, ",", []), - format_list(File, Fs, Ind); -format_list(_, [], _) -> ok. - - -print_vdb(File, [{Var,F,E}|Vs]) -> - io:format(File, "~p:~p..~p ", [Var,F,E]), - print_vdb(File, Vs); -print_vdb(_, []) -> ok. - -ind_format(File, Ind, Format, Args) -> - case get(beam_listing_nl) of - true -> - put(beam_listing_nl, false), - io:put_chars(File, Ind); - false -> ok - end, - io:format(File, Format, Args). - -nl(File) -> - case put(beam_listing_nl, true) of - true -> ok; - false -> io:nl(File) - end. diff --git a/lib/compiler/src/beam_peep.erl b/lib/compiler/src/beam_peep.erl index 6df5c02334..eb3192fe8f 100644 --- a/lib/compiler/src/beam_peep.erl +++ b/lib/compiler/src/beam_peep.erl @@ -41,8 +41,7 @@ function({function,Name,Arity,CLabel,Is0}) -> Is = beam_jump:remove_unused_labels(Is1), {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -89,15 +88,37 @@ peep([{gc_bif,_,_,_,_,Dst}=I|Is], SeenTests0, Acc) -> peep([{jump,{f,L}},{label,L}=I|Is], _, Acc) -> %% Sometimes beam_jump has missed this optimization. peep(Is, gb_sets:empty(), [I|Acc]); -peep([{select,Op,R,F,Vls0}|Is], _, Acc) -> +peep([{select,Op,R,F,Vls0}|Is], SeenTests0, Acc0) -> case prune_redundant_values(Vls0, F) of [] -> %% No values left. Must convert to plain jump. I = {jump,F}, - peep(Is, gb_sets:empty(), [I|Acc]); + peep([I|Is], gb_sets:empty(), Acc0); + [{atom,_}=Value,Lbl] when Op =:= select_val -> + %% Single value left. Convert to regular test and pop redundant tests. + Is1 = [{test,is_eq_exact,F,[R,Value]},{jump,Lbl}|Is], + case Acc0 of + [{test,is_atom,F,[R]}|Acc] -> + peep(Is1, SeenTests0, Acc); + _ -> + peep(Is1, SeenTests0, Acc0) + end; + [{integer,_}=Value,Lbl] when Op =:= select_val -> + %% Single value left. Convert to regular test and pop redundant tests. + Is1 = [{test,is_eq_exact,F,[R,Value]},{jump,Lbl}|Is], + case Acc0 of + [{test,is_integer,F,[R]}|Acc] -> + peep(Is1, SeenTests0, Acc); + _ -> + peep(Is1, SeenTests0, Acc0) + end; + [Arity,Lbl] when Op =:= select_tuple_arity -> + %% Single value left. Convert to regular test + Is1 = [{test,test_arity,F,[R,Arity]},{jump,Lbl}|Is], + peep(Is1, SeenTests0, Acc0); [_|_]=Vls -> I = {select,Op,R,F,Vls}, - peep(Is, gb_sets:empty(), [I|Acc]) + peep(Is, gb_sets:empty(), [I|Acc0]) end; peep([{test,Op,_,Ops}=I|Is], SeenTests0, Acc) -> case beam_utils:is_pure_test(I) of diff --git a/lib/compiler/src/beam_receive.erl b/lib/compiler/src/beam_receive.erl index 1403e1e05e..3c8efa577c 100644 --- a/lib/compiler/src/beam_receive.erl +++ b/lib/compiler/src/beam_receive.erl @@ -82,8 +82,7 @@ function({function,Name,Arity,Entry,Is}) -> D = beam_utils:index_labels(Is), {function,Name,Arity,Entry,opt(Is, D, [])} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -207,6 +206,8 @@ opt_update_regs({label,Lbl}, R, L) -> %% A catch label for a previously seen catch instruction is OK. {R,L} end; +opt_update_regs({'try',_,{f,Lbl}}, R, L) -> + {R,gb_sets:add(Lbl, L)}; opt_update_regs({try_end,_}, R, L) -> {R,L}; opt_update_regs({line,_}, R, L) -> diff --git a/lib/compiler/src/beam_record.erl b/lib/compiler/src/beam_record.erl index 419089b1bc..58a6de6775 100644 --- a/lib/compiler/src/beam_record.erl +++ b/lib/compiler/src/beam_record.erl @@ -15,19 +15,12 @@ %% %% %CopyrightEnd% %% -%% File: beam_record.erl -%% Author: Björn-Egil Dahlberg -%% Created: 2014-09-03 -%% - --module(beam_record). --export([module/2]). %% Rewrite the instruction stream on tagged tuple tests. -%% Tagged tuples means a tuple of any arity with an atom as its first element. -%% Typically records, ok-tuples and error-tuples. -%% -%% from: +%% Tagged tuples means a tuple of any arity with an atom as its +%% first element, such as records and error tuples. +%% +%% From: %% ... %% {test,is_tuple,Fail,[Src]}. %% {test,test_arity,Fail,[Src,Sz]}. @@ -36,13 +29,16 @@ %% ... %% {test,is_eq_exact,Fail,[Dst,Atom]}. %% ... -%% to: +%% To: %% ... %% {test,is_tagged_tuple,Fail,[Src,Sz,Atom]}. %% ... +%% +-module(beam_record). +-export([module/2]). --import(lists, [reverse/1]). +-import(lists, [reverse/1,reverse/2]). -spec module(beam_utils:module_code(), [compile:option()]) -> {'ok',beam_utils:module_code()}. @@ -51,56 +47,85 @@ module({Mod,Exp,Attr,Fs0,Lc}, _Opt) -> Fs = [function(F) || F <- Fs0], {ok,{Mod,Exp,Attr,Fs,Lc}}. -function({function,Name,Arity,CLabel,Is}) -> +function({function,Name,Arity,CLabel,Is0}) -> try - Idx = beam_utils:index_labels(Is), - {function,Name,Arity,CLabel,rewrite(Is,Idx)} + Is1 = beam_utils:anno_defs(Is0), + Idx = beam_utils:index_labels(Is1), + Is = rewrite(reverse(Is1), Idx), + {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. -rewrite(Is,Idx) -> - rewrite(Is,Idx,[]). +rewrite(Is, Idx) -> + rewrite(Is, Idx, 0, []). -rewrite([{test,is_tuple,Fail,[Src]}=I1, - {test,test_arity,Fail,[Src,N]}=I2|Is],Idx,Acc) -> - case is_tagged_tuple(Is,Fail,Src,Idx) of +rewrite([{test,test_arity,Fail,[Src,N]}=TA, + {test,is_tuple,Fail,[Src]}=TT|Is], Idx, Def, Acc0) -> + case is_tagged_tuple(Acc0, Def, Fail, Src, Idx) of no -> - rewrite(Is,Idx,[I2,I1|Acc]); - {Atom,[{block,[]}|Is1]} -> - rewrite(Is1,Idx,[{test,is_tagged_tuple,Fail,[Src,N,Atom]}|Acc]); - {Atom,Is1} -> - rewrite(Is1,Idx,[{test,is_tagged_tuple,Fail,[Src,N,Atom]}|Acc]) + rewrite(Is, Idx, 0, [TT,TA|Acc0]); + {yes,Atom,Acc} -> + I = {test,is_tagged_tuple,Fail,[Src,N,Atom]}, + rewrite(Is, Idx, Def, [I|Acc]) end; -rewrite([I|Is],Idx,Acc) -> - rewrite(Is,Idx,[I|Acc]); -rewrite([],_,Acc) -> reverse(Acc). - -is_tagged_tuple([{block,[{set,[Dst],[Src],{get_tuple_element,0}}=B|Bs]}, - {test,is_eq_exact,Fail,[Dst,{atom,_}=Atom]}|Is],Fail,Src,Idx) -> +rewrite([{block,[{'%anno',{def,Def}}|Bl]}|Is], Idx, _Def, Acc) -> + rewrite(Is, Idx, Def, [{block,Bl}|Acc]); +rewrite([{label,L}=I|Is], Idx0, Def, Acc) -> + Idx = beam_utils:index_label(L, Acc, Idx0), + rewrite(Is, Idx, Def, [I|Acc]); +rewrite([I|Is], Idx, Def, Acc) -> + rewrite(Is, Idx, Def, [I|Acc]); +rewrite([], _, _, Acc) -> Acc. - %% if Dst is killed in the instruction stream and at fail label, - %% we can safely remove get_tuple_element. - %% - %% if Dst is not killed in the stream, we cannot remove get_tuple_element - %% since it is referenced. - - case is_killed(Dst,Is,Fail,Idx) of - true -> {Atom,[{block,Bs}|Is]}; - false -> {Atom,[{block,[B|Bs]}|Is]} +is_tagged_tuple([{block,Bl}, + {test,is_eq_exact,Fail,[Dst,{atom,_}=Atom]}|Is], + Def, Fail, Src, Idx) -> + case is_tagged_tuple_1(Bl, Is, Fail, Src, Dst, Idx, Def, []) of + no -> + no; + {yes,[]} -> + {yes,Atom,Is}; + {yes,[_|_]=Block} -> + {yes,Atom,[{block,Block}|Is]} end; -is_tagged_tuple([{block,[{set,_,_,_}=B|Bs]}, - {test,is_eq_exact,_,_}=I|Is],Fail,Src,Idx) -> - case is_tagged_tuple([{block,Bs},I|Is],Fail,Src,Idx) of - {Atom,[{block,Bsr}|Isr]} -> {Atom,[{block,[B|Bsr]}|Isr]}; - no -> no +is_tagged_tuple(_, _, _, _, _) -> + no. + +is_tagged_tuple_1([{set,[Dst],[Src],{get_tuple_element,0}}=I|Bl], + Is, Fail, Src, Dst, Idx, Def, Acc) -> + %% Check usage of Dst to find out whether the get_tuple_element + %% is needed. + case usage(Dst, Is, Fail, Idx) of + killed -> + %% Safe to remove the get_tuple_element instruction. + {yes,reverse(Acc, Bl)}; + used -> + %% Actively used. Must keep instruction. + {yes,reverse(Acc, [I|Bl])}; + not_used -> + %% Not actually used (but must be initialized). + case is_defined(Dst, Def) of + false -> + %% Dst must be initialized, but the + %% actual value does not matter. + Kill = {set,[Dst],[nil],move}, + {yes,reverse(Acc, [Kill|Bl])}; + true -> + %% The register is previously initialized. + %% We can remove the instruction. + {yes,reverse(Acc, Bl)} + end end; -is_tagged_tuple(_Is,_Fail,_Src,_Idx) -> +is_tagged_tuple_1([I|Bl], Is, Fail, Src, Dst, Idx, Def, Acc) -> + is_tagged_tuple_1(Bl, Is, Fail, Src, Dst, Idx, Def, [I|Acc]); +is_tagged_tuple_1(_, _, _, _, _, _, _, _) -> no. -is_killed(Dst,Is,{_,Lbl},Idx) -> - beam_utils:is_killed(Dst,Is,Idx) andalso - beam_utils:is_killed_at(Dst,Lbl,Idx). +usage(Dst, Is, Fail, Idx) -> + beam_utils:usage(Dst, [{test,is_number,Fail,[nil]}|Is], Idx). + +is_defined({x,X}, Def) -> + (Def bsr X) band 1 =:= 1. diff --git a/lib/compiler/src/beam_reorder.erl b/lib/compiler/src/beam_reorder.erl index 910b7f6b0a..63bb57a1ac 100644 --- a/lib/compiler/src/beam_reorder.erl +++ b/lib/compiler/src/beam_reorder.erl @@ -35,8 +35,7 @@ function({function,Name,Arity,CLabel,Is0}) -> Is = reorder(Is0), {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. diff --git a/lib/compiler/src/beam_split.erl b/lib/compiler/src/beam_split.erl index d041f18806..52dd89b5bb 100644 --- a/lib/compiler/src/beam_split.erl +++ b/lib/compiler/src/beam_split.erl @@ -50,8 +50,9 @@ split_block([{set,[R],[_,_,_]=As,{bif,is_record,{f,Lbl}}}|Is], Bl, Acc) -> split_block(Is, [], [{bif,is_record,{f,Lbl},As,R}|make_block(Bl, Acc)]); split_block([{set,[R],As,{bif,N,{f,Lbl}=Fail}}|Is], Bl, Acc) when Lbl =/= 0 -> split_block(Is, [], [{bif,N,Fail,As,R}|make_block(Bl, Acc)]); -split_block([{set,[R],As,{bif,raise,{f,_}=Fail}}|Is], Bl, Acc) -> - split_block(Is, [], [{bif,raise,Fail,As,R}|make_block(Bl, Acc)]); +split_block([{set,[],[],{line,_}=Line}, + {set,[R],As,{bif,raise,{f,_}=Fail}}|Is], Bl, Acc) -> + split_block(Is, [], [{bif,raise,Fail,As,R},Line|make_block(Bl, Acc)]); split_block([{set,[R],As,{alloc,Live,{gc_bif,N,{f,Lbl}=Fail}}}|Is], Bl, Acc) when Lbl =/= 0 -> split_block(Is, [], [{gc_bif,N,Fail,Live,As,R}|make_block(Bl, Acc)]); @@ -61,8 +62,6 @@ split_block([{set,[D],[S|Puts],{alloc,R,{put_map,Op,{f,Lbl}=Fail}}}|Is], make_block(Bl, Acc)]); split_block([{set,[R],[],{try_catch,Op,L}}|Is], Bl, Acc) -> split_block(Is, [], [{Op,R,L}|make_block(Bl, Acc)]); -split_block([{set,[],[],{line,_}=Line}|Is], Bl, Acc) -> - split_block(Is, [], [Line|make_block(Bl, Acc)]); split_block([I|Is], Bl, Acc) -> split_block(Is, [I|Bl], Acc); split_block([], Bl, Acc) -> make_block(Bl, Acc). diff --git a/lib/compiler/src/beam_type.erl b/lib/compiler/src/beam_type.erl index fc2c7a991b..b83ed17b55 100644 --- a/lib/compiler/src/beam_type.erl +++ b/lib/compiler/src/beam_type.erl @@ -40,11 +40,10 @@ function({function,Name,Arity,CLabel,Asm0}) -> Asm1 = beam_utils:live_opt(Asm0), Asm2 = opt(Asm1, [], tdb_new()), Asm3 = beam_utils:live_opt(Asm2), - Asm = beam_utils:delete_live_annos(Asm3), + Asm = beam_utils:delete_annos(Asm3), {function,Name,Arity,CLabel,Asm} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -81,80 +80,99 @@ simplify(Is0, TypeDb0) -> %% Basic simplification, mostly tuples, no floating point optimizations. simplify_basic(Is, Ts) -> - simplify_basic_1(Is, Ts, []). - -simplify_basic_1([{set,[D],[{integer,Index},Reg],{bif,element,_}}=I0|Is], Ts0, Acc) -> - I = case max_tuple_size(Reg, Ts0) of - Sz when 0 < Index, Index =< Sz -> - {set,[D],[Reg],{get_tuple_element,Index-1}}; - _Other -> I0 - end, - Ts = update(I, Ts0), - simplify_basic_1(Is, Ts, [I|Acc]); -simplify_basic_1([{set,[D],[TupleReg],{get_tuple_element,0}}=I|Is0], Ts0, Acc) -> - case tdb_find(TupleReg, Ts0) of - {tuple,_,[Contents]} -> - simplify_basic_1([{set,[D],[Contents],move}|Is0], Ts0, Acc); - _ -> - Ts = update(I, Ts0), - simplify_basic_1(Is0, Ts, [I|Acc]) + simplify_basic(Is, Ts, []). + +simplify_basic([I0|Is], Ts0, Acc) -> + case simplify_instr(I0, Ts0) of + [] -> + simplify_basic(Is, Ts0, Acc); + [I] -> + Ts = update(I, Ts0), + simplify_basic(Is, Ts, [I|Acc]) + end; +simplify_basic([], Ts, Acc) -> + {reverse(Acc),Ts}. + +simplify_instr({set,[D],[{integer,Index},Reg],{bif,element,_}}=I, Ts) -> + case max_tuple_size(Reg, Ts) of + Sz when 0 < Index, Index =< Sz -> + [{set,[D],[Reg],{get_tuple_element,Index-1}}]; + _ -> [I] end; -simplify_basic_1([{set,_,_,{try_catch,_,_}}=I|Is], _Ts, Acc) -> - simplify_basic_1(Is, tdb_new(), [I|Acc]); -simplify_basic_1([{test,is_atom,_,[R]}=I|Is], Ts, Acc) -> +simplify_instr({test,is_atom,_,[R]}=I, Ts) -> case tdb_find(R, Ts) of - boolean -> simplify_basic_1(Is, Ts, Acc); - _ -> simplify_basic_1(Is, Ts, [I|Acc]) + boolean -> []; + _ -> [I] end; -simplify_basic_1([{test,is_integer,_,[R]}=I|Is], Ts, Acc) -> +simplify_instr({test,is_integer,_,[R]}=I, Ts) -> case tdb_find(R, Ts) of - integer -> simplify_basic_1(Is, Ts, Acc); - {integer,_} -> simplify_basic_1(Is, Ts, Acc); - _ -> simplify_basic_1(Is, Ts, [I|Acc]) + integer -> []; + {integer,_} -> []; + _ -> [I] end; -simplify_basic_1([{test,is_tuple,_,[R]}=I|Is], Ts, Acc) -> +simplify_instr({set,[D],[TupleReg],{get_tuple_element,0}}=I, Ts) -> + case tdb_find(TupleReg, Ts) of + {tuple,_,_,[Contents]} -> + [{set,[D],[Contents],move}]; + _ -> + [I] + end; +simplify_instr({test,is_tuple,_,[R]}=I, Ts) -> case tdb_find(R, Ts) of - {tuple,_,_} -> simplify_basic_1(Is, Ts, Acc); - _ -> simplify_basic_1(Is, Ts, [I|Acc]) + {tuple,_,_,_} -> []; + _ -> [I] end; -simplify_basic_1([{test,is_map,_,[R]}=I|Is], Ts0, Acc) -> - case tdb_find(R, Ts0) of - map -> simplify_basic_1(Is, Ts0, Acc); - _Other -> - Ts = update(I, Ts0), - simplify_basic_1(Is, Ts, [I|Acc]) +simplify_instr({test,test_arity,_,[R,Arity]}=I, Ts) -> + case tdb_find(R, Ts) of + {tuple,exact_size,Arity,_} -> []; + _ -> [I] end; -simplify_basic_1([{test,is_nonempty_list,_,[R]}=I|Is], Ts0, Acc) -> - case tdb_find(R, Ts0) of - nonempty_list -> simplify_basic_1(Is, Ts0, Acc); - _Other -> - Ts = update(I, Ts0), - simplify_basic_1(Is, Ts, [I|Acc]) - end; -simplify_basic_1([{test,is_eq_exact,Fail,[R,{atom,_}=Atom]}=I|Is0], Ts0, Acc0) -> - Acc = case tdb_find(R, Ts0) of - {atom,_}=Atom -> Acc0; - {atom,_} -> [{jump,Fail}|Acc0]; - _ -> [I|Acc0] - end, - Ts = update(I, Ts0), - simplify_basic_1(Is0, Ts, Acc); -simplify_basic_1([{select,select_val,Reg,_,_}=I0|Is], Ts, Acc) -> - I = case tdb_find(Reg, Ts) of - {integer,Range} -> - simplify_select_val_int(I0, Range); - boolean -> - simplify_select_val_bool(I0); - _ -> - I0 - end, - simplify_basic_1(Is, tdb_new(), [I|Acc]); -simplify_basic_1([I|Is], Ts0, Acc) -> - Ts = update(I, Ts0), - simplify_basic_1(Is, Ts, [I|Acc]); -simplify_basic_1([], Ts, Acc) -> - Is = reverse(Acc), - {Is,Ts}. +simplify_instr({test,is_map,_,[R]}=I, Ts) -> + case tdb_find(R, Ts) of + map -> []; + _ -> [I] + end; +simplify_instr({test,is_nonempty_list,_,[R]}=I, Ts) -> + case tdb_find(R, Ts) of + nonempty_list -> []; + _ -> [I] + end; +simplify_instr({test,is_eq_exact,Fail,[R,{atom,_}=Atom]}=I, Ts) -> + case tdb_find(R, Ts) of + {atom,_}=Atom -> []; + {atom,_} -> [{jump,Fail}]; + _ -> [I] + end; +simplify_instr({test,is_record,_,[R,{atom,_}=Tag,{integer,Arity}]}=I, Ts) -> + case tdb_find(R, Ts) of + {tuple,exact_size,Arity,[Tag]} -> []; + _ -> [I] + end; +simplify_instr({select,select_val,Reg,_,_}=I, Ts) -> + [case tdb_find(Reg, Ts) of + {integer,Range} -> + simplify_select_val_int(I, Range); + boolean -> + simplify_select_val_bool(I); + _ -> + I + end]; +simplify_instr({test,bs_test_unit,_,[Src,Unit]}=I, Ts) -> + case tdb_find(Src, Ts) of + {binary,U} when U rem Unit =:= 0 -> []; + _ -> [I] + end; +simplify_instr({test,is_binary,_,[Src]}=I, Ts) -> + case tdb_find(Src, Ts) of + {binary,U} when U rem 8 =:= 0 -> []; + _ -> [I] + end; +simplify_instr({test,is_bitstr,_,[Src]}=I, Ts) -> + case tdb_find(Src, Ts) of + {binary,_} -> []; + _ -> [I] + end; +simplify_instr(I, _) -> [I]. simplify_select_val_int({select,select_val,R,_,L0}=I, {Min,Max}) -> Vs = sort([V || {integer,V} <- L0]), @@ -285,7 +303,7 @@ clearerror([], OrigIs) -> [{set,[],[],fclearerror}|OrigIs]. %% Combine two blocks and eliminate any move instructions that assign %% to registers that are killed later in the block. %% -merge_blocks(B1, [{'%live',_,_}|B2]) -> +merge_blocks(B1, [{'%anno',_}|B2]) -> merge_blocks_1(B1++[{set,[],[],stop_here}|B2]). merge_blocks_1([{set,[],_,stop_here}|Is]) -> Is; @@ -334,29 +352,17 @@ flt_need_heap_2({set,_,_,{put_tuple,_}}, H, Fl) -> {[],H+1,Fl}; flt_need_heap_2({set,_,_,put}, H, Fl) -> {[],H+1,Fl}; -%% Then the "neutral" instructions. We just pass them. -flt_need_heap_2({set,[{fr,_}],_,_}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,[],[],fclearerror}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,[],[],fcheckerror}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,{bif,_,_}}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,move}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,{get_tuple_element,_}}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,get_list}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,{try_catch,_,_}}, H, Fl) -> - {[],H,Fl}; -flt_need_heap_2({set,_,_,init}, H, Fl) -> - {[],H,Fl}; -%% All other instructions should cause the insertion of an allocation +%% The following instructions cause the insertion of an allocation %% instruction if needed. +flt_need_heap_2({set,_,_,{alloc,_,_}}, H, Fl) -> + {flt_alloc(H, Fl),0,0}; +flt_need_heap_2({set,_,_,{set_tuple_element,_}}, H, Fl) -> + {flt_alloc(H, Fl),0,0}; +flt_need_heap_2({'%anno',_}, H, Fl) -> + {flt_alloc(H, Fl),0,0}; +%% All other instructions are "neutral". We just pass them. flt_need_heap_2(_, H, Fl) -> - {flt_alloc(H, Fl),0,0}. + {[],H,Fl}. flt_alloc(0, 0) -> []; @@ -379,7 +385,7 @@ build_alloc(Words, Floats) -> {alloc,[{words,Words},{floats,Floats}]}. %% is not continous at an allocation function (e.g. if {x,0} and {x,2} %% are live, but not {x,1}). -flt_liveness([{'%live',_Live,Regs}=LiveInstr|Is]) -> +flt_liveness([{'%anno',{used,Regs}}=LiveInstr|Is]) -> flt_liveness_1(Is, Regs, [LiveInstr]). flt_liveness_1([{set,Ds,Ss,{alloc,Live0,Alloc}}|Is], Regs0, Acc) -> @@ -391,7 +397,7 @@ flt_liveness_1([{set,Ds,Ss,{alloc,Live0,Alloc}}|Is], Regs0, Acc) -> flt_liveness_1([{set,Ds,_,_}=I|Is], Regs0, Acc) -> Regs = x_live(Ds, Regs0), flt_liveness_1(Is, Regs, [I|Acc]); -flt_liveness_1([{'%live',_,_}], _Regs, Acc) -> +flt_liveness_1([{'%anno',_}], _Regs, Acc) -> reverse(Acc). init_regs(Live) -> @@ -415,13 +421,14 @@ x_live([], Regs) -> Regs. %% Update the type database to account for executing an instruction. %% %% First the cases for instructions inside basic blocks. -update({'%live',_,_}, Ts) -> Ts; +update({'%anno',_}, Ts) -> + Ts; update({set,[D],[S],move}, Ts) -> tdb_copy(S, D, Ts); update({set,[D],[{integer,I},Reg],{bif,element,_}}, Ts0) -> - tdb_update([{Reg,{tuple,I,[]}},{D,kill}], Ts0); + tdb_update([{Reg,{tuple,min_size,I,[]}},{D,kill}], Ts0); update({set,[D],[_Index,Reg],{bif,element,_}}, Ts0) -> - tdb_update([{Reg,{tuple,0,[]}},{D,kill}], Ts0); + tdb_update([{Reg,{tuple,min_size,0,[]}},{D,kill}], Ts0); update({set,[D],Args,{bif,N,_}}, Ts0) -> Ar = length(Args), BoolOp = erl_internal:new_type_test(N, Ar) orelse @@ -470,8 +477,6 @@ update({set,[D],[S1,S2],{alloc,_,{gc_bif,Op,{f,0}}}}, Ts0) -> update({set,[],_Src,_Op}, Ts0) -> Ts0; update({set,[D],_Src,_Op}, Ts0) -> tdb_update([{D,kill}], Ts0); -update({set,[D1,D2],_Src,_Op}, Ts0) -> - tdb_update([{D1,kill},{D2,kill}], Ts0); update({kill,D}, Ts) -> tdb_update([{D,kill}], Ts); @@ -479,7 +484,7 @@ update({kill,D}, Ts) -> update({test,is_float,_Fail,[Src]}, Ts0) -> tdb_update([{Src,float}], Ts0); update({test,test_arity,_Fail,[Src,Arity]}, Ts0) -> - tdb_update([{Src,{tuple,Arity,[]}}], Ts0); + tdb_update([{Src,{tuple,exact_size,Arity,[]}}], Ts0); update({test,is_map,_Fail,[Src]}, Ts0) -> tdb_update([{Src,map}], Ts0); update({get_map_elements,_,Src,{list,Elems0}}, Ts0) -> @@ -493,15 +498,19 @@ update({test,is_eq_exact,_,[Reg,{atom,_}=Atom]}, Ts) -> error -> Ts; {tuple_element,TupleReg,0} -> - tdb_update([{TupleReg,{tuple,1,[Atom]}}], Ts); + tdb_update([{TupleReg,{tuple,min_size,1,[Atom]}}], Ts); _ -> Ts end; update({test,is_record,_Fail,[Src,Tag,{integer,Arity}]}, Ts) -> - tdb_update([{Src,{tuple,Arity,[Tag]}}], Ts); + tdb_update([{Src,{tuple,exact_size,Arity,[Tag]}}], Ts); -%% Binary matching +%% Binaries and binary matching. +update({test,is_binary,_Fail,[Src]}, Ts0) -> + tdb_update([{Src,{binary,8}}], Ts0); +update({test,is_bitstr,_Fail,[Src]}, Ts0) -> + tdb_update([{Src,{binary,1}}], Ts0); update({test,bs_get_integer2,_,_,Args,Dst}, Ts) -> tdb_update([{Dst,get_bs_integer_type(Args)}], Ts); update({test,bs_get_utf8,_,_,_,Dst}, Ts) -> @@ -510,8 +519,10 @@ update({test,bs_get_utf16,_,_,_,Dst}, Ts) -> tdb_update([{Dst,?UNICODE_INT}], Ts); update({test,bs_get_utf32,_,_,_,Dst}, Ts) -> tdb_update([{Dst,?UNICODE_INT}], Ts); +update({bs_init,_,{bs_init2,_,_},_,_,Dst}, Ts) -> + tdb_update([{Dst,{binary,8}}], Ts); update({bs_init,_,_,_,_,Dst}, Ts) -> - tdb_update([{Dst,kill}], Ts); + tdb_update([{Dst,{binary,1}}], Ts); update({bs_put,_,_,_}, Ts) -> Ts; update({bs_save2,_,_}, Ts) -> @@ -520,12 +531,19 @@ update({bs_restore2,_,_}, Ts) -> Ts; update({bs_context_to_binary,Dst}, Ts) -> tdb_update([{Dst,kill}], Ts); -update({test,bs_start_match2,_,_,_,Dst}, Ts) -> - tdb_update([{Dst,kill}], Ts); -update({test,bs_get_binary2,_,_,_,Dst}, Ts) -> - tdb_update([{Dst,kill}], Ts); +update({test,bs_start_match2,_,_,[Src,_],Dst}, Ts) -> + Type = case tdb_find(Src, Ts) of + {binary,_}=Type0 -> Type0; + _ -> {binary,1} + end, + tdb_update([{Dst,Type}], Ts); +update({test,bs_get_binary2,_,_,[_,_,Unit,_],Dst}, Ts) -> + true = is_integer(Unit), %Assertion. + tdb_update([{Dst,{binary,Unit}}], Ts); update({test,bs_get_float2,_,_,_,Dst}, Ts) -> tdb_update([{Dst,float}], Ts); +update({test,bs_test_unit,_,[Src,Unit]}, Ts) -> + tdb_update([{Src,{binary,Unit}}], Ts); update({test,_Test,_Fail,_Other}, Ts) -> Ts; @@ -540,7 +558,7 @@ update({call_ext,Ar,{extfunc,math,Math,Ar}}, Ts) -> update({call_ext,3,{extfunc,erlang,setelement,3}}, Ts0) -> Ts = tdb_kill_xregs(Ts0), case tdb_find({x,1}, Ts0) of - {tuple,Sz,_}=T0 -> + {tuple,SzKind,Sz,_}=T0 -> T = case tdb_find({x,0}, Ts0) of {integer,{I,I}} when I > 1 -> %% First element is not changed. The result @@ -549,7 +567,7 @@ update({call_ext,3,{extfunc,erlang,setelement,3}}, Ts0) -> _ -> %% Position is 1 or unknown. May change the %% first element of the tuple. - {tuple,Sz,[]} + {tuple,SzKind,Sz,[]} end, tdb_update([{{x,0},T}], Ts); _ -> @@ -562,6 +580,7 @@ update({call_fun, _}, Ts) -> tdb_kill_xregs(Ts); update({apply, _}, Ts) -> tdb_kill_xregs(Ts); update({line,_}, Ts) -> Ts; +update({'%',_}, Ts) -> Ts; %% The instruction is unknown. Kill all information. update(_I, _Ts) -> tdb_new(). @@ -633,7 +652,7 @@ possibly_numeric(_) -> false. max_tuple_size(Reg, Ts) -> case tdb_find(Reg, Ts) of - {tuple,Sz,_} -> Sz; + {tuple,_,Sz,_} -> Sz; _Other -> 0 end. @@ -789,16 +808,20 @@ checkerror_2(OrigIs) -> [{set,[],[],fcheckerror}|OrigIs]. %%% Routines for maintaining a type database. The type database %%% associates type information with registers. %%% -%%% {tuple,Size,First} means that the corresponding register contains a -%%% tuple with *at least* Size elements. An tuple with unknown -%%% size is represented as {tuple,0,[]}. First is either [] (meaning that -%%% the tuple's first element is unknown) or [FirstElement] (the contents -%%% of the first element). +%%% {tuple,min_size,Size,First} means that the corresponding register contains +%%% a tuple with *at least* Size elements (conversely, exact_size means that it +%%% contains a tuple with *exactly* Size elements). An tuple with unknown size +%%% is represented as {tuple,min_size,0,[]}. First is either [] (meaning that +%%% the tuple's first element is unknown) or [FirstElement] (the contents of +%%% the first element). %%% %%% 'float' means that the register contains a float. %%% %%% 'integer' or {integer,{Min,Max}} that the register contains an %%% integer. +%%% +%%% {binary,Unit} means that the register contains a binary/bitstring aligned +%%% to unit Unit. %% tdb_new() -> EmptyDataBase %% Creates a new, empty type database. @@ -837,7 +860,7 @@ tdb_copy(Literal, D, Ts) -> {literal,#{}} -> map; {literal,Tuple} when tuple_size(Tuple) >= 1 -> Lit = tag_literal(element(1, Tuple)), - {tuple,tuple_size(Tuple),[Lit]}; + {tuple,exact_size,tuple_size(Tuple),[Lit]}; _ -> term end, if @@ -859,14 +882,14 @@ tag_literal(Lit) -> {literal,Lit}. %% Updates a type database. If a 'kill' operation is given, the type %% information for that register will be removed from the database. %% A kill operation takes precedence over other operations for the same -%% register (i.e. [{{x,0},kill},{{x,0},{tuple,5,[]}}] means that the +%% register (i.e. [{{x,0},kill},{{x,0},{tuple,min_size,5,[]}}] means that the %% the existing type information, if any, will be discarded, and the -%% the '{tuple,5,[]}' information ignored. +%% the '{tuple,min_size,5,[]}' information ignored. %% %% If NewInfo information is given and there exists information about %% the register, the old and new type information will be merged. -%% For instance, {tuple,5,_} and {tuple,10,_} will be merged to produce -%% {tuple,10,_}. +%% For instance, {tuple,min_size,5,_} and {tuple,min_size,10,_} will be merged +%% to produce {tuple,min_size,10,_}. tdb_update(Uis0, Ts0) -> Uis1 = filter(fun ({{x,_},_Op}) -> true; @@ -904,34 +927,41 @@ tdb_kill_xregs([]) -> []. remove_key(Key, [{Key,_Op}|Ops]) -> remove_key(Key, Ops); remove_key(_, Ops) -> Ops. - + merge_type_info(I, I) -> I; -merge_type_info({tuple,Sz1,Same}, {tuple,Sz2,Same}=Max) when Sz1 < Sz2 -> +merge_type_info({tuple,min_size,Sz1,Same}, {tuple,min_size,Sz2,Same}=Max) when Sz1 < Sz2 -> Max; -merge_type_info({tuple,Sz1,Same}=Max, {tuple,Sz2,Same}) when Sz1 > Sz2 -> +merge_type_info({tuple,min_size,Sz1,Same}=Max, {tuple,min_size,Sz2,Same}) when Sz1 > Sz2 -> Max; -merge_type_info({tuple,Sz1,[]}, {tuple,_Sz2,First}=Tuple2) -> - merge_type_info({tuple,Sz1,First}, Tuple2); -merge_type_info({tuple,_Sz1,First}=Tuple1, {tuple,Sz2,_}) -> - merge_type_info(Tuple1, {tuple,Sz2,First}); +merge_type_info({tuple,exact_size,_,Same}=Exact, {tuple,_,_,Same}) -> + Exact; +merge_type_info({tuple,_,_,Same},{tuple,exact_size,_,Same}=Exact) -> + Exact; +merge_type_info({tuple,SzKind1,Sz1,[]}, {tuple,_SzKind2,_Sz2,First}=Tuple2) -> + merge_type_info({tuple,SzKind1,Sz1,First}, Tuple2); +merge_type_info({tuple,_SzKind1,_Sz1,First}=Tuple1, {tuple,SzKind2,Sz2,_}) -> + merge_type_info(Tuple1, {tuple,SzKind2,Sz2,First}); merge_type_info(integer, {integer,_}) -> integer; merge_type_info({integer,_}, integer) -> integer; merge_type_info({integer,{Min1,Max1}}, {integer,{Min2,Max2}}) -> {integer,{max(Min1, Min2),min(Max1, Max2)}}; +merge_type_info({binary,U1}, {binary,U2}) -> + {binary,max(U1, U2)}; merge_type_info(NewType, _) -> verify_type(NewType), NewType. verify_type({atom,_}) -> ok; +verify_type({binary,U}) when is_integer(U) -> ok; verify_type(boolean) -> ok; verify_type(integer) -> ok; verify_type({integer,{Min,Max}}) when is_integer(Min), is_integer(Max) -> ok; verify_type(map) -> ok; verify_type(nonempty_list) -> ok; -verify_type({tuple,Sz,[]}) when is_integer(Sz) -> ok; -verify_type({tuple,Sz,[_]}) when is_integer(Sz) -> ok; +verify_type({tuple,_,Sz,[]}) when is_integer(Sz) -> ok; +verify_type({tuple,_,Sz,[_]}) when is_integer(Sz) -> ok; verify_type({tuple_element,_,_}) -> ok; verify_type(float) -> ok. diff --git a/lib/compiler/src/beam_utils.erl b/lib/compiler/src/beam_utils.erl index e39fbdc3b7..814cfb8265 100644 --- a/lib/compiler/src/beam_utils.erl +++ b/lib/compiler/src/beam_utils.erl @@ -22,15 +22,23 @@ -module(beam_utils). -export([is_killed_block/2,is_killed/3,is_killed_at/3, - is_not_used/3, - empty_label_index/0,index_label/3,index_labels/1, + is_not_used/3,usage/3, + empty_label_index/0,index_label/3,index_labels/1,replace_labels/4, code_at/2,bif_to_test/3,is_pure_test/1, - live_opt/1,delete_live_annos/1,combine_heap_needs/2, - split_even/1]). + live_opt/1,delete_annos/1,combine_heap_needs/2, + anno_defs/1, + split_even/1 + ]). -export_type([code_index/0,module_code/0,instruction/0]). --import(lists, [member/2,sort/1,reverse/1,splitwith/2]). +-import(lists, [flatmap/2,map/2,member/2,sort/1,reverse/1,splitwith/2]). + +-define(is_const(Val), (Val =:= nil orelse + element(1, Val) =:= integer orelse + element(1, Val) =:= float orelse + element(1, Val) =:= atom orelse + element(1, Val) =:= literal)). %% instruction() describes all instructions that are used during optimzation %% (from beam_a to beam_z). @@ -54,6 +62,23 @@ {lbl :: code_index(), %Label to code index. res :: result_cache()}). %Result cache for each label. +%% usage(Register, [Instruction], State) -> killed|not_used|used. +%% Determine the usage of Register in the instruction sequence. +%% The return value is one of: +%% +%% killed - The register is not used in any way. +%% not_used - The register is referenced only by an allocating instruction +%% (the actual value does not matter). +%% used - The register is used (its value do matter). + +-spec usage(beam_asm:reg(), [instruction()], code_index()) -> + 'killed' | 'not_used' | 'used'. + +usage(R, Is, D) -> + St = #live{lbl=D,res=gb_trees:empty()}, + {Usage,_} = check_liveness(R, Is, St), + Usage. + %% is_killed_block(Register, [Instruction]) -> true|false %% Determine whether a register is killed by the instruction sequence inside @@ -70,7 +95,7 @@ is_killed_block({x,X}, [{set,_,_,{alloc,Live,_}}|_]) -> X >= Live; is_killed_block(R, [{set,Ds,Ss,_Op}|Is]) -> not member(R, Ss) andalso (member(R, Ds) orelse is_killed_block(R, Is)); -is_killed_block(R, [{'%live',_,Regs}|Is]) -> +is_killed_block(R, [{'%anno',{used,Regs}}|Is]) -> case R of {x,X} when (Regs bsr X) band 1 =:= 0 -> true; _ -> is_killed_block(R, Is) @@ -93,6 +118,7 @@ is_killed(R, Is, D) -> St = #live{lbl=D,res=gb_trees:empty()}, case check_liveness(R, Is, St) of {killed,_} -> true; + {exit_not_used,_} -> false; {_,_} -> false end. @@ -105,6 +131,7 @@ is_killed_at(R, Lbl, D) when is_integer(Lbl) -> St0 = #live{lbl=D,res=gb_trees:empty()}, case check_liveness_at(R, Lbl, St0) of {killed,_} -> true; + {exit_not_used,_} -> false; {_,_} -> false end. @@ -121,6 +148,7 @@ is_not_used(R, Is, D) -> St = #live{lbl=D,res=gb_trees:empty()}, case check_liveness(R, Is, St) of {used,_} -> false; + {exit_not_used,_} -> true; {_,_} -> true end. @@ -160,6 +188,18 @@ index_label(Lbl, Is0, Acc) -> code_at(L, Ll) -> gb_trees:get(L, Ll). +%% replace_labels(FunctionIs, Tail, ReplaceDb, Fallback) -> FunctionIs. +%% Replace all labels in instructions according to the ReplaceDb. +%% If label is not found the Fallback is called with the label to +%% produce a new one. + +-spec replace_labels([instruction()], + [instruction()], + #{beam_asm:label() => beam_asm:label()}, + fun((beam_asm:label()) -> term())) -> [instruction()]. +replace_labels(Is, Acc, D, Fb) -> + replace_labels_1(Is, Acc, D, Fb). + %% bif_to_test(Bif, [Op], Fail) -> {test,Test,Fail,[Op]} %% Convert a BIF to a test. Fail if not possible. @@ -185,10 +225,20 @@ bif_to_test('>', [A,B], Fail) -> {test,is_lt,Fail,[B,A]}; bif_to_test('<', [_,_]=Ops, Fail) -> {test,is_lt,Fail,Ops}; bif_to_test('>=', [_,_]=Ops, Fail) -> {test,is_ge,Fail,Ops}; bif_to_test('==', [A,nil], Fail) -> {test,is_nil,Fail,[A]}; +bif_to_test('==', [nil,A], Fail) -> {test,is_nil,Fail,[A]}; +bif_to_test('==', [C,A], Fail) when ?is_const(C) -> + {test,is_eq,Fail,[A,C]}; bif_to_test('==', [_,_]=Ops, Fail) -> {test,is_eq,Fail,Ops}; +bif_to_test('/=', [C,A], Fail) when ?is_const(C) -> + {test,is_ne,Fail,[A,C]}; bif_to_test('/=', [_,_]=Ops, Fail) -> {test,is_ne,Fail,Ops}; bif_to_test('=:=', [A,nil], Fail) -> {test,is_nil,Fail,[A]}; +bif_to_test('=:=', [nil,A], Fail) -> {test,is_nil,Fail,[A]}; +bif_to_test('=:=', [C,A], Fail) when ?is_const(C) -> + {test,is_eq_exact,Fail,[A,C]}; bif_to_test('=:=', [_,_]=Ops, Fail) -> {test,is_eq_exact,Fail,Ops}; +bif_to_test('=/=', [C,A], Fail) when ?is_const(C) -> + {test,is_ne_exact,Fail,[A,C]}; bif_to_test('=/=', [_,_]=Ops, Fail) -> {test,is_ne_exact,Fail,Ops}; bif_to_test(is_record, [_,_,_]=Ops, Fail) -> {test,is_record,Fail,Ops}. @@ -220,7 +270,7 @@ is_pure_test({test,Op,_,Ops}) -> %% Go through the instruction sequence in reverse execution %% order, keep track of liveness and remove 'move' instructions %% whose destination is a register that will not be used. -%% Also insert {'%live',Live,Regs} annotations at the beginning +%% Also insert {used,Regs} annotations at the beginning %% and end of each block. -spec live_opt([instruction()]) -> [instruction()]. @@ -235,35 +285,52 @@ live_opt(Is0) -> Bef ++ [Fi|live_opt(reverse(Is), 0, D, [])]. -%% delete_live_annos([Instruction]) -> [Instruction]. -%% Delete all live annotations. +%% delete_annos([Instruction]) -> [Instruction]. +%% Delete all annotations. --spec delete_live_annos([instruction()]) -> [instruction()]. +-spec delete_annos([instruction()]) -> [instruction()]. -delete_live_annos([{block,Bl0}|Is]) -> - case delete_live_annos(Bl0) of - [] -> delete_live_annos(Is); - [_|_]=Bl -> [{block,Bl}|delete_live_annos(Is)] +delete_annos([{block,Bl0}|Is]) -> + case delete_annos(Bl0) of + [] -> delete_annos(Is); + [_|_]=Bl -> [{block,Bl}|delete_annos(Is)] end; -delete_live_annos([{'%live',_,_}|Is]) -> - delete_live_annos(Is); -delete_live_annos([I|Is]) -> - [I|delete_live_annos(Is)]; -delete_live_annos([]) -> []. - +delete_annos([{'%anno',_}|Is]) -> + delete_annos(Is); +delete_annos([I|Is]) -> + [I|delete_annos(Is)]; +delete_annos([]) -> []. + %% combine_heap_needs(HeapNeed1, HeapNeed2) -> HeapNeed %% Combine the heap need for two allocation instructions. --spec combine_heap_needs(term(), term()) -> term(). +-type heap_need_tag() :: 'floats' | 'words'. +-type heap_need() :: non_neg_integer() | + {'alloc',[{heap_need_tag(),non_neg_integer()}]}. +-spec combine_heap_needs(heap_need(), heap_need()) -> heap_need(). -combine_heap_needs({alloc,Alloc1}, {alloc,Alloc2}) -> - {alloc,combine_alloc_lists(Alloc1, Alloc2)}; -combine_heap_needs({alloc,Alloc}, Words) when is_integer(Words) -> - {alloc,combine_alloc_lists(Alloc, [{words,Words}])}; -combine_heap_needs(Words, {alloc,Alloc}) when is_integer(Words) -> - {alloc,combine_alloc_lists(Alloc, [{words,Words}])}; combine_heap_needs(H1, H2) when is_integer(H1), is_integer(H2) -> - H1+H2. + H1 + H2; +combine_heap_needs(H1, H2) -> + {alloc,combine_alloc_lists([H1,H2])}. + + +%% anno_defs(Instructions) -> Instructions' +%% Add {def,RegisterBitmap} annotations to the beginning of +%% each block. Iff bit X is set in the the bitmap, it means +%% that {x,X} is defined when the block is entered. + +-spec anno_defs([instruction()]) -> [instruction()]. + +anno_defs(Is0) -> + {Bef,[Fi|Is1]} = + splitwith(fun({func_info,_,_,_}) -> false; + (_) -> true + end, Is0), + {func_info,_,_,Arity} = Fi, + Regs = init_def_regs(Arity), + Is = defs(Is1, Regs, #{}), + Bef ++ [Fi|Is]. %% split_even/1 %% [1,2,3,4,5,6] -> {[1,3,5],[2,4,6]} @@ -272,7 +339,6 @@ combine_heap_needs(H1, H2) when is_integer(H1), is_integer(H2) -> split_even(Rs) -> split_even(Rs, [], []). - %%% %%% Local functions. %%% @@ -284,12 +350,19 @@ split_even(Rs) -> split_even(Rs, [], []). %% %% killed - Reg is assigned or killed by an allocation instruction. %% not_used - the value of Reg is not used, but Reg must not be garbage +%% exit_not_used - the value of Reg is not used, but must not be garbage +%% because the stack will be scanned because an +%% exit BIF will raise an exception %% used - Reg is used check_liveness(R, [{block,Blk}|Is], St0) -> case check_liveness_block(R, Blk, St0) of {transparent,St1} -> check_liveness(R, Is, St1); + {alloc_used,St1} -> + %% Used by an allocating instruction, but value not referenced. + %% Must check the rest of the instructions. + not_used(check_liveness(R, Is, St1)); {Other,_}=Res when is_atom(Other) -> Res end; @@ -303,6 +376,8 @@ check_liveness(R, [{test,_,{f,Fail},As}|Is], St0) -> case check_liveness_at(R, Fail, St0) of {killed,St1} -> check_liveness(R, Is, St1); + {exit_not_used,St1} -> + check_liveness(R, Is, St1); {not_used,St1} -> not_used(check_liveness(R, Is, St1)); {used,_}=Used -> @@ -322,6 +397,8 @@ check_liveness(R, [{jump,{f,F}}|_], St) -> check_liveness_at(R, F, St); check_liveness(R, [{case_end,Used}|_], St) -> check_liveness_ret(R, Used, St); +check_liveness(R, [{try_case_end,Used}|_], St) -> + check_liveness_ret(R, Used, St); check_liveness(R, [{badmatch,Used}|_], St) -> check_liveness_ret(R, Used, St); check_liveness(_, [if_end|_], St) -> @@ -348,17 +425,27 @@ check_liveness(R, [{bs_init,_,_,none,Ss,Dst}|Is], St) -> check_liveness(R, [{bs_init,_,_,Live,Ss,Dst}|Is], St) -> case R of {x,X} -> - case X < Live orelse member(R, Ss) of - true -> {used,St}; - false -> {killed,St} + case member(R, Ss) of + true -> + {used,St}; + false -> + if + X < Live -> + not_used(check_liveness(R, Is, St)); + true -> + {killed,St} + end end; {y,_} -> case member(R, Ss) of true -> {used,St}; false -> + %% If the exception is taken, the stack may + %% be scanned. Therefore the register is not + %% guaranteed to be killed. if - R =:= Dst -> {killed,St}; - true -> check_liveness(R, Is, St) + R =:= Dst -> {not_used,St}; + true -> not_used(check_liveness(R, Is, St)) end end end; @@ -392,7 +479,7 @@ check_liveness(R, [{call_ext,Live,_}=I|Is], St) -> %% We must make sure we don't check beyond this %% instruction or we will fall through into random %% unrelated code and get stuck in a loop. - {killed,St} + {exit_not_used,St} end end; check_liveness(R, [{call_fun,Live}|Is], St) -> @@ -440,16 +527,12 @@ check_liveness(R, [{make_fun2,_,_,_,NumFree}|Is], St) -> {x,_} -> {killed,St}; {y,_} -> not_used(check_liveness(R, Is, St)) end; -check_liveness({x,_}=R, [{'catch',_,_}|Is], St) -> - %% All x registers will be killed if an exception occurs. - %% Therefore we only need to check the liveness for the - %% instructions following the catch instruction. - check_liveness(R, Is, St); -check_liveness({x,_}=R, [{'try',_,_}|Is], St) -> - %% All x registers will be killed if an exception occurs. - %% Therefore we only need to check the liveness for the - %% instructions inside the 'try' block. - check_liveness(R, Is, St); +check_liveness(R, [{'catch'=Op,Y,Fail}|Is], St) -> + Set = {set,[Y],[],{try_catch,Op,Fail}}, + check_liveness(R, [{block,[Set]}|Is], St); +check_liveness(R, [{'try'=Op,Y,Fail}|Is], St) -> + Set = {set,[Y],[],{try_catch,Op,Fail}}, + check_liveness(R, [{block,[Set]}|Is], St); check_liveness(R, [{try_end,Y}|Is], St) -> case R of Y -> @@ -510,15 +593,34 @@ check_liveness(R, [{get_map_elements,{f,Fail},S,{list,L}}|Is], St0) -> check_liveness(R, [{put_map,F,Op,S,D,Live,{list,Puts}}|Is], St) -> Set = {set,[D],[S|Puts],{alloc,Live,{put_map,Op,F}}}, check_liveness(R, [{block,[Set]}||Is], St); +check_liveness(R, [{put_tuple,Ar,D}|Is], St) -> + Set = {set,[D],[],{put_tuple,Ar}}, + check_liveness(R, [{block,[Set]}||Is], St); +check_liveness(R, [{put_list,S1,S2,D}|Is], St) -> + Set = {set,[D],[S1,S2],put_list}, + check_liveness(R, [{block,[Set]}||Is], St); check_liveness(R, [{test_heap,N,Live}|Is], St) -> I = {block,[{set,[],[],{alloc,Live,{nozero,nostack,N,[]}}}]}, check_liveness(R, [I|Is], St); check_liveness(R, [{allocate_zero,N,Live}|Is], St) -> I = {block,[{set,[],[],{alloc,Live,{zero,N,0,[]}}}]}, check_liveness(R, [I|Is], St); -check_liveness(R, [{get_list,S,D1,D2}|Is], St) -> - I = {block,[{set,[D1,D2],[S],get_list}]}, +check_liveness(R, [{get_hd,S,D}|Is], St) -> + I = {block,[{set,[D],[S],get_hd}]}, check_liveness(R, [I|Is], St); +check_liveness(R, [{get_tl,S,D}|Is], St) -> + I = {block,[{set,[D],[S],get_tl}]}, + check_liveness(R, [I|Is], St); +check_liveness(R, [remove_message|Is], St) -> + check_liveness(R, Is, St); +check_liveness({x,X}, [build_stacktrace|_], St) when X > 0 -> + {killed,St}; +check_liveness(R, [{recv_mark,_}|Is], St) -> + check_liveness(R, Is, St); +check_liveness(R, [{recv_set,_}|Is], St) -> + check_liveness(R, Is, St); +check_liveness(R, [{'%',_}|Is], St) -> + check_liveness(R, Is, St); check_liveness(_R, Is, St) when is_list(Is) -> %% Not implemented. Conservatively assume that the register is used. {used,St}. @@ -553,6 +655,7 @@ check_liveness_at(R, Lbl, #live{lbl=Ll,res=ResMemorized}=St0) -> {Res,St#live{res=gb_trees:insert(Lbl, Res, St#live.res)}} end. +not_used({exit_not_used,St}) -> {not_used,St}; not_used({killed,St}) -> {not_used,St}; not_used({_,_}=Res) -> Res. @@ -560,7 +663,7 @@ check_liveness_ret(R, R, St) -> {used,St}; check_liveness_ret(_, _, St) -> {killed,St}. %% check_liveness_block(Reg, [Instruction], State) -> -%% {killed | not_used | used | transparent,State'} +%% {killed | not_used | used | alloc_used | transparent,State'} %% Finds out how Reg is used in the instruction sequence inside a block. %% Returns one of: %% killed - Reg is assigned a new value or killed by an @@ -568,9 +671,10 @@ check_liveness_ret(_, _, St) -> {killed,St}. %% not_used - The value is not used, but the register is referenced %% e.g. by an allocation instruction %% transparent - Reg is neither used nor killed +%% alloc_used - Used only in an allocate instruction %% used - Reg is explicitly used by an instruction %% -%% '%live' annotations are not allowed. +%% Annotations are not allowed. %% %% (Unknown instructions will cause an exception.) @@ -580,13 +684,30 @@ check_liveness_block({x,X}=R, [{set,Ds,Ss,{alloc,Live,Op}}|Is], St0) -> {killed,St0}; true -> case check_liveness_block_1(R, Ss, Ds, Op, Is, St0) of - {killed,St} -> {not_used,St}; - {transparent,St} -> {not_used,St}; - {_,_}=Res -> Res + {transparent,St} -> {alloc_used,St}; + {_,_}=Res -> not_used(Res) end end; -check_liveness_block({y,_}=R, [{set,Ds,Ss,{alloc,_Live,Op}}|Is], St) -> - check_liveness_block_1(R, Ss, Ds, Op, Is, St); +check_liveness_block({y,_}=R, [{set,Ds,Ss,{alloc,_Live,Op}}|Is], St0) -> + case check_liveness_block_1(R, Ss, Ds, Op, Is, St0) of + {transparent,St} -> {alloc_used,St}; + {_,_}=Res -> not_used(Res) + end; +check_liveness_block({y,_}=R, [{set,Ds,Ss,{try_catch,_,Op}}|Is], St0) -> + case Ds of + [R] -> + {killed,St0}; + _ -> + case check_liveness_block_1(R, Ss, Ds, Op, Is, St0) of + {exit_not_used,St} -> + {used,St}; + {transparent,St} -> + %% Conservatively assumed that it is used. + {used,St}; + {_,_}=Res -> + Res + end + end; check_liveness_block(R, [{set,Ds,Ss,Op}|Is], St) -> check_liveness_block_1(R, Ss, Ds, Op, Is, St); check_liveness_block(_, [], St) -> {transparent,St}. @@ -602,6 +723,11 @@ check_liveness_block_1(R, Ss, Ds, Op, Is, St0) -> true -> {killed,St}; false -> check_liveness_block(R, Is, St) end; + {exit_not_used,St} -> + case member(R, Ds) of + true -> {exit_not_used,St}; + false -> check_liveness_block(R, Is, St) + end; {not_used,St} -> not_used(case member(R, Ds) of true -> {killed,St}; @@ -612,8 +738,8 @@ check_liveness_block_1(R, Ss, Ds, Op, Is, St0) -> end end. -check_liveness_block_2(R, {gc_bif,_Op,{f,Lbl}}, _Ss, St) -> - check_liveness_block_3(R, Lbl, St); +check_liveness_block_2(R, {gc_bif,Op,{f,Lbl}}, Ss, St) -> + check_liveness_block_3(R, Lbl, {Op,length(Ss)}, St); check_liveness_block_2(R, {bif,Op,{f,Lbl}}, Ss, St) -> Arity = length(Ss), case erl_internal:comp_op(Op, Arity) orelse @@ -621,16 +747,23 @@ check_liveness_block_2(R, {bif,Op,{f,Lbl}}, Ss, St) -> true -> {killed,St}; false -> - check_liveness_block_3(R, Lbl, St) + check_liveness_block_3(R, Lbl, {Op,length(Ss)}, St) end; check_liveness_block_2(R, {put_map,_Op,{f,Lbl}}, _Ss, St) -> - check_liveness_block_3(R, Lbl, St); + check_liveness_block_3(R, Lbl, {unsafe,0}, St); check_liveness_block_2(_, _, _, St) -> {killed,St}. -check_liveness_block_3(_, 0, St) -> +check_liveness_block_3({x,_}, 0, _FA, St) -> {killed,St}; -check_liveness_block_3(R, Lbl, St0) -> +check_liveness_block_3({y,_}, 0, {F,A}, St) -> + %% If the exception is thrown, the stack may be scanned, + %% thus implicitly using the y register. + case erl_bifs:is_safe(erlang, F, A) of + true -> {killed,St}; + false -> {used,St} + end; +check_liveness_block_3(R, Lbl, _FA, St0) -> check_liveness_at(R, Lbl, St0). index_labels_1([{label,Lbl}|Is0], Acc) -> @@ -643,22 +776,70 @@ index_labels_1([], Acc) -> gb_trees:from_orddict(sort(Acc)). drop_labels([{label,_}|Is]) -> drop_labels(Is); drop_labels(Is) -> Is. -%% Help functions for combine_heap_needs. -combine_alloc_lists(Al1, Al2) -> - combine_alloc_lists_1(sort(Al1++Al2)). +replace_labels_1([{test,Test,{f,Lbl},Ops}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{test,Test,{f,label(Lbl, D, Fb)},Ops}|Acc], D, Fb); +replace_labels_1([{test,Test,{f,Lbl},Live,Ops,Dst}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{test,Test,{f,label(Lbl, D, Fb)},Live,Ops,Dst}|Acc], D, Fb); +replace_labels_1([{select,I,R,{f,Fail0},Vls0}|Is], Acc, D, Fb) -> + Vls = map(fun ({f,L}) -> {f,label(L, D, Fb)}; + (Other) -> Other + end, Vls0), + Fail = label(Fail0, D, Fb), + replace_labels_1(Is, [{select,I,R,{f,Fail},Vls}|Acc], D, Fb); +replace_labels_1([{'try',R,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{'try',R,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{'catch',R,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{'catch',R,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{jump,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{jump,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{loop_rec,{f,Lbl},R}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{loop_rec,{f,label(Lbl, D, Fb)},R}|Acc], D, Fb); +replace_labels_1([{loop_rec_end,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{loop_rec_end,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{wait,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{wait,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{wait_timeout,{f,Lbl},To}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{wait_timeout,{f,label(Lbl, D, Fb)},To}|Acc], D, Fb); +replace_labels_1([{bif,Name,{f,Lbl},As,R}|Is], Acc, D, Fb) when Lbl =/= 0 -> + replace_labels_1(Is, [{bif,Name,{f,label(Lbl, D, Fb)},As,R}|Acc], D, Fb); +replace_labels_1([{gc_bif,Name,{f,Lbl},Live,As,R}|Is], Acc, D, Fb) when Lbl =/= 0 -> + replace_labels_1(Is, [{gc_bif,Name,{f,label(Lbl, D, Fb)},Live,As,R}|Acc], D, Fb); +replace_labels_1([{call,Ar,{f,Lbl}}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{call,Ar,{f,label(Lbl, D, Fb)}}|Acc], D, Fb); +replace_labels_1([{make_fun2,{f,Lbl},U1,U2,U3}|Is], Acc, D, Fb) -> + replace_labels_1(Is, [{make_fun2,{f,label(Lbl, D, Fb)},U1,U2,U3}|Acc], D, Fb); +replace_labels_1([{bs_init,{f,Lbl},Info,Live,Ss,Dst}|Is], Acc, D, Fb) when Lbl =/= 0 -> + replace_labels_1(Is, [{bs_init,{f,label(Lbl, D, Fb)},Info,Live,Ss,Dst}|Acc], D, Fb); +replace_labels_1([{bs_put,{f,Lbl},Info,Ss}|Is], Acc, D, Fb) when Lbl =/= 0 -> + replace_labels_1(Is, [{bs_put,{f,label(Lbl, D, Fb)},Info,Ss}|Acc], D, Fb); +replace_labels_1([{put_map=I,{f,Lbl},Op,Src,Dst,Live,List}|Is], Acc, D, Fb) + when Lbl =/= 0 -> + replace_labels_1(Is, [{I,{f,label(Lbl, D, Fb)},Op,Src,Dst,Live,List}|Acc], D, Fb); +replace_labels_1([{get_map_elements=I,{f,Lbl},Src,List}|Is], Acc, D, Fb) when Lbl =/= 0 -> + replace_labels_1(Is, [{I,{f,label(Lbl, D, Fb)},Src,List}|Acc], D, Fb); +replace_labels_1([I|Is], Acc, D, Fb) -> + replace_labels_1(Is, [I|Acc], D, Fb); +replace_labels_1([], Acc, _, _) -> Acc. + +label(Old, D, Fb) -> + case D of + #{Old := New} -> New; + _ -> Fb(Old) + end. -combine_alloc_lists_1([{words,W1},{words,W2}|T]) - when is_integer(W1), is_integer(W2) -> - [{words,W1+W2}|combine_alloc_lists_1(T)]; -combine_alloc_lists_1([{floats,F1},{floats,F2}|T]) - when is_integer(F1), is_integer(F2) -> - [{floats,F1+F2}|combine_alloc_lists_1(T)]; -combine_alloc_lists_1([{words,_}=W|T]) -> - [W|combine_alloc_lists_1(T)]; -combine_alloc_lists_1([{floats,_}=F|T]) -> - [F|combine_alloc_lists_1(T)]; -combine_alloc_lists_1([]) -> []. +%% Help function for combine_heap_needs. + +combine_alloc_lists(Al0) -> + Al1 = flatmap(fun(Words) when is_integer(Words) -> + [{words,Words}]; + ({alloc,List}) -> + List + end, Al0), + Al2 = sofs:relation(Al1), + Al3 = sofs:relation_to_family(Al2), + Al4 = sofs:to_external(Al3), + [{Tag,lists:sum(L)} || {Tag,L} <- Al4]. %% live_opt/4. @@ -697,10 +878,14 @@ live_opt([{test,bs_start_match2,Fail,Live,[Src,_],_}=I|Is], _, D, Acc) -> %% Other instructions. live_opt([{block,Bl0}|Is], Regs0, D, Acc) -> - Live0 = {'%live',live_regs(Regs0),Regs0}, + Live0 = make_anno({used,Regs0}), {Bl,Regs} = live_opt_block(reverse(Bl0), Regs0, D, [Live0]), - Live = {'%live',live_regs(Regs),Regs}, + Live = make_anno({used,Regs}), live_opt(Is, Regs, D, [{block,[Live|Bl]}|Acc]); +live_opt([build_stacktrace=I|Is], _, D, Acc) -> + live_opt(Is, live_call(1), D, [I|Acc]); +live_opt([raw_raise=I|Is], _, D, Acc) -> + live_opt(Is, live_call(3), D, [I|Acc]); live_opt([{label,L}=I|Is], Regs, D0, Acc) -> D = gb_trees:insert(L, Regs, D0), live_opt(Is, Regs, D, [I|Acc]); @@ -742,12 +927,19 @@ live_opt([{test,_,Fail,Live,Ss,_}=I|Is], _, D, Acc) -> Regs1 = x_live(Ss, Regs0), Regs = live_join_label(Fail, D, Regs1), live_opt(Is, Regs, D, [I|Acc]); -live_opt([{select,_,Src,Fail,List}=I|Is], Regs0, D, Acc) -> +live_opt([{select,_,Src,Fail,List}=I|Is], _, D, Acc) -> + Regs0 = 0, Regs1 = x_live([Src], Regs0), Regs = live_join_labels([Fail|List], D, Regs1), live_opt(Is, Regs, D, [I|Acc]); -live_opt([{try_case,_}=I|Is], _, D, Acc) -> - live_opt(Is, live_call(1), D, [I|Acc]); +live_opt([{try_case,Y}=I|Is], Regs0, D, Acc) -> + Regs = live_call(1), + case Regs0 of + 0 -> + live_opt(Is, Regs, D, [{try_end,Y}|Acc]); + _ -> + live_opt(Is, live_call(1), D, [I|Acc]) + end; live_opt([{loop_rec,_Fail,_Dst}=I|Is], _, D, Acc) -> live_opt(Is, 0, D, [I|Acc]); live_opt([timeout=I|Is], _, D, Acc) -> @@ -759,6 +951,25 @@ live_opt([{get_map_elements,Fail,Src,{list,List}}=I|Is], Regs0, D, Acc) -> Regs1 = x_live([Src|Ss], x_dead(Ds, Regs0)), Regs = live_join_label(Fail, D, Regs1), live_opt(Is, Regs, D, [I|Acc]); +live_opt([{gc_bif,N,F,R,As,Dst}=I|Is], Regs0, D, Acc) -> + Bl = [{set,[Dst],As,{alloc,R,{gc_bif,N,F}}}], + {_,Regs} = live_opt_block(Bl, Regs0, D, []), + live_opt(Is, Regs, D, [I|Acc]); +live_opt([{bif,N,F,As,Dst}=I|Is], Regs0, D, Acc) -> + Bl = [{set,[Dst],As,{bif,N,F}}], + {_,Regs} = live_opt_block(Bl, Regs0, D, []), + live_opt(Is, Regs, D, [I|Acc]); +live_opt([{get_tuple_element,Src,Idx,Dst}=I|Is], Regs0, D, Acc) -> + Bl = [{set,[Dst],[Src],{get_tuple_element,Idx}}], + {_,Regs} = live_opt_block(Bl, Regs0, D, []), + live_opt(Is, Regs, D, [I|Acc]); +live_opt([{move,Src,Dst}=I|Is], Regs0, D, Acc) -> + Regs = x_live([Src], x_dead([Dst], Regs0)), + live_opt(Is, Regs, D, [I|Acc]); +live_opt([{put_map,F,Op,S,Dst,R,{list,Puts}}=I|Is], Regs0, D, Acc) -> + Bl = [{set,[Dst],[S|Puts],{alloc,R,{put_map,Op,F}}}], + {_,Regs} = live_opt_block(Bl, Regs0, D, []), + live_opt(Is, Regs, D, [I|Acc]); %% Transparent instructions - they neither use nor modify x registers. live_opt([{deallocate,_}=I|Is], Regs, D, Acc) -> @@ -775,6 +986,10 @@ live_opt([{wait_timeout,_,{Tag,_}}=I|Is], Regs, D, Acc) when Tag =/= x -> live_opt(Is, Regs, D, [I|Acc]); live_opt([{line,_}=I|Is], Regs, D, Acc) -> live_opt(Is, Regs, D, [I|Acc]); +live_opt([{'catch',_,_}=I|Is], Regs, D, Acc) -> + live_opt(Is, Regs, D, [I|Acc]); +live_opt([{'try',_,_}=I|Is], Regs, D, Acc) -> + live_opt(Is, Regs, D, [I|Acc]); %% The following instructions can occur if the "compilation" has been %% started from a .S file using the 'from_asm' option. @@ -789,47 +1004,52 @@ live_opt([{recv_mark,_}=I|Is], Regs, D, Acc) -> live_opt([], _, _, Acc) -> Acc. -live_opt_block([{set,Ds,Ss,Op0}|Is], Regs0, D, Acc) -> - Regs1 = x_live(Ss, x_dead(Ds, Regs0)), - {Op, Regs} = live_opt_block_op(Op0, Regs1, D), - I = {set, Ds, Ss, Op}, - - case Ds of - [{x,X}] -> - case (not is_live(X, Regs0)) andalso Op =:= move of - true -> - live_opt_block(Is, Regs0, D, Acc); - false -> - live_opt_block(Is, Regs, D, [I|Acc]) - end; - _ -> - live_opt_block(Is, Regs, D, [I|Acc]) +live_opt_block([{set,[{x,X}]=Ds,Ss,move}=I|Is], Regs0, D, Acc) -> + Regs = x_live(Ss, x_dead(Ds, Regs0)), + case is_live(X, Regs0) of + true -> + live_opt_block(Is, Regs, D, [I|Acc]); + false -> + %% Useless move, will never be used. + live_opt_block(Is, Regs, D, Acc) end; -live_opt_block([{'%live',_,_}|Is], Regs, D, Acc) -> - live_opt_block(Is, Regs, D, Acc); -live_opt_block([], Regs, _, Acc) -> {Acc,Regs}. - -live_opt_block_op({alloc,Live0,AllocOp}, Regs0, D) -> - Regs = - case AllocOp of - {Kind, _N, Fail} when Kind =:= gc_bif; Kind =:= put_map -> - live_join_label(Fail, D, Regs0); - _ -> - Regs0 - end, +live_opt_block([{set,Ds,Ss,{alloc,Live0,AllocOp}}|Is], Regs0, D, Acc) -> + %% Calculate liveness from the point of view of the GC. + %% There will never be a GC if the instruction fails, so we should + %% ignore the failure branch. + GcRegs1 = x_dead(Ds, Regs0), + GcRegs = x_live(Ss, GcRegs1), + Live = live_regs(GcRegs), %% The life-time analysis used by the code generator is sometimes too %% conservative, so it may be possible to lower the number of live %% registers based on the exact liveness information. The main benefit is %% that more optimizations that depend on liveness information (such as the - %% beam_bool and beam_dead passes) may be applied. - Live = live_regs(Regs), - true = Live =< Live0, - {{alloc,Live,AllocOp}, live_call(Live)}; -live_opt_block_op({bif,_N,Fail} = Op, Regs, D) -> - {Op, live_join_label(Fail, D, Regs)}; -live_opt_block_op(Op, Regs, _D) -> - {Op, Regs}. + %% beam_dead pass) may be applied. + true = Live =< Live0, %Assertion. + I = {set,Ds,Ss,{alloc,Live,AllocOp}}, + + %% Calculate liveness from the point of view of the preceding instruction. + %% The liveness is the union of live registers in the GC and the live + %% registers at the failure label. + Regs1 = live_call(Live), + Regs = live_join_alloc(AllocOp, D, Regs1), + live_opt_block(Is, Regs, D, [I|Acc]); +live_opt_block([{set,Ds,Ss,{bif,_,Fail}}=I|Is], Regs0, D, Acc) -> + Regs1 = x_dead(Ds, Regs0), + Regs2 = x_live(Ss, Regs1), + Regs = live_join_label(Fail, D, Regs2), + live_opt_block(Is, Regs, D, [I|Acc]); +live_opt_block([{set,Ds,Ss,_}=I|Is], Regs0, D, Acc) -> + Regs = x_live(Ss, x_dead(Ds, Regs0)), + live_opt_block(Is, Regs, D, [I|Acc]); +live_opt_block([{'%anno',_}|Is], Regs, D, Acc) -> + live_opt_block(Is, Regs, D, Acc); +live_opt_block([], Regs, _, Acc) -> {Acc,Regs}. + +live_join_alloc({Kind,_Name,Fail}, D, Regs) when Kind =:= gc_bif; Kind =:= put_map -> + live_join_label(Fail, D, Regs); +live_join_alloc(_, _, Regs) -> Regs. live_join_labels([{f,L}|T], D, Regs0) when L =/= 0 -> Regs = gb_trees:get(L, D) bor Regs0, @@ -865,3 +1085,220 @@ split_even([], Ss, Ds) -> {reverse(Ss),reverse(Ds)}; split_even([S,D|Rs], Ss, Ds) -> split_even(Rs, [S|Ss], [D|Ds]). + +%%% +%%% Add annotations for defined registers. +%%% +%%% This analysis is done by scanning the instructions in +%%% execution order. +%%% + +defs([{apply,_}=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{bif,_,{f,Fail},_Src,Dst}=I|Is], Regs0, D) -> + Regs = def_regs([Dst], Regs0), + [I|defs(Is, Regs, update_regs(Fail, Regs0, D))]; +defs([{block,Block0}|Is], Regs0, D0) -> + {Block,Regs,D} = defs_list(Block0, Regs0, D0), + [{block,[make_anno({def,Regs0})|Block]}|defs(Is, Regs, D)]; +defs([{bs_init,{f,L},_,_,_,Dst}=I|Is], Regs0, D) -> + Regs = def_regs([Dst], Regs0), + [I|defs(Is, Regs, update_regs(L, Regs, D))]; +defs([{bs_put,{f,L},_,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, update_regs(L, Regs, D))]; +defs([build_stacktrace=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{call,_,_}=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{call_ext,_,{extfunc,M,F,A}}=I|Is], _Regs, D) -> + case erl_bifs:is_exit_bif(M, F, A) of + false -> + [I|defs(Is, 1, D)]; + true -> + [I|defs_unreachable(Is, D)] + end; +defs([{call_ext,_,_}=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{call_fun,_}=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{'catch',_,{f,L}}=I|Is], Regs, D) -> + RegsAtLabel = init_def_regs(1), + [I|defs(Is, Regs, update_regs(L, RegsAtLabel, D))]; +defs([{catch_end,_}=I|Is], _Regs, D) -> + Regs = init_def_regs(1), + [I|defs(Is, Regs, D)]; +defs([{gc_bif,_,{f,Fail},Live,_Src,Dst}=I|Is], Regs0, D) -> + true = all_defined(Live, Regs0), %Assertion. + Regs = def_regs([Dst], init_def_regs(Live)), + [I|defs(Is, Regs, update_regs(Fail, Regs0, D))]; +defs([{get_map_elements,{f,L},_Src,{list,DstList}}=I|Is], Regs0, D) -> + {_,Ds} = beam_utils:split_even(DstList), + Regs = def_regs(Ds, Regs0), + [I|defs(Is, Regs, update_regs(L, Regs0, D))]; +defs([{get_tuple_element,_,_,Dst}=I|Is], Regs0, D) -> + Regs = def_regs([Dst], Regs0), + [I|defs(Is, Regs, D)]; +defs([{jump,{f,L}}=I|Is], Regs, D) -> + [I|defs_unreachable(Is, update_regs(L, Regs, D))]; +defs([{label,L}=I|Is], Regs0, D) -> + case D of + #{L:=Regs1} -> + Regs = Regs0 band Regs1, + [I|defs(Is, Regs, D)]; + #{} -> + [I|defs(Is, Regs0, D)] + end; +defs([{loop_rec,{f,L},{x,0}}=I|Is], _Regs, D0) -> + RegsAtLabel = init_def_regs(0), + D = update_regs(L, RegsAtLabel, D0), + [I|defs(Is, init_def_regs(1), D)]; +defs([{loop_rec_end,_}=I|Is], _Regs, D) -> + [I|defs(Is, 0, D)]; +defs([{make_fun2,_,_,_,_}=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([{move,_,Dst}=I|Is], Regs0, D) -> + Regs = def_regs([Dst], Regs0), + [I|defs(Is, Regs, D)]; +defs([{put_map,{f,Fail},_,_,Dst,_,_}=I|Is], Regs0, D) -> + Regs = def_regs([Dst], Regs0), + [I|defs(Is, Regs, update_regs(Fail, Regs0, D))]; +defs([raw_raise=I|Is], _Regs, D) -> + [I|defs(Is, 1, D)]; +defs([return=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; +defs([{select,_,_Src,Fail,List}=I|Is], Regs, D0) -> + D = update_list([Fail|List], Regs, D0), + [I|defs_unreachable(Is, D)]; +defs([{test,_,{f,L},_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, update_regs(L, Regs, D))]; +defs([{test,_,{f,L},Live,_,Dst}=I|Is], Regs0, D) -> + true = all_defined(Live, Regs0), %Assertion. + Regs = def_regs([Dst], init_def_regs(Live)), + [I|defs(Is, Regs, update_regs(L, Regs0, D))]; +defs([{'try',_,{f,L}}=I|Is], Regs, D) -> + RegsAtLabel = init_def_regs(3), + [I|defs(Is, Regs, update_regs(L, RegsAtLabel, D))]; +defs([{try_case,_}=I|Is], _Regs, D) -> + [I|defs(Is, init_def_regs(3), D)]; +defs([{wait,_}=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; +defs([{wait_timeout,_,_}=I|Is], _Regs, D) -> + [I|defs(Is, 0, D)]; + +%% Exceptions. +defs([{badmatch,_}=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; +defs([{case_end,_}=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; +defs([if_end=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; +defs([{try_case_end,_}=I|Is], _Regs, D) -> + [I|defs_unreachable(Is, D)]; + +%% Neutral instructions +defs([{bs_context_to_binary,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{bs_restore2,_,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{bs_save2,_,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{deallocate,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{kill,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{line,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{recv_mark,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{recv_set,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([timeout=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{trim,_,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{try_end,_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([{'%',_}=I|Is], Regs, D) -> + [I|defs(Is, Regs, D)]; +defs([], _, _) -> []. + +defs_unreachable([{label,L}=I|Is], D) -> + case D of + #{L:=Regs} -> + [I|defs(Is, Regs, D)]; + #{} -> + defs_unreachable(Is, D) + end; +defs_unreachable([_|Is], D) -> + defs_unreachable(Is, D); +defs_unreachable([], _D) -> []. + +defs_list(Is, Regs, D) -> + defs_list(Is, Regs, D, []). + +defs_list([{set,Ds,_,{alloc,Live,Info}}=I|Is], Regs0, D0, Acc) -> + true = all_defined(Live, Regs0), %Assertion. + D = case Info of + {gc_bif,_,{f,Fail}} -> + update_regs(Fail, Regs0, D0); + {put_map,_,{f,Fail}} -> + update_regs(Fail, Regs0, D0); + _ -> + D0 + end, + Regs = def_regs(Ds, init_def_regs(Live)), + defs_list(Is, Regs, D, [I|Acc]); +defs_list([{set,Ds,_,Info}=I|Is], Regs0, D0, Acc) -> + D = case Info of + {bif,_,{f,Fail}} -> + update_regs(Fail, Regs0, D0); + {try_catch,'catch',{f,Fail}} -> + update_regs(Fail, init_def_regs(1), D0); + {try_catch,'try',{f,Fail}} -> + update_regs(Fail, init_def_regs(3), D0); + _ -> + D0 + end, + Regs = def_regs(Ds, Regs0), + defs_list(Is, Regs, D, [I|Acc]); +defs_list([], Regs, D, Acc) -> + {reverse(Acc),Regs,D}. + +init_def_regs(Arity) -> + (1 bsl Arity) - 1. + +def_regs([{x,X}|T], Regs) -> + def_regs(T, Regs bor (1 bsl X)); +def_regs([_|T], Regs) -> + def_regs(T, Regs); +def_regs([], Regs) -> Regs. + +update_list([{f,L}|T], Regs, D0) -> + D = update_regs(L, Regs, D0), + update_list(T, Regs, D); +update_list([_|T], Regs, D) -> + update_list(T, Regs, D); +update_list([], _Regs, D) -> D. + +update_regs(L, Regs0, D) -> + case D of + #{L:=Regs1} -> + Regs = Regs0 band Regs1, + D#{L:=Regs}; + #{} -> + D#{L=>Regs0} + end. + +all_defined(Live, Regs) -> + All = (1 bsl Live) - 1, + Regs band All =:= All. + +%%% +%%% Utilities. +%%% + +%% make_anno(Anno) -> WrappedAnno. +%% Wrap an annotation term. + +make_anno(Anno) -> + {'%anno',Anno}. diff --git a/lib/compiler/src/beam_validator.erl b/lib/compiler/src/beam_validator.erl index ea38969814..c30ab34ac7 100644 --- a/lib/compiler/src/beam_validator.erl +++ b/lib/compiler/src/beam_validator.erl @@ -85,8 +85,6 @@ format_error(Error) -> %%% Things currently not checked. XXX %%% %%% - Heap allocation for binaries. -%%% - That put_tuple is followed by the correct number of -%%% put instructions. %%% %% validate(Module, [Function]) -> [] | [Error] @@ -130,9 +128,8 @@ validate_0(Module, [{function,Name,Ar,Entry,Code}|Fs], Ft) -> throw:Error -> %% Controlled error. [Error|validate_0(Module, Fs, Ft)]; - Class:Error -> + Class:Error:Stack -> %% Crash. - Stack = erlang:get_stacktrace(), io:fwrite("Function: ~w/~w\n", [Name,Ar]), erlang:raise(Class, Error, Stack) end. @@ -149,7 +146,8 @@ validate_0(Module, [{function,Name,Ar,Entry,Code}|Fs], Ft) -> hf=0, %Available heap size for floats. fls=undefined, %Floating point state. ct=[], %List of hot catch/try labels - setelem=false %Previous instruction was setelement/3. + setelem=false, %Previous instruction was setelement/3. + puts_left=none %put/1 instructions left. }). -type label() :: integer(). @@ -294,6 +292,8 @@ valfun_1({bs_context_to_binary,Ctx}, #vst{current=#st{x=Xs}}=Vst) -> end; valfun_1(bs_init_writable=I, Vst) -> call(I, 1, Vst); +valfun_1(build_stacktrace=I, Vst) -> + call(I, 1, Vst); valfun_1({move,{y,_}=Src,{y,_}=Dst}, Vst) -> %% The stack trimming optimization may generate a move from an initialized %% but unassigned Y register to another Y register. @@ -339,11 +339,25 @@ valfun_1({put_list,A,B,Dst}, Vst0) -> Vst = eat_heap(2, Vst0), set_type_reg(cons, Dst, Vst); valfun_1({put_tuple,Sz,Dst}, Vst0) when is_integer(Sz) -> + Vst1 = eat_heap(1, Vst0), + Vst = set_type_reg(tuple_in_progress, Dst, Vst1), + #vst{current=St0} = Vst, + St = St0#st{puts_left={Sz,{Dst,{tuple,Sz}}}}, + Vst#vst{current=St}; +valfun_1({put,Src}, Vst0) -> + assert_term(Src, Vst0), Vst = eat_heap(1, Vst0), - set_type_reg({tuple,Sz}, Dst, Vst); -valfun_1({put,Src}, Vst) -> - assert_term(Src, Vst), - eat_heap(1, Vst); + #vst{current=St0} = Vst, + case St0 of + #st{puts_left=none} -> + error(not_building_a_tuple); + #st{puts_left={1,{Dst,Type}}} -> + St = St0#st{puts_left=none}, + set_type_reg(Type, Dst, Vst#vst{current=St}); + #st{puts_left={PutsLeft,Info}} when is_integer(PutsLeft) -> + St = St0#st{puts_left={PutsLeft-1,Info}}, + Vst#vst{current=St} + end; %% Instructions for optimization of selective receives. valfun_1({recv_mark,{f,Fail}}, Vst) when is_integer(Fail) -> Vst; @@ -523,6 +537,8 @@ valfun_4({bif,element,{f,Fail},[Pos,Tuple],Dst}, Vst0) -> valfun_4({bif,raise,{f,0},Src,_Dst}, Vst) -> validate_src(Src, Vst), kill_state(Vst); +valfun_4(raw_raise=I, Vst) -> + call(I, 3, Vst); valfun_4({bif,Op,{f,Fail},Src,Dst}, Vst0) -> validate_src(Src, Vst0), Vst = branch_state(Fail, Vst0), @@ -575,6 +591,12 @@ valfun_4({get_list,Src,D1,D2}, Vst0) -> assert_type(cons, Src, Vst0), Vst = set_type_reg(term, D1, Vst0), set_type_reg(term, D2, Vst); +valfun_4({get_hd,Src,Dst}, Vst) -> + assert_type(cons, Src, Vst), + set_type_reg(term, Dst, Vst); +valfun_4({get_tl,Src,Dst}, Vst) -> + assert_type(cons, Src, Vst), + set_type_reg(term, Dst, Vst); valfun_4({get_tuple_element,Src,I,Dst}, Vst) -> assert_type({tuple_element,I+1}, Src, Vst), set_type_reg(term, Dst, Vst); @@ -1131,6 +1153,7 @@ set_type_y(Type, {y,Y}=Reg, #vst{current=#st{y=Ys0}=St}=Vst) {value,_} -> gb_trees:update(Y, Type, Ys0) end, + check_try_catch_tags(Type, Y, Ys0), Vst#vst{current=St#st{y=Ys}}; set_type_y(Type, Reg, #vst{}) -> error({invalid_store,Reg,Type}). @@ -1138,6 +1161,29 @@ set_catch_end({y,Y}, #vst{current=#st{y=Ys0}=St}=Vst) -> Ys = gb_trees:update(Y, initialized, Ys0), Vst#vst{current=St#st{y=Ys}}. +check_try_catch_tags(Type, LastY, Ys) -> + case is_try_catch_tag(Type) of + false -> + ok; + true -> + %% Every catch or try/catch must use a lower Y register + %% number than any enclosing catch or try/catch. That will + %% ensure that when the stack is scanned when an + %% exception occurs, the innermost try/catch tag is found + %% first. + Bad = [{{y,Y},Tag} || {Y,Tag} <- gb_trees:to_list(Ys), + Y < LastY, is_try_catch_tag(Tag)], + case Bad of + [] -> + ok; + [_|_] -> + error({bad_try_catch_nesting,{y,LastY},Bad}) + end + end. + +is_try_catch_tag({catchtag,_}) -> true; +is_try_catch_tag({trytag,_}) -> true; +is_try_catch_tag(_) -> false. is_reg_defined({x,_}=Reg, Vst) -> is_type_defined_x(Reg, Vst); is_reg_defined({y,_}=Reg, Vst) -> is_type_defined_y(Reg, Vst); @@ -1271,6 +1317,7 @@ get_move_term_type(Src, Vst) -> initialized -> error({unassigned,Src}); {catchtag,_} -> error({catchtag,Src}); {trytag,_} -> error({trytag,Src}); + tuple_in_progress -> error({tuple_in_progress,Src}); Type -> Type end. @@ -1279,10 +1326,7 @@ get_move_term_type(Src, Vst) -> %% a standard Erlang type (no catch/try tags or match contexts). get_term_type(Src, Vst) -> - case get_term_type_1(Src, Vst) of - initialized -> error({unassigned,Src}); - {catchtag,_} -> error({catchtag,Src}); - {trytag,_} -> error({trytag,Src}); + case get_move_term_type(Src, Vst) of #ms{} -> error({match_context,Src}); Type -> Type end. diff --git a/lib/compiler/src/beam_z.erl b/lib/compiler/src/beam_z.erl index 787e33c142..6c3a6995d7 100644 --- a/lib/compiler/src/beam_z.erl +++ b/lib/compiler/src/beam_z.erl @@ -24,22 +24,23 @@ -export([module/2]). --import(lists, [dropwhile/2]). +-import(lists, [dropwhile/2,map/2]). -spec module(beam_utils:module_code(), [compile:option()]) -> {'ok',beam_asm:module_code()}. -module({Mod,Exp,Attr,Fs0,Lc}, _Opt) -> - Fs = [function(F) || F <- Fs0], +module({Mod,Exp,Attr,Fs0,Lc}, Opts) -> + NoGetHdTl = proplists:get_bool(no_get_hd_tl, Opts), + Fs = [function(F, NoGetHdTl) || F <- Fs0], {ok,{Mod,Exp,Attr,Fs,Lc}}. -function({function,Name,Arity,CLabel,Is0}) -> +function({function,Name,Arity,CLabel,Is0}, NoGetHdTl) -> try - Is = undo_renames(Is0), + Is1 = undo_renames(Is0), + Is = maybe_eliminate_get_hd_tl(Is1, NoGetHdTl), {function,Name,Arity,CLabel,Is} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. @@ -66,6 +67,10 @@ undo_renames([{bif,raise,_,_,_}=I|Is0]) -> (_) -> true end, Is0), [I|undo_renames(Is)]; +undo_renames([{get_hd,Src,Dst1},{get_tl,Src,Dst2}|Is]) -> + [{get_list,Src,Dst1,Dst2}|undo_renames(Is)]; +undo_renames([{get_tl,Src,Dst2},{get_hd,Src,Dst1}|Is]) -> + [{get_list,Src,Dst1,Dst2}|undo_renames(Is)]; undo_renames([I|Is]) -> [undo_rename(I)|undo_renames(Is)]; undo_renames([]) -> []. @@ -108,3 +113,17 @@ undo_rename({get_map_elements,Fail,Src,{list,List}}) -> undo_rename({select,I,Reg,Fail,List}) -> {I,Reg,Fail,{list,List}}; undo_rename(I) -> I. + +%%% +%%% Eliminate get_hd/get_tl instructions if requested by +%%% the no_get_hd_tl option. +%%% + +maybe_eliminate_get_hd_tl(Is, true) -> + map(fun({get_hd,Cons,Hd}) -> + {get_list,Cons,Hd,{x,1022}}; + ({get_tl,Cons,Tl}) -> + {get_list,Cons,{x,1022},Tl}; + (I) -> I + end, Is); +maybe_eliminate_get_hd_tl(Is, false) -> Is. diff --git a/lib/compiler/src/cerl_clauses.erl b/lib/compiler/src/cerl_clauses.erl index 7d6518c3c6..fa5104c01b 100644 --- a/lib/compiler/src/cerl_clauses.erl +++ b/lib/compiler/src/cerl_clauses.erl @@ -353,6 +353,8 @@ match(P, E, Bs) -> map -> %% The most we can do is to say "definitely no match" if a %% map pattern is matched against non-map data. + %% (Note: See the document internal_doc/cerl-notes.md for + %% information why we don't try to do more here.) case E of any -> {false, Bs}; diff --git a/lib/compiler/src/compile.erl b/lib/compiler/src/compile.erl index 50b0ba76f8..c6a0056a70 100644 --- a/lib/compiler/src/compile.erl +++ b/lib/compiler/src/compile.erl @@ -203,7 +203,12 @@ expand_opts(Opts0) -> {_,_,undefined} -> [debug_info|Opts0]; {_,_,_} -> Opts0 end, - foldr(fun expand_opt/2, [], Opts). + %% iff,unless processing is to complex... + Opts1 = case proplists:is_defined(makedep_side_effect,Opts) of + true -> proplists:delete(makedep,Opts); + false -> Opts + end, + foldr(fun expand_opt/2, [], Opts1). expand_opt(basic_validation, Os) -> [no_code_generation,to_pp,binary|Os]; @@ -214,13 +219,15 @@ expand_opt(report, Os) -> expand_opt(return, Os) -> [return_errors,return_warnings|Os]; expand_opt(r16, Os) -> - [no_record_opt,no_utf8_atoms|Os]; + [no_get_hd_tl,no_record_opt,no_utf8_atoms|Os]; expand_opt(r17, Os) -> - [no_record_opt,no_utf8_atoms|Os]; + [no_get_hd_tl,no_record_opt,no_utf8_atoms|Os]; expand_opt(r18, Os) -> - [no_record_opt,no_utf8_atoms|Os]; + [no_get_hd_tl,no_record_opt,no_utf8_atoms|Os]; expand_opt(r19, Os) -> - [no_record_opt,no_utf8_atoms|Os]; + [no_get_hd_tl,no_record_opt,no_utf8_atoms|Os]; +expand_opt(r20, Os) -> + [no_get_hd_tl,no_record_opt,no_utf8_atoms|Os]; expand_opt({debug_info_key,_}=O, Os) -> [encrypt_debug_info,O|Os]; expand_opt(no_float_opt, Os) -> @@ -674,6 +681,7 @@ select_list_passes_1([], _, Acc) -> standard_passes() -> [?pass(transform_module), + {iff,makedep_side_effect,?pass(makedep_and_output)}, {iff,makedep,[ ?pass(makedep), {unless,binary,?pass(makedep_output)} @@ -706,14 +714,16 @@ core_passes() -> [{unless,no_copt, [{core_old_inliner,fun test_old_inliner/1,fun core_old_inliner/2}, {iff,doldinline,{listing,"oldinline"}}, - {pass,sys_core_fold}, + {unless,no_fold,{pass,sys_core_fold}}, {iff,dcorefold,{listing,"corefold"}}, {core_inline_module,fun test_core_inliner/1,fun core_inline_module/2}, {iff,dinline,{listing,"inline"}}, {core_fold_after_inlining,fun test_any_inliner/1, fun core_fold_module_after_inlining/2}, + {iff,dcopt,{listing,"copt"}}, + {unless,no_alias,{pass,sys_core_alias}}, + {iff,dalias,{listing,"core_alias"}}, ?pass(core_transforms)]}, - {iff,dcopt,{listing,"copt"}}, {iff,'to_core',{done,"core"}}]} | kernel_passes()]. @@ -731,8 +741,6 @@ kernel_passes() -> ?pass(v3_kernel), {iff,dkern,{listing,"kernel"}}, {iff,'to_kernel',{done,"kernel"}}, - {pass,v3_life}, - {iff,dlife,{listing,"life"}}, {pass,v3_codegen}, {iff,dcg,{listing,"codegen"}} | asm_passes()]. @@ -769,6 +777,8 @@ asm_passes() -> {iff,drecv,{listing,"recv"}}, {unless,no_record_opt,{pass,beam_record}}, {iff,drecord,{listing,"record"}}, + {unless,no_blk2,?pass(block2)}, + {iff,dblk2,{listing,"block2"}}, {unless,no_stack_trimming,{pass,beam_trim}}, {iff,dtrim,{listing,"trim"}}, {pass,beam_flatten}]}, @@ -787,7 +797,7 @@ asm_passes() -> | binary_passes()]. binary_passes() -> - [{iff,'to_dis',{listing,"dis"}}, + [{iff,'to_dis',?pass(to_dis)}, {native_compile,fun test_native/1,fun native_compile/2}, {unless,binary,?pass(save_binary,not_werror)} ]. @@ -1128,6 +1138,16 @@ core_lint_module(Code, St) -> errors=St#compile.errors ++ Es}} end. +%% makedep + output and continue +makedep_and_output(Code0, St) -> + {ok,DepCode,St1} = makedep(Code0,St), + case makedep_output(DepCode, St1) of + {ok,_IgnoreCode,St2} -> + {ok,Code0,St2}; + {error,St2} -> + {error,St2} + end. + makedep(Code0, #compile{ifile=Ifile,ofile=Ofile,options=Opts}=St) -> %% Get the target of the Makefile rule. @@ -1334,6 +1354,10 @@ v3_kernel(Code0, #compile{options=Opts,warnings=Ws0}=St) -> {ok,Code,St} end. +block2(Code0, #compile{options=Opts}=St) -> + {ok,Code} = beam_block:module(Code0, [no_blockify|Opts]), + {ok,Code,St}. + test_old_inliner(#compile{options=Opts}) -> %% The point of this test is to avoid loading the old inliner %% if we know that it will not be used. @@ -1448,15 +1472,33 @@ save_core_code(Code, St) -> beam_asm(Code0, #compile{ifile=File,extra_chunks=ExtraChunks,options=CompilerOpts}=St) -> case debug_info(St) of {ok,DebugInfo,Opts0} -> - Source = paranoid_absname(File), Opts1 = [O || O <- Opts0, effects_code_generation(O)], Chunks = [{<<"Dbgi">>, DebugInfo} | ExtraChunks], - {ok,Code} = beam_asm:module(Code0, Chunks, Source, Opts1, CompilerOpts), + CompileInfo = compile_info(File, CompilerOpts, Opts1), + {ok,Code} = beam_asm:module(Code0, Chunks, CompileInfo, CompilerOpts), {ok,Code,St#compile{abstract_code=[]}}; {error,Es} -> {error,St#compile{errors=St#compile.errors ++ [{File,Es}]}} end. +compile_info(File, CompilerOpts, Opts) -> + IsSlim = member(slim, CompilerOpts), + IsDeterministic = member(deterministic, CompilerOpts), + Info0 = proplists:get_value(compile_info, Opts, []), + Info1 = + case paranoid_absname(File) of + [_|_] = Source when not IsSlim, not IsDeterministic -> + [{source,Source} | Info0]; + _ -> + Info0 + end, + Info2 = + case IsDeterministic of + false -> [{options,proplists:delete(compile_info, Opts)} | Info1]; + true -> Info1 + end, + Info2. + paranoid_absname(""=File) -> File; paranoid_absname(File) -> @@ -1512,15 +1554,14 @@ native_compile_1(Code, St) -> {error,St#compile{errors=St#compile.errors ++ Es}} end catch - Class:R -> - Stk = erlang:get_stacktrace(), + Class:R:Stack -> case IgnoreErrors of true -> Ws = [{St#compile.ifile, - [{none,?MODULE,{native_crash,R,Stk}}]}], + [{none,?MODULE,{native_crash,R,Stack}}]}], {ok,St#compile{warnings=St#compile.warnings ++ Ws}}; false -> - erlang:raise(Class, R, Stk) + erlang:raise(Class, R, Stack) end end. @@ -1748,6 +1789,21 @@ listing(LFun, Ext, Code, St) -> {error,St#compile{errors=St#compile.errors ++ Es}} end. +to_dis(Code, #compile{module=Module,ofile=Outfile}=St) -> + Loaded = code:is_loaded(Module), + Sticky = code:is_sticky(Module), + _ = [code:unstick_mod(Module) || Sticky], + + {module,Module} = code:load_binary(Module, "", Code), + DestDir = filename:dirname(Outfile), + DisFile = filename:join(DestDir, atom_to_list(Module) ++ ".dis"), + ok = erts_debug:dis_to_file(Module, DisFile), + + %% Restore loaded module + _ = [{module, Module} = code:load_file(Module) || Loaded =/= false], + [code:stick_mod(Module) || Sticky], + {ok,Code,St}. + output_encoding(F, #compile{encoding = none}) -> ok = io:setopts(F, [{encoding, epp:default_encoding()}]); output_encoding(F, #compile{encoding = Encoding}) -> @@ -1923,12 +1979,12 @@ pre_load() -> erl_lint, erl_parse, erl_scan, + sys_core_alias, sys_core_bsm, sys_core_dsetel, sys_core_fold, v3_codegen, v3_core, - v3_kernel, - v3_life], + v3_kernel], _ = code:ensure_modules_loaded(L), ok. diff --git a/lib/compiler/src/compiler.app.src b/lib/compiler/src/compiler.app.src index 3139d68902..cf32fd251c 100644 --- a/lib/compiler/src/compiler.app.src +++ b/lib/compiler/src/compiler.app.src @@ -58,6 +58,7 @@ core_lib, erl_bifs, rec_env, + sys_core_alias, sys_core_bsm, sys_core_dsetel, sys_core_fold, @@ -67,8 +68,7 @@ v3_codegen, v3_core, v3_kernel, - v3_kernel_pp, - v3_life + v3_kernel_pp ]}, {registered, []}, {applications, [kernel, stdlib]}, diff --git a/lib/compiler/src/core_lint.erl b/lib/compiler/src/core_lint.erl index 7d3513c0ba..6e2114be56 100644 --- a/lib/compiler/src/core_lint.erl +++ b/lib/compiler/src/core_lint.erl @@ -353,12 +353,6 @@ expr(#c_case{arg=Arg,clauses=Cs}, Def, Rt, St0) -> Pc = case_patcount(Cs), St1 = body(Arg, Def, Pc, St0), clauses(Cs, Def, Pc, Rt, St1); -expr(#c_receive{clauses=Cs,timeout=#c_literal{val=infinity}, - action=#c_literal{}}, - Def, Rt, St) -> - %% If the timeout is 'infinity', the after code can never - %% be reached. We don't care if the return count is wrong. - clauses(Cs, Def, 1, Rt, St); expr(#c_receive{clauses=Cs,timeout=T,action=A}, Def, Rt, St0) -> St1 = expr(T, Def, 1, St0), St2 = body(A, Def, Rt, St1), diff --git a/lib/compiler/src/core_pp.erl b/lib/compiler/src/core_pp.erl index cff6c7098b..2516a9a1e1 100644 --- a/lib/compiler/src/core_pp.erl +++ b/lib/compiler/src/core_pp.erl @@ -1,7 +1,7 @@ %% %% %CopyrightBegin% %% -%% Copyright Ericsson AB 1999-2016. All Rights Reserved. +%% Copyright Ericsson AB 1999-2017. All Rights Reserved. %% %% Licensed under the Apache License, Version 2.0 (the "License"); %% you may not use this file except in compliance with the License. @@ -464,7 +464,7 @@ indent(#ctxt{indent=N}) -> N =< 0 -> ""; true -> - string:chars($\t, N div ?TAB_WIDTH, spaces(N rem ?TAB_WIDTH)) + lists:duplicate(N div ?TAB_WIDTH, $\t) ++ spaces(N rem ?TAB_WIDTH) end. nl_indent(Ctxt) -> [$\n|indent(Ctxt)]. diff --git a/lib/compiler/src/core_scan.erl b/lib/compiler/src/core_scan.erl index 9f0676538f..a50a2ffa8d 100644 --- a/lib/compiler/src/core_scan.erl +++ b/lib/compiler/src/core_scan.erl @@ -200,8 +200,8 @@ pre_string(eof, Q, _, Sp, SoFar, Pos) -> pre_string_error(Q, Sp, SoFar, Pos). pre_string_error(Q, Sp, SoFar, Pos) -> - S = reverse(string:substr(SoFar, 1, string:chr(SoFar, Q)-1)), - pre_error({string,Q,string:substr(S, 1, 16)}, Sp, Pos). + [S,_] = string:split(SoFar, [Q]), + pre_error({string,Q,string:slice(string:reverse(S), 0, 16)}, Sp, Pos). pre_char([C|Cs], SoFar) -> pre_char(C, Cs, SoFar); pre_char([], _) -> more; diff --git a/lib/compiler/src/genop.tab b/lib/compiler/src/genop.tab index b5688de339..a47d4e8cf7 100755 --- a/lib/compiler/src/genop.tab +++ b/lib/compiler/src/genop.tab @@ -545,3 +545,32 @@ BEAM_FORMAT_NUMBER=0 ## Test the arity of Reg and jumps to Lbl if it is not N. ## Test the first element of the tuple and jumps to Lbl if it is not Atom. 159: is_tagged_tuple/4 + +# OTP 21 + +## @spec build_stacktrace +## @doc Given the raw stacktrace in x(0), build a cooked stacktrace suitable +## for human consumption. Store it in x(0). Destroys all other registers. +## Do a garbage collection if necessary to allocate space on the heap +## for the result. +160: build_stacktrace/0 + +## @spec raw_raise +## @doc This instruction works like the erlang:raise/3 BIF, except that the +## stacktrace in x(2) must be a raw stacktrace. +## x(0) is the class of the exception (error, exit, or throw), +## x(1) is the exception term, and x(2) is the raw stackframe. +## If x(0) is not a valid class, the instruction will not throw an +## exception, but store the atom 'badarg' in x(0) and execute the +## next instruction. +161: raw_raise/0 + +## @spec get_hd Source Head +## @doc Get the head (or car) part of a list (a cons cell) from Source and +## put it into the register Head. +162: get_hd/2 + +## @spec get_tl Source Tail +## @doc Get the tail (or cdr) part of a list (a cons cell) from Source and +## put it into the register Tail. +163: get_tl/2 diff --git a/lib/compiler/src/sys_core_alias.erl b/lib/compiler/src/sys_core_alias.erl new file mode 100644 index 0000000000..1bce1577d1 --- /dev/null +++ b/lib/compiler/src/sys_core_alias.erl @@ -0,0 +1,307 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 1999-2016. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +%% Purpose : Replace values by aliases from patterns optimisation for Core + +%% Replace expressions by aliases from patterns. For example: +%% +%% example({ok, Val}) -> +%% {ok, Val}. +%% +%% will become: +%% +%% example({ok, Val} = Tuple) -> +%% Tuple. +%% +%% Currently this pass aliases tuple and cons nodes made of literals, +%% variables and other cons. The tuple/cons may appear anywhere in the +%% pattern and it will be aliased if used later on. +%% +%% Notice a tuple/cons made only of literals is not aliased as it may +%% be part of the literal pool. + +-module(sys_core_alias). + +-export([module/2]). + +-include("core_parse.hrl"). + +-define(NOTSET, 0). + +-record(sub, {p=#{} :: #{term() => ?NOTSET | atom()}, %% Found pattern substitutions + v=cerl_sets:new() :: cerl_sets:set(cerl:var_name()), %% Variables used by patterns + t=undefined :: term()}). %% Temporary information from pre to post + +-type sub() :: #sub{}. + +-spec module(cerl:c_module(), [compile:option()]) -> + {'ok',cerl:c_module(),[]}. + +module(#c_module{defs=Ds0}=Mod, _Opts) -> + Ds1 = [def(D) || D <- Ds0], + {ok,Mod#c_module{defs=Ds1},[]}. + +def({#c_var{name={F,Arity}}=Name,B0}) -> + try + put(new_var_num, 0), + {B1,_} = cerl_trees:mapfold(fun pre/2, fun post/2, sub_new(undefined), B0), + erase(new_var_num), + {Name,B1} + catch + Class:Error:Stack -> + io:fwrite("Function: ~w/~w\n", [F,Arity]), + erlang:raise(Class, Error, Stack) + end. + +pre(#c_let{vars=Vars}=Node, Sub) -> + {Node,sub_fold(get_variables(Vars), Sub)}; + +pre(#c_fun{vars=Vars}=Node, Sub) -> + {Node,sub_fold(get_variables(Vars), Sub)}; + +pre(#c_clause{pats=Pats}=Node, Sub0) -> + VarNames = get_variables(Pats), + Sub1 = sub_fold(VarNames, Sub0), + Keys = get_pattern_keys(Pats), + Sub2 = sub_add_keys(Keys, Sub1), + + #sub{v=SubNames,t=Temp} = Sub2, + Sub3 = Sub2#sub{v=merge_variables(VarNames, SubNames), + t={clause,Pats,Keys,SubNames,Temp}}, + + {Node#c_clause{pats=[]},Sub3}; + +pre(Node, Sub0) -> + %% We cache only tuples and cons. + case cerl:is_data(Node) andalso not cerl:is_literal(Node) of + false -> + {Node,Sub0}; + true -> + Kind = cerl:data_type(Node), + Es = cerl:data_es(Node), + case sub_cache_nodes(Kind, Es, Sub0) of + {Name,Sub1} -> + {cerl:ann_c_var(cerl:get_ann(Node), Name),Sub1}; + error -> + {Node,Sub0} + end + end. + +post(#c_let{}=Node, Sub) -> + {Node,sub_unfold(Sub)}; + +post(#c_fun{}=Node, Sub) -> + {Node,sub_unfold(Sub)}; + +post(#c_clause{}=Node, #sub{t={clause,Pats0,Keys,V,T}}=Sub0) -> + {Sub1,PostKeys} = sub_take_keys(Keys, Sub0), + Pats1 = put_pattern_keys(Pats0, PostKeys), + Sub2 = sub_unfold(Sub1#sub{v=V,t=T}), + {Node#c_clause{pats=Pats1},Sub2}; + +post(Node, Sub) -> + {Node,Sub}. + +%% sub_new/1 +%% sub_add_keys/2 +%% sub_take_keys/3 +%% sub_cache_nodes/3 +%% +%% Manages the substitutions record. + +%% Builds a new sub. +-spec sub_new(term()) -> sub(). +sub_new(Temp) -> + #sub{t=Temp}. + +%% Folds the sub into a new one if the variables in nodes are not disjoint +sub_fold(VarNames, #sub{v=SubNames}=Sub) -> + case is_disjoint_variables(VarNames, SubNames) of + true -> Sub#sub{t={temp,Sub#sub.t}}; + false -> sub_new({sub,Sub}) + end. + +%% Unfolds the sub in case one was folded in the previous step +sub_unfold(#sub{t={temp,Temp}}=Sub) -> + Sub#sub{t=Temp}; +sub_unfold(#sub{t={sub,Sub}}) -> + Sub. + +%% Adds the keys extracted from patterns to the state. +-spec sub_add_keys([term()], sub()) -> sub(). +sub_add_keys(Keys, #sub{p=Pat0}=Sub) -> + Pat1 = + lists:foldl(fun(Key, Acc) -> + false = maps:is_key(Key, Acc), %Assertion. + maps:put(Key, ?NOTSET, Acc) + end, Pat0, Keys), + Sub#sub{p=Pat1}. + +%% Take the keys from the map taking into account the keys +%% that have changed as those must become aliases in the pattern. +-spec sub_take_keys([term()], sub()) -> {sub(), [{term(), atom()}]}. +sub_take_keys(Keys, #sub{p=Pat0}=Sub) -> + {Pat1,Acc} = sub_take_keys(Keys, Pat0, []), + {Sub#sub{p=Pat1},Acc}. + +sub_take_keys([K|T], Sub0, Acc) -> + case maps:take(K, Sub0) of + {?NOTSET,Sub1} -> + sub_take_keys(T, Sub1, Acc); + {Name,Sub1} -> + sub_take_keys(T, Sub1, [{K,Name}|Acc]) + end; +sub_take_keys([], Sub, Acc) -> + {Sub,Acc}. + +%% Check if the node can be cached based on the state information. +%% If it can be cached and it does not have an alias for it, we +%% build one. +-spec sub_cache_nodes(atom(), [cerl:cerl()], sub()) -> {atom(), sub()} | error. +sub_cache_nodes(Kind, Nodes, #sub{p=Pat}=Sub) -> + case nodes_to_key(Kind, Nodes) of + {ok, Key} -> + case Pat of + #{Key := ?NOTSET} -> + new_var_name(Key, Sub); + #{Key := Name} -> + {Name,Sub}; + #{} -> + error + end; + error -> + error + end. + +new_var_name(Key, #sub{p=Pat}=Sub) -> + Counter = get(new_var_num), + Name = list_to_atom("@r" ++ integer_to_list(Counter)), + put(new_var_num, Counter + 1), + {Name,Sub#sub{p=maps:put(Key, Name, Pat)}}. + +%% get_variables/1 +%% is_disjoint_variables/2 +%% merge_variables/2 + +get_variables(NodesList) -> + cerl_sets:from_list([Var || Node <- NodesList, Var <- cerl_trees:variables(Node)]). + +is_disjoint_variables(Vars1, Vars2) -> + cerl_sets:is_disjoint(Vars1, Vars2). + +merge_variables(Vars1, Vars2) -> + cerl_sets:union(Vars1, Vars2). + +%% get_pattern_keys/2 +%% put_pattern_keys/2 +%% +%% Gets keys from patterns or add them as aliases. + +get_pattern_keys(Patterns) -> + lists:foldl(fun get_pattern_keys/2, [], Patterns). + +get_pattern_keys(#c_tuple{es=Es}, Acc0) -> + Acc1 = accumulate_pattern_keys(tuple, Es, Acc0), + lists:foldl(fun get_pattern_keys/2, Acc1, Es); +get_pattern_keys(#c_cons{hd=Hd,tl=Tl}, Acc0) -> + Acc1 = accumulate_pattern_keys(cons, [Hd, Tl], Acc0), + get_pattern_keys(Tl, get_pattern_keys(Hd, Acc1)); +get_pattern_keys(#c_alias{pat=Pat}, Acc0) -> + get_pattern_keys(Pat, Acc0); +get_pattern_keys(#c_map{es=Es}, Acc0) -> + lists:foldl(fun get_pattern_keys/2, Acc0, Es); +get_pattern_keys(#c_map_pair{val=Val}, Acc0) -> + get_pattern_keys(Val, Acc0); +get_pattern_keys(_, Acc) -> + Acc. + +accumulate_pattern_keys(Kind, Nodes, Acc) -> + case nodes_to_key(Kind, Nodes) of + {ok,Key} -> [Key|Acc]; + error -> Acc + end. + +put_pattern_keys(Patterns, []) -> + Patterns; +put_pattern_keys(Patterns, Keys) -> + {NewPatterns,Map} = + lists:mapfoldl(fun alias_pattern_keys/2, maps:from_list(Keys), Patterns), + %% Check all aliases have been consumed from the map. + 0 = map_size(Map), + NewPatterns. + +alias_pattern_keys(#c_tuple{anno=Anno,es=Es0}=Node, Acc0) -> + {Es1,Acc1} = lists:mapfoldl(fun alias_pattern_keys/2, Acc0, Es0), + nodes_to_alias(tuple, Es0, Anno, Node#c_tuple{es=Es1}, Acc1); +alias_pattern_keys(#c_cons{anno=Anno,hd=Hd0,tl=Tl0}=Node, Acc0) -> + {Hd1,Acc1} = alias_pattern_keys(Hd0, Acc0), + {Tl1,Acc2} = alias_pattern_keys(Tl0, Acc1), + nodes_to_alias(cons, [Hd0, Tl0], Anno, Node#c_cons{hd=Hd1,tl=Tl1}, Acc2); +alias_pattern_keys(#c_alias{pat=Pat0}=Node, Acc0) -> + {Pat1,Acc1} = alias_pattern_keys(Pat0, Acc0), + {Node#c_alias{pat=Pat1}, Acc1}; +alias_pattern_keys(#c_map{es=Es0}=Node, Acc0) -> + {Es1,Acc1} = lists:mapfoldl(fun alias_pattern_keys/2, Acc0, Es0), + {Node#c_map{es=Es1}, Acc1}; +alias_pattern_keys(#c_map_pair{val=Val0}=Node, Acc0) -> + {Val1,Acc1} = alias_pattern_keys(Val0, Acc0), + {Node#c_map_pair{val=Val1}, Acc1}; +alias_pattern_keys(Pattern, Acc) -> + {Pattern,Acc}. + +%% Check if a node must become an alias because +%% its pattern was used later on as an expression. +nodes_to_alias(Kind, Inner, Anno, Node, Keys0) -> + case nodes_to_key(Kind, Inner) of + {ok,Key} -> + case maps:take(Key, Keys0) of + {Name,Keys1} -> + Var = cerl:ann_c_var(Anno, Name), + {cerl:ann_c_alias(Anno, Var, Node), Keys1}; + error -> + {Node,Keys0} + end; + error -> + {Node,Keys0} + end. + +%% Builds the key used to check if a value can be +%% replaced by an alias. It considers literals, +%% aliases, variables, tuples and cons recursively. +nodes_to_key(Kind, Nodes) -> + nodes_to_key(Nodes, [], Kind). + +nodes_to_key([#c_alias{var=Var}|T], Acc, Kind) -> + nodes_to_key([Var|T], Acc, Kind); +nodes_to_key([#c_var{name=Name}|T], Acc, Kind) -> + nodes_to_key(T, [[var,Name]|Acc], Kind); +nodes_to_key([Node|T], Acc0, Kind) -> + case cerl:is_data(Node) of + false -> + error; + true -> + case nodes_to_key(cerl:data_es(Node), [], cerl:data_type(Node)) of + {ok,Key} -> + nodes_to_key(T, [Key|Acc0], Kind); + error -> + error + end + end; +nodes_to_key([], Acc, Kind) -> + {ok,[Kind|Acc]}. diff --git a/lib/compiler/src/sys_core_bsm.erl b/lib/compiler/src/sys_core_bsm.erl index 3e04cc33df..65580f79e3 100644 --- a/lib/compiler/src/sys_core_bsm.erl +++ b/lib/compiler/src/sys_core_bsm.erl @@ -24,7 +24,7 @@ -export([module/2,format_error/1]). -include("core_parse.hrl"). --import(lists, [member/2,nth/2,reverse/1,usort/1]). +-import(lists, [member/2,reverse/1,usort/1]). -spec module(cerl:c_module(), [compile:option()]) -> {'ok', cerl:c_module()}. @@ -44,8 +44,7 @@ function([{#c_var{name={F,Arity}}=Name,B0}|Fs], FsAcc, Ws0) -> {B,Ws} -> function(Fs, [{Name,B}|FsAcc], Ws) catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [F,Arity]), erlang:raise(Class, Error, Stack) end; @@ -60,13 +59,6 @@ format_error(bin_opt_alias) -> format_error(bin_partition) -> "INFO: matching non-variables after a previous clause matching a variable " "will prevent delayed sub binary optimization"; -format_error(bin_left_var_used_in_guard) -> - "INFO: a variable to the left of the binary pattern is used in a guard; " - "will prevent delayed sub binary optimization"; -format_error(bin_argument_order) -> - "INFO: matching anything else but a plain variable to the left of " - "binary pattern will prevent delayed sub binary optimization; " - "SUGGEST changing argument order"; format_error(bin_var_used) -> "INFO: using a matched out sub binary will prevent " "delayed sub binary optimization"; @@ -97,46 +89,41 @@ bsm_an(#c_case{arg=#c_values{es=Es}}=Case) -> bsm_an(Other) -> {ok,Other}. -bsm_an_1(Vs, #c_case{clauses=Cs}=Case) -> - case bsm_leftmost(Cs) of - none -> {ok,Case}; - Pos -> bsm_an_2(Vs, Cs, Case, Pos) - end. - -bsm_an_2(Vs, Cs, Case, Pos) -> - case bsm_nonempty(Cs, Pos) of - true -> bsm_an_3(Vs, Cs, Case, Pos); - false -> {ok,Case} +bsm_an_1(Vs0, #c_case{clauses=Cs0}=Case) -> + case bsm_leftmost(Cs0) of + none -> + {ok,Case}; + 1 -> + bsm_an_2(Vs0, Cs0, Case); + Pos -> + Vs = move_from_col(Pos, Vs0), + Cs = [C#c_clause{pats=move_from_col(Pos, Ps)} || + #c_clause{pats=Ps}=C <- Cs0], + bsm_an_2(Vs, Cs, Case) end. -bsm_an_3(Vs, Cs, Case, Pos) -> +bsm_an_2(Vs, Cs, Case) -> try - bsm_ensure_no_partition(Cs, Pos), - {ok,bsm_do_an(Vs, Pos, Cs, Case)} + bsm_ensure_no_partition(Cs), + {ok,bsm_do_an(Vs, Cs, Case)} catch - throw:{problem,Where,What} -> - {ok,Case,{Where,What}} + throw:{problem,Where,What} -> + {ok,Case,{Where,What}} end. -bsm_do_an(Vs0, Pos, Cs0, Case) -> - case nth(Pos, Vs0) of - #c_var{name=Vname}=V0 -> - Cs = bsm_do_an_var(Vname, Pos, Cs0, []), - V = bsm_annotate_for_reuse(V0), - Bef = lists:sublist(Vs0, Pos-1), - Aft = lists:nthtail(Pos, Vs0), - case Bef ++ [V|Aft] of - [_] -> - Case#c_case{arg=V,clauses=Cs}; - Vs -> - Case#c_case{arg=#c_values{es=Vs},clauses=Cs} - end; - _ -> - Case - end. +move_from_col(Pos, L) -> + {First,[Col|Rest]} = lists:split(Pos - 1, L), + [Col|First] ++ Rest. -bsm_do_an_var(V, S, [#c_clause{pats=Ps,guard=G,body=B0}=C0|Cs], Acc) -> - case nth(S, Ps) of +bsm_do_an([#c_var{name=Vname}=V0|Vs0], Cs0, Case) -> + Cs = bsm_do_an_var(Vname, Cs0), + V = bsm_annotate_for_reuse(V0), + Vs = core_lib:make_values([V|Vs0]), + Case#c_case{arg=Vs,clauses=Cs}; +bsm_do_an(_Vs, _Cs, Case) -> Case. + +bsm_do_an_var(V, [#c_clause{pats=[P|_],guard=G,body=B0}=C0|Cs]) -> + case P of #c_var{name=VarName} -> case core_lib:is_var_used(V, G) of true -> bsm_problem(C0, orig_bin_var_used_in_guard); @@ -149,23 +136,23 @@ bsm_do_an_var(V, S, [#c_clause{pats=Ps,guard=G,body=B0}=C0|Cs], Acc) -> B1 = bsm_maybe_ctx_to_binary(VarName, B0), B = bsm_maybe_ctx_to_binary(V, B1), C = C0#c_clause{body=B}, - bsm_do_an_var(V, S, Cs, [C|Acc]); - #c_alias{}=P -> + [C|bsm_do_an_var(V, Cs)]; + #c_alias{} -> case bsm_could_match_binary(P) of false -> - bsm_do_an_var(V, S, Cs, [C0|Acc]); + [C0|bsm_do_an_var(V, Cs)]; true -> bsm_problem(C0, bin_opt_alias) end; - P -> + _ -> case bsm_could_match_binary(P) andalso bsm_is_var_used(V, G, B0) of false -> - bsm_do_an_var(V, S, Cs, [C0|Acc]); + [C0|bsm_do_an_var(V, Cs)]; true -> bsm_problem(C0, bin_var_used) end end; -bsm_do_an_var(_, _, [], Acc) -> reverse(Acc). +bsm_do_an_var(_, []) -> []. bsm_annotate_for_reuse(#c_var{anno=Anno}=Var) -> Var#c_var{anno=[reuse_for_context|Anno]}. @@ -193,131 +180,82 @@ previous_ctx_to_binary(V, Core) -> end. %% bsm_leftmost(Cs) -> none | ArgumentNumber -%% Find the leftmost argument that does binary matching. Return -%% the number of the argument (1-N). +%% Find the leftmost argument that matches a nonempty binary. +%% Return either 'none' or the argument number (1-N). bsm_leftmost(Cs) -> bsm_leftmost_1(Cs, none). +bsm_leftmost_1([_|_], 1) -> + 1; bsm_leftmost_1([#c_clause{pats=Ps}|Cs], Pos) -> bsm_leftmost_2(Ps, Cs, 1, Pos); bsm_leftmost_1([], Pos) -> Pos. bsm_leftmost_2(_, Cs, Pos, Pos) -> bsm_leftmost_1(Cs, Pos); -bsm_leftmost_2([#c_binary{}|_], Cs, N, _) -> +bsm_leftmost_2([#c_binary{segments=[_|_]}|_], Cs, N, _) -> bsm_leftmost_1(Cs, N); bsm_leftmost_2([_|Ps], Cs, N, Pos) -> bsm_leftmost_2(Ps, Cs, N+1, Pos); bsm_leftmost_2([], Cs, _, Pos) -> bsm_leftmost_1(Cs, Pos). -%% bsm_nonempty(Cs, Pos) -> true|false -%% Check if at least one of the clauses matches a non-empty -%% binary in the given argument position. +%% bsm_ensure_no_partition(Cs) -> ok (exception if problem) +%% There must only be a single bs_start_match2 instruction if we +%% are to reuse the binary variable for the match context. +%% +%% To make sure that there is only a single bs_start_match2 +%% instruction, we will check for partitions such as: %% -bsm_nonempty([#c_clause{pats=Ps}|Cs], Pos) -> - case nth(Pos, Ps) of - #c_binary{segments=[_|_]} -> - true; - _ -> - bsm_nonempty(Cs, Pos) - end; -bsm_nonempty([], _ ) -> false. - -%% bsm_ensure_no_partition(Cs, Pos) -> ok (exception if problem) -%% We must make sure that matching is not partitioned between -%% variables like this: %% foo(<<...>>) -> ... %% foo(<Variable>) when ... -> ... -%% foo(<Any non-variable pattern>) -> -%% If there is such partition, we are not allowed to reuse the binary variable -%% for the match context. +%% foo(<Non-variable pattern>) -> %% -%% Also, arguments to the left of the argument that is matched -%% against a binary, are only allowed to be simple variables, not -%% used in guards. The reason is that we must know that the binary is -%% only matched in one place (i.e. there must be only one bs_start_match2 -%% instruction emitted). +%% If there is such partition, we reject the optimization. -bsm_ensure_no_partition(Cs, Pos) -> - bsm_ensure_no_partition_1(Cs, Pos, before). +bsm_ensure_no_partition(Cs) -> + bsm_ensure_no_partition_1(Cs, before). %% Loop through each clause. -bsm_ensure_no_partition_1([#c_clause{pats=Ps,guard=G}|Cs], Pos, State0) -> - State = bsm_ensure_no_partition_2(Ps, Pos, G, simple_vars, State0), +bsm_ensure_no_partition_1([#c_clause{pats=Ps,guard=G}|Cs], State0) -> + State = bsm_ensure_no_partition_2(Ps, G, State0), case State of 'after' -> - bsm_ensure_no_partition_after(Cs, Pos); + bsm_ensure_no_partition_after(Cs); _ -> ok end, - bsm_ensure_no_partition_1(Cs, Pos, State); -bsm_ensure_no_partition_1([], _, _) -> ok. + bsm_ensure_no_partition_1(Cs, State); +bsm_ensure_no_partition_1([], _) -> ok. -%% Loop through each pattern for this clause. -bsm_ensure_no_partition_2([#c_binary{}=Where|_], 1, _, Vstate, State) -> - case State of - before when Vstate =:= simple_vars -> within; - before -> bsm_problem(Where, Vstate); - within when Vstate =:= simple_vars -> within; - within -> bsm_problem(Where, Vstate) - end; -bsm_ensure_no_partition_2([#c_alias{}=Alias|_], 1, N, Vstate, State) -> +bsm_ensure_no_partition_2([#c_binary{}|_], _, _State) -> + within; +bsm_ensure_no_partition_2([#c_alias{}=Alias|_], N, State) -> %% Retrieve the real pattern that the alias refers to and check that. P = bsm_real_pattern(Alias), - bsm_ensure_no_partition_2([P], 1, N, Vstate, State); -bsm_ensure_no_partition_2([_|_], 1, _, _Vstate, before=State) -> + bsm_ensure_no_partition_2([P], N, State); +bsm_ensure_no_partition_2([_|_], _, before=State) -> %% No binary matching yet - therefore no partition. State; -bsm_ensure_no_partition_2([P|_], 1, _, Vstate, State) -> +bsm_ensure_no_partition_2([P|_], _, State) -> case bsm_could_match_binary(P) of false -> - %% If clauses can be freely arranged (Vstate =:= simple_vars), - %% a clause that cannot match a binary will not partition the clause. - %% Example: - %% - %% a(Var, <<>>) -> ... - %% a(Var, []) -> ... - %% a(Var, <<B>>) -> ... - %% - %% But if the clauses can't be freely rearranged, as in - %% - %% b(Var, <<X>>) -> ... - %% b(1, 2) -> ... - %% - %% we do have a problem. - %% - case Vstate of - simple_vars -> State; - _ -> bsm_problem(P, Vstate) - end; + State; true -> %% The pattern P *may* match a binary, so we must update the state. %% (P must be a variable.) - case State of - within -> 'after'; - 'after' -> 'after' - end - end; -bsm_ensure_no_partition_2([#c_var{name=V}|Ps], N, G, Vstate, S) -> - case core_lib:is_var_used(V, G) of - false -> - bsm_ensure_no_partition_2(Ps, N-1, G, Vstate, S); - true -> - bsm_ensure_no_partition_2(Ps, N-1, G, bin_left_var_used_in_guard, S) - end; -bsm_ensure_no_partition_2([_|Ps], N, G, _, S) -> - bsm_ensure_no_partition_2(Ps, N-1, G, bin_argument_order, S). + 'after' + end. -bsm_ensure_no_partition_after([#c_clause{pats=Ps}=C|Cs], Pos) -> - case nth(Pos, Ps) of - #c_var{} -> - bsm_ensure_no_partition_after(Cs, Pos); - _ -> - bsm_problem(C, bin_partition) +bsm_ensure_no_partition_after([#c_clause{pats=Ps}=C|Cs]) -> + case Ps of + [#c_var{}|_] -> + bsm_ensure_no_partition_after(Cs); + _ -> + bsm_problem(C, bin_partition) end; -bsm_ensure_no_partition_after([], _) -> ok. +bsm_ensure_no_partition_after([]) -> ok. bsm_could_match_binary(#c_alias{pat=P}) -> bsm_could_match_binary(P); bsm_could_match_binary(#c_cons{}) -> false; diff --git a/lib/compiler/src/sys_core_dsetel.erl b/lib/compiler/src/sys_core_dsetel.erl index bd3eeae238..9e2df69b33 100644 --- a/lib/compiler/src/sys_core_dsetel.erl +++ b/lib/compiler/src/sys_core_dsetel.erl @@ -81,8 +81,7 @@ visit_module_1([{Name,F0}|Fs], Env, Acc) -> {F,_} -> visit_module_1(Fs, Env, [{Name,F}|Acc]) catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> #c_var{name={Func,Arity}} = Name, io:fwrite("Function: ~w/~w\n", [Func,Arity]), erlang:raise(Class, Error, Stack) diff --git a/lib/compiler/src/sys_core_fold.erl b/lib/compiler/src/sys_core_fold.erl index d73060fb7e..a9bd363ee1 100644 --- a/lib/compiler/src/sys_core_fold.erl +++ b/lib/compiler/src/sys_core_fold.erl @@ -125,8 +125,7 @@ function_1({#c_var{name={F,Arity}}=Name,B0}) -> end, B0, 20), {Name,B} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [F,Arity]), erlang:raise(Class, Error, Stack) end. @@ -146,14 +145,9 @@ find_fixpoint(OptFun, Core0, Max) -> body(Body, Sub) -> body(Body, value, Sub). -body(#c_values{anno=A,es=Es0}, Ctxt, Sub) -> - Es1 = expr_list(Es0, Ctxt, Sub), - case Ctxt of - value -> - #c_values{anno=A,es=Es1}; - effect -> - make_effect_seq(Es1, Sub) - end; +body(#c_values{anno=A,es=Es0}, value, Sub) -> + Es1 = expr_list(Es0, value, Sub), + #c_values{anno=A,es=Es1}; body(E, Ctxt, Sub) -> ?ASSERT(verify_scope(E, Sub)), expr(E, Ctxt, Sub). @@ -314,9 +308,15 @@ expr(#c_seq{arg=Arg0,body=B0}=Seq0, Ctxt, Sub) -> false -> %% Arg cannot be "values" here - only a single value %% make sense here. - case is_safe_simple(Arg, Sub) of - true -> B1; - false -> Seq0#c_seq{arg=Arg,body=B1} + case {Ctxt,is_safe_simple(Arg, Sub)} of + {effect,true} -> B1; + {effect,false} -> + case is_safe_simple(B1, Sub) of + true -> Arg; + false -> Seq0#c_seq{arg=Arg,body=B1} + end; + {value,true} -> B1; + {value,false} -> Seq0#c_seq{arg=Arg,body=B1} end end; expr(#c_let{}=Let0, Ctxt, Sub) -> @@ -380,10 +380,7 @@ expr(#c_case{}=Case0, Ctxt, Sub) -> Case = Case1#c_case{arg=Arg2,clauses=Cs2}, warn_no_clause_match(Case1, Case), Expr = eval_case(Case, Sub), - case move_case_into_arg(Case, Sub) of - impossible -> Expr; - Other -> Other - end; + move_case_into_arg(Expr, Sub); Other -> expr(Other, Ctxt, Sub) end; @@ -415,9 +412,20 @@ expr(#c_call{module=M0,name=N0}=Call0, Ctxt, Sub) -> no -> call(Call, M1, N1, Sub); {yes,Seq} -> expr(Seq, Ctxt, Sub) end; +expr(#c_primop{name=#c_literal{val=build_stacktrace}}, effect, _Sub) -> + void(); expr(#c_primop{args=As0}=Prim, _, Sub) -> As1 = expr_list(As0, value, Sub), Prim#c_primop{args=As1}; +expr(#c_catch{anno=Anno,body=B}, effect, Sub) -> + %% When the return value of the 'catch' is ignored, we can replace it + %% with a try/catch to avoid building a stack trace when an exception + %% occurs. + Var = #c_var{name='catch_value'}, + Evs = [#c_var{name='Class'},#c_var{name='Reason'},#c_var{name='Stk'}], + Try = #c_try{anno=Anno,arg=B,vars=[Var],body=Var, + evars=Evs,handler=void()}, + expr(Try, effect, Sub); expr(#c_catch{body=B0}=Catch, _, Sub) -> %% We can remove catch if the value is simple B1 = body(B0, value, Sub), @@ -2422,16 +2430,10 @@ move_let_into_expr(#c_let{vars=InnerVs0,body=InnerBody0}=Inner, Outer#c_let{vars=OuterVs,arg=Arg, body=Inner#c_let{vars=InnerVs,arg=OuterBody,body=InnerBody}}; move_let_into_expr(#c_let{vars=Lvs0,body=Lbody0}=Let, - #c_case{arg=Cexpr0,clauses=[Ca0,Cb0|Cs]}=Case, Sub0) -> - %% Test if there are no more clauses than Ca0 and Cb0, or if - %% Cb0 is guaranteed to match. - TwoClauses = Cs =:= [] orelse - case Cb0 of - #c_clause{pats=[#c_var{}],guard=#c_literal{val=true}} -> true; - _ -> false - end, - case {TwoClauses,is_failing_clause(Ca0),is_failing_clause(Cb0)} of - {true,false,true} -> + #c_case{arg=Cexpr0,clauses=[Ca0|Cs0]}=Case, Sub0) -> + case not is_failing_clause(Ca0) andalso + are_all_failing_clauses(Cs0) of + true -> %% let <Lvars> = case <Case-expr> of %% <Cpats> -> <Clause-body>; %% <OtherCpats> -> erlang:error(...) @@ -2467,8 +2469,8 @@ move_let_into_expr(#c_let{vars=Lvs0,body=Lbody0}=Let, body=Lbody}, Ca = Ca0#c_clause{pats=CaPats,guard=G,body=B}, - Cb = clause(Cb0, Cexpr, value, Sub0), - Case#c_case{arg=Cexpr,clauses=[Ca,Cb]} + Cs = [clause(C, Cexpr, value, Sub0) || C <- Cs0], + Case#c_case{arg=Cexpr,clauses=[Ca|Cs]} catch nomatch -> %% This is not a defeat. The code will eventually @@ -2476,7 +2478,7 @@ move_let_into_expr(#c_let{vars=Lvs0,body=Lbody0}=Let, %% optimizations done in this module. impossible end; - {_,_,_} -> impossible + false -> impossible end; move_let_into_expr(#c_let{vars=Lvs0,body=Lbody0}=Let, #c_seq{arg=Sarg0,body=Sbody0}=Seq, Sub0) -> @@ -2499,9 +2501,78 @@ move_let_into_expr(#c_let{vars=Lvs0,body=Lbody0}=Let, body=Lbody}}; move_let_into_expr(_Let, _Expr, _Sub) -> impossible. +are_all_failing_clauses(Cs) -> + all(fun is_failing_clause/1, Cs). + is_failing_clause(#c_clause{body=B}) -> will_fail(B). +%% opt_build_stacktrace(Let) -> Core. +%% If the stacktrace is *only* used in a call to erlang:raise/3, +%% there is no need to build a cooked stackframe using build_stacktrace/1. + +opt_build_stacktrace(#c_let{vars=[#c_var{name=Cooked}], + arg=#c_primop{name=#c_literal{val=build_stacktrace}, + args=[RawStk]}, + body=Body}=Let) -> + case Body of + #c_call{module=#c_literal{val=erlang}, + name=#c_literal{val=raise}, + args=[Class,Exp,#c_var{name=Cooked}]} -> + %% The stacktrace is only used in a call to erlang:raise/3. + %% There is no need to build the stacktrace. Replace the + %% call to erlang:raise/3 with the the raw_raise/3 instruction, + %% which will use a raw stacktrace. + #c_primop{name=#c_literal{val=raw_raise}, + args=[Class,Exp,RawStk]}; + #c_let{vars=[#c_var{name=V}],arg=Arg,body=B0} when V =/= Cooked -> + case core_lib:is_var_used(Cooked, Arg) of + false -> + %% The built stacktrace is not used in the argument, + %% so we can sink the building of the stacktrace into + %% the body of the let. + B = opt_build_stacktrace(Let#c_let{body=B0}), + Body#c_let{body=B}; + true -> + Let + end; + #c_seq{arg=Arg,body=B0} -> + case core_lib:is_var_used(Cooked, Arg) of + false -> + %% The built stacktrace is not used in the argument, + %% so we can sink the building of the stacktrace into + %% the body of the sequence. + B = opt_build_stacktrace(Let#c_let{body=B0}), + Body#c_seq{body=B}; + true -> + Let + end; + #c_case{arg=Arg,clauses=Cs0} -> + case core_lib:is_var_used(Cooked, Arg) orelse + is_used_in_any_guard(Cooked, Cs0) of + false -> + %% The built stacktrace is not used in the argument, + %% so we can sink the building of the stacktrace into + %% each arm of the case. + Cs = [begin + B = opt_build_stacktrace(Let#c_let{body=B0}), + C#c_clause{body=B} + end || #c_clause{body=B0}=C <- Cs0], + Body#c_case{clauses=Cs}; + true -> + Let + end; + _ -> + Let + end; +opt_build_stacktrace(Expr) -> + Expr. + +is_used_in_any_guard(V, Cs) -> + any(fun(#c_clause{guard=G}) -> + core_lib:is_var_used(V, G) + end, Cs). + %% opt_case_in_let(Let) -> Let' %% Try to avoid building tuples that are immediately matched. %% A common pattern is: @@ -2616,9 +2687,13 @@ delay_build_expr_1(#c_receive{clauses=Cs0, timeout=Timeout, action=A0}=Rec, TypeSig) -> Cs = delay_build_cs(Cs0, TypeSig), - A = case Timeout of - #c_literal{val=infinity} -> A0; - _ -> delay_build_expr(A0, TypeSig) + A = case {Timeout,A0} of + {#c_literal{val=infinity},#c_literal{}} -> + {_Type,Arity} = TypeSig, + Es = lists:duplicate(Arity, A0), + core_lib:make_values(Es); + _ -> + delay_build_expr(A0, TypeSig) end, Rec#c_receive{clauses=Cs,action=A}; delay_build_expr_1(#c_seq{body=B0}=Seq, TypeSig) -> @@ -2653,53 +2728,94 @@ opt_simple_let_1(#c_let{vars=Vs0,body=B0}=Let, Arg0, Ctxt, Sub0) -> %% Optimise let and add new substitutions. {Vs,Args,Sub1} = let_substs(Vs0, Arg0, Sub0), BodySub = update_let_types(Vs, Args, Sub1), + Sub = Sub1#sub{v=[],s=cerl_sets:new()}, B = body(B0, Ctxt, BodySub), Arg = core_lib:make_values(Args), - opt_simple_let_2(Let, Vs, Arg, B, B0, Ctxt, Sub1). + opt_simple_let_2(Let, Vs, Arg, B, B0, Sub). + + +%% opt_simple_let_2(Let0, Vs0, Arg0, Body, PrevBody, Ctxt, Sub) -> Core. +%% Do final simplifications of the let. +%% +%% Note that the substitutions and scope in Sub have been cleared +%% and should not be used. -opt_simple_let_2(Let0, Vs0, Arg0, Body, PrevBody, Ctxt, Sub) -> +opt_simple_let_2(Let0, Vs0, Arg0, Body, PrevBody, Sub) -> case {Vs0,Arg0,Body} of - {[#c_var{name=N1}],Arg1,#c_var{name=N2}} -> - case N1 =:= N2 of - true -> - %% let <Var> = Arg in <Var> ==> Arg - Arg1; - false -> - %% let <Var> = Arg in <OtherVar> ==> seq Arg OtherVar - Arg = maybe_suppress_warnings(Arg1, Vs0, PrevBody), - #c_seq{arg=Arg,body=Body} - end; + {[#c_var{name=V}],Arg1,#c_var{name=V}} -> + %% let <Var> = Arg in <Var> ==> Arg + Arg1; {[],#c_values{es=[]},_} -> %% No variables left. Body; - {Vs,Arg1,#c_literal{}} -> - Arg = maybe_suppress_warnings(Arg1, Vs, PrevBody), - case Ctxt of - effect -> - %% Throw away the literal body. - Arg; - value -> - %% Since the variable is not used in the body, we - %% can rewrite the let to a sequence. - %% let <Var> = Arg in Literal ==> seq Arg Literal - #c_seq{arg=Arg,body=Body} - end; - {Vs,Arg1,Body} -> - %% If none of the variables are used in the body, we can - %% rewrite the let to a sequence: - %% let <Var> = Arg in BodyWithoutVar ==> - %% seq Arg BodyWithoutVar - case is_any_var_used(Vs, Body) of - false -> - Arg = maybe_suppress_warnings(Arg1, Vs, PrevBody), - #c_seq{arg=Arg,body=Body}; - true -> - Let1 = Let0#c_let{vars=Vs,arg=Arg1,body=Body}, - opt_bool_case_in_let(Let1, Sub) + {[#c_var{name=V}=Var|Vars]=Vars0,Arg1,Body} -> + case core_lib:is_var_used(V, Body) of + false when Vars =:= [] -> + %% If the variable is not used in the body, we can + %% rewrite the let to a sequence: + %% let <Var> = Arg in BodyWithoutVar ==> + %% seq Arg BodyWithoutVar + Arg = maybe_suppress_warnings(Arg1, Var, PrevBody), + #c_seq{arg=Arg,body=Body}; + false -> + %% There are multiple values returned by the argument + %% and the first value is not used (this is a 'case' + %% with exported variables, but the return value is + %% ignored). We can remove the first variable and the + %% the first value returned from the 'let' argument. + Arg2 = remove_first_value(Arg1, Sub), + Let1 = Let0#c_let{vars=Vars,arg=Arg2,body=Body}, + post_opt_let(Let1, Sub); + true -> + Let1 = Let0#c_let{vars=Vars0,arg=Arg1,body=Body}, + post_opt_let(Let1, Sub) end end. -%% maybe_suppress_warnings(Arg, [#c_var{}], PreviousBody) -> Arg' +%% post_opt_let(Let, Sub) +%% Final optimizations of the let. +%% +%% Note that the substitutions and scope in Sub have been cleared +%% and should not be used. + +post_opt_let(Let0, Sub) -> + Let1 = opt_bool_case_in_let(Let0, Sub), + opt_build_stacktrace(Let1). + + +%% remove_first_value(Core0, Sub) -> Core. +%% Core0 is an expression that returns at least two values. +%% Remove the first value returned from Core0. + +remove_first_value(#c_values{es=[V|Vs]}, Sub) -> + Values = core_lib:make_values(Vs), + case is_safe_simple(V, Sub) of + false -> + #c_seq{arg=V,body=Values}; + true -> + Values + end; +remove_first_value(#c_case{clauses=Cs0}=Core, Sub) -> + Cs = remove_first_value_cs(Cs0, Sub), + Core#c_case{clauses=Cs}; +remove_first_value(#c_receive{clauses=Cs0,action=Act0}=Core, Sub) -> + Cs = remove_first_value_cs(Cs0, Sub), + Act = remove_first_value(Act0, Sub), + Core#c_receive{clauses=Cs,action=Act}; +remove_first_value(#c_let{body=B}=Core, Sub) -> + Core#c_let{body=remove_first_value(B, Sub)}; +remove_first_value(#c_seq{body=B}=Core, Sub) -> + Core#c_seq{body=remove_first_value(B, Sub)}; +remove_first_value(#c_primop{}=Core, _Sub) -> + Core; +remove_first_value(#c_call{}=Core, _Sub) -> + Core. + +remove_first_value_cs(Cs, Sub) -> + [C#c_clause{body=remove_first_value(B, Sub)} || + #c_clause{body=B}=C <- Cs]. + +%% maybe_suppress_warnings(Arg, #c_var{}, PreviousBody) -> Arg' %% Try to suppress false warnings when a variable is not used. %% For instance, we don't expect a warning for useless building in: %% @@ -2710,12 +2826,12 @@ opt_simple_let_2(Let0, Vs0, Arg0, Body, PrevBody, Ctxt, Sub) -> %% referenced in the original unoptimized code. If they were, we will %% consider the warning false and suppress it. -maybe_suppress_warnings(Arg, Vs, PrevBody) -> +maybe_suppress_warnings(Arg, #c_var{name=V}, PrevBody) -> case should_suppress_warning(Arg) of true -> Arg; %Already suppressed. false -> - case is_any_var_used(Vs, PrevBody) of + case core_lib:is_var_used(V, PrevBody) of true -> suppress_warning([Arg]); false -> @@ -2804,7 +2920,7 @@ move_case_into_arg(#c_case{arg=#c_case{arg=OuterArg, Outer#c_case{arg=OuterArg, clauses=[OuterCa,OuterCb]}; false -> - impossible + Inner0 end; move_case_into_arg(#c_case{arg=#c_seq{arg=OuterArg,body=InnerArg}=Outer, clauses=InnerClauses}=Inner, _Sub) -> @@ -2820,15 +2936,8 @@ move_case_into_arg(#c_case{arg=#c_seq{arg=OuterArg,body=InnerArg}=Outer, %% Outer#c_seq{arg=OuterArg, body=Inner#c_case{arg=InnerArg,clauses=InnerClauses}}; -move_case_into_arg(_, _) -> - impossible. - -is_any_var_used([#c_var{name=V}|Vs], Expr) -> - case core_lib:is_var_used(V, Expr) of - false -> is_any_var_used(Vs, Expr); - true -> true - end; -is_any_var_used([], _) -> false. +move_case_into_arg(Expr, _) -> + Expr. %%% %%% Retrieving information about types. diff --git a/lib/compiler/src/v3_codegen.erl b/lib/compiler/src/v3_codegen.erl index 47c1567f10..a8f4926e55 100644 --- a/lib/compiler/src/v3_codegen.erl +++ b/lib/compiler/src/v3_codegen.erl @@ -19,25 +19,6 @@ %% %% Purpose : Code generator for Beam. -%% The following assumptions have been made: -%% -%% 1. Matches, i.e. things with {match,M,Ret} wrappers, only return -%% values; no variables are exported. If the match would have returned -%% extra variables then these have been transformed to multiple return -%% values. -%% -%% 2. All BIF's called in guards are gc-safe so there is no need to -%% put thing on the stack in the guard. While this would in principle -%% work it would be difficult to keep track of the stack depth when -%% trimming. -%% -%% The code generation uses variable lifetime information added by -%% the v3_life module to save variables, allocate registers and -%% move registers to the stack when necessary. -%% -%% We try to use a consistent variable name scheme throughout. The -%% StackReg record is always called Bef,Int<n>,Aft. - -module(v3_codegen). %% The main interface. @@ -45,12 +26,14 @@ -import(lists, [member/2,keymember/3,keysort/2,keydelete/3, append/1,flatmap/2,filter/2,foldl/3,foldr/3,mapfoldl/3, - sort/1,reverse/1,reverse/2]). --import(v3_life, [vdb_find/2]). + sort/1,reverse/1,reverse/2,map/2]). +-import(ordsets, [add_element/2,intersection/2,union/2]). -%%-compile([export_all]). +-include("v3_kernel.hrl"). --include("v3_life.hrl"). +%% These are not defined in v3_kernel.hrl. +get_kanno(Kthing) -> element(2, Kthing). +set_kanno(Kthing, Anno) -> setelement(2, Kthing, Anno). %% Main codegen structure. -record(cg, {lcount=1, %Label counter @@ -61,38 +44,400 @@ functable=#{}, %Map of local functions: {Name,Arity}=>Label in_catch=false, %Inside a catch or not. need_frame, %Need a stack frame. - ultimate_failure %Label for ultimate match failure. - }). + ultimate_failure, %Label for ultimate match failure. + ctx %Match context. + }). %% Stack/register state record. -record(sr, {reg=[], %Register table stk=[], %Stack table - res=[]}). %Reserved regs: [{reserved,I,V}] + res=[]}). %Registers to reserve + +%% Internal records. +-record(cg_need_heap, {anno=[] :: term(), + h=0 :: integer()}). +-record(cg_block, {anno=[] :: term(), + es=[] :: [term()]}). --type life_module() :: {module(),_,_,[_]}. +-type vdb_entry() :: {atom(),non_neg_integer(),non_neg_integer()}. --spec module(life_module(), [compile:option()]) -> {'ok',beam_asm:module_code()}. +-record(l, {i=0 :: non_neg_integer(), %Op number + vdb=[] :: [vdb_entry()], %Variable database + a=[] :: [term()]}). %Core annotation -module({Mod,Exp,Attr,Forms}, _Options) -> - {Fs,St} = functions(Forms, {atom,Mod}), - {ok,{Mod,Exp,Attr,Fs,St#cg.lcount}}. +-spec module(#k_mdef{}, [compile:option()]) -> {'ok',beam_asm:module_code()}. + +module(#k_mdef{name=Mod,exports=Es,attributes=Attr,body=Forms}, _Opts) -> + {Asm,St} = functions(Forms, {atom,Mod}), + {ok,{Mod,Es,Attr,Asm,St#cg.lcount}}. functions(Forms, AtomMod) -> mapfoldl(fun (F, St) -> function(F, AtomMod, St) end, #cg{lcount=1}, Forms). -function({function,Name,Arity,Asm0,Vb,Vdb,Anno}, AtomMod, St0) -> +function(#k_fdef{anno=#k{a=Anno},func=Name,arity=Arity, + vars=As,body=Kb}, AtomMod, St0) -> + try + #k_match{} = Kb, %Assertion. + + %% Try to suppress the stack frame unless it is + %% really needed. + Body0 = avoid_stack_frame(Kb), + + %% Annotate kernel records with variable usage. + Vdb0 = init_vars(As), + {Body,_,Vdb} = body(Body0, 1, Vdb0), + + %% Generate the BEAM assembly code. + {Asm,EntryLabel,St} = cg_fun(Body, As, Vdb, AtomMod, + {Name,Arity}, Anno, St0), + Func = {function,Name,Arity,EntryLabel,Asm}, + {Func,St} + catch + Class:Error:Stack -> + io:fwrite("Function: ~w/~w\n", [Name,Arity]), + erlang:raise(Class, Error, Stack) + end. + + +%% avoid_stack_frame(Kernel) -> Kernel' +%% If possible, avoid setting up a stack frame. Functions +%% that only do matching, calls to guard BIFs, and tail-recursive +%% calls don't need a stack frame. + +avoid_stack_frame(#k_match{body=Body}=M) -> try - {Asm,EntryLabel,St} = cg_fun(Vb, Asm0, Vdb, AtomMod, - {Name,Arity}, Anno, St0), - Func = {function,Name,Arity,EntryLabel,Asm}, - {Func,St} + M#k_match{body=avoid_stack_frame_1(Body)} catch - Class:Error -> - Stack = erlang:get_stacktrace(), - io:fwrite("Function: ~w/~w\n", [Name,Arity]), - erlang:raise(Class, Error, Stack) + impossible -> + M end. +avoid_stack_frame_1(#k_alt{first=First0,then=Then0}=Alt) -> + First = avoid_stack_frame_1(First0), + Then = avoid_stack_frame_1(Then0), + Alt#k_alt{first=First,then=Then}; +avoid_stack_frame_1(#k_bif{op=Op}=Bif) -> + case Op of + #k_internal{} -> + %% Most internal BIFs clobber the X registers. + throw(impossible); + _ -> + Bif + end; +avoid_stack_frame_1(#k_break{anno=Anno,args=Args}) -> + #k_guard_break{anno=Anno,args=Args}; +avoid_stack_frame_1(#k_guard_break{}=Break) -> + Break; +avoid_stack_frame_1(#k_enter{}=Enter) -> + %% Tail-recursive calls don't need a stack frame. + Enter; +avoid_stack_frame_1(#k_guard{clauses=Cs0}=Guard) -> + Cs = avoid_stack_frame_list(Cs0), + Guard#k_guard{clauses=Cs}; +avoid_stack_frame_1(#k_guard_clause{guard=G0,body=B0}=C) -> + G = avoid_stack_frame_1(G0), + B = avoid_stack_frame_1(B0), + C#k_guard_clause{guard=G,body=B}; +avoid_stack_frame_1(#k_match{anno=A,vars=Vs,body=B0,ret=Ret}) -> + %% Use #k_guard_match{} instead to avoid saving the X registers + %% to the stack before matching. + B = avoid_stack_frame_1(B0), + #k_guard_match{anno=A,vars=Vs,body=B,ret=Ret}; +avoid_stack_frame_1(#k_guard_match{body=B0}=M) -> + B = avoid_stack_frame_1(B0), + M#k_guard_match{body=B}; +avoid_stack_frame_1(#k_protected{arg=Arg0}=Prot) -> + Arg = avoid_stack_frame_1(Arg0), + Prot#k_protected{arg=Arg}; +avoid_stack_frame_1(#k_put{}=Put) -> + Put; +avoid_stack_frame_1(#k_return{}=Ret) -> + Ret; +avoid_stack_frame_1(#k_select{var=#k_var{anno=Vanno},types=Types0}=Select) -> + case member(reuse_for_context, Vanno) of + false -> + Types = avoid_stack_frame_list(Types0), + Select#k_select{types=Types}; + true -> + %% Including binary patterns that overwrite the register containing + %% the binary with the match context may not be safe. For example, + %% bs_match_SUITE:bin_tail_e/1 with inlining will be rejected by + %% beam_validator. + %% + %% Essentially the following code is produced: + %% + %% bs_match {x,0} => {x,0} + %% ... + %% bs_match {x,0} => {x,1} %% ILLEGAL + %% + %% A bs_match instruction will only accept a match context as the + %% source operand if the source and destination registers are the + %% the same (as in the first bs_match instruction above). + %% The second bs_match instruction is therefore illegal. + %% + %% This situation is avoided if there is a stack frame: + %% + %% move {x,0} => {y,0} + %% bs_match {x,0} => {x,0} + %% ... + %% bs_match {y,0} => {x,1} %% LEGAL + %% + throw(impossible) + end; +avoid_stack_frame_1(#k_seq{arg=#k_call{anno=Anno,op=Op}=Call, + body=#k_break{args=BrArgs0}}=Seq) -> + case Op of + #k_remote{mod=#k_atom{val=Mod}, + name=#k_atom{val=Name}, + arity=Arity} -> + case erl_bifs:is_exit_bif(Mod, Name, Arity) of + false -> + %% Will clobber X registers. Must have a stack frame. + throw(impossible); + true -> + %% The call to this BIF will never return. It is safe + %% to suppress the stack frame. + Bif = #k_bif{anno=Anno, + op=#k_internal{name=guard_error,arity=1}, + args=[Call],ret=[]}, + BrArgs = lists:duplicate(length(BrArgs0), #k_nil{}), + GB = #k_guard_break{anno=#k{us=[],ns=[],a=[]},args=BrArgs}, + Seq#k_seq{arg=Bif,body=GB} + end; + _ -> + %% Will clobber X registers. Must have a stack frame. + throw(impossible) + end; +avoid_stack_frame_1(#k_seq{arg=A0,body=B0}=Seq) -> + A = avoid_stack_frame_1(A0), + B = avoid_stack_frame_1(B0), + Seq#k_seq{arg=A,body=B}; +avoid_stack_frame_1(#k_test{}=Test) -> + Test; +avoid_stack_frame_1(#k_type_clause{values=Values0}=TC) -> + Values = avoid_stack_frame_list(Values0), + TC#k_type_clause{values=Values}; +avoid_stack_frame_1(#k_val_clause{body=B0}=VC) -> + B = avoid_stack_frame_1(B0), + VC#k_val_clause{body=B}; +avoid_stack_frame_1(_Body) -> + throw(impossible). + +avoid_stack_frame_list([H|T]) -> + [avoid_stack_frame_1(H)|avoid_stack_frame_list(T)]; +avoid_stack_frame_list([]) -> []. + + +%% This pass creates beam format annotated with variable lifetime +%% information. Each thing is given an index and for each variable we +%% store the first and last index for its occurrence. The variable +%% database, VDB, attached to each thing is only relevant internally +%% for that thing. +%% +%% For nested things like matches the numbering continues locally and +%% the VDB for that thing refers to the variable usage within that +%% thing. Variables which live through a such a thing are internally +%% given a very large last index. Internally the indexes continue +%% after the index of that thing. This creates no problems as the +%% internal variable info never escapes and externally we only see +%% variable which are alive both before or after. +%% +%% This means that variables never "escape" from a thing and the only +%% way to get values from a thing is to "return" them, with 'break' or +%% 'return'. Externally these values become the return values of the +%% thing. This is no real limitation as most nested things have +%% multiple threads so working out a common best variable usage is +%% difficult. + +%% body(Kbody, I, Vdb) -> {[Expr],MaxI,Vdb}. +%% Handle a body. + +body(#k_seq{arg=Ke,body=Kb}, I, Vdb0) -> + %%ok = io:fwrite("life ~w:~p~n", [?LINE,{Ke,I,Vdb0}]), + A = get_kanno(Ke), + Vdb1 = use_vars(union(A#k.us, A#k.ns), I, Vdb0), + {Es,MaxI,Vdb2} = body(Kb, I+1, Vdb1), + E = expr(Ke, I, Vdb2), + {[E|Es],MaxI,Vdb2}; +body(Ke, I, Vdb0) -> + %%ok = io:fwrite("life ~w:~p~n", [?LINE,{Ke,I,Vdb0}]), + A = get_kanno(Ke), + Vdb1 = use_vars(union(A#k.us, A#k.ns), I, Vdb0), + E = expr(Ke, I, Vdb1), + {[E],I,Vdb1}. + +%% expr(Kexpr, I, Vdb) -> Expr. + +expr(#k_test{anno=A}=Test, I, _Vdb) -> + Test#k_test{anno=#l{i=I,a=A#k.a}}; +expr(#k_call{anno=A}=Call, I, _Vdb) -> + Call#k_call{anno=#l{i=I,a=A#k.a}}; +expr(#k_enter{anno=A}=Enter, I, _Vdb) -> + Enter#k_enter{anno=#l{i=I,a=A#k.a}}; +expr(#k_bif{anno=A}=Bif, I, _Vdb) -> + Bif#k_bif{anno=#l{i=I,a=A#k.a}}; +expr(#k_match{anno=A,body=Kb,ret=Rs}, I, Vdb) -> + %% Work out imported variables which need to be locked. + Mdb = vdb_sub(I, I+1, Vdb), + M = match(Kb, A#k.us, I+1, Mdb), + L = #l{i=I,vdb=use_vars(A#k.us, I+1, Mdb),a=A#k.a}, + #k_match{anno=L,body=M,ret=Rs}; +expr(#k_guard_match{anno=A,body=Kb,ret=Rs}, I, Vdb) -> + %% Work out imported variables which need to be locked. + Mdb = vdb_sub(I, I+1, Vdb), + M = match(Kb, A#k.us, I+1, Mdb), + L = #l{i=I,vdb=use_vars(A#k.us, I+1, Mdb),a=A#k.a}, + #k_guard_match{anno=L,body=M,ret=Rs}; +expr(#k_protected{}=Protected, I, Vdb) -> + protected(Protected, I, Vdb); +expr(#k_try{anno=A,arg=Ka,vars=Vs,body=Kb,evars=Evs,handler=Kh}=Try, I, Vdb) -> + %% Lock variables that are alive before the catch and used afterwards. + %% Don't lock variables that are only used inside the try. + Tdb0 = vdb_sub(I, I+1, Vdb), + %% This is the tricky bit. Lock variables in Arg that are used in + %% the body and handler. Add try tag 'variable'. + Ab = get_kanno(Kb), + Ah = get_kanno(Kh), + Tdb1 = use_vars(union(Ab#k.us, Ah#k.us), I+3, Tdb0), + Tdb2 = vdb_sub(I, I+2, Tdb1), + Vnames = fun (Kvar) -> Kvar#k_var.name end, %Get the variable names + {Aes,_,Adb} = body(Ka, I+2, add_var({catch_tag,I+1}, I+1, locked, Tdb2)), + {Bes,_,Bdb} = body(Kb, I+4, new_vars(sort(map(Vnames, Vs)), I+3, Tdb2)), + {Hes,_,Hdb} = body(Kh, I+4, new_vars(sort(map(Vnames, Evs)), I+3, Tdb2)), + L = #l{i=I,vdb=Tdb1,a=A#k.a}, + Try#k_try{anno=L, + arg=#cg_block{es=Aes,anno=#l{i=I+1,vdb=Adb,a=[]}}, + vars=Vs,body=#cg_block{es=Bes,anno=#l{i=I+3,vdb=Bdb,a=[]}}, + evars=Evs,handler=#cg_block{es=Hes,anno=#l{i=I+3,vdb=Hdb,a=[]}}}; +expr(#k_try_enter{anno=A,arg=Ka,vars=Vs,body=Kb,evars=Evs,handler=Kh}, I, Vdb) -> + %% Lock variables that are alive before the catch and used afterwards. + %% Don't lock variables that are only used inside the try. + Tdb0 = vdb_sub(I, I+1, Vdb), + %% This is the tricky bit. Lock variables in Arg that are used in + %% the body and handler. Add try tag 'variable'. + Ab = get_kanno(Kb), + Ah = get_kanno(Kh), + Tdb1 = use_vars(union(Ab#k.us, Ah#k.us), I+3, Tdb0), + Tdb2 = vdb_sub(I, I+2, Tdb1), + Vnames = fun (Kvar) -> Kvar#k_var.name end, %Get the variable names + {Aes,_,Adb} = body(Ka, I+2, add_var({catch_tag,I+1}, I+1, 1000000, Tdb2)), + {Bes,_,Bdb} = body(Kb, I+4, new_vars(sort(map(Vnames, Vs)), I+3, Tdb2)), + {Hes,_,Hdb} = body(Kh, I+4, new_vars(sort(map(Vnames, Evs)), I+3, Tdb2)), + L = #l{i=I,vdb=Tdb1,a=A#k.a}, + #k_try_enter{anno=L, + arg=#cg_block{es=Aes,anno=#l{i=I+1,vdb=Adb,a=[]}}, + vars=Vs,body=#cg_block{es=Bes,anno=#l{i=I+3,vdb=Bdb,a=[]}}, + evars=Evs,handler=#cg_block{es=Hes,anno=#l{i=I+3,vdb=Hdb,a=[]}}}; +expr(#k_catch{anno=A,body=Kb}=Catch, I, Vdb) -> + %% Lock variables that are alive before the catch and used afterwards. + %% Don't lock variables that are only used inside the catch. + %% Add catch tag 'variable'. + Cdb0 = vdb_sub(I, I+1, Vdb), + {Es,_,Cdb1} = body(Kb, I+1, add_var({catch_tag,I}, I, locked, Cdb0)), + L = #l{i=I,vdb=Cdb1,a=A#k.a}, + Catch#k_catch{anno=L,body=#cg_block{es=Es}}; +expr(#k_receive{anno=A,var=V,body=Kb,action=Ka}=Recv, I, Vdb) -> + %% Work out imported variables which need to be locked. + Rdb = vdb_sub(I, I+1, Vdb), + M = match(Kb, add_element(V#k_var.name, A#k.us), I+1, + new_vars([V#k_var.name], I, Rdb)), + {Tes,_,Adb} = body(Ka, I+1, Rdb), + Le = #l{i=I,vdb=use_vars(A#k.us, I+1, Vdb),a=A#k.a}, + Recv#k_receive{anno=Le,body=M, + action=#cg_block{anno=#l{i=I+1,vdb=Adb,a=[]},es=Tes}}; +expr(#k_receive_accept{anno=A}, I, _Vdb) -> + #k_receive_accept{anno=#l{i=I,a=A#k.a}}; +expr(#k_receive_next{anno=A}, I, _Vdb) -> + #k_receive_next{anno=#l{i=I,a=A#k.a}}; +expr(#k_put{anno=A}=Put, I, _Vdb) -> + Put#k_put{anno=#l{i=I,a=A#k.a}}; +expr(#k_break{anno=A}=Break, I, _Vdb) -> + Break#k_break{anno=#l{i=I,a=A#k.a}}; +expr(#k_guard_break{anno=A}=Break, I, _Vdb) -> + Break#k_guard_break{anno=#l{i=I,a=A#k.a}}; +expr(#k_return{anno=A}=Ret, I, _Vdb) -> + Ret#k_return{anno=#l{i=I,a=A#k.a}}. + +%% protected(Kprotected, I, Vdb) -> Protected. +%% Only used in guards. + +protected(#k_protected{anno=A,arg=Ts}=Prot, I, Vdb) -> + %% Lock variables that are alive before try and used afterwards. + %% Don't lock variables that are only used inside the protected + %% expression. + Pdb0 = vdb_sub(I, I+1, Vdb), + {T,MaxI,Pdb1} = body(Ts, I+1, Pdb0), + Pdb2 = use_vars(A#k.ns, MaxI+1, Pdb1), %Save "return" values + Prot#k_protected{arg=T,anno=#l{i=I,a=A#k.a,vdb=Pdb2}}. + +%% match(Kexpr, [LockVar], I, Vdb) -> Expr. +%% Convert match tree to old format. + +match(#k_alt{anno=A,first=Kf,then=Kt}, Ls, I, Vdb0) -> + Vdb1 = use_vars(union(A#k.us, Ls), I, Vdb0), + F = match(Kf, Ls, I+1, Vdb1), + T = match(Kt, Ls, I+1, Vdb1), + #k_alt{anno=[],first=F,then=T}; +match(#k_select{anno=A,types=Kts}=Select, Ls, I, Vdb0) -> + Vdb1 = use_vars(union(A#k.us, Ls), I, Vdb0), + Ts = [type_clause(Tc, Ls, I+1, Vdb1) || Tc <- Kts], + Select#k_select{anno=[],types=Ts}; +match(#k_guard{anno=A,clauses=Kcs}, Ls, I, Vdb0) -> + Vdb1 = use_vars(union(A#k.us, Ls), I, Vdb0), + Cs = [guard_clause(G, Ls, I+1, Vdb1) || G <- Kcs], + #k_guard{anno=[],clauses=Cs}; +match(Other, Ls, I, Vdb0) -> + Vdb1 = use_vars(Ls, I, Vdb0), + {B,_,Vdb2} = body(Other, I+1, Vdb1), + Le = #l{i=I,vdb=Vdb2,a=[]}, + #cg_block{anno=Le,es=B}. + +type_clause(#k_type_clause{anno=A,type=T,values=Kvs}, Ls, I, Vdb0) -> + %%ok = io:format("life ~w: ~p~n", [?LINE,{T,Kvs}]), + Vdb1 = use_vars(union(A#k.us, Ls), I+1, Vdb0), + Vs = [val_clause(Vc, Ls, I+1, Vdb1) || Vc <- Kvs], + #k_type_clause{anno=[],type=T,values=Vs}. + +val_clause(#k_val_clause{anno=A,val=V,body=Kb}, Ls0, I, Vdb0) -> + New = (get_kanno(V))#k.ns, + Bus = (get_kanno(Kb))#k.us, + %%ok = io:format("Ls0 = ~p, Used=~p\n New=~p, Bus=~p\n", [Ls0,Used,New,Bus]), + Ls1 = union(intersection(New, Bus), Ls0), %Lock for safety + Vdb1 = use_vars(union(A#k.us, Ls1), I+1, new_vars(New, I, Vdb0)), + B = match(Kb, Ls1, I+1, Vdb1), + Le = #l{i=I,vdb=use_vars(Bus, I+1, Vdb1),a=A#k.a}, + #k_val_clause{anno=Le,val=V,body=B}. + +guard_clause(#k_guard_clause{anno=A,guard=Kg,body=Kb}, Ls, I, Vdb0) -> + Vdb1 = use_vars(union(A#k.us, Ls), I+2, Vdb0), + Gdb = vdb_sub(I+1, I+2, Vdb1), + G = protected(Kg, I+1, Gdb), + B = match(Kb, Ls, I+2, Vdb1), + Le = #l{i=I,vdb=use_vars((get_kanno(Kg))#k.us, I+2, Vdb1),a=A#k.a}, + #k_guard_clause{anno=Le,guard=G,body=B}. + + +%% Here follows the code generator pass. +%% +%% The following assumptions have been made: +%% +%% 1. Matches, i.e. things with {match,M,Ret} wrappers, only return +%% values; no variables are exported. If the match would have returned +%% extra variables then these have been transformed to multiple return +%% values. +%% +%% 2. All BIF's called in guards are gc-safe so there is no need to +%% put thing on the stack in the guard. While this would in principle +%% work it would be difficult to keep track of the stack depth when +%% trimming. +%% +%% The code generation uses variable lifetime information added by +%% the previous pass to save variables, allocate registers and +%% move registers to the stack when necessary. +%% +%% We try to use a consistent variable name scheme throughout. The +%% StackReg record is always called Bef,Int<n>,Aft. + %% cg_fun([Lkexpr], [HeadVar], Vdb, State) -> {[Ainstr],State} cg_fun(Les, Hvs, Vdb, AtomMod, NameArity, Anno, St0) -> @@ -114,18 +459,18 @@ cg_fun(Les, Hvs, Vdb, AtomMod, NameArity, Anno, St0) -> %% Note that and 'if_end' instruction does not need any %% live x registers, so it will always be safe to jump to %% it. (We never ever expect the jump to be taken, and in - %% must functions there will never be any references to + %% most functions there will never be any references to %% the label in the first place.) %% {UltimateMatchFail,St3} = new_label(St2), %% Create initial stack/register state, clear unused arguments. - Bef = clear_dead(#sr{reg=foldl(fun ({var,V}, Reg) -> + Bef = clear_dead(#sr{reg=foldl(fun (#k_var{name=V}, Reg) -> put_reg(V, Reg) end, [], Hvs), stk=[]}, 0, Vdb), - {B,_Aft,St} = cg_list(Les, 0, Vdb, Bef, + {B,_Aft,St} = cg_list(Les, Vdb, Bef, St3#cg{bfail=0, ultimate_failure=UltimateMatchFail, is_top_block=true}), @@ -136,66 +481,64 @@ cg_fun(Les, Hvs, Vdb, AtomMod, NameArity, Anno, St0) -> %% cg(Lkexpr, Vdb, StackReg, State) -> {[Ainstr],StackReg,State}. %% Generate code for a kexpr. -%% Split function into two steps for clarity, not efficiency. -cg(Le, Vdb, Bef, St) -> - cg(Le#l.ke, Le, Vdb, Bef, St). - -cg({block,Es}, Le, Vdb, Bef, St) -> +cg(#cg_block{anno=Le,es=Es}, Vdb, Bef, St) -> block_cg(Es, Le, Vdb, Bef, St); -cg({match,M,Rs}, Le, Vdb, Bef, St) -> +cg(#k_match{anno=Le,body=M,ret=Rs}, Vdb, Bef, St) -> match_cg(M, Rs, Le, Vdb, Bef, St); -cg({guard_match,M,Rs}, Le, Vdb, Bef, St) -> +cg(#k_guard_match{anno=Le,body=M,ret=Rs}, Vdb, Bef, St) -> guard_match_cg(M, Rs, Le, Vdb, Bef, St); -cg({call,Func,As,Rs}, Le, Vdb, Bef, St) -> +cg(#k_call{anno=Le,op=Func,args=As,ret=Rs}, Vdb, Bef, St) -> call_cg(Func, As, Rs, Le, Vdb, Bef, St); -cg({enter,Func,As}, Le, Vdb, Bef, St) -> +cg(#k_enter{anno=Le,op=Func,args=As}, Vdb, Bef, St) -> enter_cg(Func, As, Le, Vdb, Bef, St); -cg({bif,Bif,As,Rs}, Le, Vdb, Bef, St) -> - bif_cg(Bif, As, Rs, Le, Vdb, Bef, St); -cg({gc_bif,Bif,As,Rs}, Le, Vdb, Bef, St) -> - gc_bif_cg(Bif, As, Rs, Le, Vdb, Bef, St); -cg({internal,Bif,As,Rs}, Le, Vdb, Bef, St) -> - internal_cg(Bif, As, Rs, Le, Vdb, Bef, St); -cg({receive_loop,Te,Rvar,Rm,Tes,Rs}, Le, Vdb, Bef, St) -> +cg(#k_bif{anno=Le}=Bif, Vdb, Bef, St) -> + bif_cg(Bif, Le, Vdb, Bef, St); +cg(#k_receive{anno=Le,timeout=Te,var=Rvar,body=Rm,action=Tes,ret=Rs}, + Vdb, Bef, St) -> recv_loop_cg(Te, Rvar, Rm, Tes, Rs, Le, Vdb, Bef, St); -cg(receive_next, Le, Vdb, Bef, St) -> +cg(#k_receive_next{anno=Le}, Vdb, Bef, St) -> recv_next_cg(Le, Vdb, Bef, St); -cg(receive_accept, _Le, _Vdb, Bef, St) -> {[remove_message],Bef,St}; -cg({'try',Ta,Vs,Tb,Evs,Th,Rs}, Le, Vdb, Bef, St) -> +cg(#k_receive_accept{}, _Vdb, Bef, St) -> + {[remove_message],Bef,St}; +cg(#k_try{anno=Le,arg=Ta,vars=Vs,body=Tb,evars=Evs,handler=Th,ret=Rs}, + Vdb, Bef, St) -> try_cg(Ta, Vs, Tb, Evs, Th, Rs, Le, Vdb, Bef, St); -cg({try_enter,Ta,Vs,Tb,Evs,Th}, Le, Vdb, Bef, St) -> +cg(#k_try_enter{anno=Le,arg=Ta,vars=Vs,body=Tb,evars=Evs,handler=Th}, + Vdb, Bef, St) -> try_enter_cg(Ta, Vs, Tb, Evs, Th, Le, Vdb, Bef, St); -cg({'catch',Cb,R}, Le, Vdb, Bef, St) -> +cg(#k_catch{anno=Le,body=Cb,ret=[R]}, Vdb, Bef, St) -> catch_cg(Cb, R, Le, Vdb, Bef, St); -cg({set,Var,Con}, Le, Vdb, Bef, St) -> - set_cg(Var, Con, Le, Vdb, Bef, St); -cg({return,Rs}, Le, Vdb, Bef, St) -> return_cg(Rs, Le, Vdb, Bef, St); -cg({break,Bs}, Le, Vdb, Bef, St) -> break_cg(Bs, Le, Vdb, Bef, St); -cg({guard_break,Bs,N}, Le, Vdb, Bef, St) -> - guard_break_cg(Bs, N, Le, Vdb, Bef, St); -cg({need_heap,H}, _Le, _Vdb, Bef, St) -> +cg(#k_put{anno=Le,arg=Con,ret=Var}, Vdb, Bef, St) -> + put_cg(Var, Con, Le, Vdb, Bef, St); +cg(#k_return{anno=Le,args=Rs}, Vdb, Bef, St) -> + return_cg(Rs, Le, Vdb, Bef, St); +cg(#k_break{anno=Le,args=Bs}, Vdb, Bef, St) -> + break_cg(Bs, Le, Vdb, Bef, St); +cg(#k_guard_break{anno=Le,args=Bs}, Vdb, Bef, St) -> + guard_break_cg(Bs, Le, Vdb, Bef, St); +cg(#cg_need_heap{h=H}, _Vdb, Bef, St) -> {[{test_heap,H,max_reg(Bef#sr.reg)}],Bef,St}. %% cg_list([Kexpr], FirstI, Vdb, StackReg, St) -> {[Ainstr],StackReg,St}. -cg_list(Kes, I, Vdb, Bef, St0) -> +cg_list(Kes, Vdb, Bef, St0) -> {Keis,{Aft,St1}} = flatmapfoldl(fun (Ke, {Inta,Sta}) -> {Keis,Intb,Stb} = cg(Ke, Vdb, Inta, Sta), {Keis,{Intb,Stb}} - end, {Bef,St0}, need_heap(Kes, I)), + end, {Bef,St0}, need_heap(Kes)), {Keis,Aft,St1}. %% need_heap([Lkexpr], I, St) -> [Lkexpr]. %% Insert need_heap instructions in Kexpr list. Try to be smart and %% collect them together as much as possible. -need_heap(Kes0, I) -> +need_heap(Kes0) -> {Kes,H} = need_heap_0(reverse(Kes0), 0, []), %% Prepend need_heap if necessary. - need_heap_need(I, H) ++ Kes. + need_heap_need(H) ++ Kes. need_heap_0([Ke|Kes], H0, Acc) -> {Ns,H} = need_heap_1(Ke, H0), @@ -203,27 +546,54 @@ need_heap_0([Ke|Kes], H0, Acc) -> need_heap_0([], H, Acc) -> {Acc,H}. -need_heap_1(#l{ke={set,_,{binary,_}},i=I}, H) -> - {need_heap_need(I, H),0}; -need_heap_1(#l{ke={set,_,{map,_,_,_}},i=I}, H) -> - {need_heap_need(I, H),0}; -need_heap_1(#l{ke={set,_,Val}}, H) -> +need_heap_1(#k_put{arg=#k_binary{}}, H) -> + {need_heap_need(H),0}; +need_heap_1(#k_put{arg=#k_map{}}, H) -> + {need_heap_need(H),0}; +need_heap_1(#k_put{arg=Val}, H) -> %% Just pass through adding to needed heap. {[],H + case Val of - {cons,_} -> 2; - {tuple,Es} -> 1 + length(Es); + #k_cons{} -> 2; + #k_tuple{es=Es} -> 1 + length(Es); _Other -> 0 end}; -need_heap_1(#l{ke={bif,_Bif,_As,_Rs}}, H) -> - {[],H}; -need_heap_1(#l{i=I}, H) -> +need_heap_1(#k_bif{}=Bif, H) -> + case is_gc_bif(Bif) of + false -> + {[],H}; + true -> + {need_heap_need(H),0} + end; +need_heap_1(_Ke, H) -> %% Call or call-like instruction such as set_tuple_element/3. - {need_heap_need(I, H),0}. - -need_heap_need(_I, 0) -> []; -need_heap_need(I, H) -> [#l{ke={need_heap,H},i=I}]. - -%% match_cg(Match, [Ret], Le, Vdb, StackReg, State) -> + {need_heap_need(H),0}. + +need_heap_need(0) -> []; +need_heap_need(H) -> [#cg_need_heap{h=H}]. + +%% is_gc_bif(#k_bif{}) -> true|false. +%% is_gc_bif(Name, Arity) -> true|false. +%% Determines whether the BIF Name/Arity might do a GC. + +is_gc_bif(#k_bif{op=#k_remote{name=#k_atom{val=Name}},args=Args}) -> + is_gc_bif(Name, length(Args)); +is_gc_bif(#k_bif{op=#k_internal{}}) -> + true. + +is_gc_bif(hd, 1) -> false; +is_gc_bif(tl, 1) -> false; +is_gc_bif(self, 0) -> false; +is_gc_bif(node, 0) -> false; +is_gc_bif(node, 1) -> false; +is_gc_bif(element, 2) -> false; +is_gc_bif(get, 1) -> false; +is_gc_bif(tuple_size, 1) -> false; +is_gc_bif(Bif, Arity) -> + not (erl_internal:bool_op(Bif, Arity) orelse + erl_internal:new_type_test(Bif, Arity) orelse + erl_internal:comp_op(Bif, Arity)). + +%% match_cg(Matc, [Ret], Le, Vdb, StackReg, State) -> %% {[Ainstr],StackReg,State}. %% Generate code for a match. First save all variables on the stack %% that are to survive after the match. We leave saved variables in @@ -244,7 +614,10 @@ match_cg(M, Rs, Le, Vdb, Bef, St0) -> guard_match_cg(M, Rs, Le, Vdb, Bef, St0) -> I = Le#l.i, {B,St1} = new_label(St0), - #cg{bfail=Fail} = St1, + Fail = case St0 of + #cg{bfail=0,ultimate_failure=Fail0} -> Fail0; + #cg{bfail=Fail0} -> Fail0 + end, {Mis,Aft,St2} = match_cg(M, Fail, Bef, St1#cg{break=B}), %% Update the register descriptors for the return registers. Reg = guard_match_regs(Aft#sr.reg, Rs), @@ -252,7 +625,7 @@ guard_match_cg(M, Rs, Le, Vdb, Bef, St0) -> clear_dead(Aft#sr{reg=Reg}, I, Vdb), St2#cg{break=St1#cg.break}}. -guard_match_regs([{I,gbreakvar}|Rs], [{var,V}|Vs]) -> +guard_match_regs([{I,gbreakvar}|Rs], [#k_var{name=V}|Vs]) -> [{I,V}|guard_match_regs(Rs, Vs)]; guard_match_regs([R|Rs], Vs) -> [R|guard_match_regs(Rs, Vs)]; @@ -264,17 +637,14 @@ guard_match_regs([], []) -> []. %% down as each level which uses this takes its own internal Vdb not %% the outer one. -match_cg(Le, Fail, Bef, St) -> - match_cg(Le#l.ke, Le, Fail, Bef, St). - -match_cg({alt,F,S}, _Le, Fail, Bef, St0) -> +match_cg(#k_alt{first=F,then=S}, Fail, Bef, St0) -> {Tf,St1} = new_label(St0), {Fis,Faft,St2} = match_cg(F, Tf, Bef, St1), {Sis,Saft,St3} = match_cg(S, Fail, Bef, St2), Aft = sr_merge(Faft, Saft), {Fis ++ [{label,Tf}] ++ Sis,Aft,St3}; -match_cg({select,{var,Vname}=V,Scs0}, #l{a=Anno}, Fail, Bef, St) -> - ReuseForContext = member(reuse_for_context, Anno) andalso +match_cg(#k_select{var=#k_var{anno=Vanno,name=Vname}=V,types=Scs0}, Fail, Bef, St) -> + ReuseForContext = member(reuse_for_context, Vanno) andalso find_reg(Vname, Bef#sr.reg) =/= error, Scs = case ReuseForContext of false -> Scs0; @@ -283,10 +653,10 @@ match_cg({select,{var,Vname}=V,Scs0}, #l{a=Anno}, Fail, Bef, St) -> match_fmf(fun (S, F, Sta) -> select_cg(S, V, F, Fail, Bef, Sta) end, Fail, St, Scs); -match_cg({guard,Gcs}, _Le, Fail, Bef, St) -> +match_cg(#k_guard{clauses=Gcs}, Fail, Bef, St) -> match_fmf(fun (G, F, Sta) -> guard_clause_cg(G, F, Bef, Sta) end, Fail, St, Gcs); -match_cg({block,Es}, Le, _Fail, Bef, St) -> +match_cg(#cg_block{anno=Le,es=Es}, _Fail, Bef, St) -> %% Must clear registers and stack of dead variables. Int = clear_dead(Bef, Le#l.i, Le#l.vdb), block_cg(Es, Le, Int, St). @@ -294,8 +664,8 @@ match_cg({block,Es}, Le, _Fail, Bef, St) -> %% bsm_rename_ctx([Clause], Var) -> [Clause] %% We know from an annotation that the register for a binary can %% be reused for the match context because the two are not truly -%% alive at the same time (even though the conservative life time -%% information calculated by v3_life says so). +%% alive at the same time (even though the life time information +%% says so). %% %% The easiest way to have those variables share the same register is %% to rename the variable with the shortest life-span (the match @@ -306,12 +676,14 @@ match_cg({block,Es}, Le, _Fail, Bef, St) -> %% We must also remove all information about the match context %% variable from all life-time information databases (Vdb). -bsm_rename_ctx([#l{ke={type_clause,binary, - [#l{ke={val_clause,{binary,{var,Old}},Ke0}}=L2]}}=L1|Cs], New) -> +bsm_rename_ctx([#k_type_clause{type=k_binary,values=Vcs}=TC|Cs], New) -> + [#k_val_clause{val=#k_binary{segs=#k_var{name=Old}}=Bin, + body=Ke0}=VC0] = Vcs, Ke = bsm_rename_ctx(Ke0, Old, New, false), - [L1#l{ke={type_clause,binary, - [L2#l{ke={val_clause,{binary,{var,New}},Ke}}]}}|bsm_rename_ctx(Cs, New)]; -bsm_rename_ctx([C|Cs], New) -> + VC = VC0#k_val_clause{val=Bin#k_binary{segs=#k_var{name=New}}, + body=Ke}, + [TC#k_type_clause{values=[VC]}|bsm_rename_ctx(Cs, New)]; +bsm_rename_ctx([C|Cs], New) -> [C|bsm_rename_ctx(Cs, New)]; bsm_rename_ctx([], _) -> []. @@ -321,34 +693,24 @@ bsm_rename_ctx([], _) -> []. %% only complicatate things to recurse into blocks not in a protected %% (the match context variable is not live inside them). -bsm_rename_ctx(#l{ke={select,{var,V},Cs0}}=L, Old, New, InProt) -> +bsm_rename_ctx(#k_select{var=#k_var{name=V},types=Cs0}=Sel, + Old, New, InProt) -> Cs = bsm_rename_ctx_list(Cs0, Old, New, InProt), - L#l{ke={select,{var,bsm_rename_var(V, Old, New)},Cs}}; -bsm_rename_ctx(#l{ke={type_clause,Type,Cs0}}=L, Old, New, InProt) -> + Sel#k_select{var=#k_var{name=bsm_rename_var(V, Old, New)},types=Cs}; +bsm_rename_ctx(#k_type_clause{values=Cs0}=TC, Old, New, InProt) -> Cs = bsm_rename_ctx_list(Cs0, Old, New, InProt), - L#l{ke={type_clause,Type,Cs}}; -bsm_rename_ctx(#l{ke={val_clause,{bin_end,V},Ke0}}=L, Old, New, InProt) -> + TC#k_type_clause{values=Cs}; +bsm_rename_ctx(#k_val_clause{body=Ke0}=VC, Old, New, InProt) -> Ke = bsm_rename_ctx(Ke0, Old, New, InProt), - L#l{ke={val_clause,{bin_end,bsm_rename_var(V, Old, New)},Ke}}; -bsm_rename_ctx(#l{ke={val_clause,{bin_seg,V,Sz,U,Type,Fl,Vs},Ke0}}=L, - Old, New, InProt) -> - Ke = bsm_rename_ctx(Ke0, Old, New, InProt), - L#l{ke={val_clause,{bin_seg,bsm_rename_var(V, Old, New),Sz,U,Type,Fl,Vs},Ke}}; -bsm_rename_ctx(#l{ke={val_clause,{bin_int,V,Sz,U,Fl,Val,Vs},Ke0}}=L, - Old, New, InProt) -> - Ke = bsm_rename_ctx(Ke0, Old, New, InProt), - L#l{ke={val_clause,{bin_int,bsm_rename_var(V, Old, New),Sz,U,Fl,Val,Vs},Ke}}; -bsm_rename_ctx(#l{ke={val_clause,Val,Ke0}}=L, Old, New, InProt) -> - Ke = bsm_rename_ctx(Ke0, Old, New, InProt), - L#l{ke={val_clause,Val,Ke}}; -bsm_rename_ctx(#l{ke={alt,F0,S0}}=L, Old, New, InProt) -> + VC#k_val_clause{body=Ke}; +bsm_rename_ctx(#k_alt{first=F0,then=S0}=Alt, Old, New, InProt) -> F = bsm_rename_ctx(F0, Old, New, InProt), S = bsm_rename_ctx(S0, Old, New, InProt), - L#l{ke={alt,F,S}}; -bsm_rename_ctx(#l{ke={guard,Gcs0}}=L, Old, New, InProt) -> + Alt#k_alt{first=F,then=S}; +bsm_rename_ctx(#k_guard{clauses=Gcs0}=Guard, Old, New, InProt) -> Gcs = bsm_rename_ctx_list(Gcs0, Old, New, InProt), - L#l{ke={guard,Gcs}}; -bsm_rename_ctx(#l{ke={guard_clause,G0,B0}}=L, Old, New, InProt) -> + Guard#k_guard{clauses=Gcs}; +bsm_rename_ctx(#k_guard_clause{guard=G0,body=B0}=GC, Old, New, InProt) -> G = bsm_rename_ctx(G0, Old, New, InProt), B = bsm_rename_ctx(B0, Old, New, InProt), %% A guard clause may cause unsaved variables to be saved on the stack. @@ -356,49 +718,45 @@ bsm_rename_ctx(#l{ke={guard_clause,G0,B0}}=L, Old, New, InProt) -> %% same register), it is neither in the stack nor register descriptor %% lists and we would crash when we didn't find it unless we remove %% it from the database. - bsm_forget_var(L#l{ke={guard_clause,G,B}}, Old); -bsm_rename_ctx(#l{ke={protected,Ts0,Rs}}=L, Old, New, _InProt) -> + bsm_forget_var(GC#k_guard_clause{guard=G,body=B}, Old); +bsm_rename_ctx(#k_protected{arg=Ts0}=Prot, Old, New, _InProt) -> InProt = true, Ts = bsm_rename_ctx_list(Ts0, Old, New, InProt), - bsm_forget_var(L#l{ke={protected,Ts,Rs}}, Old); -bsm_rename_ctx(#l{ke={match,Ms0,Rs}}=L, Old, New, InProt) -> + bsm_forget_var(Prot#k_protected{arg=Ts}, Old); +bsm_rename_ctx(#k_guard_match{body=Ms0}=Match, Old, New, InProt) -> Ms = bsm_rename_ctx(Ms0, Old, New, InProt), - L#l{ke={match,Ms,Rs}}; -bsm_rename_ctx(#l{ke={guard_match,Ms0,Rs}}=L, Old, New, InProt) -> - Ms = bsm_rename_ctx(Ms0, Old, New, InProt), - L#l{ke={guard_match,Ms,Rs}}; -bsm_rename_ctx(#l{ke={test,_,_,_}}=L, _, _, _) -> L; -bsm_rename_ctx(#l{ke={bif,_,_,_}}=L, _, _, _) -> L; -bsm_rename_ctx(#l{ke={gc_bif,_,_,_}}=L, _, _, _) -> L; -bsm_rename_ctx(#l{ke={set,_,_}}=L, _, _, _) -> L; -bsm_rename_ctx(#l{ke={call,_,_,_}}=L, _, _, _) -> L; -bsm_rename_ctx(#l{ke={block,_}}=L, Old, _, false) -> + Match#k_guard_match{body=Ms}; +bsm_rename_ctx(#k_test{}=Test, _, _, _) -> Test; +bsm_rename_ctx(#k_bif{}=Bif, _, _, _) -> Bif; +bsm_rename_ctx(#k_put{}=Put, _, _, _) -> Put; +bsm_rename_ctx(#k_call{}=Call, _, _, _) -> Call; +bsm_rename_ctx(#cg_block{}=Block, Old, _, false) -> %% This block is not inside a protected. The match context variable cannot %% possibly be live inside the block. - bsm_forget_var(L, Old); -bsm_rename_ctx(#l{ke={block,Bl0}}=L, Old, New, true) -> + bsm_forget_var(Block, Old); +bsm_rename_ctx(#cg_block{es=Es0}=Block, Old, New, true) -> %% A block in a protected. We must recursively rename the variable %% inside the block. - Bl = bsm_rename_ctx_list(Bl0, Old, New, true), - bsm_forget_var(L#l{ke={block,Bl}}, Old); -bsm_rename_ctx(#l{ke={guard_break,Bs,Locked0}}=L0, Old, _New, _InProt) -> - Locked = Locked0 -- [Old], - L = L0#l{ke={guard_break,Bs,Locked}}, - bsm_forget_var(L, Old). + Es = bsm_rename_ctx_list(Es0, Old, New, true), + bsm_forget_var(Block#cg_block{es=Es}, Old); +bsm_rename_ctx(#k_guard_break{}=Break, Old, _New, _InProt) -> + bsm_forget_var(Break, Old). bsm_rename_ctx_list([C|Cs], Old, New, InProt) -> [bsm_rename_ctx(C, Old, New, InProt)| bsm_rename_ctx_list(Cs, Old, New, InProt)]; bsm_rename_ctx_list([], _, _, _) -> []. - + bsm_rename_var(Old, Old, New) -> New; bsm_rename_var(V, _, _) -> V. %% bsm_forget_var(#l{}, Variable) -> #l{} %% Remove a variable from the variable life-time database. -bsm_forget_var(#l{vdb=Vdb}=L, V) -> - L#l{vdb=keydelete(V, 1, Vdb)}. +bsm_forget_var(Ke, V) -> + #l{vdb=Vdb} = L0 = get_kanno(Ke), + L = L0#l{vdb=keydelete(V, 1, Vdb)}, + set_kanno(Ke, L). %% block_cg([Kexpr], Le, Vdb, StackReg, St) -> {[Ainstr],StackReg,St}. %% block_cg([Kexpr], Le, StackReg, St) -> {[Ainstr],StackReg,St}. @@ -407,158 +765,227 @@ block_cg(Es, Le, _Vdb, Bef, St) -> block_cg(Es, Le, Bef, St). block_cg(Es, Le, Bef, #cg{is_top_block=false}=St) -> - cg_block(Es, Le#l.i, Le#l.vdb, Bef, St); -block_cg(Es, Le, Bef, St0) -> - {Is0,Aft,St} = cg_block(Es, Le#l.i, Le#l.vdb, Bef, - St0#cg{is_top_block=false,need_frame=false}), - Is = top_level_block(Is0, Aft, max_reg(Bef#sr.reg), St), - {Is,Aft,St#cg{is_top_block=true}}. - -cg_block([], _I, _Vdb, Bef, St0) -> + cg_block(Es, Le#l.vdb, Bef, St); +block_cg(Es, Le, Bef, #cg{is_top_block=true}=St0) -> + %% No stack frame has been established yet. Do we need one? + case need_stackframe(Es) of + true -> + %% We need a stack frame. Generate the code and add the + %% code for creating and deallocating the stack frame. + {Is0,Aft,St} = cg_block(Es, Le#l.vdb, Bef, + St0#cg{is_top_block=false,need_frame=false}), + Is = top_level_block(Is0, Aft, max_reg(Bef#sr.reg), St), + {Is,Aft,St#cg{is_top_block=true}}; + false -> + %% This sequence of instructions ending in a #k_match{} (a + %% 'case' or 'if') in the Erlang code does not need a + %% stack frame yet. Delay the creation (if a stack frame + %% is needed at all, it will be created inside the + %% #k_match{}). + cg_list(Es, Le#l.vdb, Bef, St0) + end. + +%% need_stackframe([Kexpr]) -> true|false. +%% Does this list of instructions need a stack frame? +%% +%% A sequence of instructions that don't clobber the X registers +%% followed by a single #k_match{} doesn't need a stack frame. + +need_stackframe([H|T]) -> + case H of + #k_bif{op=#k_internal{}} -> true; + #k_put{arg=#k_binary{}} -> true; + #k_bif{} -> need_stackframe(T); + #k_put{} -> need_stackframe(T); + #k_guard_match{} -> need_stackframe(T); + #k_match{} when T =:= [] -> false; + _ -> true + end; +need_stackframe([]) -> false. + +cg_block([], _Vdb, Bef, St0) -> {[],Bef,St0}; -cg_block(Kes0, I, Vdb, Bef, St0) -> +cg_block(Kes0, Vdb, Bef, St0) -> {Kes2,Int1,St1} = case basic_block(Kes0) of {Kes1,LastI,Args,Rest} -> - Ke = hd(Kes1), - Fb = Ke#l.i, - cg_basic_block(Kes1, Fb, LastI, Args, Vdb, Bef, St0); + cg_basic_block(Kes1, LastI, Args, Vdb, Bef, St0); {Kes1,Rest} -> - cg_list(Kes1, I, Vdb, Bef, St0) + cg_list(Kes1, Vdb, Bef, St0) end, - {Kes3,Int2,St2} = cg_block(Rest, I, Vdb, Int1, St1), + {Kes3,Int2,St2} = cg_block(Rest, Vdb, Int1, St1), {Kes2 ++ Kes3,Int2,St2}. basic_block(Kes) -> basic_block(Kes, []). -basic_block([Le|Les], Acc) -> - case collect_block(Le#l.ke) of - include -> basic_block(Les, [Le|Acc]); +basic_block([Ke|Kes], Acc) -> + case collect_block(Ke) of + include -> basic_block(Kes, [Ke|Acc]); {block_end,As} -> case Acc of [] -> - %% If the basic block does not contain any set instructions, + %% If the basic block does not contain any #k_put{} instructions, %% it serves no useful purpose to do basic block optimizations. - {[Le],Les}; + {[Ke],Kes}; _ -> - {reverse(Acc, [Le]),Le#l.i,As,Les} + #l{i=I} = get_kanno(Ke), + {reverse(Acc, [Ke]),I,As,Kes} end; - no_block -> {reverse(Acc, [Le]),Les} + no_block -> {reverse(Acc, [Ke]),Kes} end. -%% sets that may garbage collect are not allowed in basic blocks. - -collect_block({set,_,{binary,_}}) -> no_block; -collect_block({set,_,{map,_,_,_}}) -> no_block; -collect_block({set,_,_}) -> include; -collect_block({call,{var,_}=Var,As,_Rs}) -> {block_end,As++[Var]}; -collect_block({call,Func,As,_Rs}) -> {block_end,As++func_vars(Func)}; -collect_block({enter,{var,_}=Var,As})-> {block_end,As++[Var]}; -collect_block({enter,Func,As}) -> {block_end,As++func_vars(Func)}; -collect_block({return,Rs}) -> {block_end,Rs}; -collect_block({break,Bs}) -> {block_end,Bs}; -collect_block(_) -> no_block. - -func_vars({remote,M,F}) when element(1, M) =:= var; - element(1, F) =:= var -> +collect_block(#k_put{arg=Arg}) -> + %% #k_put{} instructions that may garbage collect are not allowed + %% in basic blocks. + case Arg of + #k_binary{} -> no_block; + #k_map{} -> no_block; + _ -> include + end; +collect_block(#k_call{op=Func,args=As}) -> + {block_end,As++func_vars(Func)}; +collect_block(#k_enter{op=Func,args=As}) -> + {block_end,As++func_vars(Func)}; +collect_block(#k_return{args=Rs}) -> + {block_end,Rs}; +collect_block(#k_break{args=Bs}) -> + {block_end,Bs}; +collect_block(_) -> no_block. + +func_vars(#k_var{}=Var) -> + [Var]; +func_vars(#k_remote{mod=M,name=F}) + when is_record(M, k_var); is_record(F, k_var) -> [M,F]; func_vars(_) -> []. -%% cg_basic_block([Kexpr], FirstI, LastI, As, Vdb, StackReg, State) -> +%% cg_basic_block([Kexpr], FirstI, LastI, Arguments, Vdb, StackReg, State) -> %% {[Ainstr],StackReg,State}. - -cg_basic_block(Kes, Fb, Lf, As, Vdb, Bef, St0) -> - Res = make_reservation(As, 0), - Regs0 = reserve(Res, Bef#sr.reg, Bef#sr.stk), - Stk = extend_stack(Bef, Lf, Lf+1, Vdb), - Int0 = Bef#sr{reg=Regs0,stk=Stk,res=Res}, - X0_v0 = x0_vars(As, Fb, Lf, Vdb), - {Keis,{Aft,_,St1}} = +%% +%% Do a specialized code generation for a basic block of #put{} +%% instructions (that don't do any garbage collection) followed by a +%% call, break, or return. +%% +%% 'Arguments' is a list of the variables that must be loaded into +%% consecutive X registers before the last instruction in the block. +%% The point of this specialized code generation is to try put the +%% all of the variables in 'Arguments' into the correct X register +%% to begin with, instead of putting them into the first available +%% X register and having to move them to the correct X register +%% later. +%% +%% To achieve that, we attempt to reserve the X registers that the +%% variables in 'Arguments' will need to be in when the block ends. +%% +%% To make it more likely that reservations will be successful, we +%% will try to save variables that need to be saved to the stack as +%% early as possible (if an X register needed by a variable in +%% Arguments is occupied by another variable, the value in the +%% X register can be evicted if it is saved on the stack). +%% +%% We will take care not to increase the size of stack frame compared +%% to what the standard code generator would have done (that is, to +%% save all X registers at the last possible moment). We will do that +%% by extending the stack frame to the minimal size needed to save +%% all that needs to be saved using extend_stack/4, and use +%% save_carefully/4 during code generation to only save the variables +%% that can be saved without growing the stack frame. + +cg_basic_block(Kes, Lf, As, Vdb, Bef, St0) -> + Int0 = reserve_arg_regs(As, Bef), + Int = extend_stack(Int0, Lf, Lf+1, Vdb), + {Keis,{Aft,St1}} = flatmapfoldl(fun(Ke, St) -> cg_basic_block(Ke, St, Lf, Vdb) end, - {Int0,X0_v0,St0}, need_heap(Kes, Fb)), + {Int,St0}, need_heap(Kes)), {Keis,Aft,St1}. -cg_basic_block(#l{ke={need_heap,_}}=Ke, {Inta,X0v,Sta}, _Lf, Vdb) -> - {Keis,Intb,Stb} = cg(Ke, Vdb, Inta, Sta), - {Keis, {Intb,X0v,Stb}}; -cg_basic_block(Ke, {Inta,X0_v1,Sta}, Lf, Vdb) -> - {Sis,Intb} = save_carefully(Inta, Ke#l.i, Lf+1, Vdb), - {X0_v2,Intc} = allocate_x0(X0_v1, Ke#l.i, Intb), - Intd = reserve(Intc), - {Keis,Inte,Stb} = cg(Ke, Vdb, Intd, Sta), - {Sis ++ Keis, {Inte,X0_v2,Stb}}. - -make_reservation([], _) -> []; -make_reservation([{var,V}|As], I) -> [{I,V}|make_reservation(As, I+1)]; -make_reservation([A|As], I) -> [{I,A}|make_reservation(As, I+1)]. - -reserve(Sr) -> Sr#sr{reg=reserve(Sr#sr.res, Sr#sr.reg, Sr#sr.stk)}. - -reserve([{I,V}|Rs], [free|Regs], Stk) -> [{reserved,I,V}|reserve(Rs, Regs, Stk)]; -reserve([{I,V}|Rs], [{I,V}|Regs], Stk) -> [{I,V}|reserve(Rs, Regs, Stk)]; -reserve([{I,V}|Rs], [{I,Var}|Regs], Stk) -> +cg_basic_block(#cg_need_heap{}=Ke, {Bef,St0}, _Lf, Vdb) -> + {Keis,Aft,St1} = cg(Ke, Vdb, Bef, St0), + {Keis,{Aft,St1}}; +cg_basic_block(Ke, {Bef,St0}, Lf, Vdb) -> + #l{i=I} = get_kanno(Ke), + + %% Save all we can to increase the possibility that reserving + %% registers will succeed. + {Sis,Int0} = save_carefully(Bef, I, Lf+1, Vdb), + Int1 = reserve(Int0), + {Keis,Aft,St1} = cg(Ke, Vdb, Int1, St0), + {Sis ++ Keis,{Aft,St1}}. + +%% reserve_arg_regs([Argument], Bef) -> Aft. +%% Try to reserve the X registers for all arguments. All registers +%% that we wish to reserve will be saved in Bef#sr.res. + +reserve_arg_regs(As, Bef) -> + Res = reserve_arg_regs_1(As, 0), + reserve(Bef#sr{res=Res}). + +reserve_arg_regs_1([#k_var{name=V}|As], I) -> + [{I,V}|reserve_arg_regs_1(As, I+1)]; +reserve_arg_regs_1([A|As], I) -> + [{I,A}|reserve_arg_regs_1(As, I+1)]; +reserve_arg_regs_1([], _) -> []. + +%% reserve(Bef) -> Aft. +%% Try to reserve more registers. The registers we wish to reserve +%% are found in Bef#sr.res. + +reserve(#sr{reg=Regs,stk=Stk,res=Res}=Sr) -> + Sr#sr{reg=reserve_1(Res, Regs, Stk)}. + +reserve_1([{I,V}|Rs], [free|Regs], Stk) -> + [{reserved,I,V}|reserve_1(Rs, Regs, Stk)]; +reserve_1([{I,V}|Rs], [{I,V}|Regs], Stk) -> + [{I,V}|reserve_1(Rs, Regs, Stk)]; +reserve_1([{I,V}|Rs], [{I,Var}|Regs], Stk) -> case on_stack(Var, Stk) of - true -> [{reserved,I,V}|reserve(Rs, Regs, Stk)]; - false -> [{I,Var}|reserve(Rs, Regs, Stk)] + true -> [{reserved,I,V}|reserve_1(Rs, Regs, Stk)]; + false -> [{I,Var}|reserve_1(Rs, Regs, Stk)] end; -reserve([{I,V}|Rs], [{reserved,I,_}|Regs], Stk) -> - [{reserved,I,V}|reserve(Rs, Regs, Stk)]; -%reserve([{I,V}|Rs], [Other|Regs], Stk) -> [Other|reserve(Rs, Regs, Stk)]; -reserve([{I,V}|Rs], [], Stk) -> [{reserved,I,V}|reserve(Rs, [], Stk)]; -reserve([], Regs, _) -> Regs. - -extend_stack(Bef, Fb, Lf, Vdb) -> - Stk0 = clear_dead_stk(Bef#sr.stk, Fb, Vdb), - Saves = [V || {V,F,L} <- Vdb, - F < Fb, - L >= Lf, - not on_stack(V, Stk0)], - Stk1 = foldl(fun (V, Stk) -> put_stack(V, Stk) end, Stk0, Saves), - Bef#sr.stk ++ lists:duplicate(length(Stk1) - length(Bef#sr.stk), free). - -save_carefully(Bef, Fb, Lf, Vdb) -> - Stk = Bef#sr.stk, - %% New variables that are in use but not on stack. - New = [VFL || {V,F,L} = VFL <- Vdb, - F < Fb, - L >= Lf, - not on_stack(V, Stk)], - Saves = [V || {V,_,_} <- keysort(2, New)], - save_carefully(Saves, Bef, []). - -save_carefully([], Bef, Acc) -> {reverse(Acc),Bef}; -save_carefully([V|Vs], Bef, Acc) -> - case put_stack_carefully(V, Bef#sr.stk) of - error -> {reverse(Acc),Bef}; +reserve_1([{I,V}|Rs], [{reserved,I,_}|Regs], Stk) -> + [{reserved,I,V}|reserve_1(Rs, Regs, Stk)]; +reserve_1([{I,V}|Rs], [], Stk) -> + [{reserved,I,V}|reserve_1(Rs, [], Stk)]; +reserve_1([], Regs, _) -> Regs. + +%% extend_stack(Bef, FirstBefore, LastFrom, Vdb) -> Aft. +%% Extend the stack enough to fit all variables alive past LastFrom +%% and not already on the stack. + +extend_stack(#sr{stk=Stk0}=Bef, Fb, Lf, Vdb) -> + Stk1 = clear_dead_stk(Stk0, Fb, Vdb), + New = new_not_on_stack(Stk1, Fb, Lf, Vdb), + Stk2 = foldl(fun ({V,_,_}, Stk) -> put_stack(V, Stk) end, Stk1, New), + Stk = Stk0 ++ lists:duplicate(length(Stk2) - length(Stk0), free), + Bef#sr{stk=Stk}. + +%% save_carefully(Bef, FirstBefore, LastFrom, Vdb) -> {[SaveVar],Aft}. +%% Save variables which are used past current point and which are not +%% already on the stack, but only if the variables can be saved without +%% growing the stack frame. + +save_carefully(#sr{stk=Stk}=Bef, Fb, Lf, Vdb) -> + New0 = new_not_on_stack(Stk, Fb, Lf, Vdb), + New = keysort(2, New0), + save_carefully_1(New, Bef, []). + +save_carefully_1([{V,_,_}|Vs], #sr{reg=Regs,stk=Stk0}=Bef, Acc) -> + case put_stack_carefully(V, Stk0) of + error -> + {reverse(Acc),Bef}; Stk1 -> - SrcReg = fetch_reg(V, Bef#sr.reg), + SrcReg = fetch_reg(V, Regs), Move = {move,SrcReg,fetch_stack(V, Stk1)}, {x,_} = SrcReg, %Assertion - must be X register. - save_carefully(Vs, Bef#sr{stk=Stk1}, [Move|Acc]) - end. + save_carefully_1(Vs, Bef#sr{stk=Stk1}, [Move|Acc]) + end; +save_carefully_1([], Bef, Acc) -> + {reverse(Acc),Bef}. -x0_vars([], _Fb, _Lf, _Vdb) -> []; -x0_vars([{var,V}|_], Fb, _Lf, Vdb) -> - {V,F,_L} = VFL = vdb_find(V, Vdb), - x0_vars1([VFL], Fb, F, Vdb); -x0_vars([X0|_], Fb, Lf, Vdb) -> - x0_vars1([{X0,Lf,Lf}], Fb, Lf, Vdb). - -x0_vars1(X0, Fb, Xf, Vdb) -> - Vs0 = [VFL || {_V,F,L}=VFL <- Vdb, - F >= Fb, - L < Xf], - Vs1 = keysort(3, Vs0), - keysort(2, X0++Vs1). - -allocate_x0([], _, Bef) -> {[],Bef#sr{res=[]}}; -allocate_x0([{_,_,L}|Vs], I, Bef) when L =< I -> - allocate_x0(Vs, I, Bef); -allocate_x0([{V,_F,_L}=VFL|Vs], _, Bef) -> - {[VFL|Vs],Bef#sr{res=reserve_x0(V, Bef#sr.res)}}. - -reserve_x0(V, [_|Res]) -> [{0,V}|Res]; -reserve_x0(V, []) -> [{0,V}]. +%% top_level_block([Instruction], Bef, MaxRegs, St) -> [Instruction]. +%% For the top-level block, allocate a stack frame a necessary, +%% adjust Y register numbering and instructions that return +%% from the function. top_level_block(Keis, #sr{stk=[]}, _MaxRegs, #cg{need_frame=false}) -> Keis; @@ -640,21 +1067,27 @@ turn_yreg(Other, _MaxY) -> %% wrong. These are different as in the second case there is no need %% to try the next type, it will always fail. -select_cg(#l{ke={type_clause,cons,[S]}}, {var,V}, Tf, Vf, Bef, St) -> +select_cg(#k_type_clause{type=Type,values=Vs}, Var, Tf, Vf, Bef, St) -> + #k_var{name=V} = Var, + select_cg(Type, Vs, V, Tf, Vf, Bef, St). + +select_cg(k_cons, [S], V, Tf, Vf, Bef, St) -> select_cons(S, V, Tf, Vf, Bef, St); -select_cg(#l{ke={type_clause,nil,[S]}}, {var,V}, Tf, Vf, Bef, St) -> +select_cg(k_nil, [S], V, Tf, Vf, Bef, St) -> select_nil(S, V, Tf, Vf, Bef, St); -select_cg(#l{ke={type_clause,binary,[S]}}, {var,V}, Tf, Vf, Bef, St) -> +select_cg(k_binary, [S], V, Tf, Vf, Bef, St) -> select_binary(S, V, Tf, Vf, Bef, St); -select_cg(#l{ke={type_clause,bin_seg,S}}, {var,V}, Tf, _Vf, Bef, St) -> +select_cg(k_bin_seg, S, V, Tf, _Vf, Bef, St) -> select_bin_segs(S, V, Tf, Bef, St); -select_cg(#l{ke={type_clause,bin_int,S}}, {var,V}, Tf, _Vf, Bef, St) -> +select_cg(k_bin_int, S, V, Tf, _Vf, Bef, St) -> select_bin_segs(S, V, Tf, Bef, St); -select_cg(#l{ke={type_clause,bin_end,[S]}}, {var,V}, Tf, _Vf, Bef, St) -> +select_cg(k_bin_end, [S], V, Tf, _Vf, Bef, St) -> select_bin_end(S, V, Tf, Bef, St); -select_cg(#l{ke={type_clause,map,S}}, {var,V}, Tf, Vf, Bef, St) -> +select_cg(k_map, S, V, Tf, Vf, Bef, St) -> select_map(S, V, Tf, Vf, Bef, St); -select_cg(#l{ke={type_clause,Type,Scs}}, {var,V}, Tf, Vf, Bef, St0) -> +select_cg(k_literal, S, V, Tf, Vf, Bef, St) -> + select_literal(S, V, Tf, Vf, Bef, St); +select_cg(Type, Scs, V, Tf, Vf, Bef, St0) -> {Vis,{Aft,St1}} = mapfoldl(fun (S, {Int,Sta}) -> {Val,Is,Inta,Stb} = select_val(S, V, Vf, Bef, Sta), @@ -664,22 +1097,29 @@ select_cg(#l{ke={type_clause,Type,Scs}}, {var,V}, Tf, Vf, Bef, St0) -> {Vls,Sis,St2} = select_labels(OptVls, St1, [], []), {select_val_cg(Type, fetch_var(V, Bef), Vls, Tf, Vf, Sis), Aft, St2}. -select_val_cg(tuple, R, [Arity,{f,Lbl}], Tf, Vf, [{label,Lbl}|Sis]) -> +select_val_cg(k_tuple, R, [Arity,{f,Lbl}], Tf, Vf, [{label,Lbl}|Sis]) -> [{test,is_tuple,{f,Tf},[R]},{test,test_arity,{f,Vf},[R,Arity]}|Sis]; -select_val_cg(tuple, R, Vls, Tf, Vf, Sis) -> +select_val_cg(k_tuple, R, Vls, Tf, Vf, Sis) -> [{test,is_tuple,{f,Tf},[R]},{select_tuple_arity,R,{f,Vf},{list,Vls}}|Sis]; select_val_cg(Type, R, [Val, {f,Lbl}], Fail, Fail, [{label,Lbl}|Sis]) -> - [{test,is_eq_exact,{f,Fail},[R,{Type,Val}]}|Sis]; + [{test,is_eq_exact,{f,Fail},[R,{type(Type),Val}]}|Sis]; select_val_cg(Type, R, [Val, {f,Lbl}], Tf, Vf, [{label,Lbl}|Sis]) -> [{test,select_type_test(Type),{f,Tf},[R]}, - {test,is_eq_exact,{f,Vf},[R,{Type,Val}]}|Sis]; + {test,is_eq_exact,{f,Vf},[R,{type(Type),Val}]}|Sis]; select_val_cg(Type, R, Vls0, Tf, Vf, Sis) -> - Vls1 = [case Value of {f,_Lbl} -> Value; _ -> {Type,Value} end || Value <- Vls0], + Vls1 = [case Value of + {f,_Lbl} -> Value; + _ -> {type(Type),Value} + end || Value <- Vls0], [{test,select_type_test(Type),{f,Tf},[R]}, {select_val,R,{f,Vf},{list,Vls1}}|Sis]. - -select_type_test(integer) -> is_integer; -select_type_test(atom) -> is_atom; -select_type_test(float) -> is_float. + +type(k_atom) -> atom; +type(k_float) -> float; +type(k_int) -> integer. + +select_type_test(k_int) -> is_integer; +select_type_test(k_atom) -> is_atom; +select_type_test(k_float) -> is_float. combine([{Is,Vs1}, {Is,Vs2}|Vis]) -> combine([{Is,Vs1 ++ Vs2}|Vis]); combine([V|Vis]) -> [V|combine(Vis)]; @@ -695,36 +1135,49 @@ add_vls([V|Vs], Lbl, Acc) -> add_vls(Vs, Lbl, [V, {f,Lbl}|Acc]); add_vls([], _, Acc) -> Acc. -select_cons(#l{ke={val_clause,{cons,Es},B},i=I,vdb=Vdb}, V, Tf, Vf, Bef, St0) -> +select_literal(S, V, Tf, Vf, Bef, St) -> + Reg = fetch_var(V, Bef), + F = fun(ValClause, Fail, St0) -> + {Val,Is,Aft,St1} = select_val(ValClause, V, Vf, Bef, St0), + Test = {test,is_eq_exact,{f,Fail},[Reg,{literal,Val}]}, + {[Test|Is],Aft,St1} + end, + match_fmf(F, Tf, St, S). + +select_cons(#k_val_clause{val=#k_cons{hd=Hd,tl=Tl},body=B,anno=#l{i=I,vdb=Vdb}}, + V, Tf, Vf, Bef, St0) -> + Es = [Hd,Tl], {Eis,Int,St1} = select_extract_cons(V, Es, I, Vdb, Bef, St0), {Bis,Aft,St2} = match_cg(B, Vf, Int, St1), {[{test,is_nonempty_list,{f,Tf},[fetch_var(V, Bef)]}] ++ Eis ++ Bis,Aft,St2}. -select_nil(#l{ke={val_clause,nil,B}}, V, Tf, Vf, Bef, St0) -> +select_nil(#k_val_clause{val=#k_nil{},body=B}, V, Tf, Vf, Bef, St0) -> {Bis,Aft,St1} = match_cg(B, Vf, Bef, St0), {[{test,is_nil,{f,Tf},[fetch_var(V, Bef)]}] ++ Bis,Aft,St1}. -select_binary(#l{ke={val_clause,{binary,{var,V}},B},i=I,vdb=Vdb}, - V, Tf, Vf, Bef, St0) -> +select_binary(#k_val_clause{val=#k_binary{segs=#k_var{name=V}},body=B, + anno=#l{i=I,vdb=Vdb}}, V, Tf, Vf, Bef, St0) -> + #cg{ctx=OldCtx} = St0, Int0 = clear_dead(Bef#sr{reg=Bef#sr.reg}, I, Vdb), - {Bis0,Aft,St1} = match_cg(B, Vf, Int0, St0), + {Bis0,Aft,St1} = match_cg(B, Vf, Int0, St0#cg{ctx=V}), CtxReg = fetch_var(V, Int0), Live = max_reg(Bef#sr.reg), Bis1 = [{test,bs_start_match2,{f,Tf},Live,[CtxReg,V],CtxReg}, {bs_save2,CtxReg,{V,V}}|Bis0], Bis = finish_select_binary(Bis1), - {Bis,Aft,St1}; -select_binary(#l{ke={val_clause,{binary,{var,Ivar}},B},i=I,vdb=Vdb}, - V, Tf, Vf, Bef, St0) -> + {Bis,Aft,St1#cg{ctx=OldCtx}}; +select_binary(#k_val_clause{val=#k_binary{segs=#k_var{name=Ivar}},body=B, + anno=#l{i=I,vdb=Vdb}}, V, Tf, Vf, Bef, St0) -> + #cg{ctx=OldCtx} = St0, Regs = put_reg(Ivar, Bef#sr.reg), Int0 = clear_dead(Bef#sr{reg=Regs}, I, Vdb), - {Bis0,Aft,St1} = match_cg(B, Vf, Int0, St0), + {Bis0,Aft,St1} = match_cg(B, Vf, Int0, St0#cg{ctx=Ivar}), CtxReg = fetch_var(Ivar, Int0), Live = max_reg(Bef#sr.reg), Bis1 = [{test,bs_start_match2,{f,Tf},Live,[fetch_var(V, Bef),Ivar],CtxReg}, {bs_save2,CtxReg,{Ivar,Ivar}}|Bis0], Bis = finish_select_binary(Bis1), - {Bis,Aft,St1}. + {Bis,Aft,St1#cg{ctx=OldCtx}}. finish_select_binary([{bs_save2,R,Point}=I,{bs_restore2,R,Point}|Is]) -> [I|finish_select_binary(Is)]; @@ -746,9 +1199,16 @@ select_bin_segs(Scs, Ivar, Tf, Bef, St) -> select_bin_seg(S, Ivar, Fail, Bef, Sta) end, Tf, St, Scs). -select_bin_seg(#l{ke={val_clause,{bin_seg,Ctx,Size,U,T,Fs0,Es},B},i=I,vdb=Vdb,a=A}, - Ivar, Fail, Bef, St0) -> +select_bin_seg(#k_val_clause{val=#k_bin_seg{size=Size,unit=U,type=T, + seg=Seg,flags=Fs0,next=Next}, + body=B, + anno=#l{i=I,vdb=Vdb,a=A}}, Ivar, Fail, Bef, St0) -> + Ctx = St0#cg.ctx, Fs = [{anno,A}|Fs0], + Es = case Next of + [] -> [Seg]; + _ -> [Seg,Next] + end, {Mis,Int,St1} = select_extract_bin(Es, Size, U, T, Fs, Fail, I, Vdb, Bef, Ctx, B, St0), {Bis,Aft,St2} = match_cg(B, Fail, Int, St1), @@ -761,9 +1221,12 @@ select_bin_seg(#l{ke={val_clause,{bin_seg,Ctx,Size,U,T,Fs0,Es},B},i=I,vdb=Vdb,a= [{bs_restore2,CtxReg,{Ctx,Ivar}}|Mis++Bis] end, {Is,Aft,St2}; -select_bin_seg(#l{ke={val_clause,{bin_int,Ctx,Sz,U,Fs,Val,Es},B},i=I,vdb=Vdb}, - Ivar, Fail, Bef, St0) -> - {Mis,Int,St1} = select_extract_int(Es, Val, Sz, U, Fs, Fail, +select_bin_seg(#k_val_clause{val=#k_bin_int{size=Sz,unit=U,flags=Fs, + val=Val,next=Next}, + body=B, + anno=#l{i=I,vdb=Vdb}}, Ivar, Fail, Bef, St0) -> + Ctx = St0#cg.ctx, + {Mis,Int,St1} = select_extract_int(Next, Val, Sz, U, Fs, Fail, I, Vdb, Bef, Ctx, St0), {Bis,Aft,St2} = match_cg(B, Fail, Int, St1), CtxReg = fetch_var(Ctx, Bef), @@ -784,7 +1247,7 @@ select_bin_seg(#l{ke={val_clause,{bin_int,Ctx,Sz,U,Fs,Val,Es},B},i=I,vdb=Vdb}, end, {[{bs_restore2,CtxReg,{Ctx,Ivar}}|Is],Aft,St2}. -select_extract_int([{var,Tl}], Val, {integer,Sz}, U, Fs, Vf, +select_extract_int(#k_var{name=Tl}, Val, #k_int{val=Sz}, U, Fs, Vf, I, Vdb, Bef, Ctx, St) -> Bits = U*Sz, Bin = case member(big, Fs) of @@ -805,7 +1268,7 @@ select_extract_int([{var,Tl}], Val, {integer,Sz}, U, Fs, Vf, end, {Is,clear_dead(Bef, I, Vdb),St}. -select_extract_bin([{var,Hd},{var,Tl}], Size0, Unit, Type, Flags, Vf, +select_extract_bin([#k_var{name=Hd},#k_var{name=Tl}], Size0, Unit, Type, Flags, Vf, I, Vdb, Bef, Ctx, _Body, St) -> SizeReg = get_bin_size_reg(Size0, Bef), {Es,Aft} = @@ -828,11 +1291,11 @@ select_extract_bin([{var,Hd},{var,Tl}], Size0, Unit, Type, Flags, Vf, {bs_save2,CtxReg,{Ctx,Tl}}],Int1} end, {Es,clear_dead(Aft, I, Vdb),St}; -select_extract_bin([{var,Hd}], Size, Unit, binary, Flags, Vf, +select_extract_bin([#k_var{name=Hd}], Size, Unit, binary, Flags, Vf, I, Vdb, Bef, Ctx, Body, St) -> %% Match the last segment of a binary. We KNOW that the size %% must be 'all'. - Size = {atom,all}, %Assertion. + #k_atom{val=all} = Size, %Assertion. {Es,Aft} = case vdb_find(Hd, Vdb) of {_,_,Lhd} when Lhd =< I -> @@ -857,7 +1320,7 @@ select_extract_bin([{var,Hd}], Size, Unit, binary, Flags, Vf, Name = bs_get_binary2, Live = max_reg(Bef#sr.reg), {[{test,Name,{f,Vf},Live, - [CtxReg,Size,Unit,{field_flags,Flags}],Rhd}], + [CtxReg,atomic(Size),Unit,{field_flags,Flags}],Rhd}], Int1}; true -> %% Since the matching context will not be used again, @@ -872,36 +1335,42 @@ select_extract_bin([{var,Hd}], Size, Unit, binary, Flags, Vf, Name = bs_get_binary2, Live = max_reg(Int1#sr.reg), {[{test,Name,{f,Vf},Live, - [CtxReg,Size,Unit,{field_flags,Flags}],CtxReg}], + [CtxReg,atomic(Size),Unit,{field_flags,Flags}],CtxReg}], Int1} end end, {Es,clear_dead(Aft, I, Vdb),St}. %% is_context_unused(Ke) -> true | false -%% Simple heurististic to determine whether the code that follows will -%% use the current matching context again. (The information of liveness -%% calculcated by v3_life is too conservative to be useful for this purpose.) -%% 'true' means that the code that follows will definitely not use the context -%% again (because it is a block, not guard or matching code); 'false' that we -%% are not sure (there is either a guard, or more matching, either which may -%% reference the context again). - -is_context_unused(#l{ke=Ke}) -> is_context_unused(Ke); -is_context_unused({block,_}) -> true; -is_context_unused(_) -> false. - -select_bin_end(#l{ke={val_clause,{bin_end,Ctx},B}}, - Ivar, Tf, Bef, St0) -> +%% Simple heurististic to determine whether the code that follows +%% will use the current matching context again. (The liveness +%% information is too conservative to be useful for this purpose.) +%% 'true' means that the code that follows will definitely not use +%% the context again (because it is a block, not guard or matching +%% code); 'false' that we are not sure (there could be more +%% matching). + +is_context_unused(#k_alt{then=Then}) -> + %% #k_alt{} can be used for different purposes. If the Then part + %% is a block, it means that matching has finished and is used for a guard + %% to choose between the matched clauses. + is_context_unused(Then); +is_context_unused(#cg_block{}) -> + true; +is_context_unused(_) -> + false. + +select_bin_end(#k_val_clause{val=#k_bin_end{},body=B}, Ivar, Tf, Bef, St0) -> + Ctx = St0#cg.ctx, {Bis,Aft,St2} = match_cg(B, Tf, Bef, St0), CtxReg = fetch_var(Ctx, Bef), {[{bs_restore2,CtxReg,{Ctx,Ivar}}, {test,bs_test_tail2,{f,Tf},[CtxReg,0]}|Bis],Aft,St2}. -get_bin_size_reg({var,V}, Bef) -> +get_bin_size_reg(#k_var{name=V}, Bef) -> fetch_var(V, Bef); get_bin_size_reg(Literal, _Bef) -> - Literal. + atomic(Literal). build_bs_instr(Type, Vf, CtxReg, Live, SizeReg, Unit, Flags, Rhd) -> {Format,Name} = case Type of @@ -935,11 +1404,18 @@ build_skip_instr(Type, Vf, CtxReg, Live, SizeReg, Unit, Flags) -> {test,Name,{f,Vf},[CtxReg,Live,{field_flags,Flags}]} end. -select_val(#l{ke={val_clause,{tuple,Es},B},i=I,vdb=Vdb}, V, Vf, Bef, St0) -> +select_val(#k_val_clause{val=#k_tuple{es=Es},body=B,anno=#l{i=I,vdb=Vdb}}, + V, Vf, Bef, St0) -> {Eis,Int,St1} = select_extract_tuple(V, Es, I, Vdb, Bef, St0), {Bis,Aft,St2} = match_cg(B, Vf, Int, St1), {length(Es),Eis ++ Bis,Aft,St2}; -select_val(#l{ke={val_clause,{_,Val},B}}, _V, Vf, Bef, St0) -> +select_val(#k_val_clause{val=Val0,body=B}, _V, Vf, Bef, St0) -> + Val = case Val0 of + #k_atom{val=Lit} -> Lit; + #k_float{val=Lit} -> Lit; + #k_int{val=Lit} -> Lit; + #k_literal{val=Lit} -> Lit + end, {Bis,Aft,St1} = match_cg(B, Vf, Bef, St0), {Val,Bis,Aft,St1}. @@ -948,7 +1424,7 @@ select_val(#l{ke={val_clause,{_,Val},B}}, _V, Vf, Bef, St0) -> %% Extract tuple elements, but only if they do not immediately die. select_extract_tuple(Src, Vs, I, Vdb, Bef, St) -> - F = fun ({var,V}, {Int0,Elem}) -> + F = fun (#k_var{name=V}, {Int0,Elem}) -> case vdb_find(V, Vdb) of {V,_,L} when L =< I -> {[], {Int0,Elem+1}}; _Other -> @@ -965,9 +1441,10 @@ select_extract_tuple(Src, Vs, I, Vdb, Bef, St) -> select_map(Scs, V, Tf, Vf, Bef, St0) -> Reg = fetch_var(V, Bef), {Is,Aft,St1} = - match_fmf(fun(#l{ke={val_clause,{map,exact,_,Es},B},i=I,vdb=Vdb}, Fail, St1) -> - select_map_val(V, Es, B, Fail, I, Vdb, Bef, St1) - end, Vf, St0, Scs), + match_fmf(fun(#k_val_clause{val=#k_map{op=exact,es=Es}, + body=B,anno=#l{i=I,vdb=Vdb}}, Fail, St1) -> + select_map_val(V, Es, B, Fail, I, Vdb, Bef, St1) + end, Vf, St0, Scs), {[{test,is_map,{f,Tf},[Reg]}|Is],Aft,St1}. select_map_val(V, Es, B, Fail, I, Vdb, Bef, St0) -> @@ -984,29 +1461,32 @@ select_extract_map(Src, Vs, Fail, I, Vdb, Bef, St) -> %% Assume keys are term-sorted Rsrc = fetch_var(Src, Bef), - {{HasKs,GetVs,HasVarKs,GetVarVs},Aft} = lists:foldr(fun - ({map_pair,{var,K},{var,V}},{{HasKsi,GetVsi,HasVarVsi,GetVarVsi},Int0}) -> - case vdb_find(V, Vdb) of - {V,_,L} when L =< I -> - RK = fetch_var(K,Int0), - {{HasKsi,GetVsi,[RK|HasVarVsi],GetVarVsi},Int0}; - _Other -> - Reg1 = put_reg(V, Int0#sr.reg), - Int1 = Int0#sr{reg=Reg1}, - RK = fetch_var(K,Int0), - RV = fetch_reg(V,Reg1), - {{HasKsi,GetVsi,HasVarVsi,[[RK,RV]|GetVarVsi]},Int1} - end; - ({map_pair,Key,{var,V}},{{HasKsi,GetVsi,HasVarVsi,GetVarVsi},Int0}) -> - case vdb_find(V, Vdb) of - {V,_,L} when L =< I -> - {{[Key|HasKsi],GetVsi,HasVarVsi,GetVarVsi},Int0}; - _Other -> - Reg1 = put_reg(V, Int0#sr.reg), - Int1 = Int0#sr{reg=Reg1}, - {{HasKsi,[Key,fetch_reg(V, Reg1)|GetVsi],HasVarVsi,GetVarVsi},Int1} - end - end, {{[],[],[],[]},Bef}, Vs), + {{HasKs,GetVs,HasVarKs,GetVarVs},Aft} = + foldr(fun(#k_map_pair{key=#k_var{name=K},val=#k_var{name=V}}, + {{HasKsi,GetVsi,HasVarVsi,GetVarVsi},Int0}) -> + case vdb_find(V, Vdb) of + {V,_,L} when L =< I -> + RK = fetch_var(K,Int0), + {{HasKsi,GetVsi,[RK|HasVarVsi],GetVarVsi},Int0}; + _Other -> + Reg1 = put_reg(V, Int0#sr.reg), + Int1 = Int0#sr{reg=Reg1}, + RK = fetch_var(K,Int0), + RV = fetch_reg(V,Reg1), + {{HasKsi,GetVsi,HasVarVsi,[[RK,RV]|GetVarVsi]},Int1} + end; + (#k_map_pair{key=Key,val=#k_var{name=V}}, + {{HasKsi,GetVsi,HasVarVsi,GetVarVsi},Int0}) -> + case vdb_find(V, Vdb) of + {V,_,L} when L =< I -> + {{[atomic(Key)|HasKsi],GetVsi,HasVarVsi,GetVarVsi},Int0}; + _Other -> + Reg1 = put_reg(V, Int0#sr.reg), + Int1 = Int0#sr{reg=Reg1}, + {{HasKsi,[atomic(Key),fetch_reg(V, Reg1)|GetVsi], + HasVarVsi,GetVarVsi},Int1} + end + end, {{[],[],[],[]},Bef}, Vs), Code = [{test,has_map_fields,{f,Fail},Rsrc,{list,HasKs}} || HasKs =/= []] ++ [{test,has_map_fields,{f,Fail},Rsrc,{list,[K]}} || K <- HasVarKs] ++ @@ -1015,30 +1495,36 @@ select_extract_map(Src, Vs, Fail, I, Vdb, Bef, St) -> {Code, Aft, St}. -select_extract_cons(Src, [{var,Hd}, {var,Tl}], I, Vdb, Bef, St) -> - {Es,Aft} = case {vdb_find(Hd, Vdb), vdb_find(Tl, Vdb)} of - {{_,_,Lhd}, {_,_,Ltl}} when Lhd =< I, Ltl =< I -> - %% Both head and tail are dead. No need to generate - %% any instruction. - {[], Bef}; - _ -> - %% At least one of head and tail will be used, - %% but we must always fetch both. We will call - %% clear_dead/2 to allow reuse of the register - %% in case only of them is used. - - Reg0 = put_reg(Tl, put_reg(Hd, Bef#sr.reg)), - Int0 = Bef#sr{reg=Reg0}, - Rsrc = fetch_var(Src, Int0), - Rhd = fetch_reg(Hd, Reg0), - Rtl = fetch_reg(Tl, Reg0), - Int1 = clear_dead(Int0, I, Vdb), - {[{get_list,Rsrc,Rhd,Rtl}], Int1} - end, - {Es,Aft,St}. - +select_extract_cons(Src, [#k_var{name=Hd},#k_var{name=Tl}], I, Vdb, Bef, St) -> + Rsrc = fetch_var(Src, Bef), + Int = clear_dead(Bef, I, Vdb), + {{_,_,Lhd},{_,_,Ltl}} = {vdb_find(Hd, Vdb),vdb_find(Tl, Vdb)}, + case {Lhd =< I, Ltl =< I} of + {true,true} -> + %% Both dead. + {[],Bef,St}; + {true,false} -> + %% Head dead. + Reg0 = put_reg(Tl, Bef#sr.reg), + Aft = Int#sr{reg=Reg0}, + Rtl = fetch_reg(Tl, Reg0), + {[{get_tl,Rsrc,Rtl}],Aft,St}; + {false,true} -> + %% Tail dead. + Reg0 = put_reg(Hd, Bef#sr.reg), + Aft = Int#sr{reg=Reg0}, + Rhd = fetch_reg(Hd, Reg0), + {[{get_hd,Rsrc,Rhd}],Aft,St}; + {false,false} -> + %% Both used. + Reg0 = put_reg(Tl, put_reg(Hd, Bef#sr.reg)), + Aft = Bef#sr{reg=Reg0}, + Rhd = fetch_reg(Hd, Reg0), + Rtl = fetch_reg(Tl, Reg0), + {[{get_hd,Rsrc,Rhd},{get_tl,Rsrc,Rtl}],Aft,St} + end. -guard_clause_cg(#l{ke={guard_clause,G,B},vdb=Vdb}, Fail, Bef, St0) -> +guard_clause_cg(#k_guard_clause{anno=#l{vdb=Vdb},guard=G,body=B}, Fail, Bef, St0) -> {Gis,Int,St1} = guard_cg(G, Fail, Vdb, Bef, St0), {Bis,Aft,St} = match_cg(B, Fail, Int, St1), {Gis ++ Bis,Aft,St}. @@ -1051,11 +1537,11 @@ guard_clause_cg(#l{ke={guard_clause,G,B},vdb=Vdb}, Fail, Bef, St0) -> %% the correct exit point. Primops and tests all go to the next %% instruction on success or jump to a failure label. -guard_cg(#l{ke={protected,Ts,Rs},i=I,vdb=Pdb}, Fail, _Vdb, Bef, St) -> - protected_cg(Ts, Rs, Fail, I, Pdb, Bef, St); -guard_cg(#l{ke={block,Ts},i=I,vdb=Bdb}, Fail, _Vdb, Bef, St) -> - guard_cg_list(Ts, Fail, I, Bdb, Bef, St); -guard_cg(#l{ke={test,Test,As,Inverted},i=I,vdb=_Tdb}, Fail, Vdb, Bef, St0) -> +guard_cg(#k_protected{arg=Ts,ret=Rs,anno=#l{vdb=Pdb}}, Fail, _Vdb, Bef, St) -> + protected_cg(Ts, Rs, Fail, Pdb, Bef, St); +guard_cg(#k_test{anno=#l{i=I},op=Test0,args=As,inverted=Inverted}, + Fail, Vdb, Bef, St0) -> + #k_remote{mod=#k_atom{val=erlang},name=#k_atom{val=Test}} = Test0, case Inverted of false -> test_cg(Test, As, Fail, I, Vdb, Bef, St0); @@ -1070,6 +1556,18 @@ guard_cg(G, _Fail, Vdb, Bef, St) -> %%ok = io:fwrite("cg ~w: ~p~n", [?LINE,{Aft}]), {Gis,Aft,St1}. +%% guard_cg_list([Kexpr], Fail, I, Vdb, StackReg, St) -> +%% {[Ainstr],StackReg,St}. + +guard_cg_list(Kes, Fail, Vdb, Bef, St0) -> + {Keis,{Aft,St1}} = + flatmapfoldl(fun (Ke, {Inta,Sta}) -> + {Keis,Intb,Stb} = + guard_cg(Ke, Fail, Vdb, Inta, Sta), + {Keis,{Intb,Stb}} + end, {Bef,St0}, need_heap(Kes)), + {Keis,Aft,St1}. + %% protected_cg([Kexpr], [Ret], Fail, I, Vdb, Bef, St) -> {[Ainstr],Aft,St}. %% Do a protected. Protecteds without return values are just done %% for effect, the return value is not checked, success passes on to @@ -1077,19 +1575,18 @@ guard_cg(G, _Fail, Vdb, Bef, St) -> %% return values then these must be set to 'false' on failure, %% control always passes to the next instruction. -protected_cg(Ts, [], Fail, I, Vdb, Bef, St0) -> +protected_cg(Ts, [], Fail, Vdb, Bef, St0) -> %% Protect these calls, revert when done. - {Tis,Aft,St1} = guard_cg_list(Ts, Fail, I, Vdb, Bef, - St0#cg{bfail=Fail}), + {Tis,Aft,St1} = guard_cg_list(Ts, Fail, Vdb, Bef, St0#cg{bfail=Fail}), {Tis,Aft,St1#cg{bfail=St0#cg.bfail}}; -protected_cg(Ts, Rs, _Fail, I, Vdb, Bef, St0) -> +protected_cg(Ts, Rs, _Fail, Vdb, Bef, St0) -> {Pfail,St1} = new_label(St0), {Psucc,St2} = new_label(St1), - {Tis,Aft,St3} = guard_cg_list(Ts, Pfail, I, Vdb, Bef, + {Tis,Aft,St3} = guard_cg_list(Ts, Pfail, Vdb, Bef, St2#cg{bfail=Pfail}), %%ok = io:fwrite("cg ~w: ~p~n", [?LINE,{Rs,I,Vdb,Aft}]), %% Set return values to false. - Mis = [{move,{atom,false},fetch_var(V,Aft)}||{var,V} <- Rs], + Mis = [{move,{atom,false},fetch_var(V,Aft)}||#k_var{name=V} <- Rs], {Tis ++ [{jump,{f,Psucc}}, {label,Pfail}] ++ Mis ++ [{label,Psucc}], Aft,St3#cg{bfail=St0#cg.bfail}}. @@ -1114,7 +1611,7 @@ test_cg(is_map, [A], Fail, I, Vdb, Bef, St) -> Arg = cg_reg_arg_prefer_y(A, Bef), Aft = clear_dead(Bef, I, Vdb), {[{test,is_map,{f,Fail},[Arg]}],Aft,St}; -test_cg(is_boolean, [{atom,Val}], Fail, I, Vdb, Bef, St) -> +test_cg(is_boolean, [#k_atom{val=Val}], Fail, I, Vdb, Bef, St) -> Aft = clear_dead(Bef, I, Vdb), Is = case is_boolean(Val) of true -> []; @@ -1126,18 +1623,6 @@ test_cg(Test, As, Fail, I, Vdb, Bef, St) -> Aft = clear_dead(Bef, I, Vdb), {[beam_utils:bif_to_test(Test, Args, {f,Fail})],Aft,St}. -%% guard_cg_list([Kexpr], Fail, I, Vdb, StackReg, St) -> -%% {[Ainstr],StackReg,St}. - -guard_cg_list(Kes, Fail, I, Vdb, Bef, St0) -> - {Keis,{Aft,St1}} = - flatmapfoldl(fun (Ke, {Inta,Sta}) -> - {Keis,Intb,Stb} = - guard_cg(Ke, Fail, Vdb, Inta, Sta), - {Keis,{Intb,Stb}} - end, {Bef,St0}, need_heap(Kes, I)), - {Keis,Aft,St1}. - %% match_fmf(Fun, LastFail, State, [Clause]) -> {Is,Aft,State}. %% This is a special flatmapfoldl for match code gen where we %% generate a "failure" label for each clause. The last clause uses @@ -1160,7 +1645,7 @@ match_fmf(F, LastFail, St0, [H|T]) -> %% frame size. Finally the actual call is made. Call then needs the %% return values filled in. -call_cg({var,_V} = Var, As, Rs, Le, Vdb, Bef, St0) -> +call_cg(#k_var{}=Var, As, Rs, Le, Vdb, Bef, St0) -> {Sis,Int} = cg_setup_call(As++[Var], Bef, Le#l.i, Vdb), %% Put return values in registers. Reg = load_vars(Rs, clear_regs(Int#sr.reg)), @@ -1169,9 +1654,8 @@ call_cg({var,_V} = Var, As, Rs, Le, Vdb, Bef, St0) -> {Frees,Aft} = free_dead(clear_dead(Int#sr{reg=Reg}, Le#l.i, Vdb)), {Sis ++ Frees ++ [line(Le),{call_fun,Arity}],Aft, need_stack_frame(St0)}; -call_cg({remote,Mod,Name}, As, Rs, Le, Vdb, Bef, St0) - when element(1, Mod) =:= var; - element(1, Name) =:= var -> +call_cg(#k_remote{mod=Mod,name=Name}, As, Rs, Le, Vdb, Bef, St0) + when is_record(Mod, k_var); is_record(Name, k_var) -> {Sis,Int} = cg_setup_call(As++[Mod,Name], Bef, Le#l.i, Vdb), %% Put return values in registers. Reg = load_vars(Rs, clear_regs(Int#sr.reg)), @@ -1189,8 +1673,9 @@ call_cg(Func, As, Rs, Le, Vdb, Bef, St0) -> %% %% move {atom,ok} DestReg %% jump FailureLabel - {remote,{atom,erlang},{atom,error}} = Func, %Assertion. - [{var,DestVar}] = Rs, + #k_remote{mod=#k_atom{val=erlang}, + name=#k_atom{val=error}} = Func, %Assertion. + [#k_var{name=DestVar}] = Rs, Int0 = clear_dead(Bef, Le#l.i, Vdb), Reg = put_reg(DestVar, Int0#sr.reg), Int = Int0#sr{reg=Reg}, @@ -1209,11 +1694,11 @@ call_cg(Func, As, Rs, Le, Vdb, Bef, St0) -> {Sis ++ Frees ++ [line(Le)|Call],Aft,St1} end. -build_call({remote,{atom,erlang},{atom,'!'}}, 2, St0) -> +build_call(#k_remote{mod=#k_atom{val=erlang},name=#k_atom{val='!'}}, 2, St0) -> {[send],need_stack_frame(St0)}; -build_call({remote,{atom,Mod},{atom,Name}}, Arity, St0) -> +build_call(#k_remote{mod=#k_atom{val=Mod},name=#k_atom{val=Name}}, Arity, St0) -> {[{call_ext,Arity,{extfunc,Mod,Name,Arity}}],need_stack_frame(St0)}; -build_call(Name, Arity, St0) when is_atom(Name) -> +build_call(#k_local{name=Name}, Arity, St0) when is_atom(Name) -> {Lbl,St1} = local_func_label(Name, Arity, need_stack_frame(St0)), {[{call,Arity,{f,Lbl}}],St1}. @@ -1229,16 +1714,15 @@ free_dead([Any|Stk], Y, Instr, StkAcc) -> free_dead(Stk, Y+1, Instr, [Any|StkAcc]); free_dead([], _, Instr, StkAcc) -> {Instr,reverse(StkAcc)}. -enter_cg({var,_V} = Var, As, Le, Vdb, Bef, St0) -> +enter_cg(#k_var{} = Var, As, Le, Vdb, Bef, St0) -> {Sis,Int} = cg_setup_call(As++[Var], Bef, Le#l.i, Vdb), %% Build complete code and final stack/register state. Arity = length(As), {Sis ++ [line(Le),{call_fun,Arity},return], clear_dead(Int#sr{reg=clear_regs(Int#sr.reg)}, Le#l.i, Vdb), need_stack_frame(St0)}; -enter_cg({remote,Mod,Name}, As, Le, Vdb, Bef, St0) - when element(1, Mod) =:= var; - element(1, Name) =:= var -> +enter_cg(#k_remote{mod=Mod,name=Name}, As, Le, Vdb, Bef, St0) + when is_record(Mod, k_var); is_record(Name, k_var) -> {Sis,Int} = cg_setup_call(As++[Mod,Name], Bef, Le#l.i, Vdb), %% Build complete code and final stack/register state. Arity = length(As), @@ -1256,19 +1740,19 @@ enter_cg(Func, As, Le, Vdb, Bef, St0) -> clear_dead(Int#sr{reg=clear_regs(Int#sr.reg)}, Le#l.i, Vdb), St1}. -build_enter({remote,{atom,erlang},{atom,'!'}}, 2, St0) -> +build_enter(#k_remote{mod=#k_atom{val=erlang},name=#k_atom{val='!'}}, 2, St0) -> {[send,return],need_stack_frame(St0)}; -build_enter({remote,{atom,Mod},{atom,Name}}, Arity, St0) -> +build_enter(#k_remote{mod=#k_atom{val=Mod},name=#k_atom{val=Name}}, Arity, St0) -> St1 = case trap_bif(Mod, Name, Arity) of true -> need_stack_frame(St0); false -> St0 end, {[{call_ext_only,Arity,{extfunc,Mod,Name,Arity}}],St1}; -build_enter(Name, Arity, St0) when is_atom(Name) -> +build_enter(#k_local{name=Name}, Arity, St0) when is_atom(Name) -> {Lbl,St1} = local_func_label(Name, Arity, St0), {[{call_only,Arity,{f,Lbl}}],St1}. -enter_line({remote,{atom,Mod},{atom,Name}}, Arity, Le) -> +enter_line(#k_remote{mod=#k_atom{val=Mod},name=#k_atom{val=Name}}, Arity, Le) -> case erl_bifs:is_safe(Mod, Name, Arity) of false -> %% Tail-recursive call, possibly to a BIF. @@ -1316,17 +1800,28 @@ trap_bif(erlang, group_leader, 2) -> true; trap_bif(erlang, exit, 2) -> true; trap_bif(_, _, _) -> false. +%% bif_cg(#k_bif{}, Le, Vdb, StackReg, State) -> +%% {[Ainstr],StackReg,State}. +%% Generate code a BIF. + +bif_cg(#k_bif{op=#k_internal{name=Name},args=As,ret=Rs}, Le, Vdb, Bef, St) -> + internal_cg(Name, As, Rs, Le, Vdb, Bef, St); +bif_cg(#k_bif{op=#k_remote{mod=#k_atom{val=erlang},name=#k_atom{val=Name}}, + args=As,ret=Rs}, Le, Vdb, Bef, St) -> + Ar = length(As), + case is_gc_bif(Name, Ar) of + false -> + bif_cg(Name, As, Rs, Le, Vdb, Bef, St); + true -> + gc_bif_cg(Name, As, Rs, Le, Vdb, Bef, St) + end. + %% internal_cg(Bif, [Arg], [Ret], Le, Vdb, StackReg, State) -> %% {[Ainstr],StackReg,State}. internal_cg(bs_context_to_binary=Instr, [Src0], [], Le, Vdb, Bef, St0) -> [Src] = cg_reg_args([Src0], Bef), - case is_register(Src) of - false -> - {[],clear_dead(Bef, Le#l.i, Vdb), St0}; - true -> - {[{Instr,Src}],clear_dead(Bef, Le#l.i, Vdb), St0} - end; + {[{Instr,Src}],clear_dead(Bef, Le#l.i, Vdb), St0}; internal_cg(dsetelement, [Index0,Tuple0,New0], _Rs, Le, Vdb, Bef, St0) -> [New,Tuple,{integer,Index1}] = cg_reg_args([New0,Tuple0,Index0], Bef), Index = Index1-1, @@ -1334,8 +1829,8 @@ internal_cg(dsetelement, [Index0,Tuple0,New0], _Rs, Le, Vdb, Bef, St0) -> clear_dead(Bef, Le#l.i, Vdb), St0}; internal_cg(make_fun, [Func0,Arity0|As], Rs, Le, Vdb, Bef, St0) -> %% This behaves more like a function call. - {atom,Func} = Func0, - {integer,Arity} = Arity0, + #k_atom{val=Func} = Func0, + #k_int{val=Arity} = Arity0, {Sis,Int} = cg_setup_call(As, Bef, Le#l.i, Vdb), Reg = load_vars(Rs, clear_regs(Int#sr.reg)), {FuncLbl,St1} = local_func_label(Func, Arity, St0), @@ -1348,14 +1843,35 @@ internal_cg(bs_init_writable=I, As, Rs, Le, Vdb, Bef, St) -> {Sis,Int} = cg_setup_call(As, Bef, Le#l.i, Vdb), Reg = load_vars(Rs, clear_regs(Int#sr.reg)), {Sis++[I],clear_dead(Int#sr{reg=Reg}, Le#l.i, Vdb),St}; +internal_cg(build_stacktrace=I, As, Rs, Le, Vdb, Bef, St) -> + %% This behaves like a function call. + {Sis,Int} = cg_setup_call(As, Bef, Le#l.i, Vdb), + Reg = load_vars(Rs, clear_regs(Int#sr.reg)), + {Sis++[I],clear_dead(Int#sr{reg=Reg}, Le#l.i, Vdb),St}; internal_cg(raise, As, Rs, Le, Vdb, Bef, St) -> %% raise can be treated like a guard BIF. - bif_cg(raise, As, Rs, Le, Vdb, Bef, St). + bif_cg(raise, As, Rs, Le, Vdb, Bef, St); +internal_cg(guard_error, [ExitCall], _Rs, Le, Vdb, Bef, St) -> + %% A call an exit BIF from inside a #k_guard_match{}. + %% Generate a standard call, but leave the register descriptors + %% alone, effectively pretending that there was no call. + #k_call{op=#k_remote{mod=#k_atom{val=Mod},name=#k_atom{val=Name}}, + args=As} = ExitCall, + Arity = length(As), + {Ms,_} = cg_call_args(As, Bef, Le#l.i, Vdb), + Call = {call_ext,Arity,{extfunc,Mod,Name,Arity}}, + Is = Ms++[line(Le),Call], + {Is,Bef,St}; +internal_cg(raw_raise=I, As, Rs, Le, Vdb, Bef, St) -> + %% This behaves like a function call. + {Sis,Int} = cg_setup_call(As, Bef, Le#l.i, Vdb), + Reg = load_vars(Rs, clear_regs(Int#sr.reg)), + {Sis++[I],clear_dead(Int#sr{reg=Reg}, Le#l.i, Vdb),St}. %% bif_cg(Bif, [Arg], [Ret], Le, Vdb, StackReg, State) -> %% {[Ainstr],StackReg,State}. -bif_cg(Bif, As, [{var,V}], Le, Vdb, Bef, St0) -> +bif_cg(Bif, As, [#k_var{name=V}], Le, Vdb, Bef, St0) -> Ars = cg_reg_args(As, Bef), %% If we are inside a catch and in a body (not in guard) and the @@ -1393,7 +1909,7 @@ bif_cg(Bif, As, [{var,V}], Le, Vdb, Bef, St0) -> %% gc_bif_cg(Bif, [Arg], [Ret], Le, Vdb, StackReg, State) -> %% {[Ainstr],StackReg,State}. -gc_bif_cg(Bif, As, [{var,V}], Le, Vdb, Bef, St0) -> +gc_bif_cg(Bif, As, [#k_var{name=V}], Le, Vdb, Bef, St0) -> Ars = cg_reg_args(As, Bef), %% If we are inside a catch and in a body (not in guard) and the @@ -1439,7 +1955,7 @@ recv_loop_cg(Te, Rvar, Rm, Tes, Rs, Le, Vdb, Bef, St0) -> %% cg_recv_mesg( ) -> {[Ainstr],Aft,St}. -cg_recv_mesg({var,R}, Rm, Tl, Bef, St0) -> +cg_recv_mesg(#k_var{name=R}, Rm, Tl, Bef, St0) -> Int0 = Bef#sr{reg=put_reg(R, Bef#sr.reg)}, Ret = fetch_reg(R, Int0#sr.reg), %% Int1 = clear_dead(Int0, I, Rm#l.vdb), @@ -1449,22 +1965,22 @@ cg_recv_mesg({var,R}, Rm, Tl, Bef, St0) -> %% cg_recv_wait(Te, Tes, I, Vdb, Int2, St3) -> {[Ainstr],Aft,St}. -cg_recv_wait({atom,infinity}, Tes, I, Bef, St0) -> +cg_recv_wait(#k_atom{val=infinity}, #cg_block{anno=Le,es=Tes}, I, Bef, St0) -> %% We know that the 'after' body will never be executed. %% But to keep the stack and register information up to date, %% we will generate the code for the 'after' body, and then discard it. - Int1 = clear_dead(Bef, I, Tes#l.vdb), - {_,Int2,St1} = cg_block(Tes#l.ke, Tes#l.i, Tes#l.vdb, - Int1#sr{reg=clear_regs(Int1#sr.reg)}, St0), + Int1 = clear_dead(Bef, I, Le#l.vdb), + {_,Int2,St1} = cg_block(Tes, Le#l.vdb, + Int1#sr{reg=clear_regs(Int1#sr.reg)}, St0), {[{wait,{f,St1#cg.recv}}],Int2,St1}; -cg_recv_wait({integer,0}, Tes, _I, Bef, St0) -> - {Tis,Int,St1} = cg_block(Tes#l.ke, Tes#l.i, Tes#l.vdb, Bef, St0), +cg_recv_wait(#k_int{val=0}, #cg_block{anno=Le,es=Tes}, _I, Bef, St0) -> + {Tis,Int,St1} = cg_block(Tes, Le#l.vdb, Bef, St0), {[timeout|Tis],Int,St1}; -cg_recv_wait(Te, Tes, I, Bef, St0) -> +cg_recv_wait(Te, #cg_block{anno=Le,es=Tes}, I, Bef, St0) -> Reg = cg_reg_arg(Te, Bef), %% Must have empty registers here! Bug if anything in registers. - Int0 = clear_dead(Bef, I, Tes#l.vdb), - {Tis,Int,St1} = cg_block(Tes#l.ke, Tes#l.i, Tes#l.vdb, + Int0 = clear_dead(Bef, I, Le#l.vdb), + {Tis,Int,St1} = cg_block(Tes, Le#l.vdb, Int0#sr{reg=clear_regs(Int0#sr.reg)}, St0), {[{wait_timeout,{f,St1#cg.recv},Reg},timeout] ++ Tis,Int,St1}. @@ -1482,7 +1998,7 @@ try_cg(Ta, Vs, Tb, Evs, Th, Rs, Le, Vdb, Bef, St0) -> {B,St1} = new_label(St0), %Body label {H,St2} = new_label(St1), %Handler label {E,St3} = new_label(St2), %End label - TryTag = Ta#l.i, + #l{i=TryTag} = get_kanno(Ta), Int1 = Bef#sr{stk=put_catch(TryTag, Bef#sr.stk)}, TryReg = fetch_stack({catch_tag,TryTag}, Int1#sr.stk), {Ais,Int2,St4} = cg(Ta, Vdb, Int1, St3#cg{break=B,in_catch=true}), @@ -1502,7 +2018,7 @@ try_cg(Ta, Vs, Tb, Evs, Th, Rs, Le, Vdb, Bef, St0) -> try_enter_cg(Ta, Vs, Tb, Evs, Th, Le, Vdb, Bef, St0) -> {B,St1} = new_label(St0), %Body label {H,St2} = new_label(St1), %Handler label - TryTag = Ta#l.i, + #l{i=TryTag} = get_kanno(Ta), Int1 = Bef#sr{stk=put_catch(TryTag, Bef#sr.stk)}, TryReg = fetch_stack({catch_tag,TryTag}, Int1#sr.stk), {Ais,Int2,St3} = cg(Ta, Vdb, Int1, St2#cg{break=B,in_catch=true}), @@ -1520,12 +2036,12 @@ try_enter_cg(Ta, Vs, Tb, Evs, Th, Le, Vdb, Bef, St0) -> %% catch_cg(CatchBlock, Ret, Le, Vdb, Bef, St) -> {[Ainstr],Aft,St}. -catch_cg(C, {var,R}, Le, Vdb, Bef, St0) -> +catch_cg(#cg_block{es=C}, #k_var{name=R}, Le, Vdb, Bef, St0) -> {B,St1} = new_label(St0), CatchTag = Le#l.i, Int1 = Bef#sr{stk=put_catch(CatchTag, Bef#sr.stk)}, CatchReg = fetch_stack({catch_tag,CatchTag}, Int1#sr.stk), - {Cis,Int2,St2} = cg_block(C, Le#l.i, Le#l.vdb, Int1, + {Cis,Int2,St2} = cg_block(C, Le#l.vdb, Int1, St1#cg{break=B,in_catch=true}), [] = Int2#sr.reg, %Assertion. Aft = Int2#sr{reg=[{0,R}],stk=drop_catch(CatchTag, Int2#sr.stk)}, @@ -1534,8 +2050,8 @@ catch_cg(C, {var,R}, Le, Vdb, Bef, St0) -> clear_dead(Aft, Le#l.i, Vdb), St2#cg{break=St1#cg.break,in_catch=St1#cg.in_catch}}. -%% set_cg([Var], Constr, Le, Vdb, Bef, St) -> {[Ainstr],Aft,St}. -%% We have to be careful how a 'set' works. First the structure is +%% put_cg([Var], Constr, Le, Vdb, Bef, St) -> {[Ainstr],Aft,St}. +%% We have to be careful how a 'put' works. First the structure is %% built, then it is filled and finally things can be cleared. The %% annotation must reflect this and make sure that the return %% variable is allocated first. @@ -1543,13 +2059,14 @@ catch_cg(C, {var,R}, Le, Vdb, Bef, St0) -> %% put_list and put_map are atomic instructions, both of %% which can safely resuse one of the source registers as target. -set_cg([{var,R}], {cons,Es}, Le, Vdb, Bef, St) -> - [S1,S2] = cg_reg_args(Es, Bef), +put_cg([#k_var{name=R}], #k_cons{hd=Hd,tl=Tl}, Le, Vdb, Bef, St) -> + [S1,S2] = cg_reg_args([Hd,Tl], Bef), Int0 = clear_dead(Bef, Le#l.i, Vdb), Int1 = Int0#sr{reg=put_reg(R, Int0#sr.reg)}, Ret = fetch_reg(R, Int1#sr.reg), {[{put_list,S1,S2,Ret}], Int1, St}; -set_cg([{var,R}], {binary,Segs}, Le, Vdb, Bef, #cg{bfail=Bfail}=St) -> +put_cg([#k_var{name=R}], #k_binary{segs=Segs}, Le, Vdb, Bef, + #cg{bfail=Bfail}=St) -> %% At run-time, binaries are constructed in three stages: %% 1) First the size of the binary is calculated. %% 2) Then the binary is allocated. @@ -1577,7 +2094,9 @@ set_cg([{var,R}], {binary,Segs}, Le, Vdb, Bef, #cg{bfail=Bfail}=St) -> {Sis++Code,Aft,St}; %% Map: single variable key. -set_cg([{var,R}], {map,Op,Map,[{map_pair,{var,_}=K,V}]}, Le, Vdb, Bef, St0) -> +put_cg([#k_var{name=R}], #k_map{op=Op,var=Map, + es=[#k_map_pair{key=#k_var{}=K,val=V}]}, + Le, Vdb, Bef, St0) -> {Sis,Int0} = maybe_adjust_stack(Bef, Le#l.i, Le#l.i+1, Vdb, St0), SrcReg = cg_reg_arg_prefer_y(Map, Int0), @@ -1592,22 +2111,23 @@ set_cg([{var,R}], {map,Op,Map,[{map_pair,{var,_}=K,V}]}, Le, Vdb, Bef, St0) -> Aft = Aft0#sr{reg=put_reg(R, Aft0#sr.reg)}, Target = fetch_reg(R, Aft#sr.reg), - {Is,St1} = set_cg_map(Line, Op, SrcReg, Target, Live, List, St0), + {Is,St1} = put_cg_map(Line, Op, SrcReg, Target, Live, List, St0), {Sis++Is,Aft,St1}; %% Map: (possibly) multiple literal keys. -set_cg([{var,R}], {map,Op,Map,Es}, Le, Vdb, Bef, St0) -> +put_cg([#k_var{name=R}], #k_map{op=Op,var=Map,es=Es}, Le, Vdb, Bef, St0) -> %% assert key literals - [] = [Var||{map_pair,{var,_}=Var,_} <- Es], + [] = [Var || #k_map_pair{key=#k_var{}=Var} <- Es], {Sis,Int0} = maybe_adjust_stack(Bef, Le#l.i, Le#l.i+1, Vdb, St0), SrcReg = cg_reg_arg_prefer_y(Map, Int0), Line = line(Le#l.a), %% fetch registers for values to be put into the map - Pairs = [{K,V} || {_,K,V} <- Es], - List = flatmap(fun({K,V}) -> [K,cg_reg_arg(V,Int0)] end, Pairs), + List = flatmap(fun(#k_map_pair{key=K,val=V}) -> + [atomic(K),cg_reg_arg(V, Int0)] + end, Es), Live = max_reg(Bef#sr.reg), @@ -1616,16 +2136,16 @@ set_cg([{var,R}], {map,Op,Map,Es}, Le, Vdb, Bef, St0) -> Aft = Aft0#sr{reg=put_reg(R, Aft0#sr.reg)}, Target = fetch_reg(R, Aft#sr.reg), - {Is,St1} = set_cg_map(Line, Op, SrcReg, Target, Live, List, St0), + {Is,St1} = put_cg_map(Line, Op, SrcReg, Target, Live, List, St0), {Sis++Is,Aft,St1}; %% Everything else. -set_cg([{var,R}], Con, Le, Vdb, Bef, St) -> +put_cg([#k_var{name=R}], Con, Le, Vdb, Bef, St) -> %% Find a place for the return register first. Int = Bef#sr{reg=put_reg(R, Bef#sr.reg)}, Ret = fetch_reg(R, Int#sr.reg), Ais = case Con of - {tuple,Es} -> + #k_tuple{es=Es} -> [{put_tuple,length(Es),Ret}] ++ cg_build_args(Es, Bef); Other -> [{move,cg_reg_arg(Other, Int),Ret}] @@ -1633,7 +2153,7 @@ set_cg([{var,R}], Con, Le, Vdb, Bef, St) -> {Ais,clear_dead(Int, Le#l.i, Vdb),St}. -set_cg_map(Line, Op0, SrcReg, Target, Live, List, St0) -> +put_cg_map(Line, Op0, SrcReg, Target, Live, List, St0) -> Bfail = St0#cg.bfail, Fail = {f,St0#cg.bfail}, Op = case Op0 of @@ -1811,24 +2331,44 @@ cg_gen_binsize([], _, _, _, _, Acc) -> Acc. %% cg_bin_opt(Code0) -> Code %% Optimize the size calculations for binary construction. -cg_bin_opt([{move,Size,D},{bs_append,Fail,D,Extra,Regs,U,Bin,Flags,D}|Is]) -> - cg_bin_opt([{bs_append,Fail,Size,Extra,Regs,U,Bin,Flags,D}|Is]); -cg_bin_opt([{move,Size,D},{bs_private_append,Fail,D,U,Bin,Flags,D}|Is]) -> - cg_bin_opt([{bs_private_append,Fail,Size,U,Bin,Flags,D}|Is]); -cg_bin_opt([{move,{integer,0},D},{bs_add,_,[D,{integer,_}=S,1],Dst}|Is]) -> - cg_bin_opt([{move,S,Dst}|Is]); -cg_bin_opt([{move,{integer,0},D},{bs_add,Fail,[D,S,U],Dst}|Is]) -> - cg_bin_opt([{bs_add,Fail,[{integer,0},S,U],Dst}|Is]); -cg_bin_opt([{move,{integer,Bytes},D},{Op,Fail,D,Extra,Regs,Flags,D}|Is]) +cg_bin_opt([{move,S1,{x,X}=D},{gc_bif,Op,Fail,Live0,As,Dst}|Is]) -> + Live = if + X + 1 =:= Live0 -> X; + true -> Live0 + end, + [{gc_bif,Op,Fail,Live,As,D}|cg_bin_opt([{move,S1,Dst}|Is])]; +cg_bin_opt([{move,_,_}=I1,{Op,_,_,_}=I2|Is]) + when Op =:= bs_utf8_size orelse Op =:= bs_utf16_size -> + [I2|cg_bin_opt([I1|Is])]; +cg_bin_opt([{bs_add,_,[{integer,0},Src,1],Dst}|Is]) -> + cg_bin_opt_1([{move,Src,Dst}|Is]); +cg_bin_opt([{bs_add,_,[Src,{integer,0},_],Dst}|Is]) -> + cg_bin_opt_1([{move,Src,Dst}|Is]); +cg_bin_opt(Is) -> + cg_bin_opt_1(Is). + +cg_bin_opt_1([{move,Size,D},{bs_append,Fail,D,Extra,Regs,U,Bin,Flags,D}|Is]) -> + [{bs_append,Fail,Size,Extra,Regs,U,Bin,Flags,D}|cg_bin_opt(Is)]; +cg_bin_opt_1([{move,Size,D},{bs_private_append,Fail,D,U,Bin,Flags,D}|Is]) -> + [{bs_private_append,Fail,Size,U,Bin,Flags,D}|cg_bin_opt(Is)]; +cg_bin_opt_1([{move,Size,D},{Op,Fail,D,Extra,Regs,Flags,D}|Is]) when Op =:= bs_init2; Op =:= bs_init_bits -> - cg_bin_opt([{Op,Fail,Bytes,Extra,Regs,Flags,D}|Is]); -cg_bin_opt([{move,Src1,Dst},{bs_add,Fail,[Dst,Src2,U],Dst}|Is]) -> - cg_bin_opt([{bs_add,Fail,[Src1,Src2,U],Dst}|Is]); -cg_bin_opt([I|Is]) -> + Bytes = case Size of + {integer,Int} -> Int; + _ -> Size + end, + [{Op,Fail,Bytes,Extra,Regs,Flags,D}|cg_bin_opt(Is)]; +cg_bin_opt_1([{move,S1,D},{bs_add,Fail,[D,S2,U],Dst}|Is]) -> + cg_bin_opt([{bs_add,Fail,[S1,S2,U],Dst}|Is]); +cg_bin_opt_1([{move,S1,D},{bs_add,Fail,[S2,D,U],Dst}|Is]) -> + cg_bin_opt([{bs_add,Fail,[S2,S1,U],Dst}|Is]); +cg_bin_opt_1([I|Is]) -> [I|cg_bin_opt(Is)]; -cg_bin_opt([]) -> []. +cg_bin_opt_1([]) -> + []. -cg_bin_put({bin_seg,[],S0,U,T,Fs,[E0,Next]}, Fail, Bef) -> +cg_bin_put(#k_bin_seg{size=S0,unit=U,type=T,flags=Fs,seg=E0,next=Next}, + Fail, Bef) -> S1 = cg_reg_arg(S0, Bef), E1 = cg_reg_arg(E0, Bef), {Format,Op} = case T of @@ -1845,7 +2385,7 @@ cg_bin_put({bin_seg,[],S0,U,T,Fs,[E0,Next]}, Fail, Bef) -> utf -> [{Op,Fail,{field_flags,Fs},E1}|cg_bin_put(Next, Fail, Bef)] end; -cg_bin_put({bin_end,[]}, _, _) -> []. +cg_bin_put(#k_bin_end{}, _, _) -> []. cg_build_args(As, Bef) -> [{put,cg_reg_arg(A, Bef)} || A <- As]. @@ -1865,13 +2405,12 @@ break_cg(Bs, Le, Vdb, Bef, St) -> {Ms ++ [{jump,{f,St#cg.break}}], Int#sr{reg=clear_regs(Int#sr.reg)},St}. -guard_break_cg(Bs, Locked, #l{i=I}, Vdb, #sr{reg=Reg0}=Bef, St) -> - RegLocked = get_locked_regs(Reg0, Locked), - #sr{reg=Reg1} = Int = clear_dead(Bef#sr{reg=RegLocked}, I, Vdb), +guard_break_cg(Bs, #l{i=I}, Vdb, #sr{reg=Reg0}=Bef, St) -> + #sr{reg=Reg1} = Int = clear_dead(Bef, I, Vdb), Reg2 = trim_free(Reg1), NumLocked = length(Reg2), Moves0 = gen_moves(Bs, Bef, NumLocked, []), - Moves = order_moves(Moves0, find_scratch_reg(RegLocked)), + Moves = order_moves(Moves0, find_scratch_reg(Reg0)), {BreakVars,_} = mapfoldl(fun(_, RegNum) -> {{RegNum,gbreakvar},RegNum+1} end, length(Reg2), Bs), @@ -1879,31 +2418,17 @@ guard_break_cg(Bs, Locked, #l{i=I}, Vdb, #sr{reg=Reg0}=Bef, St) -> Aft = Int#sr{reg=Reg}, {Moves ++ [{jump,{f,St#cg.break}}],Aft,St}. -get_locked_regs([R|Rs0], Preserve) -> - case {get_locked_regs(Rs0, Preserve),R} of - {[],{_,V}} -> - case lists:member(V, Preserve) of - true -> [R]; - false -> [] - end; - {[],_} -> - []; - {Rs,_} -> - [R|Rs] - end; -get_locked_regs([], _) -> []. - %% cg_reg_arg(Arg0, Info) -> Arg %% cg_reg_args([Arg0], Info) -> [Arg] %% Convert argument[s] into registers. Literal values are returned unchanged. cg_reg_args(As, Bef) -> [cg_reg_arg(A, Bef) || A <- As]. -cg_reg_arg({var,V}, Bef) -> fetch_var(V, Bef); -cg_reg_arg(Literal, _) -> Literal. +cg_reg_arg(#k_var{name=V}, Bef) -> fetch_var(V, Bef); +cg_reg_arg(Literal, _) -> atomic(Literal). -cg_reg_arg_prefer_y({var,V}, Bef) -> fetch_var_prefer_y(V, Bef); -cg_reg_arg_prefer_y(Literal, _) -> Literal. +cg_reg_arg_prefer_y(#k_var{name=V}, Bef) -> fetch_var_prefer_y(V, Bef); +cg_reg_arg_prefer_y(Literal, _) -> atomic(Literal). %% cg_setup_call([Arg], Bef, Cur, Vdb) -> {[Instr],Aft}. %% Do the complete setup for a call/enter. @@ -1941,9 +2466,9 @@ cg_call_args(As, Bef, I, Vdb) -> load_arg_regs(Regs, As) -> load_arg_regs(Regs, As, 0). -load_arg_regs([_|Rs], [{var,V}|As], I) -> [{I,V}|load_arg_regs(Rs, As, I+1)]; +load_arg_regs([_|Rs], [#k_var{name=V}|As], I) -> [{I,V}|load_arg_regs(Rs, As, I+1)]; load_arg_regs([_|Rs], [A|As], I) -> [{I,A}|load_arg_regs(Rs, As, I+1)]; -load_arg_regs([], [{var,V}|As], I) -> [{I,V}|load_arg_regs([], As, I+1)]; +load_arg_regs([], [#k_var{name=V}|As], I) -> [{I,V}|load_arg_regs([], As, I+1)]; load_arg_regs([], [A|As], I) -> [{I,A}|load_arg_regs([], As, I+1)]; load_arg_regs(Rs, [], _) -> Rs. @@ -1979,12 +2504,13 @@ move_unsaved([], _, Regs, Acc) -> {Acc,Regs}. gen_moves(As, Sr) -> gen_moves(As, Sr, 0, []). -gen_moves([{var,V}|As], Sr, I, Acc) -> +gen_moves([#k_var{name=V}|As], Sr, I, Acc) -> case fetch_var(V, Sr) of {x,I} -> gen_moves(As, Sr, I+1, Acc); Reg -> gen_moves(As, Sr, I+1, [{move,Reg,{x,I}}|Acc]) end; -gen_moves([A|As], Sr, I, Acc) -> +gen_moves([A0|As], Sr, I, Acc) -> + A = atomic(A0), gen_moves(As, Sr, I+1, [{move,A,{x,I}}|Acc]); gen_moves([], _, _, Acc) -> lists:keysort(3, Acc). @@ -2032,21 +2558,21 @@ break_up_cycle1(Dst, [M|Path], LastMove) -> %% clear_dead(Sr, Until, Vdb) -> Aft. %% Remove all variables in Sr which have died AT ALL so far. -clear_dead(Sr, Until, Vdb) -> - Sr#sr{reg=clear_dead_reg(Sr, Until, Vdb), - stk=clear_dead_stk(Sr#sr.stk, Until, Vdb)}. +clear_dead(#sr{stk=Stk}=Sr0, Until, Vdb) -> + Sr = Sr0#sr{reg=clear_dead_reg(Sr0, Until, Vdb), + stk=clear_dead_stk(Stk, Until, Vdb)}, + reserve(Sr). clear_dead_reg(Sr, Until, Vdb) -> - Reg = [case R of - {_I,V} = IV -> - case vdb_find(V, Vdb) of - {V,_,L} when L > Until -> IV; - _ -> free %Remove anything else - end; - {reserved,_I,_V} = Reserved -> Reserved; - free -> free - end || R <- Sr#sr.reg], - reserve(Sr#sr.res, Reg, Sr#sr.stk). + [case R of + {_I,V} = IV -> + case vdb_find(V, Vdb) of + {V,_,L} when L > Until -> IV; + _ -> free %Remove anything else + end; + {reserved,_I,_V}=Reserved -> Reserved; + free -> free + end || R <- Sr#sr.reg]. clear_dead_stk(Stk, Until, Vdb) -> [case S of @@ -2118,16 +2644,25 @@ adjust_stack(Bef, Fb, Lf, Vdb) -> save_stack(Stk0, Fb, Lf, Vdb) -> %% New variables that are in use but not on stack. - New = [VFL || {V,F,L} = VFL <- Vdb, - F < Fb, - L >= Lf, - not on_stack(V, Stk0)], + New = new_not_on_stack(Stk0, Fb, Lf, Vdb), + %% Add new variables that are not just dropped immediately. %% N.B. foldr works backwards from the end!! Saves = [V || {V,_,_} <- keysort(3, New)], Stk1 = foldr(fun (V, Stk) -> put_stack(V, Stk) end, Stk0, Saves), {Stk1,Saves}. +%% new_not_on_stack(Stack, FirstBefore, LastFrom, Vdb) -> +%% [{Variable,First,Last}] +%% Return information about all variables that are used past current +%% point and that are not already on the stack. + +new_not_on_stack(Stk, Fb, Lf, Vdb) -> + [VFL || {V,F,L} = VFL <- Vdb, + F < Fb, + L >= Lf, + not on_stack(V, Stk)]. + %% saves([SaveVar], Reg, Stk) -> [{move,Reg,Stk}]. %% Generate move instructions to save variables onto stack. The %% stack/reg info used is that after the new stack has been made. @@ -2153,7 +2688,7 @@ fetch_var_prefer_y(V, #sr{reg=Reg,stk=Stk}) -> end. load_vars(Vs, Regs) -> - foldl(fun ({var,V}, Rs) -> put_reg(V, Rs) end, Regs, Vs). + foldl(fun (#k_var{name=V}, Rs) -> put_reg(V, Rs) end, Regs, Vs). %% put_reg(Val, Regs) -> Regs. %% find_reg(Val, Regs) -> {ok,r{R}} | error. @@ -2233,10 +2768,6 @@ find_stack(_, [], _) -> error. on_stack(V, Stk) -> keymember(V, 1, Stk). -is_register({x,_}) -> true; -is_register({yy,_}) -> true; -is_register(_) -> false. - %% put_catch(CatchTag, Stack) -> Stack' %% drop_catch(CatchTag, Stack) -> Stack' %% Special interface for putting and removing catch tags, to ensure that @@ -2254,6 +2785,16 @@ put_catch(Tag, [Other|Stk], Acc) -> drop_catch(Tag, [{{catch_tag,Tag}}|Stk]) -> [free|Stk]; drop_catch(Tag, [Other|Stk]) -> [Other|drop_catch(Tag, Stk)]. +%% atomic(Klit) -> Lit. +%% atomic_list([Klit]) -> [Lit]. + +atomic(#k_literal{val=V}) -> {literal,V}; +atomic(#k_int{val=I}) -> {integer,I}; +atomic(#k_float{val=F}) -> {float,F}; +atomic(#k_atom{val=A}) -> {atom,A}; +%%atomic(#k_char{val=C}) -> {char,C}; +atomic(#k_nil{}) -> nil. + %% new_label(St) -> {L,St}. new_label(#cg{lcount=Next}=St) -> @@ -2296,3 +2837,86 @@ flatmapfoldl(F, Accu0, [Hd|Tail]) -> {Rs,Accu2} = flatmapfoldl(F, Accu1, Tail), {R++Rs,Accu2}; flatmapfoldl(_, Accu, []) -> {[],Accu}. + +%% Keep track of life time for variables. +%% +%% init_vars([{var,VarName}]) -> Vdb. +%% new_vars([VarName], I, Vdb) -> Vdb. +%% use_vars([VarName], I, Vdb) -> Vdb. +%% add_var(VarName, F, L, Vdb) -> Vdb. +%% +%% The list of variable names for new_vars/3 and use_vars/3 +%% must be sorted. + +init_vars(Vs) -> + vdb_new(Vs). + +new_vars([], _, Vdb) -> Vdb; +new_vars([V], I, Vdb) -> vdb_store_new(V, {V,I,I}, Vdb); +new_vars(Vs, I, Vdb) -> vdb_update_vars(Vs, Vdb, I). + +use_vars([], _, Vdb) -> + Vdb; +use_vars([V], I, Vdb) -> + case vdb_find(V, Vdb) of + {V,F,L} when I > L -> vdb_update(V, {V,F,I}, Vdb); + {V,_,_} -> Vdb; + error -> vdb_store_new(V, {V,I,I}, Vdb) + end; +use_vars(Vs, I, Vdb) -> vdb_update_vars(Vs, Vdb, I). + +add_var(V, F, L, Vdb) -> + vdb_store_new(V, {V,F,L}, Vdb). + +%% vdb + +vdb_new(Vs) -> + ordsets:from_list([{V,0,0} || #k_var{name=V} <- Vs]). + +-type var() :: atom(). + +-spec vdb_find(var(), [vdb_entry()]) -> 'error' | vdb_entry(). + +vdb_find(V, Vdb) -> + case lists:keyfind(V, 1, Vdb) of + false -> error; + Vd -> Vd + end. + +vdb_update(V, Update, [{V,_,_}|Vdb]) -> + [Update|Vdb]; +vdb_update(V, Update, [Vd|Vdb]) -> + [Vd|vdb_update(V, Update, Vdb)]. + +vdb_store_new(V, New, [{V1,_,_}=Vd|Vdb]) when V > V1 -> + [Vd|vdb_store_new(V, New, Vdb)]; +vdb_store_new(V, New, [{V1,_,_}|_]=Vdb) when V < V1 -> + [New|Vdb]; +vdb_store_new(_, New, []) -> [New]. + +vdb_update_vars([V|_]=Vs, [{V1,_,_}=Vd|Vdb], I) when V > V1 -> + [Vd|vdb_update_vars(Vs, Vdb, I)]; +vdb_update_vars([V|Vs], [{V1,_,_}|_]=Vdb, I) when V < V1 -> + %% New variable. + [{V,I,I}|vdb_update_vars(Vs, Vdb, I)]; +vdb_update_vars([V|Vs], [{_,F,L}=Vd|Vdb], I) -> + %% Existing variable. + if + I > L -> [{V,F,I}|vdb_update_vars(Vs, Vdb, I)]; + true -> [Vd|vdb_update_vars(Vs, Vdb, I)] + end; +vdb_update_vars([V|Vs], [], I) -> + %% New variable. + [{V,I,I}|vdb_update_vars(Vs, [], I)]; +vdb_update_vars([], Vdb, _) -> Vdb. + +%% vdb_sub(Min, Max, Vdb) -> Vdb. +%% Extract variables which are used before and after Min. Lock +%% variables alive after Max. + +vdb_sub(Min, Max, Vdb) -> + [ if L >= Max -> {V,F,locked}; + true -> Vd + end || {V,F,L}=Vd <- Vdb, + F < Min, + L >= Min ]. diff --git a/lib/compiler/src/v3_core.erl b/lib/compiler/src/v3_core.erl index ae650546e5..6029b91cdc 100644 --- a/lib/compiler/src/v3_core.erl +++ b/lib/compiler/src/v3_core.erl @@ -920,8 +920,9 @@ try_exception(Ecs0, St0) -> %% Note that Tag is not needed for rethrow - it is already in Info. {Evs,St1} = new_vars(3, St0), % Tag, Value, Info {Ecs1,Ceps,St2} = clauses(Ecs0, St1), + Ecs2 = try_build_stacktrace(Ecs1, hd(Evs)), [_,Value,Info] = Evs, - LA = case Ecs1 of + LA = case Ecs2 of [] -> []; [C|_] -> get_lineno_anno(C) end, @@ -930,7 +931,7 @@ try_exception(Ecs0, St0) -> body=[#iprimop{anno=#a{}, %Must have an #a{} name=#c_literal{val=raise}, args=[Info,Value]}]}, - Hs = [#icase{anno=#a{anno=LA},args=[c_tuple(Evs)],clauses=Ecs1,fc=Ec}], + Hs = [#icase{anno=#a{anno=LA},args=[c_tuple(Evs)],clauses=Ecs2,fc=Ec}], {Evs,Ceps++Hs,St2}. try_after(As, St0) -> @@ -946,6 +947,25 @@ try_after(As, St0) -> Hs = [#icase{anno=#a{},args=[c_tuple(Evs)],clauses=[],fc=Ec}], {Evs,Hs,St1}. +try_build_stacktrace([#iclause{pats=Ps0,body=B0}=C0|Cs], RawStk) -> + [#c_tuple{es=[Class,Exc,Stk]}=Tup] = Ps0, + case Stk of + #c_var{name='_'} -> + %% Stacktrace variable is not used. Nothing to do. + [C0|try_build_stacktrace(Cs, RawStk)]; + _ -> + %% Add code to build the stacktrace. + Ps = [Tup#c_tuple{es=[Class,Exc,RawStk]}], + Call = #iprimop{anno=#a{}, + name=#c_literal{val=build_stacktrace}, + args=[RawStk]}, + Iset = #iset{var=Stk,arg=Call}, + B = [Iset|B0], + C = C0#iclause{pats=Ps,body=B}, + [C|try_build_stacktrace(Cs, RawStk)] + end; +try_build_stacktrace([], _) -> []. + %% expr_bin([ArgExpr], St) -> {[Arg],[PreExpr],St}. %% Flatten the arguments of a bin. Do this straight left to right! %% Note that ibinary needs to have its annotation wrapped in a #a{} @@ -2462,9 +2482,11 @@ cexpr(#icase{anno=A,args=Largs,clauses=Lcs,fc=Lfc}, As, St0) -> cexpr(#ireceive1{anno=A,clauses=Lcs}, As, St0) -> Exp = intersection(A#a.ns, As), %Exports {Ccs,St1} = cclauses(Lcs, Exp, St0), + True = #c_literal{val=true}, + Action = core_lib:make_values(lists:duplicate(1+length(Exp), True)), {#c_receive{anno=A#a.anno, clauses=Ccs, - timeout=#c_literal{val=infinity},action=#c_literal{val=true}}, + timeout=#c_literal{val=infinity},action=Action}, Exp,A#a.us,St1}; cexpr(#ireceive2{anno=A,clauses=Lcs,timeout=Lto,action=Les}, As, St0) -> Exp = intersection(A#a.ns, As), %Exports @@ -2505,8 +2527,46 @@ cexpr(#ifun{anno=#a{us=Us0}=A0,name={named,Name},fc=#iclause{pats=Ps}}=Fun0, end; cexpr(#iapply{anno=A,op=Op,args=Args}, _As, St) -> {#c_apply{anno=A#a.anno,op=Op,args=Args},[],A#a.us,St}; -cexpr(#icall{anno=A,module=Mod,name=Name,args=Args}, _As, St) -> - {#c_call{anno=A#a.anno,module=Mod,name=Name,args=Args},[],A#a.us,St}; +cexpr(#icall{anno=A,module=Mod,name=Name,args=Args}, _As, St0) -> + Anno = A#a.anno, + case (not cerl:is_c_atom(Mod)) andalso member(tuple_calls, St0#core.opts) of + true -> + GenAnno = [compiler_generated|Anno], + + %% Generate the clause that matches on the tuple + {TupleVar,St1} = new_var(GenAnno, St0), + {TupleSizeVar, St2} = new_var(GenAnno, St1), + {TupleModVar, St3} = new_var(GenAnno, St2), + {TupleArgsVar, St4} = new_var(GenAnno, St3), + TryVar = cerl:c_var('Try'), + + TupleGuardExpr = + cerl:c_let([TupleSizeVar], + c_call_erl(tuple_size, [TupleVar]), + c_call_erl('>', [TupleSizeVar, cerl:c_int(0)])), + + TupleGuard = + cerl:c_try(TupleGuardExpr, [TryVar], TryVar, + [cerl:c_var('T'),cerl:c_var('R')], cerl:c_atom(false)), + + TupleApply = + cerl:c_let([TupleModVar], + c_call_erl(element, [cerl:c_int(1),TupleVar]), + cerl:c_let([TupleArgsVar], + cerl:make_list(Args ++ [TupleVar]), + c_call_erl(apply, [TupleModVar,Name,TupleArgsVar]))), + + TupleClause = cerl:ann_c_clause(GenAnno, [TupleVar], TupleGuard, TupleApply), + + %% Generate the fallback clause + {OtherVar,St5} = new_var(GenAnno, St4), + OtherApply = cerl:ann_c_call(GenAnno, OtherVar, Name, Args), + OtherClause = cerl:ann_c_clause(GenAnno, [OtherVar], OtherApply), + + {cerl:ann_c_case(GenAnno, Mod, [TupleClause,OtherClause]),[],A#a.us,St5}; + false -> + {#c_call{anno=Anno,module=Mod,name=Name,args=Args},[],A#a.us,St0} + end; cexpr(#iprimop{anno=A,name=Name,args=Args}, _As, St) -> {#c_primop{anno=A#a.anno,name=Name,args=Args},[],A#a.us,St}; cexpr(#iprotect{anno=A,body=Es}, _As, St0) -> @@ -2536,6 +2596,9 @@ cfun(#ifun{anno=A,id=Id,vars=Args,clauses=Lcs,fc=Lfc}, _As, St0) -> clauses=Ccs ++ [Cfc]}}, [],A#a.us,St2}. +c_call_erl(Fun, Args) -> + cerl:c_call(cerl:c_atom(erlang), cerl:c_atom(Fun), Args). + %% lit_vars(Literal) -> [Var]. lit_vars(Lit) -> lit_vars(Lit, []). diff --git a/lib/compiler/src/v3_kernel.erl b/lib/compiler/src/v3_kernel.erl index 1fc05109c5..fd73e5a7dc 100644 --- a/lib/compiler/src/v3_kernel.erl +++ b/lib/compiler/src/v3_kernel.erl @@ -82,7 +82,8 @@ -export([module/2,format_error/1]). -import(lists, [map/2,foldl/3,foldr/3,mapfoldl/3,splitwith/2,member/2, - keymember/3,keyfind/3,partition/2,droplast/1,last/1,sort/1]). + keymember/3,keyfind/3,partition/2,droplast/1,last/1,sort/1, + reverse/1]). -import(ordsets, [add_element/2,del_element/2,union/2,union/1,subtract/2]). -import(cerl, [c_tuple/1]). @@ -107,6 +108,7 @@ copy_anno(Kdst, Ksrc) -> -record(iclause, {anno=[],isub,osub,pats,guard,body}). -record(ireceive_accept, {anno=[],arg}). -record(ireceive_next, {anno=[],arg}). +-record(ignored, {anno=[]}). -type warning() :: term(). % XXX: REFINE @@ -159,11 +161,9 @@ function({#c_var{name={F,Arity}=FA},Body}, St0) -> {#ifun{anno=Ab,vars=Kvs,body=B0},[],St2} = expr(Body, new_sub(), St1), {B1,_,St3} = ubody(B0, return, St2), %%B1 = B0, St3 = St2, %Null second pass - {#k_fdef{anno=#k{us=[],ns=[],a=Ab}, - func=F,arity=Arity,vars=Kvs,body=B1},St3} + {make_fdef(#k{us=[],ns=[],a=Ab}, F, Arity, Kvs, B1),St3} catch - Class:Error -> - Stack = erlang:get_stacktrace(), + Class:Error:Stack -> io:fwrite("Function: ~w/~w\n", [F,Arity]), erlang:raise(Class, Error, Stack) end. @@ -489,7 +489,7 @@ make_alt(First0, Then0) -> Then1 = pre_seq(droplast(Then0), last(Then0)), First2 = make_protected(First1), Then2 = make_protected(Then1), - Body = #k_atom{val=ignored}, + Body = #ignored{}, First3 = #k_guard_clause{guard=First2,body=Body}, Then3 = #k_guard_clause{guard=Then2,body=Body}, First = #k_guard{clauses=[First3]}, @@ -1589,23 +1589,18 @@ match_var([U|Us], Cs0, Def, St) -> %% according to type, the order is really irrelevant but tries to be %% smart. -match_con(Us, [C], Def, St) -> - %% There is only one clause. We can keep literal tuples and - %% lists, but we must convert []/integer/float/atom literals - %% to the proper record (#k_nil{} and so on). - Cs = [expand_pat_lit_clause(C, false)], - match_con_1(Us, Cs, Def, St); match_con(Us, Cs0, Def, St) -> - %% More than one clause. Remove literals at the top level. - Cs = [expand_pat_lit_clause(C, true) || C <- Cs0], + %% Expand literals at the top level. + Cs = [expand_pat_lit_clause(C) || C <- Cs0], match_con_1(Us, Cs, Def, St). match_con_1([U|_Us] = L, Cs, Def, St0) -> %% Extract clauses for different constructors (types). %%ok = io:format("match_con ~p~n", [Cs]), - Ttcs = select_types([k_binary], Cs) ++ select_bin_con(Cs) ++ - select_types([k_cons,k_tuple,k_map,k_atom,k_float,k_int, - k_nil,k_literal], Cs), + Ttcs0 = select_types([k_binary], Cs) ++ select_bin_con(Cs) ++ + select_types([k_cons,k_tuple,k_map,k_atom,k_float, + k_int,k_nil], Cs), + Ttcs = opt_single_valued(Ttcs0), %%ok = io:format("ttcs = ~p~n", [Ttcs]), {Scs,St1} = mapfoldl(fun ({T,Tcs}, St) -> @@ -1618,28 +1613,14 @@ match_con_1([U|_Us] = L, Cs, Def, St0) -> select_types(Types, Cs) -> [{T,Tcs} || T <- Types, begin Tcs = select(T, Cs), Tcs =/= [] end]. - -expand_pat_lit_clause(#iclause{pats=[#ialias{pat=#k_literal{anno=A,val=Val}}=Alias|Ps]}=C, B) -> - P = case B of - true -> expand_pat_lit(Val, A); - false -> literal(Val, A) - end, + +expand_pat_lit_clause(#iclause{pats=[#ialias{pat=#k_literal{anno=A,val=Val}}=Alias|Ps]}=C) -> + P = expand_pat_lit(Val, A), C#iclause{pats=[Alias#ialias{pat=P}|Ps]}; -expand_pat_lit_clause(#iclause{pats=[#k_literal{anno=A,val=Val}|Ps]}=C, B) -> - P = case B of - true -> expand_pat_lit(Val, A); - false -> literal(Val, A) - end, +expand_pat_lit_clause(#iclause{pats=[#k_literal{anno=A,val=Val}|Ps]}=C) -> + P = expand_pat_lit(Val, A), C#iclause{pats=[P|Ps]}; -expand_pat_lit_clause(#iclause{pats=[#k_binary{anno=A,segs=#k_bin_end{}}|Ps]}=C, B) -> - case B of - true -> - C; - false -> - P = #k_literal{anno=A,val = <<>>}, - C#iclause{pats=[P|Ps]} - end; -expand_pat_lit_clause(C, _) -> C. +expand_pat_lit_clause(C) -> C. expand_pat_lit([H|T], A) -> #k_cons{anno=A,hd=literal(H, A),tl=literal(T, A)}; @@ -1659,6 +1640,107 @@ literal(Val, A) when is_atom(Val) -> literal(Val, A) when is_list(Val); is_tuple(Val) -> #k_literal{anno=A,val=Val}. +%% opt_singled_valued([{Type,Clauses}]) -> [{Type,Clauses}]. +%% If a type only has one clause and if the pattern is literal, +%% the matching can be done more efficiently by directly comparing +%% with the literal (that is especially true for binaries). + +opt_single_valued(Ttcs) -> + opt_single_valued(Ttcs, [], []). + +opt_single_valued([{_,[#iclause{pats=[P0|Ps]}=Tc]}=Ttc|Ttcs], TtcAcc, LitAcc) -> + try combine_lit_pat(P0) of + P -> + LitTtc = Tc#iclause{pats=[P|Ps]}, + opt_single_valued(Ttcs, TtcAcc, [LitTtc|LitAcc]) + catch + not_possible -> + opt_single_valued(Ttcs, [Ttc|TtcAcc], LitAcc) + end; +opt_single_valued([Ttc|Ttcs], TtcAcc, LitAcc) -> + opt_single_valued(Ttcs, [Ttc|TtcAcc], LitAcc); +opt_single_valued([], TtcAcc, []) -> + reverse(TtcAcc); +opt_single_valued([], TtcAcc, LitAcc) -> + Literals = {k_literal,reverse(LitAcc)}, + %% Test the literals as early as possible. + case reverse(TtcAcc) of + [{k_binary,_}=Bin|Ttcs] -> + %% The delayed creation of sub binaries requires + %% bs_start_match2 to be the first instruction in the + %% function. + [Bin,Literals|Ttcs]; + Ttcs -> + [Literals|Ttcs] + end. + +combine_lit_pat(#ialias{pat=Pat0}=Alias) -> + Pat = combine_lit_pat(Pat0), + Alias#ialias{pat=Pat}; +combine_lit_pat(Pat) -> + case do_combine_lit_pat(Pat) of + #k_literal{val=Val} when is_atom(Val) -> + throw(not_possible); + #k_literal{val=Val} when is_number(Val) -> + throw(not_possible); + #k_literal{val=[]} -> + throw(not_possible); + #k_literal{}=Lit -> + Lit + end. + +do_combine_lit_pat(#k_atom{anno=A,val=Val}) -> + #k_literal{anno=A,val=Val}; +do_combine_lit_pat(#k_float{anno=A,val=Val}) -> + #k_literal{anno=A,val=Val}; +do_combine_lit_pat(#k_int{anno=A,val=Val}) -> + #k_literal{anno=A,val=Val}; +do_combine_lit_pat(#k_nil{anno=A}) -> + #k_literal{anno=A,val=[]}; +do_combine_lit_pat(#k_binary{anno=A,segs=Segs}) -> + Bin = combine_bin_segs(Segs), + #k_literal{anno=A,val=Bin}; +do_combine_lit_pat(#k_cons{anno=A,hd=Hd0,tl=Tl0}) -> + #k_literal{val=Hd} = do_combine_lit_pat(Hd0), + #k_literal{val=Tl} = do_combine_lit_pat(Tl0), + #k_literal{anno=A,val=[Hd|Tl]}; +do_combine_lit_pat(#k_literal{}=Lit) -> + Lit; +do_combine_lit_pat(#k_tuple{anno=A,es=Es0}) -> + Es = [begin + #k_literal{val=Lit} = do_combine_lit_pat(El), + Lit + end || El <- Es0], + #k_literal{anno=A,val=list_to_tuple(Es)}; +do_combine_lit_pat(_) -> + throw(not_possible). + +combine_bin_segs(#k_bin_seg{size=Size0,unit=Unit,type=integer, + flags=[unsigned,big],seg=Seg,next=Next}) -> + #k_literal{val=Size1} = do_combine_lit_pat(Size0), + #k_literal{val=Int} = do_combine_lit_pat(Seg), + Size = Size1 * Unit, + if + 0 < Size, Size < 64 -> + Bin = <<Int:Size>>, + case Bin of + <<Int:Size>> -> + NextBin = combine_bin_segs(Next), + <<Bin/bits,NextBin/bits>>; + _ -> + %% The integer Int does not fit in the segment, + %% thus it will not match. + throw(not_possible) + end; + true -> + %% Avoid creating huge binary literals. + throw(not_possible) + end; +combine_bin_segs(#k_bin_end{}) -> + <<>>; +combine_bin_segs(_) -> + throw(not_possible). + %% select_bin_con([Clause]) -> [{Type,[Clause]}]. %% Extract clauses for the k_bin_seg constructor. As k_bin_seg %% matching can overlap, the k_bin_seg constructors cannot be @@ -2143,7 +2225,9 @@ ubody(E, return, St0) -> {Ea,Pa,St1} = force_atomic(E, St0), ubody(pre_seq(Pa, #ivalues{args=[Ea]}), return, St1) end; -ubody(E, {break,_Rs} = Break, St0) -> +ubody(#ignored{}, {break,_} = Break, St) -> + ubody(#ivalues{args=[]}, Break, St); +ubody(E, {break,[_]} = Break, St0) -> %%ok = io:fwrite("ubody ~w:~p~n", [?LINE,{E,Br}]), %% Exiting expressions need no trailing break. case is_exit_expr(E) of @@ -2151,6 +2235,16 @@ ubody(E, {break,_Rs} = Break, St0) -> false -> {Ea,Pa,St1} = force_atomic(E, St0), ubody(pre_seq(Pa, #ivalues{args=[Ea]}), Break, St1) + end; +ubody(E, {break,Rs}=Break, St0) -> + case is_exit_expr(E) of + true -> + uexpr(E, return, St0); + false -> + {Vs,St1} = new_vars(length(Rs), St0), + Iset = #iset{vars=Vs,arg=E}, + PreSeq = pre_seq([Iset], #ivalues{args=Vs}), + ubody(PreSeq, Break, St1) end. iletrec_funs(#iletrec{defs=Fs}, St0) -> @@ -2179,9 +2273,8 @@ iletrec_funs_gen(Fs, FreeVs, St) -> Arity0 = length(Vs), {Fb1,_,Lst1} = ubody(Fb0, return, Lst0#kern{ff={N,Arity0}}), Arity = Arity0 + length(FreeVs), - Fun = #k_fdef{anno=#k{us=[],ns=[],a=Fa}, - func=N,arity=Arity, - vars=Vs ++ FreeVs,body=Fb1}, + Fun = make_fdef(#k{us=[],ns=[],a=Fa}, N, Arity, + Vs++FreeVs, Fb1), Lst1#kern{funs=[Fun|Lst1#kern.funs]} end, St, Fs). @@ -2325,8 +2418,7 @@ uexpr(#ifun{anno=A,vars=Vs,body=B0}, {break,Rs}, St0) -> %% No id annotation. Must invent a fun name. new_fun_name(St1) end, - Fun = #k_fdef{anno=#k{us=[],ns=[],a=A},func=Fname,arity=Arity, - vars=Vs ++ Fvs,body=B1}, + Fun = make_fdef(#k{us=[],ns=[],a=A}, Fname, Arity, Vs++Fvs, B1), {#k_bif{anno=#k{us=Free,ns=lit_list_vars(Rs),a=A}, op=#k_internal{name=make_fun,arity=length(Free)+2}, args=[#k_atom{val=Fname},#k_int{val=Arity}|Fvs], @@ -2343,6 +2435,16 @@ uexpr(Lit, {break,Rs0}, St0) -> add_local_function(_, #kern{funs=ignore}=St) -> St; add_local_function(F, #kern{funs=Funs}=St) -> St#kern{funs=[F|Funs]}. +%% Make a #k_fdef{}, making sure that the body is always a #k_match{}. +make_fdef(Anno, Name, Arity, Vs, #k_match{}=Body) -> + #k_fdef{anno=Anno,func=Name,arity=Arity,vars=Vs,body=Body}; +make_fdef(Anno, Name, Arity, Vs, Body) -> + Ka = get_kanno(Body), + Match = #k_match{anno=#k{us=Ka#k.us,ns=[],a=Ka#k.a}, + vars=Vs,body=Body,ret=[]}, + #k_fdef{anno=Anno,func=Name,arity=Arity,vars=Vs,body=Match}. + + %% handle_reuse_annos([#k_var{}], State) -> State. %% In general, it is only safe to reuse a variable for a match context %% if the original value of the variable will no longer be needed. diff --git a/lib/compiler/src/v3_kernel_pp.erl b/lib/compiler/src/v3_kernel_pp.erl index 53097d0d7d..ac91039ae0 100644 --- a/lib/compiler/src/v3_kernel_pp.erl +++ b/lib/compiler/src/v3_kernel_pp.erl @@ -491,7 +491,7 @@ indent(Ctxt) -> indent(Ctxt#ctxt.indent, Ctxt). indent(N, _Ctxt) when N =< 0 -> ""; indent(N, Ctxt) -> T = Ctxt#ctxt.tab_width, - string:chars($\t, N div T, string:chars($\s, N rem T)). + lists:duplicate(N div T, $\t) ++ lists:duplicate(N rem T, $\s). nl_indent(Ctxt) -> [$\n|indent(Ctxt)]. @@ -508,7 +508,7 @@ unindent([$\t|T], N, Ctxt, C) -> if N >= Tab -> unindent(T, N - Tab, Ctxt, C); true -> - unindent([string:chars($\s, Tab - N)|T], 0, Ctxt, C) + unindent([lists:duplicate(Tab - N, $\s)|T], 0, Ctxt, C) end; unindent([L|T], N, Ctxt, C) when is_list(L) -> unindent(L, N, Ctxt, [T|C]); diff --git a/lib/compiler/src/v3_life.erl b/lib/compiler/src/v3_life.erl deleted file mode 100644 index be3ade47ff..0000000000 --- a/lib/compiler/src/v3_life.erl +++ /dev/null @@ -1,468 +0,0 @@ -%% -%% %CopyrightBegin% -%% -%% Copyright Ericsson AB 1999-2016. All Rights Reserved. -%% -%% Licensed under the Apache License, Version 2.0 (the "License"); -%% you may not use this file except in compliance with the License. -%% You may obtain a copy of the License at -%% -%% http://www.apache.org/licenses/LICENSE-2.0 -%% -%% Unless required by applicable law or agreed to in writing, software -%% distributed under the License is distributed on an "AS IS" BASIS, -%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -%% See the License for the specific language governing permissions and -%% limitations under the License. -%% -%% %CopyrightEnd% -%% -%% Purpose : Convert annotated kernel expressions to annotated beam format. - -%% This module creates beam format annotated with variable lifetime -%% information. Each thing is given an index and for each variable we -%% store the first and last index for its occurrence. The variable -%% database, VDB, attached to each thing is only relevant internally -%% for that thing. -%% -%% For nested things like matches the numbering continues locally and -%% the VDB for that thing refers to the variable usage within that -%% thing. Variables which live through a such a thing are internally -%% given a very large last index. Internally the indexes continue -%% after the index of that thing. This creates no problems as the -%% internal variable info never escapes and externally we only see -%% variable which are alive both before or after. -%% -%% This means that variables never "escape" from a thing and the only -%% way to get values from a thing is to "return" them, with 'break' or -%% 'return'. Externally these values become the return values of the -%% thing. This is no real limitation as most nested things have -%% multiple threads so working out a common best variable usage is -%% difficult. - --module(v3_life). - --export([module/2]). - --export([vdb_find/2]). - --import(lists, [member/2,map/2,reverse/1,sort/1]). --import(ordsets, [add_element/2,intersection/2,union/2]). - --include("v3_kernel.hrl"). --include("v3_life.hrl"). - --type fa() :: {atom(),arity()}. - -%% These are not defined in v3_kernel.hrl. -get_kanno(Kthing) -> element(2, Kthing). -%%set_kanno(Kthing, Anno) -> setelement(2, Kthing, Anno). - --spec module(#k_mdef{}, [compile:option()]) -> - {'ok',{module(),[fa()],[_],[_]}}. - -module(#k_mdef{name=M,exports=Es,attributes=As,body=Fs0}, _Opts) -> - Fs1 = functions(Fs0, []), - {ok,{M,Es,As,Fs1}}. - -functions([F|Fs], Acc) -> - functions(Fs, [function(F)|Acc]); -functions([], Acc) -> reverse(Acc). - -%% function(Kfunc) -> Func. - -function(#k_fdef{anno=#k{a=Anno},func=F,arity=Ar,vars=Vs,body=Kb}) -> - try - As = var_list(Vs), - Vdb0 = init_vars(As), - %% Force a top-level match! - B0 = case Kb of - #k_match{} -> Kb; - _ -> - Ka = get_kanno(Kb), - #k_match{anno=#k{us=Ka#k.us,ns=[],a=Ka#k.a}, - vars=Vs,body=Kb,ret=[]} - end, - {B1,_,Vdb1} = body(B0, 1, Vdb0), - {function,F,Ar,As,B1,Vdb1,Anno} - catch - Class:Error -> - Stack = erlang:get_stacktrace(), - io:fwrite("Function: ~w/~w\n", [F,Ar]), - erlang:raise(Class, Error, Stack) - end. - -%% body(Kbody, I, Vdb) -> {[Expr],MaxI,Vdb}. -%% Handle a body. - -body(#k_seq{arg=Ke,body=Kb}, I, Vdb0) -> - %%ok = io:fwrite("life ~w:~p~n", [?LINE,{Ke,I,Vdb0}]), - A = get_kanno(Ke), - Vdb1 = use_vars(union(A#k.us, A#k.ns), I, Vdb0), - {Es,MaxI,Vdb2} = body(Kb, I+1, Vdb1), - E = expr(Ke, I, Vdb2), - {[E|Es],MaxI,Vdb2}; -body(Ke, I, Vdb0) -> - %%ok = io:fwrite("life ~w:~p~n", [?LINE,{Ke,I,Vdb0}]), - A = get_kanno(Ke), - Vdb1 = use_vars(union(A#k.us, A#k.ns), I, Vdb0), - E = expr(Ke, I, Vdb1), - {[E],I,Vdb1}. - -%% protected(Kprotected, I, Vdb) -> Protected. -%% Only used in guards. - -protected(#k_protected{anno=A,arg=Ts,ret=Rs}, I, Vdb) -> - %% Lock variables that are alive before try and used afterwards. - %% Don't lock variables that are only used inside the protected - %% expression. - Pdb0 = vdb_sub(I, I+1, Vdb), - {T,MaxI,Pdb1} = body(Ts, I+1, Pdb0), - Pdb2 = use_vars(A#k.ns, MaxI+1, Pdb1), %Save "return" values - #l{ke={protected,T,var_list(Rs)},i=I,a=A#k.a,vdb=Pdb2}. - -%% expr(Kexpr, I, Vdb) -> Expr. - -expr(#k_test{anno=A,op=Op,args=As,inverted=Inverted}, I, _Vdb) -> - #l{ke={test,test_op(Op),atomic_list(As),Inverted},i=I,a=A#k.a}; -expr(#k_call{anno=A,op=Op,args=As,ret=Rs}, I, _Vdb) -> - #l{ke={call,call_op(Op),atomic_list(As),var_list(Rs)},i=I,a=A#k.a}; -expr(#k_enter{anno=A,op=Op,args=As}, I, _Vdb) -> - #l{ke={enter,call_op(Op),atomic_list(As)},i=I,a=A#k.a}; -expr(#k_bif{anno=A,op=Op,args=As,ret=Rs}, I, _Vdb) -> - Bif = k_bif(A, Op, As, Rs), - #l{ke=Bif,i=I,a=A#k.a}; -expr(#k_match{anno=A,body=Kb,ret=Rs}, I, Vdb) -> - %% Work out imported variables which need to be locked. - Mdb = vdb_sub(I, I+1, Vdb), - M = match(Kb, A#k.us, I+1, [], Mdb), - #l{ke={match,M,var_list(Rs)},i=I,vdb=use_vars(A#k.us, I+1, Mdb),a=A#k.a}; -expr(#k_guard_match{anno=A,body=Kb,ret=Rs}, I, Vdb) -> - %% Work out imported variables which need to be locked. - Mdb = vdb_sub(I, I+1, Vdb), - M = match(Kb, A#k.us, I+1, [], Mdb), - #l{ke={guard_match,M,var_list(Rs)},i=I,vdb=use_vars(A#k.us, I+1, Mdb),a=A#k.a}; -expr(#k_try{}=Try, I, Vdb) -> - body_try(Try, I, Vdb); -expr(#k_protected{}=Protected, I, Vdb) -> - protected(Protected, I, Vdb); -expr(#k_try_enter{anno=A,arg=Ka,vars=Vs,body=Kb,evars=Evs,handler=Kh}, I, Vdb) -> - %% Lock variables that are alive before the catch and used afterwards. - %% Don't lock variables that are only used inside the try. - Tdb0 = vdb_sub(I, I+1, Vdb), - %% This is the tricky bit. Lock variables in Arg that are used in - %% the body and handler. Add try tag 'variable'. - Ab = get_kanno(Kb), - Ah = get_kanno(Kh), - Tdb1 = use_vars(union(Ab#k.us, Ah#k.us), I+3, Tdb0), - Tdb2 = vdb_sub(I, I+2, Tdb1), - Vnames = fun (Kvar) -> Kvar#k_var.name end, %Get the variable names - {Aes,_,Adb} = body(Ka, I+2, add_var({catch_tag,I+1}, I+1, 1000000, Tdb2)), - {Bes,_,Bdb} = body(Kb, I+4, new_vars(sort(map(Vnames, Vs)), I+3, Tdb2)), - {Hes,_,Hdb} = body(Kh, I+4, new_vars(sort(map(Vnames, Evs)), I+3, Tdb2)), - #l{ke={try_enter,#l{ke={block,Aes},i=I+1,vdb=Adb,a=[]}, - var_list(Vs),#l{ke={block,Bes},i=I+3,vdb=Bdb,a=[]}, - var_list(Evs),#l{ke={block,Hes},i=I+3,vdb=Hdb,a=[]}}, - i=I,vdb=Tdb1,a=A#k.a}; -expr(#k_catch{anno=A,body=Kb,ret=[R]}, I, Vdb) -> - %% Lock variables that are alive before the catch and used afterwards. - %% Don't lock variables that are only used inside the catch. - %% Add catch tag 'variable'. - Cdb0 = vdb_sub(I, I+1, Vdb), - {Es,_,Cdb1} = body(Kb, I+1, add_var({catch_tag,I}, I, locked, Cdb0)), - #l{ke={'catch',Es,variable(R)},i=I,vdb=Cdb1,a=A#k.a}; -expr(#k_receive{anno=A,var=V,body=Kb,timeout=T,action=Ka,ret=Rs}, I, Vdb) -> - %% Work out imported variables which need to be locked. - Rdb = vdb_sub(I, I+1, Vdb), - M = match(Kb, add_element(V#k_var.name, A#k.us), I+1, [], - new_vars([V#k_var.name], I, Rdb)), - {Tes,_,Adb} = body(Ka, I+1, Rdb), - #l{ke={receive_loop,atomic(T),variable(V),M, - #l{ke=Tes,i=I+1,vdb=Adb,a=[]},var_list(Rs)}, - i=I,vdb=use_vars(A#k.us, I+1, Vdb),a=A#k.a}; -expr(#k_receive_accept{anno=A}, I, _Vdb) -> - #l{ke=receive_accept,i=I,a=A#k.a}; -expr(#k_receive_next{anno=A}, I, _Vdb) -> - #l{ke=receive_next,i=I,a=A#k.a}; -expr(#k_put{anno=A,arg=Arg,ret=Rs}, I, _Vdb) -> - #l{ke={set,var_list(Rs),literal(Arg, [])},i=I,a=A#k.a}; -expr(#k_break{anno=A,args=As}, I, _Vdb) -> - #l{ke={break,atomic_list(As)},i=I,a=A#k.a}; -expr(#k_guard_break{anno=A,args=As}, I, Vdb) -> - Locked = [V || {V,_,_} <- Vdb], - #l{ke={guard_break,atomic_list(As),Locked},i=I,a=A#k.a}; -expr(#k_return{anno=A,args=As}, I, _Vdb) -> - #l{ke={return,atomic_list(As)},i=I,a=A#k.a}. - -body_try(#k_try{anno=A,arg=Ka,vars=Vs,body=Kb,evars=Evs,handler=Kh,ret=Rs}, - I, Vdb) -> - %% Lock variables that are alive before the catch and used afterwards. - %% Don't lock variables that are only used inside the try. - Tdb0 = vdb_sub(I, I+1, Vdb), - %% This is the tricky bit. Lock variables in Arg that are used in - %% the body and handler. Add try tag 'variable'. - Ab = get_kanno(Kb), - Ah = get_kanno(Kh), - Tdb1 = use_vars(union(Ab#k.us, Ah#k.us), I+3, Tdb0), - Tdb2 = vdb_sub(I, I+2, Tdb1), - Vnames = fun (Kvar) -> Kvar#k_var.name end, %Get the variable names - {Aes,_,Adb} = body(Ka, I+2, add_var({catch_tag,I+1}, I+1, locked, Tdb2)), - {Bes,_,Bdb} = body(Kb, I+4, new_vars(sort(map(Vnames, Vs)), I+3, Tdb2)), - {Hes,_,Hdb} = body(Kh, I+4, new_vars(sort(map(Vnames, Evs)), I+3, Tdb2)), - #l{ke={'try',#l{ke={block,Aes},i=I+1,vdb=Adb,a=[]}, - var_list(Vs),#l{ke={block,Bes},i=I+3,vdb=Bdb,a=[]}, - var_list(Evs),#l{ke={block,Hes},i=I+3,vdb=Hdb,a=[]}, - var_list(Rs)}, - i=I,vdb=Tdb1,a=A#k.a}. - -%% call_op(Op) -> Op. -%% test_op(Op) -> Op. -%% Do any necessary name translations here to munge into beam format. - -call_op(#k_local{name=N}) -> N; -call_op(#k_remote{mod=M,name=N}) -> {remote,atomic(M),atomic(N)}; -call_op(Other) -> variable(Other). - -test_op(#k_remote{mod=#k_atom{val=erlang},name=#k_atom{val=N}}) -> N. - -%% k_bif(Anno, Op, [Arg], [Ret], Vdb) -> Expr. -%% Build bifs. - -k_bif(_A, #k_internal{name=Name}, As, Rs) -> - {internal,Name,atomic_list(As),var_list(Rs)}; -k_bif(_A, #k_remote{mod=#k_atom{val=erlang},name=#k_atom{val=Name}}, As, Rs) -> - Ar = length(As), - case is_gc_bif(Name, Ar) of - false -> - {bif,Name,atomic_list(As),var_list(Rs)}; - true -> - {gc_bif,Name,atomic_list(As),var_list(Rs)} - end. - -%% match(Kexpr, [LockVar], I, Vdb) -> Expr. -%% Convert match tree to old format. - -match(#k_alt{anno=A,first=Kf,then=Kt}, Ls, I, Ctxt, Vdb0) -> - Vdb1 = use_vars(union(A#k.us, Ls), I, Vdb0), - F = match(Kf, Ls, I+1, Ctxt, Vdb1), - T = match(Kt, Ls, I+1, Ctxt, Vdb1), - #l{ke={alt,F,T},i=I,vdb=Vdb1,a=A#k.a}; -match(#k_select{anno=A,var=V,types=Kts}, Ls0, I, Ctxt, Vdb0) -> - Vanno = get_kanno(V), - Ls1 = case member(no_usage, Vanno) of - false -> add_element(V#k_var.name, Ls0); - true -> Ls0 - end, - Anno = case member(reuse_for_context, Vanno) of - true -> [reuse_for_context|A#k.a]; - false -> A#k.a - end, - Vdb1 = use_vars(union(A#k.us, Ls1), I, Vdb0), - Ts = [type_clause(Tc, Ls1, I+1, Ctxt, Vdb1) || Tc <- Kts], - #l{ke={select,literal(V, Ctxt),Ts},i=I,vdb=Vdb1,a=Anno}; -match(#k_guard{anno=A,clauses=Kcs}, Ls, I, Ctxt, Vdb0) -> - Vdb1 = use_vars(union(A#k.us, Ls), I, Vdb0), - Cs = [guard_clause(G, Ls, I+1, Ctxt, Vdb1) || G <- Kcs], - #l{ke={guard,Cs},i=I,vdb=Vdb1,a=A#k.a}; -match(Other, Ls, I, _Ctxt, Vdb0) -> - Vdb1 = use_vars(Ls, I, Vdb0), - {B,_,Vdb2} = body(Other, I+1, Vdb1), - #l{ke={block,B},i=I,vdb=Vdb2,a=[]}. - -type_clause(#k_type_clause{anno=A,type=T,values=Kvs}, Ls, I, Ctxt, Vdb0) -> - %%ok = io:format("life ~w: ~p~n", [?LINE,{T,Kvs}]), - Vdb1 = use_vars(union(A#k.us, Ls), I+1, Vdb0), - Vs = [val_clause(Vc, Ls, I+1, Ctxt, Vdb1) || Vc <- Kvs], - #l{ke={type_clause,type(T),Vs},i=I,vdb=Vdb1,a=A#k.a}. - -val_clause(#k_val_clause{anno=A,val=V,body=Kb}, Ls0, I, Ctxt0, Vdb0) -> - New = (get_kanno(V))#k.ns, - Bus = (get_kanno(Kb))#k.us, - %%ok = io:format("Ls0 = ~p, Used=~p\n New=~p, Bus=~p\n", [Ls0,Used,New,Bus]), - Ls1 = union(intersection(New, Bus), Ls0), %Lock for safety - Vdb1 = use_vars(union(A#k.us, Ls1), I+1, new_vars(New, I, Vdb0)), - Ctxt = case V of - #k_binary{segs=#k_var{name=C0}} -> C0; - _ -> Ctxt0 - end, - B = match(Kb, Ls1, I+1, Ctxt, Vdb1), - #l{ke={val_clause,literal(V, Ctxt),B},i=I,vdb=use_vars(Bus, I+1, Vdb1),a=A#k.a}. - -guard_clause(#k_guard_clause{anno=A,guard=Kg,body=Kb}, Ls, I, Ctxt, Vdb0) -> - Vdb1 = use_vars(union(A#k.us, Ls), I+2, Vdb0), - Gdb = vdb_sub(I+1, I+2, Vdb1), - G = protected(Kg, I+1, Gdb), - B = match(Kb, Ls, I+2, Ctxt, Vdb1), - #l{ke={guard_clause,G,B}, - i=I,vdb=use_vars((get_kanno(Kg))#k.us, I+2, Vdb1), - a=A#k.a}. - -%% type(Ktype) -> Type. - -type(k_literal) -> literal; -type(k_int) -> integer; -%%type(k_char) -> integer; %Hhhmmm??? -type(k_float) -> float; -type(k_atom) -> atom; -type(k_nil) -> nil; -type(k_cons) -> cons; -type(k_tuple) -> tuple; -type(k_binary) -> binary; -type(k_bin_seg) -> bin_seg; -type(k_bin_int) -> bin_int; -type(k_bin_end) -> bin_end; -type(k_map) -> map. - -%% variable(Klit) -> Lit. -%% var_list([Klit]) -> [Lit]. - -variable(#k_var{name=N}) -> {var,N}. - -var_list(Ks) -> [variable(K) || K <- Ks]. - -%% atomic(Klit) -> Lit. -%% atomic_list([Klit]) -> [Lit]. - -atomic(#k_literal{val=V}) -> {literal,V}; -atomic(#k_var{name=N}) -> {var,N}; -atomic(#k_int{val=I}) -> {integer,I}; -atomic(#k_float{val=F}) -> {float,F}; -atomic(#k_atom{val=N}) -> {atom,N}; -%%atomic(#k_char{val=C}) -> {char,C}; -atomic(#k_nil{}) -> nil. - -atomic_list(Ks) -> [atomic(K) || K <- Ks]. - -%% literal(Klit) -> Lit. -%% literal_list([Klit]) -> [Lit]. - -literal(#k_var{name=N}, _) -> {var,N}; -literal(#k_literal{val=I}, _) -> {literal,I}; -literal(#k_int{val=I}, _) -> {integer,I}; -literal(#k_float{val=F}, _) -> {float,F}; -literal(#k_atom{val=N}, _) -> {atom,N}; -%%literal(#k_char{val=C}, _) -> {char,C}; -literal(#k_nil{}, _) -> nil; -literal(#k_cons{hd=H,tl=T}, Ctxt) -> - {cons,[literal(H, Ctxt),literal(T, Ctxt)]}; -literal(#k_binary{segs=V}, Ctxt) -> - {binary,literal(V, Ctxt)}; -literal(#k_bin_seg{size=S,unit=U,type=T,flags=Fs,seg=Seg,next=[]}, Ctxt) -> - %% Only occurs in patterns. - {bin_seg,Ctxt,literal(S, Ctxt),U,T,Fs,[literal(Seg, Ctxt)]}; -literal(#k_bin_seg{size=S,unit=U,type=T,flags=Fs,seg=Seg,next=N}, Ctxt) -> - {bin_seg,Ctxt,literal(S, Ctxt),U,T,Fs, - [literal(Seg, Ctxt),literal(N, Ctxt)]}; -literal(#k_bin_int{size=S,unit=U,flags=Fs,val=Int,next=N}, Ctxt) -> - %% Only occurs in patterns. - {bin_int,Ctxt,literal(S, Ctxt),U,Fs,Int, - [literal(N, Ctxt)]}; -literal(#k_bin_end{}, Ctxt) -> - {bin_end,Ctxt}; -literal(#k_tuple{es=Es}, Ctxt) -> - {tuple,literal_list(Es, Ctxt)}; -literal(#k_map{op=Op,var=Var,es=Es0}, Ctxt) -> - {map,Op,literal(Var, Ctxt),literal_list(Es0, Ctxt)}; -literal(#k_map_pair{key=K,val=V}, Ctxt) -> - {map_pair,literal(K, Ctxt),literal(V, Ctxt)}. - -literal_list(Ks, Ctxt) -> - [literal(K, Ctxt) || K <- Ks]. - - -%% is_gc_bif(Name, Arity) -> true|false -%% Determines whether the BIF Name/Arity might do a GC. - -is_gc_bif(hd, 1) -> false; -is_gc_bif(tl, 1) -> false; -is_gc_bif(self, 0) -> false; -is_gc_bif(node, 0) -> false; -is_gc_bif(node, 1) -> false; -is_gc_bif(element, 2) -> false; -is_gc_bif(get, 1) -> false; -is_gc_bif(tuple_size, 1) -> false; -is_gc_bif(Bif, Arity) -> - not (erl_internal:bool_op(Bif, Arity) orelse - erl_internal:new_type_test(Bif, Arity) orelse - erl_internal:comp_op(Bif, Arity)). - -%% Keep track of life time for variables. -%% -%% init_vars([{var,VarName}]) -> Vdb. -%% new_vars([VarName], I, Vdb) -> Vdb. -%% use_vars([VarName], I, Vdb) -> Vdb. -%% add_var(VarName, F, L, Vdb) -> Vdb. -%% -%% The list of variable names for new_vars/3 and use_vars/3 -%% must be sorted. - -init_vars(Vs) -> - vdb_new(Vs). - -new_vars([], _, Vdb) -> Vdb; -new_vars([V], I, Vdb) -> vdb_store_new(V, {V,I,I}, Vdb); -new_vars(Vs, I, Vdb) -> vdb_update_vars(Vs, Vdb, I). - -use_vars([], _, Vdb) -> - Vdb; -use_vars([V], I, Vdb) -> - case vdb_find(V, Vdb) of - {V,F,L} when I > L -> vdb_update(V, {V,F,I}, Vdb); - {V,_,_} -> Vdb; - error -> vdb_store_new(V, {V,I,I}, Vdb) - end; -use_vars(Vs, I, Vdb) -> vdb_update_vars(Vs, Vdb, I). - -add_var(V, F, L, Vdb) -> - vdb_store_new(V, {V,F,L}, Vdb). - -%% vdb - -vdb_new(Vs) -> - sort([{V,0,0} || {var,V} <- Vs]). - --type var() :: atom(). - --spec vdb_find(var(), [vdb_entry()]) -> 'error' | vdb_entry(). - -vdb_find(V, Vdb) -> - case lists:keyfind(V, 1, Vdb) of - false -> error; - Vd -> Vd - end. - -vdb_update(V, Update, [{V,_,_}|Vdb]) -> - [Update|Vdb]; -vdb_update(V, Update, [Vd|Vdb]) -> - [Vd|vdb_update(V, Update, Vdb)]. - -vdb_store_new(V, New, [{V1,_,_}=Vd|Vdb]) when V > V1 -> - [Vd|vdb_store_new(V, New, Vdb)]; -vdb_store_new(V, New, [{V1,_,_}|_]=Vdb) when V < V1 -> - [New|Vdb]; -vdb_store_new(_, New, []) -> [New]. - -vdb_update_vars([V|_]=Vs, [{V1,_,_}=Vd|Vdb], I) when V > V1 -> - [Vd|vdb_update_vars(Vs, Vdb, I)]; -vdb_update_vars([V|Vs], [{V1,_,_}|_]=Vdb, I) when V < V1 -> - %% New variable. - [{V,I,I}|vdb_update_vars(Vs, Vdb, I)]; -vdb_update_vars([V|Vs], [{_,F,L}=Vd|Vdb], I) -> - %% Existing variable. - if - I > L -> [{V,F,I}|vdb_update_vars(Vs, Vdb, I)]; - true -> [Vd|vdb_update_vars(Vs, Vdb, I)] - end; -vdb_update_vars([V|Vs], [], I) -> - %% New variable. - [{V,I,I}|vdb_update_vars(Vs, [], I)]; -vdb_update_vars([], Vdb, _) -> Vdb. - -%% vdb_sub(Min, Max, Vdb) -> Vdb. -%% Extract variables which are used before and after Min. Lock -%% variables alive after Max. - -vdb_sub(Min, Max, Vdb) -> - [ if L >= Max -> {V,F,locked}; - true -> Vd - end || {V,F,L}=Vd <- Vdb, F < Min, L >= Min ]. diff --git a/lib/compiler/src/v3_life.hrl b/lib/compiler/src/v3_life.hrl deleted file mode 100644 index 5c76312067..0000000000 --- a/lib/compiler/src/v3_life.hrl +++ /dev/null @@ -1,29 +0,0 @@ -%% -%% %CopyrightBegin% -%% -%% Copyright Ericsson AB 1999-2016. All Rights Reserved. -%% -%% Licensed under the Apache License, Version 2.0 (the "License"); -%% you may not use this file except in compliance with the License. -%% You may obtain a copy of the License at -%% -%% http://www.apache.org/licenses/LICENSE-2.0 -%% -%% Unless required by applicable law or agreed to in writing, software -%% distributed under the License is distributed on an "AS IS" BASIS, -%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -%% See the License for the specific language governing permissions and -%% limitations under the License. -%% -%% %CopyrightEnd% -%% -%% This record contains variable life-time annotation for a -%% kernel expression. Added by v3_life, used by v3_codegen. - --type vdb_entry() :: {atom(),non_neg_integer(),non_neg_integer()}. - --record(l, {ke, %Kernel expression - i=0 :: non_neg_integer(), %Op number - vdb=[] :: [vdb_entry()], %Variable database - a=[] :: [term()]}). %Core annotation - |