diff options
Diffstat (limited to 'lib/compiler/src/beam_validator.erl')
| -rw-r--r-- | lib/compiler/src/beam_validator.erl | 304 |
1 files changed, 42 insertions, 262 deletions
diff --git a/lib/compiler/src/beam_validator.erl b/lib/compiler/src/beam_validator.erl index c156cf79fe..e60184c929 100644 --- a/lib/compiler/src/beam_validator.erl +++ b/lib/compiler/src/beam_validator.erl @@ -22,7 +22,6 @@ %% Avoid warning for local function error/1 clashing with autoimported BIF. -compile({no_auto_import,[error/1]}). --export([file/1, files/1]). %% Interface for compiler. -export([module/2, format_error/1]). @@ -40,38 +39,12 @@ -define(DBG_FORMAT(F, D), ok). -endif. -%%% -%%% API functions. -%%% - --spec file(file:filename()) -> 'ok' | {'error', term()}. - -file(Name) when is_list(Name) -> - case case filename:extension(Name) of - ".S" -> s_file(Name); - ".beam" -> beam_file(Name) - end of - [] -> ok; - Es -> {error,Es} - end. - --spec files([file:filename()]) -> 'ok'. - -files([F|Fs]) -> - ?DBG_FORMAT("# Verifying: ~p~n", [F]), - case file(F) of - ok -> ok; - {error,Es} -> - io:format("~tp:~n~ts~n", [F,format_error(Es)]) - end, - files(Fs); -files([]) -> ok. - %% To be called by the compiler. module({Mod,Exp,Attr,Fs,Lc}=Code, _Opts) when is_atom(Mod), is_list(Exp), is_list(Attr), is_integer(Lc) -> case validate(Mod, Fs) of - [] -> {ok,Code}; + [] -> + {ok,Code}; Es0 -> Es = [{?MODULE,E} || E <- Es0], {error,[{atom_to_list(Mod),Es}]} @@ -79,12 +52,6 @@ module({Mod,Exp,Attr,Fs,Lc}=Code, _Opts) -spec format_error(term()) -> iolist(). -format_error([]) -> []; -format_error([{{M,F,A},{I,Off,Desc}}|Es]) -> - [io_lib:format(" ~p:~p/~p+~p:~n ~p - ~p~n", - [M,F,A,Off,I,Desc])|format_error(Es)]; -format_error([Error|Es]) -> - [format_error(Error)|format_error(Es)]; format_error({{_M,F,A},{I,Off,limit}}) -> io_lib:format( "function ~p/~p+~p:~n" @@ -103,8 +70,6 @@ format_error({{_M,F,A},{I,Off,Desc}}) -> " Internal consistency check failed - please report this bug.~n" " Instruction: ~p~n" " Error: ~p:~n", [F,A,Off,I,Desc]); -format_error({Module,Error}) -> - [Module:format_error(Error)]; format_error(Error) -> io_lib:format("~p~n", [Error]). @@ -112,36 +77,6 @@ format_error(Error) -> %%% Local functions follow. %%% -s_file(Name) -> - {ok,Is} = file:consult(Name), - {module,Module} = lists:keyfind(module, 1, Is), - Fs = find_functions(Is), - validate(Module, Fs). - -find_functions(Fs) -> - find_functions_1(Fs, none, [], []). - -find_functions_1([{function,Name,Arity,Entry}|Is], Func, FuncAcc, Acc0) -> - Acc = add_func(Func, FuncAcc, Acc0), - find_functions_1(Is, {Name,Arity,Entry}, [], Acc); -find_functions_1([I|Is], Func, FuncAcc, Acc) -> - find_functions_1(Is, Func, [I|FuncAcc], Acc); -find_functions_1([], Func, FuncAcc, Acc) -> - reverse(add_func(Func, FuncAcc, Acc)). - -add_func(none, _, Acc) -> Acc; -add_func({Name,Arity,Entry}, Is, Acc) -> - [{function,Name,Arity,Entry,reverse(Is)}|Acc]. - -beam_file(Name) -> - try beam_disasm:file(Name) of - {error,beam_lib,Reason} -> [{beam_lib,Reason}]; - #beam_file{module=Module, code=Code0} -> - Code = normalize_disassembled_code(Code0), - validate(Module, Code) - catch _:_ -> [disassembly_failed] - end. - %%% %%% The validator follows. %%% @@ -196,23 +131,16 @@ validate_0(Module, [{function,Name,Ar,Entry,Code}|Fs], Ft) -> try validate_1(Code, Name, Ar, Entry, Ft) of _ -> validate_0(Module, Fs, Ft) catch - Error -> + throw:Error -> + %% Controlled error. [Error|validate_0(Module, Fs, Ft)]; - error:Error -> - [validate_error(Error, Module, Name, Ar)|validate_0(Module, Fs, Ft)] + Class:Error -> + %% Crash. + Stack = erlang:get_stacktrace(), + io:fwrite("Function: ~w/~w\n", [Name,Ar]), + erlang:raise(Class, Error, Stack) end. --ifdef(DEBUG). -validate_error(Error, Module, Name, Ar) -> - exit(validate_error_1(Error, Module, Name, Ar)). --else. -validate_error(Error, Module, Name, Ar) -> - validate_error_1(Error, Module, Name, Ar). --endif. -validate_error_1(Error, Module, Name, Ar) -> - {{Module,Name,Ar}, - {internal_error,'_',{Error,erlang:get_stacktrace()}}}. - -type index() :: non_neg_integer(). -type reg_tab() :: gb_trees:tree(index(), 'none' | {'value', _}). @@ -225,7 +153,6 @@ validate_error_1(Error, Module, Name, Ar) -> hf=0, %Available heap size for floats. fls=undefined, %Floating point state. ct=[], %List of hot catch/try labels - bsm=undefined, %Bit syntax matching state. bits=undefined, %Number of bits in bit syntax binary. setelem=false %Previous instruction was setelement/3. }). @@ -308,7 +235,7 @@ labels_1([{label,L}|Is], R) -> labels_1([{line,_}|Is], R) -> labels_1(Is, R); labels_1(Is, R) -> - {lists:reverse(R),Is}. + {reverse(R),Is}. init_state(Arity) -> Xs = init_regs(Arity, term), @@ -403,10 +330,6 @@ valfun_1({init,{y,_}=Reg}, Vst) -> set_type_y(initialized, Reg, Vst); valfun_1({test_heap,Heap,Live}, Vst) -> test_heap(Heap, Live, Vst); -valfun_1({bif,_Op,nofail,Src,Dst}, Vst) -> - %% The 'nofail' atom only occurs in disassembled code. - validate_src(Src, Vst), - set_type_reg(term, Dst, Vst); valfun_1({bif,Op,{f,_},Src,Dst}=I, Vst) -> case is_bif_safe(Op, length(Src)) of false -> @@ -432,9 +355,6 @@ valfun_1({put_tuple,Sz,Dst}, Vst0) when is_integer(Sz) -> valfun_1({put,Src}, Vst) -> assert_term(Src, Vst), eat_heap(1, Vst); -valfun_1({put_string,Sz,_,Dst}, Vst0) when is_integer(Sz) -> - Vst = eat_heap(2*Sz, Vst0), - set_type_reg(cons, Dst, Vst); %% Instructions for optimization of selective receives. valfun_1({recv_mark,{f,Fail}}, Vst) when is_integer(Fail) -> Vst; @@ -602,8 +522,6 @@ valfun_4({call_ext_last,Live,Func,StkSize}, tail_call(Func, Live, Vst); valfun_4({call_ext_last,_,_,_}, #vst{current=#st{numy=NumY}}) -> error({allocated,NumY}); -valfun_4({make_fun,_,_,Live}, Vst) -> - call('fun', Live, Vst); valfun_4({make_fun2,_,_,_,Live}, Vst) -> call(make_fun, Live, Vst); %% Other BIFs @@ -620,8 +538,6 @@ valfun_4({bif,element,{f,Fail},[Pos,Tuple],Dst}, Vst0) -> TupleType = upgrade_tuple_type({tuple,[get_tuple_size(PosType)]}, TupleType0), Vst = set_type(TupleType, Tuple, Vst1), set_type_reg(term, Dst, Vst); -valfun_4({raise,{f,_}=Fail,Src,Dst}, Vst) -> - valfun_4({bif,raise,Fail,Src,Dst}, Vst); valfun_4({bif,Op,{f,Fail},Src,Dst}, Vst0) -> validate_src(Src, Vst0), Vst = branch_state(Fail, Vst0), @@ -738,32 +654,6 @@ valfun_4({bs_save2,Ctx,SavePoint}, Vst) -> valfun_4({bs_restore2,Ctx,SavePoint}, Vst) -> bsm_restore(Ctx, SavePoint, Vst); -%% Bit syntax instructions. -valfun_4({bs_start_match,{f,_Fail}=F,Src}, Vst) -> - valfun_4({test,bs_start_match,F,[Src]}, Vst); -valfun_4({test,bs_start_match,{f,Fail},[Src]}, Vst) -> - assert_term(Src, Vst), - bs_start_match(branch_state(Fail, Vst)); - -valfun_4({bs_save,SavePoint}, Vst) -> - bs_assert_state(Vst), - bs_save(SavePoint, Vst); -valfun_4({bs_restore,SavePoint}, Vst) -> - bs_assert_state(Vst), - bs_assert_savepoint(SavePoint, Vst), - Vst; -valfun_4({test,bs_skip_bits,{f,Fail},[Src,_,_]}, Vst) -> - bs_assert_state(Vst), - assert_term(Src, Vst), - branch_state(Fail, Vst); -valfun_4({test,bs_test_tail,{f,Fail},_}, Vst) -> - bs_assert_state(Vst), - branch_state(Fail, Vst); -valfun_4({test,_,{f,Fail},[_,_,_,Dst]}, Vst0) -> - bs_assert_state(Vst0), - Vst = branch_state(Fail, Vst0), - set_type_reg({integer,[]}, Dst, Vst); - %% Other test instructions. valfun_4({test,is_float,{f,Lbl},[Float]}, Vst) -> assert_term(Float, Vst), @@ -795,9 +685,6 @@ valfun_4({bs_utf8_size,{f,Fail},A,Dst}, Vst) -> valfun_4({bs_utf16_size,{f,Fail},A,Dst}, Vst) -> assert_term(A, Vst), set_type_reg({integer,[]}, Dst, branch_state(Fail, Vst)); -valfun_4({bs_bits_to_bytes,{f,Fail},Src,Dst}, Vst) -> - assert_term(Src, Vst), - set_type_reg({integer,[]}, Dst, branch_state(Fail, Vst)); valfun_4({bs_init2,{f,Fail},Sz,Heap,Live,_,Dst}, Vst0) -> verify_live(Live, Vst0), if @@ -868,16 +755,6 @@ valfun_4({bs_put_utf32,{f,Fail},_,Src}=I, Vst0) -> assert_term(Src, Vst0), Vst = bs_align_check(I, Vst0), branch_state(Fail, Vst); -%% Old bit syntax construction (before R10B). -valfun_4({bs_init,_,_}, Vst) -> - bs_zero_bits(Vst); -valfun_4({bs_need_buf,_}, Vst) -> Vst; -valfun_4({bs_final,{f,Fail},Dst}, Vst0) -> - Vst = branch_state(Fail, Vst0), - set_type_reg(binary, Dst, Vst); -valfun_4({bs_final2,Src,Dst}, Vst0) -> - assert_term(Src, Vst0), - set_type_reg(binary, Dst, Vst0); %% Map instructions. valfun_4({put_map_assoc,{f,Fail},Src,Dst,Live,{list,List}}, Vst) -> verify_put_map(Fail, Src, Dst, Live, List, Vst); @@ -891,10 +768,14 @@ valfun_4(_, _) -> verify_get_map(Fail, Src, List, Vst0) -> assert_term(Src, Vst0), Vst1 = branch_state(Fail, Vst0), - Lits = mmap(fun(L,_R) -> [L] end, List), - assert_strict_literal_termorder(Lits), + Keys = extract_map_keys(List), + assert_strict_literal_termorder(Keys), verify_get_map_pair(List,Vst0,Vst1). +extract_map_keys([Key,_Val|T]) -> + [Key|extract_map_keys(T)]; +extract_map_keys([]) -> []. + verify_get_map_pair([],_,Vst) -> Vst; verify_get_map_pair([Src,Dst|Vs],Vst0,Vsti) -> assert_term(Src, Vst0), @@ -936,9 +817,6 @@ validate_bs_skip_utf(Fail, Ctx, Live, Vst0) -> %% val_dsetel({move,_,_}, Vst) -> Vst; -val_dsetel({put_string,0,{string,""},_}, Vst) -> - %% An empty string is OK since it doesn't build anything. - Vst; val_dsetel({call_ext,3,{extfunc,erlang,setelement,3}}, #vst{current=St}=Vst) -> Vst#vst{current=St#st{setelem=true}}; val_dsetel({set_tuple_element,_,_,_}, #vst{current=#st{setelem=false}}) -> @@ -972,7 +850,7 @@ call(Name, Live, #vst{current=St}=Vst) -> Type when Type =/= exception -> %% Type is never 'exception' because it has been handled earlier. Xs = gb_trees_from_list([{0,Type}]), - Vst#vst{current=St#st{x=Xs,f=init_fregs(),bsm=undefined}} + Vst#vst{current=St#st{x=Xs,f=init_fregs()}} end. %% Tail call. @@ -1030,7 +908,7 @@ allocate(_, _, _, _, #vst{current=#st{numy=Numy}}) -> error({existing_stack_frame,{size,Numy}}). deallocate(#vst{current=St}=Vst) -> - Vst#vst{current=St#st{y=init_regs(0, initialized),numy=none,bsm=undefined}}. + Vst#vst{current=St#st{y=init_regs(0, initialized),numy=none}}. test_heap(Heap, Live, Vst0) -> verify_live(Live, Vst0), @@ -1038,7 +916,7 @@ test_heap(Heap, Live, Vst0) -> heap_alloc(Heap, Vst). heap_alloc(Heap, #vst{current=St0}=Vst) -> - St1 = kill_heap_allocation(St0#st{bsm=undefined}), + St1 = kill_heap_allocation(St0), St = heap_alloc_1(Heap, St1), Vst#vst{current=St}. @@ -1122,74 +1000,30 @@ assert_freg_set(Fr, _) -> error({bad_source,Fr}). %%% Maps -%% ensure that a list of literals has a strict -%% ascending term order (also meaning unique literals). -%% Single item lists may have registers. -assert_strict_literal_termorder([_]) -> ok; -assert_strict_literal_termorder(Ls) -> - Vs = lists:map(fun (L) -> get_literal(L) end, Ls), +%% A single item list may be either a list or a register. +%% +%% A list with more than item must contain literals in +%% ascending term order. +%% +%% An empty list is not allowed. + +assert_strict_literal_termorder([]) -> + %% There is no reason to use the get_map_elements and + %% has_map_fields instructions with empty lists. + error(empty_field_list); +assert_strict_literal_termorder([_]) -> + ok; +assert_strict_literal_termorder([_,_|_]=Ls) -> + Vs = [get_literal(L) || L <- Ls], case check_strict_value_termorder(Vs) of true -> ok; - false -> error({not_strict_order, Ls}) - end. - -%% usage: -%% mmap(fun(A,B) -> [{A,B}] end, [1,2,3,4]), -%% [{1,2},{3,4}] - -mmap(F,List) -> - {arity,Ar} = erlang:fun_info(F,arity), - mmap(F,Ar,List). -mmap(_F,_,[]) -> []; -mmap(F,Ar,List) -> - {Hd,Tl} = lists:split(Ar,List), - apply(F,Hd) ++ mmap(F,Ar,Tl). - -check_strict_value_termorder([]) -> true; -check_strict_value_termorder([_]) -> true; -check_strict_value_termorder([V1,V2]) -> - erts_internal:cmp_term(V1,V2) < 0; -check_strict_value_termorder([V1,V2|Vs]) -> - case erts_internal:cmp_term(V1,V2) < 0 of - true -> check_strict_value_termorder([V2|Vs]); - false -> false - end. - -%%% -%%% Binary matching. -%%% -%%% Possible values for the bsm field (=bit syntax matching state). -%%% -%%% undefined - Undefined (initial state). No matching instructions allowed. -%%% -%%% (gb set) - The gb set contains the defined save points. -%%% -%%% The bsm field is reset to 'undefined' by instructions that may cause a -%%% a garbage collection (might move the binary) and/or context switch -%%% (may invalidate the save points). - -bs_start_match(#vst{current=#st{bsm=undefined}=St}=Vst) -> - Vst#vst{current=St#st{bsm=gb_sets:empty()}}; -bs_start_match(Vst) -> - %% Must retain save points here - it is possible to restore back - %% to a previous binary. - Vst. - -bs_save(Reg, #vst{current=#st{bsm=Saved}=St}=Vst) - when is_integer(Reg), Reg < ?MAXREG -> - Vst#vst{current=St#st{bsm=gb_sets:add(Reg, Saved)}}; -bs_save(_, _) -> error(limit). - -bs_assert_savepoint(Reg, #vst{current=#st{bsm=Saved}}) -> - case gb_sets:is_member(Reg, Saved) of - false -> error({no_save_point,Reg}); - true -> ok + false -> error(not_strict_order) end. -bs_assert_state(#vst{current=#st{bsm=undefined}}) -> - error(no_bs_match_state); -bs_assert_state(_) -> ok. - +check_strict_value_termorder([V1|[V2|_]=Vs]) -> + erts_internal:cmp_term(V1, V2) < 0 andalso + check_strict_value_termorder(Vs); +check_strict_value_termorder([_]) -> true. %%% %%% New binary matching instructions. @@ -1525,14 +1359,13 @@ merge_states(L, St, Branched) when L =/= 0 -> {value,OtherSt} -> merge_states_1(St, OtherSt) end. -merge_states_1(#st{x=Xs0,y=Ys0,numy=NumY0,h=H0,ct=Ct0,bsm=Bsm0}=St, - #st{x=Xs1,y=Ys1,numy=NumY1,h=H1,ct=Ct1,bsm=Bsm1}) -> +merge_states_1(#st{x=Xs0,y=Ys0,numy=NumY0,h=H0,ct=Ct0}=St, + #st{x=Xs1,y=Ys1,numy=NumY1,h=H1,ct=Ct1}) -> NumY = merge_stk(NumY0, NumY1), Xs = merge_regs(Xs0, Xs1), Ys = merge_y_regs(Ys0, Ys1), Ct = merge_ct(Ct0, Ct1), - Bsm = merge_bsm(Bsm0, Bsm1), - St#st{x=Xs,y=Ys,numy=NumY,h=min(H0, H1),ct=Ct,bsm=Bsm}. + St#st{x=Xs,y=Ys,numy=NumY,h=min(H0, H1),ct=Ct}. merge_stk(S, S) -> S; merge_stk(_, _) -> undecided. @@ -1615,10 +1448,6 @@ merge_types(T1, T2) when T1 =/= T2 -> %% Too different. All we know is that the type is a 'term'. term. -merge_bsm(undefined, _) -> undefined; -merge_bsm(_, undefined) -> undefined; -merge_bsm(Bsm0, Bsm1) -> gb_sets:intersection(Bsm0, Bsm1). - tuple_sz([Sz]) -> Sz; tuple_sz(Sz) -> Sz. @@ -1840,52 +1669,3 @@ error(Error) -> exit(Error). -else. error(Error) -> throw(Error). -endif. - - -%%% -%%% Rewrite disassembled code to the same format as we used internally -%%% to not have to worry later. -%%% - -normalize_disassembled_code(Fs) -> - Index = ndc_index(Fs, []), - ndc(Fs, Index, []). - -ndc_index([{function,Name,Arity,Entry,_Code}|Fs], Acc) -> - ndc_index(Fs, [{{Name,Arity},Entry}|Acc]); -ndc_index([], Acc) -> - gb_trees:from_orddict(lists:sort(Acc)). - -ndc([{function,Name,Arity,Entry,Code0}|Fs], D, Acc) -> - Code = ndc_1(Code0, D, []), - ndc(Fs, D, [{function,Name,Arity,Entry,Code}|Acc]); -ndc([], _, Acc) -> reverse(Acc). - -ndc_1([{call=Op,A,{_,F,A}}|Is], D, Acc) -> - ndc_1(Is, D, [{Op,A,{f,gb_trees:get({F,A}, D)}}|Acc]); -ndc_1([{call_only=Op,A,{_,F,A}}|Is], D, Acc) -> - ndc_1(Is, D, [{Op,A,{f,gb_trees:get({F,A}, D)}}|Acc]); -ndc_1([{call_last=Op,A,{_,F,A},Sz}|Is], D, Acc) -> - ndc_1(Is, D, [{Op,A,{f,gb_trees:get({F,A}, D)},Sz}|Acc]); -ndc_1([{arithbif,Op,F,Src,Dst}|Is], D, Acc) -> - ndc_1(Is, D, [{bif,Op,F,Src,Dst}|Acc]); -ndc_1([{arithfbif,Op,F,Src,Dst}|Is], D, Acc) -> - ndc_1(Is, D, [{bif,Op,F,Src,Dst}|Acc]); -ndc_1([{test,bs_start_match2=Op,F,[A1,Live,A3,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3],Dst}|Acc]); -ndc_1([{test,bs_get_binary2=Op,F,[A1,Live,A3,A4,A5,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3,A4,A5],Dst}|Acc]); -ndc_1([{test,bs_get_float2=Op,F,[A1,Live,A3,A4,A5,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3,A4,A5],Dst}|Acc]); -ndc_1([{test,bs_get_integer2=Op,F,[A1,Live,A3,A4,A5,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3,A4,A5],Dst}|Acc]); -ndc_1([{test,bs_get_utf8=Op,F,[A1,Live,A3,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3],Dst}|Acc]); -ndc_1([{test,bs_get_utf16=Op,F,[A1,Live,A3,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3],Dst}|Acc]); -ndc_1([{test,bs_get_utf32=Op,F,[A1,Live,A3,Dst]}|Is], D, Acc) -> - ndc_1(Is, D, [{test,Op,F,Live,[A1,A3],Dst}|Acc]); -ndc_1([I|Is], D, Acc) -> - ndc_1(Is, D, [I|Acc]); -ndc_1([], _, Acc) -> - reverse(Acc). |