Rewrite binary creation instructions to bs_init instructions

Rewrite the five binary creation instructions to a bs_init
instruction, in order to somewhat reduce code bloat.

8 files changed, 75 insertions, 128 deletions

diff --git a/lib/compiler/src/beam_a.erl b/lib/compiler/src/beam_a.erl
index 3709b94c3c..fd76154b78 100644
--- a/lib/compiler/src/beam_a.erl
+++ b/lib/compiler/src/beam_a.erl
@@ -64,4 +64,14 @@ rename_instr({bs_utf8_size=I,F,Src,Dst}) ->
     {bif,I,F,[Src],Dst};
 rename_instr({bs_utf16_size=I,F,Src,Dst}) ->
     {bif,I,F,[Src],Dst};
+rename_instr({bs_init2=I,F,Sz,Extra,Live,Flags,Dst}) ->
+    {bs_init,F,{I,Extra,Flags},Live,[Sz],Dst};
+rename_instr({bs_init_bits=I,F,Sz,Extra,Live,Flags,Dst}) ->
+    {bs_init,F,{I,Extra,Flags},Live,[Sz],Dst};
+rename_instr({bs_append=I,F,Sz,Extra,Live,U,Src,Flags,Dst}) ->
+    {bs_init,F,{I,Extra,U,Flags},Live,[Sz,Src],Dst};
+rename_instr({bs_private_append=I,F,Sz,U,Src,Flags,Dst}) ->
+    {bs_init,F,{I,U,Flags},none,[Sz,Src],Dst};
+rename_instr(bs_init_writable=I) ->
+    {bs_init,{f,0},I,1,[{x,0}],{x,0}};
 rename_instr(I) -> I.
diff --git a/lib/compiler/src/beam_bsm.erl b/lib/compiler/src/beam_bsm.erl
index dd3b5f5c04..cfe4af7410 100644
--- a/lib/compiler/src/beam_bsm.erl
+++ b/lib/compiler/src/beam_bsm.erl
@@ -293,18 +293,8 @@ btb_reaches_match_2([{jump,{f,Lbl}}|_], Regs, #btb{index=Li}=D) ->
     btb_reaches_match_2(Is, Regs, D);
 btb_reaches_match_2([{label,_}|Is], Regs, D) ->
     btb_reaches_match_2(Is, Regs, D);
-btb_reaches_match_2([bs_init_writable|Is], Regs0, D) ->
-    Regs = btb_kill_not_live(0, Regs0),
-    btb_reaches_match_1(Is, Regs, D);
-btb_reaches_match_2([{bs_init2,{f,0},_,_,_,_,Dst}|Is], Regs, D) ->
-    btb_reaches_match_1(Is, btb_kill([Dst], Regs), D);
-btb_reaches_match_2([{bs_init_bits,{f,0},_,_,_,_,Dst}|Is], Regs, D) ->
-    btb_reaches_match_1(Is, btb_kill([Dst], Regs), D);
-btb_reaches_match_2([{bs_append,{f,0},_,_,_,_,Src,_,Dst}=I|Is], Regs, D) ->
-    btb_ensure_not_used([Src], I, Regs),
-    btb_reaches_match_1(Is, btb_kill([Dst], Regs), D);
-btb_reaches_match_2([{bs_private_append,{f,0},_,_,Src,_,Dst}=I|Is], Regs, D) ->
-    btb_ensure_not_used([Src], I, Regs),
+btb_reaches_match_2([{bs_init,{f,0},_,_,Ss,Dst}=I|Is], Regs, D) ->
+    btb_ensure_not_used(Ss, I, Regs),
     btb_reaches_match_1(Is, btb_kill([Dst], Regs), D);
 btb_reaches_match_2([{bs_put,{f,0},_,Ss}=I|Is], Regs, D) ->
     btb_ensure_not_used(Ss, I, Regs),
diff --git a/lib/compiler/src/beam_clean.erl b/lib/compiler/src/beam_clean.erl
index 083ccb32fa..39710490df 100644
--- a/lib/compiler/src/beam_clean.erl
+++ b/lib/compiler/src/beam_clean.erl
@@ -242,15 +242,10 @@ replace([{call_only,Ar,{f,Lbl}}|Is], Acc, D) ->
     replace(Is, [{call_only,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_init2,{f,Lbl},Sz,Words,R,F,Dst}|Is], Acc, D) when Lbl =/= 0 ->
-    replace(Is, [{bs_init2,{f,label(Lbl, D)},Sz,Words,R,F,Dst}|Acc], D);
-replace([{bs_init_bits,{f,Lbl},Sz,Words,R,F,Dst}|Is], Acc, D) when Lbl =/= 0 ->
-    replace(Is, [{bs_init_bits,{f,label(Lbl, D)},Sz,Words,R,F,Dst}|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([{bs_append,{f,Lbl},_,_,_,_,_,_,_}=I0|Is], Acc, D) when Lbl =/= 0 ->
-    I = setelement(2, I0, {f,label(Lbl, D)}),
-    replace(Is, [I|Acc], D);
 replace([I|Is], Acc, D) ->
     replace(Is, [I|Acc], D);
 replace([], Acc, _) -> Acc.
diff --git a/lib/compiler/src/beam_flatten.erl b/lib/compiler/src/beam_flatten.erl
index 4cca406f45..04232d8fd2 100644
--- a/lib/compiler/src/beam_flatten.erl
+++ b/lib/compiler/src/beam_flatten.erl
@@ -79,43 +79,28 @@ norm_allocate({nozero,Ns,Nh,Inits}, Regs) ->
 
 %% insert_alloc_in_bs_init(ReverseInstructionStream, AllocationInfo) ->
 %%                                  impossible | ReverseInstructionStream'
-%%   A bs_init2/6 instruction should not be followed by a test heap instruction.
+%%   A bs_init/6 instruction should not be followed by a test heap instruction.
 %%   Given the AllocationInfo from a test heap instruction, merge the
-%%   allocation amounts into the previous bs_init2/6 instruction (if any).
+%%   allocation amounts into the previous bs_init/6 instruction (if any).
 %%
-insert_alloc_in_bs_init([I|_]=Is, Alloc) ->
-    case is_bs_constructor(I) of
-	false -> impossible;
-	true -> insert_alloc_1(Is, Alloc, [])
-    end.
-
-insert_alloc_1([{bs_init2=Op,Fail,Bs,Ws1,Regs,F,Dst}|Is], {_,nostack,Ws2,[]}, Acc) ->
-    Al = beam_utils:combine_heap_needs(Ws1, Ws2),
-    I = {Op,Fail,Bs,Al,Regs,F,Dst},
-    reverse(Acc, [I|Is]);
-insert_alloc_1([{bs_init_bits=Op,Fail,Bs,Ws1,Regs,F,Dst}|Is], {_,nostack,Ws2,[]}, Acc) ->
-    Al = beam_utils:combine_heap_needs(Ws1, Ws2),
-    I = {Op,Fail,Bs,Al,Regs,F,Dst},
-    reverse(Acc, [I|Is]);
-insert_alloc_1([{bs_append,Fail,Sz,Ws1,Regs,U,Bin,Fl,Dst}|Is],
-	       {_,nostack,Ws2,[]}, Acc) ->
+insert_alloc_in_bs_init([{bs_put,_,_,_}=I|Is], Alloc) ->
+    %% The instruction sequence ends with an bs_put/4 instruction.
+    %% We'll need to search backwards for the bs_init/6 instruction.
+    insert_alloc_1(Is, Alloc, [I]);
+insert_alloc_in_bs_init(_, _) -> impossible.
+
+insert_alloc_1([{bs_init=Op,Fail,Info0,Live,Ss,Dst}|Is],
+	       {_,nostack,Ws2,[]}, Acc) when is_integer(Live) ->
+    %% The number of extra heap words is always in the second position
+    %% in the Info tuple.
+    Ws1 = element(2, Info0),
     Al = beam_utils:combine_heap_needs(Ws1, Ws2),
-    I = {bs_append,Fail,Sz,Al,Regs,U,Bin,Fl,Dst},
+    Info = setelement(2, Info0, Al),
+    I = {Op,Fail,Info,Live,Ss,Dst},
     reverse(Acc, [I|Is]);
-insert_alloc_1([I|Is], Alloc, Acc) ->
+insert_alloc_1([{bs_put,_,_,_}=I|Is], Alloc, Acc) ->
     insert_alloc_1(Is, Alloc, [I|Acc]).
 
-
-%% is_bs_constructor(Instruction) -> true|false.
-%%  Test whether the instruction is a bit syntax construction
-%%  instruction that can occur at the end of a bit syntax
-%%  construction. (Since an empty binary would be expressed
-%%  as a literal, the bs_init2/6 instruction will not occur
-%%  at the end and therefore it is no need to test for it here.)
-%%
-is_bs_constructor({bs_put,_,_,_}) -> true;
-is_bs_constructor(_) -> false.
-
 %% opt(Is0) -> Is
 %%  Simple peep-hole optimization to move a {move,Any,{x,0}} past
 %%  any kill up to the next call instruction. (To give the loader
diff --git a/lib/compiler/src/beam_jump.erl b/lib/compiler/src/beam_jump.erl
index deb7c2033f..d4a8d9e8a5 100644
--- a/lib/compiler/src/beam_jump.erl
+++ b/lib/compiler/src/beam_jump.erl
@@ -535,14 +535,10 @@ ulbl({bif,_Name,Lbl,_As,_R}, Used) ->
     mark_used(Lbl, Used);
 ulbl({gc_bif,_Name,Lbl,_Live,_As,_R}, Used) ->
     mark_used(Lbl, Used);
-ulbl({bs_init2,Lbl,_,_,_,_,_}, Used) ->
-    mark_used(Lbl, Used);
-ulbl({bs_init_bits,Lbl,_,_,_,_,_}, Used) ->
+ulbl({bs_init,Lbl,_,_,_,_}, Used) ->
     mark_used(Lbl, Used);
 ulbl({bs_put,Lbl,_,_}, Used) ->
     mark_used(Lbl, Used);
-ulbl({bs_append,Lbl,_,_,_,_,_,_,_}, Used) ->
-    mark_used(Lbl, Used);
 ulbl(_, Used) -> Used.
 
 mark_used({f,0}, Used) -> Used;
diff --git a/lib/compiler/src/beam_trim.erl b/lib/compiler/src/beam_trim.erl
index 8f694f602b..ceeed603ab 100644
--- a/lib/compiler/src/beam_trim.erl
+++ b/lib/compiler/src/beam_trim.erl
@@ -172,19 +172,10 @@ remap([{bif,Name,Fail,Ss,D}|Is], Map, Acc) ->
 remap([{gc_bif,Name,Fail,Live,Ss,D}|Is], Map, Acc) ->
     I = {gc_bif,Name,Fail,Live,[Map(S) || S <- Ss],Map(D)},
     remap(Is, Map, [I|Acc]);
-remap([{bs_append=Op,Fail,Bits,Heap,Live,Unit,Bin,Flags,D}|Is], Map, Acc) ->
-    I = {Op,Fail,Map(Bits),Heap,Live,Unit,Map(Bin),Flags,Map(D)},
-    remap(Is, Map, [I|Acc]);
-remap([{bs_private_append=Op,Fail,Bits,Unit,Bin,Flags,D}|Is], Map, Acc) ->
-    I = {Op,Fail,Map(Bits),Unit,Map(Bin),Flags,Map(D)},
-    remap(Is, Map, [I|Acc]);
-remap([bs_init_writable=I|Is], Map, Acc) ->
-    remap(Is, Map, [I|Acc]);
-remap([{bs_init2,Fail,Src,Live,U,Flags,D}|Is], Map, Acc) ->
-    I = {bs_init2,Fail,Map(Src),Live,U,Flags,Map(D)},
-    remap(Is, Map, [I|Acc]);
-remap([{bs_init_bits,Fail,Src,Live,U,Flags,D}|Is], Map, Acc) ->
-    I = {bs_init_bits,Fail,Map(Src),Live,U,Flags,Map(D)},
+remap([{bs_init,Fail,Info,Live,Ss0,Dst0}|Is], Map, Acc) ->
+    Ss = [Map(Src) || Src <- Ss0],
+    Dst = Map(Dst0),
+    I = {bs_init,Fail,Info,Live,Ss,Dst},
     remap(Is, Map, [I|Acc]);
 remap([{bs_put=Op,Fail,Info,Ss}|Is], Map, Acc) ->
     I = {Op,Fail,Info,[Map(S) || S <- Ss]},
@@ -284,15 +275,7 @@ frame_size([{test,_,{f,L},_}|Is], Safe) ->
     frame_size_branch(L, Is, Safe);
 frame_size([{test,_,{f,L},_,_,_}|Is], Safe) ->
     frame_size_branch(L, Is, Safe);
-frame_size([{bs_append,{f,L},_,_,_,_,_,_,_}|Is], Safe) ->
-    frame_size_branch(L, Is, Safe);
-frame_size([{bs_private_append,{f,L},_,_,_,_,_}|Is], Safe) ->
-    frame_size_branch(L, Is, Safe);
-frame_size([bs_init_writable|Is], Safe) ->
-    frame_size(Is, Safe);
-frame_size([{bs_init2,{f,L},_,_,_,_,_}|Is], Safe) ->
-    frame_size_branch(L, Is, Safe);
-frame_size([{bs_init_bits,{f,L},_,_,_,_,_}|Is], Safe) ->
+frame_size([{bs_init,{f,L},_,_,_,_}|Is], Safe) ->
     frame_size_branch(L, Is, Safe);
 frame_size([{bs_put,{f,L},_,_}|Is], Safe) ->
     frame_size_branch(L, Is, Safe);
diff --git a/lib/compiler/src/beam_utils.erl b/lib/compiler/src/beam_utils.erl
index be6cc56eaa..6d86fcd9f2 100644
--- a/lib/compiler/src/beam_utils.erl
+++ b/lib/compiler/src/beam_utils.erl
@@ -301,34 +301,32 @@ check_liveness(R, [{kill,R}|_], St) ->
     {killed,St};
 check_liveness(R, [{kill,_}|Is], St) ->
     check_liveness(R, Is, St);
-check_liveness(R, [bs_init_writable|Is], St) ->
-    if
-	R =:= {x,0} -> {used,St};
-	true -> check_liveness(R, Is, St)
-    end;
-check_liveness(R, [{bs_private_append,_,Bits,_,Bin,_,Dst}|Is], St) ->
-    case R of
-	Bits -> {used,St};
-	Bin -> {used,St};
-	Dst -> {killed,St};
-	_ -> check_liveness(R, Is, St)
+check_liveness(R, [{bs_init,_,_,none,Ss,Dst}|Is], St) ->
+    case member(R, Ss) of
+	true ->
+	    {used,St};
+	false ->
+	    if
+		R =:= Dst -> {killed,St};
+		true -> check_liveness(R, Is, St)
+	    end
     end;
-check_liveness(R, [{bs_append,_,Bits,_,_,_,Bin,_,Dst}|Is], St) ->
+check_liveness(R, [{bs_init,_,_,Live,Ss,Dst}|Is], St) ->
     case R of
-	Bits -> {used,St};
-	Bin -> {used,St};
-	Dst -> {killed,St};
-	_ -> check_liveness(R, Is, St)
-    end;
-check_liveness(R, [{bs_init2,_,_,_,_,_,Dst}|Is], St) ->
-    if
-	R =:= Dst -> {killed,St};
-	true -> check_liveness(R, Is, St)
-    end;
-check_liveness(R, [{bs_init_bits,_,_,_,_,_,Dst}|Is], St) ->
-    if
-	R =:= Dst -> {killed,St};
-	true -> check_liveness(R, Is, St)
+	{x,X} ->
+	    case X < Live orelse member(R, Ss) of
+		true -> {used,St};
+		false -> {killed,St}
+	    end;
+	{y,_} ->
+	    case member(R, Ss) of
+		true -> {used,St};
+		false ->
+		    if
+			R =:= Dst -> {killed,St};
+			true -> check_liveness(R, Is, St)
+		    end
+	    end
     end;
 check_liveness(R, [{deallocate,_}|Is], St) ->
     case R of
@@ -670,34 +668,18 @@ combine_alloc_lists_1([]) -> [].
 live_opt([{bs_context_to_binary,Src}=I|Is], Regs0, D, Acc) ->
     Regs = x_live([Src], Regs0),
     live_opt(Is, Regs, D, [I|Acc]);
-live_opt([{bs_init2,Fail,Sz,Extra,Live0,Fl,Dst}|Is], Regs0, D, Acc) ->
-    Regs1 = x_dead([Dst], Regs0),
-    Live = live_regs(Regs1),
-    true = Live =< Live0,	%Assertion.
-    Regs2 = live_call(Live),
-    Regs = live_join_label(Fail, D, Regs2),
-    I = {bs_init2,Fail,Sz,Extra,Live,Fl,Dst},
+live_opt([{bs_init,Fail,_,none,Ss,Dst}=I|Is], Regs0, D, Acc) ->
+    Regs1 = x_live(Ss, x_dead([Dst], Regs0)),
+    Regs = live_join_label(Fail, D, Regs1),
     live_opt(Is, Regs, D, [I|Acc]);
-live_opt([{bs_init_bits,Fail,Sz,Extra,Live0,Fl,Dst}|Is], Regs0, D, Acc) ->
+live_opt([{bs_init,Fail,Info,Live0,Ss,Dst}|Is], Regs0, D, Acc) ->
     Regs1 = x_dead([Dst], Regs0),
     Live = live_regs(Regs1),
     true = Live =< Live0,	%Assertion.
     Regs2 = live_call(Live),
-    Regs = live_join_label(Fail, D, Regs2),
-    I = {bs_init_bits,Fail,Sz,Extra,Live,Fl,Dst},
-    live_opt(Is, Regs, D, [I|Acc]);
-live_opt([{bs_append,Fail,Bits,Extra,Live0,Unit,Src,Fl,Dst}|Is],
-	 Regs0, D, Acc) ->
-    Regs1 = x_live([Src], x_dead([Dst], Regs0)),
-    Live = live_regs(Regs1),
-    true = Live =< Live0,	%Assertion.
-    Regs2 = live_call(Live),
-    Regs = live_join_label(Fail, D, Regs2),
-    I = {bs_append,Fail,Bits,Extra,Live,Unit,Src,Fl,Dst},
-    live_opt(Is, Regs, D, [I|Acc]);
-live_opt([{bs_private_append,Fail,Src1,_,Src2,_,Dst}=I|Is], Regs0, D, Acc) ->
-    Regs1 = x_live([Src1,Src2], x_dead([Dst], Regs0)),
-    Regs = live_join_label(Fail, D, Regs1),
+    Regs3 = x_live(Ss, Regs2),
+    Regs = live_join_label(Fail, D, Regs3),
+    I = {bs_init,Fail,Info,Live,Ss,Dst},
     live_opt(Is, Regs, D, [I|Acc]);
 live_opt([{bs_put,Fail,_,Ss}=I|Is], Regs0, D, Acc) ->
     Regs1 = x_live(Ss, Regs0),
@@ -743,8 +725,6 @@ live_opt([{try_case_end,Src}=I|Is], _, D, Acc) ->
 live_opt([if_end=I|Is], _, D, Acc) ->
     Regs = 0,
     live_opt(Is, Regs, D, [I|Acc]);
-live_opt([bs_init_writable=I|Is], _, D, Acc) ->
-    live_opt(Is, live_call(1), D, [I|Acc]);
 live_opt([{call,Arity,_}=I|Is], _, D, Acc) ->
     live_opt(Is, live_call(Arity), D, [I|Acc]);
 live_opt([{call_ext,Arity,_}=I|Is], _, D, Acc) ->
diff --git a/lib/compiler/src/beam_z.erl b/lib/compiler/src/beam_z.erl
index 7af6f1a74c..37087fc306 100644
--- a/lib/compiler/src/beam_z.erl
+++ b/lib/compiler/src/beam_z.erl
@@ -50,4 +50,12 @@ undo_rename({bif,bs_utf8_size=I,F,[Src],Dst}) ->
     {I,F,Src,Dst};
 undo_rename({bif,bs_utf16_size=I,F,[Src],Dst}) ->
     {I,F,Src,Dst};
+undo_rename({bs_init,F,{I,U,Flags},none,[Sz,Src],Dst}) ->
+    {I,F,Sz,U,Src,Flags,Dst};
+undo_rename({bs_init,F,{I,Extra,Flags},Live,[Sz],Dst}) ->
+    {I,F,Sz,Extra,Live,Flags,Dst};
+undo_rename({bs_init,F,{I,Extra,U,Flags},Live,[Sz,Src],Dst}) ->
+    {I,F,Sz,Extra,Live,U,Src,Flags,Dst};
+undo_rename({bs_init,_,bs_init_writable=I,_,_,_}) ->
+    I;
 undo_rename(I) -> I.