From 47a2ff39ac75c69b41f90f05a63fd9a2e6c0b36a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bj=C3=B6rn=20Gustavsson?= <bjorn@erlang.org>
Date: Sat, 1 Sep 2018 10:57:33 +0200
Subject: Introduce a put_tuple2 instruction

Sometimes when building a tuple, there is no way to avoid an
extra `move` instruction. Consider this code:

    make_tuple(A) -> {ok,A}.

The corresponding BEAM code looks like this:

    {test_heap,3,1}.
    {put_tuple,2,{x,1}}.
    {put,{atom,ok}}.
    {put,{x,0}}.
    {move,{x,1},{x,0}}.
    return.

To avoid overwriting the source register `{x,0}`, a `move`
instruction is necessary.

The problem doesn't exist when building a list:

    %% build_list(A) -> [A].
    {test_heap,2,1}.
    {put_list,{x,0},nil,{x,0}}.
    return.

Introduce a new `put_tuple2` instruction that builds a tuple in a
single instruction, so that the `move` instruction can be eliminated:

    %% make_tuple(A) -> {ok,A}.
    {test_heap,3,1}.
    {put_tuple2,{x,0},{list,[{atom,ok},{x,0}]}}.
    return.

Note that the BEAM loader already combines `put_tuple` and `put`
instructions into an internal instruction similar to `put_tuple2`.
Therefore the introduction of the new instruction will not speed up
execution of tuple building itself, but it will be less work for
the loader to load the new instruction.
---
 erts/emulator/beam/beam_debug.c |  4 ++--
 erts/emulator/beam/instrs.tab   | 17 ++++++++++-------
 erts/emulator/beam/ops.tab      | 15 ++++++++++++---
 3 files changed, 24 insertions(+), 12 deletions(-)

(limited to 'erts/emulator')

diff --git a/erts/emulator/beam/beam_debug.c b/erts/emulator/beam/beam_debug.c
index 9633de2021..b33aab7eee 100644
--- a/erts/emulator/beam/beam_debug.c
+++ b/erts/emulator/beam/beam_debug.c
@@ -786,8 +786,8 @@ print_op(fmtfn_t to, void *to_arg, int op, int size, BeamInstr* addr)
 	    }
 	}
 	break;
-    case op_i_put_tuple_xI:
-    case op_i_put_tuple_yI:
+    case op_put_tuple2_xI:
+    case op_put_tuple2_yI:
     case op_new_map_dtI:
     case op_update_map_assoc_sdtI:
     case op_update_map_exact_jsdtI:
diff --git a/erts/emulator/beam/instrs.tab b/erts/emulator/beam/instrs.tab
index 42c1168f85..da1dd3dc45 100644
--- a/erts/emulator/beam/instrs.tab
+++ b/erts/emulator/beam/instrs.tab
@@ -559,17 +559,19 @@ update_list(Hd, Dst) {
     HTOP += 2;
 }
 
-i_put_tuple := i_put_tuple.make.fill;
-
-i_put_tuple.make(Dst) {
-    $Dst = make_tuple(HTOP);
-}
-
-i_put_tuple.fill(Arity) {
+put_tuple2(Dst, Arity) {
     Eterm* hp = HTOP;
     Eterm arity = $Arity;
 
+    /*
+     * If operands are not packed (in the 32-bit VM),
+     * is is not safe to use $Dst directly after I
+     * has been updated.
+     */
+    Eterm* dst_ptr = &($Dst);
+
     //| -no_next
+    ASSERT(arity != 0);
     *hp++ = make_arityval(arity);
     I = $NEXT_INSTRUCTION;
     do {
@@ -586,6 +588,7 @@ i_put_tuple.fill(Arity) {
             break;
         }
     } while (--arity != 0);
+    *dst_ptr = make_tuple(HTOP);
     HTOP = hp;
     ASSERT(VALID_INSTR(* (Eterm *)I));
     Goto(*I);
diff --git a/erts/emulator/beam/ops.tab b/erts/emulator/beam/ops.tab
index d71594235e..d859c4bb24 100644
--- a/erts/emulator/beam/ops.tab
+++ b/erts/emulator/beam/ops.tab
@@ -483,9 +483,16 @@ is_eq f? s s
 is_ne f? s s
 
 #
-# Putting things.
+# Putting tuples.
+#
+# Code compiled with OTP 22 and later uses put_tuple2 to
+# to construct a tuple.
+#
+# Code compiled before OTP 22 uses put_tuple + one put instruction
+# per element. Translate to put_tuple2.
 #
 
+i_put_tuple/2
 put_tuple Arity Dst => i_put_tuple Dst u
 
 i_put_tuple Dst Arity Puts=* | put S1 | put S2 | \
@@ -495,10 +502,12 @@ i_put_tuple Dst Arity Puts=* | put S1 | put S2 | \
 i_put_tuple Dst Arity Puts=* | put S => \
 	    tuple_append_put(Arity, Dst, Puts, S)
 
-i_put_tuple/2
+i_put_tuple Dst Arity Puts=* => put_tuple2 Dst Arity Puts
 
-i_put_tuple xy I
+put_tuple2 xy I
 
+#
+# Putting lists.
 #
 # The instruction "put_list Const [] Dst" were generated in rare
 # circumstances up to and including OTP 18. Starting with OTP 19,
-- 
cgit v1.2.3