Emit special instruction for array literal + .(hash|min|max)

This commit introduces a new instruction `opt_newarray_send` which is
used when there is an array literal followed by either the `hash`,
`min`, or `max` method.

```
[a, b, c].hash
```

Will emit an `opt_newarray_send` instruction.  This instruction falls
back to a method call if the "interested" method has been monkey
patched.

Here are some examples of the instructions generated:

```
$ ./miniruby --dump=insns -e '[@a, @b].max'
== disasm: #<ISeq:<main>@-e:1 (1,0)-(1,12)> (catch: FALSE)
0000 getinstancevariable                    :@a, <is:0>               (   1)[Li]
0003 getinstancevariable                    :@b, <is:1>
0006 opt_newarray_send                      2, :max
0009 leave
$ ./miniruby --dump=insns -e '[@a, @b].min'
== disasm: #<ISeq:<main>@-e:1 (1,0)-(1,12)> (catch: FALSE)
0000 getinstancevariable                    :@a, <is:0>               (   1)[Li]
0003 getinstancevariable                    :@b, <is:1>
0006 opt_newarray_send                      2, :min
0009 leave
$ ./miniruby --dump=insns -e '[@a, @b].hash'
== disasm: #<ISeq:<main>@-e:1 (1,0)-(1,13)> (catch: FALSE)
0000 getinstancevariable                    :@a, <is:0>               (   1)[Li]
0003 getinstancevariable                    :@b, <is:1>
0006 opt_newarray_send                      2, :hash
0009 leave
```

[Feature #18897] [ruby-core:109147]

Co-authored-by: John Hawthorn <[email protected]>

9 files changed, 76 insertions, 46 deletions

diff --git a/array.c b/array.c
index 1221b43923..b3b334dd68 100644
--- a/array.c
+++ b/array.c
@@ -5350,6 +5350,23 @@ rb_ary_eql(VALUE ary1, VALUE ary2)
     return rb_exec_recursive_paired(recursive_eql, ary1, ary2, ary2);
 }
 
+VALUE
+rb_ary_hash_values(long len, const VALUE *elements)
+{
+    long i;
+    st_index_t h;
+    VALUE n;
+
+    h = rb_hash_start(len);
+    h = rb_hash_uint(h, (st_index_t)rb_ary_hash_values);
+    for (i=0; i<len; i++) {
+	n = rb_hash(elements[i]);
+	h = rb_hash_uint(h, NUM2LONG(n));
+    }
+    h = rb_hash_end(h);
+    return ST2FIX(h);
+}
+
 /*
  *  call-seq:
  *    array.hash -> integer
@@ -5366,18 +5383,7 @@ rb_ary_eql(VALUE ary1, VALUE ary2)
 static VALUE
 rb_ary_hash(VALUE ary)
 {
-    long i;
-    st_index_t h;
-    VALUE n;
-
-    h = rb_hash_start(RARRAY_LEN(ary));
-    h = rb_hash_uint(h, (st_index_t)rb_ary_hash);
-    for (i=0; i<RARRAY_LEN(ary); i++) {
-        n = rb_hash(RARRAY_AREF(ary, i));
-        h = rb_hash_uint(h, NUM2LONG(n));
-    }
-    h = rb_hash_end(h);
-    return ST2FIX(h);
+    return rb_ary_hash_values(RARRAY_LEN(ary), RARRAY_CONST_PTR(ary));
 }
 
 /*
diff --git a/bootstraptest/test_insns.rb b/bootstraptest/test_insns.rb
index 91fba9b011..d2e799f855 100644
--- a/bootstraptest/test_insns.rb
+++ b/bootstraptest/test_insns.rb
@@ -214,9 +214,11 @@ tests = [
     'true'.freeze
   },
 
-  [ 'opt_newarray_max', %q{ [ ].max.nil? }, ],
-  [ 'opt_newarray_max', %q{ [1, x = 2, 3].max == 3 }, ],
-  [ 'opt_newarray_max', <<-'},', ], # {
+  [ 'opt_newarray_send', %q{ ![ ].hash.nil? }, ],
+
+  [ 'opt_newarray_send', %q{ [ ].max.nil? }, ],
+  [ 'opt_newarray_send', %q{ [1, x = 2, 3].max == 3 }, ],
+  [ 'opt_newarray_send', <<-'},', ], # {
     class Array
       def max
         true
@@ -224,9 +226,9 @@ tests = [
     end
     [1, x = 2, 3].max
   },
-  [ 'opt_newarray_min', %q{ [ ].min.nil? }, ],
-  [ 'opt_newarray_min', %q{ [3, x = 2, 1].min == 1 }, ],
-  [ 'opt_newarray_min', <<-'},', ], # {
+  [ 'opt_newarray_send', %q{ [ ].min.nil? }, ],
+  [ 'opt_newarray_send', %q{ [3, x = 2, 1].min == 1 }, ],
+  [ 'opt_newarray_send', <<-'},', ], # {
     class Array
       def min
         true
diff --git a/compile.c b/compile.c
index 819137006c..be5fd9c345 100644
--- a/compile.c
+++ b/compile.c
@@ -3706,18 +3706,23 @@ iseq_specialized_instruction(rb_iseq_t *iseq, INSN *iobj)
         if (IS_INSN_ID(niobj, send)) {
             const struct rb_callinfo *ci = (struct rb_callinfo *)OPERAND_AT(niobj, 0);
             if ((vm_ci_flag(ci) & VM_CALL_ARGS_SIMPLE) && vm_ci_argc(ci) == 0) {
-                switch (vm_ci_mid(ci)) {
-                  case idMax:
-                    iobj->insn_id = BIN(opt_newarray_max);
-                    ELEM_REMOVE(&niobj->link);
-                    return COMPILE_OK;
-                  case idMin:
-                    iobj->insn_id = BIN(opt_newarray_min);
-                    ELEM_REMOVE(&niobj->link);
-                    return COMPILE_OK;
-                }
-            }
-        }
+		switch (vm_ci_mid(ci)) {
+		  case idMax:
+		  case idMin:
+		  case idHash:
+                      {
+                          rb_num_t num = (rb_num_t)iobj->operands[0];
+                          iobj->insn_id = BIN(opt_newarray_send);
+                          iobj->operands = compile_data_calloc2(iseq, insn_len(iobj->insn_id) - 1, sizeof(VALUE));
+                          iobj->operands[0] = (VALUE)num;
+                          iobj->operands[1] = (VALUE)rb_id2sym(vm_ci_mid(ci));
+                          iobj->operand_size = insn_len(iobj->insn_id) - 1;
+                          ELEM_REMOVE(&niobj->link);
+                          return COMPILE_OK;
+                      }
+		}
+	    }
+	}
     }
 
     if (IS_INSN_ID(iobj, send)) {
diff --git a/defs/id.def b/defs/id.def
index c8d7b6af3e..2ddde7be70 100644
--- a/defs/id.def
+++ b/defs/id.def
@@ -2,6 +2,7 @@
 firstline, predefined = __LINE__+1, %[\
   max
   min
+  hash
   freeze
   nil?
   inspect
diff --git a/insns.def b/insns.def
index ba16d6365c..a457372112 100644
--- a/insns.def
+++ b/insns.def
@@ -899,8 +899,8 @@ opt_str_uminus
 }
 
 DEFINE_INSN
-opt_newarray_max
-(rb_num_t num)
+opt_newarray_send
+(rb_num_t num, ID method)
 (...)
 (VALUE val)
 /* This instruction typically has no funcalls.  But it compares array
@@ -909,20 +909,21 @@ opt_newarray_max
  * cannot but mark it being not leaf. */
 // attr bool leaf = false; /* has rb_funcall() */
 // attr rb_snum_t sp_inc = 1 - (rb_snum_t)num;
-{
-    val = vm_opt_newarray_max(ec, num, STACK_ADDR_FROM_TOP(num));
-}
-
-DEFINE_INSN
-opt_newarray_min
-(rb_num_t num)
-(...)
-(VALUE val)
-/* Same discussion as opt_newarray_max. */
-// attr bool leaf = false; /* has rb_funcall() */
-// attr rb_snum_t sp_inc = 1 - (rb_snum_t)num;
-{
-    val = vm_opt_newarray_min(ec, num, STACK_ADDR_FROM_TOP(num));
+// attr rb_snum_t comptime_sp_inc = 1 - (rb_snum_t)num;
+{
+    switch(method) {
+      case idHash:
+        val = vm_opt_newarray_hash(ec, num, STACK_ADDR_FROM_TOP(num));
+        break;
+      case idMin:
+        val = vm_opt_newarray_min(ec, num, STACK_ADDR_FROM_TOP(num));
+        break;
+      case idMax:
+        val = vm_opt_newarray_max(ec, num, STACK_ADDR_FROM_TOP(num));
+        break;
+      default:
+        rb_bug("unreachable");
+    }
 }
 
 /* super(args) # args.size => num */
diff --git a/internal/array.h b/internal/array.h
index 3aeb1be2dd..7e55b66a87 100644
--- a/internal/array.h
+++ b/internal/array.h
@@ -23,6 +23,7 @@
 #define RARRAY_PTR_IN_USE_FLAG  FL_USER14
 
 /* array.c */
+VALUE rb_ary_hash_values(long len, const VALUE *elements);
 VALUE rb_ary_last(int, const VALUE *, VALUE);
 void rb_ary_set_len(VALUE, long);
 void rb_ary_delete_same(VALUE, VALUE);
diff --git a/internal/basic_operators.h b/internal/basic_operators.h
index 12a0475990..a59403631e 100644
--- a/internal/basic_operators.h
+++ b/internal/basic_operators.h
@@ -31,6 +31,7 @@ enum ruby_basic_operators {
     BOP_UMINUS,
     BOP_MAX,
     BOP_MIN,
+    BOP_HASH,
     BOP_CALL,
     BOP_AND,
     BOP_OR,
diff --git a/vm.c b/vm.c
index bc007e21e5..b3fce8d4fb 100644
--- a/vm.c
+++ b/vm.c
@@ -2051,6 +2051,7 @@ vm_init_redefined_flag(void)
     OP(UMinus, UMINUS), (C(String));
     OP(Max, MAX), (C(Array));
     OP(Min, MIN), (C(Array));
+    OP(Hash, HASH), (C(Array));
     OP(Call, CALL), (C(Proc));
     OP(And, AND), (C(Integer));
     OP(Or, OR), (C(Integer));
diff --git a/vm_insnhelper.c b/vm_insnhelper.c
index f4894225d7..c9805377ff 100644
--- a/vm_insnhelper.c
+++ b/vm_insnhelper.c
@@ -5377,6 +5377,18 @@ rb_vm_opt_newarray_min(rb_execution_context_t *ec, rb_num_t num, const VALUE *pt
     return vm_opt_newarray_min(ec, num, ptr);
 }
 
+static VALUE
+vm_opt_newarray_hash(rb_execution_context_t *ec, rb_num_t num, const VALUE *ptr)
+{
+    // If Array#hash is _not_ monkeypatched, use the optimized call
+    if (BASIC_OP_UNREDEFINED_P(BOP_HASH, ARRAY_REDEFINED_OP_FLAG)) {
+        return rb_ary_hash_values(num, ptr);
+    }
+    else {
+        return rb_vm_call_with_refinements(ec, rb_ary_new4(num, ptr), idHash, 0, NULL, RB_NO_KEYWORDS);
+    }
+}
+
 #undef id_cmp
 
 #define IMEMO_CONST_CACHE_SHAREABLE IMEMO_FL_USER0