Remove has_source_hash from iseq_body

We can use a special value of 0 in source_hash to denote that it does not
have source_hash. This removes the has_source_hash field which saves us 8
bytes per iseq.

compile.c

@@ -1506,7 +1506,7 @@ new_child_iseq(rb_iseq_t *iseq, const NODE *const node,
 
     // The child AST wrapper does not carry the source hash, so copy it from
     // the enclosing iseq before compiling, for grandchildren to inherit it.
-    if (ISEQ_BODY(iseq)->has_source_hash) {
+    if (ISEQ_BODY(iseq)->source_hash) {
         rb_ast_t *child_ast = rb_ruby_ast_data_get(ast_value);
         child_ast->body.source_hash = ISEQ_BODY(iseq)->source_hash;
         child_ast->body.has_source_hash = 1;
@@ -12481,7 +12481,6 @@ rb_iseq_build_from_ary(rb_iseq_t *iseq, VALUE misc, VALUE locals, VALUE params,
     VALUE source_hash = rb_hash_aref(misc, ID2SYM(rb_intern("source_hash")));
     if (!NIL_P(source_hash)) {
         ISEQ_BODY(iseq)->source_hash = NUM2ULL(source_hash);
-        ISEQ_BODY(iseq)->has_source_hash = true;
     }
 
     VALUE node_ids = Qfalse;
@@ -13790,12 +13789,10 @@ ibf_dump_iseq_each(struct ibf_dump *dump, const rb_iseq_t *iseq)
     ibf_dump_write_small_value(dump, location_label_index);
     ibf_dump_write_small_value(dump, body->location.first_lineno);
     ibf_dump_write_small_value(dump, body->location.node_id);
-    /* Dump the source hash in two 32-bit halves, because VALUE may be
-     * 32 bits wide. */
-    uint64_t source_hash = body->has_source_hash ? body->source_hash : 0;
-    ibf_dump_write_small_value(dump, (VALUE)(uint32_t)(source_hash >> 32));
-    ibf_dump_write_small_value(dump, (VALUE)(uint32_t)source_hash);
-    ibf_dump_write_small_value(dump, body->has_source_hash ? 1 : 0);
+    /* Dump the source hash (0 if unavailable) in two 32-bit halves, because
+     * VALUE may be 32 bits wide. */
+    ibf_dump_write_small_value(dump, (VALUE)(uint32_t)(body->source_hash >> 32));
+    ibf_dump_write_small_value(dump, (VALUE)(uint32_t)body->source_hash);
     ibf_dump_write_small_value(dump, body->location.code_location.beg_pos.lineno);
     ibf_dump_write_small_value(dump, body->location.code_location.beg_pos.column);
     ibf_dump_write_small_value(dump, body->location.code_location.end_pos.lineno);
@@ -13911,7 +13908,6 @@ ibf_load_iseq_each(struct ibf_load *load, rb_iseq_t *iseq, ibf_offset_t offset)
     const uint64_t source_hash_hi = (uint64_t)ibf_load_small_value(load, &reading_pos);
     const uint64_t source_hash_lo = (uint64_t)ibf_load_small_value(load, &reading_pos);
     const uint64_t source_hash = (source_hash_hi << 32) | (uint32_t)source_hash_lo;
-    const bool has_source_hash = ibf_load_small_value(load, &reading_pos) != 0;
     const int location_code_location_beg_pos_lineno = (int)ibf_load_small_value(load, &reading_pos);
     const int location_code_location_beg_pos_column = (int)ibf_load_small_value(load, &reading_pos);
     const int location_code_location_end_pos_lineno = (int)ibf_load_small_value(load, &reading_pos);
@@ -14013,7 +14009,6 @@ ibf_load_iseq_each(struct ibf_load *load, rb_iseq_t *iseq, ibf_offset_t offset)
     load_body->location.first_lineno = location_first_lineno;
     load_body->location.node_id = location_node_id;
     load_body->source_hash = source_hash;
-    load_body->has_source_hash = has_source_hash;
     load_body->location.code_location.beg_pos.lineno = location_code_location_beg_pos_lineno;
     load_body->location.code_location.beg_pos.column = location_code_location_beg_pos_column;
     load_body->location.code_location.end_pos.lineno = location_code_location_end_pos_lineno;

iseq.c

@@ -1140,7 +1140,7 @@ rb_iseq_new_with_opt(VALUE ast_value, VALUE name, VALUE path, VALUE realpath,
 
     if (body && body->has_source_hash) {
         ISEQ_BODY(iseq)->source_hash = body->source_hash;
-        ISEQ_BODY(iseq)->has_source_hash = true;
+        RUBY_ASSERT(ISEQ_BODY(iseq)->source_hash != 0);
     }
 
     rb_iseq_compile_node(iseq, node);
@@ -1164,7 +1164,7 @@ pm_iseq_build(pm_scope_node_t *node, VALUE name, VALUE path, VALUE realpath,
     ISEQ_BODY(iseq)->prism = true;
 
     ISEQ_BODY(iseq)->source_hash = node->source_hash;
-    ISEQ_BODY(iseq)->has_source_hash = true;
+    RUBY_ASSERT(ISEQ_BODY(iseq)->source_hash != 0);
 
     rb_compile_option_t next_option;
     if (!option) option = &COMPILE_OPTION_DEFAULT;
@@ -3764,7 +3764,7 @@ iseq_data_to_ary(const rb_iseq_t *iseq)
     rb_hash_aset(misc, ID2SYM(rb_intern("local_size")), INT2FIX(iseq_body->local_table_size));
     rb_hash_aset(misc, ID2SYM(rb_intern("stack_max")), INT2FIX(iseq_body->stack_max));
     rb_hash_aset(misc, ID2SYM(rb_intern("node_id")), INT2FIX(iseq_body->location.node_id));
-    rb_hash_aset(misc, ID2SYM(rb_intern("source_hash")), iseq_body->has_source_hash ? ULL2NUM(iseq_body->source_hash) : Qnil);
+    rb_hash_aset(misc, ID2SYM(rb_intern("source_hash")), iseq_body->source_hash ? ULL2NUM(iseq_body->source_hash) : Qnil);
     rb_hash_aset(misc, ID2SYM(rb_intern("code_location")),
             rb_ary_new_from_args(4,
                 INT2FIX(iseq_body->location.code_location.beg_pos.lineno),
@@ -4607,7 +4607,7 @@ static VALUE
 iseqw_source_hash(VALUE self)
 {
     const rb_iseq_t *iseq = iseqw_check(self);
-    if (!ISEQ_BODY(iseq)->has_source_hash) return Qnil;
+    if (!ISEQ_BODY(iseq)->source_hash) return Qnil;
     return ULL2NUM(ISEQ_BODY(iseq)->source_hash);
 }
 

ruby_parser.c

@@ -1119,7 +1119,8 @@ rb_source_hash_update(rb_source_hash_state_t *state, const uint8_t *ptr, size_t
 uint64_t
 rb_source_hash_finalize(const rb_source_hash_state_t *state)
 {
-    return state->hash;
+    /* A hash of 0 means no source hash, so remap it to another value. */
+    return state->hash == 0 ? 1 : state->hash;
 }
 
 VALUE

vm_backtrace.c

@@ -601,7 +601,7 @@ location_source_range_m(VALUE self)
     if (node_id == -1) {
         rb_raise(rb_eRuntimeError, "cannot get source range for location without a node ID");
     }
-    if (!ISEQ_BODY(iseq)->has_source_hash) {
+    if (!ISEQ_BODY(iseq)->source_hash) {
         rb_raise(rb_eRuntimeError, "cannot get source range because the source hash is unavailable");
     }
     uint64_t source_hash = ISEQ_BODY(iseq)->source_hash;

vm_core.h

@@ -580,10 +580,10 @@ struct rb_iseq_constant_body {
     void *zjit_payload;
 #endif
 
-    // Hash of the source this iseq was compiled from. Meaningful only when
-    // has_source_hash is set.
+    // Hash of the source this iseq was compiled from, or 0 if it is
+    // unavailable. A computed hash of 0 is remapped to another value, so
+    // 0 never denotes a real hash.
     uint64_t source_hash;
-    bool has_source_hash;
 };
 
 /* T_IMEMO/iseq */