compile.c: mark catch_except_p of iseq

which has throw insn, not only ancestor iseqs of it.
I think we should remove catch_except_p flag and try to simplify the
catch table itself, to prevent similar bugs in the future.

test_jit.rb: add test to prevent the bug

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63320 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

1 files changed, 8 insertions, 10 deletions

diff --git a/compile.c b/compile.c
index 2fadf643ad..4fa3e473e2 100644
--- a/compile.c
+++ b/compile.c
@@ -1281,20 +1281,18 @@ update_catch_except_flags(struct rb_iseq_constant_body *body)
 
     /* This assumes that a block has parent_iseq which may catch an exception from the block, and that
        BREAK/NEXT/REDO catch table entries are used only when `throw` insn is used in the block. */
-    if (body->parent_iseq != NULL) {
-        pos = 0;
-        while (pos < body->iseq_size) {
+    pos = 0;
+    while (pos < body->iseq_size) {
 #if OPT_DIRECT_THREADED_CODE || OPT_CALL_THREADED_CODE
-            insn = rb_vm_insn_addr2insn((void *)body->iseq_encoded[pos]);
+        insn = rb_vm_insn_addr2insn((void *)body->iseq_encoded[pos]);
 #else
-            insn = (int)body->iseq_encoded[pos];
+        insn = (int)body->iseq_encoded[pos];
 #endif
-            if (insn == BIN(throw)) {
-                struct rb_iseq_constant_body *parent_body = body->parent_iseq->body;
-                set_catch_except_p(parent_body);
-            }
-            pos += insn_len(insn);
+        if (insn == BIN(throw)) {
+            set_catch_except_p(body);
+            break;
         }
+        pos += insn_len(insn);
     }
 
     if (ct == NULL)