YJIT: Fix ruby2_keywords splat+rest and drop bogus checks

YJIT didn't guard for ruby2_keywords hash in case of splat calls that
land in methods with a rest parameter, creating incorrect results.

The compile-time checks didn't correspond to any actual effects of
ruby2_keywords, so it was masking this bug and YJIT was needlessly
refusing to compile some code. About 16% of fallback reasons in
`lobsters` was due to the ISeq check.

We already handle the tagging part with
exit_if_supplying_kw_and_has_no_kw() and should now have a dynamic guard
for all splat cases.

Note for backporting: You also need 7f51959ff1.

[Bug #20195]

1 files changed, 13 insertions, 0 deletions

diff --git a/yjit.c b/yjit.c
index d5071cb952..cf6d81092f 100644
--- a/yjit.c
+++ b/yjit.c
@@ -871,6 +871,19 @@ rb_yjit_fix_mod_fix(VALUE recv, VALUE obj)
     return rb_fix_mod_fix(recv, obj);
 }
 
+// Return non-zero when `obj` is an array and its last item is a
+// `ruby2_keywords` hash. We don't support this kind of splat.
+size_t
+rb_yjit_ruby2_keywords_splat_p(VALUE obj)
+{
+    if (!RB_TYPE_P(obj, T_ARRAY)) return 0;
+    long len = RARRAY_LEN(obj);
+    if (len == 0) return 0;
+    VALUE last = RARRAY_AREF(obj, len - 1);
+    if (!RB_TYPE_P(last, T_HASH)) return 0;
+    return FL_TEST_RAW(last, RHASH_PASS_AS_KEYWORDS);
+}
+
 // Print the Ruby source location of some ISEQ for debugging purposes
 void
 rb_yjit_dump_iseq_loc(const rb_iseq_t *iseq, uint32_t insn_idx)