[pull] master from ruby:master

.github/workflows/zjit-macos.yml

@@ -158,7 +158,7 @@ jobs:
         include:
           # Test --call-threshold=2 with 2 iterations in total
           - ruby_opts: '--zjit-call-threshold=2'
-            bench_opts: '--warmup=1 --bench=1 --excludes=fluentd,psych-load,railsbench'
+            bench_opts: '--warmup=1 --bench=1'
             configure: '--enable-zjit=dev_nodebug' # --enable-zjit=dev is too slow
 
     runs-on: macos-14

.github/workflows/zjit-ubuntu.yml

@@ -215,7 +215,7 @@ jobs:
         include:
           # Test --call-threshold=2 with 2 iterations in total
           - ruby_opts: '--zjit-call-threshold=2'
-            bench_opts: '--warmup=1 --bench=1 --excludes=fluentd,psych-load,railsbench'
+            bench_opts: '--warmup=1 --bench=1'
             configure: '--enable-zjit=dev_nodebug' # --enable-zjit=dev is too slow
 
     runs-on: ubuntu-24.04

bootstraptest/test_yjit.rb

@@ -5435,3 +5435,18 @@ def jit_caller = test_body("session_id")
   alias some_method binding # induce environment escape
   test_body(:symbol)
 }
+
+# regression test for missing check in identity method inlining
+assert_normal_exit %q{
+  # Use dead code (if false) to create a local
+  # without initialization instructions.
+  def foo(a)
+    if false
+      x = nil
+    end
+    x
+  end
+  def test = foo(1)
+  test
+  test
+}

string.c

@@ -10263,8 +10263,6 @@ rb_str_lstrip_bang(int argc, VALUE *argv, VALUE str)
     char *start, *s;
     long olen, loffset;
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     str_modify_keep_cr(str);
     enc = STR_ENC_GET(str);
     RSTRING_GETMEM(str, start, olen);
@@ -10324,8 +10322,6 @@ rb_str_lstrip(int argc, VALUE *argv, VALUE str)
     char *start;
     long len, loffset;
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     RSTRING_GETMEM(str, start, len);
     if (argc > 0) {
         char table[TR_TABLE_SIZE];
@@ -10413,8 +10409,6 @@ rb_str_rstrip_bang(int argc, VALUE *argv, VALUE str)
     char *start;
     long olen, roffset;
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     str_modify_keep_cr(str);
     enc = STR_ENC_GET(str);
     RSTRING_GETMEM(str, start, olen);
@@ -10472,8 +10466,6 @@ rb_str_rstrip(int argc, VALUE *argv, VALUE str)
     char *start;
     long olen, roffset;
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     enc = STR_ENC_GET(str);
     RSTRING_GETMEM(str, start, olen);
     if (argc > 0) {
@@ -10510,8 +10502,6 @@ rb_str_strip_bang(int argc, VALUE *argv, VALUE str)
     long olen, loffset, roffset;
     rb_encoding *enc;
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     str_modify_keep_cr(str);
     enc = STR_ENC_GET(str);
     RSTRING_GETMEM(str, start, olen);
@@ -10577,8 +10567,6 @@ rb_str_strip(int argc, VALUE *argv, VALUE str)
     long olen, loffset, roffset;
     rb_encoding *enc = STR_ENC_GET(str);
 
-    rb_check_arity(argc, 0, UNLIMITED_ARGUMENTS);
-
     RSTRING_GETMEM(str, start, olen);
 
     if (argc > 0) {

test/ruby/test_zjit.rb

@@ -355,6 +355,21 @@ def test = foo(1)
     }, call_threshold: 2
   end
 
+  def test_nonparam_local_nil_in_jit_call
+    # Non-parameter locals must be initialized to nil in JIT-to-JIT calls.
+    # Use dead code (if false) to create locals without initialization instructions.
+    # Then eval a string that accesses the uninitialized locals.
+    assert_compiles '["x", "x", "x", "x"]', %q{
+      def f(a)
+        a ||= 1
+        if false; b = 1; end
+        eval("-> { p 'x#{b}' }")
+      end
+
+      4.times.map { f(1).call }
+    }, call_threshold: 2
+  end
+
   def test_setlocal_on_eval
     assert_compiles '1', %q{
       @b = binding

yjit/src/codegen.rs

@@ -7466,6 +7466,12 @@ fn iseq_get_return_value(iseq: IseqPtr, captured_opnd: Option<Opnd>, block: Opti
             let ep_offset = unsafe { *rb_iseq_pc_at_idx(iseq, 1) }.as_u32();
             let local_idx = ep_offset_to_local_idx(iseq, ep_offset);
 
+            // Only inline getlocal on a parameter. DCE in the IESQ builder can
+            // make a two-instruction ISEQ that does not return a parameter.
+            if local_idx >= unsafe { get_iseq_body_param_size(iseq) } {
+                return None;
+            }
+
             if unsafe { rb_simple_iseq_p(iseq) } {
                 return Some(IseqReturn::LocalVariable(local_idx));
             } else if unsafe { rb_iseq_only_kwparam_p(iseq) } {

zjit/src/codegen.rs

@@ -1391,6 +1391,16 @@ fn gen_send_without_block_direct(
         0
     };
 
+    // Fill non-parameter locals with nil (they may be read by eval before being written)
+    let num_params = params.size.to_usize();
+    if local_size > num_params {
+        asm_comment!(asm, "initialize non-parameter locals to nil");
+        for local_idx in num_params..local_size {
+            let offset = local_size_and_idx_to_bp_offset(local_size, local_idx);
+            asm.store(Opnd::mem(64, SP, -offset * SIZEOF_VALUE_I32), Qnil.into());
+        }
+    }
+
     // Make a method call. The target address will be rewritten once compiled.
     let iseq_call = IseqCall::new(iseq, num_optionals_passed);
     let dummy_ptr = cb.get_write_ptr().raw_ptr(cb);
@@ -2369,8 +2379,8 @@ c_callable! {
             let pc = unsafe { rb_iseq_pc_at_idx(iseq, entry_insn_idxs[iseq_call.jit_entry_idx.to_usize()]) };
             unsafe { rb_set_cfp_pc(cfp, pc) };
 
-            // JIT-to-JIT calls don't set SP or fill nils to uninitialized (non-argument) locals.
-            // We need to set them if we side-exit from function_stub_hit.
+            // Successful JIT-to-JIT calls fill nils to non-parameter locals in generated code.
+            // If we side-exit from function_stub_hit (before JIT code runs), we need to set them here.
             fn prepare_for_exit(iseq: IseqPtr, cfp: CfpPtr, sp: *mut VALUE, compile_error: &CompileError) {
                 unsafe {
                     // Set SP which gen_push_frame() doesn't set