8331935: Add support for primitive array C1 clone intrinsic in PPC

Reviewed-by: mdoerr, amitkumar

src/hotspot/cpu/ppc/c1_LIRAssembler_ppc.cpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2000, 2024, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2023 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2024 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -1827,18 +1827,17 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
 
   int flags = op->flags();
   ciArrayKlass* default_type = op->expected_type();
-  BasicType basic_type = default_type != nullptr ? default_type->element_type()->basic_type() : T_ILLEGAL;
+  BasicType basic_type = (default_type != nullptr) ? default_type->element_type()->basic_type() : T_ILLEGAL;
   if (basic_type == T_ARRAY) basic_type = T_OBJECT;
 
   // Set up the arraycopy stub information.
   ArrayCopyStub* stub = op->stub();
-  const int frame_resize = frame::native_abi_reg_args_size - sizeof(frame::java_abi); // C calls need larger frame.
 
   // Always do stub if no type information is available. It's ok if
   // the known type isn't loaded since the code sanity checks
   // in debug mode and the type isn't required when we know the exact type
   // also check that the type is an array type.
-  if (op->expected_type() == nullptr) {
+  if (default_type == nullptr) {
     assert(src->is_nonvolatile() && src_pos->is_nonvolatile() && dst->is_nonvolatile() && dst_pos->is_nonvolatile() &&
            length->is_nonvolatile(), "must preserve");
     address copyfunc_addr = StubRoutines::generic_arraycopy();
@@ -1873,7 +1872,7 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
     return;
   }
 
-  assert(default_type != nullptr && default_type->is_array_klass(), "must be true at this point");
+  assert(default_type != nullptr && default_type->is_array_klass() && default_type->is_loaded(), "must be true at this point");
   Label cont, slow, copyfunc;
 
   bool simple_check_flag_set = flags & (LIR_OpArrayCopy::src_null_check |
@@ -1968,7 +1967,11 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
   int shift = shift_amount(basic_type);
 
   if (!(flags & LIR_OpArrayCopy::type_check)) {
-    __ b(cont);
+    if (stub != nullptr) {
+      __ b(cont);
+      __ bind(slow);
+      __ b(*stub->entry());
+    }
   } else {
     // We don't know the array types are compatible.
     if (basic_type != T_OBJECT) {
@@ -2089,9 +2092,9 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
         __ add(dst_pos, tmp, dst_pos);
       }
     }
+    __ bind(slow);
+    __ b(*stub->entry());
   }
-  __ bind(slow);
-  __ b(*stub->entry());
   __ bind(cont);
 
 #ifdef ASSERT
@@ -2104,7 +2107,7 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
     // subtype which we can't check or src is the same array as dst
     // but not necessarily exactly of type default_type.
     Label known_ok, halt;
-    metadata2reg(op->expected_type()->constant_encoding(), tmp);
+    metadata2reg(default_type->constant_encoding(), tmp);
     if (UseCompressedClassPointers) {
       // Tmp holds the default type. It currently comes uncompressed after the
       // load of a constant, so encode it.
@@ -2180,7 +2183,9 @@ void LIR_Assembler::emit_arraycopy(LIR_OpArrayCopy* op) {
   __ mr(len, length);
   __ call_c_with_frame_resize(entry, /*stub does not need resized frame*/ 0);
 
-  __ bind(*stub->continuation());
+  if (stub != nullptr) {
+    __ bind(*stub->continuation());
+  }
 }
 
 
@@ -2301,7 +2306,8 @@ void LIR_Assembler::emit_alloc_array(LIR_OpAllocArray* op) {
                       arrayOopDesc::base_offset_in_bytes(op->type()),
                       type2aelembytes(op->type()),
                       op->klass()->as_register(),
-                      *op->stub()->entry());
+                      *op->stub()->entry(),
+                      op->zero_array());
   }
   __ bind(*op->stub()->continuation());
 }

src/hotspot/cpu/ppc/c1_LIRGenerator_ppc.cpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2005, 2024, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2023 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2024 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -758,7 +758,13 @@ void LIRGenerator::do_ArrayCopy(Intrinsic* x) {
   assert(x->number_of_arguments() == 5, "wrong type");
 
   // Make all state_for calls early since they can emit code.
-  CodeEmitInfo* info = state_for(x, x->state());
+  CodeEmitInfo* info = nullptr;
+  if (x->state_before() != nullptr && x->state_before()->force_reexecute()) {
+    info = state_for(x, x->state_before());
+    info->set_force_reexecute();
+  } else {
+    info = state_for(x, x->state());
+  }
 
   LIRItem src     (x->argument_at(0), this);
   LIRItem src_pos (x->argument_at(1), this);
@@ -778,7 +784,9 @@ void LIRGenerator::do_ArrayCopy(Intrinsic* x) {
   int flags;
   ciArrayKlass* expected_type;
   arraycopy_helper(x, &flags, &expected_type);
-
+  if (x->check_flag(Instruction::OmitChecksFlag)) {
+    flags = 0;
+  }
   __ arraycopy(src.result(), src_pos.result(), dst.result(), dst_pos.result(),
                length.result(), tmp,
                expected_type, flags, info);
@@ -903,7 +911,13 @@ void LIRGenerator::do_NewInstance(NewInstance* x) {
 
 void LIRGenerator::do_NewTypeArray(NewTypeArray* x) {
   // Evaluate state_for early since it may emit code.
-  CodeEmitInfo* info = state_for(x, x->state());
+  CodeEmitInfo* info = nullptr;
+  if (x->state_before() != nullptr && x->state_before()->force_reexecute()) {
+    info = state_for(x, x->state_before());
+    info->set_force_reexecute();
+  } else {
+    info = state_for(x, x->state());
+  }
 
   LIRItem length(x->length(), this);
   length.load_item();
@@ -921,7 +935,7 @@ void LIRGenerator::do_NewTypeArray(NewTypeArray* x) {
   __ metadata2reg(ciTypeArrayKlass::make(elem_type)->constant_encoding(), klass_reg);
 
   CodeStub* slow_path = new NewTypeArrayStub(klass_reg, len, reg, info);
-  __ allocate_array(reg, len, tmp1, tmp2, tmp3, tmp4, elem_type, klass_reg, slow_path);
+  __ allocate_array(reg, len, tmp1, tmp2, tmp3, tmp4, elem_type, klass_reg, slow_path, x->zero_array());
 
   // Must prevent reordering of stores for object initialization
   // with stores that publish the new object.

src/hotspot/cpu/ppc/c1_MacroAssembler_ppc.cpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 1999, 2024, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2018 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2024 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -310,7 +310,8 @@ void C1_MacroAssembler::allocate_array(
   int      base_offset_in_bytes,       // elements offset in bytes
   int      elt_size,                   // element size in bytes
   Register klass,                      // object klass
-  Label&   slow_case                   // continuation point if fast allocation fails
+  Label&   slow_case,                  // continuation point if fast allocation fails
+  bool     zero_array                  // zero the allocated array or not
 ) {
   assert_different_registers(obj, len, t1, t2, t3, klass);
 
@@ -346,23 +347,25 @@ void C1_MacroAssembler::allocate_array(
   try_allocate(obj, arr_size, 0, t2, t3, slow_case);
   initialize_header(obj, klass, len, t2, t3);
 
-  // Initialize body.
+  if (zero_array) {
-  const Register base  = t2;
+    // Initialize body.
-  const Register index = t3;
+    const Register base  = t2;
-  addi(base, obj, base_offset_in_bytes);               // compute address of first element
+    const Register index = t3;
-  addi(index, arr_size, -(base_offset_in_bytes));      // compute index = number of bytes to clear
+    addi(base, obj, base_offset_in_bytes);               // compute address of first element
+    addi(index, arr_size, -(base_offset_in_bytes));      // compute index = number of bytes to clear
 
-  // Zero first 4 bytes, if start offset is not word aligned.
+    // Zero first 4 bytes, if start offset is not word aligned.
-  if (!is_aligned(base_offset_in_bytes, BytesPerWord)) {
+    if (!is_aligned(base_offset_in_bytes, BytesPerWord)) {
-    assert(is_aligned(base_offset_in_bytes, BytesPerInt), "must be 4-byte aligned");
+      assert(is_aligned(base_offset_in_bytes, BytesPerInt), "must be 4-byte aligned");
-    li(t1, 0);
+      li(t1, 0);
-    stw(t1, 0, base);
+      stw(t1, 0, base);
-    addi(base, base, BytesPerInt);
+      addi(base, base, BytesPerInt);
-    // Note: initialize_body will align index down, no need to correct it here.
+      // Note: initialize_body will align index down, no need to correct it here.
+    }
+
+    initialize_body(base, index);
   }
 
-  initialize_body(base, index);
-
   if (CURRENT_ENV->dtrace_alloc_probes()) {
     Unimplemented();
     //assert(obj == O0, "must be");

src/hotspot/cpu/ppc/c1_MacroAssembler_ppc.hpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 1999, 2024, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2024 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -83,7 +83,8 @@
     int      base_offset_in_bytes,     // elements offset in bytes
     int      elt_size,                 // element size in bytes
     Register klass,                    // object klass
-    Label&   slow_case                 // continuation point if fast allocation fails
+    Label&   slow_case,                // continuation point if fast allocation fails
+    bool     zero_array                // zero the allocated array or not
   );
 
   void null_check(Register r, Label *Lnull = nullptr);

src/hotspot/share/c1/c1_Compiler.cpp

@@ -235,7 +235,7 @@ bool Compiler::is_intrinsic_supported(vmIntrinsics::ID id) {
   case vmIntrinsics::_counterTime:
 #endif
   case vmIntrinsics::_getObjectSize:
-#if defined(X86) || defined(AARCH64) || defined(S390) || defined(RISCV)
+#if defined(X86) || defined(AARCH64) || defined(S390) || defined(RISCV) || defined(PPC64)
   case vmIntrinsics::_clone:
 #endif
     break;

src/hotspot/share/c1/c1_LIR.cpp

@@ -351,7 +351,7 @@ LIR_OpArrayCopy::LIR_OpArrayCopy(LIR_Opr src, LIR_Opr src_pos, LIR_Opr dst, LIR_
   , _tmp(tmp)
   , _expected_type(expected_type)
   , _flags(flags) {
-#if defined(X86) || defined(AARCH64) || defined(S390) || defined(RISCV)
+#if defined(X86) || defined(AARCH64) || defined(S390) || defined(RISCV) || defined(PPC64)
   if (expected_type != nullptr && flags == 0) {
     _stub = nullptr;
   } else {