Implement WIldcardfinder Plugin directly into the JDK

Reviewed-by: lmesnik, alanb, coleenp

.github/workflows/main.yml

@@ -36,7 +36,7 @@ on:
       platforms:
         description: 'Platform(s) to execute on (comma separated, e.g. "linux-x64, macos, aarch64")'
         required: true
-        default: 'linux-x64, linux-x86-hs, linux-x64-variants, linux-cross-compile, alpine-linux-x64, macos-x64, macos-aarch64, windows-x64, windows-aarch64, docs'
+        default: 'linux-x64, linux-x64-variants, linux-cross-compile, alpine-linux-x64, macos-x64, macos-aarch64, windows-x64, windows-aarch64, docs'
       configure-arguments:
         description: 'Additional configure arguments'
         required: false
@@ -62,7 +62,6 @@ jobs:
       EXCLUDED_PLATFORMS: 'alpine-linux-x64'
     outputs:
       linux-x64: ${{ steps.include.outputs.linux-x64 }}
-      linux-x86-hs: ${{ steps.include.outputs.linux-x86-hs }}
       linux-x64-variants: ${{ steps.include.outputs.linux-x64-variants }}
       linux-cross-compile: ${{ steps.include.outputs.linux-cross-compile }}
       alpine-linux-x64: ${{ steps.include.outputs.alpine-linux-x64 }}
@@ -145,7 +144,6 @@ jobs:
           }
 
           echo "linux-x64=$(check_platform linux-x64 linux x64)" >> $GITHUB_OUTPUT
-          echo "linux-x86-hs=$(check_platform linux-x86-hs linux x86)" >> $GITHUB_OUTPUT
           echo "linux-x64-variants=$(check_platform linux-x64-variants variants)" >> $GITHUB_OUTPUT
           echo "linux-cross-compile=$(check_platform linux-cross-compile cross-compile)" >> $GITHUB_OUTPUT
           echo "alpine-linux-x64=$(check_platform alpine-linux-x64 alpine-linux x64)" >> $GITHUB_OUTPUT
@@ -170,24 +168,6 @@ jobs:
       make-arguments: ${{ github.event.inputs.make-arguments }}
     if: needs.prepare.outputs.linux-x64 == 'true'
 
-  build-linux-x86-hs:
-    name: linux-x86-hs
-    needs: prepare
-    uses: ./.github/workflows/build-linux.yml
-    with:
-      platform: linux-x86
-      make-target: 'hotspot'
-      gcc-major-version: '10'
-      gcc-package-suffix: '-multilib'
-      apt-architecture: 'i386'
-      # Some multilib libraries do not have proper inter-dependencies, so we have to
-      # install their dependencies manually.
-      apt-extra-packages: 'libfreetype-dev:i386 libtiff-dev:i386 libcupsimage2-dev:i386 libffi-dev:i386'
-      extra-conf-options: '--with-target-bits=32 --enable-fallback-linker --enable-libffi-bundling'
-      configure-arguments: ${{ github.event.inputs.configure-arguments }}
-      make-arguments: ${{ github.event.inputs.make-arguments }}
-    if: needs.prepare.outputs.linux-x86-hs == 'true'
-
   build-linux-x64-hs-nopch:
     name: linux-x64-hs-nopch
     needs: prepare

doc/building.html

@@ -329,8 +329,8 @@ GB of free disk space is required.</p>
 <p>Even for 32-bit builds, it is recommended to use a 64-bit build
 machine, and instead create a 32-bit target using
 <code>--with-target-bits=32</code>.</p>
-<p>Note: The Windows 32-bit x86 port is deprecated and may be removed in
+<p>Note: The 32-bit x86 port is deprecated and may be removed in a
-a future release.</p>
+future release.</p>
 <h3 id="building-on-aarch64">Building on aarch64</h3>
 <p>At a minimum, a machine with 8 cores is advisable, as well as 8 GB of
 RAM. (The more cores to use, the more memory you need.) At least 6 GB of
@@ -393,8 +393,7 @@ Build Platforms</a>. From time to time, this is updated by contributors
 to list successes or failures of building on different platforms.</p>
 <h3 id="windows">Windows</h3>
 <p>Windows XP is not a supported platform, but all newer Windows should
-be able to build the JDK. (Note: The Windows 32-bit x86 port is
+be able to build the JDK.</p>
-deprecated and may be removed in a future release.)</p>
 <p>On Windows, it is important that you pay attention to the
 instructions in the <a href="#special-considerations">Special
 Considerations</a>.</p>

doc/building.md

@@ -134,8 +134,7 @@ space is required.
 Even for 32-bit builds, it is recommended to use a 64-bit build machine, and
 instead create a 32-bit target using `--with-target-bits=32`.
 
-Note: The Windows 32-bit x86 port is deprecated and may be removed in a future
+Note: The 32-bit x86 port is deprecated and may be removed in a future release.
-release.
 
 ### Building on aarch64
 
@@ -191,8 +190,7 @@ on different platforms.
 ### Windows
 
 Windows XP is not a supported platform, but all newer Windows should be able to
-build the JDK. (Note: The Windows 32-bit x86 port is deprecated and may be
+build the JDK.
-removed in a future release.)
 
 On Windows, it is important that you pay attention to the instructions in the
 [Special Considerations](#special-considerations).

make/Main.gmk

@@ -1344,9 +1344,13 @@ test-hotspot-jtreg-native: test-hotspot_native_sanity
 test-hotspot-gtest: exploded-test-gtest
 test-jdk-jtreg-native: test-jdk_native_sanity
 
+# Set dependencies for doc tests
+$(eval $(call AddTestDependency, docs_all, docs-jdk))
+test-docs: test-docs_all
+
 ALL_TARGETS += $(RUN_TEST_TARGETS) run-test exploded-run-test check \
     test-hotspot-jtreg test-hotspot-jtreg-native test-hotspot-gtest \
-    test-jdk-jtreg-native
+    test-jdk-jtreg-native test-docs
 
 ################################################################################
 ################################################################################

make/MainSupport.gmk

@@ -137,6 +137,15 @@ define CleanModule
   $(call Clean-include, $1)
 endef
 
+define AddTestDependency
+  test-$(strip $1): $2
+
+  exploded-test-$(strip $1): $2
+
+  ifneq ($(filter $(TEST), $1), )
+    TEST_DEPS += $2
+  endif
+endef
 
 ################################################################################
 

make/autoconf/platform.m4

@@ -666,14 +666,14 @@ AC_DEFUN([PLATFORM_CHECK_DEPRECATION],
 [
   AC_ARG_ENABLE(deprecated-ports, [AS_HELP_STRING([--enable-deprecated-ports@<:@=yes/no@:>@],
       [Suppress the error when configuring for a deprecated port @<:@no@:>@])])
-  # if test "x$OPENJDK_TARGET_CPU" = xx86; then
+  if test "x$OPENJDK_TARGET_CPU" = xx86; then
-  #   if test "x$enable_deprecated_ports" = "xyes"; then
+    if test "x$enable_deprecated_ports" = "xyes"; then
-  #     AC_MSG_WARN([The x86 port is deprecated and may be removed in a future release.])
+      AC_MSG_WARN([The 32-bit x86 port is deprecated and may be removed in a future release.])
-  #   else
+    else
-  #     AC_MSG_ERROR(m4_normalize([The 32-bit x86 port is deprecated and may be removed in a future release.
+      AC_MSG_ERROR(m4_normalize([The 32-bit x86 port is deprecated and may be removed in a future release.
-  #       Use --enable-deprecated-ports=yes to suppress this error.]))
+        Use --enable-deprecated-ports=yes to suppress this error.]))
-  #   fi
+    fi
-  # fi
+  fi
 ])
 
 AC_DEFUN_ONCE([PLATFORM_SETUP_OPENJDK_BUILD_OS_VERSION],

make/hotspot/lib/JvmOverrideFiles.gmk

@@ -37,6 +37,10 @@ ifeq ($(TOOLCHAIN_TYPE), gcc)
     # Need extra inlining to collapse shared marking code into the hot marking loop
     BUILD_LIBJVM_shenandoahMark.cpp_CXXFLAGS := --param inline-unit-growth=1000
   endif
+  # disable lto in g1ParScanThreadState because of special inlining/flattening used there
+  ifeq ($(call check-jvm-feature, link-time-opt), true)
+    BUILD_LIBJVM_g1ParScanThreadState.cpp_CXXFLAGS := -fno-lto
+  endif
 endif
 
 LIBJVM_FDLIBM_COPY_OPT_FLAG := $(CXX_O_FLAG_NONE)

make/modules/java.desktop/Java.gmk

@@ -46,18 +46,6 @@ EXCLUDE_FILES += \
     javax/swing/plaf/nimbus/SpinnerPainter.java \
     javax/swing/plaf/nimbus/SplitPanePainter.java \
     javax/swing/plaf/nimbus/TabbedPanePainter.java \
-    sun/awt/resources/security-icon-bw16.png \
-    sun/awt/resources/security-icon-bw24.png \
-    sun/awt/resources/security-icon-bw32.png \
-    sun/awt/resources/security-icon-bw48.png \
-    sun/awt/resources/security-icon-interim16.png \
-    sun/awt/resources/security-icon-interim24.png \
-    sun/awt/resources/security-icon-interim32.png \
-    sun/awt/resources/security-icon-interim48.png \
-    sun/awt/resources/security-icon-yellow16.png \
-    sun/awt/resources/security-icon-yellow24.png \
-    sun/awt/resources/security-icon-yellow32.png \
-    sun/awt/resources/security-icon-yellow48.png \
     sun/awt/X11/java-icon16.png \
     sun/awt/X11/java-icon24.png \
     sun/awt/X11/java-icon32.png \

make/modules/java.desktop/gensrc/GensrcIcons.gmk

@@ -37,23 +37,6 @@ GENSRC_AWT_ICONS_SRC += \
     $(X11_ICONS_PATH_PREFIX)/classes/sun/awt/X11/java-icon32.png \
     $(X11_ICONS_PATH_PREFIX)/classes/sun/awt/X11/java-icon48.png
 
-
-AWT_ICONPATH := $(MODULE_SRC)/share/classes/sun/awt/resources
-
-GENSRC_AWT_ICONS_SRC += \
-    $(AWT_ICONPATH)/security-icon-bw16.png \
-    $(AWT_ICONPATH)/security-icon-interim16.png \
-    $(AWT_ICONPATH)/security-icon-yellow16.png \
-    $(AWT_ICONPATH)/security-icon-bw24.png \
-    $(AWT_ICONPATH)/security-icon-interim24.png \
-    $(AWT_ICONPATH)/security-icon-yellow24.png \
-    $(AWT_ICONPATH)/security-icon-bw32.png \
-    $(AWT_ICONPATH)/security-icon-interim32.png \
-    $(AWT_ICONPATH)/security-icon-yellow32.png \
-    $(AWT_ICONPATH)/security-icon-bw48.png \
-    $(AWT_ICONPATH)/security-icon-interim48.png \
-    $(AWT_ICONPATH)/security-icon-yellow48.png
-
 GENSRC_AWT_ICONS_FILES := $(notdir $(GENSRC_AWT_ICONS_SRC))
 
 GENSRC_AWT_ICONS_SHORT_NAME = $(subst .,_,$(subst -,_,$(1)))

src/demo/share/jfc/Font2DTest/FontPanel.java

@@ -1114,12 +1114,6 @@ public final class FontPanel extends JPanel implements AdjustmentListener {
             /// Position and set size of zoom window as needed
             zoomWindow.setLocation( canvasLoc.x + zoomAreaX, canvasLoc.y + zoomAreaY );
             if ( !nowZooming ) {
-                if ( zoomWindow.getWarningString() != null )
-                  /// If this is not opened as a "secure" window,
-                  /// it has a banner below the zoom dialog which makes it look really BAD
-                  /// So enlarge it by a bit
-                  zoomWindow.setSize( zoomAreaWidth + 1, zoomAreaHeight + 20 );
-                else
                 zoomWindow.setSize( zoomAreaWidth + 1, zoomAreaHeight + 1 );
             }
 

src/hotspot/cpu/aarch64/aarch64.ad

@@ -2632,6 +2632,23 @@ bool Matcher::pd_clone_node(Node* n, Node* m, Matcher::MStack& mstack) {
 // to be subsumed into complex addressing expressions or compute them
 // into registers?
 bool Matcher::pd_clone_address_expressions(AddPNode* m, Matcher::MStack& mstack, VectorSet& address_visited) {
+
+  // Loads and stores with indirect memory input (e.g., volatile loads and
+  // stores) do not subsume the input into complex addressing expressions. If
+  // the addressing expression is input to at least one such load or store, do
+  // not clone the addressing expression. Query needs_acquiring_load and
+  // needs_releasing_store as a proxy for indirect memory input, as it is not
+  // possible to directly query for indirect memory input at this stage.
+  for (DUIterator_Fast imax, i = m->fast_outs(imax); i < imax; i++) {
+    Node* n = m->fast_out(i);
+    if (n->is_Load() && needs_acquiring_load(n)) {
+      return false;
+    }
+    if (n->is_Store() && needs_releasing_store(n)) {
+      return false;
+    }
+  }
+
   if (clone_base_plus_offset_address(m, mstack, address_visited)) {
     return true;
   }

src/hotspot/cpu/aarch64/c2_globals_aarch64.hpp

@@ -66,6 +66,7 @@ define_pd_global(bool, OptoScheduling,               false);
 define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoRegScheduling,            false);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  true);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 8);
 define_pd_global(bool, IdealizeClearArrayNode,       true);
 
 define_pd_global(intx, ReservedCodeCacheSize,        48*M);

src/hotspot/cpu/aarch64/templateInterpreterGenerator_aarch64.cpp

@@ -620,7 +620,7 @@ address TemplateInterpreterGenerator::generate_cont_resume_interpreter_adapter()
   // Restore Java expression stack pointer
   __ ldr(rscratch1, Address(rfp, frame::interpreter_frame_last_sp_offset * wordSize));
   __ lea(esp, Address(rfp, rscratch1, Address::lsl(Interpreter::logStackElementSize)));
-  // and NULL it as marker that esp is now tos until next java call
+  // and null it as marker that esp is now tos until next java call
   __ str(zr, Address(rfp, frame::interpreter_frame_last_sp_offset * wordSize));
 
   // Restore machine SP

src/hotspot/cpu/arm/c2_globals_arm.hpp

@@ -64,6 +64,7 @@ define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoScheduling,               true);
 define_pd_global(bool, OptoRegScheduling,            false);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  false);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 16);
 define_pd_global(bool, IdealizeClearArrayNode,       true);
 
 #ifdef _LP64

src/hotspot/cpu/ppc/c2_globals_ppc.hpp

@@ -59,6 +59,7 @@ define_pd_global(bool, UseCISCSpill,                 false);
 define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoRegScheduling,            false);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  true);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 16);
 // GL:
 // Detected a problem with unscaled compressed oops and
 // narrow_oop_use_complex_address() == false.

src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.cpp

@@ -2339,83 +2339,6 @@ void C2_MacroAssembler::signum_fp_v(VectorRegister dst, VectorRegister one, Basi
   vfsgnj_vv(dst, one, dst, v0_t);
 }
 
-void C2_MacroAssembler::compress_bits_v(Register dst, Register src, Register mask, bool is_long) {
-  Assembler::SEW sew = is_long ? Assembler::e64 : Assembler::e32;
-  // intrinsic is enabled when MaxVectorSize >= 16
-  Assembler::LMUL lmul = is_long ? Assembler::m4 : Assembler::m2;
-  long len = is_long ? 64 : 32;
-
-  // load the src data(in bits) to be compressed.
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_s_x(v0, src);
-  // reset the src data(in bytes) to zero.
-  mv(t0, len);
-  vsetvli(x0, t0, Assembler::e8, lmul);
-  vmv_v_i(v4, 0);
-  // convert the src data from bits to bytes.
-  vmerge_vim(v4, v4, 1); // v0 as the implicit mask register
-  // reset the dst data(in bytes) to zero.
-  vmv_v_i(v8, 0);
-  // load the mask data(in bits).
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_s_x(v0, mask);
-  // compress the src data(in bytes) to dst(in bytes).
-  vsetvli(x0, t0, Assembler::e8, lmul);
-  vcompress_vm(v8, v4, v0);
-  // convert the dst data from bytes to bits.
-  vmseq_vi(v0, v8, 1);
-  // store result back.
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_x_s(dst, v0);
-}
-
-void C2_MacroAssembler::compress_bits_i_v(Register dst, Register src, Register mask) {
-  compress_bits_v(dst, src, mask, /* is_long */ false);
-}
-
-void C2_MacroAssembler::compress_bits_l_v(Register dst, Register src, Register mask) {
-  compress_bits_v(dst, src, mask, /* is_long */ true);
-}
-
-void C2_MacroAssembler::expand_bits_v(Register dst, Register src, Register mask, bool is_long) {
-  Assembler::SEW sew = is_long ? Assembler::e64 : Assembler::e32;
-  // intrinsic is enabled when MaxVectorSize >= 16
-  Assembler::LMUL lmul = is_long ? Assembler::m4 : Assembler::m2;
-  long len = is_long ? 64 : 32;
-
-  // load the src data(in bits) to be expanded.
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_s_x(v0, src);
-  // reset the src data(in bytes) to zero.
-  mv(t0, len);
-  vsetvli(x0, t0, Assembler::e8, lmul);
-  vmv_v_i(v4, 0);
-  // convert the src data from bits to bytes.
-  vmerge_vim(v4, v4, 1); // v0 as implicit mask register
-  // reset the dst data(in bytes) to zero.
-  vmv_v_i(v12, 0);
-  // load the mask data(in bits).
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_s_x(v0, mask);
-  // expand the src data(in bytes) to dst(in bytes).
-  vsetvli(x0, t0, Assembler::e8, lmul);
-  viota_m(v8, v0);
-  vrgather_vv(v12, v4, v8, VectorMask::v0_t); // v0 as implicit mask register
-  // convert the dst data from bytes to bits.
-  vmseq_vi(v0, v12, 1);
-  // store result back.
-  vsetivli(x0, 1, sew, Assembler::m1);
-  vmv_x_s(dst, v0);
-}
-
-void C2_MacroAssembler::expand_bits_i_v(Register dst, Register src, Register mask) {
-  expand_bits_v(dst, src, mask, /* is_long */ false);
-}
-
-void C2_MacroAssembler::expand_bits_l_v(Register dst, Register src, Register mask) {
-  expand_bits_v(dst, src, mask, /* is_long */ true);
-}
-
 // j.l.Math.round(float)
 //  Returns the closest int to the argument, with ties rounding to positive infinity.
 // We need to handle 3 special cases defined by java api spec:

src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.hpp

@@ -39,9 +39,6 @@
                        VectorRegister vrs,
                        bool is_latin, Label& DONE, Assembler::LMUL lmul);
 
-  void compress_bits_v(Register dst, Register src, Register mask, bool is_long);
-  void expand_bits_v(Register dst, Register src, Register mask, bool is_long);
-
  public:
   // Code used by cmpFastLock and cmpFastUnlock mach instructions in .ad file.
   void fast_lock(Register object, Register box,
@@ -184,13 +181,6 @@
 
   // intrinsic methods implemented by rvv instructions
 
-  // compress bits, i.e. j.l.Integer/Long::compress.
-  void compress_bits_i_v(Register dst, Register src, Register mask);
-  void compress_bits_l_v(Register dst, Register src, Register mask);
-  // expand bits, i.e. j.l.Integer/Long::expand.
-  void expand_bits_i_v(Register dst, Register src, Register mask);
-  void expand_bits_l_v(Register dst, Register src, Register mask);
-
   void java_round_float_v(VectorRegister dst, VectorRegister src, FloatRegister ftmp, BasicType bt, uint vector_length);
   void java_round_double_v(VectorRegister dst, VectorRegister src, FloatRegister ftmp, BasicType bt, uint vector_length);
 

src/hotspot/cpu/riscv/c2_globals_riscv.hpp

@@ -66,6 +66,7 @@ define_pd_global(bool, OptoScheduling,               true);
 define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoRegScheduling,            false);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  true);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 16);
 define_pd_global(bool, IdealizeClearArrayNode,       true);
 
 define_pd_global(intx, ReservedCodeCacheSize,        48*M);

src/hotspot/cpu/riscv/interp_masm_riscv.cpp

@@ -179,15 +179,10 @@ void InterpreterMacroAssembler::check_and_handle_earlyret(Register java_thread)
 
 void InterpreterMacroAssembler::get_unsigned_2_byte_index_at_bcp(Register reg, int bcp_offset) {
   assert(bcp_offset >= 0, "bcp is still pointing to start of bytecode");
-  if (AvoidUnalignedAccesses && (bcp_offset % 2)) {
   lbu(t1, Address(xbcp, bcp_offset));
   lbu(reg, Address(xbcp, bcp_offset + 1));
   slli(t1, t1, 8);
   add(reg, reg, t1);
-  } else {
-    lhu(reg, Address(xbcp, bcp_offset));
-    revb_h_h_u(reg, reg);
-  }
 }
 
 void InterpreterMacroAssembler::get_dispatch() {
@@ -200,15 +195,7 @@ void InterpreterMacroAssembler::get_cache_index_at_bcp(Register index,
                                                        size_t index_size) {
   assert(bcp_offset > 0, "bcp is still pointing to start of bytecode");
   if (index_size == sizeof(u2)) {
-    if (AvoidUnalignedAccesses) {
+    load_short_misaligned(index, Address(xbcp, bcp_offset), tmp, false);
-      assert_different_registers(index, tmp);
-      load_unsigned_byte(index, Address(xbcp, bcp_offset));
-      load_unsigned_byte(tmp, Address(xbcp, bcp_offset + 1));
-      slli(tmp, tmp, 8);
-      add(index, index, tmp);
-    } else {
-      load_unsigned_short(index, Address(xbcp, bcp_offset));
-    }
   } else if (index_size == sizeof(u4)) {
     load_int_misaligned(index, Address(xbcp, bcp_offset), tmp, false);
   } else if (index_size == sizeof(u1)) {
@@ -441,7 +428,14 @@ void InterpreterMacroAssembler::dispatch_base(TosState state,
                                               Register Rs) {
   // Pay attention to the argument Rs, which is acquiesce in t0.
   if (VerifyActivationFrameSize) {
-    Unimplemented();
+    Label L;
+    sub(t1, fp, esp);
+    int min_frame_size =
+      (frame::link_offset - frame::interpreter_frame_initial_sp_offset + frame::metadata_words) * wordSize;
+    sub(t1, t1, min_frame_size);
+    bgez(t1, L);
+    stop("broken stack frame");
+    bind(L);
   }
   if (verifyoop && state == atos) {
     verify_oop(x10);

src/hotspot/cpu/riscv/macroAssembler_riscv.cpp

@@ -3313,14 +3313,11 @@ void MacroAssembler::store_conditional(Register dst,
 }
 
 
-void MacroAssembler::cmpxchg_narrow_value_helper(Register addr, Register expected,
+void MacroAssembler::cmpxchg_narrow_value_helper(Register addr, Register expected, Register new_val,
-                                                 Register new_val,
                                                  enum operand_size size,
-                                                 Register tmp1, Register tmp2, Register tmp3) {
+                                                 Register shift, Register mask, Register aligned_addr) {
   assert(size == int8 || size == int16, "unsupported operand size");
 
-  Register aligned_addr = t1, shift = tmp1, mask = tmp2, not_mask = tmp3;
-
   andi(shift, addr, 3);
   slli(shift, shift, 3);
 
@@ -3335,8 +3332,6 @@ void MacroAssembler::cmpxchg_narrow_value_helper(Register addr, Register expecte
   }
   sll(mask, mask, shift);
 
-  notr(not_mask, mask);
-
   sll(expected, expected, shift);
   andr(expected, expected, mask);
 
@@ -3353,35 +3348,46 @@ void MacroAssembler::cmpxchg_narrow_value(Register addr, Register expected,
                                           Assembler::Aqrl acquire, Assembler::Aqrl release,
                                           Register result, bool result_as_bool,
                                           Register tmp1, Register tmp2, Register tmp3) {
-  Register aligned_addr = t1, shift = tmp1, mask = tmp2, not_mask = tmp3, old = result, tmp = t0;
+  assert_different_registers(addr, expected, new_val, result, tmp1, tmp2, tmp3, t0, t1);
-  assert_different_registers(addr, old, mask, not_mask, new_val, expected, shift, tmp);
+
-  cmpxchg_narrow_value_helper(addr, expected, new_val, size, tmp1, tmp2, tmp3);
+  Register scratch0 = t0, aligned_addr = t1;
+  Register shift = tmp1, mask = tmp2, scratch1 = tmp3;
+
+  cmpxchg_narrow_value_helper(addr, expected, new_val, size, shift, mask, aligned_addr);
 
   Label retry, fail, done;
 
+  if (UseZacas) {
+    lw(result, aligned_addr);
+
+    bind(retry); // amocas loads the current value into result
+    notr(scratch1, mask);
+
+    andr(scratch0, result, scratch1);  // scratch0 = word - cas bits
+    orr(scratch1, expected, scratch0); // scratch1 = non-cas bits + cas bits
+    bne(result, scratch1, fail);       // cas bits differ, cas failed
+
+    // result is the same as expected, use as expected value.
+
+    // scratch0 is still = word - cas bits
+    // Or in the new value to create complete new value.
+    orr(scratch0, scratch0, new_val);
+
+    mv(scratch1, result); // save our expected value
+    atomic_cas(result, scratch0, aligned_addr, operand_size::int32, acquire, release);
+    bne(scratch1, result, retry);
+  } else {
+    notr(scratch1, mask);
     bind(retry);
 
-  if (UseZacas) {
+    lr_w(result, aligned_addr, acquire);
-    lw(old, aligned_addr);
+    andr(scratch0, result, mask);
+    bne(scratch0, expected, fail);
 
-    // if old & mask != expected
+    andr(scratch0, result, scratch1); // scratch1 is ~mask
-    andr(tmp, old, mask);
+    orr(scratch0, scratch0, new_val);
-    bne(tmp, expected, fail);
+    sc_w(scratch0, scratch0, aligned_addr, release);
-
+    bnez(scratch0, retry);
-    andr(tmp, old, not_mask);
-    orr(tmp, tmp, new_val);
-
-    atomic_cas(old, tmp, aligned_addr, operand_size::int32, acquire, release);
-    bne(tmp, old, retry);
-  } else {
-    lr_w(old, aligned_addr, acquire);
-    andr(tmp, old, mask);
-    bne(tmp, expected, fail);
-
-    andr(tmp, old, not_mask);
-    orr(tmp, tmp, new_val);
-    sc_w(tmp, tmp, aligned_addr, release);
-    bnez(tmp, retry);
   }
 
   if (result_as_bool) {
@@ -3393,10 +3399,10 @@ void MacroAssembler::cmpxchg_narrow_value(Register addr, Register expected,
 
     bind(done);
   } else {
-    andr(tmp, old, mask);
-
     bind(fail);
-    srl(result, tmp, shift);
+
+    andr(scratch0, result, mask);
+    srl(result, scratch0, shift);
 
     if (size == int8) {
       sign_extend(result, result, 8);
@@ -3416,33 +3422,44 @@ void MacroAssembler::weak_cmpxchg_narrow_value(Register addr, Register expected,
                                                Assembler::Aqrl acquire, Assembler::Aqrl release,
                                                Register result,
                                                Register tmp1, Register tmp2, Register tmp3) {
-  Register aligned_addr = t1, shift = tmp1, mask = tmp2, not_mask = tmp3, old = result, tmp = t0;
+  assert_different_registers(addr, expected, new_val, result, tmp1, tmp2, tmp3, t0, t1);
-  assert_different_registers(addr, old, mask, not_mask, new_val, expected, shift, tmp);
+
-  cmpxchg_narrow_value_helper(addr, expected, new_val, size, tmp1, tmp2, tmp3);
+  Register scratch0 = t0, aligned_addr = t1;
+  Register shift = tmp1, mask = tmp2, scratch1 = tmp3;
+
+  cmpxchg_narrow_value_helper(addr, expected, new_val, size, shift, mask, aligned_addr);
 
   Label fail, done;
 
   if (UseZacas) {
-    lw(old, aligned_addr);
+    lw(result, aligned_addr);
 
-    // if old & mask != expected
+    notr(scratch1, mask);
-    andr(tmp, old, mask);
-    bne(tmp, expected, fail);
 
-    andr(tmp, old, not_mask);
+    andr(scratch0, result, scratch1);  // scratch0 = word - cas bits
-    orr(tmp, tmp, new_val);
+    orr(scratch1, expected, scratch0); // scratch1 = non-cas bits + cas bits
+    bne(result, scratch1, fail);       // cas bits differ, cas failed
 
-    atomic_cas(tmp, new_val, addr, operand_size::int32, acquire, release);
+    // result is the same as expected, use as expected value.
-    bne(tmp, old, fail);
+
+    // scratch0 is still = word - cas bits
+    // Or in the new value to create complete new value.
+    orr(scratch0, scratch0, new_val);
+
+    mv(scratch1, result); // save our expected value
+    atomic_cas(result, scratch0, aligned_addr, operand_size::int32, acquire, release);
+    bne(scratch1, result, fail); // This weak, so just bail-out.
   } else {
-    lr_w(old, aligned_addr, acquire);
+    notr(scratch1, mask);
-    andr(tmp, old, mask);
-    bne(tmp, expected, fail);
 
-    andr(tmp, old, not_mask);
+    lr_w(result, aligned_addr, acquire);
-    orr(tmp, tmp, new_val);
+    andr(scratch0, result, mask);
-    sc_w(tmp, tmp, aligned_addr, release);
+    bne(scratch0, expected, fail);
-    bnez(tmp, fail);
+
+    andr(scratch0, result, scratch1); // scratch1 is ~mask
+    orr(scratch0, scratch0, new_val);
+    sc_w(scratch0, scratch0, aligned_addr, release);
+    bnez(scratch0, fail);
   }
 
   // Success
@@ -3466,6 +3483,17 @@ void MacroAssembler::cmpxchg(Register addr, Register expected,
   assert_different_registers(expected, t0);
   assert_different_registers(new_val, t0);
 
+  // NOTE:
+  // Register _result_ may be the same register as _new_val_ or _expected_.
+  // Hence do NOT use _result_ until after 'cas'.
+  //
+  // Register _expected_ may be the same register as _new_val_ and is assumed to be preserved.
+  // Hence do NOT change _expected_ or _new_val_.
+  //
+  // Having _expected_ and _new_val_ being the same register is a very puzzling cas.
+  //
+  // TODO: Address these issues.
+
   if (UseZacas) {
     if (result_as_bool) {
       mv(t0, expected);
@@ -3473,8 +3501,9 @@ void MacroAssembler::cmpxchg(Register addr, Register expected,
       xorr(t0, t0, expected);
       seqz(result, t0);
     } else {
-      mv(result, expected);
+      mv(t0, expected);
-      atomic_cas(result, new_val, addr, size, acquire, release);
+      atomic_cas(t0, new_val, addr, size, acquire, release);
+      mv(result, t0);
     }
     return;
   }
@@ -3510,15 +3539,16 @@ void MacroAssembler::cmpxchg_weak(Register addr, Register expected,
                                   enum operand_size size,
                                   Assembler::Aqrl acquire, Assembler::Aqrl release,
                                   Register result) {
-  if (UseZacas) {
-    cmpxchg(addr, expected, new_val, size, acquire, release, result, true);
-    return;
-  }
 
   assert_different_registers(addr, t0);
   assert_different_registers(expected, t0);
   assert_different_registers(new_val, t0);
 
+  if (UseZacas) {
+    cmpxchg(addr, expected, new_val, size, acquire, release, result, true);
+    return;
+  }
+
   Label fail, done;
   load_reserved(t0, addr, size, acquire);
   bne(t0, expected, fail);
@@ -3581,83 +3611,18 @@ ATOMIC_XCHGU(xchgalwu, xchgalw)
 
 #undef ATOMIC_XCHGU
 
-#define ATOMIC_CAS(OP, AOP, ACQUIRE, RELEASE)                                        \
+void MacroAssembler::atomic_cas(Register prev, Register newv, Register addr,
-void MacroAssembler::atomic_##OP(Register prev, Register newv, Register addr) {      \
+                                enum operand_size size, Assembler::Aqrl acquire, Assembler::Aqrl release) {
-  assert(UseZacas, "invariant");                                                     \
-  prev = prev->is_valid() ? prev : zr;                                               \
-  AOP(prev, addr, newv, (Assembler::Aqrl)(ACQUIRE | RELEASE));                       \
-  return;                                                                            \
-}
-
-ATOMIC_CAS(cas, amocas_d, Assembler::relaxed, Assembler::relaxed)
-ATOMIC_CAS(casw, amocas_w, Assembler::relaxed, Assembler::relaxed)
-ATOMIC_CAS(casl, amocas_d, Assembler::relaxed, Assembler::rl)
-ATOMIC_CAS(caslw, amocas_w, Assembler::relaxed, Assembler::rl)
-ATOMIC_CAS(casal, amocas_d, Assembler::aq, Assembler::rl)
-ATOMIC_CAS(casalw, amocas_w, Assembler::aq, Assembler::rl)
-
-#undef ATOMIC_CAS
-
-#define ATOMIC_CASU(OP1, OP2)                                                        \
-void MacroAssembler::atomic_##OP1(Register prev, Register newv, Register addr) {     \
-  atomic_##OP2(prev, newv, addr);                                                    \
-  zero_extend(prev, prev, 32);                                                       \
-  return;                                                                            \
-}
-
-ATOMIC_CASU(caswu, casw)
-ATOMIC_CASU(caslwu, caslw)
-ATOMIC_CASU(casalwu, casalw)
-
-#undef ATOMIC_CASU
-
-void MacroAssembler::atomic_cas(
-    Register prev, Register newv, Register addr, enum operand_size size, Assembler::Aqrl acquire, Assembler::Aqrl release) {
   switch (size) {
     case int64:
-      switch ((Assembler::Aqrl)(acquire | release)) {
+      amocas_d(prev, addr, newv, (Assembler::Aqrl)(acquire | release));
-        case Assembler::relaxed:
-          atomic_cas(prev, newv, addr);
-          break;
-        case Assembler::rl:
-          atomic_casl(prev, newv, addr);
-          break;
-        case Assembler::aqrl:
-          atomic_casal(prev, newv, addr);
-          break;
-        default:
-          ShouldNotReachHere();
-      }
       break;
     case int32:
-      switch ((Assembler::Aqrl)(acquire | release)) {
+      amocas_w(prev, addr, newv, (Assembler::Aqrl)(acquire | release));
-        case Assembler::relaxed:
-          atomic_casw(prev, newv, addr);
-          break;
-        case Assembler::rl:
-          atomic_caslw(prev, newv, addr);
-          break;
-        case Assembler::aqrl:
-          atomic_casalw(prev, newv, addr);
-          break;
-        default:
-          ShouldNotReachHere();
-      }
       break;
     case uint32:
-      switch ((Assembler::Aqrl)(acquire | release)) {
+      amocas_w(prev, addr, newv, (Assembler::Aqrl)(acquire | release));
-        case Assembler::relaxed:
+      zero_extend(prev, prev, 32);
-          atomic_caswu(prev, newv, addr);
-          break;
-        case Assembler::rl:
-          atomic_caslwu(prev, newv, addr);
-          break;
-        case Assembler::aqrl:
-          atomic_casalwu(prev, newv, addr);
-          break;
-        default:
-          ShouldNotReachHere();
-      }
       break;
     default:
       ShouldNotReachHere();

src/hotspot/cpu/riscv/macroAssembler_riscv.hpp

@@ -1146,10 +1146,9 @@ public:
                     enum operand_size size,
                     Assembler::Aqrl acquire, Assembler::Aqrl release,
                     Register result);
-  void cmpxchg_narrow_value_helper(Register addr, Register expected,
+  void cmpxchg_narrow_value_helper(Register addr, Register expected, Register new_val,
-                                   Register new_val,
                                    enum operand_size size,
-                                   Register tmp1, Register tmp2, Register tmp3);
+                                   Register shift, Register mask, Register aligned_addr);
   void cmpxchg_narrow_value(Register addr, Register expected,
                             Register new_val,
                             enum operand_size size,
@@ -1175,16 +1174,6 @@ public:
   void atomic_xchgwu(Register prev, Register newv, Register addr);
   void atomic_xchgalwu(Register prev, Register newv, Register addr);
 
-  void atomic_cas(Register prev, Register newv, Register addr);
-  void atomic_casw(Register prev, Register newv, Register addr);
-  void atomic_casl(Register prev, Register newv, Register addr);
-  void atomic_caslw(Register prev, Register newv, Register addr);
-  void atomic_casal(Register prev, Register newv, Register addr);
-  void atomic_casalw(Register prev, Register newv, Register addr);
-  void atomic_caswu(Register prev, Register newv, Register addr);
-  void atomic_caslwu(Register prev, Register newv, Register addr);
-  void atomic_casalwu(Register prev, Register newv, Register addr);
-
   void atomic_cas(Register prev, Register newv, Register addr, enum operand_size size,
               Assembler::Aqrl acquire = Assembler::relaxed, Assembler::Aqrl release = Assembler::relaxed);
 

src/hotspot/cpu/riscv/riscv.ad

@@ -942,26 +942,6 @@ reg_class v11_reg(
     V11, V11_H, V11_J, V11_K
 );
 
-// class for vector register v12
-reg_class v12_reg(
-    V12, V12_H, V12_J, V12_K
-);
-
-// class for vector register v13
-reg_class v13_reg(
-    V13, V13_H, V13_J, V13_K
-);
-
-// class for vector register v14
-reg_class v14_reg(
-    V14, V14_H, V14_J, V14_K
-);
-
-// class for vector register v15
-reg_class v15_reg(
-    V15, V15_H, V15_J, V15_K
-);
-
 // class for condition codes
 reg_class reg_flags(RFLAGS);
 
@@ -1896,9 +1876,6 @@ bool Matcher::match_rule_supported(int opcode) {
       }
       break;
 
-    case Op_ExpandBits:        // fall through
-    case Op_CompressBits:      // fall through
-      guarantee(UseRVV == (MaxVectorSize >= 16), "UseRVV and MaxVectorSize not matched");
     case Op_StrCompressedCopy: // fall through
     case Op_StrInflatedCopy:   // fall through
     case Op_CountPositives:    // fall through
@@ -3541,46 +3518,6 @@ operand vReg_V11()
   interface(REG_INTER);
 %}
 
-operand vReg_V12()
-%{
-  constraint(ALLOC_IN_RC(v12_reg));
-  match(VecA);
-  match(vReg);
-  op_cost(0);
-  format %{ %}
-  interface(REG_INTER);
-%}
-
-operand vReg_V13()
-%{
-  constraint(ALLOC_IN_RC(v13_reg));
-  match(VecA);
-  match(vReg);
-  op_cost(0);
-  format %{ %}
-  interface(REG_INTER);
-%}
-
-operand vReg_V14()
-%{
-  constraint(ALLOC_IN_RC(v14_reg));
-  match(VecA);
-  match(vReg);
-  op_cost(0);
-  format %{ %}
-  interface(REG_INTER);
-%}
-
-operand vReg_V15()
-%{
-  constraint(ALLOC_IN_RC(v15_reg));
-  match(VecA);
-  match(vReg);
-  op_cost(0);
-  format %{ %}
-  interface(REG_INTER);
-%}
-
 operand vRegMask()
 %{
   constraint(ALLOC_IN_RC(vmask_reg));

src/hotspot/cpu/riscv/riscv_v.ad

@@ -3843,116 +3843,6 @@ instruct vclearArray_reg_reg(iRegL_R29 cnt, iRegP_R28 base, Universe dummy,
   ins_pipe(pipe_class_memory);
 %}
 
-// CompressBits of Long & Integer
-
-instruct compressBitsI(iRegINoSp dst, iRegIorL2I src, iRegIorL2I mask, vRegMask_V0 v0,
-                       vReg_V4 v4, vReg_V5 v5, vReg_V8 v8, vReg_V9 v9) %{
-  match(Set dst (CompressBits src mask));
-  effect(TEMP v0, TEMP v4, TEMP v5, TEMP v8, TEMP v9);
-  format %{ "vsetivli x0, 1, e32, m1, tu, mu\t#@compressBitsI\n\t"
-            "vmv.s.x $v0, $src\n\t"
-            "mv t0, 32\n\t"
-            "vsetvli x0, t0, e8, m2, tu, mu\n\t"
-            "vmv.v.i $v4, 0\n\t"
-            "vmerge.vim $v4, $v4, 1, $v0\n\t"
-            "vmv.v.i $v8, 0\n\t"
-            "vsetivli x0, 1, e32, m1, tu, mu\n\t"
-            "vmv.s.x $v0, $mask\n\t"
-            "vsetvli x0, t0, e8, m2, tu, mu\n\t"
-            "vcompress.vm $v8, $v4, $v0\n\t"
-            "vmseq.vi $v0, $v8, 1\n\t"
-            "vsetivli x0, 1, e32, m1, tu, mu\n\t"
-            "vmv.x.s $dst, $v0\t#@compressBitsI\n\t"
-          %}
-  ins_encode %{
-    __ compress_bits_i_v(as_Register($dst$$reg), as_Register($src$$reg), as_Register($mask$$reg));
-  %}
-  ins_pipe(pipe_slow);
-%}
-
-instruct compressBitsL(iRegLNoSp dst, iRegL src, iRegL mask, vRegMask_V0 v0,
-                       vReg_V4 v4, vReg_V5 v5, vReg_V6 v6, vReg_V7 v7,
-                       vReg_V8 v8, vReg_V9 v9, vReg_V10 v10, vReg_V11 v11) %{
-  match(Set dst (CompressBits src mask));
-  effect(TEMP v0, TEMP v4, TEMP v5, TEMP v6, TEMP v7, TEMP v8, TEMP v9, TEMP v10, TEMP v11);
-  format %{ "vsetivli x0, 1, e64, m1, tu, mu\t#@compressBitsL\n\t"
-            "vmv.s.x $v0, $src\n\t"
-            "mv t0, 64\n\t"
-            "vsetvli x0, t0, e8, m4, tu, mu\n\t"
-            "vmv.v.i $v4, 0\n\t"
-            "vmerge.vim $v4, $v4, 1, $v0\n\t"
-            "vmv.v.i $v8, 0\n\t"
-            "vsetivli x0, 1, e64, m1, tu, mu\n\t"
-            "vmv.s.x $v0, $mask\n\t"
-            "vsetvli x0, t0, e8, m4, tu, mu\n\t"
-            "vcompress.vm $v8, $v4, $v0\n\t"
-            "vmseq.vi $v0, $v8, 1\n\t"
-            "vsetivli x0, 1, e64, m1, tu, mu\n\t"
-            "vmv.x.s $dst, $v0\t#@compressBitsL\n\t"
-          %}
-  ins_encode %{
-    __ compress_bits_l_v(as_Register($dst$$reg), as_Register($src$$reg), as_Register($mask$$reg));
-  %}
-  ins_pipe(pipe_slow);
-%}
-
-// ExpandBits of Long & Integer
-
-instruct expandBitsI(iRegINoSp dst, iRegIorL2I src, iRegIorL2I mask, vRegMask_V0 v0,
-                     vReg_V4 v4, vReg_V5 v5, vReg_V8 v8, vReg_V9 v9, vReg_V12 v12, vReg_V13 v13) %{
-  match(Set dst (ExpandBits src mask));
-  effect(TEMP v0, TEMP v4, TEMP v5, TEMP v8, TEMP v9, TEMP v12, TEMP v13);
-  format %{ "vsetivli x0, 1, e32, m1, tu, mu\t#@expandBitsI\n\t"
-            "vmv.s.x $v0, $src\n\t"
-            "mv t0, 32\n\t"
-            "vsetvli x0, t0, e8, m2, tu, mu\n\t"
-            "vmv.v.i $v4, 0\n\t"
-            "vmerge.vim $v4, $v4, 1, $v0\n\t"
-            "vmv.v.i $v12, 0\n\t"
-            "vsetivli x0, 1, e32, m1, tu, mu\n\t"
-            "vmv.s.x $v0, $mask\n\t"
-            "vsetvli x0, t0, e8, m2, tu, mu\n\t"
-            "viota.m $v8, $v0\n\t"
-            "vrgather.vv $v12, $v4, $v8, $v0.t\n\t"
-            "vmseq.vi $v0, $v12, 1\n\t"
-            "vsetivli x0, 1, e32, m1, tu, mu\n\t"
-            "vmv.x.s $dst, $v0\t#@expandBitsI\n\t"
-          %}
-  ins_encode %{
-    __ expand_bits_i_v(as_Register($dst$$reg), as_Register($src$$reg), as_Register($mask$$reg));
-  %}
-  ins_pipe(pipe_slow);
-%}
-
-instruct expandBitsL(iRegLNoSp dst, iRegL src, iRegL mask, vRegMask_V0 v0,
-                      vReg_V4 v4, vReg_V5 v5, vReg_V6 v6, vReg_V7 v7,
-                      vReg_V8 v8, vReg_V9 v9, vReg_V10 v10, vReg_V11 v11,
-                      vReg_V12 v12, vReg_V13 v13, vReg_V14 v14, vReg_V15 v15) %{
-  match(Set dst (ExpandBits src mask));
-  effect(TEMP v0, TEMP v4, TEMP v5, TEMP v6, TEMP v7, TEMP v8, TEMP v9, TEMP v10, TEMP v11,
-         TEMP v12, TEMP v13, TEMP v14, TEMP v15);
-  format %{ "vsetivli x0, 1, e64, m1, tu, mu\t#@expandBitsL\n\t"
-            "vmv.s.x $v0, $src\n\t"
-            "mv t0, 64\n\t"
-            "vsetvli x0, t0, e8, m4, tu, mu\n\t"
-            "vmv.v.i $v4, 0\n\t"
-            "vmerge.vim $v4, $v4, 1, $v0\n\t"
-            "vmv.v.i $v12, 0\n\t"
-            "vsetivli x0, 1, e64, m1, tu, mu\n\t"
-            "vmv.s.x $v0, $mask\n\t"
-            "vsetvli x0, t0, e8, m4, tu, mu\n\t"
-            "viota.m $v8, $v0\n\t"
-            "vrgather.vv $v12, $v4, $v8, $v0.t\n\t"
-            "vmseq.vi $v0, $v12, 1\n\t"
-            "vsetivli x0, 1, e64, m1, tu, mu\n\t"
-            "vmv.x.s $dst, $v0\t#@expandBitsL\n\t"
-          %}
-  ins_encode %{
-    __ expand_bits_l_v(as_Register($dst$$reg), as_Register($src$$reg), as_Register($mask$$reg));
-  %}
-  ins_pipe(pipe_slow);
-%}
-
 // Vector Load Const
 instruct vloadcon(vReg dst, immI0 src) %{
   match(Set dst (VectorLoadConst src));

src/hotspot/cpu/riscv/stubGenerator_riscv.cpp

@@ -2112,7 +2112,7 @@ class StubGenerator: public StubCodeGenerator {
 
     // Remaining count is less than 8 bytes. Fill it by a single store.
     // Note that the total length is no less than 8 bytes.
-    if (t == T_BYTE || t == T_SHORT) {
+    if (!AvoidUnalignedAccesses && (t == T_BYTE || t == T_SHORT)) {
       __ beqz(count, L_exit1);
       __ shadd(to, count, to, tmp_reg, shift); // points to the end
       __ sd(value, Address(to, -8)); // overwrite some elements

src/hotspot/cpu/riscv/templateInterpreterGenerator_riscv.cpp

@@ -552,7 +552,7 @@ address TemplateInterpreterGenerator::generate_cont_resume_interpreter_adapter()
   // Restore Java expression stack pointer
   __ ld(t0, Address(fp, frame::interpreter_frame_last_sp_offset * wordSize));
   __ shadd(esp, t0, fp, t0, Interpreter::logStackElementSize);
-  // and NULL it as marker that esp is now tos until next java call
+  // and null it as marker that esp is now tos until next java call
   __ sd(zr, Address(fp, frame::interpreter_frame_last_sp_offset * wordSize));
 
   // Restore machine SP

src/hotspot/cpu/riscv/templateTable_riscv.cpp

@@ -292,15 +292,10 @@ void TemplateTable::bipush() {
 
 void TemplateTable::sipush() {
   transition(vtos, itos);
-  if (AvoidUnalignedAccesses) {
   __ load_signed_byte(x10, at_bcp(1));
   __ load_unsigned_byte(t1, at_bcp(2));
   __ slli(x10, x10, 8);
   __ add(x10, x10, t1);
-  } else {
-    __ load_unsigned_short(x10, at_bcp(1));
-    __ revb_h_h(x10, x10); // reverse bytes in half-word and sign-extend
-  }
 }
 
 void TemplateTable::ldc(LdcType type) {
@@ -1626,15 +1621,10 @@ void TemplateTable::branch(bool is_jsr, bool is_wide) {
 
   // load branch displacement
   if (!is_wide) {
-    if (AvoidUnalignedAccesses) {
     __ lb(x12, at_bcp(1));
     __ lbu(t1, at_bcp(2));
     __ slli(x12, x12, 8);
     __ add(x12, x12, t1);
-    } else {
-      __ lhu(x12, at_bcp(1));
-      __ revb_h_h(x12, x12); // reverse bytes in half-word and sign-extend
-    }
   } else {
     __ lwu(x12, at_bcp(1));
     __ revb_w_w(x12, x12); // reverse bytes in word and sign-extend

src/hotspot/cpu/s390/c2_globals_s390.hpp

@@ -61,6 +61,7 @@ define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoScheduling,               false);
 define_pd_global(bool, OptoRegScheduling,            false);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  true);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 16);
 // On s390x, we can clear the array with a single instruction,
 // so don't idealize it.
 define_pd_global(bool, IdealizeClearArrayNode,       false);

src/hotspot/cpu/x86/c2_globals_x86.hpp

@@ -76,6 +76,7 @@ define_pd_global(bool, OptoScheduling,               false);
 define_pd_global(bool, OptoBundling,                 false);
 define_pd_global(bool, OptoRegScheduling,            true);
 define_pd_global(bool, SuperWordLoopUnrollAnalysis,  true);
+define_pd_global(uint, SuperWordStoreToLoadForwardingFailureDetection, 16);
 define_pd_global(bool, IdealizeClearArrayNode,       true);
 
 define_pd_global(uintx, ReservedCodeCacheSize,       48*M);

src/hotspot/cpu/x86/macroAssembler_x86.cpp

@@ -4912,6 +4912,10 @@ void MacroAssembler::population_count(Register dst, Register src,
     }
     bind(done);
   }
+#ifdef ASSERT
+  mov64(scratch1, 0xCafeBabeDeadBeef);
+  movq(scratch2, scratch1);
+#endif
 }
 
 // Ensure that the inline code and the stub are using the same registers.
@@ -5113,6 +5117,7 @@ void MacroAssembler::lookup_secondary_supers_table_var(Register r_sub_klass,
   const Register r_array_base = *available_regs++;
 
   // Get the first array index that can contain super_klass into r_array_index.
+  // Note: Clobbers r_array_base and slot.
   population_count(r_array_index, r_array_index, /*temp2*/r_array_base, /*temp3*/slot);
 
   // NB! r_array_index is off by 1. It is compensated by keeping r_array_base off by 1 word.
@@ -5130,7 +5135,7 @@ void MacroAssembler::lookup_secondary_supers_table_var(Register r_sub_klass,
   jccb(Assembler::equal, L_success);
 
   // Restore slot to its true value
-  xorl(slot, (u1)(Klass::SECONDARY_SUPERS_TABLE_SIZE - 1)); // slot ^ 63 === 63 - slot (mod 64)
+  movb(slot, Address(r_super_klass, Klass::hash_slot_offset()));
 
   // Linear probe. Rotate the bitmap so that the next bit to test is
   // in Bit 1.

src/hotspot/cpu/x86/templateInterpreterGenerator_x86.cpp

@@ -400,7 +400,7 @@ address TemplateInterpreterGenerator::generate_cont_resume_interpreter_adapter()
   // Restore stack bottom
   __ movptr(rcx, Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize));
   __ lea(rsp, Address(rbp, rcx, Address::times_ptr));
-  // and NULL it as marker that esp is now tos until next java call
+  // and null it as marker that esp is now tos until next java call
   __ movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), NULL_WORD);
 
   __ jmp(rax);

src/hotspot/cpu/x86/x86.ad

@@ -6179,6 +6179,7 @@ instruct evmulL_reg(vec dst, vec src1, vec src2) %{
              VM_Version::supports_avx512dq()) ||
             VM_Version::supports_avx512vldq());
   match(Set dst (MulVL src1 src2));
+  ins_cost(500);
   format %{ "evpmullq $dst,$src1,$src2\t! mul packedL" %}
   ins_encode %{
     assert(UseAVX > 2, "required");
@@ -6195,6 +6196,7 @@ instruct evmulL_mem(vec dst, vec src, memory mem) %{
              VM_Version::supports_avx512vldq()));
   match(Set dst (MulVL src (LoadVector mem)));
   format %{ "evpmullq $dst,$src,$mem\t! mul packedL" %}
+  ins_cost(500);
   ins_encode %{
     assert(UseAVX > 2, "required");
     int vlen_enc = vector_length_encoding(this);
@@ -6206,6 +6208,7 @@ instruct evmulL_mem(vec dst, vec src, memory mem) %{
 instruct vmulL(vec dst, vec src1, vec src2, vec xtmp) %{
   predicate(UseAVX == 0);
   match(Set dst (MulVL src1 src2));
+  ins_cost(500);
   effect(TEMP dst, TEMP xtmp);
   format %{ "mulVL   $dst, $src1, $src2\t! using $xtmp as TEMP" %}
   ins_encode %{
@@ -6232,6 +6235,7 @@ instruct vmulL_reg(vec dst, vec src1, vec src2, vec xtmp1, vec xtmp2) %{
               !VM_Version::supports_avx512vldq())));
   match(Set dst (MulVL src1 src2));
   effect(TEMP xtmp1, TEMP xtmp2);
+  ins_cost(500);
   format %{ "vmulVL  $dst, $src1, $src2\t! using $xtmp1, $xtmp2 as TEMP" %}
   ins_encode %{
     int vlen_enc = vector_length_encoding(this);
@@ -6248,6 +6252,30 @@ instruct vmulL_reg(vec dst, vec src1, vec src2, vec xtmp1, vec xtmp2) %{
   ins_pipe( pipe_slow );
 %}
 
+instruct vmuludq_reg(vec dst, vec src1, vec src2) %{
+  predicate(UseAVX > 0 && n->as_MulVL()->has_uint_inputs());
+  match(Set dst (MulVL src1 src2));
+  ins_cost(100);
+  format %{ "vpmuludq $dst,$src1,$src2\t! muludq packedL" %}
+  ins_encode %{
+    int vlen_enc = vector_length_encoding(this);
+    __ vpmuludq($dst$$XMMRegister, $src1$$XMMRegister, $src2$$XMMRegister, vlen_enc);
+  %}
+  ins_pipe( pipe_slow );
+%}
+
+instruct vmuldq_reg(vec dst, vec src1, vec src2) %{
+  predicate(UseAVX > 0 && n->as_MulVL()->has_int_inputs());
+  match(Set dst (MulVL src1 src2));
+  ins_cost(100);
+  format %{ "vpmuldq $dst,$src1,$src2\t! muldq packedL" %}
+  ins_encode %{
+    int vlen_enc = vector_length_encoding(this);
+    __ vpmuldq($dst$$XMMRegister, $src1$$XMMRegister, $src2$$XMMRegister, vlen_enc);
+  %}
+  ins_pipe( pipe_slow );
+%}
+
 // Floats vector mul
 instruct vmulF(vec dst, vec src) %{
   predicate(UseAVX == 0);

src/hotspot/os/bsd/os_bsd.cpp

@@ -229,8 +229,6 @@ size_t os::rss() {
 // Cpu architecture string
 #if   defined(ZERO)
 static char cpu_arch[] = ZERO_LIBARCH;
-#elif defined(IA64)
-static char cpu_arch[] = "ia64";
 #elif defined(IA32)
 static char cpu_arch[] = "i386";
 #elif defined(AMD64)
@@ -1192,8 +1190,6 @@ void * os::dll_load(const char *filename, char *ebuf, int ebuflen) {
   static  Elf32_Half running_arch_code=EM_386;
   #elif   (defined AMD64)
   static  Elf32_Half running_arch_code=EM_X86_64;
-  #elif  (defined IA64)
-  static  Elf32_Half running_arch_code=EM_IA_64;
   #elif  (defined __powerpc64__)
   static  Elf32_Half running_arch_code=EM_PPC64;
   #elif  (defined __powerpc__)
@@ -1214,7 +1210,7 @@ void * os::dll_load(const char *filename, char *ebuf, int ebuflen) {
   static  Elf32_Half running_arch_code=EM_68K;
   #else
     #error Method os::dll_load requires that one of following is defined:\
-         IA32, AMD64, IA64, __powerpc__, ARM, S390, ALPHA, MIPS, MIPSEL, PARISC, M68K
+         IA32, AMD64, __powerpc__, ARM, S390, ALPHA, MIPS, MIPSEL, PARISC, M68K
   #endif
 
   // Identify compatibility class for VM's architecture and library's architecture

src/hotspot/os/linux/hugepages.cpp

@@ -55,8 +55,7 @@ static size_t scan_default_hugepagesize() {
 
   // large_page_size on Linux is used to round up heap size. x86 uses either
   // 2M or 4M page, depending on whether PAE (Physical Address Extensions)
-  // mode is enabled. AMD64/EM64T uses 2M page in 64bit mode. IA64 can use
+  // mode is enabled. AMD64/EM64T uses 2M page in 64bit mode.
-  // page as large as 1G.
   //
   // Here we try to figure out page size by parsing /proc/meminfo and looking
   // for a line with the following format:

src/hotspot/os/linux/mallocInfoDcmd.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2023, 2024, Oracle and/or its affiliates. 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
@@ -41,10 +41,6 @@ public:
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = { "java.lang.management.ManagementPermission", "monitor", nullptr };
-    return p;
-  }
   void execute(DCmdSource source, TRAPS) override;
 };
 

src/hotspot/os/linux/os_linux.cpp

@@ -461,26 +461,17 @@ bool os::Linux::get_tick_information(CPUPerfTicks* pticks, int which_logical_cpu
 }
 
 #ifndef SYS_gettid
-// i386: 224, ia64: 1105, amd64: 186, sparc: 143
+// i386: 224, amd64: 186, sparc: 143
-  #ifdef __ia64__
+  #if defined(__i386__)
-    #define SYS_gettid 1105
-  #else
-    #ifdef __i386__
     #define SYS_gettid 224
-    #else
+  #elif defined(__amd64__)
-      #ifdef __amd64__
     #define SYS_gettid 186
-      #else
+  #elif defined(__sparc__)
-        #ifdef __sparc__
     #define SYS_gettid 143
   #else
-          #error define gettid for the arch
+    #error "Define SYS_gettid for this architecture"
   #endif
-      #endif
+#endif // SYS_gettid
-    #endif
-  #endif
-#endif
-
 
 // pid_t gettid()
 //
@@ -1778,8 +1769,6 @@ void * os::dll_load(const char *filename, char *ebuf, int ebuflen) {
   static  Elf32_Half running_arch_code=EM_386;
 #elif   (defined AMD64) || (defined X32)
   static  Elf32_Half running_arch_code=EM_X86_64;
-#elif  (defined IA64)
-  static  Elf32_Half running_arch_code=EM_IA_64;
 #elif  (defined __sparc) && (defined _LP64)
   static  Elf32_Half running_arch_code=EM_SPARCV9;
 #elif  (defined __sparc) && (!defined _LP64)
@@ -1812,7 +1801,7 @@ void * os::dll_load(const char *filename, char *ebuf, int ebuflen) {
   static  Elf32_Half running_arch_code=EM_LOONGARCH;
 #else
     #error Method os::dll_load requires that one of following is defined:\
-        AARCH64, ALPHA, ARM, AMD64, IA32, IA64, LOONGARCH64, M68K, MIPS, MIPSEL, PARISC, __powerpc__, __powerpc64__, RISCV, S390, SH, __sparc
+        AARCH64, ALPHA, ARM, AMD64, IA32, LOONGARCH64, M68K, MIPS, MIPSEL, PARISC, __powerpc__, __powerpc64__, RISCV, S390, SH, __sparc
 #endif
 
   // Identify compatibility class for VM's architecture and library's architecture
@@ -2719,8 +2708,6 @@ void os::get_summary_cpu_info(char* cpuinfo, size_t length) {
   strncpy(cpuinfo, "ARM", length);
 #elif defined(IA32)
   strncpy(cpuinfo, "x86_32", length);
-#elif defined(IA64)
-  strncpy(cpuinfo, "IA64", length);
 #elif defined(PPC)
   strncpy(cpuinfo, "PPC64", length);
 #elif defined(RISCV)

src/hotspot/os/linux/trimCHeapDCmd.hpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2021 SAP SE. All rights reserved.
- * Copyright (c) 2021, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2021, 2024, Oracle and/or its affiliates. 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
@@ -42,10 +42,6 @@ public:
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = { "java.lang.management.ManagementPermission", "control", nullptr };
-    return p;
-  }
   virtual void execute(DCmdSource source, TRAPS);
 };
 

src/hotspot/os/posix/os_posix.cpp

@@ -2111,7 +2111,7 @@ void os::shutdown() {
 // easily trigger secondary faults in those threads. To reduce the likelihood
 // of that we use _exit rather than exit, so that no atexit hooks get run.
 // But note that os::shutdown() could also trigger secondary faults.
-void os::abort(bool dump_core, void* siginfo, const void* context) {
+void os::abort(bool dump_core, const void* siginfo, const void* context) {
   os::shutdown();
   if (dump_core) {
     LINUX_ONLY(if (DumpPrivateMappingsInCore) ClassLoader::close_jrt_image();)
@@ -2186,3 +2186,43 @@ char* os::pd_map_memory(int fd, const char* unused,
 bool os::pd_unmap_memory(char* addr, size_t bytes) {
   return munmap(addr, bytes) == 0;
 }
+
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+static ucontext_t _saved_assert_context;
+static bool _has_saved_context = false;
+#endif // CAN_SHOW_REGISTERS_ON_ASSERT
+
+void os::save_assert_context(const void* ucVoid) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  assert(ucVoid != nullptr, "invariant");
+  assert(!_has_saved_context, "invariant");
+  memcpy(&_saved_assert_context, ucVoid, sizeof(ucontext_t));
+  // on Linux ppc64, ucontext_t contains pointers into itself which have to be patched up
+  //  after copying the context (see comment in sys/ucontext.h):
+#if defined(PPC64)
+  *((void**)&_saved_assert_context.uc_mcontext.regs) = &(_saved_assert_context.uc_mcontext.gp_regs);
+#elif defined(AMD64)
+  // In the copied version, fpregs should point to the copied contents.
+  // Sanity check: fpregs should point into the context.
+  if ((address)((const ucontext_t*)ucVoid)->uc_mcontext.fpregs > (address)ucVoid) {
+    size_t fpregs_offset = pointer_delta(((const ucontext_t*)ucVoid)->uc_mcontext.fpregs, ucVoid, 1);
+    if (fpregs_offset < sizeof(ucontext_t)) {
+      // Preserve the offset.
+      *((void**)&_saved_assert_context.uc_mcontext.fpregs) = (void*)((address)(void*)&_saved_assert_context + fpregs_offset);
+    }
+  }
+#endif
+  _has_saved_context = true;
+#endif // CAN_SHOW_REGISTERS_ON_ASSERT
+}
+
+const void* os::get_saved_assert_context(const void** sigInfo) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  assert(sigInfo != nullptr, "invariant");
+  *sigInfo = nullptr;
+  return _has_saved_context ? &_saved_assert_context : nullptr;
+#endif
+  *sigInfo = nullptr;
+  return nullptr;
+}
+

src/hotspot/os/posix/signals_posix.cpp

@@ -578,9 +578,8 @@ int JVM_HANDLE_XXX_SIGNAL(int sig, siginfo_t* info,
 
   // Handle assertion poison page accesses.
 #ifdef CAN_SHOW_REGISTERS_ON_ASSERT
-  if (!signal_was_handled &&
+  if (VMError::was_assert_poison_crash(info)) {
-      ((sig == SIGSEGV || sig == SIGBUS) && info != nullptr && info->si_addr == g_assert_poison)) {
+    signal_was_handled = handle_assert_poison_fault(ucVoid);
-    signal_was_handled = handle_assert_poison_fault(ucVoid, info->si_addr);
   }
 #endif
 
@@ -961,10 +960,6 @@ static bool get_signal_code_description(const siginfo_t* si, enum_sigcode_desc_t
     { SIGILL,  ILL_PRVREG,   "ILL_PRVREG",   "Privileged register." },
     { SIGILL,  ILL_COPROC,   "ILL_COPROC",   "Coprocessor error." },
     { SIGILL,  ILL_BADSTK,   "ILL_BADSTK",   "Internal stack error." },
-#if defined(IA64) && defined(LINUX)
-    { SIGILL,  ILL_BADIADDR, "ILL_BADIADDR", "Unimplemented instruction address" },
-    { SIGILL,  ILL_BREAK,    "ILL_BREAK",    "Application Break instruction" },
-#endif
     { SIGFPE,  FPE_INTDIV,   "FPE_INTDIV",   "Integer divide by zero." },
     { SIGFPE,  FPE_INTOVF,   "FPE_INTOVF",   "Integer overflow." },
     { SIGFPE,  FPE_FLTDIV,   "FPE_FLTDIV",   "Floating-point divide by zero." },
@@ -978,9 +973,6 @@ static bool get_signal_code_description(const siginfo_t* si, enum_sigcode_desc_t
 #if defined(AIX)
     // no explanation found what keyerr would be
     { SIGSEGV, SEGV_KEYERR,  "SEGV_KEYERR",  "key error" },
-#endif
-#if defined(IA64) && !defined(AIX)
-    { SIGSEGV, SEGV_PSTKOVF, "SEGV_PSTKOVF", "Paragraph stack overflow" },
 #endif
     { SIGBUS,  BUS_ADRALN,   "BUS_ADRALN",   "Invalid address alignment." },
     { SIGBUS,  BUS_ADRERR,   "BUS_ADRERR",   "Nonexistent physical address." },
@@ -1136,8 +1128,16 @@ static const char* get_signal_name(int sig, char* out, size_t outlen) {
 }
 
 void os::print_siginfo(outputStream* os, const void* si0) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  // If we are here because of an assert/guarantee, we suppress
+  // printing the siginfo, because it is only an implementation
+  // detail capturing the context for said assert/guarantee.
+  if (VMError::was_assert_poison_crash(si0)) {
+    return;
+  }
+#endif
 
-  const siginfo_t* const si = (const siginfo_t*) si0;
+  const siginfo_t* const si = (const siginfo_t*)si0;
 
   char buf[20];
   os->print("siginfo:");

src/hotspot/os/posix/vmError_posix.cpp

@@ -84,8 +84,8 @@ static void crash_handler(int sig, siginfo_t* info, void* context) {
 
   // Needed because asserts may happen in error handling too.
 #ifdef CAN_SHOW_REGISTERS_ON_ASSERT
-  if ((sig == SIGSEGV || sig == SIGBUS) && info != nullptr && info->si_addr == g_assert_poison) {
+  if (VMError::was_assert_poison_crash(info)) {
-    if (handle_assert_poison_fault(context, info->si_addr)) {
+    if (handle_assert_poison_fault(context)) {
       return;
     }
   }
@@ -127,3 +127,14 @@ void VMError::check_failing_cds_access(outputStream* st, const void* siginfo) {
   }
 #endif
 }
+
+bool VMError::was_assert_poison_crash(const void* siginfo) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  if (siginfo == nullptr) {
+    return false;
+  }
+  const siginfo_t* const si = (siginfo_t*)siginfo;
+  return (si->si_signo == SIGSEGV || si->si_signo == SIGBUS) && si->si_addr == g_assert_poison_read_only;
+#endif
+  return false;
+}

src/hotspot/os/windows/os_windows.cpp

@@ -72,6 +72,7 @@
 #include "services/runtimeService.hpp"
 #include "symbolengine.hpp"
 #include "utilities/align.hpp"
+#include "utilities/debug.hpp"
 #include "utilities/decoder.hpp"
 #include "utilities/defaultStream.hpp"
 #include "utilities/events.hpp"
@@ -1317,7 +1318,7 @@ void os::check_core_dump_prerequisites(char* buffer, size_t bufferSize, bool che
   }
 }
 
-void os::abort(bool dump_core, void* siginfo, const void* context) {
+void os::abort(bool dump_core, const void* siginfo, const void* context) {
   EXCEPTION_POINTERS ep;
   MINIDUMP_EXCEPTION_INFORMATION mei;
   MINIDUMP_EXCEPTION_INFORMATION* pmei;
@@ -2112,7 +2113,17 @@ bool os::signal_sent_by_kill(const void* siginfo) {
 }
 
 void os::print_siginfo(outputStream *st, const void* siginfo) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  // If we are here because of an assert/guarantee, we suppress
+  // printing the siginfo, because it is only an implementation
+  // detail capturing the context for said assert/guarantee.
+  if (VMError::was_assert_poison_crash(siginfo)) {
+    return;
+  }
+#endif
+
   const EXCEPTION_RECORD* const er = (EXCEPTION_RECORD*)siginfo;
+
   st->print("siginfo:");
 
   char tmp[64];
@@ -2625,6 +2636,14 @@ LONG WINAPI topLevelExceptionFilter(struct _EXCEPTION_POINTERS* exceptionInfo) {
 #endif
 #endif
 
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  if (VMError::was_assert_poison_crash(exception_record)) {
+    if (handle_assert_poison_fault(exceptionInfo)) {
+      return EXCEPTION_CONTINUE_EXECUTION;
+    }
+  }
+#endif
+
   if (t != nullptr && t->is_Java_thread()) {
     JavaThread* thread = JavaThread::cast(t);
     bool in_java = thread->thread_state() == _thread_in_Java;
@@ -6165,3 +6184,26 @@ void os::print_user_info(outputStream* st) {
 void os::print_active_locale(outputStream* st) {
   // not implemented yet
 }
+
+static CONTEXT _saved_assert_context;
+static EXCEPTION_RECORD _saved_exception_record;
+static bool _has_saved_context = false;
+
+void os::save_assert_context(const void* ucVoid) {
+  assert(ucVoid != nullptr, "invariant");
+  assert(!_has_saved_context, "invariant");
+  const EXCEPTION_POINTERS* ep = static_cast<const EXCEPTION_POINTERS*>(ucVoid);
+  memcpy(&_saved_assert_context, ep->ContextRecord, sizeof(CONTEXT));
+  memcpy(&_saved_exception_record, ep->ExceptionRecord, sizeof(EXCEPTION_RECORD));
+  _has_saved_context = true;
+}
+
+const void* os::get_saved_assert_context(const void** sigInfo) {
+  assert(sigInfo != nullptr, "invariant");
+  if (_has_saved_context) {
+    *sigInfo = &_saved_exception_record;
+    return &_saved_assert_context;
+  }
+  *sigInfo = nullptr;
+  return nullptr;
+}

src/hotspot/os/windows/vmError_windows.cpp

@@ -28,6 +28,7 @@
 #include "runtime/arguments.hpp"
 #include "runtime/javaThread.hpp"
 #include "runtime/os.hpp"
+#include "utilities/debug.hpp"
 #include "utilities/vmError.hpp"
 
 LONG WINAPI crash_handler(struct _EXCEPTION_POINTERS* exceptionInfo) {
@@ -67,10 +68,23 @@ void VMError::check_failing_cds_access(outputStream* st, const void* siginfo) {
 void VMError::reporting_started() {}
 void VMError::interrupt_reporting_thread() {}
 
-void VMError::raise_fail_fast(void* exrecord, void* context) {
+void VMError::raise_fail_fast(const void* exrecord, const void* context) {
   DWORD flags = (exrecord == nullptr) ? FAIL_FAST_GENERATE_EXCEPTION_ADDRESS : 0;
-  RaiseFailFastException(static_cast<PEXCEPTION_RECORD>(exrecord),
+  PEXCEPTION_RECORD exception_record = static_cast<PEXCEPTION_RECORD>(const_cast<void*>(exrecord));
-                         static_cast<PCONTEXT>(context),
+  PCONTEXT ctx = static_cast<PCONTEXT>(const_cast<void*>(context));
-                         flags);
+  RaiseFailFastException(exception_record, ctx, flags);
   ::abort();
 }
+
+bool VMError::was_assert_poison_crash(const void* siginfo) {
+#ifdef CAN_SHOW_REGISTERS_ON_ASSERT
+  if (siginfo == nullptr) {
+    return false;
+  }
+  const EXCEPTION_RECORD* const er = (EXCEPTION_RECORD*)siginfo;
+  if (er->ExceptionCode == EXCEPTION_ACCESS_VIOLATION && er->NumberParameters >= 2) {
+    return (void*)er->ExceptionInformation[1] == g_assert_poison_read_only;
+  }
+#endif
+  return false;
+}

src/hotspot/share/cds/archiveBuilder.cpp

@@ -1294,10 +1294,15 @@ class ArchiveBuilder::CDSMapLogger : AllStatic {
 
       if (source_oop != nullptr) {
         // This is a regular oop that got archived.
-        print_oop_with_requested_addr_cr(&st, source_oop, false);
+        // Don't print the requested addr again as we have just printed it at the beginning of the line.
+        // Example:
+        // 0x00000007ffd27938: @@ Object (0xfffa4f27) java.util.HashMap
+        print_oop_info_cr(&st, source_oop, /*print_requested_addr=*/false);
         byte_size = source_oop->size() * BytesPerWord;
       } else if ((byte_size = ArchiveHeapWriter::get_filler_size_at(start)) > 0) {
         // We have a filler oop, which also does not exist in BufferOffsetToSourceObjectTable.
+        // Example:
+        // 0x00000007ffc3ffd8: @@ Object filler 40 bytes
         st.print_cr("filler " SIZE_FORMAT " bytes", byte_size);
       } else {
         ShouldNotReachHere();
@@ -1315,7 +1320,7 @@ class ArchiveBuilder::CDSMapLogger : AllStatic {
 
   // ArchivedFieldPrinter is used to print the fields of archived objects. We can't
   // use _source_obj->print_on(), because we want to print the oop fields
-  // in _source_obj with their requested addresses using print_oop_with_requested_addr_cr().
+  // in _source_obj with their requested addresses using print_oop_info_cr().
   class ArchivedFieldPrinter : public FieldClosure {
     ArchiveHeapInfo* _heap_info;
     outputStream* _st;
@@ -1331,8 +1336,14 @@ class ArchiveBuilder::CDSMapLogger : AllStatic {
       switch (ft) {
       case T_ARRAY:
       case T_OBJECT:
+        {
           fd->print_on(_st); // print just the name and offset
-        print_oop_with_requested_addr_cr(_st, _source_obj->obj_field(fd->offset()));
+          oop obj = _source_obj->obj_field(fd->offset());
+          if (java_lang_Class::is_instance(obj)) {
+            obj = HeapShared::scratch_java_mirror(obj);
+          }
+          print_oop_info_cr(_st, obj);
+        }
         break;
       default:
         if (ArchiveHeapWriter::is_marked_as_native_pointer(_heap_info, _source_obj, fd->offset())) {
@@ -1388,37 +1399,78 @@ class ArchiveBuilder::CDSMapLogger : AllStatic {
         objArrayOop source_obj_array = objArrayOop(source_oop);
         for (int i = 0; i < source_obj_array->length(); i++) {
           st.print(" -%4d: ", i);
-          print_oop_with_requested_addr_cr(&st, source_obj_array->obj_at(i));
+          oop obj = source_obj_array->obj_at(i);
+          if (java_lang_Class::is_instance(obj)) {
+            obj = HeapShared::scratch_java_mirror(obj);
+          }
+          print_oop_info_cr(&st, obj);
         }
       } else {
         st.print_cr(" - fields (" SIZE_FORMAT " words):", source_oop->size());
         ArchivedFieldPrinter print_field(heap_info, &st, source_oop, buffered_addr);
         InstanceKlass::cast(source_klass)->print_nonstatic_fields(&print_field);
+
+        if (java_lang_Class::is_instance(source_oop)) {
+          oop scratch_mirror = source_oop;
+          st.print(" - signature: ");
+          print_class_signature_for_mirror(&st, scratch_mirror);
+          st.cr();
+
+          Klass* src_klass = java_lang_Class::as_Klass(scratch_mirror);
+          if (src_klass != nullptr && src_klass->is_instance_klass()) {
+            oop rr = HeapShared::scratch_resolved_references(InstanceKlass::cast(src_klass)->constants());
+            st.print(" - archived_resolved_references: ");
+            print_oop_info_cr(&st, rr);
+
+            // We need to print the fields in the scratch_mirror, not the original mirror.
+            // (if a class is not aot-initialized, static fields in its scratch mirror will be cleared).
+            assert(scratch_mirror == HeapShared::scratch_java_mirror(src_klass->java_mirror()), "sanity");
+            st.print_cr("- ---- static fields (%d):", java_lang_Class::static_oop_field_count(scratch_mirror));
+            InstanceKlass::cast(src_klass)->do_local_static_fields(&print_field);
           }
         }
       }
+    }
+  }
+
+  static void print_class_signature_for_mirror(outputStream* st, oop scratch_mirror) {
+    assert(java_lang_Class::is_instance(scratch_mirror), "sanity");
+    if (java_lang_Class::is_primitive(scratch_mirror)) {
+      for (int i = T_BOOLEAN; i < T_VOID+1; i++) {
+        BasicType bt = (BasicType)i;
+        if (!is_reference_type(bt) && scratch_mirror == HeapShared::scratch_java_mirror(bt)) {
+          oop orig_mirror = Universe::java_mirror(bt);
+          java_lang_Class::print_signature(orig_mirror, st);
+          return;
+        }
+      }
+      ShouldNotReachHere();
+    }
+    java_lang_Class::print_signature(scratch_mirror, st);
+  }
 
   static void log_heap_roots() {
     LogStreamHandle(Trace, cds, map, oops) st;
     if (st.is_enabled()) {
       for (int i = 0; i < HeapShared::pending_roots()->length(); i++) {
         st.print("roots[%4d]: ", i);
-        print_oop_with_requested_addr_cr(&st, HeapShared::pending_roots()->at(i));
+        print_oop_info_cr(&st, HeapShared::pending_roots()->at(i));
       }
     }
   }
 
-  // The output looks like this. The first number is the requested address. The second number is
+  // Example output:
-  // the narrowOop version of the requested address.
+  // - The first number is the requested address (if print_requested_addr == true)
-  //     0x00000007ffc7e840 (0xfff8fd08) java.lang.Class
+  // - The second number is the narrowOop version of the requested address (if UseCompressedOops == true)
+  //     0x00000007ffc7e840 (0xfff8fd08) java.lang.Class Ljava/util/Array;
   //     0x00000007ffc000f8 (0xfff8001f) [B length: 11
-  static void print_oop_with_requested_addr_cr(outputStream* st, oop source_oop, bool print_addr = true) {
+  static void print_oop_info_cr(outputStream* st, oop source_oop, bool print_requested_addr = true) {
     if (source_oop == nullptr) {
       st->print_cr("null");
     } else {
       ResourceMark rm;
       oop requested_obj = ArchiveHeapWriter::source_obj_to_requested_obj(source_oop);
-      if (print_addr) {
+      if (print_requested_addr) {
         st->print(PTR_FORMAT " ", p2i(requested_obj));
       }
       if (UseCompressedOops) {
@@ -1428,7 +1480,27 @@ class ArchiveBuilder::CDSMapLogger : AllStatic {
         int array_len = arrayOop(source_oop)->length();
         st->print_cr("%s length: %d", source_oop->klass()->external_name(), array_len);
       } else {
-        st->print_cr("%s", source_oop->klass()->external_name());
+        st->print("%s", source_oop->klass()->external_name());
+
+        if (java_lang_String::is_instance(source_oop)) {
+          st->print(" ");
+          java_lang_String::print(source_oop, st);
+        } else if (java_lang_Class::is_instance(source_oop)) {
+          oop scratch_mirror = source_oop;
+
+          st->print(" ");
+          print_class_signature_for_mirror(st, scratch_mirror);
+
+          Klass* src_klass = java_lang_Class::as_Klass(scratch_mirror);
+          if (src_klass != nullptr && src_klass->is_instance_klass()) {
+            InstanceKlass* buffered_klass =
+              ArchiveBuilder::current()->get_buffered_addr(InstanceKlass::cast(src_klass));
+            if (buffered_klass->has_aot_initialized_mirror()) {
+              st->print(" (aot-inited)");
+            }
+          }
+        }
+        st->cr();
       }
     }
   }

src/hotspot/share/cds/archiveBuilder.hpp

@@ -343,10 +343,20 @@ public:
 
   template <typename T>
   u4 any_to_offset_u4(T p) const {
+    assert(p != nullptr, "must not be null");
     uintx offset = any_to_offset((address)p);
     return to_offset_u4(offset);
   }
 
+  template <typename T>
+  u4 any_or_null_to_offset_u4(T p) const {
+    if (p == nullptr) {
+      return 0;
+    } else {
+      return any_to_offset_u4<T>(p);
+    }
+  }
+
   template <typename T>
   T offset_to_buffered(u4 offset) const {
     return (T)offset_to_buffered_address(offset);

src/hotspot/share/cds/archiveUtils.cpp

@@ -399,188 +399,3 @@ size_t HeapRootSegments::segment_offset(size_t seg_idx) {
   return _base_offset + seg_idx * _max_size_in_bytes;
 }
 
-ArchiveWorkers ArchiveWorkers::_workers;
-
-ArchiveWorkers::ArchiveWorkers() :
-        _start_semaphore(0),
-        _end_semaphore(0),
-        _num_workers(0),
-        _started_workers(0),
-        _waiting_workers(0),
-        _running_workers(0),
-        _state(NOT_READY),
-        _task(nullptr) {
-}
-
-void ArchiveWorkers::initialize() {
-  assert(Atomic::load(&_state) == NOT_READY, "Should be");
-
-  Atomic::store(&_num_workers, max_workers());
-  Atomic::store(&_state, READY);
-
-  // Kick off pool startup by creating a single worker.
-  start_worker_if_needed();
-}
-
-int ArchiveWorkers::max_workers() {
-  // The pool is used for short-lived bursty tasks. We do not want to spend
-  // too much time creating and waking up threads unnecessarily. Plus, we do
-  // not want to overwhelm large machines. This is why we want to be very
-  // conservative about the number of workers actually needed.
-  return MAX2(0, log2i_graceful(os::active_processor_count()));
-}
-
-bool ArchiveWorkers::is_parallel() {
-  return _num_workers > 0;
-}
-
-void ArchiveWorkers::shutdown() {
-  while (true) {
-    State state = Atomic::load(&_state);
-    if (state == SHUTDOWN) {
-      // Already shut down.
-      return;
-    }
-    if (Atomic::cmpxchg(&_state, state, SHUTDOWN, memory_order_relaxed) == state) {
-      if (is_parallel()) {
-        // Execute a shutdown task and block until all workers respond.
-        run_task(&_shutdown_task);
-      }
-    }
-  }
-}
-
-void ArchiveWorkers::start_worker_if_needed() {
-  while (true) {
-    int cur = Atomic::load(&_started_workers);
-    if (cur >= _num_workers) {
-      return;
-    }
-    if (Atomic::cmpxchg(&_started_workers, cur, cur + 1, memory_order_relaxed) == cur) {
-      new ArchiveWorkerThread(this);
-      return;
-    }
-  }
-}
-
-void ArchiveWorkers::signal_worker_if_needed() {
-  while (true) {
-    int cur = Atomic::load(&_waiting_workers);
-    if (cur == 0) {
-      return;
-    }
-    if (Atomic::cmpxchg(&_waiting_workers, cur, cur - 1, memory_order_relaxed) == cur) {
-      _start_semaphore.signal(1);
-      return;
-    }
-  }
-}
-
-void ArchiveWorkers::run_task(ArchiveWorkerTask* task) {
-  assert((Atomic::load(&_state) == READY) ||
-         ((Atomic::load(&_state) == SHUTDOWN) && (task == &_shutdown_task)),
-         "Should be in correct state");
-  assert(Atomic::load(&_task) == nullptr, "Should not have running tasks");
-
-  if (is_parallel()) {
-    run_task_multi(task);
-  } else {
-    run_task_single(task);
-  }
-}
-
-void ArchiveWorkers::run_task_single(ArchiveWorkerTask* task) {
-  // Single thread needs no chunking.
-  task->configure_max_chunks(1);
-
-  // Execute the task ourselves, as there are no workers.
-  task->work(0, 1);
-}
-
-void ArchiveWorkers::run_task_multi(ArchiveWorkerTask* task) {
-  // Multiple threads can work with multiple chunks.
-  task->configure_max_chunks(_num_workers * CHUNKS_PER_WORKER);
-
-  // Set up the run and publish the task.
-  Atomic::store(&_waiting_workers, _num_workers);
-  Atomic::store(&_running_workers, _num_workers);
-  Atomic::release_store(&_task, task);
-
-  // Kick off pool wakeup by signaling a single worker, and proceed
-  // immediately to executing the task locally.
-  signal_worker_if_needed();
-
-  // Execute the task ourselves, while workers are catching up.
-  // This allows us to hide parts of task handoff latency.
-  task->run();
-
-  // Done executing task locally, wait for any remaining workers to complete,
-  // and then do the final housekeeping.
-  _end_semaphore.wait();
-  Atomic::store(&_task, (ArchiveWorkerTask *) nullptr);
-  OrderAccess::fence();
-
-  assert(Atomic::load(&_waiting_workers) == 0, "All workers were signaled");
-  assert(Atomic::load(&_running_workers) == 0, "No workers are running");
-}
-
-void ArchiveWorkerTask::run() {
-  while (true) {
-    int chunk = Atomic::load(&_chunk);
-    if (chunk >= _max_chunks) {
-      return;
-    }
-    if (Atomic::cmpxchg(&_chunk, chunk, chunk + 1, memory_order_relaxed) == chunk) {
-      assert(0 <= chunk && chunk < _max_chunks, "Sanity");
-      work(chunk, _max_chunks);
-    }
-  }
-}
-
-void ArchiveWorkerTask::configure_max_chunks(int max_chunks) {
-  if (_max_chunks == 0) {
-    _max_chunks = max_chunks;
-  }
-}
-
-bool ArchiveWorkers::run_as_worker() {
-  assert(is_parallel(), "Should be in parallel mode");
-  _start_semaphore.wait();
-
-  // Avalanche wakeups: each worker signals two others.
-  signal_worker_if_needed();
-  signal_worker_if_needed();
-
-  ArchiveWorkerTask* task = Atomic::load_acquire(&_task);
-  task->run();
-
-  // All work done in threads should be visible to caller.
-  OrderAccess::fence();
-
-  // Signal the pool the tasks are complete, if this is the last worker.
-  if (Atomic::sub(&_running_workers, 1, memory_order_relaxed) == 0) {
-    _end_semaphore.signal();
-  }
-
-  // Continue if task was not a termination task.
-  return (task != &_shutdown_task);
-}
-
-ArchiveWorkerThread::ArchiveWorkerThread(ArchiveWorkers* pool) : NamedThread(), _pool(pool) {
-  set_name("ArchiveWorkerThread");
-  os::create_thread(this, os::os_thread);
-  os::start_thread(this);
-}
-
-void ArchiveWorkerThread::run() {
-  // Avalanche thread startup: each starting worker starts two others.
-  _pool->start_worker_if_needed();
-  _pool->start_worker_if_needed();
-
-  // Set ourselves up.
-  os::set_priority(this, NearMaxPriority);
-
-  while (_pool->run_as_worker()) {
-    // Work until terminated.
-  }
-}

src/hotspot/share/cds/archiveUtils.hpp

@@ -33,8 +33,6 @@
 #include "utilities/bitMap.hpp"
 #include "utilities/exceptions.hpp"
 #include "utilities/macros.hpp"
-#include "runtime/nonJavaThread.hpp"
-#include "runtime/semaphore.hpp"
 
 class BootstrapInfo;
 class ReservedSpace;
@@ -164,7 +162,7 @@ public:
   DumpRegion(const char* name, uintx max_delta = 0)
     : _name(name), _base(nullptr), _top(nullptr), _end(nullptr),
       _max_delta(max_delta), _is_packed(false),
-      _rs(NULL), _vs(NULL) {}
+      _rs(nullptr), _vs(nullptr) {}
 
   char* expand_top_to(char* newtop);
   char* allocate(size_t num_bytes, size_t alignment = 0);
@@ -261,16 +259,33 @@ public:
   static bool has_aot_initialized_mirror(InstanceKlass* src_ik);
   template <typename T> static Array<T>* archive_array(GrowableArray<T>* tmp_array);
 
+  // The following functions translate between a u4 offset and an address in the
+  // the range of the mapped CDS archive (e.g., Metaspace::is_in_shared_metaspace()).
+  // Since the first 16 bytes in this range are dummy data (see ArchiveBuilder::reserve_buffer()),
+  // we know that offset 0 never represents a valid object. As a result, an offset of 0
+  // is used to encode a nullptr.
+  //
+  // Use the "archived_address_or_null" variants if a nullptr may be encoded.
+
   // offset must represent an object of type T in the mapped shared space. Return
   // a direct pointer to this object.
-  template <typename T> T static from_offset(u4 offset) {
+  template <typename T> T static offset_to_archived_address(u4 offset) {
+    assert(offset != 0, "sanity");
     T p = (T)(SharedBaseAddress + offset);
     assert(Metaspace::is_in_shared_metaspace(p), "must be");
     return p;
   }
 
+  template <typename T> T static offset_to_archived_address_or_null(u4 offset) {
+    if (offset == 0) {
+      return nullptr;
+    } else {
+      return offset_to_archived_address<T>(offset);
+    }
+  }
+
   // p must be an archived object. Get its offset from SharedBaseAddress
-  template <typename T> static u4 to_offset(T p) {
+  template <typename T> static u4 archived_address_to_offset(T p) {
     uintx pn = (uintx)p;
     uintx base = (uintx)SharedBaseAddress;
     assert(Metaspace::is_in_shared_metaspace(p), "must be");
@@ -279,6 +294,14 @@ public:
     assert(offset <= MAX_SHARED_DELTA, "range check");
     return static_cast<u4>(offset);
   }
+
+  template <typename T> static u4 archived_address_or_null_to_offset(T p) {
+    if (p == nullptr) {
+      return 0;
+    } else {
+      return archived_address_to_offset<T>(p);
+    }
+  }
 };
 
 class HeapRootSegments {
@@ -321,95 +344,4 @@ public:
   HeapRootSegments& operator=(const HeapRootSegments&) = default;
 };
 
-class ArchiveWorkers;
-
-// A task to be worked on by worker threads
-class ArchiveWorkerTask : public CHeapObj<mtInternal> {
-  friend class ArchiveWorkers;
-  friend class ArchiveWorkerShutdownTask;
-private:
-  const char* _name;
-  int _max_chunks;
-  volatile int _chunk;
-
-  void run();
-
-  void configure_max_chunks(int max_chunks);
-
-public:
-  ArchiveWorkerTask(const char* name) :
-      _name(name), _max_chunks(0), _chunk(0) {}
-  const char* name() const { return _name; }
-  virtual void work(int chunk, int max_chunks) = 0;
-};
-
-class ArchiveWorkerThread : public NamedThread {
-  friend class ArchiveWorkers;
-private:
-  ArchiveWorkers* const _pool;
-
-public:
-  ArchiveWorkerThread(ArchiveWorkers* pool);
-  const char* type_name() const override { return "Archive Worker Thread"; }
-  void run() override;
-};
-
-class ArchiveWorkerShutdownTask : public ArchiveWorkerTask {
-public:
-  ArchiveWorkerShutdownTask() : ArchiveWorkerTask("Archive Worker Shutdown") {
-    // This task always have only one chunk.
-    configure_max_chunks(1);
-  }
-  void work(int chunk, int max_chunks) override {
-    // Do nothing.
-  }
-};
-
-// Special worker pool for archive workers. The goal for this pool is to
-// startup fast, distribute spiky workloads efficiently, and being able to
-// shutdown after use. This makes the implementation quite different from
-// the normal GC worker pool.
-class ArchiveWorkers {
-  friend class ArchiveWorkerThread;
-private:
-  // Target number of chunks per worker. This should be large enough to even
-  // out work imbalance, and small enough to keep bookkeeping overheads low.
-  static constexpr int CHUNKS_PER_WORKER = 4;
-  static int max_workers();
-
-  // Global shared instance. Can be uninitialized, can be shut down.
-  static ArchiveWorkers _workers;
-
-  ArchiveWorkerShutdownTask _shutdown_task;
-  Semaphore _start_semaphore;
-  Semaphore _end_semaphore;
-
-  int _num_workers;
-  int _started_workers;
-  int _waiting_workers;
-  int _running_workers;
-
-  typedef enum { NOT_READY, READY, SHUTDOWN } State;
-  volatile State _state;
-
-  ArchiveWorkerTask* _task;
-
-  bool run_as_worker();
-  void start_worker_if_needed();
-  void signal_worker_if_needed();
-
-  void run_task_single(ArchiveWorkerTask* task);
-  void run_task_multi(ArchiveWorkerTask* task);
-
-  bool is_parallel();
-
-  ArchiveWorkers();
-
-public:
-  static ArchiveWorkers* workers() { return &_workers; }
-  void initialize();
-  void shutdown();
-  void run_task(ArchiveWorkerTask* task);
-};
-
 #endif // SHARE_CDS_ARCHIVEUTILS_HPP

src/hotspot/share/cds/cds_globals.hpp

@@ -117,10 +117,7 @@
   product(bool, AOTClassLinking, false,                                     \
           "Load/link all archived classes for the boot/platform/app "       \
           "loaders before application main")                                \
-                                                                            \
+
-  product(bool, AOTCacheParallelRelocation, true, DIAGNOSTIC,               \
-          "Use parallel relocation code to speed up startup.")              \
-                                                                            \
 // end of CDS_FLAGS
 
 DECLARE_FLAGS(CDS_FLAGS)

src/hotspot/share/cds/filemap.cpp

@@ -53,15 +53,18 @@
 #include "memory/oopFactory.hpp"
 #include "memory/universe.hpp"
 #include "nmt/memTracker.hpp"
+#include "oops/access.hpp"
 #include "oops/compressedOops.hpp"
 #include "oops/compressedOops.inline.hpp"
 #include "oops/compressedKlass.hpp"
 #include "oops/objArrayOop.hpp"
 #include "oops/oop.inline.hpp"
+#include "oops/typeArrayKlass.hpp"
 #include "prims/jvmtiExport.hpp"
 #include "runtime/arguments.hpp"
 #include "runtime/globals_extension.hpp"
 #include "runtime/java.hpp"
+#include "runtime/javaCalls.hpp"
 #include "runtime/mutexLocker.hpp"
 #include "runtime/os.hpp"
 #include "runtime/vm_version.hpp"
@@ -1972,32 +1975,6 @@ char* FileMapInfo::map_bitmap_region() {
   return bitmap_base;
 }
 
-class SharedDataRelocationTask : public ArchiveWorkerTask {
-private:
-  BitMapView* const _rw_bm;
-  BitMapView* const _ro_bm;
-  SharedDataRelocator* const _rw_reloc;
-  SharedDataRelocator* const _ro_reloc;
-
-public:
-  SharedDataRelocationTask(BitMapView* rw_bm, BitMapView* ro_bm, SharedDataRelocator* rw_reloc, SharedDataRelocator* ro_reloc) :
-                           ArchiveWorkerTask("Shared Data Relocation"),
-                           _rw_bm(rw_bm), _ro_bm(ro_bm), _rw_reloc(rw_reloc), _ro_reloc(ro_reloc) {}
-
-  void work(int chunk, int max_chunks) override {
-    work_on(chunk, max_chunks, _rw_bm, _rw_reloc);
-    work_on(chunk, max_chunks, _ro_bm, _ro_reloc);
-  }
-
-  void work_on(int chunk, int max_chunks, BitMapView* bm, SharedDataRelocator* reloc) {
-    BitMap::idx_t size  = bm->size();
-    BitMap::idx_t start = MIN2(size, size * chunk / max_chunks);
-    BitMap::idx_t end   = MIN2(size, size * (chunk + 1) / max_chunks);
-    assert(end > start, "Sanity: no empty slices");
-    bm->iterate(reloc, start, end);
-  }
-};
-
 // This is called when we cannot map the archive at the requested[ base address (usually 0x800000000).
 // We relocate all pointers in the 2 core regions (ro, rw).
 bool FileMapInfo::relocate_pointers_in_core_regions(intx addr_delta) {
@@ -2036,14 +2013,8 @@ bool FileMapInfo::relocate_pointers_in_core_regions(intx addr_delta) {
                                 valid_new_base, valid_new_end, addr_delta);
     SharedDataRelocator ro_patcher((address*)ro_patch_base + header()->ro_ptrmap_start_pos(), (address*)ro_patch_end, valid_old_base, valid_old_end,
                                 valid_new_base, valid_new_end, addr_delta);
-
-    if (AOTCacheParallelRelocation) {
-      SharedDataRelocationTask task(&rw_ptrmap, &ro_ptrmap, &rw_patcher, &ro_patcher);
-      ArchiveWorkers::workers()->run_task(&task);
-    } else {
     rw_ptrmap.iterate(&rw_patcher);
     ro_ptrmap.iterate(&ro_patcher);
-    }
 
     // The MetaspaceShared::bm region will be unmapped in MetaspaceShared::initialize_shared_spaces().
 
@@ -2710,11 +2681,44 @@ ClassFileStream* FileMapInfo::open_stream_for_jvmti(InstanceKlass* ik, Handle cl
   const char* const file_name = ClassLoader::file_name_for_class_name(class_name,
                                                                       name->utf8_length());
   ClassLoaderData* loader_data = ClassLoaderData::class_loader_data(class_loader());
-  ClassFileStream* cfs = cpe->open_stream_for_loader(THREAD, file_name, loader_data);
+  ClassFileStream* cfs;
+  if (class_loader() != nullptr && !cpe->is_modules_image()) {
+    cfs = get_stream_from_class_loader(class_loader, cpe, file_name, CHECK_NULL);
+  } else {
+    cfs = cpe->open_stream_for_loader(THREAD, file_name, loader_data);
+  }
   assert(cfs != nullptr, "must be able to read the classfile data of shared classes for built-in loaders.");
   log_debug(cds, jvmti)("classfile data for %s [%d: %s] = %d bytes", class_name, path_index,
                         cfs->source(), cfs->length());
   return cfs;
 }
 
+ClassFileStream* FileMapInfo::get_stream_from_class_loader(Handle class_loader,
+                                                           ClassPathEntry* cpe,
+                                                           const char* file_name,
+                                                           TRAPS) {
+  JavaValue result(T_OBJECT);
+  TempNewSymbol class_name_sym = SymbolTable::new_symbol(file_name);
+  Handle ext_class_name = java_lang_String::externalize_classname(class_name_sym, CHECK_NULL);
+
+  // byte[] ClassLoader.getResourceAsByteArray(String name)
+  JavaCalls::call_virtual(&result,
+                          class_loader,
+                          vmClasses::ClassLoader_klass(),
+                          vmSymbols::getResourceAsByteArray_name(),
+                          vmSymbols::getResourceAsByteArray_signature(),
+                          ext_class_name,
+                          CHECK_NULL);
+  assert(result.get_type() == T_OBJECT, "just checking");
+  oop obj = result.get_oop();
+  assert(obj != nullptr, "ClassLoader.getResourceAsByteArray should not return null");
+
+  // copy from byte[] to a buffer
+  typeArrayOop ba = typeArrayOop(obj);
+  jint len = ba->length();
+  u1* buffer = NEW_RESOURCE_ARRAY(u1, len);
+  ArrayAccess<>::arraycopy_to_native<>(ba, typeArrayOopDesc::element_offset<jbyte>(0), buffer, len);
+
+  return new ClassFileStream(buffer, len, cpe->name());
+}
 #endif

src/hotspot/share/cds/filemap.hpp

@@ -507,6 +507,10 @@ public:
 #if INCLUDE_JVMTI
   // Caller needs a ResourceMark because parts of the returned cfs are resource-allocated.
   static ClassFileStream* open_stream_for_jvmti(InstanceKlass* ik, Handle class_loader, TRAPS);
+  static ClassFileStream* get_stream_from_class_loader(Handle class_loader,
+                                                       ClassPathEntry* cpe,
+                                                       const char* file_name,
+                                                       TRAPS);
 #endif
 
   static SharedClassPathEntry* shared_path(int index) {

src/hotspot/share/cds/lambdaFormInvokers.cpp

@@ -254,7 +254,7 @@ void LambdaFormInvokers::read_static_archive_invokers() {
   if (_static_archive_invokers != nullptr) {
     for (int i = 0; i < _static_archive_invokers->length(); i++) {
       u4 offset = _static_archive_invokers->at(i);
-      Array<char>* line = ArchiveUtils::from_offset<Array<char>*>(offset);
+      Array<char>* line = ArchiveUtils::offset_to_archived_address<Array<char>*>(offset);
       char* str = line->adr_at(0);
       append(str);
     }

src/hotspot/share/cds/lambdaProxyClassDictionary.hpp

@@ -143,7 +143,7 @@ public:
     u4 invoked_name             = b->any_to_offset_u4(key.invoked_name());
     u4 invoked_type             = b->any_to_offset_u4(key.invoked_type());
     u4 method_type              = b->any_to_offset_u4(key.method_type());
-    u4 member_method            = b->any_to_offset_u4(key.member_method());
+    u4 member_method            = b->any_or_null_to_offset_u4(key.member_method()); // could be null
     u4 instantiated_method_type = b->any_to_offset_u4(key.instantiated_method_type());
 
     return RunTimeLambdaProxyClassKey(caller_ik, invoked_name, invoked_type, method_type,
@@ -158,12 +158,12 @@ public:
                                                      Symbol*        instantiated_method_type) {
     // All parameters must be in shared space, or else you'd get an assert in
     // ArchiveUtils::to_offset().
-    return RunTimeLambdaProxyClassKey(ArchiveUtils::to_offset(caller_ik),
+    return RunTimeLambdaProxyClassKey(ArchiveUtils::archived_address_to_offset(caller_ik),
-                                      ArchiveUtils::to_offset(invoked_name),
+                                      ArchiveUtils::archived_address_to_offset(invoked_name),
-                                      ArchiveUtils::to_offset(invoked_type),
+                                      ArchiveUtils::archived_address_to_offset(invoked_type),
-                                      ArchiveUtils::to_offset(method_type),
+                                      ArchiveUtils::archived_address_to_offset(method_type),
-                                      ArchiveUtils::to_offset(member_method),
+                                      ArchiveUtils::archived_address_or_null_to_offset(member_method), // could be null
-                                      ArchiveUtils::to_offset(instantiated_method_type));
+                                      ArchiveUtils::archived_address_to_offset(instantiated_method_type));
   }
 
   unsigned int hash() const;

src/hotspot/share/cds/metaspaceShared.cpp

@@ -315,7 +315,7 @@ static GrowableArrayCHeap<OopHandle, mtClassShared>* _extra_interned_strings = n
 // Extra Symbols to be added to the archive
 static GrowableArrayCHeap<Symbol*, mtClassShared>* _extra_symbols = nullptr;
 // Methods managed by SystemDictionary::find_method_handle_intrinsic() to be added to the archive
-static GrowableArray<Method*>* _pending_method_handle_intrinsics = NULL;
+static GrowableArray<Method*>* _pending_method_handle_intrinsics = nullptr;
 
 void MetaspaceShared::read_extra_data(JavaThread* current, const char* filename) {
   _extra_interned_strings = new GrowableArrayCHeap<OopHandle, mtClassShared>(10000);
@@ -1088,9 +1088,6 @@ void MetaspaceShared::initialize_runtime_shared_and_meta_spaces() {
   assert(CDSConfig::is_using_archive(), "Must be called when UseSharedSpaces is enabled");
   MapArchiveResult result = MAP_ARCHIVE_OTHER_FAILURE;
 
-  // We are about to open the archives. Initialize workers now.
-  ArchiveWorkers::workers()->initialize();
-
   FileMapInfo* static_mapinfo = open_static_archive();
   FileMapInfo* dynamic_mapinfo = nullptr;
 
@@ -1682,9 +1679,6 @@ void MetaspaceShared::initialize_shared_spaces() {
     dynamic_mapinfo->unmap_region(MetaspaceShared::bm);
   }
 
-  // Archive was fully read. Workers are no longer needed.
-  ArchiveWorkers::workers()->shutdown();
-
   LogStreamHandle(Info, cds) lsh;
   if (lsh.is_enabled()) {
     lsh.print("Using AOT-linked classes: %s (static archive: %s aot-linked classes",

src/hotspot/share/cds/runTimeClassInfo.cpp

@@ -79,7 +79,7 @@ InstanceKlass* RunTimeClassInfo::klass() const {
   if (ArchiveBuilder::is_active() && ArchiveBuilder::current()->is_in_buffer_space((address)this)) {
     return ArchiveBuilder::current()->offset_to_buffered<InstanceKlass*>(_klass_offset);
   } else {
-    return ArchiveUtils::from_offset<InstanceKlass*>(_klass_offset);
+    return ArchiveUtils::offset_to_archived_address<InstanceKlass*>(_klass_offset);
   }
 }
 

src/hotspot/share/cds/runTimeClassInfo.hpp

@@ -52,15 +52,15 @@ public:
   struct RTVerifierConstraint {
     u4 _name;
     u4 _from_name;
-    Symbol* name() { return ArchiveUtils::from_offset<Symbol*>(_name); }
+    Symbol* name() { return ArchiveUtils::offset_to_archived_address<Symbol*>(_name); }
-    Symbol* from_name() { return ArchiveUtils::from_offset<Symbol*>(_from_name); }
+    Symbol* from_name() { return ArchiveUtils::offset_to_archived_address<Symbol*>(_from_name); }
   };
 
   struct RTLoaderConstraint {
     u4   _name;
     char _loader_type1;
     char _loader_type2;
-    Symbol* constraint_name() { return ArchiveUtils::from_offset<Symbol*>(_name); }
+    Symbol* constraint_name() { return ArchiveUtils::offset_to_archived_address<Symbol*>(_name); }
   };
   struct RTEnumKlassStaticFields {
     int _num;
@@ -177,11 +177,7 @@ public:
 
   InstanceKlass* nest_host() {
     assert(!ArchiveBuilder::is_active(), "not called when dumping archive");
-    if (_nest_host_offset == 0) {
+    return ArchiveUtils::offset_to_archived_address_or_null<InstanceKlass*>(_nest_host_offset);
-      return nullptr;
-    } else {
-      return ArchiveUtils::from_offset<InstanceKlass*>(_nest_host_offset);
-    }
   }
 
   RTLoaderConstraint* loader_constraints() {

src/hotspot/share/classfile/classFileError.cpp

@@ -34,6 +34,10 @@
 PRAGMA_DIAG_PUSH
 PRAGMA_FORMAT_NONLITERAL_IGNORED
 
+// None of the error routines below take in a free-form, potentially unbounded
+// string, and names are all limited to < 64K, so we know that all formatted
+// strings passed to fthrow will not be excessively large.
+
 void ClassFileParser::classfile_parse_error(const char* msg, TRAPS) const {
   assert(_class_name != nullptr, "invariant");
   ResourceMark rm(THREAD);

src/hotspot/share/classfile/classFileParser.cpp

@@ -1794,6 +1794,7 @@ void ClassFileParser::throwIllegalSignature(const char* type,
   assert(sig != nullptr, "invariant");
 
   ResourceMark rm(THREAD);
+  // Names are all known to be < 64k so we know this formatted message is not excessively large.
   Exceptions::fthrow(THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
       "%s \"%s\" in class %s has illegal signature \"%s\"", type,
@@ -4073,6 +4074,8 @@ void ClassFileParser::check_super_class_access(const InstanceKlass* this_klass,
       char* msg = Reflection::verify_class_access_msg(this_klass,
                                                       InstanceKlass::cast(super),
                                                       vca_result);
+
+      // Names are all known to be < 64k so we know this formatted message is not excessively large.
       if (msg == nullptr) {
         bool same_module = (this_klass->module() == super->module());
         Exceptions::fthrow(
@@ -4121,6 +4124,8 @@ void ClassFileParser::check_super_interface_access(const InstanceKlass* this_kla
       char* msg = Reflection::verify_class_access_msg(this_klass,
                                                       k,
                                                       vca_result);
+
+      // Names are all known to be < 64k so we know this formatted message is not excessively large.
       if (msg == nullptr) {
         bool same_module = (this_klass->module() == k->module());
         Exceptions::fthrow(
@@ -4217,6 +4222,8 @@ static void check_illegal_static_method(const InstanceKlass* this_klass, TRAPS)
     // if m is static and not the init method, throw a verify error
     if ((m->is_static()) && (m->name() != vmSymbols::class_initializer_name())) {
       ResourceMark rm(THREAD);
+
+      // Names are all known to be < 64k so we know this formatted message is not excessively large.
       Exceptions::fthrow(
         THREAD_AND_LOCATION,
         vmSymbols::java_lang_VerifyError(),
@@ -4236,6 +4243,7 @@ void ClassFileParser::verify_legal_class_modifiers(jint flags, TRAPS) const {
   assert(_major_version >= JAVA_9_VERSION || !is_module, "JVM_ACC_MODULE should not be set");
   if (is_module) {
     ResourceMark rm(THREAD);
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_NoClassDefFoundError(),
@@ -4259,6 +4267,7 @@ void ClassFileParser::verify_legal_class_modifiers(jint flags, TRAPS) const {
       (is_interface && major_gte_1_5 && (is_super || is_enum)) ||
       (!is_interface && major_gte_1_5 && is_annotation)) {
     ResourceMark rm(THREAD);
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -4295,6 +4304,7 @@ void ClassFileParser::verify_class_version(u2 major, u2 minor, Symbol* class_nam
   }
 
   if (major > max_version) {
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_UnsupportedClassVersionError(),
@@ -4310,6 +4320,7 @@ void ClassFileParser::verify_class_version(u2 major, u2 minor, Symbol* class_nam
 
   if (minor == JAVA_PREVIEW_MINOR_VERSION) {
     if (major != max_version) {
+      // Names are all known to be < 64k so we know this formatted message is not excessively large.
       Exceptions::fthrow(
         THREAD_AND_LOCATION,
         vmSymbols::java_lang_UnsupportedClassVersionError(),
@@ -4362,6 +4373,7 @@ void ClassFileParser::verify_legal_field_modifiers(jint flags,
 
   if (is_illegal) {
     ResourceMark rm(THREAD);
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -4445,6 +4457,7 @@ void ClassFileParser::verify_legal_method_modifiers(jint flags,
 
   if (is_illegal) {
     ResourceMark rm(THREAD);
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -4686,6 +4699,7 @@ void ClassFileParser::verify_legal_class_name(const Symbol* name, TRAPS) const {
   if (!legal) {
     ResourceMark rm(THREAD);
     assert(_class_name != nullptr, "invariant");
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -4719,6 +4733,7 @@ void ClassFileParser::verify_legal_field_name(const Symbol* name, TRAPS) const {
   if (!legal) {
     ResourceMark rm(THREAD);
     assert(_class_name != nullptr, "invariant");
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -4756,6 +4771,7 @@ void ClassFileParser::verify_legal_method_name(const Symbol* name, TRAPS) const
   if (!legal) {
     ResourceMark rm(THREAD);
     assert(_class_name != nullptr, "invariant");
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_ClassFormatError(),
@@ -5527,6 +5543,7 @@ void ClassFileParser::parse_stream(const ClassFileStream* const stream,
     if (_class_name != class_name_in_cp) {
       if (_class_name != vmSymbols::unknown_class_name()) {
         ResourceMark rm(THREAD);
+        // Names are all known to be < 64k so we know this formatted message is not excessively large.
         Exceptions::fthrow(THREAD_AND_LOCATION,
                            vmSymbols::java_lang_NoClassDefFoundError(),
                            "%s (wrong name: %s)",

src/hotspot/share/classfile/classLoaderHierarchyDCmd.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2018, 2024, Oracle and/or its affiliates. All rights reserved.
  * Copyright (c) 2018 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -46,11 +46,6 @@ public:
   static const char* impact() {
       return "Medium: Depends on number of class loaders and classes loaded.";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission",
-                        "monitor", nullptr};
-    return p;
-  }
   static int num_arguments() { return 3; }
   virtual void execute(DCmdSource source, TRAPS);
 

src/hotspot/share/classfile/classLoaderStats.hpp

@@ -58,12 +58,6 @@ public:
   static int num_arguments() {
     return 0;
   }
-
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission",
-                        "monitor", nullptr};
-    return p;
-  }
 };
 
 

src/hotspot/share/classfile/dictionary.cpp

@@ -44,7 +44,7 @@ const size_t REHASH_LEN = 100;
 Dictionary::Dictionary(ClassLoaderData* loader_data, size_t table_size)
   : _number_of_entries(0), _loader_data(loader_data) {
 
-  size_t start_size_log_2 = MAX2(ceil_log2(table_size), 2); // 2 is minimum size even though some dictionaries only have one entry
+  size_t start_size_log_2 = MAX2(log2i_ceil(table_size), 2); // 2 is minimum size even though some dictionaries only have one entry
   size_t current_size = ((size_t)1) << start_size_log_2;
   log_info(class, loader, data)("Dictionary start size: " SIZE_FORMAT " (" SIZE_FORMAT ")",
                                 current_size, start_size_log_2);

src/hotspot/share/classfile/stringTable.cpp

@@ -309,7 +309,7 @@ public:
 };
 
 void StringTable::create_table() {
-  size_t start_size_log_2 = ceil_log2(StringTableSize);
+  size_t start_size_log_2 = log2i_ceil(StringTableSize);
   _current_size = ((size_t)1) << start_size_log_2;
   log_trace(stringtable)("Start size: " SIZE_FORMAT " (" SIZE_FORMAT ")",
                          _current_size, start_size_log_2);

src/hotspot/share/classfile/symbolTable.cpp

@@ -212,7 +212,7 @@ private:
 };
 
 void SymbolTable::create_table ()  {
-  size_t start_size_log_2 = ceil_log2(SymbolTableSize);
+  size_t start_size_log_2 = log2i_ceil(SymbolTableSize);
   _current_size = ((size_t)1) << start_size_log_2;
   log_trace(symboltable)("Start size: " SIZE_FORMAT " (" SIZE_FORMAT ")",
                          _current_size, start_size_log_2);

src/hotspot/share/classfile/verifier.cpp

@@ -105,11 +105,13 @@ static verify_byte_codes_fn_t verify_byte_codes_fn() {
 
 // Methods in Verifier
 
+// This method determines whether we run the verifier and class file format checking code.
 bool Verifier::should_verify_for(oop class_loader) {
   return class_loader == nullptr ?
     BytecodeVerificationLocal : BytecodeVerificationRemote;
 }
 
+// This method determines whether we allow package access in access checks in reflection.
 bool Verifier::relax_access_for(oop loader) {
   bool trusted = java_lang_ClassLoader::is_trusted_loader(loader);
   bool need_verify =
@@ -120,6 +122,21 @@ bool Verifier::relax_access_for(oop loader) {
   return !need_verify;
 }
 
+// Callers will pass should_verify_class as true, depending on the results of should_verify_for() above,
+// or pass true for redefinition of any class.
+static bool is_eligible_for_verification(InstanceKlass* klass, bool should_verify_class) {
+  Symbol* name = klass->name();
+
+  return (should_verify_class &&
+    // Can not verify the bytecodes for shared classes because they have
+    // already been rewritten to contain constant pool cache indices,
+    // which the verifier can't understand.
+    // Shared classes shouldn't have stackmaps either.
+    // However, bytecodes for shared old classes can be verified because
+    // they have not been rewritten.
+    !(klass->is_shared() && klass->is_rewritten()));
+}
+
 void Verifier::trace_class_resolution(Klass* resolve_class, InstanceKlass* verify_class) {
   assert(verify_class != nullptr, "Unexpected null verify_class");
   ResourceMark rm;
@@ -273,27 +290,6 @@ bool Verifier::verify(InstanceKlass* klass, bool should_verify_class, TRAPS) {
   }
 }
 
-bool Verifier::is_eligible_for_verification(InstanceKlass* klass, bool should_verify_class) {
-  Symbol* name = klass->name();
-
-  return (should_verify_class &&
-    // return if the class is a bootstrapping class
-    // or defineClass specified not to verify by default (flags override passed arg)
-    // We need to skip the following four for bootstraping
-    name != vmSymbols::java_lang_Object() &&
-    name != vmSymbols::java_lang_Class() &&
-    name != vmSymbols::java_lang_String() &&
-    name != vmSymbols::java_lang_Throwable() &&
-
-    // Can not verify the bytecodes for shared classes because they have
-    // already been rewritten to contain constant pool cache indices,
-    // which the verifier can't understand.
-    // Shared classes shouldn't have stackmaps either.
-    // However, bytecodes for shared old classes can be verified because
-    // they have not been rewritten.
-    !(klass->is_shared() && klass->is_rewritten()));
-}
-
 Symbol* Verifier::inference_verify(
     InstanceKlass* klass, char* message, size_t message_len, TRAPS) {
   JavaThread* thread = THREAD;

src/hotspot/share/classfile/verifier.hpp

@@ -61,7 +61,6 @@ class Verifier : AllStatic {
   static void trace_class_resolution(Klass* resolve_class, InstanceKlass* verify_class);
 
  private:
-  static bool is_eligible_for_verification(InstanceKlass* klass, bool should_verify_class);
   static Symbol* inference_verify(
     InstanceKlass* klass, char* msg, size_t msg_len, TRAPS);
 };

src/hotspot/share/classfile/vmSymbols.hpp

@@ -723,6 +723,8 @@ class SerializeClosure;
   template(dumpSharedArchive_signature,                     "(ZLjava/lang/String;)Ljava/lang/String;")            \
   template(generateLambdaFormHolderClasses,                 "generateLambdaFormHolderClasses")                    \
   template(generateLambdaFormHolderClasses_signature,       "([Ljava/lang/String;)[Ljava/lang/Object;")           \
+  template(getResourceAsByteArray_name,                     "getResourceAsByteArray")                             \
+  template(getResourceAsByteArray_signature,                "(Ljava/lang/String;)[B")                             \
   template(java_lang_Enum,                                  "java/lang/Enum")                                     \
   template(java_lang_invoke_Invokers_Holder,                "java/lang/invoke/Invokers$Holder")                   \
   template(java_lang_invoke_DirectMethodHandle_Holder,      "java/lang/invoke/DirectMethodHandle$Holder")         \

src/hotspot/share/compiler/methodMatcher.cpp

@@ -219,21 +219,22 @@ bool MethodMatcher::match(Symbol* candidate, Symbol* match, Mode match_mode) con
 
 static MethodMatcher::Mode check_mode(char name[], const char*& error_msg) {
   int match = MethodMatcher::Exact;
+  size_t len = strlen(name);
   if (name[0] == '*') {
-    if (strlen(name) == 1) {
+    if (len == 1) {
       return MethodMatcher::Any;
     }
     match |= MethodMatcher::Suffix;
-    memmove(name, name + 1, strlen(name + 1) + 1);
+    memmove(name, name + 1, len); // Include terminating nul in move.
+    len--;
   }
 
-  size_t len = strlen(name);
   if (len > 0 && name[len - 1] == '*') {
     match |= MethodMatcher::Prefix;
     name[--len] = '\0';
   }
 
-  if (strlen(name) == 0) {
+  if (len == 0) {
     error_msg = "** Not a valid pattern";
     return MethodMatcher::Any;
   }

src/hotspot/share/gc/g1/g1Arguments.cpp

@@ -244,10 +244,7 @@ void G1Arguments::initialize() {
   if (max_parallel_refinement_threads > UINT_MAX / divisor) {
     vm_exit_during_initialization("Too large parallelism for remembered sets.");
   }
-}
 
-void G1Arguments::initialize_heap_flags_and_sizes() {
-  GCArguments::initialize_heap_flags_and_sizes();
   FullGCForwarding::initialize_flags(heap_reserved_size_bytes());
 }
 

src/hotspot/share/gc/g1/g1Arguments.hpp

@@ -39,7 +39,6 @@ class G1Arguments : public GCArguments {
   static void parse_verification_type(const char* type);
 
   virtual void initialize_alignments();
-  virtual void initialize_heap_flags_and_sizes();
 
   virtual void initialize();
   virtual size_t conservative_max_heap_alignment();

src/hotspot/share/gc/parallel/parallelArguments.cpp

@@ -83,6 +83,8 @@ void ParallelArguments::initialize() {
   if (FLAG_IS_DEFAULT(ParallelRefProcEnabled) && ParallelGCThreads > 1) {
     FLAG_SET_DEFAULT(ParallelRefProcEnabled, true);
   }
+
+  FullGCForwarding::initialize_flags(heap_reserved_size_bytes());
 }
 
 // The alignment used for boundary between young gen and old gen
@@ -128,7 +130,6 @@ void ParallelArguments::initialize_heap_flags_and_sizes() {
     // Redo everything from the start
     initialize_heap_flags_and_sizes_one_pass();
   }
-  FullGCForwarding::initialize_flags(heap_reserved_size_bytes());
 }
 
 size_t ParallelArguments::heap_reserved_size_bytes() {

src/hotspot/share/gc/parallel/psParallelCompact.cpp

@@ -362,8 +362,9 @@ HeapWord* ParallelCompactData::summarize_split_space(size_t src_region,
 
     split_info.record(split_region, overflowing_obj, preceding_live_words);
 
-    HeapWord* src_region_start = region_to_addr(src_region);
+    // The [overflowing_obj, src_region_start) part has been accounted for, so
-    HeapWord* new_top = destination - pointer_delta(src_region_start, overflowing_obj);
+    // must move back the new_top, now that this overflowing obj is deferred.
+    HeapWord* new_top = destination - pointer_delta(region_to_addr(src_region), overflowing_obj);
 
     // If the overflowing obj was relocated to its original destination,
     // those destination regions would have their source_region set. Now that
@@ -890,7 +891,7 @@ void PSParallelCompact::summary_phase()
       _summary_data.summarize_dense_prefix(old_space->bottom(), dense_prefix_end);
     }
 
-    // Compacting objs inn [dense_prefix_end, old_space->top())
+    // Compacting objs in [dense_prefix_end, old_space->top())
     _summary_data.summarize(_space_info[id].split_info(),
                             dense_prefix_end, old_space->top(), nullptr,
                             dense_prefix_end, old_space->end(),
@@ -1596,9 +1597,9 @@ void PSParallelCompact::forward_to_new_addr() {
                                  &start_region, &end_region);
         for (size_t cur_region = start_region; cur_region < end_region; ++cur_region) {
           RegionData* region_ptr = _summary_data.region(cur_region);
-          size_t live_words = region_ptr->partial_obj_size();
+          size_t partial_obj_size = region_ptr->partial_obj_size();
 
-          if (live_words == ParallelCompactData::RegionSize) {
+          if (partial_obj_size == ParallelCompactData::RegionSize) {
             // No obj-start
             continue;
           }
@@ -1606,19 +1607,18 @@ void PSParallelCompact::forward_to_new_addr() {
           HeapWord* region_start = _summary_data.region_to_addr(cur_region);
           HeapWord* region_end = region_start + ParallelCompactData::RegionSize;
 
-
           if (split_info.is_split(cur_region)) {
             // Part 1: will be relocated to space-1
             HeapWord* preceding_destination = split_info.preceding_destination();
             HeapWord* split_point = split_info.split_point();
-            forward_objs_in_range(cm, region_start + live_words, split_point, preceding_destination + live_words);
+            forward_objs_in_range(cm, region_start + partial_obj_size, split_point, preceding_destination + partial_obj_size);
 
             // Part 2: will be relocated to space-2
             HeapWord* destination = region_ptr->destination();
             forward_objs_in_range(cm, split_point, region_end, destination);
           } else {
             HeapWord* destination = region_ptr->destination();
-            forward_objs_in_range(cm, region_start + live_words, region_end, destination + live_words);
+            forward_objs_in_range(cm, region_start + partial_obj_size, region_end, destination + partial_obj_size);
           }
         }
       }
@@ -1984,11 +1984,11 @@ HeapWord* PSParallelCompact::skip_live_words(HeapWord* beg, HeapWord* end, size_
   }
 }
 
-// On filling a destination region (dest-region), we need to know the location
+// On starting to fill a destination region (dest-region), we need to know the
-// of the word that will be at the start of the dest-region after compaction.
+// location of the word that will be at the start of the dest-region after
-// A dest-region can have one or more source regions, but only the first
+// compaction. A dest-region can have one or more source regions, but only the
-// source-region contains this location. This location is retrieved by calling
+// first source-region contains this location. This location is retrieved by
-// `first_src_addr` on a dest-region.
+// calling `first_src_addr` on a dest-region.
 // Conversely, a source-region has a dest-region which holds the destination of
 // the first live word on this source-region, based on which the destination
 // for the rest of live words can be derived.
@@ -2017,9 +2017,9 @@ HeapWord* PSParallelCompact::skip_live_words(HeapWord* beg, HeapWord* end, size_
 //              ^                  ^
 //              | old-space-end    | eden-space-start
 //
-// Therefore, in this example, region-n will have two dest-regions, one for
+// Therefore, in this example, region-n will have two dest-regions:
-// the final region in old-space and the other for the first region in
+// 1. the final region in old-space
-// eden-space.
+// 2. the first region in eden-space.
 // To handle this special case, we introduce the concept of split-region, whose
 // contents are relocated to two spaces. `SplitInfo` captures all necessary
 // info about the split, the first part, spliting-point, and the second part.

src/hotspot/share/gc/serial/serialArguments.cpp

@@ -24,12 +24,12 @@
 
 #include "precompiled.hpp"
 #include "gc/shared/fullGCForwarding.hpp"
-#include "gc/shared/genArguments.hpp"
+#include "gc/shared/gcArguments.hpp"
 #include "gc/serial/serialArguments.hpp"
 #include "gc/serial/serialHeap.hpp"
 
-void SerialArguments::initialize_heap_flags_and_sizes() {
+void SerialArguments::initialize() {
-  GenArguments::initialize_heap_flags_and_sizes();
+  GCArguments::initialize();
   FullGCForwarding::initialize_flags(MaxHeapSize);
 }
 

src/hotspot/share/gc/serial/serialArguments.hpp

@@ -31,8 +31,8 @@ class CollectedHeap;
 
 class SerialArguments : public GenArguments {
 private:
+  virtual void initialize();
   virtual CollectedHeap* create_heap();
-  virtual void initialize_heap_flags_and_sizes();
 };
 
 #endif // SHARE_GC_SERIAL_SERIALARGUMENTS_HPP

src/hotspot/share/gc/shared/locationPrinter.inline.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2019, 2024, Oracle and/or its affiliates. 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
@@ -54,7 +54,7 @@ bool BlockLocationPrinter<CollectedHeapT>::print_location(outputStream* st, void
   // Check if addr points into Java heap.
   bool in_heap = CollectedHeapT::heap()->is_in(addr);
   if (in_heap) {
-    // base_oop_or_null() might be unimplemented and return NULL for some GCs/generations
+    // base_oop_or_null() might be unimplemented and return null for some GCs/generations
     oop o = base_oop_or_null(addr);
     if (o != nullptr) {
       if ((void*)o == addr) {

src/hotspot/share/gc/shenandoah/shenandoahArguments.cpp

@@ -176,6 +176,8 @@ void ShenandoahArguments::initialize() {
   if (FLAG_IS_DEFAULT(TLABAllocationWeight)) {
     FLAG_SET_DEFAULT(TLABAllocationWeight, 90);
   }
+
+  FullGCForwarding::initialize_flags(MaxHeapSize);
 }
 
 size_t ShenandoahArguments::conservative_max_heap_alignment() {
@@ -199,11 +201,6 @@ void ShenandoahArguments::initialize_alignments() {
   HeapAlignment = align;
 }
 
-void ShenandoahArguments::initialize_heap_flags_and_sizes() {
-  GCArguments::initialize_heap_flags_and_sizes();
-  FullGCForwarding::initialize_flags(MaxHeapSize);
-}
-
 CollectedHeap* ShenandoahArguments::create_heap() {
   return new ShenandoahHeap(new ShenandoahCollectorPolicy());
 }

src/hotspot/share/gc/shenandoah/shenandoahArguments.hpp

@@ -35,7 +35,6 @@ private:
 
   virtual void initialize();
   virtual size_t conservative_max_heap_alignment();
-  virtual void initialize_heap_flags_and_sizes();
   virtual CollectedHeap* create_heap();
 };
 

src/hotspot/share/gc/shenandoah/shenandoahPhaseTimings.cpp

@@ -185,33 +185,33 @@ void ShenandoahPhaseTimings::flush_par_workers_to_cycle() {
   for (uint pi = 0; pi < _num_phases; pi++) {
     Phase phase = Phase(pi);
     if (is_worker_phase(phase)) {
-      double s = uninitialized();
+      double sum = uninitialized();
       for (uint i = 1; i < _num_par_phases; i++) {
         ShenandoahWorkerData* wd = worker_data(phase, ParPhase(i));
-        double ws = uninitialized();
+        double worker_sum = uninitialized();
         for (uint c = 0; c < _max_workers; c++) {
-          double v = wd->get(c);
+          double worker_time = wd->get(c);
-          if (v != ShenandoahWorkerData::uninitialized()) {
+          if (worker_time != ShenandoahWorkerData::uninitialized()) {
-            if (ws == uninitialized()) {
+            if (worker_sum == uninitialized()) {
-              ws = v;
+              worker_sum = worker_time;
             } else {
-              ws += v;
+              worker_sum += worker_time;
             }
           }
         }
-        if (ws != uninitialized()) {
+        if (worker_sum != uninitialized()) {
           // add to each line in phase
-          set_cycle_data(Phase(phase + i + 1), ws);
+          set_cycle_data(Phase(phase + i + 1), worker_sum);
-          if (s == uninitialized()) {
+          if (sum == uninitialized()) {
-            s = ws;
+            sum = worker_sum;
           } else {
-            s += ws;
+            sum += worker_sum;
           }
         }
       }
-      if (s != uninitialized()) {
+      if (sum != uninitialized()) {
         // add to total for phase
-        set_cycle_data(Phase(phase + 1), s);
+        set_cycle_data(Phase(phase + 1), sum);
       }
     }
   }

src/hotspot/share/include/jvm.h

@@ -1147,13 +1147,6 @@ JVM_GetClassFileVersion(JNIEnv *env, jclass current);
 JNIEXPORT jboolean JNICALL
 JVM_PrintWarningAtDynamicAgentLoad(void);
 
-#define JNI_ONLOAD_SYMBOLS   {"JNI_OnLoad"}
-#define JNI_ONUNLOAD_SYMBOLS {"JNI_OnUnload"}
-#define JVM_ONLOAD_SYMBOLS      {"JVM_OnLoad"}
-#define AGENT_ONLOAD_SYMBOLS    {"Agent_OnLoad"}
-#define AGENT_ONUNLOAD_SYMBOLS  {"Agent_OnUnload"}
-#define AGENT_ONATTACH_SYMBOLS  {"Agent_OnAttach"}
-
 /*
  * This structure is used by the launcher to get the default thread
  * stack size from the VM using JNI_GetDefaultJavaVMInitArgs() with a

src/hotspot/share/interpreter/linkResolver.cpp

@@ -323,6 +323,9 @@ void LinkResolver::check_klass_accessibility(Klass* ref_klass, Klass* sel_klass,
     char* msg = Reflection::verify_class_access_msg(ref_klass,
                                                     InstanceKlass::cast(base_klass),
                                                     vca_result);
+
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
+
     bool same_module = (base_klass->module() == ref_klass->module());
     if (msg == nullptr) {
       Exceptions::fthrow(
@@ -615,6 +618,7 @@ void LinkResolver::check_method_accessability(Klass* ref_klass,
       print_nest_host_error_on(&ss, ref_klass, sel_klass);
     }
 
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(THREAD_AND_LOCATION,
                        vmSymbols::java_lang_IllegalAccessError(),
                        "%s",
@@ -968,6 +972,7 @@ void LinkResolver::check_field_accessability(Klass* ref_klass,
     if (fd.is_private()) {
       print_nest_host_error_on(&ss, ref_klass, sel_klass);
     }
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(THREAD_AND_LOCATION,
                        vmSymbols::java_lang_IllegalAccessError(),
                        "%s",
@@ -1187,6 +1192,7 @@ Method* LinkResolver::linktime_resolve_special_method(const LinkInfo& link_info,
     ss.print(" %s(", resolved_method->name()->as_C_string());
     resolved_method->signature()->print_as_signature_external_parameters(&ss);
     ss.print(")' not found");
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(
       THREAD_AND_LOCATION,
       vmSymbols::java_lang_NoSuchMethodError(),

src/hotspot/share/jfr/dcmd/jfrDcmds.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2012, 2024, Oracle and/or its affiliates. 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
@@ -59,10 +59,6 @@ class JfrStartFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Medium: Depending on the settings for a recording, the impact can range from low to high.";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdStart";
   }
@@ -84,10 +80,6 @@ class JfrDumpFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdDump";
   }
@@ -109,10 +101,6 @@ class JfrCheckFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdCheck";
   }
@@ -134,10 +122,6 @@ class JfrStopFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdStop";
   }
@@ -159,10 +143,6 @@ class JfrViewFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Medium";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdView";
   }
@@ -184,10 +164,6 @@ class JfrQueryFlightRecordingDCmd : public JfrDCmd {
   static const char* impact() {
     return "Medium";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   virtual const char* javaClass() const {
     return "jdk/jfr/internal/dcmd/DCmdQuery";
   }
@@ -225,10 +201,6 @@ class JfrConfigureFlightRecorderDCmd : public DCmdWithParser {
   static const char* impact() {
     return "Low";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "monitor", nullptr};
-    return p;
-  }
   static int num_arguments() { return 10; }
   virtual void execute(DCmdSource source, TRAPS);
   virtual void print_help(const char* name) const;

src/hotspot/share/jfr/leakprofiler/sampling/objectSampler.cpp

@@ -258,12 +258,25 @@ void ObjectSampler::add(HeapWord* obj, size_t allocated, traceid thread_id, bool
       // quick reject, will not fit
       return;
     }
-    sample = _list->reuse(_priority_queue->pop());
+    ObjectSample* popped = _priority_queue->pop();
+    size_t popped_span = popped->span();
+    ObjectSample* previous = popped->prev();
+    sample = _list->reuse(popped);
+    assert(sample != nullptr, "invariant");
+    if (previous != nullptr) {
+      push_span(previous, popped_span);
+      sample->set_span(span);
+    } else {
+      // The removed sample was the youngest sample in the list, which means the new sample is now the youngest
+      // sample. It should cover the spans of both.
+      sample->set_span(span + popped_span);
+    }
   } else {
     sample = _list->get();
+    assert(sample != nullptr, "invariant");
+    sample->set_span(span);
   }
 
-  assert(sample != nullptr, "invariant");
   signal_unresolved_entry();
   sample->set_thread_id(thread_id);
   if (virtual_thread) {
@@ -278,7 +291,6 @@ void ObjectSampler::add(HeapWord* obj, size_t allocated, traceid thread_id, bool
     sample->set_stack_trace_hash(stacktrace_hash);
   }
 
-  sample->set_span(allocated);
   sample->set_object(cast_to_oop(obj));
   sample->set_allocated(allocated);
   sample->set_allocation_time(JfrTicks::now());
@@ -305,14 +317,18 @@ void ObjectSampler::remove_dead(ObjectSample* sample) {
   ObjectSample* const previous = sample->prev();
   // push span onto previous
   if (previous != nullptr) {
-    _priority_queue->remove(previous);
+    push_span(previous, sample->span());
-    previous->add_span(sample->span());
-    _priority_queue->push(previous);
   }
   _priority_queue->remove(sample);
   _list->release(sample);
 }
 
+void ObjectSampler::push_span(ObjectSample* sample, size_t span) {
+    _priority_queue->remove(sample);
+    sample->add_span(span);
+    _priority_queue->push(sample);
+}
+
 ObjectSample* ObjectSampler::last() const {
   return _list->last();
 }

src/hotspot/share/jfr/leakprofiler/sampling/objectSampler.hpp

@@ -64,6 +64,7 @@ class ObjectSampler : public CHeapObj<mtTracing> {
   void add(HeapWord* object, size_t size, traceid thread_id, bool virtual_thread, const JfrBlobHandle& bh, JavaThread* thread);
   void scavenge();
   void remove_dead(ObjectSample* sample);
+  void push_span(ObjectSample* sample, size_t span);
 
   const ObjectSample* item_at(int index) const;
   ObjectSample* item_at(int index);

src/hotspot/share/logging/logDiagnosticCommand.hpp

@@ -58,11 +58,6 @@ class LogDiagnosticCommand : public DCmdWithParser {
   static const char* description() {
     return "Lists current log configuration, enables/disables/configures a log output, or rotates all logs.";
   }
-
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission", "control", nullptr};
-    return p;
-  }
 };
 
 #endif // SHARE_LOGGING_LOGDIAGNOSTICCOMMAND_HPP

src/hotspot/share/memory/allocation.hpp

@@ -353,7 +353,7 @@ class MetaspaceObj {
   void* operator new(size_t size, ClassLoaderData* loader_data,
                      size_t word_size,
                      Type type) throw();
-  void operator delete(void* p) { ShouldNotCallThis(); }
+  void operator delete(void* p) = delete;
 
   // Declare a *static* method with the same signature in any subclass of MetaspaceObj
   // that should be read-only by default. See symbol.hpp for an example. This function

src/hotspot/share/memory/metaspace/metaspaceDCmd.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2018, 2024, Oracle and/or its affiliates. All rights reserved.
  * Copyright (c) 2018, 2020 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -52,11 +52,6 @@ public:
   static const char* impact() {
       return "Medium: Depends on number of classes loaded.";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission",
-                        "monitor", nullptr};
-    return p;
-  }
   static int num_arguments() { return 8; }
   virtual void execute(DCmdSource source, TRAPS);
 };

src/hotspot/share/memory/universe.cpp

@@ -62,6 +62,7 @@
 #include "oops/instanceMirrorKlass.hpp"
 #include "oops/klass.inline.hpp"
 #include "oops/objArrayOop.inline.hpp"
+#include "oops/objLayout.hpp"
 #include "oops/oop.inline.hpp"
 #include "oops/oopHandle.inline.hpp"
 #include "oops/typeArrayKlass.hpp"
@@ -868,6 +869,8 @@ jint universe_init() {
   // Initialize CPUTimeCounters object, which must be done before creation of the heap.
   CPUTimeCounters::initialize();
 
+  ObjLayout::initialize();
+
 #ifdef _LP64
   MetaspaceShared::adjust_heap_sizes_for_dumping();
 #endif // _LP64

src/hotspot/share/nmt/memoryFileTracker.cpp

@@ -179,15 +179,11 @@ const GrowableArrayCHeap<MemoryFileTracker::MemoryFile*, mtNMT>& MemoryFileTrack
 };
 
 void MemoryFileTracker::summary_snapshot(VirtualMemorySnapshot* snapshot) const {
-  for (int d = 0; d < _files.length(); d++) {
+  iterate_summary([&](MemTag tag, const VirtualMemory* current) {
-    const MemoryFile* file = _files.at(d);
+    VirtualMemory* snap = snapshot->by_type(tag);
-    for (int i = 0; i < mt_number_of_tags; i++) {
-      VirtualMemory* snap = snapshot->by_type(NMTUtil::index_to_tag(i));
-      const VirtualMemory* current = file->_summary.by_type(NMTUtil::index_to_tag(i));
     // Only account the committed memory.
     snap->commit_memory(current->committed());
-    }
+  });
-  }
 }
 
 void MemoryFileTracker::Instance::summary_snapshot(VirtualMemorySnapshot* snapshot) {

src/hotspot/share/nmt/memoryFileTracker.hpp

@@ -39,6 +39,8 @@
 // The MemoryFileTracker tracks memory of 'memory files',
 // storage with its own memory space separate from the process.
 // A typical example of such a file is a memory mapped file.
+// All memory is accounted as committed, there is no reserved memory.
+// Any reserved memory is expected to exist in the VirtualMemoryTracker.
 class MemoryFileTracker {
   friend class NMTMemoryFileTrackerTest;
 
@@ -72,6 +74,16 @@ public:
   MemoryFile* make_file(const char* descriptive_name);
   void free_file(MemoryFile* file);
 
+  template<typename F>
+  void iterate_summary(F f) const {
+    for (int d = 0; d < _files.length(); d++) {
+      const MemoryFile* file = _files.at(d);
+      for (int i = 0; i < mt_number_of_tags; i++) {
+        f(NMTUtil::index_to_tag(i), file->_summary.by_type(NMTUtil::index_to_tag(i)));
+      }
+    }
+  }
+
   void summary_snapshot(VirtualMemorySnapshot* snapshot) const;
 
   // Print detailed report of file
@@ -99,6 +111,11 @@ public:
                                 const NativeCallStack& stack, MemTag mem_tag);
     static void free_memory(MemoryFile* device, size_t offset, size_t size);
 
+    template<typename F>
+    static void iterate_summary(F f) {
+      _tracker->iterate_summary(f);
+    };
+
     static void summary_snapshot(VirtualMemorySnapshot* snapshot);
 
     static void print_report_on(const MemoryFile* device, outputStream* stream, size_t scale);

src/hotspot/share/nmt/nmtDCmd.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2012, 2024, Oracle and/or its affiliates. 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
@@ -53,11 +53,6 @@ class NMTDCmd: public DCmdWithParser {
   static const char* impact() {
     return "Medium";
   }
-  static const JavaPermission permission() {
-    JavaPermission p = {"java.lang.management.ManagementPermission",
-                        "monitor", nullptr};
-    return p;
-  }
   virtual void execute(DCmdSource source, TRAPS);
 
  private:

src/hotspot/share/nmt/nmtUsage.cpp

@@ -24,6 +24,7 @@
 
 #include "precompiled.hpp"
 #include "nmt/mallocTracker.hpp"
+#include "nmt/memoryFileTracker.hpp"
 #include "nmt/nmtCommon.hpp"
 #include "nmt/nmtUsage.hpp"
 #include "nmt/threadStackTracker.hpp"
@@ -90,6 +91,16 @@ void NMTUsage::update_vm_usage() {
     _vm_total.reserved += vm->reserved();
     _vm_total.committed += vm->committed();
   }
+
+  { // MemoryFileTracker addition
+    using MFT = MemoryFileTracker::Instance;
+    MFT::Locker lock;
+    MFT::iterate_summary([&](MemTag tag, const VirtualMemory* vm) {
+      int i = NMTUtil::tag_to_index(tag);
+      _vm_by_type[i].committed += vm->committed();
+      _vm_total.committed += vm->committed();
+    });
+  }
 }
 
 void NMTUsage::refresh() {

src/hotspot/share/nmt/nmtUsage.hpp

@@ -26,6 +26,7 @@
 #define SHARE_NMT_NMTUSAGE_HPP
 
 #include "memory/allocation.hpp"
+#include "nmt/memTag.hpp"
 #include "utilities/globalDefinitions.hpp"
 
 struct NMTUsagePair {

src/hotspot/share/oops/constantPool.cpp

@@ -428,7 +428,7 @@ void ConstantPool::restore_unshareable_info(TRAPS) {
   assert(is_shared(), "should always be set for shared constant pools");
   if (is_for_method_handle_intrinsic()) {
     // See the same check in remove_unshareable_info() below.
-    assert(cache() == NULL, "must not have cpCache");
+    assert(cache() == nullptr, "must not have cpCache");
     return;
   }
   assert(_cache != nullptr, "constant pool _cache should not be null");
@@ -474,7 +474,7 @@ void ConstantPool::remove_unshareable_info() {
     // This CP was created by Method::make_method_handle_intrinsic() and has nothing
     // that need to be removed/restored. It has no cpCache since the intrinsic methods
     // don't have any bytecodes.
-    assert(cache() == NULL, "must not have cpCache");
+    assert(cache() == nullptr, "must not have cpCache");
     return;
   }
 
@@ -1266,6 +1266,7 @@ oop ConstantPool::resolve_constant_at_impl(const constantPoolHandle& this_cp,
                  cp_index,
                  callee->is_interface() ? "CONSTANT_MethodRef" : "CONSTANT_InterfaceMethodRef",
                  callee->is_interface() ? "CONSTANT_InterfaceMethodRef" : "CONSTANT_MethodRef");
+        // Names are all known to be < 64k so we know this formatted message is not excessively large.
         Exceptions::fthrow(THREAD_AND_LOCATION, vmSymbols::java_lang_IncompatibleClassChangeError(), "%s", ss.as_string());
         save_and_throw_exception(this_cp, cp_index, tag, CHECK_NULL);
       }

src/hotspot/share/oops/cpCache.cpp

@@ -829,9 +829,15 @@ oop ConstantPoolCache::appendix_if_resolved(ResolvedMethodEntry* method_entry) c
 void ConstantPoolCache::print_on(outputStream* st) const {
   st->print_cr("%s", internal_name());
   // print constant pool cache entries
+  if (_resolved_field_entries != nullptr) {
     print_resolved_field_entries(st);
+  }
+  if (_resolved_method_entries != nullptr) {
     print_resolved_method_entries(st);
+  }
+  if (_resolved_indy_entries != nullptr) {
     print_resolved_indy_entries(st);
+  }
 }
 
 void ConstantPoolCache::print_resolved_field_entries(outputStream* st) const {

src/hotspot/share/oops/instanceKlass.cpp

@@ -899,6 +899,7 @@ bool InstanceKlass::link_class_impl(TRAPS) {
     // if we are executing Java code. This is not a problem for CDS dumping phase since
     // it doesn't execute any Java code.
     ResourceMark rm(THREAD);
+    // Names are all known to be < 64k so we know this formatted message is not excessively large.
     Exceptions::fthrow(THREAD_AND_LOCATION,
                        vmSymbols::java_lang_NoClassDefFoundError(),
                        "Class %s, or one of its supertypes, failed class initialization",
@@ -919,6 +920,7 @@ bool InstanceKlass::link_class_impl(TRAPS) {
   if (super_klass != nullptr) {
     if (super_klass->is_interface()) {  // check if super class is an interface
       ResourceMark rm(THREAD);
+      // Names are all known to be < 64k so we know this formatted message is not excessively large.
       Exceptions::fthrow(
         THREAD_AND_LOCATION,
         vmSymbols::java_lang_IncompatibleClassChangeError(),
@@ -3286,6 +3288,7 @@ InstanceKlass* InstanceKlass::compute_enclosing_class(bool* inner_is_member, TRA
         // If the outer class is not an instance klass then it cannot have
         // declared any inner classes.
         ResourceMark rm(THREAD);
+        // Names are all known to be < 64k so we know this formatted message is not excessively large.
         Exceptions::fthrow(
           THREAD_AND_LOCATION,
           vmSymbols::java_lang_IncompatibleClassChangeError(),

src/hotspot/share/oops/instanceKlass.hpp

@@ -45,7 +45,6 @@
 class ConstantPool;
 class DeoptimizationScope;
 class klassItable;
-class Monitor;
 class RecordComponent;
 
 // An InstanceKlass is the VM level representation of a Java class.
@@ -68,7 +67,6 @@ class ClassFileStream;
 class KlassDepChange;
 class DependencyContext;
 class fieldDescriptor;
-class jniIdMapBase;
 class JNIid;
 class JvmtiCachedClassFieldMap;
 class nmethodBucket;

src/hotspot/share/oops/klass.hpp

@@ -25,8 +25,6 @@
 #ifndef SHARE_OOPS_KLASS_HPP
 #define SHARE_OOPS_KLASS_HPP
 
-#include "memory/iterator.hpp"
-#include "memory/memRegion.hpp"
 #include "oops/klassFlags.hpp"
 #include "oops/markWord.hpp"
 #include "oops/metadata.hpp"
@@ -60,8 +58,6 @@ class fieldDescriptor;
 class klassVtable;
 class ModuleEntry;
 class PackageEntry;
-class ParCompactionManager;
-class PSPromotionManager;
 class vtableEntry;
 
 class Klass : public Metadata {

src/hotspot/share/oops/method.cpp

@@ -335,7 +335,8 @@ int Method::bci_from(address bcp) const {
 
 
 int Method::validate_bci(int bci) const {
-  return (bci == 0 || bci < code_size()) ? bci : -1;
+  // Called from the verifier, and should return -1 if not valid.
+  return ((is_native() && bci == 0) || (!is_native() && 0 <= bci && bci < code_size())) ? bci : -1;
 }
 
 // Return bci if it appears to be a valid bcp

src/hotspot/share/oops/method.hpp

@@ -32,7 +32,6 @@
 #include "oops/methodFlags.hpp"
 #include "oops/instanceKlass.hpp"
 #include "oops/oop.hpp"
-#include "oops/typeArrayOop.hpp"
 #include "utilities/accessFlags.hpp"
 #include "utilities/align.hpp"
 #include "utilities/growableArray.hpp"

src/hotspot/share/opto/block.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2024, Oracle and/or its affiliates. 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
@@ -181,10 +181,11 @@ int Block::is_Empty() const {
     return success_result;
   }
 
-  // Ideal nodes are allowable in empty blocks: skip them  Only MachNodes
+  // Ideal nodes (except BoxLock) are allowable in empty blocks: skip them. Only
-  // turn directly into code, because only MachNodes have non-trivial
+  // Mach and BoxLock nodes turn directly into code via emit().
-  // emit() functions.
+  while ((end_idx > 0) &&
-  while ((end_idx > 0) && !get_node(end_idx)->is_Mach()) {
+         !get_node(end_idx)->is_Mach() &&
+         !get_node(end_idx)->is_BoxLock()) {
     end_idx--;
   }
 

src/hotspot/share/opto/c2_globals.hpp

@@ -355,6 +355,12 @@
   product(bool, SuperWordReductions, true,                                  \
           "Enable reductions support in superword.")                        \
                                                                             \
+  product_pd(uint, SuperWordStoreToLoadForwardingFailureDetection, DIAGNOSTIC, \
+          "if >0, auto-vectorization detects possible store-to-load "       \
+          "forwarding failures. The number specifies over how many "        \
+          "loop iterations this detection spans.")                          \
+          range(0, 4096)                                                    \
+                                                                            \
   product(bool, UseCMoveUnconditionally, false,                             \
           "Use CMove (scalar and vector) ignoring profitability test.")     \
                                                                             \

src/hotspot/share/opto/chaitin.cpp

@@ -956,7 +956,6 @@ void PhaseChaitin::gather_lrg_masks( bool after_aggressive ) {
           // Each entry is reg_pressure_per_value,number_of_regs
           //         RegL  RegI  RegFlags   RegF RegD    INTPRESSURE  FLOATPRESSURE
           // IA32     2     1     1          1    1          6           6
-          // IA64     1     1     1          1    1         50          41
           // SPARC    2     2     2          2    2         48 (24)     52 (26)
           // SPARCV9  2     2     2          2    2         48 (24)     52 (26)
           // AMD64    1     1     1          1    1         14          15

src/hotspot/share/opto/compile.cpp

@@ -4335,6 +4335,9 @@ Compile::TracePhase::TracePhase(PhaseTraceId id)
 
 Compile::TracePhase::~TracePhase() {
   if (_compile->failing_internal()) {
+    if (_log != nullptr) {
+      _log->done("phase");
+    }
     return; // timing code, not stressing bailouts.
   }
 #ifdef ASSERT

src/hotspot/share/opto/generateOptoStub.cpp

@@ -224,10 +224,6 @@ void GraphKit::gen_stub(address C_function,
   store_to_memory(control(), adr_sp, null(), T_ADDRESS, MemNode::unordered);
   // Clear last_Java_pc
   store_to_memory(control(), adr_last_Java_pc, null(), T_ADDRESS, MemNode::unordered);
-#if (defined(IA64) && !defined(AIX))
-  Node* adr_last_Java_fp = basic_plus_adr(top(), thread, in_bytes(JavaThread::last_Java_fp_offset()));
-  store_to_memory(control(), adr_last_Java_fp, null(), T_ADDRESS, MemNode::unordered);
-#endif
 
   // For is-fancy-jump, the C-return value is also the branch target
   Node* target = map()->in(TypeFunc::Parms);