src/cpu/x86/vm/sharedRuntime_x86_32.cpp

*** 39,48 ****
--- 39,49 ----
  #include "c1/c1_Runtime1.hpp"
  #endif
  #ifdef COMPILER2
  #include "opto/runtime.hpp"
  #endif
+ #include "vm_version_x86.hpp"
  
  #define __ masm->
  
  const int StackAlignmentInSlots = StackAlignmentInBytes / VMRegImpl::stack_slot_size;
  
*** 118,129 ****
    int num_xmm_regs = XMMRegisterImpl::number_of_registers;
    int ymm_bytes = num_xmm_regs * 16;
    int zmm_bytes = num_xmm_regs * 32;
  #ifdef COMPILER2
    if (save_vectors) {
!     assert(UseAVX > 0, "up to 512bit vectors are supported with EVEX");
!     assert(MaxVectorSize <= 64, "up to 512bit vectors are supported now");
      // Save upper half of YMM registers
      int vect_bytes = ymm_bytes;
      if (UseAVX > 2) {
        // Save upper half of ZMM registers as well
        vect_bytes += zmm_bytes;
--- 119,130 ----
    int num_xmm_regs = XMMRegisterImpl::number_of_registers;
    int ymm_bytes = num_xmm_regs * 16;
    int zmm_bytes = num_xmm_regs * 32;
  #ifdef COMPILER2
    if (save_vectors) {
!     assert(UseAVX > 0, "Vectors larger than 16 byte long are supported only with AVX");
!     assert(MaxVectorSize <= 64, "Only up to 64 byte long vectors are supported");
      // Save upper half of YMM registers
      int vect_bytes = ymm_bytes;
      if (UseAVX > 2) {
        // Save upper half of ZMM registers as well
        vect_bytes += zmm_bytes;
*** 217,226 ****
--- 218,228 ----
        for (int n = 0; n < num_xmm_regs; n++) {
          __ vextractf64x4_high(Address(rsp, n*32), as_XMMRegister(n));
        }
      }
    }
+   __ vzeroupper();
  
    // Set an oopmap for the call site.  This oopmap will map all
    // oop-registers and debug-info registers as callee-saved.  This
    // will allow deoptimization at this safepoint to find all possible
    // debug-info recordings, as well as let GC find all oops.
*** 267,278 ****
    int zmm_bytes = num_xmm_regs * 32;
    // Recover XMM & FPU state
    int additional_frame_bytes = 0;
  #ifdef COMPILER2
    if (restore_vectors) {
!     assert(UseAVX > 0, "up to 512bit vectors are supported with EVEX");
!     assert(MaxVectorSize <= 64, "up to 512bit vectors are supported now");
      // Save upper half of YMM registers
      additional_frame_bytes = ymm_bytes;
      if (UseAVX > 2) {
        // Save upper half of ZMM registers as well
        additional_frame_bytes += zmm_bytes;
--- 269,280 ----
    int zmm_bytes = num_xmm_regs * 32;
    // Recover XMM & FPU state
    int additional_frame_bytes = 0;
  #ifdef COMPILER2
    if (restore_vectors) {
!     assert(UseAVX > 0, "Vectors larger than 16 byte long are supported only with AVX");
!     assert(MaxVectorSize <= 64, "Only up to 64 byte long vectors are supported");
      // Save upper half of YMM registers
      additional_frame_bytes = ymm_bytes;
      if (UseAVX > 2) {
        // Save upper half of ZMM registers as well
        additional_frame_bytes += zmm_bytes;
*** 283,292 ****
--- 285,296 ----
  #endif
  
    int off = xmm0_off;
    int delta = xmm1_off - off;
  
+   __ vzeroupper();
+ 
    if (UseSSE == 1) {
      // Restore XMM registers
      assert(additional_frame_bytes == 0, "");
      for (int n = 0; n < num_xmm_regs; n++) {
        __ movflt(as_XMMRegister(n), Address(rsp, off*wordSize));
*** 2121,2130 ****
--- 2125,2136 ----
      // and never return here preventing us from clearing _last_native_pc down below.
      // Also can't use call_VM_leaf either as it will check to see if rsi & rdi are
      // preserved and correspond to the bcp/locals pointers. So we do a runtime call
      // by hand.
      //
+     __ vzeroupper();
+ 
      save_native_result(masm, ret_type, stack_slots);
      __ push(thread);
      if (!is_critical_native) {
        __ call(RuntimeAddress(CAST_FROM_FN_PTR(address,
                                                JavaThread::check_special_condition_for_native_trans)));
*** 2305,2315 ****
  
      // END Slow path lock
  
      // BEGIN Slow path unlock
      __ bind(slow_path_unlock);
! 
      // Slow path unlock
  
      if (ret_type == T_FLOAT || ret_type == T_DOUBLE ) {
        save_native_result(masm, ret_type, stack_slots);
      }
--- 2311,2321 ----
  
      // END Slow path lock
  
      // BEGIN Slow path unlock
      __ bind(slow_path_unlock);
!     __ vzeroupper();
      // Slow path unlock
  
      if (ret_type == T_FLOAT || ret_type == T_DOUBLE ) {
        save_native_result(masm, ret_type, stack_slots);
      }
*** 2350,2359 ****
--- 2356,2366 ----
    }
  
    // SLOW PATH Reguard the stack if needed
  
    __ bind(reguard);
+   __ vzeroupper();
    save_native_result(masm, ret_type, stack_slots);
    {
      __ call(RuntimeAddress(CAST_FROM_FN_PTR(address, SharedRuntime::reguard_yellow_pages)));
    }
    restore_native_result(masm, ret_type, stack_slots);