--- old/src/hotspot/cpu/x86/c2_MacroAssembler_x86.hpp	2020-04-02 18:03:44.944854099 -0700
+++ new/src/hotspot/cpu/x86/c2_MacroAssembler_x86.hpp	2020-04-02 18:03:44.784854099 -0700
@@ -28,6 +28,8 @@
 // C2_MacroAssembler contains high-level macros for C2
 
 public:
+  Assembler::AvxVectorLen vector_length_encoding(int vlen_in_bytes);
+
   // special instructions for EVEX
   void setvectmask(Register dst, Register src);
   void restorevectmask();
@@ -71,18 +73,61 @@
   void vabsnegd(int opcode, XMMRegister dst, XMMRegister src, int vector_len, Register scr);
   void vabsnegf(int opcode, XMMRegister dst, XMMRegister src, Register scr);
   void vabsnegf(int opcode, XMMRegister dst, XMMRegister src, int vector_len, Register scr);
+
+  void pminmax(int opcode, BasicType elem_bt, XMMRegister dst, XMMRegister src,
+               XMMRegister tmp = xnoreg);
+  void vpminmax(int opcode, BasicType elem_bt,
+                XMMRegister dst, XMMRegister src1, XMMRegister src2,
+                int vlen_enc);
+
+  void vminmax_fp(int opcode, BasicType elem_bt,
+                  XMMRegister dst, XMMRegister a, XMMRegister b,
+                  XMMRegister tmp, XMMRegister atmp, XMMRegister btmp,
+                  int vlen_enc);
+  void evminmax_fp(int opcode, BasicType elem_bt,
+                   XMMRegister dst, XMMRegister a, XMMRegister b,
+                   KRegister ktmp, XMMRegister atmp, XMMRegister btmp,
+                   int vlen_enc);
+
   void vextendbw(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
   void vextendbw(bool sign, XMMRegister dst, XMMRegister src);
-  void vshiftd(int opcode, XMMRegister dst, XMMRegister src);
-  void vshiftd(int opcode, XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
-  void vshiftw(int opcode, XMMRegister dst, XMMRegister src);
-  void vshiftw(int opcode, XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
-  void vshiftq(int opcode, XMMRegister dst, XMMRegister src);
-  void vshiftq(int opcode, XMMRegister dst, XMMRegister nds, XMMRegister src, int vector_len);
+  void vextendbd(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
+  void vextendwd(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
 
-  // Reductions for vectors of ints, longs, floats, and doubles.
+  void vshiftd(int opcode, XMMRegister dst, XMMRegister shift);
+  void vshiftd(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
+  void vshiftw(int opcode, XMMRegister dst, XMMRegister shift);
+  void vshiftw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
+  void vshiftq(int opcode, XMMRegister dst, XMMRegister shift);
+  void vshiftq(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
+  void varshiftd(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
+  void varshiftw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
+  void varshiftq(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc, XMMRegister vtmp = xnoreg);
+  void varshiftbw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len, XMMRegister vtmp, Register scratch);
+  void evarshiftb(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len, XMMRegister vtmp, Register scratch);
+
+  void insert(BasicType typ, XMMRegister dst, Register val, int idx);
+  void vinsert(BasicType typ, XMMRegister dst, XMMRegister src, Register val, int idx);
+  void vgather(BasicType typ, XMMRegister dst, Register base, XMMRegister idx, XMMRegister mask, int vector_len);
+  void evgather(BasicType typ, XMMRegister dst, KRegister mask, Register base, XMMRegister idx, int vector_len);
+  void evscatter(BasicType typ, Register base, XMMRegister idx, KRegister mask, XMMRegister src, int vector_len);
+
+  // extract
+  void extract(BasicType typ, Register dst, XMMRegister src, int idx);
+  XMMRegister get_lane(BasicType typ, XMMRegister dst, XMMRegister src, int elemindex);
+  void get_elem(BasicType typ, Register dst, XMMRegister src, int elemindex);
+  void get_elem(BasicType typ, XMMRegister dst, XMMRegister src, int elemindex, Register tmp = noreg, XMMRegister vtmp = xnoreg);
+
+  // blend
+  void evpcmp(BasicType typ, KRegister kdmask, KRegister ksmask, XMMRegister src1, AddressLiteral adr, int comparison, int vector_len, Register scratch = rscratch1);
+  void evpblend(BasicType typ, XMMRegister dst, KRegister kmask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
 
-  // dst = src1 + reduce(op, src2) using vtmp as temps
+  void load_vector_mask(XMMRegister dst, XMMRegister src, int vlen_in_bytes, BasicType elem_bt);
+  void load_iota_indices(XMMRegister dst, Register scratch, int vlen_in_bytes);
+
+  // Reductions for vectors of bytes, shorts, ints, longs, floats, and doubles.
+
+  // dst = src1  reduce(op, src2) using vtmp as temps
   void reduceI(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
 #ifdef _LP64
   void reduceL(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
@@ -92,32 +137,62 @@
   void reduce_fp(int opcode, int vlen,
                  XMMRegister dst, XMMRegister src,
                  XMMRegister vtmp1, XMMRegister vtmp2 = xnoreg);
+  void reduceB(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void mulreduceB(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduceS(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduceFloatMinMax(int opcode, int vlen, bool is_dst_valid,
+                         XMMRegister dst, XMMRegister src,
+                         XMMRegister tmp, XMMRegister atmp, XMMRegister btmp, XMMRegister xmm_0, XMMRegister xmm_1 = xnoreg);
+  void reduceDoubleMinMax(int opcode, int vlen, bool is_dst_valid,
+                          XMMRegister dst, XMMRegister src,
+                          XMMRegister tmp, XMMRegister atmp, XMMRegister btmp, XMMRegister xmm_0, XMMRegister xmm_1 = xnoreg);
  private:
   void reduceF(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduceD(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
 
+  // Int Reduction
   void reduce2I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce4I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce8I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce16I(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
 
+  // Byte Reduction
+  void reduce8B (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce16B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce32B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce64B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void mulreduce8B (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void mulreduce16B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void mulreduce32B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void mulreduce64B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+
+  // Short Reduction
+  void reduce4S (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce8S (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce16S(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+  void reduce32S(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
+
+  // Long Reduction
 #ifdef _LP64
   void reduce2L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce4L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce8L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
 #endif // _LP64
 
+  // Float Reduction
   void reduce2F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
   void reduce4F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
   void reduce8F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce16F(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
 
+  // Double Reduction
   void reduce2D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
   void reduce4D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
   void reduce8D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
 
-  void reduce_operation_128(int opcode, XMMRegister dst, XMMRegister src);
-  void reduce_operation_256(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2);
+  // Base reduction instruction
+  void reduce_operation_128(BasicType typ, int opcode, XMMRegister dst, XMMRegister src);
+  void reduce_operation_256(BasicType typ, int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2);
 
  public: