--- a/hotspot/src/cpu/aarch64/vm/aarch64.ad	Sat Feb 20 15:15:35 2016 +0000
+++ b/hotspot/src/cpu/aarch64/vm/aarch64.ad	Sat Mar 05 22:22:37 2016 +0800
@@ -4683,6 +4683,7 @@
       __ br(Assembler::EQ, cont);
     } else {
       Label retry_load;
+      __ prfm(Address(oop), PSTL1STRM);
       __ bind(retry_load);
       __ ldaxr(tmp, oop);
       __ cmp(tmp, disp_hdr);
@@ -4737,6 +4738,7 @@
         __ cmp(rscratch1, disp_hdr);
       } else {
         Label retry_load, fail;
+        __ prfm(Address(tmp), PSTL1STRM);
         __ bind(retry_load);
         __ ldaxr(rscratch1, tmp);
         __ cmp(disp_hdr, rscratch1);
@@ -4830,6 +4832,7 @@
         __ cmp(tmp, box);
       } else {
         Label retry_load;
+        __ prfm(Address(oop), PSTL1STRM);
         __ bind(retry_load);
         __ ldxr(tmp, oop);
         __ cmp(box, tmp);

--- a/hotspot/src/cpu/aarch64/vm/c1_LIRAssembler_aarch64.cpp	Sat Feb 20 15:15:35 2016 +0000
+++ b/hotspot/src/cpu/aarch64/vm/c1_LIRAssembler_aarch64.cpp	Sat Mar 05 22:22:37 2016 +0800
@@ -1565,6 +1565,7 @@
     Label retry_load, nope;
     // flush and load exclusive from the memory location
     // and fail if it is not what we expect
+    __ prfm(Address(addr), PSTL1STRM);
     __ bind(retry_load);
     __ ldaxrw(rscratch1, addr);
     __ cmpw(rscratch1, cmpval);
@@ -1590,6 +1591,7 @@
     Label retry_load, nope;
     // flush and load exclusive from the memory location
     // and fail if it is not what we expect
+    __ prfm(Address(addr), PSTL1STRM);
     __ bind(retry_load);
     __ ldaxr(rscratch1, addr);
     __ cmp(rscratch1, cmpval);
@@ -3170,6 +3172,7 @@
       }
       Label again;
       __ lea(tmp, addr);
+      __ prfm(Address(tmp), PSTL1STRM);
       __ bind(again);
       (_masm->*lda)(dst, tmp);
       (_masm->*add)(rscratch1, dst, inc);
@@ -3189,6 +3192,7 @@
       assert_different_registers(obj, addr.base(), tmp, rscratch2, dst);
       Label again;
       __ lea(tmp, addr);
+      __ prfm(Address(tmp), PSTL1STRM);
       __ bind(again);
       (_masm->*lda)(dst, tmp);
       (_masm->*stl)(rscratch2, obj, tmp);

--- a/hotspot/src/cpu/aarch64/vm/macroAssembler_aarch64.cpp	Sat Feb 20 15:15:35 2016 +0000
+++ b/hotspot/src/cpu/aarch64/vm/macroAssembler_aarch64.cpp	Sat Mar 05 22:22:37 2016 +0800
@@ -1638,6 +1638,7 @@
 
 void MacroAssembler::atomic_incw(Register counter_addr, Register tmp, Register tmp2) {
   Label retry_load;
+  prfm(Address(counter_addr), PSTL1STRM);
   bind(retry_load);
   // flush and load exclusive from the memory location
   ldxrw(tmp, counter_addr);
@@ -2078,7 +2079,7 @@
     membar(AnyAny);
   } else {
     Label retry_load, nope;
-
+    prfm(Address(addr), PSTL1STRM);
     bind(retry_load);
     // flush and load exclusive from the memory location
     // and fail if it is not what we expect
@@ -2114,7 +2115,7 @@
     membar(AnyAny);
   } else {
     Label retry_load, nope;
-
+    prfm(Address(addr), PSTL1STRM);
     bind(retry_load);
     // flush and load exclusive from the memory location
     // and fail if it is not what we expect
@@ -2149,6 +2150,7 @@
   } else {
     BLOCK_COMMENT("cmpxchg {");
     Label retry_load, done;
+    prfm(Address(addr), PSTL1STRM);
     bind(retry_load);
     load_exclusive(tmp, addr, size, acquire);
     if (size == xword)
@@ -2177,6 +2179,7 @@
     result = different(prev, incr, addr) ? prev : rscratch2;            \
                                                                         \
   Label retry_load;                                                     \
+  prfm(Address(addr), PSTL1STRM);                                       \
   bind(retry_load);                                                     \
   LDXR(result, addr);                                                   \
   OP(rscratch1, result, incr);                                          \
@@ -2199,6 +2202,7 @@
     result = different(prev, newv, addr) ? prev : rscratch2;            \
                                                                         \
   Label retry_load;                                                     \
+  prfm(Address(addr), PSTL1STRM);                                       \
   bind(retry_load);                                                     \
   LDXR(result, addr);                                                   \
   STXR(rscratch1, newv, addr);                                          \

--- a/hotspot/src/cpu/aarch64/vm/templateInterpreterGenerator_aarch64.cpp	Sat Feb 20 15:15:35 2016 +0000
+++ b/hotspot/src/cpu/aarch64/vm/templateInterpreterGenerator_aarch64.cpp	Sat Mar 05 22:22:37 2016 +0800
@@ -1984,6 +1984,7 @@
   __ push(rscratch3);
   Label L;
   __ mov(rscratch2, (address) &BytecodeCounter::_counter_value);
+  __ prfm(Address(rscratch2), PSTL1STRM);
   __ bind(L);
   __ ldxr(rscratch1, rscratch2);
   __ add(rscratch1, rscratch1, 1);