8211387: [Zero] atomic_copy64: Use ldrexd for atomic reads on ARMv7
Reviewed-by: dholmes, aph
diff --git a/hotspot/src/os_cpu/linux_zero/vm/os_linux_zero.hpp b/hotspot/src/os_cpu/linux_zero/vm/os_linux_zero.hpp
index 0f1ad16..e1a80d7 100644
--- a/hotspot/src/os_cpu/linux_zero/vm/os_linux_zero.hpp
+++ b/hotspot/src/os_cpu/linux_zero/vm/os_linux_zero.hpp
@@ -54,6 +54,16 @@
"std %0, %1\n"
: "=&f"(tmp), "=Q"(*(volatile double*)dst)
: "Q"(*(volatile double*)src));
+#elif defined(__ARM_ARCH_7A__)
+ // Note that a ldrexd + clrex combination is only needed for
+ // correctness on the OS level (context-switches). In this
+ // case, clrex *may* be beneficial for performance. For now
+ // don't bother with clrex as this is Zero.
+ jlong tmp;
+ asm volatile ("ldrexd %0, [%1]\n"
+ : "=r"(tmp)
+ : "r"(src), "m"(src));
+ *(jlong *) dst = tmp;
#else
*(jlong *) dst = *(jlong *) src;
#endif
