eal/arm: disable NEON for 32-bit memcpy

The new flag CONFIG_RTE_ARCH_ARM_NEON_MEMCPY is used to enable memcpy
optimizations in EAL.
As it is not always the performance benefit, the feature is disabled.

Signed-off-by: Jan Viktorin <viktorin@rehivetech.com>

config/defconfig_arm-armv7a-linuxapp-gcc

@@ -37,6 +37,10 @@ CONFIG_RTE_ARCH_ARM=y
 CONFIG_RTE_ARCH_ARMv7=y
 CONFIG_RTE_ARCH_ARM_TUNE="cortex-a9"
 
+# Accelerate memcpy operations. Consider enabling for Cortex-A15.
+# For Cortex-A7 and Cortex-A9, It might accelerate short data copies (< 64 B).
+CONFIG_RTE_ARCH_ARM_NEON_MEMCPY=n
+
 CONFIG_RTE_FORCE_INTRINSICS=y
 CONFIG_RTE_ARCH_STRICT_ALIGN=y
 

lib/librte_eal/common/include/arch/arm/rte_memcpy_32.h

@@ -42,7 +42,11 @@ extern "C" {
 
 #include "generic/rte_memcpy.h"
 
-#ifdef __ARM_NEON_FP
+#ifdef RTE_ARCH_ARM_NEON_MEMCPY
+
+#ifndef RTE_MACHINE_CPUFLAG_NEON
+#error "Cannot optimize memcpy by NEON as the CPU seems to not support this"
+#endif
 
 /* ARM NEON Intrinsics are used to copy data */
 #include <arm_neon.h>
@@ -325,7 +329,7 @@ rte_memcpy_func(void *dst, const void *src, size_t n)
 	return memcpy(dst, src, n);
 }
 
-#endif /* __ARM_NEON_FP */
+#endif /* RTE_ARCH_ARM_NEON_MEMCPY */
 
 #ifdef __cplusplus
 }