csky: Add C based string functions

Try to access RAM with the largest bit width possible, but without
doing unaligned accesses.

A further improvement could be to use multiple read and writes as the
assembly version was trying to do.

Tested on a BeagleV Starlight with a SiFive U74 core, where the
improvement is noticeable.

Signed-off-by: Matteo Croce <mcroce@microsoft.com>
Co-developed-by: Guo Ren <guoren@linux.alibaba.com>
Signed-off-by: Guo Ren <guoren@linux.alibaba.com>

arch/csky/Kconfig

@@ -321,6 +321,14 @@ config HOTPLUG_CPU
 	  controlled through /sys/devices/system/cpu/cpu1/hotplug/target.
 
 	  Say N if you want to disable CPU hotplug.
+
+config HAVE_EFFICIENT_UNALIGNED_ACCESS
+	bool "Enable EFFICIENT_UNALIGNED_ACCESS for abiv2"
+	depends on CPU_CK807 || CPU_CK810 || CPU_CK860
+	help
+	  Say Y here to enable EFFICIENT_UNALIGNED_ACCESS. Some CPU models could
+	  deal with unaligned access by hardware.
+
 endmenu
 
 source "arch/csky/Kconfig.platforms"

arch/csky/abiv1/Makefile

@@ -4,5 +4,4 @@ obj-y					+= bswapdi.o
 obj-y					+= bswapsi.o
 obj-y					+= cacheflush.o
 obj-y					+= mmap.o
-obj-y					+= memcpy.o
 obj-y					+= strksyms.o

arch/csky/abiv2/Makefile

@@ -2,9 +2,11 @@
 obj-y				+= cacheflush.o
 obj-$(CONFIG_CPU_HAS_FPU)	+= fpu.o
 obj-y				+= memcmp.o
+ifeq ($(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS), y)
 obj-y				+= memcpy.o
 obj-y				+= memmove.o
 obj-y				+= memset.o
+endif
 obj-y				+= strcmp.o
 obj-y				+= strcpy.o
 obj-y				+= strlen.o

arch/csky/lib/Makefile

@@ -1,3 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0-only
 lib-y  := usercopy.o delay.o
 obj-$(CONFIG_FUNCTION_ERROR_INJECTION) += error-inject.o
+ifneq ($(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS), y)
+lib-y  += string.o
+endif