[ARM] Newer binutils want -mcpu=xscale not -mxscale

arch/arm/Makefile

@@ -34,7 +34,7 @@ comma = ,
 # Note that GCC does not numerically define an architecture version
 # macro, but instead defines a whole series of macros which makes
 # testing for a specific architecture or later rather impossible.
-arch-$(CONFIG_CPU_32v5)		:=-D__LINUX_ARM_ARCH__=5 $(call check_gcc,-march=armv5te,-march=armv4 -Wa$(comma)-mxscale)
+arch-$(CONFIG_CPU_32v5)		:=-D__LINUX_ARM_ARCH__=5 $(call check_gcc,-march=armv5te,-march=armv4)
 arch-$(CONFIG_CPU_32v4)		:=-D__LINUX_ARM_ARCH__=4 -march=armv4
 arch-$(CONFIG_CPU_32v3)		:=-D__LINUX_ARM_ARCH__=3 -march=armv3
 
@@ -47,8 +47,9 @@ tune-$(CONFIG_CPU_ARM922T)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_ARM926T)	:=-mtune=arm9tdmi
 tune-$(CONFIG_CPU_SA110)	:=-mtune=strongarm110
 tune-$(CONFIG_CPU_SA1100)	:=-mtune=strongarm1100
-tune-$(CONFIG_CPU_XSCALE)	:=$(call check_gcc,-mtune=xscale,-mtune=strongarm110)
+tune-$(CONFIG_CPU_XSCALE)	:=$(call check_gcc,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
 
+# Need -Uarm for gcc < 3.x
 CFLAGS_BOOT	:=-mapcs-32 $(arch-y) $(tune-y) -mshort-load-bytes -msoft-float -Wa,-mno-fpu -Uarm
 CFLAGS		+=-mapcs-32 $(arch-y) $(tune-y) -mshort-load-bytes -msoft-float -Wa,-mno-fpu -Uarm
 AFLAGS		+=-mapcs-32 $(arch-y) $(tune-y) -msoft-float -Wa,-mno-fpu