diff --git a/arch/arm/cpu/armv8/cache.S b/arch/arm/cpu/armv8/cache.S
new file mode 100644
index 0000000..546a83e
--- /dev/null
+++ b/arch/arm/cpu/armv8/cache.S
@@ -0,0 +1,136 @@
+/*
+ * (C) Copyright 2013
+ * David Feng <fenghua@phytium.com.cn>
+ *
+ * This file is based on sample code from ARMv8 ARM.
+ *
+ * SPDX-License-Identifier:	GPL-2.0+
+ */
+
+#include <asm-offsets.h>
+#include <config.h>
+#include <version.h>
+#include <asm/macro.h>
+#include <linux/linkage.h>
+
+/*
+ * void __asm_flush_dcache_level(level)
+ *
+ * clean and invalidate one level cache.
+ *
+ * x0: cache level
+ * x1~x9: clobbered
+ */
+ENTRY(__asm_flush_dcache_level)
+	lsl	x1, x0, #1
+	msr	csselr_el1, x1		/* select cache level */
+	isb				/* sync change of cssidr_el1 */
+	mrs	x6, ccsidr_el1		/* read the new cssidr_el1 */
+	and	x2, x6, #7		/* x2 <- log2(cache line size)-4 */
+	add	x2, x2, #4		/* x2 <- log2(cache line size) */
+	mov	x3, #0x3ff
+	and	x3, x3, x6, lsr #3	/* x3 <- max number of #ways */
+	add	w4, w3, w3
+	sub	w4, w4, 1		/* round up log2(#ways + 1) */
+	clz	w5, w4			/* bit position of #ways */
+	mov	x4, #0x7fff
+	and	x4, x4, x6, lsr #13	/* x4 <- max number of #sets */
+	/* x1 <- cache level << 1 */
+	/* x2 <- line length offset */
+	/* x3 <- number of cache ways - 1 */
+	/* x4 <- number of cache sets - 1 */
+	/* x5 <- bit position of #ways */
+
+loop_set:
+	mov	x6, x3			/* x6 <- working copy of #ways */
+loop_way:
+	lsl	x7, x6, x5
+	orr	x9, x1, x7		/* map way and level to cisw value */
+	lsl	x7, x4, x2
+	orr	x9, x9, x7		/* map set number to cisw value */
+	dc	cisw, x9		/* clean & invalidate by set/way */
+	subs	x6, x6, #1		/* decrement the way */
+	b.ge	loop_way
+	subs	x4, x4, #1		/* decrement the set */
+	b.ge	loop_set
+
+	ret
+ENDPROC(__asm_flush_dcache_level)
+
+/*
+ * void __asm_flush_dcache_all(void)
+ *
+ * clean and invalidate all data cache by SET/WAY.
+ */
+ENTRY(__asm_flush_dcache_all)
+	dsb	sy
+	mrs	x10, clidr_el1		/* read clidr_el1 */
+	lsr	x11, x10, #24
+	and	x11, x11, #0x7		/* x11 <- loc */
+	cbz	x11, finished		/* if loc is 0, exit */
+	mov	x15, lr
+	mov	x0, #0			/* start flush at cache level 0 */
+	/* x0  <- cache level */
+	/* x10 <- clidr_el1 */
+	/* x11 <- loc */
+	/* x15 <- return address */
+
+loop_level:
+	lsl	x1, x0, #1
+	add	x1, x1, x0		/* x0 <- tripled cache level */
+	lsr	x1, x10, x1
+	and	x1, x1, #7		/* x1 <- cache type */
+	cmp	x1, #2
+	b.lt	skip			/* skip if no cache or icache */
+	bl	__asm_flush_dcache_level
+skip:
+	add	x0, x0, #1		/* increment cache level */
+	cmp	x11, x0
+	b.gt	loop_level
+
+	mov	x0, #0
+	msr	csselr_el1, x0		/* resotre csselr_el1 */
+	dsb	sy
+	isb
+	mov	lr, x15
+
+finished:
+	ret
+ENDPROC(__asm_flush_dcache_all)
+
+/*
+ * void __asm_flush_dcache_range(start, end)
+ *
+ * clean & invalidate data cache in the range
+ *
+ * x0: start address
+ * x1: end address
+ */
+ENTRY(__asm_flush_dcache_range)
+	mrs	x3, ctr_el0
+	lsr	x3, x3, #16
+	and	x3, x3, #0xf
+	mov	x2, #4
+	lsl	x2, x2, x3		/* cache line size */
+
+	/* x2 <- minimal cache line size in cache system */
+	sub	x3, x2, #1
+	bic	x0, x0, x3
+1:	dc	civac, x0	/* clean & invalidate data or unified cache */
+	add	x0, x0, x2
+	cmp	x0, x1
+	b.lo	1b
+	dsb	sy
+	ret
+ENDPROC(__asm_flush_dcache_range)
+
+/*
+ * void __asm_invalidate_icache_all(void)
+ *
+ * invalidate all tlb entries.
+ */
+ENTRY(__asm_invalidate_icache_all)
+	ic	ialluis
+	isb	sy
+	ret
+ENDPROC(__asm_invalidate_icache_all)