powerpc: POWER7 optimised memcpy using VMX and enhanced prefetch
authorAnton Blanchard <anton@samba.org>
Wed, 30 May 2012 20:22:09 +0000 (20:22 +0000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Tue, 3 Jul 2012 04:14:46 +0000 (14:14 +1000)
Implement a POWER7 optimised memcpy using VMX and enhanced prefetch
instructions.

This is a copy of the POWER7 optimised copy_to_user/copy_from_user
loop. Detailed implementation and performance details can be found in
commit a66086b8197d (powerpc: POWER7 optimised
copy_to_user/copy_from_user using VMX).

I noticed memcpy issues when profiling a RAID6 workload:

.memcpy
.async_memcpy
.async_copy_data
.__raid_run_ops
.handle_stripe
.raid5d
.md_thread

I created a simplified testcase by building a RAID6 array with 4 1GB
ramdisks (booting with brd.rd_size=1048576):

# mdadm -CR -e 1.2 /dev/md0 --level=6 -n4 /dev/ram[0-3]

I then timed how long it took to write to the entire array:

# dd if=/dev/zero of=/dev/md0 bs=1M

Before: 892 MB/s
After:  999 MB/s

A 12% improvement.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/lib/Makefile
arch/powerpc/lib/memcpy_64.S
arch/powerpc/lib/memcpy_power7.S [new file with mode: 0644]

index 873805ec3ed6dcbe0b983907f5c29a12fbb153a1..746e0c895cd77e866938abef1a6dcb1214edc6a4 100644 (file)
@@ -17,7 +17,8 @@ obj-$(CONFIG_HAS_IOMEM)       += devres.o
 obj-$(CONFIG_PPC64)    += copypage_64.o copyuser_64.o \
                           memcpy_64.o usercopy_64.o mem_64.o string.o \
                           checksum_wrappers_64.o hweight_64.o \
-                          copyuser_power7.o string_64.o copypage_power7.o
+                          copyuser_power7.o string_64.o copypage_power7.o \
+                          memcpy_power7.o
 obj-$(CONFIG_XMON)     += sstep.o ldstfp.o
 obj-$(CONFIG_KPROBES)  += sstep.o ldstfp.o
 obj-$(CONFIG_HAVE_HW_BREAKPOINT)       += sstep.o ldstfp.o
index 82fea3963e15d36f8c30577fce53e7bc30196ac3..d2bbbc8d7dc0b045aab751d79264a2f4f959fe61 100644 (file)
 
        .align  7
 _GLOBAL(memcpy)
+BEGIN_FTR_SECTION
        std     r3,48(r1)       /* save destination pointer for return value */
+FTR_SECTION_ELSE
+       b       memcpy_power7
+ALT_FTR_SECTION_END_IFCLR(CPU_FTR_VMX_COPY)
        PPC_MTOCRF(0x01,r5)
        cmpldi  cr1,r5,16
        neg     r6,r3           # LS 3 bits = # bytes to 8-byte dest bdry
diff --git a/arch/powerpc/lib/memcpy_power7.S b/arch/powerpc/lib/memcpy_power7.S
new file mode 100644 (file)
index 0000000..84674d8
--- /dev/null
@@ -0,0 +1,650 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright (C) IBM Corporation, 2012
+ *
+ * Author: Anton Blanchard <anton@au.ibm.com>
+ */
+#include <asm/ppc_asm.h>
+
+#define STACKFRAMESIZE 256
+#define STK_REG(i)     (112 + ((i)-14)*8)
+
+_GLOBAL(memcpy_power7)
+#ifdef CONFIG_ALTIVEC
+       cmpldi  r5,16
+       cmpldi  cr1,r5,4096
+
+       std     r3,48(r1)
+
+       blt     .Lshort_copy
+       bgt     cr1,.Lvmx_copy
+#else
+       cmpldi  r5,16
+
+       std     r3,48(r1)
+
+       blt     .Lshort_copy
+#endif
+
+.Lnonvmx_copy:
+       /* Get the source 8B aligned */
+       neg     r6,r4
+       mtocrf  0x01,r6
+       clrldi  r6,r6,(64-3)
+
+       bf      cr7*4+3,1f
+       lbz     r0,0(r4)
+       addi    r4,r4,1
+       stb     r0,0(r3)
+       addi    r3,r3,1
+
+1:     bf      cr7*4+2,2f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+2:     bf      cr7*4+1,3f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+3:     sub     r5,r5,r6
+       cmpldi  r5,128
+       blt     5f
+
+       mflr    r0
+       stdu    r1,-STACKFRAMESIZE(r1)
+       std     r14,STK_REG(r14)(r1)
+       std     r15,STK_REG(r15)(r1)
+       std     r16,STK_REG(r16)(r1)
+       std     r17,STK_REG(r17)(r1)
+       std     r18,STK_REG(r18)(r1)
+       std     r19,STK_REG(r19)(r1)
+       std     r20,STK_REG(r20)(r1)
+       std     r21,STK_REG(r21)(r1)
+       std     r22,STK_REG(r22)(r1)
+       std     r0,STACKFRAMESIZE+16(r1)
+
+       srdi    r6,r5,7
+       mtctr   r6
+
+       /* Now do cacheline (128B) sized loads and stores. */
+       .align  5
+4:
+       ld      r0,0(r4)
+       ld      r6,8(r4)
+       ld      r7,16(r4)
+       ld      r8,24(r4)
+       ld      r9,32(r4)
+       ld      r10,40(r4)
+       ld      r11,48(r4)
+       ld      r12,56(r4)
+       ld      r14,64(r4)
+       ld      r15,72(r4)
+       ld      r16,80(r4)
+       ld      r17,88(r4)
+       ld      r18,96(r4)
+       ld      r19,104(r4)
+       ld      r20,112(r4)
+       ld      r21,120(r4)
+       addi    r4,r4,128
+       std     r0,0(r3)
+       std     r6,8(r3)
+       std     r7,16(r3)
+       std     r8,24(r3)
+       std     r9,32(r3)
+       std     r10,40(r3)
+       std     r11,48(r3)
+       std     r12,56(r3)
+       std     r14,64(r3)
+       std     r15,72(r3)
+       std     r16,80(r3)
+       std     r17,88(r3)
+       std     r18,96(r3)
+       std     r19,104(r3)
+       std     r20,112(r3)
+       std     r21,120(r3)
+       addi    r3,r3,128
+       bdnz    4b
+
+       clrldi  r5,r5,(64-7)
+
+       ld      r14,STK_REG(r14)(r1)
+       ld      r15,STK_REG(r15)(r1)
+       ld      r16,STK_REG(r16)(r1)
+       ld      r17,STK_REG(r17)(r1)
+       ld      r18,STK_REG(r18)(r1)
+       ld      r19,STK_REG(r19)(r1)
+       ld      r20,STK_REG(r20)(r1)
+       ld      r21,STK_REG(r21)(r1)
+       ld      r22,STK_REG(r22)(r1)
+       addi    r1,r1,STACKFRAMESIZE
+
+       /* Up to 127B to go */
+5:     srdi    r6,r5,4
+       mtocrf  0x01,r6
+
+6:     bf      cr7*4+1,7f
+       ld      r0,0(r4)
+       ld      r6,8(r4)
+       ld      r7,16(r4)
+       ld      r8,24(r4)
+       ld      r9,32(r4)
+       ld      r10,40(r4)
+       ld      r11,48(r4)
+       ld      r12,56(r4)
+       addi    r4,r4,64
+       std     r0,0(r3)
+       std     r6,8(r3)
+       std     r7,16(r3)
+       std     r8,24(r3)
+       std     r9,32(r3)
+       std     r10,40(r3)
+       std     r11,48(r3)
+       std     r12,56(r3)
+       addi    r3,r3,64
+
+       /* Up to 63B to go */
+7:     bf      cr7*4+2,8f
+       ld      r0,0(r4)
+       ld      r6,8(r4)
+       ld      r7,16(r4)
+       ld      r8,24(r4)
+       addi    r4,r4,32
+       std     r0,0(r3)
+       std     r6,8(r3)
+       std     r7,16(r3)
+       std     r8,24(r3)
+       addi    r3,r3,32
+
+       /* Up to 31B to go */
+8:     bf      cr7*4+3,9f
+       ld      r0,0(r4)
+       ld      r6,8(r4)
+       addi    r4,r4,16
+       std     r0,0(r3)
+       std     r6,8(r3)
+       addi    r3,r3,16
+
+9:     clrldi  r5,r5,(64-4)
+
+       /* Up to 15B to go */
+.Lshort_copy:
+       mtocrf  0x01,r5
+       bf      cr7*4+0,12f
+       lwz     r0,0(r4)        /* Less chance of a reject with word ops */
+       lwz     r6,4(r4)
+       addi    r4,r4,8
+       stw     r0,0(r3)
+       stw     r6,4(r3)
+       addi    r3,r3,8
+
+12:    bf      cr7*4+1,13f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+13:    bf      cr7*4+2,14f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+14:    bf      cr7*4+3,15f
+       lbz     r0,0(r4)
+       stb     r0,0(r3)
+
+15:    ld      r3,48(r1)
+       blr
+
+.Lunwind_stack_nonvmx_copy:
+       addi    r1,r1,STACKFRAMESIZE
+       b       .Lnonvmx_copy
+
+#ifdef CONFIG_ALTIVEC
+.Lvmx_copy:
+       mflr    r0
+       std     r4,56(r1)
+       std     r5,64(r1)
+       std     r0,16(r1)
+       stdu    r1,-STACKFRAMESIZE(r1)
+       bl      .enter_vmx_copy
+       cmpwi   r3,0
+       ld      r0,STACKFRAMESIZE+16(r1)
+       ld      r3,STACKFRAMESIZE+48(r1)
+       ld      r4,STACKFRAMESIZE+56(r1)
+       ld      r5,STACKFRAMESIZE+64(r1)
+       mtlr    r0
+
+       /*
+        * We prefetch both the source and destination using enhanced touch
+        * instructions. We use a stream ID of 0 for the load side and
+        * 1 for the store side.
+        */
+       clrrdi  r6,r4,7
+       clrrdi  r9,r3,7
+       ori     r9,r9,1         /* stream=1 */
+
+       srdi    r7,r5,7         /* length in cachelines, capped at 0x3FF */
+       cmpldi  cr1,r7,0x3FF
+       ble     cr1,1f
+       li      r7,0x3FF
+1:     lis     r0,0x0E00       /* depth=7 */
+       sldi    r7,r7,7
+       or      r7,r7,r0
+       ori     r10,r7,1        /* stream=1 */
+
+       lis     r8,0x8000       /* GO=1 */
+       clrldi  r8,r8,32
+
+.machine push
+.machine "power4"
+       dcbt    r0,r6,0b01000
+       dcbt    r0,r7,0b01010
+       dcbtst  r0,r9,0b01000
+       dcbtst  r0,r10,0b01010
+       eieio
+       dcbt    r0,r8,0b01010   /* GO */
+.machine pop
+
+       beq     .Lunwind_stack_nonvmx_copy
+
+       /*
+        * If source and destination are not relatively aligned we use a
+        * slower permute loop.
+        */
+       xor     r6,r4,r3
+       rldicl. r6,r6,0,(64-4)
+       bne     .Lvmx_unaligned_copy
+
+       /* Get the destination 16B aligned */
+       neg     r6,r3
+       mtocrf  0x01,r6
+       clrldi  r6,r6,(64-4)
+
+       bf      cr7*4+3,1f
+       lbz     r0,0(r4)
+       addi    r4,r4,1
+       stb     r0,0(r3)
+       addi    r3,r3,1
+
+1:     bf      cr7*4+2,2f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+2:     bf      cr7*4+1,3f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+3:     bf      cr7*4+0,4f
+       ld      r0,0(r4)
+       addi    r4,r4,8
+       std     r0,0(r3)
+       addi    r3,r3,8
+
+4:     sub     r5,r5,r6
+
+       /* Get the desination 128B aligned */
+       neg     r6,r3
+       srdi    r7,r6,4
+       mtocrf  0x01,r7
+       clrldi  r6,r6,(64-7)
+
+       li      r9,16
+       li      r10,32
+       li      r11,48
+
+       bf      cr7*4+3,5f
+       lvx     vr1,r0,r4
+       addi    r4,r4,16
+       stvx    vr1,r0,r3
+       addi    r3,r3,16
+
+5:     bf      cr7*4+2,6f
+       lvx     vr1,r0,r4
+       lvx     vr0,r4,r9
+       addi    r4,r4,32
+       stvx    vr1,r0,r3
+       stvx    vr0,r3,r9
+       addi    r3,r3,32
+
+6:     bf      cr7*4+1,7f
+       lvx     vr3,r0,r4
+       lvx     vr2,r4,r9
+       lvx     vr1,r4,r10
+       lvx     vr0,r4,r11
+       addi    r4,r4,64
+       stvx    vr3,r0,r3
+       stvx    vr2,r3,r9
+       stvx    vr1,r3,r10
+       stvx    vr0,r3,r11
+       addi    r3,r3,64
+
+7:     sub     r5,r5,r6
+       srdi    r6,r5,7
+
+       std     r14,STK_REG(r14)(r1)
+       std     r15,STK_REG(r15)(r1)
+       std     r16,STK_REG(r16)(r1)
+
+       li      r12,64
+       li      r14,80
+       li      r15,96
+       li      r16,112
+
+       mtctr   r6
+
+       /*
+        * Now do cacheline sized loads and stores. By this stage the
+        * cacheline stores are also cacheline aligned.
+        */
+       .align  5
+8:
+       lvx     vr7,r0,r4
+       lvx     vr6,r4,r9
+       lvx     vr5,r4,r10
+       lvx     vr4,r4,r11
+       lvx     vr3,r4,r12
+       lvx     vr2,r4,r14
+       lvx     vr1,r4,r15
+       lvx     vr0,r4,r16
+       addi    r4,r4,128
+       stvx    vr7,r0,r3
+       stvx    vr6,r3,r9
+       stvx    vr5,r3,r10
+       stvx    vr4,r3,r11
+       stvx    vr3,r3,r12
+       stvx    vr2,r3,r14
+       stvx    vr1,r3,r15
+       stvx    vr0,r3,r16
+       addi    r3,r3,128
+       bdnz    8b
+
+       ld      r14,STK_REG(r14)(r1)
+       ld      r15,STK_REG(r15)(r1)
+       ld      r16,STK_REG(r16)(r1)
+
+       /* Up to 127B to go */
+       clrldi  r5,r5,(64-7)
+       srdi    r6,r5,4
+       mtocrf  0x01,r6
+
+       bf      cr7*4+1,9f
+       lvx     vr3,r0,r4
+       lvx     vr2,r4,r9
+       lvx     vr1,r4,r10
+       lvx     vr0,r4,r11
+       addi    r4,r4,64
+       stvx    vr3,r0,r3
+       stvx    vr2,r3,r9
+       stvx    vr1,r3,r10
+       stvx    vr0,r3,r11
+       addi    r3,r3,64
+
+9:     bf      cr7*4+2,10f
+       lvx     vr1,r0,r4
+       lvx     vr0,r4,r9
+       addi    r4,r4,32
+       stvx    vr1,r0,r3
+       stvx    vr0,r3,r9
+       addi    r3,r3,32
+
+10:    bf      cr7*4+3,11f
+       lvx     vr1,r0,r4
+       addi    r4,r4,16
+       stvx    vr1,r0,r3
+       addi    r3,r3,16
+
+       /* Up to 15B to go */
+11:    clrldi  r5,r5,(64-4)
+       mtocrf  0x01,r5
+       bf      cr7*4+0,12f
+       ld      r0,0(r4)
+       addi    r4,r4,8
+       std     r0,0(r3)
+       addi    r3,r3,8
+
+12:    bf      cr7*4+1,13f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+13:    bf      cr7*4+2,14f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+14:    bf      cr7*4+3,15f
+       lbz     r0,0(r4)
+       stb     r0,0(r3)
+
+15:    addi    r1,r1,STACKFRAMESIZE
+       ld      r3,48(r1)
+       b       .exit_vmx_copy          /* tail call optimise */
+
+.Lvmx_unaligned_copy:
+       /* Get the destination 16B aligned */
+       neg     r6,r3
+       mtocrf  0x01,r6
+       clrldi  r6,r6,(64-4)
+
+       bf      cr7*4+3,1f
+       lbz     r0,0(r4)
+       addi    r4,r4,1
+       stb     r0,0(r3)
+       addi    r3,r3,1
+
+1:     bf      cr7*4+2,2f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+2:     bf      cr7*4+1,3f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+3:     bf      cr7*4+0,4f
+       lwz     r0,0(r4)        /* Less chance of a reject with word ops */
+       lwz     r7,4(r4)
+       addi    r4,r4,8
+       stw     r0,0(r3)
+       stw     r7,4(r3)
+       addi    r3,r3,8
+
+4:     sub     r5,r5,r6
+
+       /* Get the desination 128B aligned */
+       neg     r6,r3
+       srdi    r7,r6,4
+       mtocrf  0x01,r7
+       clrldi  r6,r6,(64-7)
+
+       li      r9,16
+       li      r10,32
+       li      r11,48
+
+       lvsl    vr16,0,r4       /* Setup permute control vector */
+       lvx     vr0,0,r4
+       addi    r4,r4,16
+
+       bf      cr7*4+3,5f
+       lvx     vr1,r0,r4
+       vperm   vr8,vr0,vr1,vr16
+       addi    r4,r4,16
+       stvx    vr8,r0,r3
+       addi    r3,r3,16
+       vor     vr0,vr1,vr1
+
+5:     bf      cr7*4+2,6f
+       lvx     vr1,r0,r4
+       vperm   vr8,vr0,vr1,vr16
+       lvx     vr0,r4,r9
+       vperm   vr9,vr1,vr0,vr16
+       addi    r4,r4,32
+       stvx    vr8,r0,r3
+       stvx    vr9,r3,r9
+       addi    r3,r3,32
+
+6:     bf      cr7*4+1,7f
+       lvx     vr3,r0,r4
+       vperm   vr8,vr0,vr3,vr16
+       lvx     vr2,r4,r9
+       vperm   vr9,vr3,vr2,vr16
+       lvx     vr1,r4,r10
+       vperm   vr10,vr2,vr1,vr16
+       lvx     vr0,r4,r11
+       vperm   vr11,vr1,vr0,vr16
+       addi    r4,r4,64
+       stvx    vr8,r0,r3
+       stvx    vr9,r3,r9
+       stvx    vr10,r3,r10
+       stvx    vr11,r3,r11
+       addi    r3,r3,64
+
+7:     sub     r5,r5,r6
+       srdi    r6,r5,7
+
+       std     r14,STK_REG(r14)(r1)
+       std     r15,STK_REG(r15)(r1)
+       std     r16,STK_REG(r16)(r1)
+
+       li      r12,64
+       li      r14,80
+       li      r15,96
+       li      r16,112
+
+       mtctr   r6
+
+       /*
+        * Now do cacheline sized loads and stores. By this stage the
+        * cacheline stores are also cacheline aligned.
+        */
+       .align  5
+8:
+       lvx     vr7,r0,r4
+       vperm   vr8,vr0,vr7,vr16
+       lvx     vr6,r4,r9
+       vperm   vr9,vr7,vr6,vr16
+       lvx     vr5,r4,r10
+       vperm   vr10,vr6,vr5,vr16
+       lvx     vr4,r4,r11
+       vperm   vr11,vr5,vr4,vr16
+       lvx     vr3,r4,r12
+       vperm   vr12,vr4,vr3,vr16
+       lvx     vr2,r4,r14
+       vperm   vr13,vr3,vr2,vr16
+       lvx     vr1,r4,r15
+       vperm   vr14,vr2,vr1,vr16
+       lvx     vr0,r4,r16
+       vperm   vr15,vr1,vr0,vr16
+       addi    r4,r4,128
+       stvx    vr8,r0,r3
+       stvx    vr9,r3,r9
+       stvx    vr10,r3,r10
+       stvx    vr11,r3,r11
+       stvx    vr12,r3,r12
+       stvx    vr13,r3,r14
+       stvx    vr14,r3,r15
+       stvx    vr15,r3,r16
+       addi    r3,r3,128
+       bdnz    8b
+
+       ld      r14,STK_REG(r14)(r1)
+       ld      r15,STK_REG(r15)(r1)
+       ld      r16,STK_REG(r16)(r1)
+
+       /* Up to 127B to go */
+       clrldi  r5,r5,(64-7)
+       srdi    r6,r5,4
+       mtocrf  0x01,r6
+
+       bf      cr7*4+1,9f
+       lvx     vr3,r0,r4
+       vperm   vr8,vr0,vr3,vr16
+       lvx     vr2,r4,r9
+       vperm   vr9,vr3,vr2,vr16
+       lvx     vr1,r4,r10
+       vperm   vr10,vr2,vr1,vr16
+       lvx     vr0,r4,r11
+       vperm   vr11,vr1,vr0,vr16
+       addi    r4,r4,64
+       stvx    vr8,r0,r3
+       stvx    vr9,r3,r9
+       stvx    vr10,r3,r10
+       stvx    vr11,r3,r11
+       addi    r3,r3,64
+
+9:     bf      cr7*4+2,10f
+       lvx     vr1,r0,r4
+       vperm   vr8,vr0,vr1,vr16
+       lvx     vr0,r4,r9
+       vperm   vr9,vr1,vr0,vr16
+       addi    r4,r4,32
+       stvx    vr8,r0,r3
+       stvx    vr9,r3,r9
+       addi    r3,r3,32
+
+10:    bf      cr7*4+3,11f
+       lvx     vr1,r0,r4
+       vperm   vr8,vr0,vr1,vr16
+       addi    r4,r4,16
+       stvx    vr8,r0,r3
+       addi    r3,r3,16
+
+       /* Up to 15B to go */
+11:    clrldi  r5,r5,(64-4)
+       addi    r4,r4,-16       /* Unwind the +16 load offset */
+       mtocrf  0x01,r5
+       bf      cr7*4+0,12f
+       lwz     r0,0(r4)        /* Less chance of a reject with word ops */
+       lwz     r6,4(r4)
+       addi    r4,r4,8
+       stw     r0,0(r3)
+       stw     r6,4(r3)
+       addi    r3,r3,8
+
+12:    bf      cr7*4+1,13f
+       lwz     r0,0(r4)
+       addi    r4,r4,4
+       stw     r0,0(r3)
+       addi    r3,r3,4
+
+13:    bf      cr7*4+2,14f
+       lhz     r0,0(r4)
+       addi    r4,r4,2
+       sth     r0,0(r3)
+       addi    r3,r3,2
+
+14:    bf      cr7*4+3,15f
+       lbz     r0,0(r4)
+       stb     r0,0(r3)
+
+15:    addi    r1,r1,STACKFRAMESIZE
+       ld      r3,48(r1)
+       b       .exit_vmx_copy          /* tail call optimise */
+#endif /* CONFiG_ALTIVEC */