Break bionic implementations into arch versions. DO NOT MERGE
Move arch specific code for arm, mips, x86 into separate
makefiles.
In addition, add different arm cpu versions of memcpy/memset.
Bug: 8005082
(cherry picked from commit acdde8c1cf8e8beed98c052757d96695b820b50c)
Change-Id: I0108d432af9f6283ae99adfc92a3399e5ab3e31d
diff --git a/libc/arch-arm/cortex-a9/bionic/memset.S b/libc/arch-arm/cortex-a9/bionic/memset.S
new file mode 100644
index 0000000..b58aa45
--- /dev/null
+++ b/libc/arch-arm/cortex-a9/bionic/memset.S
@@ -0,0 +1,152 @@
+/*
+ * Copyright (C) 2008 The Android Open Source Project
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+ * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
+ * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
+ * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+ * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
+ * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#include <machine/cpu-features.h>
+#include <machine/asm.h>
+
+/*
+ * This code assumes it is running on a processor that supports all arm v7
+ * instructions and that supports neon instructions.
+ */
+
+ .fpu neon
+
+ENTRY(bzero)
+ mov r2, r1
+ mov r1, #0
+END(bzero)
+
+/* memset() returns its first argument. */
+ENTRY(memset)
+ # The neon memset only wins for less than 132.
+ cmp r2, #132
+ bhi 11f
+
+ .save {r0}
+ stmfd sp!, {r0}
+
+ vdup.8 q0, r1
+
+ /* make sure we have at least 32 bytes to write */
+ subs r2, r2, #32
+ blo 2f
+ vmov q1, q0
+
+1: /* The main loop writes 32 bytes at a time */
+ subs r2, r2, #32
+ vst1.8 {d0 - d3}, [r0]!
+ bhs 1b
+
+2: /* less than 32 left */
+ add r2, r2, #32
+ tst r2, #0x10
+ beq 3f
+
+ // writes 16 bytes, 128-bits aligned
+ vst1.8 {d0, d1}, [r0]!
+3: /* write up to 15-bytes (count in r2) */
+ movs ip, r2, lsl #29
+ bcc 1f
+ vst1.8 {d0}, [r0]!
+1: bge 2f
+ vst1.32 {d0[0]}, [r0]!
+2: movs ip, r2, lsl #31
+ strmib r1, [r0], #1
+ strcsb r1, [r0], #1
+ strcsb r1, [r0], #1
+ ldmfd sp!, {r0}
+ bx lr
+11:
+ /* compute the offset to align the destination
+ * offset = (4-(src&3))&3 = -src & 3
+ */
+
+ .save {r0, r4-r7, lr}
+ stmfd sp!, {r0, r4-r7, lr}
+ rsb r3, r0, #0
+ ands r3, r3, #3
+ cmp r3, r2
+ movhi r3, r2
+
+ /* splat r1 */
+ mov r1, r1, lsl #24
+ orr r1, r1, r1, lsr #8
+ orr r1, r1, r1, lsr #16
+
+ movs r12, r3, lsl #31
+ strcsb r1, [r0], #1 /* can't use strh (alignment unknown) */
+ strcsb r1, [r0], #1
+ strmib r1, [r0], #1
+ subs r2, r2, r3
+ ldmlsfd sp!, {r0, r4-r7, lr} /* return */
+ bxls lr
+
+ /* align the destination to a cache-line */
+ mov r12, r1
+ mov lr, r1
+ mov r4, r1
+ mov r5, r1
+ mov r6, r1
+ mov r7, r1
+
+ rsb r3, r0, #0
+ ands r3, r3, #0x1C
+ beq 3f
+ cmp r3, r2
+ andhi r3, r2, #0x1C
+ sub r2, r2, r3
+
+ /* conditionally writes 0 to 7 words (length in r3) */
+ movs r3, r3, lsl #28
+ stmcsia r0!, {r1, lr}
+ stmcsia r0!, {r1, lr}
+ stmmiia r0!, {r1, lr}
+ movs r3, r3, lsl #2
+ strcs r1, [r0], #4
+
+3:
+ subs r2, r2, #32
+ mov r3, r1
+ bmi 2f
+1: subs r2, r2, #32
+ stmia r0!, {r1,r3,r4,r5,r6,r7,r12,lr}
+ bhs 1b
+2: add r2, r2, #32
+
+ /* conditionally stores 0 to 31 bytes */
+ movs r2, r2, lsl #28
+ stmcsia r0!, {r1,r3,r12,lr}
+ stmmiia r0!, {r1, lr}
+ movs r2, r2, lsl #2
+ strcs r1, [r0], #4
+ strmih r1, [r0], #2
+ movs r2, r2, lsl #2
+ strcsb r1, [r0]
+ ldmfd sp!, {r0, r4-r7, lr}
+ bx lr
+END(memset)