generic-library/bionic
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge "Use unified syntax to compile with both llvm and gcc."

Browse Source
This commit is contained in:
Chih-Hung Hsieh 2015-05-12 04:31:09 +00:00 committed by Gerrit Code Review
commit 972d7fe63a
9 changed files with 89 additions and 91 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
libc/Android.mk
View File

@ -941,10 +941,6 @@ LOCAL_SRC_FILES := $(libc_bionic_src_files)
LOCAL_CFLAGS := $(libc_common_cflags) \ LOCAL_CFLAGS := $(libc_common_cflags) \
-Wframe-larger-than=2048 \ -Wframe-larger-than=2048 \
# memcpy.S, memchr.S, etc. do not compile with Clang.
LOCAL_CLANG_ASFLAGS_arm += -no-integrated-as
LOCAL_CLANG_ASFLAGS_arm64 += -no-integrated-as
LOCAL_CONLYFLAGS := $(libc_common_conlyflags) LOCAL_CONLYFLAGS := $(libc_common_conlyflags)
LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast
LOCAL_C_INCLUDES := $(libc_common_c_includes) bionic/libstdc++/include LOCAL_C_INCLUDES := $(libc_common_c_includes) bionic/libstdc++/include
@ -972,10 +968,6 @@ LOCAL_SRC_FILES := $(libc_bionic_ndk_src_files)
LOCAL_CFLAGS := $(libc_common_cflags) \ LOCAL_CFLAGS := $(libc_common_cflags) \
-Wframe-larger-than=2048 \ -Wframe-larger-than=2048 \
# memcpy.S, memchr.S, etc. do not compile with Clang.
LOCAL_CLANG_ASFLAGS_arm += -no-integrated-as
LOCAL_CLANG_ASFLAGS_arm64 += -no-integrated-as
LOCAL_CONLYFLAGS := $(libc_common_conlyflags) LOCAL_CONLYFLAGS := $(libc_common_conlyflags)
LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast
LOCAL_C_INCLUDES := $(libc_common_c_includes) bionic/libstdc++/include LOCAL_C_INCLUDES := $(libc_common_c_includes) bionic/libstdc++/include
@ -1024,10 +1016,6 @@ LOCAL_SRC_FILES := $(libc_pthread_src_files)
LOCAL_CFLAGS := $(libc_common_cflags) \ LOCAL_CFLAGS := $(libc_common_cflags) \
-Wframe-larger-than=2048 \ -Wframe-larger-than=2048 \
# memcpy.S, memchr.S, etc. do not compile with Clang.
LOCAL_CLANG_ASFLAGS_arm += -no-integrated-as
LOCAL_CLANG_ASFLAGS_arm64 += -no-integrated-as
LOCAL_CONLYFLAGS := $(libc_common_conlyflags) LOCAL_CONLYFLAGS := $(libc_common_conlyflags)
LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast LOCAL_CPPFLAGS := $(libc_common_cppflags) -Wold-style-cast
LOCAL_C_INCLUDES := $(libc_common_c_includes) LOCAL_C_INCLUDES := $(libc_common_c_includes)

29
libc/arch-arm/cortex-a9/bionic/memset.S
View File

@ -35,6 +35,7 @@
*/ */
.fpu neon .fpu neon
.syntax unified
ENTRY(__memset_chk) ENTRY(__memset_chk)
cmp r2, r3 cmp r2, r3
@ -100,9 +101,9 @@ ENTRY(memset)
1: bge 2f 1: bge 2f
vst1.32 {d0[0]}, [r0]! vst1.32 {d0[0]}, [r0]!
2: movs ip, r2, lsl #31 2: movs ip, r2, lsl #31
strmib r1, [r0], #1 strbmi r1, [r0], #1
strcsb r1, [r0], #1 strbcs r1, [r0], #1
strcsb r1, [r0], #1 strbcs r1, [r0], #1
ldmfd sp!, {r0} ldmfd sp!, {r0}
bx lr bx lr
END(memset) END(memset)
@ -131,11 +132,11 @@ ENTRY_PRIVATE(__memset_large_copy)
orr r1, r1, r1, lsr #16 orr r1, r1, r1, lsr #16
movs r12, r3, lsl #31 movs r12, r3, lsl #31
strcsb r1, [r0], #1 /* can't use strh (alignment unknown) */ strbcs r1, [r0], #1 /* can't use strh (alignment unknown) */
strcsb r1, [r0], #1 strbcs r1, [r0], #1
strmib r1, [r0], #1 strbmi r1, [r0], #1
subs r2, r2, r3 subs r2, r2, r3
ldmlsfd sp!, {r0, r4-r7, lr} /* return */ popls {r0, r4-r7, lr} /* return */
bxls lr bxls lr
/* align the destination to a cache-line */ /* align the destination to a cache-line */
@ -155,9 +156,9 @@ ENTRY_PRIVATE(__memset_large_copy)
/* conditionally writes 0 to 7 words (length in r3) */ /* conditionally writes 0 to 7 words (length in r3) */
movs r3, r3, lsl #28 movs r3, r3, lsl #28
stmcsia r0!, {r1, lr} stmcs r0!, {r1, lr}
stmcsia r0!, {r1, lr} stmcs r0!, {r1, lr}
stmmiia r0!, {r1, lr} stmmi r0!, {r1, lr}
movs r3, r3, lsl #2 movs r3, r3, lsl #2
strcs r1, [r0], #4 strcs r1, [r0], #4
@ -172,13 +173,13 @@ ENTRY_PRIVATE(__memset_large_copy)
/* conditionally stores 0 to 31 bytes */ /* conditionally stores 0 to 31 bytes */
movs r2, r2, lsl #28 movs r2, r2, lsl #28
stmcsia r0!, {r1,r3,r12,lr} stmcs r0!, {r1,r3,r12,lr}
stmmiia r0!, {r1, lr} stmmi r0!, {r1, lr}
movs r2, r2, lsl #2 movs r2, r2, lsl #2
strcs r1, [r0], #4 strcs r1, [r0], #4
strmih r1, [r0], #2 strhmi r1, [r0], #2
movs r2, r2, lsl #2 movs r2, r2, lsl #2
strcsb r1, [r0] strbcs r1, [r0]
ldmfd sp!, {r0, r4-r7, lr} ldmfd sp!, {r0, r4-r7, lr}
bx lr bx lr
END(__memset_large_copy) END(__memset_large_copy)

24
libc/arch-arm/generic/bionic/memcmp.S
View File

@ -40,6 +40,8 @@
* Optimized memcmp() for Cortex-A9. * Optimized memcmp() for Cortex-A9.
*/ */
.syntax unified
ENTRY(memcmp) ENTRY(memcmp)
pld [r0, #(CACHE_LINE_SIZE * 0)] pld [r0, #(CACHE_LINE_SIZE * 0)]
pld [r0, #(CACHE_LINE_SIZE * 1)] pld [r0, #(CACHE_LINE_SIZE * 1)]
@ -161,25 +163,25 @@ ENTRY(memcmp)
eors r0, r0, ip eors r0, r0, ip
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq ip, [r1, #4]! ldreq ip, [r1, #4]!
eoreqs r0, r0, lr eorseq r0, r0, lr
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq lr, [r1, #4]! ldreq lr, [r1, #4]!
eoreqs r0, r0, ip eorseq r0, r0, ip
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq ip, [r1, #4]! ldreq ip, [r1, #4]!
eoreqs r0, r0, lr eorseq r0, r0, lr
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq lr, [r1, #4]! ldreq lr, [r1, #4]!
eoreqs r0, r0, ip eorseq r0, r0, ip
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq ip, [r1, #4]! ldreq ip, [r1, #4]!
eoreqs r0, r0, lr eorseq r0, r0, lr
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq lr, [r1, #4]! ldreq lr, [r1, #4]!
eoreqs r0, r0, ip eorseq r0, r0, ip
ldreq r0, [r4], #4 ldreq r0, [r4], #4
ldreq ip, [r1, #4]! ldreq ip, [r1, #4]!
eoreqs r0, r0, lr eorseq r0, r0, lr
bne 2f bne 2f
subs r2, r2, #32 subs r2, r2, #32
bhs 0b bhs 0b
@ -263,17 +265,17 @@ ENTRY(memcmp)
ldreq lr, [r1], #4 ldreq lr, [r1], #4
ldreq r0, [r4], #4 ldreq r0, [r4], #4
orreq ip, ip, lr, lsl #16 orreq ip, ip, lr, lsl #16
eoreqs r0, r0, ip eorseq r0, r0, ip
moveq ip, lr, lsr #16 moveq ip, lr, lsr #16
ldreq lr, [r1], #4 ldreq lr, [r1], #4
ldreq r0, [r4], #4 ldreq r0, [r4], #4
orreq ip, ip, lr, lsl #16 orreq ip, ip, lr, lsl #16
eoreqs r0, r0, ip eorseq r0, r0, ip
moveq ip, lr, lsr #16 moveq ip, lr, lsr #16
ldreq lr, [r1], #4 ldreq lr, [r1], #4
ldreq r0, [r4], #4 ldreq r0, [r4], #4
orreq ip, ip, lr, lsl #16 orreq ip, ip, lr, lsl #16
eoreqs r0, r0, ip eorseq r0, r0, ip
bne 7f bne 7f
subs r2, r2, #16 subs r2, r2, #16
bhs 6b bhs 6b
@ -317,7 +319,7 @@ ENTRY(memcmp)
ldreq r7, [r1], #4 ldreq r7, [r1], #4
ldreq r0, [r4], #4 ldreq r0, [r4], #4
orreq ip, ip, r7, lsl r6 orreq ip, ip, r7, lsl r6
eoreqs r0, r0, ip eorseq r0, r0, ip
bne 7f bne 7f
subs r2, r2, #8 subs r2, r2, #8
bhs 6b bhs 6b

62
libc/arch-arm/generic/bionic/memcpy.S
View File

@ -37,6 +37,8 @@
* so we have to preserve R0. * so we have to preserve R0.
*/ */
.syntax unified
ENTRY(__memcpy_chk) ENTRY(__memcpy_chk)
cmp r2, r3 cmp r2, r3
bhi __memcpy_chk_fail bhi __memcpy_chk_fail
@ -81,12 +83,12 @@ ENTRY(memcpy)
*/ */
movs r12, r3, lsl #31 movs r12, r3, lsl #31
sub r2, r2, r3 /* we know that r3 <= r2 because r2 >= 4 */ sub r2, r2, r3 /* we know that r3 <= r2 because r2 >= 4 */
ldrmib r3, [r1], #1 ldrbmi r3, [r1], #1
ldrcsb r4, [r1], #1 ldrbcs r4, [r1], #1
ldrcsb r12,[r1], #1 ldrbcs r12,[r1], #1
strmib r3, [r0], #1 strbmi r3, [r0], #1
strcsb r4, [r0], #1 strbcs r4, [r0], #1
strcsb r12,[r0], #1 strbcs r12,[r0], #1
.Lsrc_aligned: .Lsrc_aligned:
@ -109,10 +111,10 @@ ENTRY(memcpy)
/* conditionally copies 0 to 7 words (length in r3) */ /* conditionally copies 0 to 7 words (length in r3) */
movs r12, r3, lsl #28 movs r12, r3, lsl #28
ldmcsia r1!, {r4, r5, r6, r7} /* 16 bytes */ ldmcs r1!, {r4, r5, r6, r7} /* 16 bytes */
ldmmiia r1!, {r8, r9} /* 8 bytes */ ldmmi r1!, {r8, r9} /* 8 bytes */
stmcsia r0!, {r4, r5, r6, r7} stmcs r0!, {r4, r5, r6, r7}
stmmiia r0!, {r8, r9} stmmi r0!, {r8, r9}
tst r3, #0x4 tst r3, #0x4
ldrne r10,[r1], #4 /* 4 bytes */ ldrne r10,[r1], #4 /* 4 bytes */
strne r10,[r0], #4 strne r10,[r0], #4
@ -177,18 +179,18 @@ ENTRY(memcpy)
/* conditionnaly copies 0 to 31 bytes */ /* conditionnaly copies 0 to 31 bytes */
movs r12, r2, lsl #28 movs r12, r2, lsl #28
ldmcsia r1!, {r4, r5, r6, r7} /* 16 bytes */ ldmcs r1!, {r4, r5, r6, r7} /* 16 bytes */
ldmmiia r1!, {r8, r9} /* 8 bytes */ ldmmi r1!, {r8, r9} /* 8 bytes */
stmcsia r0!, {r4, r5, r6, r7} stmcs r0!, {r4, r5, r6, r7}
stmmiia r0!, {r8, r9} stmmi r0!, {r8, r9}
movs r12, r2, lsl #30 movs r12, r2, lsl #30
ldrcs r3, [r1], #4 /* 4 bytes */ ldrcs r3, [r1], #4 /* 4 bytes */
ldrmih r4, [r1], #2 /* 2 bytes */ ldrhmi r4, [r1], #2 /* 2 bytes */
strcs r3, [r0], #4 strcs r3, [r0], #4
strmih r4, [r0], #2 strhmi r4, [r0], #2
tst r2, #0x1 tst r2, #0x1
ldrneb r3, [r1] /* last byte */ ldrbne r3, [r1] /* last byte */
strneb r3, [r0] strbne r3, [r0]
/* we're done! restore everything and return */ /* we're done! restore everything and return */
1: ldmfd sp!, {r5-r11} 1: ldmfd sp!, {r5-r11}
@ -228,11 +230,11 @@ ENTRY(memcpy)
* becomes aligned to 32 bits (r5 = nb of words to copy for alignment) * becomes aligned to 32 bits (r5 = nb of words to copy for alignment)
*/ */
movs r5, r5, lsl #31 movs r5, r5, lsl #31
strmib r3, [r0], #1 strbmi r3, [r0], #1
movmi r3, r3, lsr #8 movmi r3, r3, lsr #8
strcsb r3, [r0], #1 strbcs r3, [r0], #1
movcs r3, r3, lsr #8 movcs r3, r3, lsr #8
strcsb r3, [r0], #1 strbcs r3, [r0], #1
movcs r3, r3, lsr #8 movcs r3, r3, lsr #8
cmp r2, #4 cmp r2, #4
@ -363,23 +365,23 @@ ENTRY(memcpy)
.Lpartial_word_tail: .Lpartial_word_tail:
/* we have a partial word in the input buffer */ /* we have a partial word in the input buffer */
movs r5, lr, lsl #(31-3) movs r5, lr, lsl #(31-3)
strmib r3, [r0], #1 strbmi r3, [r0], #1
movmi r3, r3, lsr #8 movmi r3, r3, lsr #8
strcsb r3, [r0], #1 strbcs r3, [r0], #1
movcs r3, r3, lsr #8 movcs r3, r3, lsr #8
strcsb r3, [r0], #1 strbcs r3, [r0], #1
/* Refill spilled registers from the stack. Don't update sp. */ /* Refill spilled registers from the stack. Don't update sp. */
ldmfd sp, {r5-r11} ldmfd sp, {r5-r11}
.Lcopy_last_3_and_return: .Lcopy_last_3_and_return:
movs r2, r2, lsl #31 /* copy remaining 0, 1, 2 or 3 bytes */ movs r2, r2, lsl #31 /* copy remaining 0, 1, 2 or 3 bytes */
ldrmib r2, [r1], #1 ldrbmi r2, [r1], #1
ldrcsb r3, [r1], #1 ldrbcs r3, [r1], #1
ldrcsb r12,[r1] ldrbcs r12,[r1]
strmib r2, [r0], #1 strbmi r2, [r0], #1
strcsb r3, [r0], #1 strbcs r3, [r0], #1
strcsb r12,[r0] strbcs r12,[r0]
/* we're done! restore sp and spilled registers and return */ /* we're done! restore sp and spilled registers and return */
add sp, sp, #28 add sp, sp, #28

24
libc/arch-arm/generic/bionic/memset.S
View File

@ -35,6 +35,8 @@
* memset() returns its first argument. * memset() returns its first argument.
*/ */
.syntax unified
ENTRY(__memset_chk) ENTRY(__memset_chk)
cmp r2, r3 cmp r2, r3
bls done bls done
@ -76,11 +78,11 @@ ENTRY(memset)
orr r1, r1, r1, lsr #16 orr r1, r1, r1, lsr #16
movs r12, r3, lsl #31 movs r12, r3, lsl #31
strcsb r1, [r0], #1 /* can't use strh (alignment unknown) */ strbcs r1, [r0], #1 /* can't use strh (alignment unknown) */
strcsb r1, [r0], #1 strbcs r1, [r0], #1
strmib r1, [r0], #1 strbmi r1, [r0], #1
subs r2, r2, r3 subs r2, r2, r3
ldmlsfd sp!, {r0, r4-r7, lr} /* return */ popls {r0, r4-r7, lr} /* return */
bxls lr bxls lr
/* align the destination to a cache-line */ /* align the destination to a cache-line */
@ -100,9 +102,9 @@ ENTRY(memset)
/* conditionally writes 0 to 7 words (length in r3) */ /* conditionally writes 0 to 7 words (length in r3) */
movs r3, r3, lsl #28 movs r3, r3, lsl #28
stmcsia r0!, {r1, lr} stmcs r0!, {r1, lr}
stmcsia r0!, {r1, lr} stmcs r0!, {r1, lr}
stmmiia r0!, {r1, lr} stmmi r0!, {r1, lr}
movs r3, r3, lsl #2 movs r3, r3, lsl #2
strcs r1, [r0], #4 strcs r1, [r0], #4
@ -117,13 +119,13 @@ ENTRY(memset)
/* conditionally stores 0 to 31 bytes */ /* conditionally stores 0 to 31 bytes */
movs r2, r2, lsl #28 movs r2, r2, lsl #28
stmcsia r0!, {r1,r3,r12,lr} stmcs r0!, {r1,r3,r12,lr}
stmmiia r0!, {r1, lr} stmmi r0!, {r1, lr}
movs r2, r2, lsl #2 movs r2, r2, lsl #2
strcs r1, [r0], #4 strcs r1, [r0], #4
strmih r1, [r0], #2 strhmi r1, [r0], #2
movs r2, r2, lsl #2 movs r2, r2, lsl #2
strcsb r1, [r0] strbcs r1, [r0]
ldmfd sp!, {r0, r4-r7, lr} ldmfd sp!, {r0, r4-r7, lr}
bx lr bx lr
END(memset) END(memset)

10
libc/arch-arm/generic/bionic/strcpy.S
View File

@ -32,6 +32,8 @@
#include <machine/cpu-features.h> #include <machine/cpu-features.h>
#include <private/bionic_asm.h> #include <private/bionic_asm.h>
.syntax unified
ENTRY(strcpy) ENTRY(strcpy)
pld [r1, #0] pld [r1, #0]
eor r2, r0, r1 eor r2, r0, r1
@ -108,15 +110,15 @@ ENTRY(strcpy)
#ifdef __ARMEB__ #ifdef __ARMEB__
tst r2, #0xff00 tst r2, #0xff00
iteet ne iteet ne
strneh r2, [ip], #2 strhne r2, [ip], #2
lsreq r2, r2, #8 lsreq r2, r2, #8
streqb r2, [ip] strbeq r2, [ip]
tstne r2, #0xff tstne r2, #0xff
#else #else
tst r2, #0xff tst r2, #0xff
itet ne itet ne
strneh r2, [ip], #2 strhne r2, [ip], #2
streqb r2, [ip] strbeq r2, [ip]
tstne r2, #0xff00 tstne r2, #0xff00
#endif #endif
bne 5b bne 5b

7
libc/arch-arm/krait/bionic/memset.S
View File

@ -37,6 +37,7 @@
*/ */
.fpu neon .fpu neon
.syntax unified
ENTRY(__memset_chk) ENTRY(__memset_chk)
cmp r2, r3 cmp r2, r3
@ -98,9 +99,9 @@ ENTRY(memset)
1: bge 2f 1: bge 2f
vst1.32 {d0[0]}, [r0]! vst1.32 {d0[0]}, [r0]!
2: movs ip, r2, lsl #31 2: movs ip, r2, lsl #31
strmib r1, [r0], #1 strbmi r1, [r0], #1
strcsb r1, [r0], #1 strbcs r1, [r0], #1
strcsb r1, [r0], #1 strbcs r1, [r0], #1
ldmfd sp!, {r0} ldmfd sp!, {r0}
bx lr bx lr
END(memset) END(memset)

6
libc/arch-arm64/generic/bionic/memchr.S
View File

@ -101,7 +101,7 @@ ENTRY(memchr)
and vhas_chr2.16b, vhas_chr2.16b, vrepmask.16b and vhas_chr2.16b, vhas_chr2.16b, vrepmask.16b
addp vend.16b, vhas_chr1.16b, vhas_chr2.16b /* 256->128 */ addp vend.16b, vhas_chr1.16b, vhas_chr2.16b /* 256->128 */
addp vend.16b, vend.16b, vend.16b /* 128->64 */ addp vend.16b, vend.16b, vend.16b /* 128->64 */
mov synd, vend.2d[0] mov synd, vend.d[0]
/* Clear the soff*2 lower bits */ /* Clear the soff*2 lower bits */
lsl tmp, soff, #1 lsl tmp, soff, #1
lsr synd, synd, tmp lsr synd, synd, tmp
@ -121,7 +121,7 @@ ENTRY(memchr)
/* Use a fast check for the termination condition */ /* Use a fast check for the termination condition */
orr vend.16b, vhas_chr1.16b, vhas_chr2.16b orr vend.16b, vhas_chr1.16b, vhas_chr2.16b
addp vend.2d, vend.2d, vend.2d addp vend.2d, vend.2d, vend.2d
mov synd, vend.2d[0] mov synd, vend.d[0]
/* We're not out of data, loop if we haven't found the character */ /* We're not out of data, loop if we haven't found the character */
cbz synd, .Lloop cbz synd, .Lloop
@ -131,7 +131,7 @@ ENTRY(memchr)
and vhas_chr2.16b, vhas_chr2.16b, vrepmask.16b and vhas_chr2.16b, vhas_chr2.16b, vrepmask.16b
addp vend.16b, vhas_chr1.16b, vhas_chr2.16b /* 256->128 */ addp vend.16b, vhas_chr1.16b, vhas_chr2.16b /* 256->128 */
addp vend.16b, vend.16b, vend.16b /* 128->64 */ addp vend.16b, vend.16b, vend.16b /* 128->64 */
mov synd, vend.2d[0] mov synd, vend.d[0]
/* Only do the clear for the last possible block */ /* Only do the clear for the last possible block */
b.hi .Ltail b.hi .Ltail

6
libc/arch-arm64/generic/bionic/strchr.S
View File

@ -109,7 +109,7 @@ ENTRY(strchr)
addp vend1.16b, vend1.16b, vend2.16b // 128->64 addp vend1.16b, vend1.16b, vend2.16b // 128->64
lsr tmp1, tmp3, tmp1 lsr tmp1, tmp3, tmp1
mov tmp3, vend1.2d[0] mov tmp3, vend1.d[0]
bic tmp1, tmp3, tmp1 // Mask padding bits. bic tmp1, tmp3, tmp1 // Mask padding bits.
cbnz tmp1, .Ltail cbnz tmp1, .Ltail
@ -124,7 +124,7 @@ ENTRY(strchr)
orr vend2.16b, vhas_nul2.16b, vhas_chr2.16b orr vend2.16b, vhas_nul2.16b, vhas_chr2.16b
orr vend1.16b, vend1.16b, vend2.16b orr vend1.16b, vend1.16b, vend2.16b
addp vend1.2d, vend1.2d, vend1.2d addp vend1.2d, vend1.2d, vend1.2d
mov tmp1, vend1.2d[0] mov tmp1, vend1.d[0]
cbz tmp1, .Lloop cbz tmp1, .Lloop
/* Termination condition found. Now need to establish exactly why /* Termination condition found. Now need to establish exactly why
@ -138,7 +138,7 @@ ENTRY(strchr)
addp vend1.16b, vend1.16b, vend2.16b // 256->128 addp vend1.16b, vend1.16b, vend2.16b // 256->128
addp vend1.16b, vend1.16b, vend2.16b // 128->64 addp vend1.16b, vend1.16b, vend2.16b // 128->64
mov tmp1, vend1.2d[0] mov tmp1, vend1.d[0]
.Ltail: .Ltail:
/* Count the trailing zeros, by bit reversing... */ /* Count the trailing zeros, by bit reversing... */
rbit tmp1, tmp1 rbit tmp1, tmp1