~linaro-toolchain-dev/cortex-strings/trunk : revision 110

1

2

3

4

Redistribution and use in source and binary forms, with or without

5

modification, are permitted provided that the following conditions

6

are met:

7

8

* Redistributions of source code must retain the above copyright

9

notice, this list of conditions and the following disclaimer.

10

11

* Redistributions in binary form must reproduce the above copyright

12

notice, this list of conditions and the following disclaimer in the

13

documentation and/or other materials provided with the distribution.

14

15

* Neither the name of Linaro Limited nor the names of its

16

contributors may be used to endorse or promote products derived

17

from this software without specific prior written permission.

18

19

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS

20

"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT

21

LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR

22

A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT

23

HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,

24

SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT

25

LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,

26

DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY

27

THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT

28

(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE

29

OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

30

31

Assumes:

32

ARMv7, AArch32

33

*/

34

35

.macro def_fn f p2align=0

36

.text

37

.p2align \p2align

38

.global \f

39

.type \f, %function

40

\f:

41

.endm

42

43

#ifdef __ARMEB__

44

#define S2LO lsl

45

#define S2HI lsr

46

#else

47

#define S2LO lsr

48

#define S2HI lsl

49

#endif

50

51

/* This code requires Thumb. */

52

.thumb

53

.syntax unified

54

55

/* Parameters and result. */

56

#define srcin r0

57

#define result r0

58

59

/* Internal variables. */

60

#define src r1

61

#define data1a r2

62

#define data1b r3

63

#define const_m1 r12

64

#define const_0 r4

65

#define tmp1 r4 /* Overlaps const_0 */

66

#define tmp2 r5

67

68

def_fn strlen p2align=6

69

pld [srcin, #0]

70

strd r4, r5, [sp, #-8]!

71

bic src, srcin, #7

72

mvn const_m1, #0

73

ands tmp1, srcin, #7 /* (8 - bytes) to alignment. */

74

pld [src, #32]

75

bne.w .Lmisaligned8

76

mov const_0, #0

77

mov result, #-8

78

.Lloop_aligned:

79

/* Bytes 0-7. */

80

ldrd data1a, data1b, [src]

81

pld [src, #64]

82

add result, result, #8

83

.Lstart_realigned:

84

uadd8 data1a, data1a, const_m1 /* Saturating GE<0:3> set. */

85

sel data1a, const_0, const_m1 /* Select based on GE<0:3>. */

86

uadd8 data1b, data1b, const_m1

87

sel data1b, data1a, const_m1 /* Only used if d1a == 0. */

88

cbnz data1b, .Lnull_found

89

90

/* Bytes 8-15. */

91

ldrd data1a, data1b, [src, #8]

92

uadd8 data1a, data1a, const_m1 /* Saturating GE<0:3> set. */

93

add result, result, #8

94

sel data1a, const_0, const_m1 /* Select based on GE<0:3>. */

95

uadd8 data1b, data1b, const_m1

96

sel data1b, data1a, const_m1 /* Only used if d1a == 0. */

97

cbnz data1b, .Lnull_found

98

99

/* Bytes 16-23. */

100

ldrd data1a, data1b, [src, #16]

101

uadd8 data1a, data1a, const_m1 /* Saturating GE<0:3> set. */

102

add result, result, #8

103

sel data1a, const_0, const_m1 /* Select based on GE<0:3>. */

104

uadd8 data1b, data1b, const_m1

105

sel data1b, data1a, const_m1 /* Only used if d1a == 0. */

106

cbnz data1b, .Lnull_found

107

108

/* Bytes 24-31. */

109

ldrd data1a, data1b, [src, #24]

110

add src, src, #32

111

uadd8 data1a, data1a, const_m1 /* Saturating GE<0:3> set. */

112

add result, result, #8

113

sel data1a, const_0, const_m1 /* Select based on GE<0:3>. */

114

uadd8 data1b, data1b, const_m1

115

sel data1b, data1a, const_m1 /* Only used if d1a == 0. */

116

cmp data1b, #0

117

beq .Lloop_aligned

118

119

.Lnull_found:

120

cmp data1a, #0

121

itt eq

122

addeq result, result, #4

123

moveq data1a, data1b

124

#ifndef __ARMEB__

125

rev data1a, data1a

126

#endif

127

clz data1a, data1a

128

ldrd r4, r5, [sp], #8

129

add result, result, data1a, lsr #3 /* Bits -> Bytes. */

130

bx lr

131

132

.Lmisaligned8:

133

ldrd data1a, data1b, [src]

134

and tmp2, tmp1, #3

135

rsb result, tmp1, #0

136

lsl tmp2, tmp2, #3 /* Bytes -> bits. */

137

tst tmp1, #4

138

pld [src, #64]

139

S2HI tmp2, const_m1, tmp2

140

orn data1a, data1a, tmp2

141

itt ne

142

ornne data1b, data1b, tmp2

143

movne data1a, const_m1

144

mov const_0, #0

145

b .Lstart_realigned

146

.size strlen, . - strlen

147