blob: 04c7438c079b9dff5c0b9325ae85e4a49f1a458a (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
|
#
# For a description of the syntax of this configuration file,
# see scripts/config/Kconfig-language.txt
#
menu "BigInt Options"
depends on !CONFIG_SSL_SKELETON_MODE
choice
prompt "Reduction Algorithm"
default CONFIG_BIGINT_BARRETT
config CONFIG_BIGINT_CLASSICAL
bool "Classical"
help
Classical uses standard division. It has no limitations and is
theoretically the slowest due to the divisions used. For this particular
implementation it is surprisingly quite fast.
config CONFIG_BIGINT_MONTGOMERY
bool "Montgomery"
help
Montgomery uses simple addition and multiplication to achieve its
performance. In this implementation it is slower than classical,
and it has the limitation that 0 <= x, y < m, and so is not used
when CRT is active.
This option will not be normally selected.
config CONFIG_BIGINT_BARRETT
bool "Barrett"
help
Barrett performs expensive precomputation before reduction and partial
multiplies for computational speed. It can't be used with some of the
calculations when CRT is used, and so defaults to classical when this
occurs.
It is about 40% faster than Classical/Montgomery with the expense of
about 2kB, and so this option is normally selected.
endchoice
config CONFIG_BIGINT_CRT
bool "Chinese Remainder Theorem (CRT)"
default y
help
Allow the Chinese Remainder Theorem (CRT) to be used.
Uses a number of extra coefficients from the private key to improve the
performance of a decryption. This feature is one of the most
significant performance improvements (it reduces a decryption time by
over 3 times).
This option should be selected.
config CONFIG_BIGINT_KARATSUBA
bool "Karatsuba Multiplication"
default n
help
Allow Karasuba multiplication to be used.
Uses 3 multiplications (plus a number of additions/subtractions)
instead of 4. Multiplications are O(N^2) but addition/subtraction
is O(N) hence for large numbers is beneficial. For this project, the
effect was only useful for 4096 bit keys. As these aren't likely to
be used, the feature is disabled by default.
It costs about 2kB to enable it.
config MUL_KARATSUBA_THRESH
int "Karatsuba Multiplication Theshold"
default 20
depends on CONFIG_BIGINT_KARATSUBA
help
The minimum number of components needed before Karasuba muliplication
is used.
This is very dependent on the speed/implementation of bi_add()/
bi_subtract(). There is a bit of trial and error here and will be
at a different point for different architectures.
config SQU_KARATSUBA_THRESH
int "Karatsuba Square Threshold"
default 40
depends on CONFIG_BIGINT_KARATSUBA && CONFIG_BIGINT_SQUARE
help
The minimum number of components needed before Karatsuba squaring
is used.
This is very dependent on the speed/implementation of bi_add()/
bi_subtract(). There is a bit of trial and error here and will be
at a different point for different architectures.
config CONFIG_BIGINT_SLIDING_WINDOW
bool "Sliding Window Exponentiation"
default y
help
Allow Sliding-Window Exponentiation to be used.
Potentially processes more than 1 bit at a time when doing
exponentiation. The sliding-window technique reduces the number of
precomputations compared to other precomputed techniques.
It results in a considerable performance improvement with it enabled
(it halves the decryption time) and so should be selected.
config CONFIG_BIGINT_SQUARE
bool "Square Algorithm"
default y
help
Allow squaring to be used instead of a multiplication.
Squaring is theoretically 50% faster than a standard multiply
(but is actually about 25% faster).
It gives a 20% speed improvement and so should be selected.
config CONFIG_BIGINT_CHECK_ON
bool "BigInt Integrity Checking"
default n if !CONFIG_DEBUG
default y if CONFIG_DEBUG
help
This is used when developing bigint algorithms. It performs a sanity
check on all operations at the expense of speed.
This option is only selected when developing and should normally be
turned off.
endmenu
|
