2 * Copyright (c) 2016 Thomas Pornin <pornin@bolet.org>
4 * Permission is hereby granted, free of charge, to any person obtaining
5 * a copy of this software and associated documentation files (the
6 * "Software"), to deal in the Software without restriction, including
7 * without limitation the rights to use, copy, modify, merge, publish,
8 * distribute, sublicense, and/or sell copies of the Software, and to
9 * permit persons to whom the Software is furnished to do so, subject to
10 * the following conditions:
12 * The above copyright notice and this permission notice shall be
13 * included in all copies or substantial portions of the Software.
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
16 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
17 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
18 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
19 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
20 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
21 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
35 * On MSVC, disable the warning about applying unary minus on an
36 * unsigned type: it is standard, we do it all the time, and for
40 #pragma warning( disable : 4146 )
44 * Maximum size for a RSA modulus (in bits). Allocated stack buffers
45 * depend on that size, so this value should be kept small. Currently,
46 * 2048-bit RSA keys offer adequate security, and should still do so for
47 * the next few decades; however, a number of widespread PKI have
48 * already set their root keys to RSA-4096, so we should be able to
51 * This value MUST be a multiple of 64.
53 #define BR_MAX_RSA_SIZE 4096
56 * Maximum size for a RSA factor (in bits). This is for RSA private-key
57 * operations. Default is to support factors up to a bit more than half
58 * the maximum modulus size.
60 * This value MUST be a multiple of 32.
62 #define BR_MAX_RSA_FACTOR ((BR_MAX_RSA_SIZE + 64) >> 1)
65 * Maximum size for an EC curve (modulus or order), in bits. Size of
66 * stack buffers depends on that parameter. This size MUST be a multiple
67 * of 8 (so that decoding an integer with that many bytes does not
70 #define BR_MAX_EC_SIZE 528
73 * Some macros to recognize the current architecture. Right now, we are
74 * interested into automatically recognizing architecture with efficient
75 * 64-bit types so that we may automatically use implementations that
76 * use 64-bit registers in that case. Future versions may detect, e.g.,
77 * availability of SSE2 intrinsics.
79 * If 'unsigned long' is a 64-bit type, then we assume that 64-bit types
80 * are efficient. Otherwise, we rely on macros that depend on compiler,
81 * OS and architecture. In any case, failure to detect the architecture
82 * as 64-bit means that the 32-bit code will be used, and that code
83 * works also on 64-bit architectures (the 64-bit code may simply be
86 * The test on 'unsigned long' should already catch most cases, the one
87 * notable exception being Windows code where 'unsigned long' is kept to
88 * 32-bit for compatbility with all the legacy code that liberally uses
89 * the 'DWORD' type for 32-bit values.
91 * Macro names are taken from: http://nadeausoftware.com/articles/2012/02/c_c_tip_how_detect_processor_type_using_compiler_predefined_macros
94 #if ((ULONG_MAX >> 31) >> 31) == 3
96 #elif defined(__ia64) || defined(__itanium__) || defined(_M_IA64)
98 #elif defined(__powerpc64__) || defined(__ppc64__) || defined(__PPC64__) \
99 || defined(__64BIT__) || defined(_LP64) || defined(__LP64__)
101 #elif defined(__sparc64__)
103 #elif defined(__x86_64__) || defined(_M_X64)
109 * Set BR_LOMUL on platforms where it makes sense.
112 #if BR_ARMEL_CORTEX_GCC
118 * Determine whether x86 AES instructions are understood by the compiler.
122 #if (__i386__ || __x86_64__) \
123 && ((__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)) \
124 || (__clang_major__ > 3 \
125 || (__clang_major__ == 3 && __clang_minor__ >= 7)))
126 #define BR_AES_X86NI 1
127 #elif (_M_IX86 || _M_X64) && (_MSC_VER >= 1700)
128 #define BR_AES_X86NI 1
133 * If we use x86 AES instruction, determine the compiler brand.
136 #ifndef BR_AES_X86NI_GCC
138 #define BR_AES_X86NI_GCC 1
141 #ifndef BR_AES_X86NI_MSC
143 #define BR_AES_X86NI_MSC 1
149 * A macro to tag a function with a "target" attribute (for GCC and Clang).
152 #define BR_TARGET(x) __attribute__((target(x)))
158 * GCC versions from 4.4 to 4.8 (inclusive) must use a special #pragma
159 * to activate extra opcodes before including the relevant intrinsic
160 * headers. But these don't work with Clang (which does not need them
163 #if BR_AES_X86NI_GCC && !defined BR_AES_X86NI_GCC_OLD
164 #if __GNUC__ == 4 && __GNUC_MINOR__ >= 4 && __GNUC_MINOR__ <= 8 && !__clang__
165 #define BR_AES_X86NI_GCC_OLD 1
170 * POWER8 crypto support. We rely on compiler macros for the
171 * architecture, since we do not have a reliable, simple way to detect
172 * the required support at runtime (we could try running an opcode, and
173 * trapping the exception or signal on illegal instruction, but this
174 * induces some non-trivial OS dependencies that we would prefer to
175 * avoid if possible).
178 #if __GNUC__ && ((_ARCH_PWR8 || _ARCH_PPC) && __CRYPTO__)
184 * Detect endinanness on POWER8.
187 #if defined BR_POWER8_LE
190 #define BR_POWER8_BE 0
192 #define BR_POWER8_BE 1
194 #elif defined BR_POWER8_BE
197 #define BR_POWER8_LE 0
199 #define BR_POWER8_LE 1
202 #if __LITTLE_ENDIAN__
203 #define BR_POWER8_LE 1
204 #define BR_POWER8_BE 0
206 #define BR_POWER8_LE 0
207 #define BR_POWER8_BE 1
212 /* ==================================================================== */
214 * Encoding/decoding functions.
216 * 32-bit and 64-bit decoding, both little-endian and big-endian, is
217 * implemented with the inline functions below. These functions are
218 * generic: they don't depend on the architecture natural endianness,
219 * and they can handle unaligned accesses. Optimized versions for some
220 * specific architectures may be implemented at a later time.
224 br_enc16le(void *dst
, unsigned x
)
229 buf
[0] = (unsigned char)x
;
230 buf
[1] = (unsigned char)(x
>> 8);
234 br_enc16be(void *dst
, unsigned x
)
239 buf
[0] = (unsigned char)(x
>> 8);
240 buf
[1] = (unsigned char)x
;
243 static inline unsigned
244 br_dec16le(const void *src
)
246 const unsigned char *buf
;
249 return (unsigned)buf
[0] | ((unsigned)buf
[1] << 8);
252 static inline unsigned
253 br_dec16be(const void *src
)
255 const unsigned char *buf
;
258 return ((unsigned)buf
[0] << 8) | (unsigned)buf
[1];
262 br_enc32le(void *dst
, uint32_t x
)
267 buf
[0] = (unsigned char)x
;
268 buf
[1] = (unsigned char)(x
>> 8);
269 buf
[2] = (unsigned char)(x
>> 16);
270 buf
[3] = (unsigned char)(x
>> 24);
274 br_enc32be(void *dst
, uint32_t x
)
279 buf
[0] = (unsigned char)(x
>> 24);
280 buf
[1] = (unsigned char)(x
>> 16);
281 buf
[2] = (unsigned char)(x
>> 8);
282 buf
[3] = (unsigned char)x
;
285 static inline uint32_t
286 br_dec32le(const void *src
)
288 const unsigned char *buf
;
291 return (uint32_t)buf
[0]
292 | ((uint32_t)buf
[1] << 8)
293 | ((uint32_t)buf
[2] << 16)
294 | ((uint32_t)buf
[3] << 24);
297 static inline uint32_t
298 br_dec32be(const void *src
)
300 const unsigned char *buf
;
303 return ((uint32_t)buf
[0] << 24)
304 | ((uint32_t)buf
[1] << 16)
305 | ((uint32_t)buf
[2] << 8)
310 br_enc64le(void *dst
, uint64_t x
)
315 br_enc32le(buf
, (uint32_t)x
);
316 br_enc32le(buf
+ 4, (uint32_t)(x
>> 32));
320 br_enc64be(void *dst
, uint64_t x
)
325 br_enc32be(buf
, (uint32_t)(x
>> 32));
326 br_enc32be(buf
+ 4, (uint32_t)x
);
329 static inline uint64_t
330 br_dec64le(const void *src
)
332 const unsigned char *buf
;
335 return (uint64_t)br_dec32le(buf
)
336 | ((uint64_t)br_dec32le(buf
+ 4) << 32);
339 static inline uint64_t
340 br_dec64be(const void *src
)
342 const unsigned char *buf
;
345 return ((uint64_t)br_dec32be(buf
) << 32)
346 | (uint64_t)br_dec32be(buf
+ 4);
350 * Range decoding and encoding (for several successive values).
352 void br_range_dec16le(uint16_t *v
, size_t num
, const void *src
);
353 void br_range_dec16be(uint16_t *v
, size_t num
, const void *src
);
354 void br_range_enc16le(void *dst
, const uint16_t *v
, size_t num
);
355 void br_range_enc16be(void *dst
, const uint16_t *v
, size_t num
);
357 void br_range_dec32le(uint32_t *v
, size_t num
, const void *src
);
358 void br_range_dec32be(uint32_t *v
, size_t num
, const void *src
);
359 void br_range_enc32le(void *dst
, const uint32_t *v
, size_t num
);
360 void br_range_enc32be(void *dst
, const uint32_t *v
, size_t num
);
362 void br_range_dec64le(uint64_t *v
, size_t num
, const void *src
);
363 void br_range_dec64be(uint64_t *v
, size_t num
, const void *src
);
364 void br_range_enc64le(void *dst
, const uint64_t *v
, size_t num
);
365 void br_range_enc64be(void *dst
, const uint64_t *v
, size_t num
);
368 * Byte-swap a 32-bit integer.
370 static inline uint32_t
371 br_swap32(uint32_t x
)
373 x
= ((x
& (uint32_t)0x00FF00FF) << 8)
374 | ((x
>> 8) & (uint32_t)0x00FF00FF);
375 return (x
<< 16) | (x
>> 16);
378 /* ==================================================================== */
380 * Support code for hash functions.
384 * IV for MD5, SHA-1, SHA-224 and SHA-256.
386 extern const uint32_t br_md5_IV
[];
387 extern const uint32_t br_sha1_IV
[];
388 extern const uint32_t br_sha224_IV
[];
389 extern const uint32_t br_sha256_IV
[];
392 * Round functions for MD5, SHA-1, SHA-224 and SHA-256 (SHA-224 and
393 * SHA-256 use the same round function).
395 void br_md5_round(const unsigned char *buf
, uint32_t *val
);
396 void br_sha1_round(const unsigned char *buf
, uint32_t *val
);
397 void br_sha2small_round(const unsigned char *buf
, uint32_t *val
);
400 * The core function for the TLS PRF. It computes
401 * P_hash(secret, label + seed), and XORs the result into the dst buffer.
403 void br_tls_phash(void *dst
, size_t len
,
404 const br_hash_class
*dig
,
405 const void *secret
, size_t secret_len
,
406 const char *label
, const void *seed
, size_t seed_len
);
409 * Copy all configured hash implementations from a multihash context
413 br_multihash_copyimpl(br_multihash_context
*dst
,
414 const br_multihash_context
*src
)
416 memcpy((void *)dst
->impl
, src
->impl
, sizeof src
->impl
);
419 /* ==================================================================== */
421 * Constant-time primitives. These functions manipulate 32-bit values in
422 * order to provide constant-time comparisons and multiplexers.
424 * Boolean values (the "ctl" bits) MUST have value 0 or 1.
426 * Implementation notes:
427 * =====================
429 * The uintN_t types are unsigned and with width exactly N bits; the C
430 * standard guarantees that computations are performed modulo 2^N, and
431 * there can be no overflow. Negation (unary '-') works on unsigned types
434 * The intN_t types are guaranteed to have width exactly N bits, with no
435 * padding bit, and using two's complement representation. Casting
436 * intN_t to uintN_t really is conversion modulo 2^N. Beware that intN_t
437 * types, being signed, trigger implementation-defined behaviour on
438 * overflow (including raising some signal): with GCC, while modular
439 * arithmetics are usually applied, the optimizer may assume that
440 * overflows don't occur (unless the -fwrapv command-line option is
441 * added); Clang has the additional -ftrapv option to explicitly trap on
442 * integer overflow or underflow.
448 static inline uint32_t
455 * Multiplexer: returns x if ctl == 1, y if ctl == 0.
457 static inline uint32_t
458 MUX(uint32_t ctl
, uint32_t x
, uint32_t y
)
460 return y
^ (-ctl
& (x
^ y
));
464 * Equality check: returns 1 if x == y, 0 otherwise.
466 static inline uint32_t
467 EQ(uint32_t x
, uint32_t y
)
472 return NOT((q
| -q
) >> 31);
476 * Inequality check: returns 1 if x != y, 0 otherwise.
478 static inline uint32_t
479 NEQ(uint32_t x
, uint32_t y
)
484 return (q
| -q
) >> 31;
488 * Comparison: returns 1 if x > y, 0 otherwise.
490 static inline uint32_t
491 GT(uint32_t x
, uint32_t y
)
494 * If both x < 2^31 and x < 2^31, then y-x will have its high
495 * bit set if x > y, cleared otherwise.
497 * If either x >= 2^31 or y >= 2^31 (but not both), then the
498 * result is the high bit of x.
500 * If both x >= 2^31 and y >= 2^31, then we can virtually
501 * subtract 2^31 from both, and we are back to the first case.
502 * Since (y-2^31)-(x-2^31) = y-x, the subtraction is already
508 return (z
^ ((x
^ y
) & (x
^ z
))) >> 31;
512 * Other comparisons (greater-or-equal, lower-than, lower-or-equal).
514 #define GE(x, y) NOT(GT(y, x))
515 #define LT(x, y) GT(y, x)
516 #define LE(x, y) NOT(GT(x, y))
519 * General comparison: returned value is -1, 0 or 1, depending on
520 * whether x is lower than, equal to, or greater than y.
522 static inline int32_t
523 CMP(uint32_t x
, uint32_t y
)
525 return (int32_t)GT(x
, y
) | -(int32_t)GT(y
, x
);
529 * Returns 1 if x == 0, 0 otherwise. Take care that the operand is signed.
531 static inline uint32_t
537 return ~(q
| -q
) >> 31;
541 * Returns 1 if x > 0, 0 otherwise. Take care that the operand is signed.
543 static inline uint32_t
547 * High bit of -x is 0 if x == 0, but 1 if x > 0.
552 return (~q
& -q
) >> 31;
556 * Returns 1 if x >= 0, 0 otherwise. Take care that the operand is signed.
558 static inline uint32_t
561 return ~(uint32_t)x
>> 31;
565 * Returns 1 if x < 0, 0 otherwise. Take care that the operand is signed.
567 static inline uint32_t
570 return (uint32_t)x
>> 31;
574 * Returns 1 if x <= 0, 0 otherwise. Take care that the operand is signed.
576 static inline uint32_t
582 * ~-x has its high bit set if and only if -x is nonnegative (as
583 * a signed int), i.e. x is in the -(2^31-1) to 0 range. We must
584 * do an OR with x itself to account for x = -2^31.
587 return (q
| ~-q
) >> 31;
591 * Conditional copy: src[] is copied into dst[] if and only if ctl is 1.
592 * dst[] and src[] may overlap completely (but not partially).
594 void br_ccopy(uint32_t ctl
, void *dst
, const void *src
, size_t len
);
596 #define CCOPY br_ccopy
599 * Compute the bit length of a 32-bit integer. Returned value is between 0
600 * and 32 (inclusive).
602 static inline uint32_t
603 BIT_LENGTH(uint32_t x
)
608 c
= GT(x
, 0xFFFF); x
= MUX(c
, x
>> 16, x
); k
+= c
<< 4;
609 c
= GT(x
, 0x00FF); x
= MUX(c
, x
>> 8, x
); k
+= c
<< 3;
610 c
= GT(x
, 0x000F); x
= MUX(c
, x
>> 4, x
); k
+= c
<< 2;
611 c
= GT(x
, 0x0003); x
= MUX(c
, x
>> 2, x
); k
+= c
<< 1;
617 * Compute the minimum of x and y.
619 static inline uint32_t
620 MIN(uint32_t x
, uint32_t y
)
622 return MUX(GT(x
, y
), y
, x
);
626 * Compute the maximum of x and y.
628 static inline uint32_t
629 MAX(uint32_t x
, uint32_t y
)
631 return MUX(GT(x
, y
), x
, y
);
635 * Multiply two 32-bit integers, with a 64-bit result. This default
636 * implementation assumes that the basic multiplication operator
637 * yields constant-time code.
639 #define MUL(x, y) ((uint64_t)(x) * (uint64_t)(y))
644 * Alternate implementation of MUL31, that will be constant-time on some
645 * (old) platforms where the default MUL31 is not. Unfortunately, it is
646 * also substantially slower, and yields larger code, on more modern
647 * platforms, which is why it is deactivated by default.
649 * MUL31_lo() must do some extra work because on some platforms, the
650 * _signed_ multiplication may return early if the top bits are 1.
651 * Simply truncating (casting) the output of MUL31() would not be
652 * sufficient, because the compiler may notice that we keep only the low
653 * word, and then replace automatically the unsigned multiplication with
654 * a signed multiplication opcode.
656 #define MUL31(x, y) ((uint64_t)((x) | (uint32_t)0x80000000) \
657 * (uint64_t)((y) | (uint32_t)0x80000000) \
658 - ((uint64_t)(x) << 31) - ((uint64_t)(y) << 31) \
659 - ((uint64_t)1 << 62))
660 static inline uint32_t
661 MUL31_lo(uint32_t x
, uint32_t y
)
666 xl
= (x
& 0xFFFF) | (uint32_t)0x80000000;
667 xh
= (x
>> 16) | (uint32_t)0x80000000;
668 yl
= (y
& 0xFFFF) | (uint32_t)0x80000000;
669 yh
= (y
>> 16) | (uint32_t)0x80000000;
670 return (xl
* yl
+ ((xl
* yh
+ xh
* yl
) << 16)) & (uint32_t)0x7FFFFFFF;
676 * Multiply two 31-bit integers, with a 62-bit result. This default
677 * implementation assumes that the basic multiplication operator
678 * yields constant-time code.
679 * The MUL31_lo() macro returns only the low 31 bits of the product.
681 #define MUL31(x, y) ((uint64_t)(x) * (uint64_t)(y))
682 #define MUL31_lo(x, y) (((uint32_t)(x) * (uint32_t)(y)) & (uint32_t)0x7FFFFFFF)
687 * Multiply two words together; the sum of the lengths of the two
688 * operands must not exceed 31 (for instance, one operand may use 16
689 * bits if the other fits on 15). If BR_CT_MUL15 is non-zero, then the
690 * macro will contain some extra operations that help in making the
691 * operation constant-time on some platforms, where the basic 32-bit
692 * multiplication is not constant-time.
695 #define MUL15(x, y) (((uint32_t)(x) | (uint32_t)0x80000000) \
696 * ((uint32_t)(y) | (uint32_t)0x80000000) \
697 & (uint32_t)0x7FFFFFFF)
699 #define MUL15(x, y) ((uint32_t)(x) * (uint32_t)(y))
703 * Arithmetic right shift (sign bit is copied). What happens when
704 * right-shifting a negative value is _implementation-defined_, so it
705 * does not trigger undefined behaviour, but it is still up to each
706 * compiler to define (and document) what it does. Most/all compilers
707 * will do an arithmetic shift, the sign bit being used to fill the
708 * holes; this is a native operation on the underlying CPU, and it would
709 * make little sense for the compiler to do otherwise. GCC explicitly
710 * documents that it follows that convention.
712 * Still, if BR_NO_ARITH_SHIFT is defined (and non-zero), then an
713 * alternate version will be used, that does not rely on such
714 * implementation-defined behaviour. Unfortunately, it is also slower
715 * and yields bigger code, which is why it is deactivated by default.
717 #if BR_NO_ARITH_SHIFT
718 #define ARSH(x, n) (((uint32_t)(x) >> (n)) \
719 | ((-((uint32_t)(x) >> 31)) << (32 - (n))))
721 #define ARSH(x, n) ((*(int32_t *)&(x)) >> (n))
725 * Constant-time division. The dividend hi:lo is divided by the
726 * divisor d; the quotient is returned and the remainder is written
727 * in *r. If hi == d, then the quotient does not fit on 32 bits;
728 * returned value is thus truncated. If hi > d, returned values are
731 uint32_t br_divrem(uint32_t hi
, uint32_t lo
, uint32_t d
, uint32_t *r
);
734 * Wrapper for br_divrem(); the remainder is returned, and the quotient
737 static inline uint32_t
738 br_rem(uint32_t hi
, uint32_t lo
, uint32_t d
)
742 br_divrem(hi
, lo
, d
, &r
);
747 * Wrapper for br_divrem(); the quotient is returned, and the remainder
750 static inline uint32_t
751 br_div(uint32_t hi
, uint32_t lo
, uint32_t d
)
755 return br_divrem(hi
, lo
, d
, &r
);
758 /* ==================================================================== */
764 * The 'i32' functions implement computations on big integers using
765 * an internal representation as an array of 32-bit integers. For
767 * -- x[0] contains the "announced bit length" of the integer
768 * -- x[1], x[2]... contain the value in little-endian order (x[1]
769 * contains the least significant 32 bits)
771 * Multiplications rely on the elementary 32x32->64 multiplication.
773 * The announced bit length specifies the number of bits that are
774 * significant in the subsequent 32-bit words. Unused bits in the
775 * last (most significant) word are set to 0; subsequent words are
776 * uninitialized and need not exist at all.
778 * The execution time and memory access patterns of all computations
779 * depend on the announced bit length, but not on the actual word
780 * values. For modular integers, the announced bit length of any integer
781 * modulo n is equal to the actual bit length of n; thus, computations
782 * on modular integers are "constant-time" (only the modulus length may
787 * Compute the actual bit length of an integer. The argument x should
788 * point to the first (least significant) value word of the integer.
789 * The len 'xlen' contains the number of 32-bit words to access.
791 * CT: value or length of x does not leak.
793 uint32_t br_i32_bit_length(uint32_t *x
, size_t xlen
);
796 * Decode an integer from its big-endian unsigned representation. The
797 * "true" bit length of the integer is computed, but all words of x[]
798 * corresponding to the full 'len' bytes of the source are set.
800 * CT: value or length of x does not leak.
802 void br_i32_decode(uint32_t *x
, const void *src
, size_t len
);
805 * Decode an integer from its big-endian unsigned representation. The
806 * integer MUST be lower than m[]; the announced bit length written in
807 * x[] will be equal to that of m[]. All 'len' bytes from the source are
810 * Returned value is 1 if the decode value fits within the modulus, 0
811 * otherwise. In the latter case, the x[] buffer will be set to 0 (but
812 * still with the announced bit length of m[]).
814 * CT: value or length of x does not leak. Memory access pattern depends
815 * only of 'len' and the announced bit length of m. Whether x fits or
816 * not does not leak either.
818 uint32_t br_i32_decode_mod(uint32_t *x
,
819 const void *src
, size_t len
, const uint32_t *m
);
822 * Reduce an integer (a[]) modulo another (m[]). The result is written
823 * in x[] and its announced bit length is set to be equal to that of m[].
825 * x[] MUST be distinct from a[] and m[].
827 * CT: only announced bit lengths leak, not values of x, a or m.
829 void br_i32_reduce(uint32_t *x
, const uint32_t *a
, const uint32_t *m
);
832 * Decode an integer from its big-endian unsigned representation, and
833 * reduce it modulo the provided modulus m[]. The announced bit length
834 * of the result is set to be equal to that of the modulus.
836 * x[] MUST be distinct from m[].
838 void br_i32_decode_reduce(uint32_t *x
,
839 const void *src
, size_t len
, const uint32_t *m
);
842 * Encode an integer into its big-endian unsigned representation. The
843 * output length in bytes is provided (parameter 'len'); if the length
844 * is too short then the integer is appropriately truncated; if it is
845 * too long then the extra bytes are set to 0.
847 void br_i32_encode(void *dst
, size_t len
, const uint32_t *x
);
850 * Multiply x[] by 2^32 and then add integer z, modulo m[]. This
851 * function assumes that x[] and m[] have the same announced bit
852 * length, and the announced bit length of m[] matches its true
855 * x[] and m[] MUST be distinct arrays.
857 * CT: only the common announced bit length of x and m leaks, not
858 * the values of x, z or m.
860 void br_i32_muladd_small(uint32_t *x
, uint32_t z
, const uint32_t *m
);
863 * Extract one word from an integer. The offset is counted in bits.
864 * The word MUST entirely fit within the word elements corresponding
865 * to the announced bit length of a[].
867 static inline uint32_t
868 br_i32_word(const uint32_t *a
, uint32_t off
)
873 u
= (size_t)(off
>> 5) + 1;
874 j
= (unsigned)off
& 31;
878 return (a
[u
] >> j
) | (a
[u
+ 1] << (32 - j
));
883 * Test whether an integer is zero.
885 uint32_t br_i32_iszero(const uint32_t *x
);
888 * Add b[] to a[] and return the carry (0 or 1). If ctl is 0, then a[]
889 * is unmodified, but the carry is still computed and returned. The
890 * arrays a[] and b[] MUST have the same announced bit length.
892 * a[] and b[] MAY be the same array, but partial overlap is not allowed.
894 uint32_t br_i32_add(uint32_t *a
, const uint32_t *b
, uint32_t ctl
);
897 * Subtract b[] from a[] and return the carry (0 or 1). If ctl is 0,
898 * then a[] is unmodified, but the carry is still computed and returned.
899 * The arrays a[] and b[] MUST have the same announced bit length.
901 * a[] and b[] MAY be the same array, but partial overlap is not allowed.
903 uint32_t br_i32_sub(uint32_t *a
, const uint32_t *b
, uint32_t ctl
);
906 * Compute d+a*b, result in d. The initial announced bit length of d[]
907 * MUST match that of a[]. The d[] array MUST be large enough to
908 * accommodate the full result, plus (possibly) an extra word. The
909 * resulting announced bit length of d[] will be the sum of the announced
910 * bit lengths of a[] and b[] (therefore, it may be larger than the actual
911 * bit length of the numerical result).
913 * a[] and b[] may be the same array. d[] must be disjoint from both a[]
916 void br_i32_mulacc(uint32_t *d
, const uint32_t *a
, const uint32_t *b
);
919 * Zeroize an integer. The announced bit length is set to the provided
920 * value, and the corresponding words are set to 0.
923 br_i32_zero(uint32_t *x
, uint32_t bit_len
)
926 memset(x
, 0, ((bit_len
+ 31) >> 5) * sizeof *x
);
930 * Compute -(1/x) mod 2^32. If x is even, then this function returns 0.
932 uint32_t br_i32_ninv32(uint32_t x
);
935 * Convert a modular integer to Montgomery representation. The integer x[]
936 * MUST be lower than m[], but with the same announced bit length.
938 void br_i32_to_monty(uint32_t *x
, const uint32_t *m
);
941 * Convert a modular integer back from Montgomery representation. The
942 * integer x[] MUST be lower than m[], but with the same announced bit
943 * length. The "m0i" parameter is equal to -(1/m0) mod 2^32, where m0 is
944 * the least significant value word of m[] (this works only if m[] is
947 void br_i32_from_monty(uint32_t *x
, const uint32_t *m
, uint32_t m0i
);
950 * Compute a modular Montgomery multiplication. d[] is filled with the
951 * value of x*y/R modulo m[] (where R is the Montgomery factor). The
952 * array d[] MUST be distinct from x[], y[] and m[]. x[] and y[] MUST be
953 * numerically lower than m[]. x[] and y[] MAY be the same array. The
954 * "m0i" parameter is equal to -(1/m0) mod 2^32, where m0 is the least
955 * significant value word of m[] (this works only if m[] is an odd
958 void br_i32_montymul(uint32_t *d
, const uint32_t *x
, const uint32_t *y
,
959 const uint32_t *m
, uint32_t m0i
);
962 * Compute a modular exponentiation. x[] MUST be an integer modulo m[]
963 * (same announced bit length, lower value). m[] MUST be odd. The
964 * exponent is in big-endian unsigned notation, over 'elen' bytes. The
965 * "m0i" parameter is equal to -(1/m0) mod 2^32, where m0 is the least
966 * significant value word of m[] (this works only if m[] is an odd
967 * integer). The t1[] and t2[] parameters must be temporary arrays,
968 * each large enough to accommodate an integer with the same size as m[].
970 void br_i32_modpow(uint32_t *x
, const unsigned char *e
, size_t elen
,
971 const uint32_t *m
, uint32_t m0i
, uint32_t *t1
, uint32_t *t2
);
973 /* ==================================================================== */
979 * The 'i31' functions implement computations on big integers using
980 * an internal representation as an array of 32-bit integers. For
982 * -- x[0] encodes the array length and the "announced bit length"
983 * of the integer: namely, if the announced bit length is k,
984 * then x[0] = ((k / 31) << 5) + (k % 31).
985 * -- x[1], x[2]... contain the value in little-endian order, 31
986 * bits per word (x[1] contains the least significant 31 bits).
987 * The upper bit of each word is 0.
989 * Multiplications rely on the elementary 32x32->64 multiplication.
991 * The announced bit length specifies the number of bits that are
992 * significant in the subsequent 32-bit words. Unused bits in the
993 * last (most significant) word are set to 0; subsequent words are
994 * uninitialized and need not exist at all.
996 * The execution time and memory access patterns of all computations
997 * depend on the announced bit length, but not on the actual word
998 * values. For modular integers, the announced bit length of any integer
999 * modulo n is equal to the actual bit length of n; thus, computations
1000 * on modular integers are "constant-time" (only the modulus length may
1005 * Test whether an integer is zero.
1007 uint32_t br_i31_iszero(const uint32_t *x
);
1010 * Add b[] to a[] and return the carry (0 or 1). If ctl is 0, then a[]
1011 * is unmodified, but the carry is still computed and returned. The
1012 * arrays a[] and b[] MUST have the same announced bit length.
1014 * a[] and b[] MAY be the same array, but partial overlap is not allowed.
1016 uint32_t br_i31_add(uint32_t *a
, const uint32_t *b
, uint32_t ctl
);
1019 * Subtract b[] from a[] and return the carry (0 or 1). If ctl is 0,
1020 * then a[] is unmodified, but the carry is still computed and returned.
1021 * The arrays a[] and b[] MUST have the same announced bit length.
1023 * a[] and b[] MAY be the same array, but partial overlap is not allowed.
1025 uint32_t br_i31_sub(uint32_t *a
, const uint32_t *b
, uint32_t ctl
);
1028 * Compute the ENCODED actual bit length of an integer. The argument x
1029 * should point to the first (least significant) value word of the
1030 * integer. The len 'xlen' contains the number of 32-bit words to
1031 * access. The upper bit of each value word MUST be 0.
1032 * Returned value is ((k / 31) << 5) + (k % 31) if the bit length is k.
1034 * CT: value or length of x does not leak.
1036 uint32_t br_i31_bit_length(uint32_t *x
, size_t xlen
);
1039 * Decode an integer from its big-endian unsigned representation. The
1040 * "true" bit length of the integer is computed and set in the encoded
1041 * announced bit length (x[0]), but all words of x[] corresponding to
1042 * the full 'len' bytes of the source are set.
1044 * CT: value or length of x does not leak.
1046 void br_i31_decode(uint32_t *x
, const void *src
, size_t len
);
1049 * Decode an integer from its big-endian unsigned representation. The
1050 * integer MUST be lower than m[]; the (encoded) announced bit length
1051 * written in x[] will be equal to that of m[]. All 'len' bytes from the
1054 * Returned value is 1 if the decode value fits within the modulus, 0
1055 * otherwise. In the latter case, the x[] buffer will be set to 0 (but
1056 * still with the announced bit length of m[]).
1058 * CT: value or length of x does not leak. Memory access pattern depends
1059 * only of 'len' and the announced bit length of m. Whether x fits or
1060 * not does not leak either.
1062 uint32_t br_i31_decode_mod(uint32_t *x
,
1063 const void *src
, size_t len
, const uint32_t *m
);
1066 * Zeroize an integer. The announced bit length is set to the provided
1067 * value, and the corresponding words are set to 0. The ENCODED bit length
1071 br_i31_zero(uint32_t *x
, uint32_t bit_len
)
1074 memset(x
, 0, ((bit_len
+ 31) >> 5) * sizeof *x
);
1078 * Right-shift an integer. The shift amount must be lower than 31
1081 void br_i31_rshift(uint32_t *x
, int count
);
1084 * Reduce an integer (a[]) modulo another (m[]). The result is written
1085 * in x[] and its announced bit length is set to be equal to that of m[].
1087 * x[] MUST be distinct from a[] and m[].
1089 * CT: only announced bit lengths leak, not values of x, a or m.
1091 void br_i31_reduce(uint32_t *x
, const uint32_t *a
, const uint32_t *m
);
1094 * Decode an integer from its big-endian unsigned representation, and
1095 * reduce it modulo the provided modulus m[]. The announced bit length
1096 * of the result is set to be equal to that of the modulus.
1098 * x[] MUST be distinct from m[].
1100 void br_i31_decode_reduce(uint32_t *x
,
1101 const void *src
, size_t len
, const uint32_t *m
);
1104 * Multiply x[] by 2^31 and then add integer z, modulo m[]. This
1105 * function assumes that x[] and m[] have the same announced bit
1106 * length, the announced bit length of m[] matches its true
1109 * x[] and m[] MUST be distinct arrays. z MUST fit in 31 bits (upper
1112 * CT: only the common announced bit length of x and m leaks, not
1113 * the values of x, z or m.
1115 void br_i31_muladd_small(uint32_t *x
, uint32_t z
, const uint32_t *m
);
1118 * Encode an integer into its big-endian unsigned representation. The
1119 * output length in bytes is provided (parameter 'len'); if the length
1120 * is too short then the integer is appropriately truncated; if it is
1121 * too long then the extra bytes are set to 0.
1123 void br_i31_encode(void *dst
, size_t len
, const uint32_t *x
);
1126 * Compute -(1/x) mod 2^31. If x is even, then this function returns 0.
1128 uint32_t br_i31_ninv31(uint32_t x
);
1131 * Compute a modular Montgomery multiplication. d[] is filled with the
1132 * value of x*y/R modulo m[] (where R is the Montgomery factor). The
1133 * array d[] MUST be distinct from x[], y[] and m[]. x[] and y[] MUST be
1134 * numerically lower than m[]. x[] and y[] MAY be the same array. The
1135 * "m0i" parameter is equal to -(1/m0) mod 2^31, where m0 is the least
1136 * significant value word of m[] (this works only if m[] is an odd
1139 void br_i31_montymul(uint32_t *d
, const uint32_t *x
, const uint32_t *y
,
1140 const uint32_t *m
, uint32_t m0i
);
1143 * Convert a modular integer to Montgomery representation. The integer x[]
1144 * MUST be lower than m[], but with the same announced bit length.
1146 void br_i31_to_monty(uint32_t *x
, const uint32_t *m
);
1149 * Convert a modular integer back from Montgomery representation. The
1150 * integer x[] MUST be lower than m[], but with the same announced bit
1151 * length. The "m0i" parameter is equal to -(1/m0) mod 2^32, where m0 is
1152 * the least significant value word of m[] (this works only if m[] is
1155 void br_i31_from_monty(uint32_t *x
, const uint32_t *m
, uint32_t m0i
);
1158 * Compute a modular exponentiation. x[] MUST be an integer modulo m[]
1159 * (same announced bit length, lower value). m[] MUST be odd. The
1160 * exponent is in big-endian unsigned notation, over 'elen' bytes. The
1161 * "m0i" parameter is equal to -(1/m0) mod 2^31, where m0 is the least
1162 * significant value word of m[] (this works only if m[] is an odd
1163 * integer). The t1[] and t2[] parameters must be temporary arrays,
1164 * each large enough to accommodate an integer with the same size as m[].
1166 void br_i31_modpow(uint32_t *x
, const unsigned char *e
, size_t elen
,
1167 const uint32_t *m
, uint32_t m0i
, uint32_t *t1
, uint32_t *t2
);
1170 * Compute d+a*b, result in d. The initial announced bit length of d[]
1171 * MUST match that of a[]. The d[] array MUST be large enough to
1172 * accommodate the full result, plus (possibly) an extra word. The
1173 * resulting announced bit length of d[] will be the sum of the announced
1174 * bit lengths of a[] and b[] (therefore, it may be larger than the actual
1175 * bit length of the numerical result).
1177 * a[] and b[] may be the same array. d[] must be disjoint from both a[]
1180 void br_i31_mulacc(uint32_t *d
, const uint32_t *a
, const uint32_t *b
);
1182 /* ==================================================================== */
1185 * FIXME: document "i15" functions.
1189 br_i15_zero(uint16_t *x
, uint16_t bit_len
)
1192 memset(x
, 0, ((bit_len
+ 15) >> 4) * sizeof *x
);
1195 uint32_t br_i15_iszero(const uint16_t *x
);
1197 uint16_t br_i15_ninv15(uint16_t x
);
1199 uint32_t br_i15_add(uint16_t *a
, const uint16_t *b
, uint32_t ctl
);
1201 uint32_t br_i15_sub(uint16_t *a
, const uint16_t *b
, uint32_t ctl
);
1203 void br_i15_muladd_small(uint16_t *x
, uint16_t z
, const uint16_t *m
);
1205 void br_i15_montymul(uint16_t *d
, const uint16_t *x
, const uint16_t *y
,
1206 const uint16_t *m
, uint16_t m0i
);
1208 void br_i15_to_monty(uint16_t *x
, const uint16_t *m
);
1210 void br_i15_modpow(uint16_t *x
, const unsigned char *e
, size_t elen
,
1211 const uint16_t *m
, uint16_t m0i
, uint16_t *t1
, uint16_t *t2
);
1213 uint32_t br_i15_modpow_opt(uint16_t *x
, const unsigned char *e
, size_t elen
,
1214 const uint16_t *m
, uint16_t m0i
, uint16_t *tmp
, size_t twlen
);
1216 void br_i15_encode(void *dst
, size_t len
, const uint16_t *x
);
1218 uint32_t br_i15_decode_mod(uint16_t *x
,
1219 const void *src
, size_t len
, const uint16_t *m
);
1221 void br_i15_rshift(uint16_t *x
, int count
);
1223 uint32_t br_i15_bit_length(uint16_t *x
, size_t xlen
);
1225 void br_i15_decode(uint16_t *x
, const void *src
, size_t len
);
1227 void br_i15_from_monty(uint16_t *x
, const uint16_t *m
, uint16_t m0i
);
1229 void br_i15_decode_reduce(uint16_t *x
,
1230 const void *src
, size_t len
, const uint16_t *m
);
1232 void br_i15_reduce(uint16_t *x
, const uint16_t *a
, const uint16_t *m
);
1234 void br_i15_mulacc(uint16_t *d
, const uint16_t *a
, const uint16_t *b
);
1236 /* ==================================================================== */
1238 static inline size_t
1239 br_digest_size(const br_hash_class
*digest_class
)
1241 return (size_t)(digest_class
->desc
>> BR_HASHDESC_OUT_OFF
)
1242 & BR_HASHDESC_OUT_MASK
;
1246 * Get the output size (in bytes) of a hash function.
1248 size_t br_digest_size_by_ID(int digest_id
);
1251 * Get the OID (encoded OBJECT IDENTIFIER value, without tag and length)
1252 * for a hash function. If digest_id is not a supported digest identifier
1253 * (in particular if it is equal to 0, i.e. br_md5sha1_ID), then NULL is
1254 * returned and *len is set to 0.
1256 const unsigned char *br_digest_OID(int digest_id
, size_t *len
);
1258 /* ==================================================================== */
1260 * DES support functions.
1264 * Apply DES Initial Permutation.
1266 void br_des_do_IP(uint32_t *xl
, uint32_t *xr
);
1269 * Apply DES Final Permutation (inverse of IP).
1271 void br_des_do_invIP(uint32_t *xl
, uint32_t *xr
);
1274 * Key schedule unit: for a DES key (8 bytes), compute 16 subkeys. Each
1275 * subkey is two 28-bit words represented as two 32-bit words; the PC-2
1276 * bit extration is NOT applied.
1278 void br_des_keysched_unit(uint32_t *skey
, const void *key
);
1281 * Reversal of 16 DES sub-keys (for decryption).
1283 void br_des_rev_skey(uint32_t *skey
);
1286 * DES/3DES key schedule for 'des_tab' (encryption direction). Returned
1287 * value is the number of rounds.
1289 unsigned br_des_tab_keysched(uint32_t *skey
, const void *key
, size_t key_len
);
1292 * DES/3DES key schedule for 'des_ct' (encryption direction). Returned
1293 * value is the number of rounds.
1295 unsigned br_des_ct_keysched(uint32_t *skey
, const void *key
, size_t key_len
);
1298 * DES/3DES subkey decompression (from the compressed bitsliced subkeys).
1300 void br_des_ct_skey_expand(uint32_t *sk_exp
,
1301 unsigned num_rounds
, const uint32_t *skey
);
1304 * DES/3DES block encryption/decryption ('des_tab').
1306 void br_des_tab_process_block(unsigned num_rounds
,
1307 const uint32_t *skey
, void *block
);
1310 * DES/3DES block encryption/decryption ('des_ct').
1312 void br_des_ct_process_block(unsigned num_rounds
,
1313 const uint32_t *skey
, void *block
);
1315 /* ==================================================================== */
1317 * AES support functions.
1321 * The AES S-box (256-byte table).
1323 extern const unsigned char br_aes_S
[];
1326 * AES key schedule. skey[] is filled with n+1 128-bit subkeys, where n
1327 * is the number of rounds (10 to 14, depending on key size). The number
1328 * of rounds is returned. If the key size is invalid (not 16, 24 or 32),
1329 * then 0 is returned.
1331 * This implementation uses a 256-byte table and is NOT constant-time.
1333 unsigned br_aes_keysched(uint32_t *skey
, const void *key
, size_t key_len
);
1336 * AES key schedule for decryption ('aes_big' implementation).
1338 unsigned br_aes_big_keysched_inv(uint32_t *skey
,
1339 const void *key
, size_t key_len
);
1342 * AES block encryption with the 'aes_big' implementation (fast, but
1343 * not constant-time). This function encrypts a single block "in place".
1345 void br_aes_big_encrypt(unsigned num_rounds
, const uint32_t *skey
, void *data
);
1348 * AES block decryption with the 'aes_big' implementation (fast, but
1349 * not constant-time). This function decrypts a single block "in place".
1351 void br_aes_big_decrypt(unsigned num_rounds
, const uint32_t *skey
, void *data
);
1354 * AES block encryption with the 'aes_small' implementation (small, but
1355 * slow and not constant-time). This function encrypts a single block
1358 void br_aes_small_encrypt(unsigned num_rounds
,
1359 const uint32_t *skey
, void *data
);
1362 * AES block decryption with the 'aes_small' implementation (small, but
1363 * slow and not constant-time). This function decrypts a single block
1366 void br_aes_small_decrypt(unsigned num_rounds
,
1367 const uint32_t *skey
, void *data
);
1370 * The constant-time implementation is "bitsliced": the 128-bit state is
1371 * split over eight 32-bit words q* in the following way:
1373 * -- Input block consists in 16 bytes:
1374 * a00 a10 a20 a30 a01 a11 a21 a31 a02 a12 a22 a32 a03 a13 a23 a33
1375 * In the terminology of FIPS 197, this is a 4x4 matrix which is read
1378 * -- Each byte is split into eight bits which are distributed over the
1379 * eight words, at the same rank. Thus, for a byte x at rank k, bit 0
1380 * (least significant) of x will be at rank k in q0 (if that bit is b,
1381 * then it contributes "b << k" to the value of q0), bit 1 of x will be
1382 * at rank k in q1, and so on.
1384 * -- Ranks given to bits are in "row order" and are either all even, or
1385 * all odd. Two independent AES states are thus interleaved, one using
1386 * the even ranks, the other the odd ranks. Row order means:
1387 * a00 a01 a02 a03 a10 a11 a12 a13 a20 a21 a22 a23 a30 a31 a32 a33
1389 * Converting input bytes from two AES blocks to bitslice representation
1390 * is done in the following way:
1391 * -- Decode first block into the four words q0 q2 q4 q6, in that order,
1392 * using little-endian convention.
1393 * -- Decode second block into the four words q1 q3 q5 q7, in that order,
1394 * using little-endian convention.
1395 * -- Call br_aes_ct_ortho().
1397 * Converting back to bytes is done by using the reverse operations. Note
1398 * that br_aes_ct_ortho() is its own inverse.
1402 * Perform bytewise orthogonalization of eight 32-bit words. Bytes
1403 * of q0..q7 are spread over all words: for a byte x that occurs
1404 * at rank i in q[j] (byte x uses bits 8*i to 8*i+7 in q[j]), the bit
1405 * of rank k in x (0 <= k <= 7) goes to q[k] at rank 8*i+j.
1407 * This operation is an involution.
1409 void br_aes_ct_ortho(uint32_t *q
);
1412 * The AES S-box, as a bitsliced constant-time version. The input array
1413 * consists in eight 32-bit words; 32 S-box instances are computed in
1414 * parallel. Bits 0 to 7 of each S-box input (bit 0 is least significant)
1415 * are spread over the words 0 to 7, at the same rank.
1417 void br_aes_ct_bitslice_Sbox(uint32_t *q
);
1420 * Like br_aes_bitslice_Sbox(), but for the inverse S-box.
1422 void br_aes_ct_bitslice_invSbox(uint32_t *q
);
1425 * Compute AES encryption on bitsliced data. Since input is stored on
1426 * eight 32-bit words, two block encryptions are actually performed
1429 void br_aes_ct_bitslice_encrypt(unsigned num_rounds
,
1430 const uint32_t *skey
, uint32_t *q
);
1433 * Compute AES decryption on bitsliced data. Since input is stored on
1434 * eight 32-bit words, two block decryptions are actually performed
1437 void br_aes_ct_bitslice_decrypt(unsigned num_rounds
,
1438 const uint32_t *skey
, uint32_t *q
);
1441 * AES key schedule, constant-time version. skey[] is filled with n+1
1442 * 128-bit subkeys, where n is the number of rounds (10 to 14, depending
1443 * on key size). The number of rounds is returned. If the key size is
1444 * invalid (not 16, 24 or 32), then 0 is returned.
1446 unsigned br_aes_ct_keysched(uint32_t *comp_skey
,
1447 const void *key
, size_t key_len
);
1450 * Expand AES subkeys as produced by br_aes_ct_keysched(), into
1451 * a larger array suitable for br_aes_ct_bitslice_encrypt() and
1452 * br_aes_ct_bitslice_decrypt().
1454 void br_aes_ct_skey_expand(uint32_t *skey
,
1455 unsigned num_rounds
, const uint32_t *comp_skey
);
1458 * For the ct64 implementation, the same bitslicing technique is used,
1459 * but four instances are interleaved. First instance uses bits 0, 4,
1460 * 8, 12,... of each word; second instance uses bits 1, 5, 9, 13,...
1465 * Perform bytewise orthogonalization of eight 64-bit words. Bytes
1466 * of q0..q7 are spread over all words: for a byte x that occurs
1467 * at rank i in q[j] (byte x uses bits 8*i to 8*i+7 in q[j]), the bit
1468 * of rank k in x (0 <= k <= 7) goes to q[k] at rank 8*i+j.
1470 * This operation is an involution.
1472 void br_aes_ct64_ortho(uint64_t *q
);
1475 * Interleave bytes for an AES input block. If input bytes are
1476 * denoted 0123456789ABCDEF, and have been decoded with little-endian
1477 * convention (w[0] contains 0123, with '3' being most significant;
1478 * w[1] contains 4567, and so on), then output word q0 will be
1479 * set to 08192A3B (again little-endian convention) and q1 will
1480 * be set to 4C5D6E7F.
1482 void br_aes_ct64_interleave_in(uint64_t *q0
, uint64_t *q1
, const uint32_t *w
);
1485 * Perform the opposite of br_aes_ct64_interleave_in().
1487 void br_aes_ct64_interleave_out(uint32_t *w
, uint64_t q0
, uint64_t q1
);
1490 * The AES S-box, as a bitsliced constant-time version. The input array
1491 * consists in eight 64-bit words; 64 S-box instances are computed in
1492 * parallel. Bits 0 to 7 of each S-box input (bit 0 is least significant)
1493 * are spread over the words 0 to 7, at the same rank.
1495 void br_aes_ct64_bitslice_Sbox(uint64_t *q
);
1498 * Like br_aes_bitslice_Sbox(), but for the inverse S-box.
1500 void br_aes_ct64_bitslice_invSbox(uint64_t *q
);
1503 * Compute AES encryption on bitsliced data. Since input is stored on
1504 * eight 64-bit words, four block encryptions are actually performed
1507 void br_aes_ct64_bitslice_encrypt(unsigned num_rounds
,
1508 const uint64_t *skey
, uint64_t *q
);
1511 * Compute AES decryption on bitsliced data. Since input is stored on
1512 * eight 64-bit words, four block decryptions are actually performed
1515 void br_aes_ct64_bitslice_decrypt(unsigned num_rounds
,
1516 const uint64_t *skey
, uint64_t *q
);
1519 * AES key schedule, constant-time version. skey[] is filled with n+1
1520 * 128-bit subkeys, where n is the number of rounds (10 to 14, depending
1521 * on key size). The number of rounds is returned. If the key size is
1522 * invalid (not 16, 24 or 32), then 0 is returned.
1524 unsigned br_aes_ct64_keysched(uint64_t *comp_skey
,
1525 const void *key
, size_t key_len
);
1528 * Expand AES subkeys as produced by br_aes_ct64_keysched(), into
1529 * a larger array suitable for br_aes_ct64_bitslice_encrypt() and
1530 * br_aes_ct64_bitslice_decrypt().
1532 void br_aes_ct64_skey_expand(uint64_t *skey
,
1533 unsigned num_rounds
, const uint64_t *comp_skey
);
1536 * Test support for AES-NI opcodes.
1538 int br_aes_x86ni_supported(void);
1541 * AES key schedule, using x86 AES-NI instructions. This yields the
1542 * subkeys in the encryption direction. Number of rounds is returned.
1543 * Key size MUST be 16, 24 or 32 bytes; otherwise, 0 is returned.
1545 unsigned br_aes_x86ni_keysched_enc(unsigned char *skni
,
1546 const void *key
, size_t len
);
1549 * AES key schedule, using x86 AES-NI instructions. This yields the
1550 * subkeys in the decryption direction. Number of rounds is returned.
1551 * Key size MUST be 16, 24 or 32 bytes; otherwise, 0 is returned.
1553 unsigned br_aes_x86ni_keysched_dec(unsigned char *skni
,
1554 const void *key
, size_t len
);
1557 * Test support for AES POWER8 opcodes.
1559 int br_aes_pwr8_supported(void);
1562 * AES key schedule, using POWER8 instructions. This yields the
1563 * subkeys in the encryption direction. Number of rounds is returned.
1564 * Key size MUST be 16, 24 or 32 bytes; otherwise, 0 is returned.
1566 unsigned br_aes_pwr8_keysched(unsigned char *skni
,
1567 const void *key
, size_t len
);
1569 /* ==================================================================== */
1575 * Apply proper PKCS#1 v1.5 padding (for signatures). 'hash_oid' is
1576 * the encoded hash function OID, or NULL.
1578 uint32_t br_rsa_pkcs1_sig_pad(const unsigned char *hash_oid
,
1579 const unsigned char *hash
, size_t hash_len
,
1580 uint32_t n_bitlen
, unsigned char *x
);
1583 * Check PKCS#1 v1.5 padding (for signatures). 'hash_oid' is the encoded
1584 * hash function OID, or NULL. The provided 'sig' value is _after_ the
1585 * modular exponentiation, i.e. it should be the padded hash. On
1586 * success, the hashed message is extracted.
1588 uint32_t br_rsa_pkcs1_sig_unpad(const unsigned char *sig
, size_t sig_len
,
1589 const unsigned char *hash_oid
, size_t hash_len
,
1590 unsigned char *hash_out
);
1592 /* ==================================================================== */
1598 * Type for generic EC parameters: curve order (unsigned big-endian
1599 * encoding) and encoded conventional generator.
1603 const unsigned char *order
;
1605 const unsigned char *generator
;
1606 size_t generator_len
;
1609 extern const br_ec_curve_def br_secp256r1
;
1610 extern const br_ec_curve_def br_secp384r1
;
1611 extern const br_ec_curve_def br_secp521r1
;
1614 * For Curve25519, the advertised "order" really is 2^255-1, since the
1615 * point multipliction function really works over arbitrary 255-bit
1616 * scalars. This value is only meant as a hint for ECDH key generation;
1617 * only ECDSA uses the exact curve order, and ECDSA is not used with
1618 * that specific curve.
1620 extern const br_ec_curve_def br_curve25519
;
1623 * Decode some bytes as an i31 integer, with truncation (corresponding
1624 * to the 'bits2int' operation in RFC 6979). The target ENCODED bit
1625 * length is provided as last parameter. The resulting value will have
1626 * this declared bit length, and consists the big-endian unsigned decoding
1627 * of exactly that many bits in the source (capped at the source length).
1629 void br_ecdsa_i31_bits2int(uint32_t *x
,
1630 const void *src
, size_t len
, uint32_t ebitlen
);
1633 * Decode some bytes as an i15 integer, with truncation (corresponding
1634 * to the 'bits2int' operation in RFC 6979). The target ENCODED bit
1635 * length is provided as last parameter. The resulting value will have
1636 * this declared bit length, and consists the big-endian unsigned decoding
1637 * of exactly that many bits in the source (capped at the source length).
1639 void br_ecdsa_i15_bits2int(uint16_t *x
,
1640 const void *src
, size_t len
, uint32_t ebitlen
);
1642 /* ==================================================================== */
1644 * SSL/TLS support functions.
1650 #define BR_SSL_CHANGE_CIPHER_SPEC 20
1651 #define BR_SSL_ALERT 21
1652 #define BR_SSL_HANDSHAKE 22
1653 #define BR_SSL_APPLICATION_DATA 23
1656 * Handshake message types.
1658 #define BR_SSL_HELLO_REQUEST 0
1659 #define BR_SSL_CLIENT_HELLO 1
1660 #define BR_SSL_SERVER_HELLO 2
1661 #define BR_SSL_CERTIFICATE 11
1662 #define BR_SSL_SERVER_KEY_EXCHANGE 12
1663 #define BR_SSL_CERTIFICATE_REQUEST 13
1664 #define BR_SSL_SERVER_HELLO_DONE 14
1665 #define BR_SSL_CERTIFICATE_VERIFY 15
1666 #define BR_SSL_CLIENT_KEY_EXCHANGE 16
1667 #define BR_SSL_FINISHED 20
1672 #define BR_LEVEL_WARNING 1
1673 #define BR_LEVEL_FATAL 2
1676 * Low-level I/O state.
1678 #define BR_IO_FAILED 0
1681 #define BR_IO_INOUT 3
1684 * Mark a SSL engine as failed. The provided error code is recorded if
1685 * the engine was not already marked as failed. If 'err' is 0, then the
1686 * engine is marked as closed (without error).
1688 void br_ssl_engine_fail(br_ssl_engine_context
*cc
, int err
);
1691 * Test whether the engine is closed (normally or as a failure).
1694 br_ssl_engine_closed(const br_ssl_engine_context
*cc
)
1696 return cc
->iomode
== BR_IO_FAILED
;
1700 * Configure a new maximum fragment length. If possible, the maximum
1701 * length for outgoing records is immediately adjusted (if there are
1702 * not already too many buffered bytes for that).
1704 void br_ssl_engine_new_max_frag_len(
1705 br_ssl_engine_context
*rc
, unsigned max_frag_len
);
1708 * Test whether the current incoming record has been fully received
1709 * or not. This functions returns 0 only if a complete record header
1710 * has been received, but some of the (possibly encrypted) payload
1711 * has not yet been obtained.
1713 int br_ssl_engine_recvrec_finished(const br_ssl_engine_context
*rc
);
1716 * Flush the current record (if not empty). This is meant to be called
1717 * from the handshake processor only.
1719 void br_ssl_engine_flush_record(br_ssl_engine_context
*cc
);
1722 * Test whether there is some accumulated payload to send.
1725 br_ssl_engine_has_pld_to_send(const br_ssl_engine_context
*rc
)
1727 return rc
->oxa
!= rc
->oxb
&& rc
->oxa
!= rc
->oxc
;
1731 * Initialize RNG in engine. Returned value is 1 on success, 0 on error.
1732 * This function will try to use the OS-provided RNG, if available. If
1733 * there is no OS-provided RNG, or if it failed, and no entropy was
1734 * injected by the caller, then a failure will be reported. On error,
1735 * the context error code is set.
1737 int br_ssl_engine_init_rand(br_ssl_engine_context
*cc
);
1740 * Reset the handshake-related parts of the engine.
1742 void br_ssl_engine_hs_reset(br_ssl_engine_context
*cc
,
1743 void (*hsinit
)(void *), void (*hsrun
)(void *));
1746 * Get the PRF to use for this context, for the provided PRF hash
1749 br_tls_prf_impl
br_ssl_engine_get_PRF(br_ssl_engine_context
*cc
, int prf_id
);
1752 * Consume the provided pre-master secret and compute the corresponding
1753 * master secret. The 'prf_id' is the ID of the hash function to use
1754 * with the TLS 1.2 PRF (ignored if the version is TLS 1.0 or 1.1).
1756 void br_ssl_engine_compute_master(br_ssl_engine_context
*cc
,
1757 int prf_id
, const void *pms
, size_t len
);
1760 * Switch to CBC decryption for incoming records.
1761 * cc the engine context
1762 * is_client non-zero for a client, zero for a server
1763 * prf_id id of hash function for PRF (ignored if not TLS 1.2+)
1764 * mac_id id of hash function for HMAC
1765 * bc_impl block cipher implementation (CBC decryption)
1766 * cipher_key_len block cipher key length (in bytes)
1768 void br_ssl_engine_switch_cbc_in(br_ssl_engine_context
*cc
,
1769 int is_client
, int prf_id
, int mac_id
,
1770 const br_block_cbcdec_class
*bc_impl
, size_t cipher_key_len
);
1773 * Switch to CBC encryption for outgoing records.
1774 * cc the engine context
1775 * is_client non-zero for a client, zero for a server
1776 * prf_id id of hash function for PRF (ignored if not TLS 1.2+)
1777 * mac_id id of hash function for HMAC
1778 * bc_impl block cipher implementation (CBC encryption)
1779 * cipher_key_len block cipher key length (in bytes)
1781 void br_ssl_engine_switch_cbc_out(br_ssl_engine_context
*cc
,
1782 int is_client
, int prf_id
, int mac_id
,
1783 const br_block_cbcenc_class
*bc_impl
, size_t cipher_key_len
);
1786 * Switch to GCM decryption for incoming records.
1787 * cc the engine context
1788 * is_client non-zero for a client, zero for a server
1789 * prf_id id of hash function for PRF
1790 * bc_impl block cipher implementation (CTR)
1791 * cipher_key_len block cipher key length (in bytes)
1793 void br_ssl_engine_switch_gcm_in(br_ssl_engine_context
*cc
,
1794 int is_client
, int prf_id
,
1795 const br_block_ctr_class
*bc_impl
, size_t cipher_key_len
);
1798 * Switch to GCM encryption for outgoing records.
1799 * cc the engine context
1800 * is_client non-zero for a client, zero for a server
1801 * prf_id id of hash function for PRF
1802 * bc_impl block cipher implementation (CTR)
1803 * cipher_key_len block cipher key length (in bytes)
1805 void br_ssl_engine_switch_gcm_out(br_ssl_engine_context
*cc
,
1806 int is_client
, int prf_id
,
1807 const br_block_ctr_class
*bc_impl
, size_t cipher_key_len
);
1810 * Switch to ChaCha20+Poly1305 decryption for incoming records.
1811 * cc the engine context
1812 * is_client non-zero for a client, zero for a server
1813 * prf_id id of hash function for PRF
1815 void br_ssl_engine_switch_chapol_in(br_ssl_engine_context
*cc
,
1816 int is_client
, int prf_id
);
1819 * Switch to ChaCha20+Poly1305 encryption for outgoing records.
1820 * cc the engine context
1821 * is_client non-zero for a client, zero for a server
1822 * prf_id id of hash function for PRF
1824 void br_ssl_engine_switch_chapol_out(br_ssl_engine_context
*cc
,
1825 int is_client
, int prf_id
);
1828 * Calls to T0-generated code.
1830 void br_ssl_hs_client_init_main(void *ctx
);
1831 void br_ssl_hs_client_run(void *ctx
);
1832 void br_ssl_hs_server_init_main(void *ctx
);
1833 void br_ssl_hs_server_run(void *ctx
);
1836 * Get the hash function to use for signatures, given a bit mask of
1837 * supported hash functions. This implements a strict choice order
1838 * (namely SHA-256, SHA-384, SHA-512, SHA-224, SHA-1). If the mask
1839 * does not document support of any of these hash functions, then this
1840 * functions returns 0.
1842 int br_ssl_choose_hash(unsigned bf
);
1844 /* ==================================================================== */
1847 * PowerPC / POWER assembly stuff. The special BR_POWER_ASM_MACROS macro
1848 * must be defined before including this file; this is done by source
1849 * files that use some inline assembly for PowerPC / POWER machines.
1852 #if BR_POWER_ASM_MACROS
1854 #define lxvw4x(xt, ra, rb) lxvw4x_(xt, ra, rb)
1855 #define stxvw4x(xt, ra, rb) stxvw4x_(xt, ra, rb)
1857 #define bdnz(foo) bdnz_(foo)
1858 #define beq(foo) beq_(foo)
1860 #define li(rx, value) li_(rx, value)
1861 #define addi(rx, ra, imm) addi_(rx, ra, imm)
1862 #define cmpldi(rx, imm) cmpldi_(rx, imm)
1863 #define mtctr(rx) mtctr_(rx)
1864 #define vspltb(vrt, vrb, uim) vspltb_(vrt, vrb, uim)
1865 #define vspltw(vrt, vrb, uim) vspltw_(vrt, vrb, uim)
1866 #define vspltisb(vrt, imm) vspltisb_(vrt, imm)
1867 #define vspltisw(vrt, imm) vspltisw_(vrt, imm)
1868 #define vrlw(vrt, vra, vrb) vrlw_(vrt, vra, vrb)
1869 #define vsbox(vrt, vra) vsbox_(vrt, vra)
1870 #define vxor(vrt, vra, vrb) vxor_(vrt, vra, vrb)
1871 #define vand(vrt, vra, vrb) vand_(vrt, vra, vrb)
1872 #define vsro(vrt, vra, vrb) vsro_(vrt, vra, vrb)
1873 #define vsl(vrt, vra, vrb) vsl_(vrt, vra, vrb)
1874 #define vsldoi(vt, va, vb, sh) vsldoi_(vt, va, vb, sh)
1875 #define vsr(vrt, vra, vrb) vsr_(vrt, vra, vrb)
1876 #define vadduwm(vrt, vra, vrb) vadduwm_(vrt, vra, vrb)
1877 #define vsububm(vrt, vra, vrb) vsububm_(vrt, vra, vrb)
1878 #define vsubuwm(vrt, vra, vrb) vsubuwm_(vrt, vra, vrb)
1879 #define vsrw(vrt, vra, vrb) vsrw_(vrt, vra, vrb)
1880 #define vcipher(vt, va, vb) vcipher_(vt, va, vb)
1881 #define vcipherlast(vt, va, vb) vcipherlast_(vt, va, vb)
1882 #define vncipher(vt, va, vb) vncipher_(vt, va, vb)
1883 #define vncipherlast(vt, va, vb) vncipherlast_(vt, va, vb)
1884 #define vperm(vt, va, vb, vc) vperm_(vt, va, vb, vc)
1885 #define vpmsumd(vt, va, vb) vpmsumd_(vt, va, vb)
1886 #define xxpermdi(vt, va, vb, d) xxpermdi_(vt, va, vb, d)
1888 #define lxvw4x_(xt, ra, rb) "\tlxvw4x\t" #xt "," #ra "," #rb "\n"
1889 #define stxvw4x_(xt, ra, rb) "\tstxvw4x\t" #xt "," #ra "," #rb "\n"
1891 #define label(foo) #foo "%=:\n"
1892 #define bdnz_(foo) "\tbdnz\t" #foo "%=\n"
1893 #define beq_(foo) "\tbeq\t" #foo "%=\n"
1895 #define li_(rx, value) "\tli\t" #rx "," #value "\n"
1896 #define addi_(rx, ra, imm) "\taddi\t" #rx "," #ra "," #imm "\n"
1897 #define cmpldi_(rx, imm) "\tcmpldi\t" #rx "," #imm "\n"
1898 #define mtctr_(rx) "\tmtctr\t" #rx "\n"
1899 #define vspltb_(vrt, vrb, uim) "\tvspltb\t" #vrt "," #vrb "," #uim "\n"
1900 #define vspltw_(vrt, vrb, uim) "\tvspltw\t" #vrt "," #vrb "," #uim "\n"
1901 #define vspltisb_(vrt, imm) "\tvspltisb\t" #vrt "," #imm "\n"
1902 #define vspltisw_(vrt, imm) "\tvspltisw\t" #vrt "," #imm "\n"
1903 #define vrlw_(vrt, vra, vrb) "\tvrlw\t" #vrt "," #vra "," #vrb "\n"
1904 #define vsbox_(vrt, vra) "\tvsbox\t" #vrt "," #vra "\n"
1905 #define vxor_(vrt, vra, vrb) "\tvxor\t" #vrt "," #vra "," #vrb "\n"
1906 #define vand_(vrt, vra, vrb) "\tvand\t" #vrt "," #vra "," #vrb "\n"
1907 #define vsro_(vrt, vra, vrb) "\tvsro\t" #vrt "," #vra "," #vrb "\n"
1908 #define vsl_(vrt, vra, vrb) "\tvsl\t" #vrt "," #vra "," #vrb "\n"
1909 #define vsldoi_(vt, va, vb, sh) "\tvsldoi\t" #vt "," #va "," #vb "," #sh "\n"
1910 #define vsr_(vrt, vra, vrb) "\tvsr\t" #vrt "," #vra "," #vrb "\n"
1911 #define vadduwm_(vrt, vra, vrb) "\tvadduwm\t" #vrt "," #vra "," #vrb "\n"
1912 #define vsububm_(vrt, vra, vrb) "\tvsububm\t" #vrt "," #vra "," #vrb "\n"
1913 #define vsubuwm_(vrt, vra, vrb) "\tvsubuwm\t" #vrt "," #vra "," #vrb "\n"
1914 #define vsrw_(vrt, vra, vrb) "\tvsrw\t" #vrt "," #vra "," #vrb "\n"
1915 #define vcipher_(vt, va, vb) "\tvcipher\t" #vt "," #va "," #vb "\n"
1916 #define vcipherlast_(vt, va, vb) "\tvcipherlast\t" #vt "," #va "," #vb "\n"
1917 #define vncipher_(vt, va, vb) "\tvncipher\t" #vt "," #va "," #vb "\n"
1918 #define vncipherlast_(vt, va, vb) "\tvncipherlast\t" #vt "," #va "," #vb "\n"
1919 #define vperm_(vt, va, vb, vc) "\tvperm\t" #vt "," #va "," #vb "," #vc "\n"
1920 #define vpmsumd_(vt, va, vb) "\tvpmsumd\t" #vt "," #va "," #vb "\n"
1921 #define xxpermdi_(vt, va, vb, d) "\txxpermdi\t" #vt "," #va "," #vb "," #d "\n"
1925 /* ==================================================================== */