1
0
mirror of https://git.tartarus.org/simon/putty.git synced 2025-01-24 16:52:24 +00:00

Reorganise SHA-512 to match SHA-256.

This builds on the previous refactoring by reworking the SHA-512
vtables and block layer to look more like the SHA-256 version, in
which the block and padding structure is a subroutine of the top-level
vtable methods instead of an owning layer around them.

This also organises the code in a way that makes it easy to drop in
hardware-accelerated versions alongside it: the block layer and the
big arrays of constants are now nicely separate from the inner
block-transform part.
This commit is contained in:
Simon Tatham 2020-12-24 15:20:03 +00:00
parent 43cdc3d910
commit c6d921add5
2 changed files with 201 additions and 252 deletions

1
ssh.h
View File

@ -748,6 +748,7 @@ struct ssh_hashalg {
const char *text_basename; /* the semantic name of the hash */
const char *annotation; /* extra info, e.g. which of multiple impls */
const char *text_name; /* both combined, e.g. "SHA-n (unaccelerated)" */
const void *extra; /* private to the hash implementation */
};
static inline ssh_hash *ssh_hash_new(const ssh_hashalg *alg)

View File

@ -9,73 +9,7 @@
#include <assert.h>
#include "ssh.h"
#define BLKSIZE 128
typedef struct {
uint64_t h[8];
unsigned char block[BLKSIZE];
int blkused;
uint64_t lenhi, lenlo;
BinarySink_IMPLEMENTATION;
} SHA512_State;
/* ----------------------------------------------------------------------
* Core SHA512 algorithm: processes 16-doubleword blocks into a
* message digest.
*/
static inline uint64_t ror(uint64_t x, unsigned y)
{
return (x << (63 & -y)) | (x >> (63 & y));
}
static inline uint64_t Ch(uint64_t ctrl, uint64_t if1, uint64_t if0)
{
return if0 ^ (ctrl & (if1 ^ if0));
}
static inline uint64_t Maj(uint64_t x, uint64_t y, uint64_t z)
{
return (x & y) | (z & (x | y));
}
static inline uint64_t Sigma_0(uint64_t x)
{
return ror(x,28) ^ ror(x,34) ^ ror(x,39);
}
static inline uint64_t Sigma_1(uint64_t x)
{
return ror(x,14) ^ ror(x,18) ^ ror(x,41);
}
static inline uint64_t sigma_0(uint64_t x)
{
return ror(x,1) ^ ror(x,8) ^ (x >> 7);
}
static inline uint64_t sigma_1(uint64_t x)
{
return ror(x,19) ^ ror(x,61) ^ (x >> 6);
}
static inline void SHA512_Round(
unsigned round_index, const uint64_t *round_constants,
const uint64_t *schedule,
uint64_t *a, uint64_t *b, uint64_t *c, uint64_t *d,
uint64_t *e, uint64_t *f, uint64_t *g, uint64_t *h)
{
uint64_t t1 = *h + Sigma_1(*e) + Ch(*e,*f,*g) +
round_constants[round_index] + schedule[round_index];
uint64_t t2 = Sigma_0(*a) + Maj(*a,*b,*c);
*d += t1;
*h = t1 + t2;
}
static void SHA512_Core_Init(SHA512_State *s) {
static const uint64_t iv[] = {
static const uint64_t sha512_initial_state[] = {
0x6a09e667f3bcc908ULL,
0xbb67ae8584caa73bULL,
0x3c6ef372fe94f82bULL,
@ -84,14 +18,9 @@ static void SHA512_Core_Init(SHA512_State *s) {
0x9b05688c2b3e6c1fULL,
0x1f83d9abfb41bd6bULL,
0x5be0cd19137e2179ULL,
};
int i;
for (i = 0; i < 8; i++)
s->h[i] = iv[i];
}
};
static void SHA384_Core_Init(SHA512_State *s) {
static const uint64_t iv[] = {
static const uint64_t sha384_initial_state[] = {
0xcbbb9d5dc1059ed8ULL,
0x629a292a367cd507ULL,
0x9159015a3070dd17ULL,
@ -100,16 +29,9 @@ static void SHA384_Core_Init(SHA512_State *s) {
0x8eb44a8768581511ULL,
0xdb0c2e0d64f98fa7ULL,
0x47b5481dbefa4fa4ULL,
};
int i;
for (i = 0; i < 8; i++)
s->h[i] = iv[i];
}
};
static void SHA512_Block(SHA512_State *s, uint64_t *block) {
uint64_t w[80];
uint64_t a,b,c,d,e,f,g,h;
static const uint64_t k[] = {
static const uint64_t sha512_round_constants[] = {
0x428a2f98d728ae22ULL, 0x7137449123ef65cdULL,
0xb5c0fbcfec4d3b2fULL, 0xe9b5dba58189dbbcULL,
0x3956c25bf348b538ULL, 0x59f111f1b605d019ULL,
@ -150,203 +72,229 @@ static void SHA512_Block(SHA512_State *s, uint64_t *block) {
0x3c9ebe0a15c9bebcULL, 0x431d67c49c100d4cULL,
0x4cc5d4becb3e42b6ULL, 0x597f299cfc657e2aULL,
0x5fcb6fab3ad6faecULL, 0x6c44198c4a475817ULL,
};
};
#define SHA512_ROUNDS 80
typedef struct sha512_block sha512_block;
struct sha512_block {
uint8_t block[128];
size_t used;
uint64_t lenhi, lenlo;
};
static inline void sha512_block_setup(sha512_block *blk)
{
blk->used = 0;
blk->lenhi = blk->lenlo = 0;
}
static inline bool sha512_block_write(
sha512_block *blk, const void **vdata, size_t *len)
{
size_t blkleft = sizeof(blk->block) - blk->used;
size_t chunk = *len < blkleft ? *len : blkleft;
const uint8_t *p = *vdata;
memcpy(blk->block + blk->used, p, chunk);
*vdata = p + chunk;
*len -= chunk;
blk->used += chunk;
size_t chunkbits = chunk << 3;
blk->lenlo += chunkbits;
blk->lenhi += (blk->lenlo < chunkbits);
if (blk->used == sizeof(blk->block)) {
blk->used = 0;
return true;
}
return false;
}
static inline void sha512_block_pad(sha512_block *blk, BinarySink *bs)
{
uint64_t final_lenhi = blk->lenhi;
uint64_t final_lenlo = blk->lenlo;
size_t pad = 127 & (111 - blk->used);
put_byte(bs, 0x80);
put_padding(bs, pad, 0);
put_uint64(bs, final_lenhi);
put_uint64(bs, final_lenlo);
assert(blk->used == 0 && "Should have exactly hit a block boundary");
}
static inline uint64_t ror(uint64_t x, unsigned y)
{
return (x << (63 & -y)) | (x >> (63 & y));
}
static inline uint64_t Ch(uint64_t ctrl, uint64_t if1, uint64_t if0)
{
return if0 ^ (ctrl & (if1 ^ if0));
}
static inline uint64_t Maj(uint64_t x, uint64_t y, uint64_t z)
{
return (x & y) | (z & (x | y));
}
static inline uint64_t Sigma_0(uint64_t x)
{
return ror(x,28) ^ ror(x,34) ^ ror(x,39);
}
static inline uint64_t Sigma_1(uint64_t x)
{
return ror(x,14) ^ ror(x,18) ^ ror(x,41);
}
static inline uint64_t sigma_0(uint64_t x)
{
return ror(x,1) ^ ror(x,8) ^ (x >> 7);
}
static inline uint64_t sigma_1(uint64_t x)
{
return ror(x,19) ^ ror(x,61) ^ (x >> 6);
}
static inline void sha512_sw_round(
unsigned round_index, const uint64_t *schedule,
uint64_t *a, uint64_t *b, uint64_t *c, uint64_t *d,
uint64_t *e, uint64_t *f, uint64_t *g, uint64_t *h)
{
uint64_t t1 = *h + Sigma_1(*e) + Ch(*e,*f,*g) +
sha512_round_constants[round_index] + schedule[round_index];
uint64_t t2 = Sigma_0(*a) + Maj(*a,*b,*c);
*d += t1;
*h = t1 + t2;
}
static void sha512_sw_block(uint64_t *core, const uint8_t *block)
{
uint64_t w[SHA512_ROUNDS];
uint64_t a,b,c,d,e,f,g,h;
int t;
for (t = 0; t < 16; t++)
w[t] = block[t];
w[t] = GET_64BIT_MSB_FIRST(block + 8*t);
for (t = 16; t < 80; t++)
for (t = 16; t < SHA512_ROUNDS; t++)
w[t] = w[t-16] + w[t-7] + sigma_0(w[t-15]) + sigma_1(w[t-2]);
a = s->h[0]; b = s->h[1]; c = s->h[2]; d = s->h[3];
e = s->h[4]; f = s->h[5]; g = s->h[6]; h = s->h[7];
a = core[0]; b = core[1]; c = core[2]; d = core[3];
e = core[4]; f = core[5]; g = core[6]; h = core[7];
for (t = 0; t < 80; t+=8) {
SHA512_Round(t+0, k,w, &a,&b,&c,&d,&e,&f,&g,&h);
SHA512_Round(t+1, k,w, &h,&a,&b,&c,&d,&e,&f,&g);
SHA512_Round(t+2, k,w, &g,&h,&a,&b,&c,&d,&e,&f);
SHA512_Round(t+3, k,w, &f,&g,&h,&a,&b,&c,&d,&e);
SHA512_Round(t+4, k,w, &e,&f,&g,&h,&a,&b,&c,&d);
SHA512_Round(t+5, k,w, &d,&e,&f,&g,&h,&a,&b,&c);
SHA512_Round(t+6, k,w, &c,&d,&e,&f,&g,&h,&a,&b);
SHA512_Round(t+7, k,w, &b,&c,&d,&e,&f,&g,&h,&a);
for (t = 0; t < SHA512_ROUNDS; t+=8) {
sha512_sw_round(t+0, w, &a,&b,&c,&d,&e,&f,&g,&h);
sha512_sw_round(t+1, w, &h,&a,&b,&c,&d,&e,&f,&g);
sha512_sw_round(t+2, w, &g,&h,&a,&b,&c,&d,&e,&f);
sha512_sw_round(t+3, w, &f,&g,&h,&a,&b,&c,&d,&e);
sha512_sw_round(t+4, w, &e,&f,&g,&h,&a,&b,&c,&d);
sha512_sw_round(t+5, w, &d,&e,&f,&g,&h,&a,&b,&c);
sha512_sw_round(t+6, w, &c,&d,&e,&f,&g,&h,&a,&b);
sha512_sw_round(t+7, w, &b,&c,&d,&e,&f,&g,&h,&a);
}
s->h[0] += a; s->h[1] += b; s->h[2] += c; s->h[3] += d;
s->h[4] += e; s->h[5] += f; s->h[6] += g; s->h[7] += h;
core[0] += a; core[1] += b; core[2] += c; core[3] += d;
core[4] += e; core[5] += f; core[6] += g; core[7] += h;
smemclr(w, sizeof(w));
}
/* ----------------------------------------------------------------------
* Outer SHA512 algorithm: take an arbitrary length byte string,
* convert it into 16-doubleword blocks with the prescribed padding
* at the end, and pass those blocks to the core SHA512 algorithm.
*/
static void SHA512_BinarySink_write(BinarySink *bs,
const void *p, size_t len);
static void SHA512_Init(SHA512_State *s) {
SHA512_Core_Init(s);
s->blkused = 0;
s->lenhi = s->lenlo = 0;
BinarySink_INIT(s, SHA512_BinarySink_write);
}
static void SHA384_Init(SHA512_State *s) {
SHA384_Core_Init(s);
s->blkused = 0;
s->lenhi = s->lenlo = 0;
BinarySink_INIT(s, SHA512_BinarySink_write);
}
static void SHA512_BinarySink_write(BinarySink *bs,
const void *p, size_t len)
{
SHA512_State *s = BinarySink_DOWNCAST(bs, SHA512_State);
unsigned char *q = (unsigned char *)p;
uint64_t wordblock[16];
int i;
/*
* Update the length field.
*/
s->lenlo += len;
s->lenhi += (s->lenlo < len);
if (s->blkused && s->blkused+len < BLKSIZE) {
/*
* Trivial case: just add to the block.
*/
memcpy(s->block + s->blkused, q, len);
s->blkused += len;
} else {
/*
* We must complete and process at least one block.
*/
while (s->blkused + len >= BLKSIZE) {
memcpy(s->block + s->blkused, q, BLKSIZE - s->blkused);
q += BLKSIZE - s->blkused;
len -= BLKSIZE - s->blkused;
/* Now process the block. Gather bytes big-endian into words */
for (i = 0; i < 16; i++)
wordblock[i] = GET_64BIT_MSB_FIRST(s->block + i*8);
SHA512_Block(s, wordblock);
s->blkused = 0;
}
memcpy(s->block, q, len);
s->blkused = len;
}
}
static void SHA512_Final(SHA512_State *s, unsigned char *digest) {
int i;
int pad;
unsigned char c[BLKSIZE];
uint64_t lenhi, lenlo;
if (s->blkused >= BLKSIZE-16)
pad = (BLKSIZE-16) + BLKSIZE - s->blkused;
else
pad = (BLKSIZE-16) - s->blkused;
lenhi = (s->lenhi << 3) | (s->lenlo >> (32-3));
lenlo = (s->lenlo << 3);
memset(c, 0, pad);
c[0] = 0x80;
put_data(s, &c, pad);
put_uint64(s, lenhi);
put_uint64(s, lenlo);
for (i = 0; i < 8; i++)
PUT_64BIT_MSB_FIRST(digest + i*8, s->h[i]);
}
static void SHA384_Final(SHA512_State *s, unsigned char *digest) {
unsigned char biggerDigest[512 / 8];
SHA512_Final(s, biggerDigest);
memcpy(digest, biggerDigest, 384 / 8);
}
/*
* Thin abstraction for things where hashes are pluggable.
*/
struct sha512_hash {
SHA512_State state;
typedef struct sha512_sw {
uint64_t core[8];
sha512_block blk;
BinarySink_IMPLEMENTATION;
ssh_hash hash;
};
} sha512_sw;
static ssh_hash *sha512_new(const ssh_hashalg *alg)
static void sha512_sw_write(BinarySink *bs, const void *vp, size_t len);
static ssh_hash *sha512_sw_new(const ssh_hashalg *alg)
{
struct sha512_hash *h = snew(struct sha512_hash);
h->hash.vt = alg;
BinarySink_DELEGATE_INIT(&h->hash, &h->state);
return ssh_hash_reset(&h->hash);
sha512_sw *s = snew(sha512_sw);
s->hash.vt = alg;
BinarySink_INIT(s, sha512_sw_write);
BinarySink_DELEGATE_INIT(&s->hash, s);
return &s->hash;
}
static void sha512_reset(ssh_hash *hash)
static void sha512_sw_reset(ssh_hash *hash)
{
struct sha512_hash *h = container_of(hash, struct sha512_hash, hash);
SHA512_Init(&h->state);
sha512_sw *s = container_of(hash, sha512_sw, hash);
/* The 'extra' field in the ssh_hashalg indicates which
* initialisation vector we're using */
memcpy(s->core, hash->vt->extra, sizeof(s->core));
sha512_block_setup(&s->blk);
}
static void sha512_copyfrom(ssh_hash *hashnew, ssh_hash *hashold)
static void sha512_sw_copyfrom(ssh_hash *hcopy, ssh_hash *horig)
{
struct sha512_hash *hold = container_of(hashold, struct sha512_hash, hash);
struct sha512_hash *hnew = container_of(hashnew, struct sha512_hash, hash);
sha512_sw *copy = container_of(hcopy, sha512_sw, hash);
sha512_sw *orig = container_of(horig, sha512_sw, hash);
hnew->state = hold->state;
BinarySink_COPIED(&hnew->state);
memcpy(copy, orig, sizeof(*copy));
BinarySink_COPIED(copy);
BinarySink_DELEGATE_INIT(&copy->hash, copy);
}
static void sha512_free(ssh_hash *hash)
static void sha512_sw_free(ssh_hash *hash)
{
struct sha512_hash *h = container_of(hash, struct sha512_hash, hash);
sha512_sw *s = container_of(hash, sha512_sw, hash);
smemclr(h, sizeof(*h));
sfree(h);
smemclr(s, sizeof(*s));
sfree(s);
}
static void sha512_digest(ssh_hash *hash, unsigned char *output)
static void sha512_sw_write(BinarySink *bs, const void *vp, size_t len)
{
struct sha512_hash *h = container_of(hash, struct sha512_hash, hash);
SHA512_Final(&h->state, output);
sha512_sw *s = BinarySink_DOWNCAST(bs, sha512_sw);
while (len > 0)
if (sha512_block_write(&s->blk, &vp, &len))
sha512_sw_block(s->core, s->blk.block);
}
static void sha512_sw_digest(ssh_hash *hash, uint8_t *digest)
{
sha512_sw *s = container_of(hash, sha512_sw, hash);
sha512_block_pad(&s->blk, BinarySink_UPCAST(s));
for (size_t i = 0; i < hash->vt->hlen / 8; i++)
PUT_64BIT_MSB_FIRST(digest + 8*i, s->core[i]);
}
const ssh_hashalg ssh_sha512 = {
.new = sha512_new,
.reset = sha512_reset,
.copyfrom = sha512_copyfrom,
.digest = sha512_digest,
.free = sha512_free,
.new = sha512_sw_new,
.reset = sha512_sw_reset,
.copyfrom = sha512_sw_copyfrom,
.digest = sha512_sw_digest,
.free = sha512_sw_free,
.hlen = 64,
.blocklen = BLKSIZE,
HASHALG_NAMES_BARE("SHA-512"),
.blocklen = 128,
HASHALG_NAMES_ANNOTATED("SHA-512", "unaccelerated"),
.extra = sha512_initial_state,
};
static void sha384_reset(ssh_hash *hash)
{
struct sha512_hash *h = container_of(hash, struct sha512_hash, hash);
SHA384_Init(&h->state);
}
static void sha384_digest(ssh_hash *hash, unsigned char *output)
{
struct sha512_hash *h = container_of(hash, struct sha512_hash, hash);
SHA384_Final(&h->state, output);
}
const ssh_hashalg ssh_sha384 = {
.new = sha512_new,
.reset = sha384_reset,
.copyfrom = sha512_copyfrom,
.digest = sha384_digest,
.free = sha512_free,
.new = sha512_sw_new,
.reset = sha512_sw_reset,
.copyfrom = sha512_sw_copyfrom,
.digest = sha512_sw_digest,
.free = sha512_sw_free,
.hlen = 48,
.blocklen = BLKSIZE,
HASHALG_NAMES_BARE("SHA-384"),
.blocklen = 128,
HASHALG_NAMES_ANNOTATED("SHA-384", "unaccelerated"),
.extra = sha384_initial_state,
};