mirror/ git
1
0
Fork 0
mirror of https://github.com/git/git.git synced 2024-05-09 23:06:09 +02:00
Code
git/block-sha1/sha1.c
Linus Torvalds 926172c5e4 block-sha1: improve code on large-register-set machines 
For x86 performance (especially in 32-bit mode) I added that hack to write
the SHA1 internal temporary hash using a volatile pointer, in order to get
gcc to not try to cache the array contents. Because gcc will do all the
wrong things, and then spill things in insane random ways.

But on architectures like PPC, where you have 32 registers, it's actually
perfectly reasonable to put the whole temporary array[] into the register
set, and gcc can do so.

So make the 'volatile unsigned int *' cast be dependent on a
SMALL_REGISTER_SET preprocessor symbol, and enable it (currently) on just
x86 and x86-64.  With that, the routine is fairly reasonable even when
compared to the hand-scheduled PPC version. Ben Herrenschmidt reports on
a G5:

 * Paulus asm version:       about 3.67s
 * Yours with no change:     about 5.74s
 * Yours without "volatile": about 3.78s

so with this the C version is within about 3% of the asm one.

And add a lot of commentary on what the heck is going on.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-08-10 17:26:51 -07:00

246 lines
6.6 KiB
C
Raw Blame History

/*
* Based on the Mozilla SHA1 (see mozilla-sha1/sha1.c),
* optimized to do word accesses rather than byte accesses,
* and to avoid unnecessary copies into the context array.
*/
#include <string.h>
#include <arpa/inet.h>
#include "sha1.h"
/* Hash one 64-byte block of data */
static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data);
void blk_SHA1_Init(blk_SHA_CTX *ctx)
{
ctx->size = 0;
/* Initialize H with the magic constants (see FIPS180 for constants)
*/
ctx->H[0] = 0x67452301;
ctx->H[1] = 0xefcdab89;
ctx->H[2] = 0x98badcfe;
ctx->H[3] = 0x10325476;
ctx->H[4] = 0xc3d2e1f0;
}
void blk_SHA1_Update(blk_SHA_CTX *ctx, const void *data, unsigned long len)
{
int lenW = ctx->size & 63;
ctx->size += len;
/* Read the data into W and process blocks as they get full
*/
if (lenW) {
int left = 64 - lenW;
if (len < left)
left = len;
memcpy(lenW + (char *)ctx->W, data, left);
lenW = (lenW + left) & 63;
len -= left;
data += left;
if (lenW)
return;
blk_SHA1Block(ctx, ctx->W);
}
while (len >= 64) {
blk_SHA1Block(ctx, data);
data += 64;
len -= 64;
}
if (len)
memcpy(ctx->W, data, len);
}
void blk_SHA1_Final(unsigned char hashout[20], blk_SHA_CTX *ctx)
{
static const unsigned char pad[64] = { 0x80 };
unsigned int padlen[2];
int i;
/* Pad with a binary 1 (ie 0x80), then zeroes, then length
*/
padlen[0] = htonl(ctx->size >> 29);
padlen[1] = htonl(ctx->size << 3);
i = ctx->size & 63;
blk_SHA1_Update(ctx, pad, 1+ (63 & (55 - i)));
blk_SHA1_Update(ctx, padlen, 8);
/* Output hash
*/
for (i = 0; i < 5; i++)
((unsigned int *)hashout)[i] = htonl(ctx->H[i]);
}
#if defined(__i386__) || defined(__x86_64__)
#define SHA_ASM(op, x, n) ({ unsigned int __res; __asm__(op " %1,%0":"=r" (__res):"i" (n), "0" (x)); __res; })
#define SHA_ROL(x,n) SHA_ASM("rol", x, n)
#define SHA_ROR(x,n) SHA_ASM("ror", x, n)
#define SMALL_REGISTER_SET
#else
#define SHA_ROT(X,l,r) (((X) << (l)) | ((X) >> (r)))
#define SHA_ROL(X,n) SHA_ROT(X,n,32-(n))
#define SHA_ROR(X,n) SHA_ROT(X,32-(n),n)
#endif
/* This "rolls" over the 512-bit array */
#define W(x) (array[(x)&15])
/*
* If you have 32 registers or more, the compiler can (and should)
* try to change the array[] accesses into registers. However, on
* machines with less than ~25 registers, that won't really work,
* and at least gcc will make an unholy mess of it.
*
* So to avoid that mess which just slows things down, we force
* the stores to memory to actually happen (we might be better off
* with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
* suggested by Artur Skawina - that will also make gcc unable to
* try to do the silly "optimize away loads" part because it won't
* see what the value will be).
*
* Ben Herrenschmidt reports that on PPC, the C version comes close
* to the optimized asm with this (ie on PPC you don't want that
* 'volatile', since there are lots of registers).
*/
#ifdef SMALL_REGISTER_SET
#define setW(x, val) (*(volatile unsigned int *)&W(x) = (val))
#else
#define setW(x, val) (W(x) = (val))
#endif
/*
* Where do we get the source from? The first 16 iterations get it from
* the input data, the next mix it from the 512-bit array.
*/
#define SHA_SRC(t) htonl(data[t])
#define SHA_MIX(t) SHA_ROL(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
#define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
unsigned int TEMP = input(t); setW(t, TEMP); \
E += TEMP + SHA_ROL(A,5) + (fn) + (constant); \
B = SHA_ROR(B, 2); } while (0)
#define T_0_15(t, A, B, C, D, E) SHA_ROUND(t, SHA_SRC, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
#define T_16_19(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
#define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
#define T_40_59(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, ((B&C)+(D&(B^C))) , 0x8f1bbcdc, A, B, C, D, E )
#define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
static void blk_SHA1Block(blk_SHA_CTX *ctx, const unsigned int *data)
{
unsigned int A,B,C,D,E;
unsigned int array[16];
A = ctx->H[0];
B = ctx->H[1];
C = ctx->H[2];
D = ctx->H[3];
E = ctx->H[4];
/* Round 1 - iterations 0-16 take their input from 'data' */
T_0_15( 0, A, B, C, D, E);
T_0_15( 1, E, A, B, C, D);
T_0_15( 2, D, E, A, B, C);
T_0_15( 3, C, D, E, A, B);
T_0_15( 4, B, C, D, E, A);
T_0_15( 5, A, B, C, D, E);
T_0_15( 6, E, A, B, C, D);
T_0_15( 7, D, E, A, B, C);
T_0_15( 8, C, D, E, A, B);
T_0_15( 9, B, C, D, E, A);
T_0_15(10, A, B, C, D, E);
T_0_15(11, E, A, B, C, D);
T_0_15(12, D, E, A, B, C);
T_0_15(13, C, D, E, A, B);
T_0_15(14, B, C, D, E, A);
T_0_15(15, A, B, C, D, E);
/* Round 1 - tail. Input from 512-bit mixing array */
T_16_19(16, E, A, B, C, D);
T_16_19(17, D, E, A, B, C);
T_16_19(18, C, D, E, A, B);
T_16_19(19, B, C, D, E, A);
/* Round 2 */
T_20_39(20, A, B, C, D, E);
T_20_39(21, E, A, B, C, D);
T_20_39(22, D, E, A, B, C);
T_20_39(23, C, D, E, A, B);
T_20_39(24, B, C, D, E, A);
T_20_39(25, A, B, C, D, E);
T_20_39(26, E, A, B, C, D);
T_20_39(27, D, E, A, B, C);
T_20_39(28, C, D, E, A, B);
T_20_39(29, B, C, D, E, A);
T_20_39(30, A, B, C, D, E);
T_20_39(31, E, A, B, C, D);
T_20_39(32, D, E, A, B, C);
T_20_39(33, C, D, E, A, B);
T_20_39(34, B, C, D, E, A);
T_20_39(35, A, B, C, D, E);
T_20_39(36, E, A, B, C, D);
T_20_39(37, D, E, A, B, C);
T_20_39(38, C, D, E, A, B);
T_20_39(39, B, C, D, E, A);
/* Round 3 */
T_40_59(40, A, B, C, D, E);
T_40_59(41, E, A, B, C, D);
T_40_59(42, D, E, A, B, C);
T_40_59(43, C, D, E, A, B);
T_40_59(44, B, C, D, E, A);
T_40_59(45, A, B, C, D, E);
T_40_59(46, E, A, B, C, D);
T_40_59(47, D, E, A, B, C);
T_40_59(48, C, D, E, A, B);
T_40_59(49, B, C, D, E, A);
T_40_59(50, A, B, C, D, E);
T_40_59(51, E, A, B, C, D);
T_40_59(52, D, E, A, B, C);
T_40_59(53, C, D, E, A, B);
T_40_59(54, B, C, D, E, A);
T_40_59(55, A, B, C, D, E);
T_40_59(56, E, A, B, C, D);
T_40_59(57, D, E, A, B, C);
T_40_59(58, C, D, E, A, B);
T_40_59(59, B, C, D, E, A);
/* Round 4 */
T_60_79(60, A, B, C, D, E);
T_60_79(61, E, A, B, C, D);
T_60_79(62, D, E, A, B, C);
T_60_79(63, C, D, E, A, B);
T_60_79(64, B, C, D, E, A);
T_60_79(65, A, B, C, D, E);
T_60_79(66, E, A, B, C, D);
T_60_79(67, D, E, A, B, C);
T_60_79(68, C, D, E, A, B);
T_60_79(69, B, C, D, E, A);
T_60_79(70, A, B, C, D, E);
T_60_79(71, E, A, B, C, D);
T_60_79(72, D, E, A, B, C);
T_60_79(73, C, D, E, A, B);
T_60_79(74, B, C, D, E, A);
T_60_79(75, A, B, C, D, E);
T_60_79(76, E, A, B, C, D);
T_60_79(77, D, E, A, B, C);
T_60_79(78, C, D, E, A, B);
T_60_79(79, B, C, D, E, A);
ctx->H[0] += A;
ctx->H[1] += B;
ctx->H[2] += C;
ctx->H[3] += D;
ctx->H[4] += E;
}
View Git Blame Copy Permalink