[master] 1f631aa Update our vsha256 with the latest improvements from Colin Percival:

Poul-Henning Kamp phk at FreeBSD.org
Sat May 6 00:52:05 CEST 2017


commit 1f631aa61dd170aa3bdd6485fbdb801c7cc36eb0
Author: Poul-Henning Kamp <phk at FreeBSD.org>
Date:   Fri May 5 22:50:03 2017 +0000

    Update our vsha256 with the latest improvements from Colin Percival:
    
    Retune SHA2 code for improved performance on CPUs with more ILP and
    a preference for memory load instructions over large code footprints
    with embedded immediate variables.
    
    On amd64 CPUs from 2007-2008 there is not a significant change, but
    amd64 CPUs from 2009-2010 get roughly 10% more throughput with this
    code; amd64 CPUs from 2011-2012 get roughly 15% more throughput; and
    AMD64 CPUs from 2013-2015 get 20-25% more throughput.  The Raspberry
    Pi 2 increases its throughput by 6-8%.
    
    Sponsored by:   Tarsnap Backup Inc.

diff --git a/include/vsha256.h b/include/vsha256.h
index bf42c11..e3852e9 100644
--- a/include/vsha256.h
+++ b/include/vsha256.h
@@ -30,6 +30,7 @@
 #define _SHA256_H_
 
 #define SHA256_LEN		32
+#define SHA256_DIGEST_LENGTH	32
 
 typedef struct SHA256Context {
 	uint32_t state[8];
diff --git a/lib/libvarnish/flint.lnt b/lib/libvarnish/flint.lnt
index 0a014ba..636828a 100644
--- a/lib/libvarnish/flint.lnt
+++ b/lib/libvarnish/flint.lnt
@@ -1,5 +1,10 @@
 -efunc(835, SHA256_Transform)	// A zero has been given as ___ argument to operator '___'
 -efunc(778, SHA256_Transform)	// Constant expression evaluates to 0 in operation '___'
+-efunc(661, SHA256_Transform)	// Possible access of out-of-bounds pointer
+-efunc(662, SHA256_Transform)	// Possible creation of out-of-bounds pointer
+-efunc(662, SHA256_Update)	// Possible creation of out-of-bounds pointer
+-efunc(662, be32dec_vect)	// Possible creation of out-of-bounds pointer
+-efunc(670, SHA256_Update)	// Possible access beyond array for function '___',
 +fan			// Allow anon unions
 
 -e713	// Loss of precision (assignment) (unsigned long long to long long)
@@ -7,3 +12,4 @@
 -dVARNISH_STATE_DIR="foo"
 
 --emacro((835),BINHEAP_NOIDX)
+--emacro((835),O_CLOEXEC)
diff --git a/lib/libvarnish/vsha256.c b/lib/libvarnish/vsha256.c
index 7e73710..c668474 100644
--- a/lib/libvarnish/vsha256.c
+++ b/lib/libvarnish/vsha256.c
@@ -23,7 +23,8 @@
  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  * SUCH DAMAGE.
  *
- * From: $FreeBSD: head/lib/libmd/sha256c.c 154479 2006-01-17 15:35:57Z phk $
+ * From:
+ * $FreeBSD: sys/crypto/sha2/sha256c.c 300966 2016-05-29 17:26:40Z cperciva $
  */
 
 #include "config.h"
@@ -84,7 +85,27 @@ be32dec_vect(uint32_t *dst, const unsigned char *src, size_t len)
 		dst[i] = vbe32dec(src + i * 4);
 }
 
-#endif
+#endif /* BYTE_ORDER != BIG_ENDIAN */
+
+/* SHA256 round constants. */
+static const uint32_t K[64] = {
+	0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
+	0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+	0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3,
+	0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+	0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc,
+	0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+	0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7,
+	0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+	0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13,
+	0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+	0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3,
+	0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+	0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5,
+	0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+	0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208,
+	0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+};
 
 /* Elementary functions used by SHA256 */
 #define Ch(x, y, z)	((x & (y ^ z)) ^ z)
@@ -97,19 +118,28 @@ be32dec_vect(uint32_t *dst, const unsigned char *src, size_t len)
 #define s1(x)		(ROTR(x, 17) ^ ROTR(x, 19) ^ SHR(x, 10))
 
 /* SHA256 round function */
-#define RND(a, b, c, d, e, f, g, h, k)			\
-	t0 = h + S1(e) + Ch(e, f, g) + k;		\
-	t1 = S0(a) + Maj(a, b, c);			\
-	d += t0;					\
-	h  = t0 + t1;
+#define RND(a, b, c, d, e, f, g, h, k)				\
+	do {							\
+		h += S1(e) + Ch(e, f, g) + (k);			\
+		d += h;						\
+		h += S0(a) + Maj(a, b, c);			\
+	} while (0)
 
 /* Adjusted round function for rotating state */
-#define RNDr(S, W, i, k)			\
+#define RNDr(S, W, i, ii)			\
 	RND(S[(64 - i) % 8], S[(65 - i) % 8],	\
 	    S[(66 - i) % 8], S[(67 - i) % 8],	\
 	    S[(68 - i) % 8], S[(69 - i) % 8],	\
 	    S[(70 - i) % 8], S[(71 - i) % 8],	\
-	    (W[i] + k))
+	    W[i + ii] + K[i + ii])
+
+/* Message schedule computation */
+#define MSCH(W, ii, i)						\
+	do {							\
+		W[i + ii + 16] =				\
+		    s1(W[i + ii + 14]) + W[i + ii + 9] +	\
+		    s0(W[i + ii + 1]) + W[i + ii];		\
+	} while (0)
 
 /*
  * SHA256 block compression function.  The 256-bit state is transformed via
@@ -120,82 +150,52 @@ SHA256_Transform(uint32_t * state, const unsigned char block[64])
 {
 	uint32_t W[64];
 	uint32_t S[8];
-	uint32_t t0, t1;
 	int i;
 
-	/* 1. Prepare message schedule W. */
+	/* 1. Prepare the first part of the message schedule W. */
 	be32dec_vect(W, block, 64);
-	for (i = 16; i < 64; i++)
-		W[i] = s1(W[i - 2]) + W[i - 7] + s0(W[i - 15]) + W[i - 16];
 
 	/* 2. Initialize working variables. */
 	memcpy(S, state, 32);
 
 	/* 3. Mix. */
-	RNDr(S, W, 0, 0x428a2f98);
-	RNDr(S, W, 1, 0x71374491);
-	RNDr(S, W, 2, 0xb5c0fbcf);
-	RNDr(S, W, 3, 0xe9b5dba5);
-	RNDr(S, W, 4, 0x3956c25b);
-	RNDr(S, W, 5, 0x59f111f1);
-	RNDr(S, W, 6, 0x923f82a4);
-	RNDr(S, W, 7, 0xab1c5ed5);
-	RNDr(S, W, 8, 0xd807aa98);
-	RNDr(S, W, 9, 0x12835b01);
-	RNDr(S, W, 10, 0x243185be);
-	RNDr(S, W, 11, 0x550c7dc3);
-	RNDr(S, W, 12, 0x72be5d74);
-	RNDr(S, W, 13, 0x80deb1fe);
-	RNDr(S, W, 14, 0x9bdc06a7);
-	RNDr(S, W, 15, 0xc19bf174);
-	RNDr(S, W, 16, 0xe49b69c1);
-	RNDr(S, W, 17, 0xefbe4786);
-	RNDr(S, W, 18, 0x0fc19dc6);
-	RNDr(S, W, 19, 0x240ca1cc);
-	RNDr(S, W, 20, 0x2de92c6f);
-	RNDr(S, W, 21, 0x4a7484aa);
-	RNDr(S, W, 22, 0x5cb0a9dc);
-	RNDr(S, W, 23, 0x76f988da);
-	RNDr(S, W, 24, 0x983e5152);
-	RNDr(S, W, 25, 0xa831c66d);
-	RNDr(S, W, 26, 0xb00327c8);
-	RNDr(S, W, 27, 0xbf597fc7);
-	RNDr(S, W, 28, 0xc6e00bf3);
-	RNDr(S, W, 29, 0xd5a79147);
-	RNDr(S, W, 30, 0x06ca6351);
-	RNDr(S, W, 31, 0x14292967);
-	RNDr(S, W, 32, 0x27b70a85);
-	RNDr(S, W, 33, 0x2e1b2138);
-	RNDr(S, W, 34, 0x4d2c6dfc);
-	RNDr(S, W, 35, 0x53380d13);
-	RNDr(S, W, 36, 0x650a7354);
-	RNDr(S, W, 37, 0x766a0abb);
-	RNDr(S, W, 38, 0x81c2c92e);
-	RNDr(S, W, 39, 0x92722c85);
-	RNDr(S, W, 40, 0xa2bfe8a1);
-	RNDr(S, W, 41, 0xa81a664b);
-	RNDr(S, W, 42, 0xc24b8b70);
-	RNDr(S, W, 43, 0xc76c51a3);
-	RNDr(S, W, 44, 0xd192e819);
-	RNDr(S, W, 45, 0xd6990624);
-	RNDr(S, W, 46, 0xf40e3585);
-	RNDr(S, W, 47, 0x106aa070);
-	RNDr(S, W, 48, 0x19a4c116);
-	RNDr(S, W, 49, 0x1e376c08);
-	RNDr(S, W, 50, 0x2748774c);
-	RNDr(S, W, 51, 0x34b0bcb5);
-	RNDr(S, W, 52, 0x391c0cb3);
-	RNDr(S, W, 53, 0x4ed8aa4a);
-	RNDr(S, W, 54, 0x5b9cca4f);
-	RNDr(S, W, 55, 0x682e6ff3);
-	RNDr(S, W, 56, 0x748f82ee);
-	RNDr(S, W, 57, 0x78a5636f);
-	RNDr(S, W, 58, 0x84c87814);
-	RNDr(S, W, 59, 0x8cc70208);
-	RNDr(S, W, 60, 0x90befffa);
-	RNDr(S, W, 61, 0xa4506ceb);
-	RNDr(S, W, 62, 0xbef9a3f7);
-	RNDr(S, W, 63, 0xc67178f2);
+	for (i = 0; i < 64; i += 16) {
+		RNDr(S, W, 0, i);
+		RNDr(S, W, 1, i);
+		RNDr(S, W, 2, i);
+		RNDr(S, W, 3, i);
+		RNDr(S, W, 4, i);
+		RNDr(S, W, 5, i);
+		RNDr(S, W, 6, i);
+		RNDr(S, W, 7, i);
+		RNDr(S, W, 8, i);
+		RNDr(S, W, 9, i);
+		RNDr(S, W, 10, i);
+		RNDr(S, W, 11, i);
+		RNDr(S, W, 12, i);
+		RNDr(S, W, 13, i);
+		RNDr(S, W, 14, i);
+		RNDr(S, W, 15, i);
+
+		if (i == 48)
+			break;
+		MSCH(W, 0, i);
+		MSCH(W, 1, i);
+		MSCH(W, 2, i);
+		MSCH(W, 3, i);
+		MSCH(W, 4, i);
+		MSCH(W, 5, i);
+		MSCH(W, 6, i);
+		MSCH(W, 7, i);
+		MSCH(W, 8, i);
+		MSCH(W, 9, i);
+		MSCH(W, 10, i);
+		MSCH(W, 11, i);
+		MSCH(W, 12, i);
+		MSCH(W, 13, i);
+		MSCH(W, 14, i);
+		MSCH(W, 15, i);
+	}
 
 	/* 4. Mix local working variables into global state */
 	for (i = 0; i < 8; i++)
@@ -213,23 +213,29 @@ static const unsigned char PAD[64] = {
 static void
 SHA256_Pad(SHA256_CTX * ctx)
 {
-	unsigned char len[8];
-	uint32_t r, plen;
-
-	/*
-	 * Convert length to bits and encode as a vector of bytes
-	 * -- we do this now rather than later because the length
-	 * will change after we pad.
-	 */
-	vbe64enc(len, ctx->count << 3);
-
-	/* Add 1--64 bytes so that the resulting length is 56 mod 64 */
-	r = ctx->count & 0x3f;
-	plen = (r < 56) ? (56 - r) : (120 - r);
-	SHA256_Update(ctx, PAD, (size_t)plen);
-
-	/* Add the terminating bit-count */
-	SHA256_Update(ctx, len, 8);
+	size_t r;
+
+	/* Figure out how many bytes we have buffered. */
+	r = (ctx->count >> 3) & 0x3f;
+
+	/* Pad to 56 mod 64, transforming if we finish a block en route. */
+	if (r < 56) {
+		/* Pad to 56 mod 64. */
+		memcpy(&ctx->buf[r], PAD, 56 - r);
+	} else {
+		/* Finish the current block and mix. */
+		memcpy(&ctx->buf[r], PAD, 64 - r);
+		SHA256_Transform(ctx->state, ctx->buf);
+
+		/* The start of the final block is all zeroes. */
+		memset(&ctx->buf[0], 0, 56);
+	}
+
+	/* Add the terminating bit-count. */
+	be64enc(&ctx->buf[56], ctx->count);
+
+	/* Mix in the final block. */
+	SHA256_Transform(ctx->state, ctx->buf);
 }
 
 /* SHA-256 initialization.  Begins a SHA-256 operation. */
@@ -255,23 +261,40 @@ SHA256_Init(SHA256_CTX * ctx)
 void
 SHA256_Update(SHA256_CTX * ctx, const void *in, size_t len)
 {
-	uint32_t r, l;
+	uint64_t bitlen;
+	uint32_t r;
 	const unsigned char *src = in;
 
 	/* Number of bytes left in the buffer from previous updates */
-	r = ctx->count & 0x3f;
-	while (len > 0) {
-		l = 64 - r;
-		if (l > len)
-			l = len;
-		memcpy(&ctx->buf[r], src, l);
-		len -= l;
-		src += l;
-		ctx->count += l;
-		r = ctx->count & 0x3f;
-		if (r == 0)
-			SHA256_Transform(ctx->state, ctx->buf);
+	r = (ctx->count >> 3) & 0x3f;
+
+	/* Convert the length into a number of bits */
+	bitlen = len << 3;
+
+	/* Update number of bits */
+	ctx->count += bitlen;
+
+	/* Handle the case where we don't need to perform any transforms */
+	if (len < 64 - r) {
+		memcpy(&ctx->buf[r], src, len);
+		return;
 	}
+
+	/* Finish the current block */
+	memcpy(&ctx->buf[r], src, 64 - r);
+	SHA256_Transform(ctx->state, ctx->buf);
+	src += 64 - r;
+	len -= 64 - r;
+
+	/* Perform complete blocks */
+	while (len >= 64) {
+		SHA256_Transform(ctx->state, src);
+		src += 64;
+		len -= 64;
+	}
+
+	/* Copy left over data into buffer */
+	memcpy(ctx->buf, src, len);
 }
 
 /*
@@ -279,14 +302,14 @@ SHA256_Update(SHA256_CTX * ctx, const void *in, size_t len)
  * and clears the context state.
  */
 void
-SHA256_Final(unsigned char digest[32], SHA256_CTX * ctx)
+SHA256_Final(unsigned char digest[static SHA256_DIGEST_LENGTH], SHA256_CTX *ctx)
 {
 
 	/* Add padding */
 	SHA256_Pad(ctx);
 
 	/* Write the hash */
-	be32enc_vect(digest, ctx->state, 32);
+	be32enc_vect(digest, ctx->state, SHA256_DIGEST_LENGTH);
 
 	/* Clear the context state */
 	memset((void *)ctx, 0, sizeof(*ctx));
diff --git a/tools/audit_vsha256.sh b/tools/audit_vsha256.sh
new file mode 100644
index 0000000..a81c8c0
--- /dev/null
+++ b/tools/audit_vsha256.sh
@@ -0,0 +1,11 @@
+#!/bin/sh
+#
+# Script to compare vsha256.c with FreeBSD's sha256c.c
+#
+# Run this on a up-to-date FreeBSD source tree
+
+sed '
+s/vbe32/be32/g
+' lib/libvarnish/vsha256.c |
+    diff -ub /usr/src/sys/crypto/sha2/sha256c.c -
+



More information about the varnish-commit mailing list