Merge pull request #513 from fkautz/pr_out_adding_software_fallback

pkg/utils/crypto/sha256/sha256_linux.go

@@ -65,6 +65,8 @@ func block(dig *digest, p []byte) {
 		blockAVX(dig, p)
 	case cpu.HasSSE41() == true:
 		blockSSE(dig, p)
+	default:
+		blockSoftware(dig, p)
 	}
 }
 

pkg/utils/crypto/sha256/sha256block_linux.go

@@ -14,7 +14,13 @@
 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 // See the License for the specific language governing permissions and
 // limitations under the License.
-//
+
+// Software block transform are provided by The Go Authors:
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file of
+// Golang project:
+//    https://github.com/golang/go/blob/master/LICENSE
 
 package sha256
 
@@ -37,3 +43,120 @@ func blockAVX(dig *digest, p []byte) {
 func blockAVX2(dig *digest, p []byte) {
 	C.sha256_transform_rorx((*C.char)(unsafe.Pointer(&p[0])), (*C.uint32_t)(unsafe.Pointer(&dig.h[0])), (C.ulong)(len(p)/64))
 }
+
+func blockSoftware(dig *digest, p []byte) {
+	var w [64]uint32
+	h0, h1, h2, h3, h4, h5, h6, h7 := dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7]
+	for len(p) >= chunk {
+		// Can interlace the computation of w with the
+		// rounds below if needed for speed.
+		for i := 0; i < 16; i++ {
+			j := i * 4
+			w[i] = uint32(p[j])<<24 | uint32(p[j+1])<<16 | uint32(p[j+2])<<8 | uint32(p[j+3])
+		}
+		for i := 16; i < 64; i++ {
+			v1 := w[i-2]
+			t1 := (v1>>17 | v1<<(32-17)) ^ (v1>>19 | v1<<(32-19)) ^ (v1 >> 10)
+			v2 := w[i-15]
+			t2 := (v2>>7 | v2<<(32-7)) ^ (v2>>18 | v2<<(32-18)) ^ (v2 >> 3)
+			w[i] = t1 + w[i-7] + t2 + w[i-16]
+		}
+
+		a, b, c, d, e, f, g, h := h0, h1, h2, h3, h4, h5, h6, h7
+
+		for i := 0; i < 64; i++ {
+			t1 := h + ((e>>6 | e<<(32-6)) ^ (e>>11 | e<<(32-11)) ^ (e>>25 | e<<(32-25))) + ((e & f) ^ (^e & g)) + _K[i] + w[i]
+
+			t2 := ((a>>2 | a<<(32-2)) ^ (a>>13 | a<<(32-13)) ^ (a>>22 | a<<(32-22))) + ((a & b) ^ (a & c) ^ (b & c))
+
+			h = g
+			g = f
+			f = e
+			e = d + t1
+			d = c
+			c = b
+			b = a
+			a = t1 + t2
+		}
+
+		h0 += a
+		h1 += b
+		h2 += c
+		h3 += d
+		h4 += e
+		h5 += f
+		h6 += g
+		h7 += h
+
+		p = p[chunk:]
+	}
+
+	dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7] = h0, h1, h2, h3, h4, h5, h6, h7
+}
+
+var _K = []uint32{
+	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,
+}

pkg/utils/crypto/sha512/sha512_linux.go

@@ -52,6 +52,8 @@ func block(dig *digest, p []byte) {
 		blockAVX(dig, p)
 	case cpu.HasSSE41() == true:
 		blockSSE(dig, p)
+	default:
+		blockSoftware(dig, p)
 	}
 }
 

pkg/utils/crypto/sha512/sha512block_linux.go

@@ -15,6 +15,12 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 //
+// Software block transform are provided by The Go Authors:
+// Copyright 2009 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file of
+// Golang project:
+//    https://github.com/golang/go/blob/master/LICENSE
 
 package sha512
 
@@ -24,7 +30,9 @@ package sha512
 // void sha512_transform_avx (const void* M, void* D, uint64_t L);
 // void sha512_transform_rorx (const void* M, void* D, uint64_t L);
 import "C"
-import "unsafe"
+import (
+	"unsafe"
+)
 
 func blockSSE(dig *digest, p []byte) {
 	C.sha512_transform_ssse3(unsafe.Pointer(&p[0]), unsafe.Pointer(&dig.h[0]), (C.uint64_t)(len(p)/chunk))
@@ -37,3 +45,136 @@ func blockAVX(dig *digest, p []byte) {
 func blockAVX2(dig *digest, p []byte) {
 	C.sha512_transform_rorx(unsafe.Pointer(&p[0]), unsafe.Pointer(&dig.h[0]), (C.uint64_t)(len(p)/chunk))
 }
+
+func blockSoftware(dig *digest, p []byte) {
+	var w [80]uint64
+	h0, h1, h2, h3, h4, h5, h6, h7 := dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7]
+	for len(p) >= chunk {
+		for i := 0; i < 16; i++ {
+			j := i * 8
+			w[i] = uint64(p[j])<<56 | uint64(p[j+1])<<48 | uint64(p[j+2])<<40 | uint64(p[j+3])<<32 |
+				uint64(p[j+4])<<24 | uint64(p[j+5])<<16 | uint64(p[j+6])<<8 | uint64(p[j+7])
+		}
+		for i := 16; i < 80; i++ {
+			v1 := w[i-2]
+			t1 := (v1>>19 | v1<<(64-19)) ^ (v1>>61 | v1<<(64-61)) ^ (v1 >> 6)
+			v2 := w[i-15]
+			t2 := (v2>>1 | v2<<(64-1)) ^ (v2>>8 | v2<<(64-8)) ^ (v2 >> 7)
+
+			w[i] = t1 + w[i-7] + t2 + w[i-16]
+		}
+
+		a, b, c, d, e, f, g, h := h0, h1, h2, h3, h4, h5, h6, h7
+
+		for i := 0; i < 80; i++ {
+			t1 := h + ((e>>14 | e<<(64-14)) ^ (e>>18 | e<<(64-18)) ^ (e>>41 | e<<(64-41))) + ((e & f) ^ (^e & g)) + _K[i] + w[i]
+
+			t2 := ((a>>28 | a<<(64-28)) ^ (a>>34 | a<<(64-34)) ^ (a>>39 | a<<(64-39))) + ((a & b) ^ (a & c) ^ (b & c))
+
+			h = g
+			g = f
+			f = e
+			e = d + t1
+			d = c
+			c = b
+			b = a
+			a = t1 + t2
+		}
+
+		h0 += a
+		h1 += b
+		h2 += c
+		h3 += d
+		h4 += e
+		h5 += f
+		h6 += g
+		h7 += h
+
+		p = p[chunk:]
+	}
+
+	dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7] = h0, h1, h2, h3, h4, h5, h6, h7
+}
+
+var _K = []uint64{
+	0x428a2f98d728ae22,
+	0x7137449123ef65cd,
+	0xb5c0fbcfec4d3b2f,
+	0xe9b5dba58189dbbc,
+	0x3956c25bf348b538,
+	0x59f111f1b605d019,
+	0x923f82a4af194f9b,
+	0xab1c5ed5da6d8118,
+	0xd807aa98a3030242,
+	0x12835b0145706fbe,
+	0x243185be4ee4b28c,
+	0x550c7dc3d5ffb4e2,
+	0x72be5d74f27b896f,
+	0x80deb1fe3b1696b1,
+	0x9bdc06a725c71235,
+	0xc19bf174cf692694,
+	0xe49b69c19ef14ad2,
+	0xefbe4786384f25e3,
+	0x0fc19dc68b8cd5b5,
+	0x240ca1cc77ac9c65,
+	0x2de92c6f592b0275,
+	0x4a7484aa6ea6e483,
+	0x5cb0a9dcbd41fbd4,
+	0x76f988da831153b5,
+	0x983e5152ee66dfab,
+	0xa831c66d2db43210,
+	0xb00327c898fb213f,
+	0xbf597fc7beef0ee4,
+	0xc6e00bf33da88fc2,
+	0xd5a79147930aa725,
+	0x06ca6351e003826f,
+	0x142929670a0e6e70,
+	0x27b70a8546d22ffc,
+	0x2e1b21385c26c926,
+	0x4d2c6dfc5ac42aed,
+	0x53380d139d95b3df,
+	0x650a73548baf63de,
+	0x766a0abb3c77b2a8,
+	0x81c2c92e47edaee6,
+	0x92722c851482353b,
+	0xa2bfe8a14cf10364,
+	0xa81a664bbc423001,
+	0xc24b8b70d0f89791,
+	0xc76c51a30654be30,
+	0xd192e819d6ef5218,
+	0xd69906245565a910,
+	0xf40e35855771202a,
+	0x106aa07032bbd1b8,
+	0x19a4c116b8d2d0c8,
+	0x1e376c085141ab53,
+	0x2748774cdf8eeb99,
+	0x34b0bcb5e19b48a8,
+	0x391c0cb3c5c95a63,
+	0x4ed8aa4ae3418acb,
+	0x5b9cca4f7763e373,
+	0x682e6ff3d6b2b8a3,
+	0x748f82ee5defb2fc,
+	0x78a5636f43172f60,
+	0x84c87814a1f0ab72,
+	0x8cc702081a6439ec,
+	0x90befffa23631e28,
+	0xa4506cebde82bde9,
+	0xbef9a3f7b2c67915,
+	0xc67178f2e372532b,
+	0xca273eceea26619c,
+	0xd186b8c721c0c207,
+	0xeada7dd6cde0eb1e,
+	0xf57d4f7fee6ed178,
+	0x06f067aa72176fba,
+	0x0a637dc5a2c898a6,
+	0x113f9804bef90dae,
+	0x1b710b35131c471b,
+	0x28db77f523047d84,
+	0x32caab7b40c72493,
+	0x3c9ebe0a15c9bebc,
+	0x431d67c49c100d4c,
+	0x4cc5d4becb3e42b6,
+	0x597f299cfc657e2a,
+	0x5fcb6fab3ad6faec,
+	0x6c44198c4a475817,
+}