Add a (working, optimized) x86 version of MD4

3 files changed, 182 insertions, 2 deletions

diff --git a/modules/alg_ia32/md4.cpp b/modules/alg_ia32/md4.cpp
new file mode 100644
index 000000000..fb6c77a31
--- /dev/null
+++ b/modules/alg_ia32/md4.cpp
@@ -0,0 +1,43 @@
+/*************************************************
+* MD4 Source File                                *
+* (C) 1999-2006 The Botan Project                *
+*************************************************/
+
+#include <botan/md4.h>
+#include <botan/bit_ops.h>
+
+namespace Botan {
+
+extern "C" void md4_core(u32bit[4], const byte[64], u32bit[16]);
+
+/*************************************************
+* MD4 Compression Function                       *
+*************************************************/
+void MD4::hash(const byte input[])
+   {
+   md4_core(digest, input, M);
+   }
+
+/*************************************************
+* Copy out the digest                            *
+*************************************************/
+void MD4::copy_out(byte output[])
+   {
+   for(u32bit j = 0; j != OUTPUT_LENGTH; ++j)
+      output[j] = get_byte(3 - (j % 4), digest[j/4]);
+   }
+
+/*************************************************
+* Clear memory of sensitive data                 *
+*************************************************/
+void MD4::clear() throw()
+   {
+   MDx_HashFunction::clear();
+   M.clear();
+   digest[0] = 0x67452301;
+   digest[1] = 0xEFCDAB89;
+   digest[2] = 0x98BADCFE;
+   digest[3] = 0x10325476;
+   }
+
+}
diff --git a/modules/alg_ia32/md4core.S b/modules/alg_ia32/md4core.S
new file mode 100644
index 000000000..5689d810c
--- /dev/null
+++ b/modules/alg_ia32/md4core.S
@@ -0,0 +1,134 @@
+/*************************************************
+* MD4 Source File                                *
+* (C) 1999-2006 The Botan Project                *
+*************************************************/
+
+#include <botan/asm_macr.h>
+
+START_LISTING(md4core.S)
+
+START_FUNCTION(md4_core)
+   SPILL_REGS()
+
+#define PUSHED 4
+
+   ASSIGN(EBP, ARG(2)) /* input block */
+   ASSIGN(EDI, ARG(3)) /* expanded words */
+
+   ZEROIZE(ESI)
+
+START_LOOP(.LOAD_INPUT)
+   ADD_IMM(ESI, 4)
+
+   ASSIGN(EAX, ARRAY4(EBP, 0))
+   ASSIGN(EBX, ARRAY4(EBP, 1))
+   ASSIGN(ECX, ARRAY4(EBP, 2))
+   ASSIGN(EDX, ARRAY4(EBP, 3))
+
+   ADD_IMM(EBP, 16)
+
+   ASSIGN(ARRAY4_INDIRECT(EDI,ESI,-4), EAX)
+   ASSIGN(ARRAY4_INDIRECT(EDI,ESI,-3), EBX)
+   ASSIGN(ARRAY4_INDIRECT(EDI,ESI,-2), ECX)
+   ASSIGN(ARRAY4_INDIRECT(EDI,ESI,-1), EDX)
+LOOP_UNTIL(ESI, IMM(16), .LOAD_INPUT)
+
+   ASSIGN(EBP, ARG(1))
+   ASSIGN(EAX, ARRAY4(EBP, 0))
+   ASSIGN(EBX, ARRAY4(EBP, 1))
+   ASSIGN(ECX, ARRAY4(EBP, 2))
+   ASSIGN(EDX, ARRAY4(EBP, 3))
+
+#define T1 ESI
+#define T2 EBP
+
+#define FF(A, B, C, D, MSG, S)        \
+   ASSIGN(T1, ARRAY4(EDI, MSG))     ; \
+   ASSIGN(T2, C)                    ; \
+   XOR(T2, D)                       ; \
+   AND(T2, B)                       ; \
+   XOR(T2, D)                       ; \
+   ADD(A, T1)                       ; \
+   ADD(A, T2)                       ; \
+   ROTL_IMM(A, S)                   ;
+
+#define GG(A, B, C, D, MSG, S) \
+   ASSIGN(T1, ARRAY4(EDI, MSG))     ; \
+   ASSIGN(T2, B)                    ; \
+   OR(T2, C)                        ; \
+   AND(T2, D)                       ; \
+   ADD3_IMM(A, T1, 0x5A827999)      ; \
+   ASSIGN(T1, B)                    ; \
+   AND(T1, C)                       ; \
+   OR(T2, T1)                       ; \
+   ADD(A, T2)                       ; \
+   ROTL_IMM(A, S)                   ;
+
+#define HH(A, B, C, D, MSG, S) \
+   ASSIGN(T1, ARRAY4(EDI, MSG))     ; \
+   ASSIGN(T2, B)                    ; \
+   XOR(T2, C)                       ; \
+   XOR(T2, D)                       ; \
+   ADD3_IMM(A, T1, 0x6ED9EBA1)      ; \
+   ADD(A, T2)                       ; \
+   ROTL_IMM(A, S)                   ;
+
+   FF(EAX,EBX,ECX,EDX, 0, 3);
+   FF(EDX,EAX,EBX,ECX, 1, 7);
+   FF(ECX,EDX,EAX,EBX, 2,11);
+   FF(EBX,ECX,EDX,EAX, 3,19);
+   FF(EAX,EBX,ECX,EDX, 4, 3);
+   FF(EDX,EAX,EBX,ECX, 5, 7);
+   FF(ECX,EDX,EAX,EBX, 6,11);
+   FF(EBX,ECX,EDX,EAX, 7,19);
+   FF(EAX,EBX,ECX,EDX, 8, 3);
+   FF(EDX,EAX,EBX,ECX, 9, 7);
+   FF(ECX,EDX,EAX,EBX,10,11);
+   FF(EBX,ECX,EDX,EAX,11,19);
+   FF(EAX,EBX,ECX,EDX,12, 3);
+   FF(EDX,EAX,EBX,ECX,13, 7);
+   FF(ECX,EDX,EAX,EBX,14,11);
+   FF(EBX,ECX,EDX,EAX,15,19);
+
+   GG(EAX,EBX,ECX,EDX, 0, 3);
+   GG(EDX,EAX,EBX,ECX, 4, 5);
+   GG(ECX,EDX,EAX,EBX, 8, 9);
+   GG(EBX,ECX,EDX,EAX,12,13);
+   GG(EAX,EBX,ECX,EDX, 1, 3);
+   GG(EDX,EAX,EBX,ECX, 5, 5);
+   GG(ECX,EDX,EAX,EBX, 9, 9);
+   GG(EBX,ECX,EDX,EAX,13,13);
+   GG(EAX,EBX,ECX,EDX, 2, 3);
+   GG(EDX,EAX,EBX,ECX, 6, 5);
+   GG(ECX,EDX,EAX,EBX,10, 9);
+   GG(EBX,ECX,EDX,EAX,14,13);
+   GG(EAX,EBX,ECX,EDX, 3, 3);
+   GG(EDX,EAX,EBX,ECX, 7, 5);
+   GG(ECX,EDX,EAX,EBX,11, 9);
+   GG(EBX,ECX,EDX,EAX,15,13);
+
+   HH(EAX,EBX,ECX,EDX, 0, 3);
+   HH(EDX,EAX,EBX,ECX, 8, 9);
+   HH(ECX,EDX,EAX,EBX, 4,11);
+   HH(EBX,ECX,EDX,EAX,12,15);
+   HH(EAX,EBX,ECX,EDX, 2, 3);
+   HH(EDX,EAX,EBX,ECX,10, 9);
+   HH(ECX,EDX,EAX,EBX, 6,11);
+   HH(EBX,ECX,EDX,EAX,14,15);
+   HH(EAX,EBX,ECX,EDX, 1, 3);
+   HH(EDX,EAX,EBX,ECX, 9, 9);
+   HH(ECX,EDX,EAX,EBX, 5,11);
+   HH(EBX,ECX,EDX,EAX,13,15);
+   HH(EAX,EBX,ECX,EDX, 3, 3);
+   HH(EDX,EAX,EBX,ECX,11, 9);
+   HH(ECX,EDX,EAX,EBX, 7,11);
+   HH(EBX,ECX,EDX,EAX,15,15);
+
+   ASSIGN(EBP, ARG(1))
+   ADD(ARRAY4(EBP, 0), EAX)
+   ADD(ARRAY4(EBP, 1), EBX)
+   ADD(ARRAY4(EBP, 2), ECX)
+   ADD(ARRAY4(EBP, 3), EDX)
+
+   RESTORE_REGS()
+END_FUNCTION(sha160_core)
diff --git a/modules/alg_ia32/modinfo.txt b/modules/alg_ia32/modinfo.txt
index 7d30fb624..5864c574c 100644
--- a/modules/alg_ia32/modinfo.txt
+++ b/modules/alg_ia32/modinfo.txt
@@ -1,11 +1,14 @@
 realname "Algorithm x86 Assembler"
 
-replace_file sha160.cpp
+replace_file md4.cpp
 replace_file md5.cpp
+replace_file sha160.cpp
 
 add_file asm_macr.h
-add_file sha1core.S
+
+add_file md4core.S
 add_file md5core.S
+add_file sha1core.S
 
 <arch>
 ia32