Rename alg_amd64 to asm_amd64

Namespace exposed SHA-1 function with botan_ prefix.

Add start of mp_monty for x86-64 assembly

6 files changed, 117 insertions, 8 deletions

diff --git a/modules/alg_amd64/asm_macr.h b/modules/asm_amd64/asm_macr.h
index 89e5bd028..087f3f44e 100644
--- a/modules/alg_amd64/asm_macr.h
+++ b/modules/asm_amd64/asm_macr.h
@@ -49,10 +49,12 @@ func_name:
 #define R0  %rax
 #define R1  %rbx
 #define R2  %rcx
+#define R2_32  %ecx
 #define R3  %rdx
 #define R4  %rsp
 #define R5  %rbp
 #define R6  %rsi
+#define R6_32  %esi
 #define R7  %rdi
 #define R8  %r8
 #define R9  %r9
@@ -66,8 +68,10 @@ func_name:
 
 #define ARG_1 R7
 #define ARG_2 R6
+#define ARG_2_32 R6_32
 #define ARG_3 R3
 #define ARG_4 R2
+#define ARG_4_32 R2_32
 #define ARG_5 R8
 #define ARG_6 R9
 
diff --git a/modules/alg_amd64/modinfo.txt b/modules/asm_amd64/modinfo.txt
index 929d696d0..625c9f162 100644
--- a/modules/alg_amd64/modinfo.txt
+++ b/modules/asm_amd64/modinfo.txt
@@ -9,13 +9,15 @@ sha160.cpp
 </replace>
 
 <ignore>
-#mp_mulop.cpp
+mp_mulop.cpp
+#mp_monty.cpp
 </ignore>
 
 <add>
 asm_macr.h
-#mp_mulop.S
-sha1core.S
+mp_mulop.S
+#mp_monty.S
+sha1_asm.S
 </add>
 
 <arch>
diff --git a/modules/asm_amd64/mp_monty.S b/modules/asm_amd64/mp_monty.S
new file mode 100644
index 000000000..9c94d2736
--- /dev/null
+++ b/modules/asm_amd64/mp_monty.S
@@ -0,0 +1,103 @@
+/*************************************************
+* Montgomery Reduction Source File               *
+* (C) 2008 Jack Lloyd                            *
+*************************************************/
+
+#include <botan/asm_macr.h>
+
+START_LISTING(mp_monty.S)
+
+START_FUNCTION(bigint_monty_redc)
+
+#define Z_ARR    ARG_1 // rdi
+#define Z_SIZE   ARG_2_32 // esi
+#define X_ARR    ARG_3 // rdx
+#define X_SIZE   ARG_4_32 // ecx
+#define U        ARG_5 // r8
+
+/*
+     We need all arguments for a while (we can reuse U eventually)
+   So only temp registers are
+     TEMP_1 %r10
+     TEMP_2 %r11
+     TEMP_3 = ARG_6 = %r9
+   void return, so also
+     R0 %rax (aka TEMP_9)
+   is free
+
+   Can push:
+     %rbx (base pointer, callee saved)
+     %rpb (frame pointer, callee saved)
+     %r12-%r15 (callee saved)
+
+  Can push base/frame pointers since this is a leaf function
+  and does not reference any data.
+*/
+
+   push %r12
+   push %r13
+   push %r14
+   push %r15
+
+#define LOOP_CTR R0
+
+#define CARRY    TEMP_1
+#define Z_WORD   TEMP_2
+#define MUL_LO   %rax
+#define MUL_HI   %rdx
+
+   /*
+   ZEROIZE(CARRY)
+
+   ASSIGN(LOOP_CTR, X_SIZE)
+
+   JUMP_IF_ZERO(LOOP_CTR, .L_MULADD_DONE)
+   JUMP_IF_LT(LOOP_CTR, 8, .LOOP_MULADD1)
+
+#define MULADD_OP(N)                  \
+   ASSIGN(MUL_LO, ARRAY8(X_ARR, N)) ; \
+   ASSIGN(Z_WORD, ARRAY8(Z_ARR, N)) ; \
+   MUL(Y)                           ; \
+   ADD(Z_WORD, CARRY)               ; \
+   ASSIGN(CARRY, MUL_HI)            ; \
+   ADD_LAST_CARRY(CARRY)            ; \
+   ADD(Z_WORD, MUL_LO)              ; \
+   ADD_LAST_CARRY(CARRY)            ; \
+   ASSIGN(ARRAY8(Z_ARR, N), Z_WORD)
+
+ALIGN
+.LOOP_MULADD8:
+   MULADD_OP(0)
+   MULADD_OP(1)
+   MULADD_OP(2)
+   MULADD_OP(3)
+   MULADD_OP(4)
+   MULADD_OP(5)
+   MULADD_OP(6)
+   MULADD_OP(7)
+
+   SUB_IMM(LOOP_CTR, 8)
+   ADD_IMM(Z_ARR, 64)
+   ADD_IMM(X_ARR, 64)
+   cmp IMM(8), LOOP_CTR
+   jge .LOOP_MULADD8
+
+   JUMP_IF_ZERO(LOOP_CTR, .L_MULADD_DONE)
+
+ALIGN
+.LOOP_MULADD1:
+   MULADD_OP(0)
+
+   SUB_IMM(LOOP_CTR, 1)
+   ADD_IMM(Z_ARR, 8)
+   ADD_IMM(X_ARR, 8)
+
+   cmp IMM(0), LOOP_CTR
+   jne .LOOP_MULADD1
+*/
+
+   pop %r15
+   pop %r14
+   pop %r13
+   pop %r12
+END_FUNCTION(bigint_monty_redc)
diff --git a/modules/alg_amd64/mp_mulop.S b/modules/asm_amd64/mp_mulop.S
index 983a34a82..983a34a82 100644
--- a/modules/alg_amd64/mp_mulop.S
+++ b/modules/asm_amd64/mp_mulop.S
diff --git a/modules/alg_amd64/sha160.cpp b/modules/asm_amd64/sha160.cpp
index 97ec9d522..cfac02f45 100644
--- a/modules/alg_amd64/sha160.cpp
+++ b/modules/asm_amd64/sha160.cpp
@@ -8,14 +8,14 @@
 
 namespace Botan {
 
-extern "C" void botan_asm_sha160_core(u32bit[5], const byte[64], u32bit[80]);
+extern "C" void botan_sha160_asm_amd64(u32bit[5], const byte[64], u32bit[80]);
 
 /*************************************************
 * SHA-160 Compression Function                   *
 *************************************************/
 void SHA_160::hash(const byte input[])
    {
-   botan_asm_sha160_core(digest, input, W);
+   botan_sha160_asm_amd64(digest, input, W);
    }
 
 /*************************************************
diff --git a/modules/alg_amd64/sha1core.S b/modules/asm_amd64/sha1_asm.S
index f145f8517..ecf4a18ce 100644
--- a/modules/alg_amd64/sha1core.S
+++ b/modules/asm_amd64/sha1_asm.S
@@ -5,9 +5,9 @@
 
 #include <botan/asm_macr.h>
 
-START_LISTING(sha1core.S)
+START_LISTING(sha1_asm.S)
 
-START_FUNCTION(botan_asm_sha160_core)
+START_FUNCTION(botan_sha160_asm_amd64)
 
 #define DIGEST_ARR %rdi
 #define INPUT %rsi
@@ -255,4 +255,4 @@ ALIGN;
    ADD(ARRAY4(DIGEST_ARR, 3), B)
    ADD(ARRAY4(DIGEST_ARR, 4), C)
 
-END_FUNCTION(botan_asm_sha160_core)
+END_FUNCTION(botan_sha160_asm_amd64)