use SC_LIMB in generated tables. Add a new "TODO" category, UNIFY for code that might be unified with other code

Makefile

@@ -266,7 +266,7 @@ doc: Doxyfile $(BUILD_OBJ)/timestamp $(HEADERS)
 # 	(cd $(BATNAME)/.. && tar czf $(BATBASE).tgz $(BATBASE) )
 	
 # Finds todo items in .h and .c files
-TODO_TYPES ?= HACK TODO FIXME BUG XXX PERF FUTURE REMOVE MAGIC
+TODO_TYPES ?= HACK TODO FIXME BUG XXX PERF FUTURE REMOVE MAGIC UNIFY
 TODO_LOCATIONS ?= src test Makefile Doxyfile
 todo::
 	@(find $(TODO_LOCATIONS) -name '*.h' -or -name '*.c' -or -name '*.cxx' -or -name '*.hxx') | xargs egrep --color=auto -w \

src/decaf.c

@@ -38,10 +38,6 @@
 extern const gf SQRT_MINUS_ONE;
 #endif
 
-#if COFACTOR == 8
-extern const gf SQRT_ONE_MINUS_D; /* TODO: Intern this? */
-#endif
-
 /* FIXME: this can be different from DECAF_WORD_BITS, and word_t can be different from decaf_word_t,
  * eg when mixing and matching implementations for different curves.  Homogenize this.
  */

src/decaf_gen_tables.c

@@ -37,18 +37,28 @@ void API_NS(precompute_wnafs) (
     const API_NS(point_t) base
 );
 
-/* TODO: use SC_LIMB? */
-static void scalar_print(const char *name, const API_NS(scalar_t) sc) {
+static void scalar_print(const char *name, const API_NS(scalar_t) sc) { /* UNIFY */
     printf("const API_NS(scalar_t) %s = {{{\n", name);
-    unsigned i;
-    for (i=0; i<sizeof(API_NS(scalar_t))/sizeof(decaf_word_t); i++) {
-        if (i) printf(", ");
-        printf("0x%0*llxull", (int)sizeof(decaf_word_t)*2, (unsigned long long)sc->limb[i] );
+    const int SCALAR_BYTES = (SCALAR_BITS + 7) / 8;
+    unsigned char ser[SCALAR_BYTES];
+    API_NS(scalar_encode)(ser,sc);
+    int b=0, i, comma=0;
+    unsigned long long limb = 0;
+    for (i=0; i<SCALAR_BYTES; i++) {
+        limb |= ((uint64_t)ser[i])<<b;
+        b += 8;
+        if (b == 64 || i==SCALAR_BYTES-1) {
+            b = 0;
+            if (comma) printf(",");
+            comma = 1;
+            printf("SC_LIMB(0x%016llx)", limb);
+            limb = ((uint64_t)ser[i])>>(8-b);
+        }
     }
     printf("}}};\n\n");
 }
 
-static void field_print(const gf f) {
+static void field_print(const gf f) { /* UNIFY */
     const int GF_SER_BYTES = (GF_BITS + 7) / 8;
     unsigned char ser[GF_SER_BYTES];
     gf_serialize(ser,f);
@@ -58,7 +68,7 @@ static void field_print(const gf f) {
     for (i=0; i<GF_SER_BYTES; i++) {
         limb |= ((uint64_t)ser[i])<<b;
         b += 8;
-        if (b >= GF_LIT_LIMB_BITS) {
+        if (b >= GF_LIT_LIMB_BITS || i == GF_SER_BYTES-1) {
             limb &= (1ull<<GF_LIT_LIMB_BITS) -1;
             b -= GF_LIT_LIMB_BITS;
             if (comma) printf(",");

src/include/arch_x86_64/arch_intrinsics.h

@@ -9,10 +9,7 @@
 
 #include <stdint.h>
 
-/* FUTURE: non x86-64 versions of these.
- * FUTURE: autogenerate
- */
-
+/* FUTURE: autogenerate */
 static __inline__ __uint128_t widemul(const uint64_t *a, const uint64_t *b) {
   uint64_t c,d;
   #ifndef __BMI2__