optimize floatscan downscaler to skip results that won't be needed

[musl] / src / internal / floatscan.c

diff --git a/src/internal/floatscan.c b/src/internal/floatscan.c
index 3aa5408..b231329 100644 (file)
--- a/src/internal/floatscan.c
+++ b/src/internal/floatscan.c
@@ -77,7 +77,7 @@ static long double decfloat(FILE *f, int c, int bits, int emin, int sign, int po
                if (c == '.') {
                        if (lrp!=-1) break;
                        lrp = dc;
-               } else if (k < KMAX) {
+               } else if (k < KMAX-2) {
                        dc++;
                        if (j) x[k] = x[k]*10 + c-'0';
                        else x[k] = c-'0';
@@ -88,7 +88,7 @@ static long double decfloat(FILE *f, int c, int bits, int emin, int sign, int po
                        gotdig=1;
                } else {
                        dc++;
-                       x[KMAX-1] |= c-'0';
+                       if (c!='0') x[KMAX-3] |= 1;
                }
        }
        if (lrp==-1) lrp=dc;
@@ -138,7 +138,28 @@ static long double decfloat(FILE *f, int c, int bits, int emin, int sign, int po
        e2 = 0;
        rp = lrp;
 
-       while (rp < 18+9*LD_B1B_DIG) {
+       if (rp % 9) {
+               static const int p10s[] = {
+                       100000000, 10000000, 1000000, 100000,
+                       10000, 1000, 100, 10
+               };
+               int rpm9 = rp>=0 ? rp%9 : rp%9+9;
+               int p10 = p10s[rpm9-1];
+               uint32_t carry = 0;
+               for (k=a; k!=z; k++) {
+                       uint32_t tmp = x[k] % p10;
+                       x[k] = x[k]/p10 + carry;
+                       carry = 1000000000/p10 * tmp;
+                       if (k==a && !x[k]) {
+                               a = (a+1 & MASK);
+                               rp -= 9;
+                       }
+               }
+               if (carry) x[z++] = carry;
+               rp += 9-rpm9;
+       }
+
+       while (rp < 9*LD_B1B_DIG || (rp == 9*LD_B1B_DIG && x[0]<th[0])) {
                uint32_t carry = 0;
                e2 -= 29;
                for (k=(z-1 & MASK); ; k=(k-1 & MASK)) {
@@ -164,32 +185,6 @@ static long double decfloat(FILE *f, int c, int bits, int emin, int sign, int po
                }
        }
 
-       if (rp % 9) {
-               static const int p10s[] = {
-                       100000000, 10000000, 1000000, 100000,
-                       10000, 1000, 100, 10
-               };
-               int rpm9 = rp % 9;
-               int p10 = p10s[rpm9-1];
-               uint32_t carry = 0;
-               for (k=a; k!=z; k=(k+1 & MASK)) {
-                       uint32_t tmp = x[k] % p10;
-                       x[k] = x[k]/p10 + carry;
-                       carry = 1000000000/p10 * tmp;
-                       if (k==a && !x[k]) {
-                               a = (a+1 & MASK);
-                               rp -= 9;
-                       }
-               }
-               if (carry) {
-                       if ((z+1 & MASK) != a) {
-                               x[z] = carry;
-                               z = (z+1 & MASK);
-                       } else x[z-1 & MASK] |= 1;
-               }
-               rp += 9-rpm9;
-       }
-
        for (;;) {
                uint32_t carry = 0;
                int sh = 1;
@@ -205,16 +200,17 @@ static long double decfloat(FILE *f, int c, int bits, int emin, int sign, int po
                /* FIXME: find a way to compute optimal sh */
                if (rp > 9+9*LD_B1B_DIG) sh = 9;
                e2 += sh;
-               for (k=a; k!=z; k=(k+1 & MASK)) {
+               for (i=0; (k=(a+i & MASK))!=z && i<LD_B1B_DIG+3; i++) {
                        uint32_t tmp = x[k] & (1<<sh)-1;
                        x[k] = (x[k]>>sh) + carry;
                        carry = (1000000000>>sh) * tmp;
                        if (k==a && !x[k]) {
                                a = (a+1 & MASK);
+                               i--;
                                rp -= 9;
                        }
                }
-               if (carry) {
+               if (carry && k==z) {
                        if ((z+1 & MASK) != a) {
                                x[z] = carry;
                                z = (z+1 & MASK);