fmtlib/fmt
1
0
Fork 1
mirror of https://github.com/fmtlib/fmt.git synced 2025-07-30 02:37:36 +02:00
Code Issues Packages Projects Releases Wiki Activity

Avoid unnecessary unsigned overflows (#1515)

Browse Source
This commit is contained in:
Victor Zverovich
2020-01-18 08:19:27 -08:00
parent 9bd9738da0
commit 75765bfad5
1 changed files with 7 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
include/fmt/format-inl.h
View File

@ -386,7 +386,8 @@ class fp {
const int exponent_bias = (1 << exponent_size) - limits::max_exponent - 1; const int exponent_bias = (1 << exponent_size) - limits::max_exponent - 1;
auto u = bit_cast<uint64_t>(d); auto u = bit_cast<uint64_t>(d);
f = u & significand_mask; f = u & significand_mask;
auto biased_e = (u & exponent_mask) >> double_significand_size; int biased_e =
static_cast<int>((u & exponent_mask) >> double_significand_size);
// Predecessor is closer if d is a normalized power of 2 (f == 0) other than // Predecessor is closer if d is a normalized power of 2 (f == 0) other than
// the smallest normalized number (biased_e > 1). // the smallest normalized number (biased_e > 1).
bool is_predecessor_closer = f == 0 && biased_e > 1; bool is_predecessor_closer = f == 0 && biased_e > 1;
@ -394,7 +395,7 @@ class fp {
f += implicit_bit; f += implicit_bit;
else else
biased_e = 1; // Subnormals use biased exponent 1 (min exponent). biased_e = 1; // Subnormals use biased exponent 1 (min exponent).
e = static_cast<int>(biased_e - exponent_bias - double_significand_size); e = biased_e - exponent_bias - double_significand_size;
return is_predecessor_closer; return is_predecessor_closer;
} }
@ -458,13 +459,11 @@ inline fp operator*(fp x, fp y) { return {multiply(x.f, y.f), x.e + y.e + 64}; }
// Returns a cached power of 10 `c_k = c_k.f * pow(2, c_k.e)` such that its // Returns a cached power of 10 `c_k = c_k.f * pow(2, c_k.e)` such that its
// (binary) exponent satisfies `min_exponent <= c_k.e <= min_exponent + 28`. // (binary) exponent satisfies `min_exponent <= c_k.e <= min_exponent + 28`.
inline fp get_cached_power(int min_exponent, int& pow10_exponent) { inline fp get_cached_power(int min_exponent, int& pow10_exponent) {
const uint64_t one_over_log2_10 = 0x4d104d42; // round(pow(2, 32) / log2(10)) const int64_t one_over_log2_10 = 0x4d104d42; // round(pow(2, 32) / log2(10))
int index = static_cast<int>( int index = static_cast<int>(
static_cast<int64_t>( ((min_exponent + fp::significand_size - 1) * one_over_log2_10 +
(min_exponent + fp::significand_size - 1) * one_over_log2_10 + ((int64_t(1) << 32) - 1)) // ceil
((uint64_t(1) << 32) - 1) // ceil >> 32 // arithmetic shift
) >>
32 // arithmetic shift
); );
// Decimal exponent of the first (smallest) cached power of 10. // Decimal exponent of the first (smallest) cached power of 10.
const int first_dec_exp = -348; const int first_dec_exp = -348;