Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
| Comment: | Restore FP control word on conversion of zero values.
Sneak path existed that failed to restore the floating point control word when scanning a zero value. The result was that floating point rounding was incorrectly left set to 53-bit significance, round-to-even, which is incompatible with the math library from musl. |
|---|---|
| Downloads: | Tarball | ZIP archive | SQL archive |
| Timelines: | family | ancestors | descendants | both | bug-713653b951 |
| Files: | files | file ages | folders |
| SHA3-256: |
50ecdc3e87fb5d772490b5490090d3ce |
| User & Date: | kbk 2022-07-15 17:11:41 |
Context
|
2022-07-17
| ||
| 12:57 | Fix [713653b951]: Floating point precision problems on x86 musl check-in: 5e9b6b560e user: jan.nijtmans tags: core-8-5-branch | |
|
2022-07-15
| ||
| 17:11 |
Restore FP control word on conversion of zero values.
Sneak path existed that failed to restore the... Closed-Leaf check-in: 50ecdc3e87 user: kbk tags: bug-713653b951 | |
|
2022-07-12
| ||
| 08:20 | Code cleanup (use {} in if/else statemenets) check-in: ba232277ee user: jan.nijtmans tags: core-8-6-branch | |
Changes
Changes to generic/tclStrToD.c.
| ︙ | ︙ | |||
1623 1624 1625 1626 1627 1628 1629 |
static double
MakeLowPrecisionDouble(
int signum, /* 1 if the number is negative, 0 otherwise */
Tcl_WideUInt significand, /* Significand of the number */
int numSigDigs, /* Number of digits in the significand */
long exponent) /* Power of ten */
{
| < | > | < | > < > > > > > > > | 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 |
static double
MakeLowPrecisionDouble(
int signum, /* 1 if the number is negative, 0 otherwise */
Tcl_WideUInt significand, /* Significand of the number */
int numSigDigs, /* Number of digits in the significand */
long exponent) /* Power of ten */
{
mp_int significandBig; /* Significand expressed as a bignum. */
/*
* With gcc on x86, the floating point rounding mode is double-extended.
* This causes the result of double-precision calculations to be rounded
* twice: once to the precision of double-extended and then again to the
* precision of double. Double-rounding introduces gratuitous errors of 1
* ulp, so we need to change rounding mode to 53-bits. We also make
* 'retval' volatile, so that it doesn't get promoted to a register.
*/
volatile double retval; /* Value of the number. */
/*
* Test for zero significand, which requires explicit construction
* of -0.0. (Unary minus returns a positive zero.)
*/
if (significand == 0) {
return copysign(0.0, -signum);
}
/*
* Set the FP control word for 53 bits, WARNING: It must be reset
* before returning.
*/
TCL_IEEE_DOUBLE_ROUNDING;
if (numSigDigs <= QUICK_MAX) {
if (exponent >= 0) {
if (exponent <= mmaxpow) {
/*
* The significand is an exact integer, and so is
* 10**exponent. The product will be correct to within 1/2 ulp
* without special handling.
|
| ︙ | ︙ | |||
1740 1741 1742 1743 1744 1745 1746 |
static double
MakeHighPrecisionDouble(
int signum, /* 1=negative, 0=nonnegative */
mp_int *significand, /* Exact significand of the number */
int numSigDigs, /* Number of significant digits */
long exponent) /* Power of 10 by which to multiply */
{
| < | > > > > > > > > > > > > > > > | < < < < | 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 |
static double
MakeHighPrecisionDouble(
int signum, /* 1=negative, 0=nonnegative */
mp_int *significand, /* Exact significand of the number */
int numSigDigs, /* Number of significant digits */
long exponent) /* Power of 10 by which to multiply */
{
int machexp; /* Machine exponent of a power of 10. */
/*
* With gcc on x86, the floating point rounding mode is double-extended.
* This causes the result of double-precision calculations to be rounded
* twice: once to the precision of double-extended and then again to the
* precision of double. Double-rounding introduces gratuitous errors of 1
* ulp, so we need to change rounding mode to 53-bits. We also make
* 'retval' volatile to make sure that it doesn't get promoted to a
* register.
*/
volatile double retval;
/*
* A zero significand requires explicit construction of -0.0.
* (Unary minus returns positive zero.)
*/
if (mp_iszero(significand)) {
return copysign(0.0, -signum);
}
/*
* Set the 53-bit rounding mode. WARNING: It must be reset before
* returning.
*/
TCL_IEEE_DOUBLE_ROUNDING;
/*
* Make quick checks for over/underflow. Be careful to avoid
* integer overflow when calculating with 'exponent'.
*/
if (exponent >= 0 && exponent-1 > maxDigits-numSigDigs) {
retval = HUGE_VAL;
goto returnValue;
} else if (exponent < 0 && numSigDigs+exponent < minDigits+1) {
retval = 0.0;
goto returnValue;
}
|
| ︙ | ︙ |