| Age | Commit message (Collapse) | Author |
|---|
|
therefore affected by the same bugs I've fixed a week ago.
The high part was being clipped for all exponents greater or equal
to 52. Fix this to use RESTYPE_BITS instead; also make the code
consistent.
|
|
|
|
|
|
Therefore, in nextafter() and nextafterf():
- if(x==y) return x; /* x=y, return x */
- if(x==y) return y; /* x=y, return y */
This matters for negative zero inputs of x or y.
|
|
bits, because "behavior is implementation defined in this case".
However, this is wrong; behavior is undefined if the right operand
is greater than or equal to the width of the promoted left operand.
This broke {,l}lrint{,f} (64-bit architectures), and llrint{,f}
(32-bit architectures) where results are actually 64-bit values.
The high part was clipped for all exponents greater or equal to 52.
Fix this to use RESTYPE_BITS instead; {,l}lrint{,f} are now able
to pass our regression tests, and I think are right now.
|
|
|
|
we round towards -inf or +inf.
|
|
OK kettenis@ jmc@
|
|
ok millert@
|
|
argument is a float. Eliminate a cast rendered superfluous as a result.
ok kettenis@, millert@
|
|
|
|
ok miod@
|
|
|
|
* do not use low-level roff macros like .if and .ds
* add the missing .Os to the preamble
* move unqualified text out of .Bl
ok jmc@
|
|
files or directories when applicable.
The inspiration and name of MACHINE_CPU come from NetBSD, although the way to
provide it to Makefiles is completely different.
ok kettenis@
|
|
ok pirofti@
|
|
ok otto@
|
|
unmaintainable). these days, people use source. these id's do not provide
any benefit, and do hurt the small install media
(the 33,000 line diff is essentially mechanical)
ok with the idea millert, ok dms
|
|
|
|
is defined on all architectures.
ok jmc@ (man pages) and martynas@
|
|
ok otto millert martynas
|
|
where appropriate. pointed out by jmc@
|
|
double => 4.3BSD
float => NetBSD 1.1
long double => OpenBSD 4.5
scalbln, scalblnf, scalblnl => OpenBSD 4.7
|
|
is getting ridiculous. split them into groups of copysign, ilogb,
nextafter, scalbn. discussed long ago with millert@
|
|
|
|
|
|
formats. which even for 80-bit & 128-bit long doubles is only 15
bits. therefore, scalbln, scalblnf, scalblnl are essentially the
same as scalbn, scalbnf, scalbnl with bounds checking so that
LONG_MIN..INT_MIN, and INT_MAX..LONG_MAX ranges properly raise
exceptions & yield correct values. looks good to millert@
|
|
|
|
|
|
ok martynas@, deraadt@
|
|
float/double prologue/epilogue, since the second argument is not
float/double. were never working properly (always returned inf).
actually, should help other math funcs (like pow) too, since they
use it internally
tested by kurt@; devel/boost fp regresses are working better now.
ok kurt@
|
|
overriden by md source. spotted by kurt@
actually; on amd64 scalbnf, scalbn (and therefore now frexpf) have
always been broken since second argument is not float. fix is under
reviewal / will be committed separately
ok kurt@, kettenis@, millert@. tested by kurt@
|
|
versions; spotted by kettenis@
while here also remove unused ldexp; it lives in libc
ok kettenis@, "looks good" millert@
|
|
accurately, hence the tricks in libm, using machine representation
of constants. remove kludges and switch to use decimal constants,
much simplifying the code. since, the compiler converts them
accurately. generated values match on vaxfp
discussed with millert@, and miod@
testing todd@, and myself
ok millert@
|
|
|
|
|
|
object code matches so this got to be oki
|
|
|
|
faster that way. tested by myself. discussed w/ & ok miod@, millert@
|
|
|
|
since its fp does not have distinguished values for qnans. tested
by naddy@; fixes libnova. ok theo
|
|
|
|
|
|
|
|
is already there;
ok martynas
|
|
|
|
from asm.h. discussed w/ millert@
|
|
|
|
important functions: acosl, asinl, atanl, atan2l, cosl,
sinl, tanl, exp2l, frexpl, ilogbl, ldexpl, logbl, scalbnl,
fabsl, hypotl, powl, sqrtl, rintl, copysignl, nanl, fdiml,
fmaxl, fminl. mostly taken from freebsd, needed alot of
changes to adapt. note, these are all c versions; and are
quite slow when architectures have, e.g. sqrt. assembly
versions will be added afterwards
- make them .weak/__weak_alias to the double precision
versions on other archs
- no need to have two finites. finite() and finitef() are
non-standard 3BSD obsolete versions of isfinite. remove
from libm. make them weak_alias in libc to __isfinite and
__isfinitef instead. similarly make 3BSD obsolete versions
of isinf, isinff, isnan, isnanf weak_aliases to C99's
__isinf, __isinff, __isnan, __isnanf
- remove unused infinity.c. the c library has infinities
for each supported platform
- use STRICT_ASSIGN cast hack for _kernel_rem_pio2, so that
the double version has a chance of working on i386 with
extra precision
- avoid storing multiple copies of the pi/2 array, since
it won't vary
- bump major due to removed finite/finitef. although they
will be in libc, which anything is linked to, minor bump
might be enough
ok millert@. tested by sthen@, jsg@, ajacoutot@, kili@, naddy@
|
|
|
