Commit Graph

539 Commits

Author SHA1 Message Date
bde
d9c878777f Compute (1 - x^2) as ((1 - x) * (1 + x)) instead of as (1 - x * x) to
avoid easily avoidable loss of precision when |x| is nearly 1.

Extended (64-bit) precision only moves the meaning of "nearly" here.

This probably could be done better by splitting up the range into
|x| <= 0.5 and |x| > 0.5 like the C version.  However, ucbtest
does't report any errors in this version.  Perhaps the C version
should be used anyway.  It's only 25% slower now on a P5, provided
the C version of sqrt() isn't used, and the C version could be
optimized better.

Errors checked by:	ucbtest
1997-02-20 12:37:49 +00:00
bde
9e8f1bd4ec Select between the generic math functions and the i387-specific ones
at runtime.

etc/make.conf:
Nuked HAVE_FPU option.

lib/msun/Makefile:
Always build the i387 objects.  Copy the i387 source files at build
time so that the i387 objects have different names.  This is simpler
than renaming the files in the cvs repository or repeating half of
bsd.lib.mk to add explicit rules.

lib/msun/src/*.c:
Renamed all functions that have an i387-specific version by adding
`__generic_' to their names.

lib/msun/src/get_hw_float.c:
New file for getting machdep.hw_float from the kernel.

sys/i386/include/asmacros.h:
Abuse the ENTRY() macro to generate jump vectors and associated code.
This works much like PIC PLT dynamic initialization.  The PIC case is
messy.  The old i387 entry points are renamed.  Renaming is easier
here because the names are given by macro expansions.
1997-02-16 18:26:31 +00:00
bde
8a3827cb4c Fixed the i87 version of exp(). It returned NaN for args +-Inf. It had
some small (one or two ULP) inaccuracies.

Found by:	ucbtest
1997-02-16 17:38:11 +00:00
jkh
7f556ff462 Put back .endif clobbered by the previous commit, breaking the
build.
1997-02-15 14:05:42 +00:00
bde
1dd609a3d1 Disabled the i387 version if log1p(). It just evaluates log(1 + x).
This defeats the point of log1p().  ucbtest reports errors of +-5e+15
ULPs.  A correct version would use the i387 fyl2xp1 instruction for
small x and maybe scale to small x.  The C version does the scaling
reasonably efficiently, and fyl2px1 is slow (at least on P5s), so not
much is lost by always using the C version (only 25% for small x even
with the broken i387 version; 50% for large x).
1997-02-15 05:21:16 +00:00
wosch
6d0dbd7455 Sort cross references. 1997-01-20 23:23:22 +00:00
jkh
808a36ef65 Make the long-awaited change from $Id$ to $FreeBSD$
This will make a number of things easier in the future, as well as (finally!)
avoiding the Id-smashing problem which has plagued developers for so long.

Boy, I'm glad we're not using sup anymore.  This update would have been
insane otherwise.
1997-01-14 07:20:47 +00:00
wollman
f24b5e576f Delete -D_POSIX_MODE and -D_MULTI_LIBM from CFLAGS. They never had any effect
because _IEEE_LIBM always takes priority, so the definition just served
to confuse.

Reviewed by:	bde
1997-01-08 13:22:28 +00:00
wosch
0426c0af88 comma typos 1996-12-23 23:09:55 +00:00
bde
ff59515db9 Removed references to nonexistent functions log2() and log2f(). 1996-12-18 17:47:03 +00:00
bde
d859e7c0e5 Fixed fiddling with the control word. Use the stack space reserved for
that purpose instead of space below the stack.
1996-12-14 16:43:53 +00:00
bde
14480319f6 Clean up the FPU stack before returning. One stack slot was leaked on
every call.  The damage was sometimes limited by other routines using
and freeing stack slots that should have been empty to being with.
1996-12-14 16:24:42 +00:00
peter
48809684a1 consistancy fixup
Submitted by: "Philippe Charnier" <charnier@xp11.frmug.org>
1996-08-30 21:21:52 +00:00
peter
3f709de38c cmp -s || install -c ==> install -C 1996-08-30 02:12:07 +00:00
bde
f68534c325 Made rintf() actually work. It was completely broken (when s_rint.c
was compiled with -O) by the precision bug in the i386 version of
gcc (assignments and casts don't clip the precision).  E.g.,
rintf(12.3456789) was 12.125.

Avoid the same bug in rint().  It was only broken for the unusual
case when the i387 precision is 64 bits.  FreeBSD defaults to 53
bit precision to avoid problems like this, but the standard math
emulator always uses 64 bit precision.
1996-08-28 16:34:36 +00:00
mpp
d63efd5bb2 Fix up the NAME lines forthe ceil and floor man pages to be
less confusing.

Reviewed by:	bde
Partially obtained from: NetBSD-bugs
1996-08-21 21:23:47 +00:00
jkh
c4d4a99d31 General -Wall warning cleanup, part I.
Submitted-By: Kent Vander Velden <graphix@iastate.edu>
1996-07-12 18:57:58 +00:00
bde
cbb7e33d7c Clean up the FP stack before returning. The i387 exp() leaked an FP
register on its first call.  Subsequent calls reused the register so
the leak didn't accumulate.  Fixes PR 1275.
1996-06-01 18:00:07 +00:00
bde
5af01f7f7a Don't trash %ebp.
Obtained from: NetBSD
1996-02-25 20:29:46 +00:00
mpp
69247d533f Fixed some minor formatting problems to silence manck some more.
Corrected some bogus cross references to man pages that we don't/won't
have and either deleted them, or found a more appropriate man page
that we do have.  Various other minor changes to silence manck.

Manck is currently down to about 200 lines of errors, down from
the 500 - 600+ when I started all this.
1996-02-12 01:20:38 +00:00
mpp
a7ed5d12fe Correct one small typo in previous commit. 1996-02-09 00:48:52 +00:00
mpp
ac7bc7dc91 Added some missing MLINKS for section 3 man pages.
Also corrected a few minor formatting errors, file location and cross
references in some of the section 3 man pages.

This shuts up a lot of the output from "manck" for section 3.
1996-02-09 00:45:45 +00:00
bde
2de67f54f7 Undo the the changes in the previous revision (MANSRC now works right again).
Use ${INSTALL} instead of install.
1995-10-22 19:16:01 +00:00
bde
286f90726e Fixed use of too many args for `.Em'.
Fixed description of domain of y*().
Fixed description of error domain.  (This description is still half
redundant and half wrong, as in many other math man pages.  fdlibm
doesn't support the VAX or Tahoe.)
Fixed capitalization of `Bessel'.
1995-10-22 18:16:39 +00:00
wollman
6db5ef09ae Compress manual pages (if desired) in the obj directory rather
than in the installation destination.  Should make release-building
substantially faster.  The msun Makefile changes simple adapt to the new
scheme.
1995-10-02 20:02:05 +00:00
rgrimes
f05428e4cd Remove trailing whitespace. 1995-05-30 05:51:47 +00:00
bde
2cc0b4033d Submitted by: J.T. Conklin <jtc@wimsey.com>
Second part of update to fdlibm 5.2: speed up argument reduction for trig
functions in the case pi/4 < |x| < 3pi/4.

Remove unused static constants ("one").
1995-04-07 23:23:27 +00:00
bde
07041fdd38 Submitted by: J.T. Conklin <jtc@wimsey.com>
First part of update to fdlibm 5.2: fix jn(n, x) and jnf(n, x).
jn(-1, x) was too large by a factor of 3.
1995-04-07 23:13:43 +00:00
bde
056d7133a8 Obtained from: NetBSD
Remove common sources from ${SRCS} when they are replaced by arch-specific
sources.
1995-03-08 01:41:40 +00:00
jkh
4d71750d42 Additions from Thomas Graichen to mention each functions' floating point
counterpart.
Submitted by:	Thomas Graichen <graichen@sirius.physik.fu-berlin.de>
1995-03-01 05:06:48 +00:00
ljo
287e9adfce Add missing z_abs. In BSD tradition this is in libm.a. 1994-11-11 12:56:27 +00:00
bde
c817f8d6ca Install math.h. 1994-09-08 10:40:23 +00:00
jkh
d9d09e74e2 Latest fix from jtc:
The fyl2xp1 instruction has such a limited range:
   -(1 - (sqrt(2) / 2)) <= x <= sqrt(2) - 1
it's not worth trying to use it.

Also, I'm not sure fyl2xp1's extra precision will
matter once the result is converted from extended
real (80 bits) back to double real (64 bits).

Reviewed by:	jkh
Submitted by:	jtc
1994-08-19 23:52:29 +00:00
jkh
fd411c4b09 Make this puppy actually compile now.
Submitted by:	jkh
1994-08-19 11:15:40 +00:00
jkh
9f0650ff29 Do all the includes: <machine/asm.h> -> <machine/asmacros.h>
Reviewed by:
Submitted by:
1994-08-19 11:14:32 +00:00
jkh
5d58375ab0 Change includes to reference <machine/asmacros.h>.
Submitted by:	jkh
1994-08-19 11:12:52 +00:00
jkh
3a5f48d7f8 This commit was generated by cvs2svn to compensate for changes in r2116,
which included commits to RCS files with non-trunk default branches.
1994-08-19 09:40:01 +00:00
jkh
2a8fd4fc31 J.T. Conklin's latest version of the Sun math library.
-- Begin comments from J.T. Conklin:
The most significant improvement is the addition of "float" versions
of the math functions that take float arguments, return floats, and do
all operations in floating point.  This doesn't help (performance)
much on the i386, but they are still nice to have.

The float versions were orginally done by Cygnus' Ian Taylor when
fdlibm was integrated into the libm we support for embedded systems.
I gave Ian a copy of my libm as a starting point since I had already
fixed a lot of bugs & problems in Sun's original code.  After he was
done, I cleaned it up a bit and integrated the changes back into my
libm.
-- End comments

Reviewed by:	jkh
Submitted by:	jtc
1994-08-19 09:40:01 +00:00
rgrimes
cb2f2e712b BSD 4.4 Lite Lib Sources 1994-05-27 05:00:24 +00:00