branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

Trailing whitespace

Fix feraiseexcept.

- Don't touch the trap flags (though on all ARMv8 I know they have no
effect anyway).

- Don't clear any existing raised exception flags; just add to them.

XXX atf test
XXX pullup-9

Fix fesetenv and feupdateenv.

- fesetenv is supposed to set the stored rounding mode (and stored trap
settings, but they have no effect on any ARMv8 I know).

- feupdateenv is supposed to re-raise the exceptions that were raised
in the environment when it was called.

XXX atf test
XXX pullup-9

branches: 1.4.2;
When hardware subnormal support is available, disable flush-to-zero.

Similarly, when hardware NaN propagation is available, disable
default-NaN substitution.

This enables IEEE 754 semantics on any hardware that supports it by
default. Programs that want flush-to-zero or default-NaN substitution
can enable them explicitly.

ok ryo@

branches: 1.3.10; 1.3.12;
provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.2.2; 1.2.4;
Fix consistent typo: it is FE_ALL_EXCEPT instead of FE_EXCEPT_ALL (this
went unnoticed as the undefined macro evaluated to 0 and the expression
was always used with ~)

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

branches: 1.1.4;
Preliminary files for AARCH64 (64-bit ARM) support.
Enough for a distribution build.

Fix usage of sysarch(ALPHA_FPGETMASK)

branches: 1.2.4;
provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.1.2; 1.1.4;
fenv.h for alpha

Ensure that the stack is always 16-byte aligned by rounding sizes as needed.

All changes from rin@ except swapcontext.S, which is from me (added
symbolic constants to make the code clearer).

add licenses

allocate stack space for temporary storage

also provide an alpha assembler implemtation of lrint()

It's copysign, not _copysign.

Add copysignl aliases.

branches: 1.5.40;
Remove clause 3 and 4 from TNF licenses

branches: 1.4.40;
More trailing white space.

Fix files using old TNF copyright notice

Use <machine/asm.h>

Alpha versions of copysign() and copysignf()

Remove clause 3 and 4 from TNF licenses

branches: 1.3.44;
Fix files using old TNF copyright notice

Use <machine/asm.h>

Alpha versions of copysign() and copysignf()

branches: 1.1.4; 1.1.10;
Add support for the ARM hardfloat instructions vsqrt and vabs.

branches: 1.1.4; 1.1.10;
Add support for the ARM hardfloat instructions vsqrt and vabs.

Complete the namespace dance

Add missing namespace.h include

branches: 1.7.2;
provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.6.2; 1.6.4;
Fix exception status vs mask confusion, of course feenableexcept(),
fedisableexcept() and fegetexcept() all operate on the mask.

Add missing return in __SOFTFP__ variant, pointed out by htodd.

Add our common non standard extensions and fix typo (FE_EXCEPT_ALL ->
FE_ALL_EXCEPT).

branches: 1.3.2; 1.3.8;
Fix feclearexcept return value (should be 0).

Make this compile for earmhf

Add a C99 <fenv.h> implementation for arm.

lrint() and lrintf() are supposed to use the current rounding mode,
so use vcvtr rather than vcvt.

branches: 1.1.4; 1.1.10; 1.1.16; 1.1.20;
Add VFP versions of lrint/lrintf/lrintl for earmhf*

lrint() and lrintf() are supposed to use the current rounding mode,
so use vcvtr rather than vcvt.

branches: 1.1.4; 1.1.10; 1.1.16; 1.1.20;
Add VFP versions of lrint/lrintf/lrintl for earmhf*

branches: 1.2.4; 1.2.10;
Remove .fpu since it's not needed

Add support for the ARM hardfloat instructions vsqrt and vabs.

branches: 1.1.4; 1.1.10;
Add fma/fmaf/fmal for arm hard float which uses the VFP vmla instruction.

branches: 1.1.4; 1.1.10;
Add fma/fmaf/fmal for arm hard float which uses the VFP vmla instruction.

fix typos in comments, mainly s/extention/extension/ and s/sufficent/sufficient/

provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.3.2; 1.3.4;
Update {read,write}fpsr to work everywhere

branches: 1.2.2;
Simplify, now that rounding mode defines match the hardware bits.

fenv(3) support for hppa

Use a dummy assembly file so we won't try produce lint stuff.

- Fold libm387 into libm.
- Leave libm387 as an empty shell.

branches: 1.9.2;
avoid shadow variables, and enable lint.

branches: 1.8.18;
merge libc/arch/{i386,x86_64}/gen/modf.S into one
(not quite a merge actually, just an #ifdef for now)
and build it in the i386 case (amd64 should work, just needs to be tested)
That way, a program linking against libm should get the optimized
version as expected.

activate e_atan2f.S which was there forever but not built, noticed
by Joerg Sonnenberger

Add i387 FPU routines for log2(3) and log2f(3).

ok christos@

branches: 1.5.2;
provide an assembler implementation of lrint() for i386 and amd64

Explicitly move setting of NOxxx and USE_SHLIBDIR to the top of the
Makefile (before including <bsd.own.mk>)

* Add user-controlled mk.conf variables
- SHLIBDIR Location to install shared libraries if ${USE_SHLIBDIR}
is "yes". Defaults to "/usr/lib".

- USE_SHLIBDIR If "yes", install shared libraries in ${SHLIBDIR}
instead of ${LIBDIR}. Defaults to "no".
Sets ${_LIBSODIR} to the appropriate value.
This may be set by individual Makefiles as well.

- SHLINKDIR Location of shared linker. Defaults to "/usr/libexec".
If != "/usr/libexec", change the dynamic-linker
encoded in shared programs

* Set USE_SHLIBDIR for libraries used by /bin and /sbin:
libc libcrypt libcrypto libedit libipsec libkvm libm libmi387
libtermcap libutil libz

* If ${_LIBSODIR} != ${LIBDIR}, add symlinks from ${LIBDIR}/${LIB}.so*
to ${_LIBSODIR}/${LIB}.so* for compatibility.

* Always install /sbin/init statically (for now)


The net effect of these changes depends on how the variables are set:

1.) If nothing is set or changed, there is no change from the
current behaviour:
- Static /bin, /sbin, and bits of /usr/*
- Dynamic rest
- Shared linker is /usr/libexec/ld*so

2.) If the following make variables are set:
LDSTATIC=
SHLINKDIR=/lib
SHLIBDIR=/lib
Then the behaviour becomes:
- Dynamic tools
- .so libraries used by /bin and /sbin are installed to /lib,
with symlinks from /usr/lib/lib*so to -> /lib/lib*so
where appropriate
- Shared linker is /lib/ld*so

3.) As per 2.), but add the following variable:
USE_SHLIBDIR=yes
This forces all .so's to be instaleld in /lib (with compat
symlinks), not just those tagged by their Makefiles to be.
Again, compat symlinks are installed

Readability cleanups; MKfoo=no -> NOfoo.

branches: 1.1.2;
Makefile and shared library for libm387.

Pass PICFLAGS down to cc-as-as and use __PIC__ to decide if it is small
vs big PIC mode. Retire -DPIC and -DBIGPIC.

branches: 1.7.2; 1.7.8;
Use correct stack register as noticed by christos

SSE2 registers are only used for float and double arguments, so update
comment to reflect reality. Add some accessor macros for long double
arguments for i386 and x86_64.

branches: 1.5.18;
-fix an obvious mistake in fpcw check
-replace ARG_DOUBLE_ONE_HALF by _MSW/_LSW because this reflects the
intention and also matches the terms used in C code,
also make the code where the fpcw overwrites the argument a bit
self-documenting
(this abstraction sucks because it forces to write inefficient code)

add macros for the upper half of the double arguments.

branches: 1.3.32;
Typo fix, from skrueger at europe com.

Make log1p and log1pf work in shared libraries again.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Drop in a fabs() after the fsqrt().
While it may seem pointless, it the rouding mode is set to round towards
minus infinity then acos(-1) calculates atan2(sqrt(1.0 - 1.0), -1) the subtract
generates -0.0 which sqrt() preserves.
atan2(-0, -1) is -pi, but acos(-1) is expected to be +pi.
This might 'fix' the test failures seen in some environments, but they
are not failing due to an obvously incorrent x87 rounding mode.

branches: 1.8.54; 1.8.60;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Make these idempotent as far as the FPU register stack is concerned.

branches: 1.5.8;
More trailing white space.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Make these idempotent as far as the FPU register stack is concerned.

branches: 1.4.16;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

i387-specific math functions.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

i387 float version of atan2()

-fix an obvious mistake in fpcw check
-replace ARG_DOUBLE_ONE_HALF by _MSW/_LSW because this reflects the
intention and also matches the terms used in C code,
also make the code where the fpcw overwrites the argument a bit
self-documenting
(this abstraction sucks because it forces to write inefficient code)

Share the amd64 and i386 exp versions. Fixes a problem with exp(-Inf) != 0.

branches: 1.12.34;
PR/15748: Ray Brownrigg: exp(log(0)) is broken.
Use FreeBSD's version. Handles inf/nan and rounding modes.

%rbp -> %rsp in 2 cases.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Make these idempotent as far as the FPU register stack is concerned.

branches: 1.8.8;
More trailing white space.

use one less fp stack slot

It's necessary to use 64 bit precision to avoid errors in the least
significant bit(s) (for some operands).

Fix from Bruce Evans. Commited to FreeBSD earlier this month, but was
never forwarded to me :-(.

> 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.

branches: 1.4.6;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

as in exp(), treat +/-Inf and NaN arguments specially, should fix
eg expf(-Inf) which was Nan previously
(We could avoid touching the i387 on amd64 in these cases, but we'd
need to bypass the ABI abstraction macros, so leave it the old way
for now.)

branches: 1.5.32;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Make these idempotent as far as the FPU register stack is concerned.

branches: 1.2.8;
More trailing white space.

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

netbsd.org->NetBSD.org

Oops, when I replaced sahf with a btw, I made it test the wrong bit
(2 vs. 10).

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 libm support.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

Add i387 FPU routines for log2(3) and log2f(3).

ok christos@

Add i387 FPU routines for log2(3) and log2f(3).

ok christos@

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Fix typo

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

netbsd.org->NetBSD.org

2 more bitnumber mistakes fixed while removing sahf.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 libm support.

netbsd.org->NetBSD.org

2 more bitnumber mistakes fixed while removing sahf.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Added i387 remainderf and sqrtf

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

The scalb* functions left data on the floting point register stack.
Remove it before leaving the function.

The solution is taken from OpenBSD.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Added i387 remainderf and sqrtf

branches: 1.1.2; 1.1.4;
Use a dummy assembly file so we won't try produce lint stuff.

Use a dummy assembly file so we won't try produce lint stuff.

Forgot to add this yesterday.

fix typos in comments, mainly s/extention/extension/ and s/sufficent/sufficient/

use named initializers

provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.7.2; 1.7.4;
PR/50822: David Binderman: Fix copy length.

Preserve the initial x87 control word in the global FP environment
in preparation for adjusting the default based on the main binary
version.

Fix sense of fegetexcept on x86.

Somehow I overlooked this when I fixed feenableexcept and
fedisableexcept last summer.

XXX pullup to 6

branches: 1.4.2;
Mask off the result of feenableexcept and fedisableexcept on x86.

These shouldn't return bits outside FE_ALL_EXCEPT.

branches: 1.3.6; 1.3.8;
Fix compile error.

Fix merge error: drop machine from an older workaround for machine.sse
under compat32, drop the assertion, initialize oldlen.

Add support for fenv.h interface for i386 and amd64.

Submitted by Stathis Kamperis as part of GSoC 2010 and ported from
FreeBSD.

add licenses

provide an assembler implementation of lrint() for i386 and amd64

branches: 1.1.4;
libm387: Add (intentionally empty!) exported symbol list.

This library has been an empty stub since 2008, probably won't serve
any useful purpose in the future, but let's make sure it doesn't
accidentally sprout anything.

PR lib/58838: shared libraries in base should all have expsym lists

libm: Sprinkle whitespace fixes and rcsids on inverse trig functions.

No functional change intended.

libm: Do the weak alias dance for asin, acos, atan.

These are used internally by the complex trig functions, so they need
weak aliases.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

libm: Sprinkle whitespace fixes and rcsids on inverse trig functions.

No functional change intended.

libm: Do the weak alias dance for asin, acos, atan.

These are used internally by the complex trig functions, so they need
weak aliases.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

fix various typos, mainly in comments.

Kill redundant ;

branches: 1.7.52;
netbsd.org->NetBSD.org

Fix stack usage for temporary storage. From Stephen Ma in PR xsrc/15404.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

branches: 1.4.16;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

fix various typos, mainly in comments.

Kill redundant ;

branches: 1.8.46;
amd64: stack sanity, leave the return PC alone

branches: 1.7.2;
netbsd.org->NetBSD.org

Fix mistake in previous.

Fix stack usage for temporary storage. From Stephen Ma in PR xsrc/15404.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

branches: 1.3.16;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Switch to SSE code, since our gas supports it now.

branches: 1.6.52;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

Actually used SSE branch. Noticed by enami@

Switch to SSE code, since our gas supports it now.

branches: 1.5.52;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.8.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

More trailing white space.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

Optimize the common case (theta < 2^63).

i387-specific math functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.6.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

More trailing white space.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

this simple adaptation exp() does not work

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

this simple adaptation exp() does not work

Assembly versions of expf(), logf(),log10f(), scalbf(), expm1() and expm1f().

libm/arch/i387: Do the weak alias dance for finite and finitef.

These are used internally by some things.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Use setne instead of setnel, new versions of gas balk at the bad mnemonic.
Wrong mask was used in s_finitef.S.

branches: 1.4.6;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

libm/arch/i387: Do the weak alias dance for finite and finitef.

These are used internally by some things.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Use setne instead of setnel, new versions of gas balk at the bad mnemonic.
Wrong mask was used in s_finitef.S.

branches: 1.3.6;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

fix various typos, mainly in comments.

Kill redundant ;

branches: 1.8.52;
netbsd.org->NetBSD.org

Fix stack usage for temporary storage. From Stephen Ma in PR xsrc/15404.

%rbp -> %rsp in 2 cases.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

branches: 1.4.16;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

fix various typos, mainly in comments.

Kill redundant ;

branches: 1.7.46;
amd64: stack sanity, leave the return PC alone

branches: 1.6.2;
netbsd.org->NetBSD.org

Fix stack usage for temporary storage. From Stephen Ma in PR xsrc/15404.

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

branches: 1.3.16;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.7.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Fix from Jochen Pohl, i387 version of ilogb didn't set up the stack
frame correctly (PR #1590).

branches: 1.4.2;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

i387 specific ilogb()

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.6.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Setup stack-frame correctly (same as s_ilogb.S).

branches: 1.3.2;
Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Add ilogbl(3).

libm: Fix log-related symbol aliases.

- Do the weak alias dance for log, log2, log10, log1p, and their
f/l variants.

(Not strictly necessary for log2 and log10 since nothing in the
math library itself uses them -- only a computer scientist would
use log2 and only an engineer would use log10 -- but let's just do
it the same for all our forestry.)

- Provide long double aliases on vax.

- Sprinkle RCS ids.

branches: 1.13.108;
Avoid the PIC dance.

Make log1p and log1pf work in shared libraries again.

Reimplement s_log1p.S and s_log1pf.S to use the fyl2xp1 instruction
where necessary.

The log1p() function is provided to compute an accurate value of
log(1 + x), even for tiny values of x. The i387 FPU provides the
fyl2xp1 instruction for this purpose.

However, since the range of the fyl2xp1 function is limited to
-(1 - (sqrt(2) / 2)) <= x <= sqrt(2) - 1
(-0.292893 <= x <= 0.414214)
we need to check if the argument is in the valid range.

In order to reduce the cost for testing the range, we only use
fyl2xp1 if the argument is in the range
-0.25 <= x <= 0.25
which can be checked with just one conditional branch.

Fixes PR lib/22599 by Ray Brownrigg.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

More trailing white space.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Remove lie (in a comment) about conversion between 80 bit ot 64 bit reals,
since we set up the fpu to do all calculations in 64 bit mode at this time.

The fyl2xp1 instruction has a limited range:
-(1 - (sqrt(2) / 2)) <= x <= sqrt(2) - 1
so we can't 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).

Major Bogon: change %ebp to %esp, as I don't set up a stack frame.

branches: 1.2.2;
Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

libm: Fix log-related symbol aliases.

- Do the weak alias dance for log, log2, log10, log1p, and their
f/l variants.

(Not strictly necessary for log2 and log10 since nothing in the
math library itself uses them -- only a computer scientist would
use log2 and only an engineer would use log10 -- but let's just do
it the same for all our forestry.)

- Provide long double aliases on vax.

- Sprinkle RCS ids.

branches: 1.10.108;
Avoid the PIC dance.

Make log1p and log1pf work in shared libraries again.

Reimplement s_log1p.S and s_log1pf.S to use the fyl2xp1 instruction
where necessary.

The log1p() function is provided to compute an accurate value of
log(1 + x), even for tiny values of x. The i387 FPU provides the
fyl2xp1 instruction for this purpose.

However, since the range of the fyl2xp1 function is limited to
-(1 - (sqrt(2) / 2)) <= x <= sqrt(2) - 1
(-0.292893 <= x <= 0.414214)
we need to check if the argument is in the valid range.

In order to reduce the cost for testing the range, we only use
fyl2xp1 if the argument is in the range
-0.25 <= x <= 0.25
which can be checked with just one conditional branch.

Fixes PR lib/22599 by Ray Brownrigg.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

More trailing white space.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Remove lie (in a comment) about conversion between 80 bit ot 64 bit reals,
since we set up the fpu to do all calculations in 64 bit mode at this time.

Assembly language versions of some of the "float" functions.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Add logbl(3).

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.1.84;
merge libc/arch/{i386,x86_64}/gen/modf.S into one
(not quite a merge actually, just an #ifdef for now)
and build it in the i386 case (amd64 should work, just needs to be tested)
That way, a program linking against libm should get the optimized
version as expected.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

FPU specific math code.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

libm: New s_rintl.S for x86.

Prompted by PR lib/58054.

Drop s_scalbnl.c (and s_scalbln.S as side effect) and provide scalbnl in
s_scalbln.c.

Add a dummy s_scalbln.S to avoid the generic implementations of scalbln
and co on i386 and amd64. They are implemented next to scalbn and co.

Fix ldexp aliases.

Use correct alias for scalbn(). Tested by martin@.

Add aliases for scalbn* to ldexp, they are identical for FLT_RADIX=2.

Use the correct register on AMD64 for the argument.

Add long double version of x86 scalb functions

branches: 1.9.6; 1.9.12;
-build ldexp/frexp/modf locally again rather than relying on libc
reaching over, to get things into a sane state,
-do the usual namespace protection game with scalbn(f) because it
is used internally
(more symbol renamings are necessary)

The weak_alias change for mc68881 is untested.

The second argument is not a float, so the "abi.h" abstraction
doesn't work.
Open-code the argument handling instead.
Should fix the ldexp() bug reported by Brendon Costa in port-amd64.

branches: 1.7.4; 1.7.6; 1.7.10;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

The scalb* functions left data on the floting point register stack.
Remove it before leaving the function.

The solution is taken from OpenBSD.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

More i387 math support, too bad we can't enable it

Correctly alias the ldexp* functions to the scalbn equivalents
that take an int argument.

Add aliases for scalbn* to ldexp, they are identical for FLT_RADIX=2.

Use the correct register on AMD64 for the argument.

Fix the strong alias correctly this time.

Fix strong_alias

Add long double version of x86 scalb functions

branches: 1.8.6; 1.8.12;
-build ldexp/frexp/modf locally again rather than relying on libc
reaching over, to get things into a sane state,
-do the usual namespace protection game with scalbn(f) because it
is used internally
(more symbol renamings are necessary)

The weak_alias change for mc68881 is untested.

The second argument is not a float, so the "abi.h" abstraction
doesn't work.
Open-code the argument handling instead.
Should fix the ldexp() bug reported by Brendon Costa in port-amd64.

branches: 1.6.4; 1.6.6; 1.6.10;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

The scalb* functions left data on the floting point register stack.
Remove it before leaving the function.

The solution is taken from OpenBSD.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Correctly alias the ldexp* functions to the scalbn equivalents
that take an int argument.

Use the correct register on AMD64 for the argument.

provide ldexpl

Add long double version of x86 scalb functions

branches: 1.1.2; 1.1.8;
Add optimised version of scalbnl for x86.

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

Added i387 version of significand().

netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.7.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

Optimize the common case (theta < 2^63).

i387-specific math functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.5.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.7.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Added RCS ID's. ID's use new RCSID macro from <machine/asm.h>.

Optimize the common case (theta < 2^63).

i387-specific math functions.

Remove unused 387 implementations.

These were removed from the build for being wrong, but the implementation
stayed around. This is confusing, we have the attic for old code,
let's delete the unused implementations.

branches: 1.5.96;
netbsd.org->NetBSD.org

Modify the i387 code so that it can be shared between the i386 port
and the x86_64 port. XXX some files should be distinct.

Id -> NetBSD

Winning Strategies has placed this code into the Public Domain

Assembly language versions of some of the "float" functions.

Bump minor version for log2{,f}.

branches: 1.1.2;
Makefile and shared library for libm387.

branches: 1.3.2;
expose the BSD/linux fenv extensions in libm on ia64 too.

provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().

branches: 1.1.2;
Add fenv for ia64, imported from FreeBSD, compile tested only.

libm/arch/m68060/Makefile: bsd.sys.mk -> bsd.host.mk

Somewhat belatedly, after bsd.sys.mk was forbidden to use directly
from Makefiles back in 2014.

Use ${HOST_SH} instead of `sh'.

If necessary, pull in <bsd.sys.mk> to get the definition of HOST_SH;
Makefiles that pull in one of (most of) <bsd.*.mk> will get this anyway.

Use ${NETBSDSRCDIR}/some/path instead of ${.CURDIR}/../../some/path (etc).
(Reduces make output by ~ 20%)

Move original Motorola files to the "dist" subdirectory, in response
to PR 9626 by Kris Shannon.

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

Clarify instructions

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.

*gamma* need k_sin*/k_cos*

wrap lines in Makefile.list

branches: 1.1.1;
Initial revision

lib/libm/arch/m68060: regen

lib/libm/arch/m68060: regen

Jump through the PLT in PIC mode to fix the DT_TEXTREL warning on creating
the shared object.

Regenerate, after changing internal label used for linkage to the wrapper
scripts.

Regenerated files after this change:
Move each wrapper function to its own source file, such that redefining one
function doesn't give link errors when another one is pulled in from the
library.
Problem discovered by Klaus Klein.