New port: mac68ksf - Macintosh 68k softfloat.

This commit is all the Makefile glue for the new port which is the mac68k
port with MKSOFTFLOAT set to "yes".

As posted here:
https://mail-index.netbsd.org/port-mac68k/2025/10/14/msg001037.html

Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

ARM EABI needs left shift logical too.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.14.2;
earm* need __lshldi3

branches: 1.13.2;
add support for __floatunsisf().

Update to C89 style.
Fix name of floatundidf (uns -> un).

Assign quad / softfloat specific ${SRCS} to ${SRCS.quad} / ${SRCS.softfloat}
then append them to ${SRCS}. Those will be used to generate quad / softfloat
symbol lists.

Compile ashldi3.S and lshrdi3.S on m68000.

branches: 1.9.12;
use ${ARCHDIR} instead of ${.CURDIR}/arch/${MACHINE_ARCH} where appropriate
libc builds on pmax again

don't include lshrdi3.c in MI sources, since they're already in MD sources

Note functions that probably can be removed

Fix typo in last change

Enable use of assembly language shift routines on the m68k

clean up Ids; no substantive changes.

update for .S and MACHINE_ARCH

make an arch directory in libc, and move all machine dirs into it

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.3.14;
Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.5.12;
Make shifting by 0 work even when shifting by 32 is equivalent to shifting
by 0. Fixes lib/4845 (KNFification of the fix given therein).

branches: 1.4.2;
Fix RCSID's

Don't special case shift counts > QUAD_BITS or == 0. The former is
undefined behavior (so we can do whatever we want) while the latter
doesn't happen much in actual practice.

These changes make these functions considerably smaller and faster.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.7.10;
The last commit changed the behavious wrt sign extension.
Back it out and replace with LINTED comments.

Cast signed long shifts to unsigned long to avoid lint portability warnings.
The comment above the code mentions that it should work even on machines
that shift in zeroes.

Make shifting by 0 work even when shifting by 32 is equivalent to shifting
by 0. Fixes lib/4845 (KNFification of the fix given therein).

branches: 1.4.2;
Fix RCSID's

Don't special case shift counts > QUAD_BITS or == 0. The former is
undefined behavior (so we can do whatever we want) while the latter
doesn't happen much in actual practice.

These changes make these functions considerably smaller and faster.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

Re-arranged code a bit so that gcc can optimize this a bit better.
This change results in better code on the i386, m68k, & ns32k, and
equivalent code on the sparc.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.6.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

branches: 1.5.2;
Update to C89 style.
Fix name of floatundidf (uns -> un).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.1.4; 1.1.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.5.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

branches: 1.4.56;
Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix declaration to use quad_t, not long long.
Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.1.4; 1.1.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.10.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.8.2;
Update to C89 style.
Fix name of floatundidf (uns -> un).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.5.12;
Corrected an off-by-one error (lib/6314 Torbjorn Granlund)

Fix RCSID's

clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.1.4; 1.1.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.8.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.6.56;
Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.4.14;
Fix RCSID's

clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.1.4; 1.1.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.9.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.7.2;
Update to C89 style.
Fix name of floatundidf (uns -> un).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.4.14;
Fix RCSID's

clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.1.4; 1.1.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.9.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.7.2;
Update to C89 style.
Fix name of floatundidf (uns -> un).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.4.14;
Fix RCSID's

clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

branches: 1.2.4; 1.2.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Add IEEE 754 versions of __float{,}disf which use no floating point.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.3.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.1.2;
Update to C89 style.
Fix name of floatundidf (uns -> un).

branches: 1.2.4; 1.2.8;
Don't return 0x1.0p31 for every input value 2^N.
There is also no need to special case the value 1.

Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Adapt to earm hard float by including softfloat-for-gcc.h even if __ARM_EABI__
so we can get the __aeabi aliases.

branches: 1.3.2;
If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.1.2;
add support for __floatunsisf().

branches: 1.2.4; 1.2.8;
Add IEEE 754 versions of
fixdfdi.c fixsfdi.c fixunsdfdi.c fixunssfdi.c
floatdidf.c floatdisf.c floatundidf.c floatundisf.c
These don't FP so are well suited to softfloat implementation.

Add IEEE 754 versions of __float{,}disf which use no floating point.

If compiling for SOFTFLOAT, include "softfloat/softfloat-for-gcc.h" so we
can pick any redefinitions of the function (for ARM EABI).

- change from size_t to unsigned int because we don't need the extra width
- cast to unsigned int results of quad ops

branches: 1.2.2;
Appease lint and don't use C99 features.

Add ieee754 versions of the compiler runtime functions
__fixuns{sf,df,tf}{si,di}.
Add an ieee754 version of the compiler runtime function __floatuntfdi

Update to C89 style.
Fix name of floatundidf (uns -> un).

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.4.14;
Fix RCSID's

clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.5.12;
Make shifting by 0 work even when shifting by 32 is equivalent to shifting
by 0. Fixes lib/4845 (KNFification of the fix given therein).

branches: 1.4.2;
Fix RCSID's

Don't special case shift counts > QUAD_BITS or == 0. The former is
undefined behavior (so we can do whatever we want) while the latter
doesn't happen much in actual practice.

These changes make these functions considerably smaller and faster.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.5.12;
Make shifting by 0 work even when shifting by 32 is equivalent to shifting
by 0. Fixes lib/4845 (KNFification of the fix given therein).

branches: 1.4.2;
Fix RCSID's

Don't special case shift counts > QUAD_BITS or == 0. The former is
undefined behavior (so we can do whatever we want) while the latter
doesn't happen much in actual practice.

These changes make these functions considerably smaller and faster.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

make sure result of % operator is identical for quad_t, int and long;
fix provided by Chris Torek <torek@BSDI.COM> in a message to tech-kern.

branches: 1.4.2;
Fix RCSID's

Re-arranged code a bit so that gcc can optimize this a bit better.
This change results in better code on the i386, m68k, & ns32k, and
equivalent code on the sparc.

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.4.12;
remove obsolete register declarations

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.10.6;
Delint.

Revert previous change: the assertion
* is made in a utility function with file scope only, and
* will never fail.

* use _DIAGASSERT() to check pointer arguments against NULL and file
descriptors against -1 (as appropriate).
* add actual checks which to detect stuff that would trigger_DIAGASSERT(),
and attempt to return a sane error condition.
* knf some code
* remove some `register' decls.

the first two items result in the addition of code similar to the
following in various functions:

_DIAGASSERT(path != NULL)
#ifdef _DIAGNOSTIC
if (path == NULL) {
errno = EFAULT;
return (-1);
}
#endif

Cast things to (digit) where the LHS is (digit) to avoid loss of precision
lint warnings. This change should be harmless.

remove obsolete register declarations

Fix RCSID's

clean up Ids; no substantive changes.

pull in changes from libkern.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.9.2;
remove duplicate declaration.

Use __GNUC_PREREQ__().

change an #if __GNUC >= 2 to #if (__GNUC__ >= 2) || defined(lint)
to get the right definition of qshift_t.

use machine/limits.h if _KERNEL or _STANDALONE

Add missing prototypes

Fix from Jochen Pohl which fixes case where quad multiply
goes awry if bit 31 of at least one of the operands is set
(PR #1591).

branches: 1.3.2;
clean up Ids; no substantive changes.

constants are normally ints, remember!

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.3.14;
Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

Use reach-over sources from common/lib/libc

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Fix RCSID's

clean up Ids; no substantive changes.

branches: 1.1.1;
Torek's quad stuff from 4.4.

clean up Ids; no substantive changes.

branches: 1.1.1;
Initial revision

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.2.22;
clean up Ids; no substantive changes.

branches: 1.1.1;
Initial revision

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

As discussed (briefly) on tech-userlevel, fix our quad support to work
correctly on LP64 platforms. This is mostly just s/long/int/ where
appropriate.

branches: 1.2.22;
clean up Ids; no substantive changes.

branches: 1.1.1;
Initial revision