add support COMPAT_LINUX for aarch64

Refactor pipe1() and correct a bug in sys_pipe2() (SYS_pipe2)

sys_pipe2() returns two integers (values), the 2nd one is a copy of the 2nd
file descriptor that lands in fildes[2]. This is a side effect of reusing
the code for sys_pipe() (SYS_pipe) and not cleaning it up.

The first returned value is (on success) 0.

Introduced a small refactoring in pipe1() that it does not operate over
retval[], but on an array int[2]. A user sets retval[] for pipe() when
desired and needed.

This refactoring touches compat code: netbsd32, linux, linux32.

Before the changes on NetBSD/amd64:

$ ktruss -i ./a.out
[...]
15131 1 a.out pipe2(0x7f7fff2e62b8, 0) = 0, 4
[...]

After the changes:

$ ktruss -i ./a.out
[...]
782 1 a.out pipe2(0x7f7fff97e850, 0) = 0
[...]

There should not be a visible change for current users.

Sponsored by <The NetBSD Foundation>

Do not uselessly include <sys/malloc.h>.

branches: 1.66.2;
Cleanup pipe(2) flags, now that native handle them.

branches: 1.65.14; 1.65.28;
move dup3 to a more appropriate place because pipe is "special". Gotta love
linux.

- implement dup3 and pipe2
- eliminate amd64 ifdef

branches: 1.63.18; 1.63.24;
Include <sys/sched.h> before <sys/syscallargs.h> for cpuset_t.

branches: 1.62.2; 1.62.4;
Remove clause 3 and 4 from TNF licenses

branches: 1.61.6; 1.61.8; 1.61.10;
Convert all the system call entry points from:
int foo(struct lwp *l, void *v, register_t *retval)
to:
int foo(struct lwp *l, const struct foo_args *uap, register_t *retval)
Fixup compat code to not write into 'uap' and (in some cases) to actually
pass a correctly formatted 'uap' structure with the right name to the
next routine.
A few 'compat' routines that just call standard ones have been deleted.
All the 'compat' code compiles (along with the kernels required to test
build it).
98% done by automated scripts.

branches: 1.60.4;
ANSIfy most of the function definitions in sys/compat (but not ndis).
All by the magic of sed ...

branches: 1.59.4;
Add compat_linux and exec_linux_elf lkm support for amd64:
- Add needed COMPAT_OSSAUDIO to GENERIC.
- Add missing includes needed by linux_syscallargs.h.
- Add lkm building.

branches: 1.58.18; 1.58.20; 1.58.24;
Merge newlock2 to head.

branches: 1.57.20;
merge ktrace-lwp.

Fix a messed up patch

register_t is not an int on amd64

Fix pipe linux emulation on amd64

branches: 1.53.2; 1.53.18; 1.53.20;
Merge the nathanw_sa branch.

add RCSIDs (including regeneration of files as appropriate)

branches: 1.51.2; 1.51.4;
remove include of <vm/vm.h>

remove redundant vm includes.

branches: 1.49.12;
Assign my copyright to TNF. Merge with others were appropriate.
Regen syscall files after script change.

Attempt to fix the mess.

Wheeee!: 8-way split of compat/linux/linux_misc.c into common code and multi-architechture functions.

Substantial signal handling changes:
* Increase the size of sigset_t to accomodate 128 signals -- adding new
versions of sys_setprocmask(), sys_sigaction(), sys_sigpending() and
sys_sigsuspend() to handle the changed arguments.
* Abstract the guts of sys_sigaltstack(), sys_setprocmask(), sys_sigaction(),
sys_sigpending() and sys_sigsuspend() into separate functions, and call them
from all the emulations rather than hard-coding everything. (Avoids uses
the stackgap crap for these system calls.)
* Add a new flag (p_checksig) to indicate that a process may have signals
pending and userret() needs to do the full (slow) check.
* Eliminate SAS_ALTSTACK; it's exactly the inverse of SS_DISABLE.
* Correct emulation bugs with restoring SS_ONSTACK.
* Make the signal mask in the sigcontext always use the emulated mask format.
* Store signals internally in sigaction structures, rather than maintaining a
bunch of little sigsets for each SA_* bit.
* Keep track of where we put the signal trampoline, rather than figuring it out
in *_sendsig().
* Issue a warning when a non-emulated sigaction bit is observed.
* Add missing emulated signals, and a native SIGPWR (currently not used).
* Implement the `not reset when caught' semantics for relevant signals.

Note: Only code touched by the i386 port has been modified. Other ports and
emulations need to be updated.

ufs_readdir no longer checks if vp->v_type != VDIR so
the individual emulated readdirs must check.
Since netbsd and freebsd return EINVAL for the error
and I don't know what the other platforms do, return
EINVAL for them too.

Add some braces to make egcs happy.

Oops, fix typo, pointed out by Urban Boquist <boquist@cs.chalmers.se>.

Implement the shrinking and no-change bits of the Linux mremap(2) system
call, provided by Urban Boquist <boquist@cs.chalmers.se> in PR #5693.

Add a comment about how one might implement the growing bit of mremap(2).

Fix typo in utime().

Only free cookiebuf when sure that the vop actually allocated it.

Merge with Lite2 + local changes

Linux's getsid(2) does in fact return a pid, not a pointer.

Add a comment clearly describing that Linux's getsid(2) differs from
the XPG4.2 definition in that it returns the kernel virtual address
of the session structure, rather than the process group ID of the
session leader.

branches: 1.36.2;
msync -> __msync13

Fix braino in my interpretation of the eof flag to VOP_READDIR.

Emulate mremap(2) by returning ENOMEM for now. This should be sufficient for
realloc(3) and other uses. From Matthias Scheler in PR 3920.

Signal error when getting a 64 bit offset cookie that can't be stored
in 32 bits. Provide an error message to the user, and return EINVAL.
Also, pay attention to the EOF flag from VOP_READDIR. Correct a
misspell in a panic message.

Fix "missing entries" bug when reading large directories: When reading a
directory and running out of space in the dest buffer, off should point to the
current entry (which was not saved) and not to the next.
I discovered this bug using linux and SunOS emulation over NFS, but seems to
affect other emulations as well.

branches: 1.31.2;
Changes to the sudio system:
- It is now possible to handle devices that want "looping" DMA,
e.g. the SoundBlaster correctly. The WSS and SB drivers use this.
To do this several new methods were introduced in audio_hw_if.
- Different silence handling (forced by previous change).
- The audio driver can now be mmap()-ed, but due to problems in
the VM system only for writing for now.
- The OSS (Linux) audio emulation takes advantage of some of the
new features.

Add Linux nice() syscall.

sys/dir.h -> sys/dirent.h

Catch up with setre[ug]id() changes.

branches: 1.27.4;
Update list of Linux syscalls and implement the ones that are likely to be
used and doable.

- Prototyping changes
- Fix bug that gcc -Wall found in linux_sys_getpgid()

Add explicit return types.

msync, readv and writev were recently added to the Linux kernel; add them
to the emulation as well.

We don't need a private version of fork() here any more.

Use cookies for directory offset, mostly from Greg Hudson.

Prefix names of system call implementation functions with `sys_'.

Make system calls conform to a standard prototype and bring those
prototypes into scope.

Update offset elsewhere in linux_getdents(); still emulate a bug in
Linux, but avoid endless loops.

Fix some bugs & typing problems.

Emulate some syscalls that appear in Linux 1.3: a select() with normal args,
getdents() and flock().

Check for status NULL argument in wait4 and waitpid. From PR #1392
by Thomas EberHardt.

Implement the old and older uname(2) calls.

If updating the timeval would make it negative, zero it instead.

Allow bogus timevals, per Linux.

Export bsd_to_linux_sig[], and use it.

Rearrange #includes.

Linux doesn't allow select(2) to be restarted.

Use compat_util.[ch].

* Changed to use generic ELF code from kern/exec_elf.c (which was taken
from the svr4 exec code plus the linux compat mods)
* Include file change
* Add multicast sockopts (from John Brezak)

Linux select() returns the amount of time left before a timeout would have
occured in the timeval structure. Emulate this behavior. Reported by
Theo de Raadt.

* Make Linux ELF binaries work. Lots of code duplication, but it will have to do
for now.
* Add a sanity check in linux_uselib

Rename the dirent fields to match Linux. Make sure the requested buffer
is no larger than MAXBSIZE.

Emulate the return value of times() better.

Update to use timer{add,sub}().

Added a few more system calls.
More consistent alternate root file semantics throughout.

Added Linux compat code.

Refactor pipe1() and correct a bug in sys_pipe2() (SYS_pipe2)

sys_pipe2() returns two integers (values), the 2nd one is a copy of the 2nd
file descriptor that lands in fildes[2]. This is a side effect of reusing
the code for sys_pipe() (SYS_pipe) and not cleaning it up.

The first returned value is (on success) 0.

Introduced a small refactoring in pipe1() that it does not operate over
retval[], but on an array int[2]. A user sets retval[] for pipe() when
desired and needed.

This refactoring touches compat code: netbsd32, linux, linux32.

Before the changes on NetBSD/amd64:

$ ktruss -i ./a.out
[...]
15131 1 a.out pipe2(0x7f7fff2e62b8, 0) = 0, 4
[...]

After the changes:

$ ktruss -i ./a.out
[...]
782 1 a.out pipe2(0x7f7fff97e850, 0) = 0
[...]

There should not be a visible change for current users.

Sponsored by <The NetBSD Foundation>

Do not uselessly include <sys/malloc.h>.

branches: 1.66.2;
Cleanup pipe(2) flags, now that native handle them.

branches: 1.65.14; 1.65.28;
move dup3 to a more appropriate place because pipe is "special". Gotta love
linux.

- implement dup3 and pipe2
- eliminate amd64 ifdef

branches: 1.63.18; 1.63.24;
Include <sys/sched.h> before <sys/syscallargs.h> for cpuset_t.

branches: 1.62.2; 1.62.4;
Remove clause 3 and 4 from TNF licenses

branches: 1.61.6; 1.61.8; 1.61.10;
Convert all the system call entry points from:
int foo(struct lwp *l, void *v, register_t *retval)
to:
int foo(struct lwp *l, const struct foo_args *uap, register_t *retval)
Fixup compat code to not write into 'uap' and (in some cases) to actually
pass a correctly formatted 'uap' structure with the right name to the
next routine.
A few 'compat' routines that just call standard ones have been deleted.
All the 'compat' code compiles (along with the kernels required to test
build it).
98% done by automated scripts.

branches: 1.60.4;
ANSIfy most of the function definitions in sys/compat (but not ndis).
All by the magic of sed ...

branches: 1.59.4;
Add compat_linux and exec_linux_elf lkm support for amd64:
- Add needed COMPAT_OSSAUDIO to GENERIC.
- Add missing includes needed by linux_syscallargs.h.
- Add lkm building.

branches: 1.58.18; 1.58.20; 1.58.24;
Merge newlock2 to head.

branches: 1.57.20;
merge ktrace-lwp.

Fix a messed up patch

register_t is not an int on amd64

Fix pipe linux emulation on amd64

branches: 1.53.2; 1.53.18; 1.53.20;
Merge the nathanw_sa branch.

add RCSIDs (including regeneration of files as appropriate)

branches: 1.51.2; 1.51.4;
remove include of <vm/vm.h>

remove redundant vm includes.

branches: 1.49.12;
Assign my copyright to TNF. Merge with others were appropriate.
Regen syscall files after script change.

Attempt to fix the mess.

Wheeee!: 8-way split of compat/linux/linux_misc.c into common code and multi-architechture functions.

Substantial signal handling changes:
* Increase the size of sigset_t to accomodate 128 signals -- adding new
versions of sys_setprocmask(), sys_sigaction(), sys_sigpending() and
sys_sigsuspend() to handle the changed arguments.
* Abstract the guts of sys_sigaltstack(), sys_setprocmask(), sys_sigaction(),
sys_sigpending() and sys_sigsuspend() into separate functions, and call them
from all the emulations rather than hard-coding everything. (Avoids uses
the stackgap crap for these system calls.)
* Add a new flag (p_checksig) to indicate that a process may have signals
pending and userret() needs to do the full (slow) check.
* Eliminate SAS_ALTSTACK; it's exactly the inverse of SS_DISABLE.
* Correct emulation bugs with restoring SS_ONSTACK.
* Make the signal mask in the sigcontext always use the emulated mask format.
* Store signals internally in sigaction structures, rather than maintaining a
bunch of little sigsets for each SA_* bit.
* Keep track of where we put the signal trampoline, rather than figuring it out
in *_sendsig().
* Issue a warning when a non-emulated sigaction bit is observed.
* Add missing emulated signals, and a native SIGPWR (currently not used).
* Implement the `not reset when caught' semantics for relevant signals.

Note: Only code touched by the i386 port has been modified. Other ports and
emulations need to be updated.

ufs_readdir no longer checks if vp->v_type != VDIR so
the individual emulated readdirs must check.
Since netbsd and freebsd return EINVAL for the error
and I don't know what the other platforms do, return
EINVAL for them too.

Add some braces to make egcs happy.

Oops, fix typo, pointed out by Urban Boquist <boquist@cs.chalmers.se>.

Implement the shrinking and no-change bits of the Linux mremap(2) system
call, provided by Urban Boquist <boquist@cs.chalmers.se> in PR #5693.

Add a comment about how one might implement the growing bit of mremap(2).

Fix typo in utime().

Only free cookiebuf when sure that the vop actually allocated it.

Merge with Lite2 + local changes

Linux's getsid(2) does in fact return a pid, not a pointer.

Add a comment clearly describing that Linux's getsid(2) differs from
the XPG4.2 definition in that it returns the kernel virtual address
of the session structure, rather than the process group ID of the
session leader.

branches: 1.36.2;
msync -> __msync13

Fix braino in my interpretation of the eof flag to VOP_READDIR.

Emulate mremap(2) by returning ENOMEM for now. This should be sufficient for
realloc(3) and other uses. From Matthias Scheler in PR 3920.

Signal error when getting a 64 bit offset cookie that can't be stored
in 32 bits. Provide an error message to the user, and return EINVAL.
Also, pay attention to the EOF flag from VOP_READDIR. Correct a
misspell in a panic message.

Fix "missing entries" bug when reading large directories: When reading a
directory and running out of space in the dest buffer, off should point to the
current entry (which was not saved) and not to the next.
I discovered this bug using linux and SunOS emulation over NFS, but seems to
affect other emulations as well.

branches: 1.31.2;
Changes to the sudio system:
- It is now possible to handle devices that want "looping" DMA,
e.g. the SoundBlaster correctly. The WSS and SB drivers use this.
To do this several new methods were introduced in audio_hw_if.
- Different silence handling (forced by previous change).
- The audio driver can now be mmap()-ed, but due to problems in
the VM system only for writing for now.
- The OSS (Linux) audio emulation takes advantage of some of the
new features.

Add Linux nice() syscall.

sys/dir.h -> sys/dirent.h

Catch up with setre[ug]id() changes.

branches: 1.27.4;
Update list of Linux syscalls and implement the ones that are likely to be
used and doable.

- Prototyping changes
- Fix bug that gcc -Wall found in linux_sys_getpgid()

Add explicit return types.

msync, readv and writev were recently added to the Linux kernel; add them
to the emulation as well.

We don't need a private version of fork() here any more.

Use cookies for directory offset, mostly from Greg Hudson.

Prefix names of system call implementation functions with `sys_'.

Make system calls conform to a standard prototype and bring those
prototypes into scope.

Update offset elsewhere in linux_getdents(); still emulate a bug in
Linux, but avoid endless loops.

Fix some bugs & typing problems.

Emulate some syscalls that appear in Linux 1.3: a select() with normal args,
getdents() and flock().

Check for status NULL argument in wait4 and waitpid. From PR #1392
by Thomas EberHardt.

Implement the old and older uname(2) calls.

If updating the timeval would make it negative, zero it instead.

Allow bogus timevals, per Linux.

Export bsd_to_linux_sig[], and use it.

Rearrange #includes.

Linux doesn't allow select(2) to be restarted.

Use compat_util.[ch].

* Changed to use generic ELF code from kern/exec_elf.c (which was taken
from the svr4 exec code plus the linux compat mods)
* Include file change
* Add multicast sockopts (from John Brezak)

Linux select() returns the amount of time left before a timeout would have
occured in the timeval structure. Emulate this behavior. Reported by
Theo de Raadt.

* Make Linux ELF binaries work. Lots of code duplication, but it will have to do
for now.
* Add a sanity check in linux_uselib

Rename the dirent fields to match Linux. Make sure the requested buffer
is no larger than MAXBSIZE.

Emulate the return value of times() better.

Update to use timer{add,sub}().

Added a few more system calls.
More consistent alternate root file semantics throughout.

Added Linux compat code.