Remove enable_intr() and disable_intr(), which are unused.

(phx)

RFC 3542 (section 10.1) states that optlen should only be checked when
opt != NULL (Eitan Adler)

(christos)

Like amd64, use -msoft-float.

(joerg)

If we just want _end, load it directly and skip the gp

(matt)

fix vax build.

(christos)

fix the second copy of this file. XXX: merge them.

(christos)

add missing prototype; remove names from args

(christos)

bump

(christos)

Always supply LIB_PATH in case one of the included scripts set it to
something inappropriate.

(matt)

Remove an annoying printf.  And to answer the question:  VFS_VGET() gets
used by NFS V3 server for readdirplus.

(hannken)

USE PCI_BAR[0-5]. Part of PR#48016.

(msaitoh)

bump the date

(rmind)

cdbw(3) man page: fix the header file name and use .Fa for function arguments.

(rmind)

Don't unset LIB_PATH, just override it.

(matt)

Pass the correct buffer size.

(joerg)

Move fstrans_start()/fstrans_done() into genfs_insane_rename() to protect
the complete rename operation like we do for all other vnode operations.

(hannken)

Update count. Add serial comma.

(wiz)

Import a copy of jetlib in othersrc. This is exactly the same as the
version I've had posted on my web page the last five years; it is from
October 17, 2008 and was written in a day or so after irritation at
one of the endless proplib flamewars boiled over.

This should be considered a straw man only; it is not quite finished
and there are some unresolved issues to sort out (particularly
pertaining to iteration) before it can be used for real. There are
also probably things I'd do differently if I wrote it again now, and I
haven't attended to any of those or even looked at the thing in any
detail.

Also IIRC it's never actually been run.

(dholland)

__HAVE_ATOMIC_AS_MEMBAR is spelled with two leading underscores.

This underscores the need to replace this error-prone cpp API by
unconditionally defined {pre,post}atomic_membar_*.

This change should only remove unnecessary membar_producers on x86.

(riastradh)

Add support for CDB based NPF tables.

(rmind)

Apparantly clang doesn't like -mno-80387 or -mno-fp-ret-in-387
-msoft-float is equivalent to the former, the latter is probably unlikely
to stop anything not picked up elsewhere.

(dsl)

fix incorrect variable; (hi matt)

(christos)

make this compile.

(christos)

put back line accidentally deleted before (hi matt)

(christos)

Change the compiler options to explicitly specify:
  -mno-mmx -mno-sse -mno-avx -mno-80387 -mno-fp-ret-in-387
Since no-sse implies no-sse2 that should ensure that the compiler really
  doesn't emit any instructions that might trap trying to use the FPU.
On amd64 at least some of those are needed to stop the compiler
  saving the registers to stack on every varargs function.
It might be that -mno-sse did that before.

(dsl)

give correct results when running the digest when an mmap(2) fails but
read(2) succeeds

(agc)

Err on the safe side for small memory allocations and provide 4byte
alignment by default.  Override this to 8byte alignment for alpha and arm
eabi.

Someone (tm) can review this change once the standards and compiler(s)
have sorted themselves out.

(skrll)

npftest: fix previous harder - pass and use libc's random(3).

(rmind)

npftest: fix the failure of NAT test -- adjust for RUMP's conversion to
the in-kernel CPRNG (hi pooka!).

(rmind)

Add another set of parens

(matt)

mask the shift by __n with NBBY*sizeof(uintmax-t)-1
to suppress the following diagnostic from clang:
error: shift count >= width of type [-Werror,-Wshift-count-overflow]
    PAR_ATTR            = __BITS(63,56),// F=0 memory attributes
                          ^~~~~~~~~~~~~
src/sys/sys/cdefs.h:554:4: note: expanded from macro '__BITS'
        ((__BIT(MAX((__m), (__n)) + 1) - 1) ^ (__BIT(MIN((__m), (__n))) - 1))
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/sys/sys/cdefs.h:550:73: note: expanded from macro '__BIT'
    (((uintmax_t)(__n) >= NBBY * sizeof(uintmax_t)) ? 0 : ((uintmax_t)1 << (uintmax_t)(__n)))
                                                                        ^  ~~~~~~~~~~~~~~~~

(matt)

Now that the sv_xmm and sv_87 fields in the pcb are the same size as the
  hardware ones, and those in the mcontext, ptrace and core files we
  can really safely cast and copy between the types.
Remove a couple of temporary buffers and add CTASSERT()s that the sizes
  do actually match.

(dsl)

Wait for all the interfaces specified on the command line to be configured
before daemonizing, not just the first one. Perhaps we should introduce a
separate flag for this behavior?

(christos)

Remove the fields from 'struct save87' and 'struct fxsave' that were used
  to save the control word and tag word on fp faults.
Nothing I could find looked at them before I removed the code that wrote them.
It might appear that these fields got exported to userspace (via mcontext),
  but that code carefully avoided those fields.
I've also deleted some very historic fields from the union inside __fpregset_t
  this doesn't affect its size.
The __fpregset_t fields are now the same size as the ones in the pcb.
Which is good because they get cast to each other in a few places.

(dsl)

There is no need to check for recursive calls into fpudna().
Rename the associated ci_fpsaving field to 'unused'.
I'm not sure they could ever happen, you could get unwanted calls into
  the fpu trap code while saving state when using INT13 - but these are
  different.
The return value from the i386 fpudna() was always 1 - possibly a historic
  relic of the kernel fp emulation. Remove and don't check in trap.S.
The amd64 and i386 fpudna() code is now almost identical.

(dsl)

Correct number of output files in example.
From Jason McIntyre <jmc@OpenBSD>.

While here, also explain about the extra file in the example and
put .Ex into its own section in the correct place. Bump date.

(wiz)

_setsoftintr is long dead.  reap any leftovers from arm.

(matt)

#define __BYTE_ORDER BYTE_ORDER etc.
now endianness detection actually works and colours look right on my Gdium

(macallan)

Handle another declaration after statement case for lint in c99 mode.
Add the corresponding testcase.

(njoly)

When building up a PGP public key from an ssh key, do not use the key
file's mtime as the birthtime of the signature - rather, just zero
this value (the key birthtime is an inherent part of calculating the
key id)

This allows public key files to be copied to other file systems, machines
or directories, and still produce the same key id.

(agc)

consistency fixes.

(christos)

Undo previous - it still needs a lot more work.

For now, we'll use the hand-crafted cf* structures and directly
call all the config routines.

(pgoyette)

Since we always run with CR0.NE set (internal fpu using vector 0x10)
npxintr() is only generated when a process executes an FP instruction
and the TS set interrupt takes precedence - so we know the current lwp
owns the fp registers.
It is also then impossible to get a splurious error while saving
the registers.
Ths lets the npxintr() code be simplified somewhat.
XXX: I'm not at all sure it really DTRT if the process actually wished
to fixup anything in the signal handler (or even if the signal is masked).

(dsl)

Provide scalbnl, scalbnf, logbl, logbf, fmaxl for vax.

(martin)

From Igor Sobrado in private email (based on his OpenBSD commit):

    improve POSIX compliance by continuing to process the remaining file
    operands after not finding an input file.

    from the IEEE Std 1003.1-2008 (``POSIX.1'') rationale:

    "Unlike other utilities, some historical implementations of cut
    exit after not finding an input file, rather than continuing to
    process the remaining file operands.  This behavior is prohibited
    by this volume of POSIX.1-2008, where only the exit status is
    affected by this problem."

    joint work with jmc@, who identified the compliance issue, and millert@

    ok millert@, jmc@

(wiz)

Provide a simple version of _bus_dmamap_load_raw.  If each segments can
be mapped by XKPHYS/KSEGn, then the load will succeed.  If it would
require a bounce buffer or being mapped into the kernel's address space,
the load will fail.

(matt)

Properly translate struct swapent for COMPAT_NETBSD32  (missing commit)

(manu)

pthread__mutex_lock_slow: fix the handling of a potential race with the
non-interlocked CAS in the fast unlock path -- it is unsafe to test for
the waiters-bit while the owner thread is running, we have to spin for
the owner or its state change to be sure about the presence of the bit.
Split off the logic into the pthread__mutex_setwaiters() routine.

This is a partial fix to the named lockup problem (also see PR/44756).
It seems there is another race which can be reproduced on faster CPUs.

(rmind)

Try improving a comment.

(wiz)

For ancient GCC, disable the missing prototype warning completely.

(joerg)

Properly translate struct swapent for COMPAT_NETBSD32

(manu)

Ticket #1023

(sborrill)

Pull up the following revisions(s) (requested by martin in ticket #1023):
lib/libc/gen/posix_spawn.3: revision 1.5
lib/libc/gen/posix_spawn_file_actions_addopen.3: revision 1.4
lib/libc/gen/posix_spawn_file_actions_init.3: revision 1.4
lib/libc/gen/posix_spawn_fileactions.c: revision 1.3
sys/compat/netbsd32/netbsd32_execve.c: revision 1.38
sys/kern/kern_exec.c: revision 1.373

Limit the amount of kernel memory a posix_spawn syscall can use (for
handling the file action list) by limiting the maximum number of file
actions to twice the current file descriptor limit. Fix a few bugs in
the support functions and document the new limit. From Maxime Villard.

(sborrill)

Ticket #1023

(sborrill)

Pull up the following revisions(s) (requested by martin in ticket #1023):
lib/libc/gen/posix_spawn.3: revision 1.5
lib/libc/gen/posix_spawn_file_actions_addopen.3: revision 1.4
lib/libc/gen/posix_spawn_file_actions_init.3: revision 1.4
lib/libc/gen/posix_spawn_fileactions.c: revision 1.3
sys/compat/netbsd32/netbsd32_execve.c: revision 1.38
sys/kern/kern_exec.c: revision 1.373

Limit the amount of kernel memory a posix_spawn syscall can use (for
handling the file action list) by limiting the maximum number of file
actions to twice the current file descriptor limit. Fix a few bugs in
the support functions and document the new limit. From Maxime Villard.

(sborrill)

Ticket #1023

(sborrill)

Pull up the following revisions(s) (requested by martin in ticket #1023):
lib/libc/gen/posix_spawn.3: revision 1.5
lib/libc/gen/posix_spawn_file_actions_addopen.3: revision 1.4
lib/libc/gen/posix_spawn_file_actions_init.3: revision 1.4
lib/libc/gen/posix_spawn_fileactions.c: revision 1.3
sys/compat/netbsd32/netbsd32_execve.c: revision 1.38
sys/kern/kern_exec.c: revision 1.373

Limit the amount of kernel memory a posix_spawn syscall can use (for
handling the file action list) by limiting the maximum number of file
actions to twice the current file descriptor limit. Fix a few bugs in
the support functions and document the new limit. From Maxime Villard.

(sborrill)

Carefully set IFF_OACTIVE in mvgbe_start(). Even if mvgbe_encap()
returns other than 0, the TX ring might not full. Check whether
the TX ring has one or more packets. If the ring is empty,
dont' set IFF_OACTIVE because an TX complete interrupt never
occur and IFF_OACTIVE flags is left. The interface's timer
isn't reset, so a device timeout desn't occur.

Fixes a bug that IFF_OACTIVE flag is left on heavy load.
Part of PR#48568.

(msaitoh)

- npfctl: fix table IDs (breakage since the table naming was added).
- libnpf: remove npf_table_exists_p() from public API.

(rmind)

remove stale comment and explain what we are doing. thanks dsl

(christos)

Minor fpu initialisation cleanups:
Set default CR) so that the FPU is enabled (unset CR0_EM) and initialise
  i386_fpu_present to 1.
No need to call the npx trap indirectly, rename to fpunda() to match amd64.
Remove the i386_fpu_exception variable and sysctl (It used to indicate
  which irq was used for fpu exceptions, but we only support 'internal'
  now).  Hopefully no one cares.
fpuinit() now only needs to clear TS before the fninit(). Apart from the
  checks for 486SX and the 'fdiv bug' this matches the amd64 version.
Exclude fpuinit() from XEN kernels, they don't call it - which rather begs
  the question as to whether it is needed at all!

(dsl)

Fix threading bug again.

(christos)

bump date for previous

(dholland)

Reorg second paragraph too, to group related info together.

(dholland)

Rework description for clarity; prompted by chat comments from bad@.

(dholland)

Sort errors. Bump date for previous.

(wiz)

Bump date for previous.

(wiz)

Misc cleanup of sio drivers.

- register softc of child devices as an argument of the sio interrupt
  handler to avoid device_lookup_private() calls on every interrupt
- change type of sc_reg from (void *) to (struct sioreg *)
  to avoid weird address offset calculations with redundant casts
- rename struct sio_softc members for consistency
- use a channel number passed via sio_attach_args in lunaws_attach()

No functional chnages.  Tested on LUNA.

(tsutsui)

Remove paranthesis from return operands.

(martin)

Cosmetics: return is an operator, not a function: remove ().

(martin)

Limit the amount of kernel memory a posix_spawn syscall can use (for handling
the file action list) by limiting the maximum number of file actions to
twice the current file descriptor limit.
Fix a few bugs in the support functions and document the new limit.
From Maxime Villard.

(martin)

Wording. New sentence, new line. filesystem -> file system.

(wiz)

Mark up HZ.

(wiz)

Delete duplicate symbol definition introduced in 1.371. Now builds again.

(dogcow)

Add a few more test values from the range that sparc64 previously would
have got wrong.

(martin)

Fix unsigned 64 bit int to long double conversion for numbers that would not
fit into a signed 64 bit int. Found by latest t_floatunditf test case
failuer, hint from Matt Thomas.

(martin)

Add EMUL_NATIVEROOT so that native binaries can be told to search an
"emulation" directory before the real root. This makes easier to test
an amd64 kernel on the top of an i386 root filesystem prior a full
migration.

(manu)

display the tempdir for exported files as well.

(mrg)

call gzip directly for distfile export target; it's usually run outside
of the normal build.

(mrg)

add lua-bozo.c.

(mrg)

update for 2014-02-01.

(mrg)

mention how HZ affects us

(christos)

Pass DBG to crunchgen so it can get into the created mk file.

(matt)

Add missing Kyuafile for usr.bin/vmstat tests.

(jmmv)

Since powerpc passes 8 arguments in registers and the syscall number in r0,
shuffle register argument so the kernel won't need to access the stack to
retrieve that last argument.

(matt)

Revert previous: calling fpuinit() leads to a panic, as a domU is not
allowed to manipulate cr0 directly. Xen doesn't need this, the fpu is
handled by the hypervisor.

(bouyer)

Stopgap measure: set i386_fpu_present to 1 for XEN.
Stop a stream of
npxintr: came from nowherenpxintr: l = 0xc17ba560, curproc = 0xc177d880, fpu_present = 0
when running some ATF tests:
http://www-soc.lip6.fr/~bouyer/NetBSD-tests/xen/HEAD/i386/201401300920Z_anita.txt

XXX this is probably still broken for native i386, at last for older CPUs.
It looks like on newer CPUs (such as the one emulated by qemu) this code path
is not used.

(bouyer)

Add a direct call to fpuinit().
I'm not sure this is architecturally the best place, but I think it
  is where npxattach() used to get called.
Might fix FP support in domu

(dsl)

Make sljit usable on sparc64 32-bit kernels.

(nakayama)

All EXT_EXP defines must be conditionized.  Back out previous change which
removed that.

(matt)

Export __HAVE_LONG_DOUBLE 128 to kernel for emulation

(matt)

Move wrongly placed ifdef.

(matt)

Skip testcase inside its body for architectures w/o long double support

(martin)

Improve the compiler_rt entry slightly

(skrll)

Advise a clean build for the compiler-rt migration.

(joerg)

sparc and sparc64 have moved to new world order, remove legacy files.

(martin)

Follow rev. 1.54, 1.55 of libexec/ld.elf_so/arch/sparc64/mdreloc.c.

(nakayama)

Make the constants describing extended format available even if we do
not have long doubles.
This fixes the sparc kernel build (which uses the constants in FPU emulation
code).

(martin)

Print a slightly more helpfull message in case of test failure

(martin)

Fix definition of SYCALL_ARG_PTR
Add SYCALL_ARG_PTR_P(sy)

(matt)

Deal with unaligned DIR32 relocs as found in debug information.

(skrll)

Only define ieee_ext_u of __HAVE_LONG_DOUBLE == 128

(matt)

Typos.

(joerg)

Fix RCS ID.

(joerg)

Be nice to LLVM and use an explicit immediate.

(joerg)

Fix __HAVE_LONG_DOUBLE tests

(matt)

Catch up with _lwp_park change.

(joerg)

Restore -Wno-error for clang build. Just don't try to force GCC 4.1 for
clang, okay?

(joerg)

Fix prototype in SYNOPSIS, mark up NULL with Dv.

(wiz)

bump libc for ___lwp_park60

(christos)

expose __lwp_park60

(christos)

__lwp_park60

(christos)

remove compatibility code for handling CLOCK_MONOTONIC and handle it in the
syscall directly.

(christos)

Make sure to define NOMAN before including <bsd.own.mk>.

(tsutsui)

Just use EXT_EXP_INFNAN instead of EXT_EXP_INF or EXT_EXP_NAN.

(matt)

Consolidate the 128-bit long double defintions to <sys/ieee754.h>
Each arch that uses it now defines __HAVE_LONG_DOUBLE to 128.
<machine/ieee.h> is now just include the machine's math.h followed
by <sys/ieee754.h>

(matt)

Workaround weird "nbmake[8]: Graph cycles through `clean'" errors.

(tsutsui)

PR/44756: Sad Clouds: Prevent leakage of errno = ESRCH from _lwp_park. This
has two parts:
- in pthread_cond_timedwait() if the thread we are trying to unpark
  exited, retry the the _lwp_park call without it.
- pthread_mutex() was affecting errno since it is calling _lwp_park()
  from pthread_mutex_lock_slow(). preserve the original errno.
Note that the example problem still causes an occassional deadlock on machines
with many CPUs and it is the same deadlock we observe with named.

(christos)

defined(LKM) -> defined(MODULAR)

(tsutsui)

defined(LKM) -> defined(MODULAR)

(tsutsui)

Fix build botched in previous.

Christos, please at least compile before commit per the guideline.

(tsutsui)

Move the startup routine to the beginning of the source as other m68k ports.

No particular comments on port-x68k@.

(tsutsui)

Switch ews4800mips and newsmips to Xorg server.

There is ~no user of X11R6 Xservers except a few developers on these ports
and Xorg wsfb server should work with manual keymap settings.

(tsutsui)

Update md.ews4800mips list for xorg-server 1.9.2 merge.

(tsutsui)

Use ioconf file to generate config data structures

(pgoyette)

Provide ioconf file to generate config data structures

(pgoyette)

Replace home-grown config with standardized calls to
config_{init,fini}_component()

(pgoyette)

Since N32/N64 pass the first 8 arguments in registers, we can load all
the arguments into those registers before invoking the syscall having
the kernel syscall handler from having to do that.

(matt)

Add WSDISPLAYIO_LINEBYTES ioctl (for Xorg wsfb driver).

(tsutsui)

Remove unnecessary #if 1/#endif.

(tsutsui)

Add inclusion protection.
Convert to ieee_ext to use uint64_t
Format uses implicit NBIT

(matt)

Remove old compiler based crt stuff for alpha mips powerpc and vax

(matt)

Convert to uint64_t

(matt)

Fix the naked char __INT8_TYPE__ and unsigned char case.

(matt)

Add a letter, fix a word.

(wiz)

Add a note for the libc/compile_rt changes requiring a partial clean.

(martin)

Move back call to pci_mode_detect() outside of #ifdef PCI_BUS_FIXUP.
Even if mode is not used, the call to pci_mode_detect() is mandatory to
initialize the PCI subsystem.
Fix "panic booting -current DOM0" reported by Patrick Welche on port-xen.

(bouyer)

-Wno-error=missing-prototypes is not available on gcc 4.1

(martin)

fix a problem to set sample rate of AC97 codec.

(ryo)

More symbols

(skrll)

Fix TARGET_LINKER_EABI_SUFFIX to actually use the right suffix.  Since this
now depends on the default float-abi, pick from two strings to select the
right suffix.

(matt)

Regen for missing emulations.

(matt)

Use separate emulations for EABI hardfloat / softfloat.

(matt)

Add armelf{,b}_nbsd_eabihf emulations for MKCOMPAT to use.

(matt)

Append -Os to DBG, not overwrite in case DBG has more than just -Ox from
sys.mk

(matt)

No unused functions if there is no long double support.

(joerg)

Turn off attribute warning errors for clang in
pixman-mmx.c (__artificial__).

(riz)

Add unordtf2.c

(matt)

Make confiditon on FLOAT128

(matt)

unord for float128 long double

(matt)

Make wsdisplay_vcons likelier to be modularizable too.

(riastradh)

Redo amd64/NO_DRM config and add amd64/DRMKMS.

(riastradh)

Tweak genfb tweaks so they work.

(riastradh)

Put EHABI exception tables at the end of text.
provide __exidx_start and __exidx_end symbols.

(matt)

Provide a simple floatunditf implementation, at least mips64 needs one.

(joerg)

Use fixunssfsi and fixunsdfsi from compiler-rt.

(joerg)

Add test for uint64 to real long double conversion.

(joerg)

Add missing cast (to make it compilable)

(martin)

Install a pkg-config file for zlib.

(wiz)

switch to unified syntax

(matt)

ldrht is v6t2 or later so use ldrt and extract the right halfword.

(matt)

move a variable into the scope of its use, where the assignment
has already validated the pointers used.  fixes a bug reported
in private email from dogcow@.

(mrg)

Make {,U}INTPTR_TYPE long

(matt)

Print uboot_args using PRIxPTR

(matt)

For BE kernels, pass --be8 to the linker

(matt)

Deal with the case when __UINT<N>_TYPE__ is defined as well
__INT<N>_TYPE__ is defined.  We make the assumption that if the former is
defined, the latter is also defined.

(matt)

Include compiler-rt in libc, libm and libkern.

(joerg)

For the tools build, always use the Bison-created m68 asm parser.
Fixes build on Linux, where the glibc guards trip up the
"must-be-included-first" check in config.h.

(joerg)