Add latent AArch64 support.

(matt)

If __ARM_PCS_AAPCS64 is defined, define struct ieee_ext and union ieee_ext_u.
Use uint64_t so we can just have ext_frach and ext_fracl.

(matt)

Fix inverted GCC version test

(matt)

if __ARM_PCS_AAPCS64 is defined, expose LDBL_* and DECIMAL_DIGIT

(matt)

Also test for __AARCH64EB__ as well as __ARMEB__ for big-endian.

(matt)

AAPCS64 supports 16-byte long doubles

(matt)

Make multiple inclusion symbols always start with _ARM_ (not _ARM32_ or
_MACHINE_).

(matt)

replace spaces in SHRT_MIN definition with tabs

(matt)

SIZE_T is unsigned so SIZE_T_MAX needs to be ULONG_MAX

(matt)

Add AAPCS64 fenv_t

(matt)

Add a check for minimum compiler compatible with gcc 4.1
__ARM_PCS_AAPCS64 also requires __ALIGNBYTES to be 7

(matt)

change include protection from _ANSI_H_ to _ARM_ANSI_H_

(matt)

Don't use _lock_cas explicitly, use atomic_cas_ulong which could be one its
aliases.

(matt)

Mark a potentially unused variable (depending on #ifdef)

(martin)

Allow building with clang.
XXX __RENAME is currently not allowed for the kernel, so use plain asm.
Clang rejects definitions of builtins, so it doesn't work without.

(joerg)

Deal with __ARM_ARCH_8A__

(matt)

One __RCSID is enough. Don't define it when building catalog as tool.

(joerg)

As discussed on tech-toolchain: untill gcc is fixed (or our ABI changed, ...)
make stack protection warnings non-fatal if using gcc 4.8 on sh3/m68k/vax.

(martin)

Remove an unused variable

(martin)

Simplify netbsd32_vm_default_addr (for now) to make it compilable.

(martin)

only limit bottom-up allocations, put back 3G limit on datasize.

(christos)

Note National Semiconductor PC87560 ohci(4) support.

(skrll)

Support the OHCI in the NS "SUPERIO" chip as found in hppa machines by
twiddling the same bits as Linux.

(skrll)

Replace the RET macro with mov pc,lr unconditionally - this is armv4
specific source.
Ok: matt@

(martin)

no need to duplicate the generic fabs_ieee754.c here

(macallan)

Clarify that -l params must be given after options.

(pooka)

* make -h actually work
* give usage for invalid usage

(pooka)

rcs id police *uuu-aaa, uuu-aaa*

(pooka)

split fabs.c into hardfloat .S and softfloat .c as requested by joerg

(macallan)

Initialize a variable earlier, so it is not used unitialized in an
error path.

(martin)

Bogus gcc 4.8 maybe-used-uninitialized warning

(martin)

Bogus gcc 4.8 maybe use uninitialized warning

(martin)

Mark a diagnostic-only variable

(martin)

Mark a diagnostic-only variable

(martin)

Remove an unused variable

(martin)

Quell a gcc 4.8 maybe-unitialized false positive

(martin)

Quell maybe-uninitialized false positives from gcc 4.8

(martin)

Quell a (bogus) "may be used unintialized" warning from gcc 4.8

(martin)

Mark a diagnostic only variable

(martin)

Remove unused variables

(martin)

Remove an unused variable

(martin)

Remove an unused variable

(martin)

Unused variable police

(martin)

Build with IPv6 support.

(njoly)

HAVE_GDB can trigger a false positive if MKGDB=no so add a check for it
like we do for HAVE_GCC.

(matt)

Cut down MAXDSIZE from 3G to 2.5G otherwise bottomup allocation ends up
supplying an out of bounds hint for sigcode (c001e000 > bf000000). Makes
a.out binaries work again.

(christos)

add a struct for the 4.3BSD struct direct

(christos)

Fix the compat-4.3 getdirentries call (pre d_type). This is used in NetBSD-0.9.

(christos)

Stopgap to prevent genfb from stealing console.  Revisit later.

(jakllsch)

Pass the -n flag to gzip invocations.

This prevents it from embedding a timestamp in the output.  We pass
"-n" unconditionally, not conditional on MKREPRO, because many other
invocations of gzip already passed the -n flag unconditionally.

(apb)

This is 6.1.3_PATCH, not (yet) 6.1.4

(bouyer)

Welcome to 5.2.2_PATCH (missing bits)

(bouyer)

Welcome to 5.1.4_PATCH (missing pieces)

(bouyer)

Welcome to 6.1.4_PATCH.

(bouyer)

Welcome to 6.0.4_PATCH.

(bouyer)

Update CIOGFSESSION argument spec to match reality

(pgoyette)

Welcome to 5.2.2_PATCH.

(bouyer)

Welcome to 5.1.4_PATCH

(bouyer)

add today's releases:  5.1.4, 5.2.2, 6.0.4, and 6.1.3

(jnemeth)

Add missing __RCSID("$NetBSD$");

(matt)

fix c&p bug.

(matt)

Return old

(matt)

Add basic IPV6 level socket options support (IPV6_V6ONLY).

(njoly)

Fix tpyo (add->and)

(matt)

fix tpyos.

(matt)

Add __sync_<OP>_and_fetch_<N> aliases

(matt)

Rework so that __sync_fetch_and_<OP>_<N> actually returns a value.

(matt)

Add __sync_fetch_and_<OP>_<N> aliases

(matt)

Add __sync_fetch_and_add_8 alias

(matt)

Add __sync_fetch_and_add_4 alias

(matt)

Add _atomic_cas_16_up and _atomic_cas_8_up

(matt)

Add _atomic_cas_16 and _atomic_cas_8 and their aliases
__sync_val_compare_and_swap_2 & __sync_val_compare_and_swap_1

(matt)

Add SO_SNDLOWAT, SO_RCVLOWAT and SO_ACCEPTCONN support for socket
options.

(njoly)

back out 1.35 since it completely breaks the mechanism for wsdisplay drivers
to determine wether to become console or not.
See http://mail-index.netbsd.org/tech-kern/2014/01/15/msg016421.html
for fixing this properly.

(macallan)

Use ${TOOL_GZIP} instead of just gzip in all Makefiles
outside */dist/* subdirectories.

When USE_PIGZGZIP=yes, bsd.own.mk sets TOOL_GZIP=${TOOL_PIGZ},
so there's no need to test USE_PIGZGZIP in these Makefiles.

(apb)

Document TOOL_GZIP.

(apb)

explain why forks fail

(christos)

little manual fixing

(christos)

add netbsd rcsids (does not handle the regex files well)

(christos)

Fix RCSID's

(christos)

Adjust trigger variable for unwind.h to follow changes in libc.

(joerg)

Fix http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57260 (patch from
gcc's repository). This bugs show up with Xorg, preventing  xf86-video-sis and
xf86-video-silicon modules from being loaded.

(bouyer)

Complete WSDISPLAYIO_SMODE/WSDISPLAYIO_GMODE support.
In sisfb_mmap(), allow mapping the framebuffer and mmio and io registers
when in WSDISPLAYIO_MODE_MAPPED mode. This can be used by X11.

(bouyer)

Support using fenv instead of fpsetmask if HAVE_FENV is defined.

(matt)

Make this real even though it isn't yet used.

(matt)

sun4v: Add handling of spill/fill and dtsb miss traps (with a XXX to be handled later)

(palle)

Remove support for 'external' floating point units and the MS-DOS
  compatible method of handling floating point exceptions.
Make kernel support for teh fpu non-optional (486SX should still work).
Only 386 cpus support external fpu, and i386 support was removed years ago.
This means that the npx code no longer uses port 0xf0 or interupt 13.
All the "npx at isa" lines go from the configs, arch/i386/isa/npx.c
  is now mandatory for all i386 kernels.
I've renamed npxinit() to fpuinit() and npxinit_cpu() to fpuinit_cpu()
  to match the very similar amd64 functions.
The fpu of the boot cpu is now initialised by a direct call from
  cpu_configure(), this enables FP emulation for a 486SX.
  (for amd64 the cr0 values are set in locore.S and similar).
This fixes a long-standing bug in linux_setregs() - which did not
  save the fpu regsiters if they were active.
I've test booted a single cpu i386 kernel (using anita).
amd64 builds - none of teh changes should affect it.
The i386 XEN kernels build, but I'm not sure where they set cr0, and
  it might have got lost!

(dsl)

Recognize coldfire

(matt)

Add BUILDSEED to RELEASEVARS.

(apb)

For MKREPRO, remove the "%%CreationDate:" comment line from
PostScript output files.

(apb)

Add MKREPRO and MKARZERO to the list of variables in RELEASEVARS.

(apb)

Fix build problem when there is no com(4) but ucom(4).

(taca)

PUCCN improvements:
- Fix a bug that the puc cn mechanism doesn't use the UART's frequency
  in pucdata.c's table.

- Add a new option PUC_CNAUTO. If this option is set, consinit() in
  x86/x86/consinit.c checks puc com device to use it as console.
  Without this option, the behavior is the same as before.

- Add a new config parameter PUC_CNBUS. The old code scans bus #0 only.
  If PUC_CNBUS is set, the specified number's bus will be scanned.

- Rename comcnprobe() to puc_cnprobe() to make it clear.

- Rename comcninit() to puc_cninit() to make it clear.

- Add code for a device that a device's com register is MMIO (#if0 ed).

(msaitoh)

Add some OXPCIe952 devices.

(msaitoh)

Regen.

(msaitoh)

Add some OXPCIe952 devices.

(msaitoh)

spelling correction - recieved -> received

(plunky)

don't re-specify the defaults

(christos)

stoa() already includes the scope, no need to print it twice.

(mlelstv)

Fix the "separate root" test. If we get a permission error check for separate
root. But if we get ENOENT, always succeed. Otherwise if someone removes
the db files or they get corrupted, you can never recreate them.

(christos)

whitespace

(christos)

kill VM_DEFAULT_ADDRESS use.

(christos)

kill VM_DEFAULT_ADDRESS use

(christos)

kill the topdown flag only if we succeed.

(christos)

Clear the VM_TOPDOWN flag only when we succeed.

(christos)

use the topdown hint.

(christos)

fix unused

(christos)

fix unused

(christos)

remove unused

(christos)

add __USE

(christos)

add a lint comment

(christos)

Change the way the x87 fp register is defined so that the copies between
  fsave and fxsave layouts can be done by structure assignments.

(dsl)

Rewrite the code that converts the x87 tag word between its full and
  abridged versions.
The code in xmm_to_s87_tag() looked at the wrong register when trying
  to generate 0b01 (zero) and 0b10 (NaN etc) because the registers are
  relative to the stack top, but the tag words are absolute.
Since even a real 387 generates the correct tag values when the tag word
  (or register value) is loaded there is no point generating other than
  0b00 (used) and 0b11 (unused).
The compiler also made a pig's breakfast of the register copy loop.
When convering from fsave to (the larger) fxsave format don't bother
  copying the fp register values if none are in use.
Converting the other way we do copy them all because the target buffer
  is often written to userspace (and has no padding).
i386 core dumps still seem ok - and do use this code.

(dsl)

__USING_TOPDOWN_VM is no more, __USE_TOPDOWN_VM...

(christos)

__USING_TOPDOWN_VM is gone.

(christos)

help lint.

(christos)

Add a TODO file to document things to be done for the sparc64 port

(palle)

add myself

(wiedi)

Improve PCU/VFP handling to the point that the atf tests don't trigger
KASSERTs on the Raspberry PI and its arm1176jzf-s.

XXX Need to emulate bounce instructions to get correct exception codes,
XXX etc.

(skrll)

make this compile.

(christos)

Add compat_10, open NULL == open "."

(christos)

deal with COMPAT_10 issue.

(christos)

regen

(christos)

Get SIZE_MAX and OFF_MAX straight...

(riastradh)

Factor out an offtab_compute_window_position routine.

(riastradh)

Fix some more integer overflow/truncation issues.

Arithmetic in C is hard.  Let's go shopping!

(riastradh)

handle non-topdown binaries

(christos)

delete VM_DEFAULT_ADDRESS and commentary which is no longer used/true.

(christos)

delete VM_DEFAULT_ADDRESS; some of those should be GC'ed because they match
the default definition.

(christos)

Update linux socket socket options (SO_*). Add common version to its
own file (linux_socket_generic.h); use it for all but alpha and mips.

(njoly)

Add now needed linux_pipe.c sources.

(njoly)

Regen for pipe2 and lchown.

(njoly)

Add pipe2 and lchown syscalls.

(njoly)

More alignment spellos

(skrll)

Use English spelling of alignment.

(skrll)

a.out binaries can't handle topdown.
Now 1.0 binaries work correctly on NetBSD-current.

(christos)

a.out binaries can't handle topdown.

(christos)

provide proper defaults for topdown and bottomup allocation.
XXX: Ports that provide their own VM_DEFAULT_ADDRESS() need to provide the
two new flavors, otherwise they get the default ones now.

(christos)

provide propert address defaults for topdown and bottomup allocation

(christos)

ifdef debug printf

(christos)

expose do_open

(christos)

compat_10 fix: open(NULL) == open(".");

(christos)

fix monitor mode channel.

(christos)

Increase VM_KERNEL_PT_PAGES from 2 to 4 for x68k.

X68030 with 8MB or 12MB RAM (which is default for most X68030s)
now gets "pmap_enter_ptpage: can't get KPT page" panic again
with VM_KERNEL_PT_PAGES=2 while it works with >=14MB.
3 is enough to boot for now, but I also add a spare.

I guess this is because the post netbsd-6 vmem(9) bootstrap
changes require more kernel memory before MD pmap_init().

Note1: netbsd-6 (which also pulls physmem detection changes
that affect nptpages) doesn't have this problem.
Note2: other hp300 pmap based m68k ports might have the same issue,
but at least 12MB luna68k works with VM_KERNEL_PT_PAGES=2
so I think it depends on kernel size and/or iomapsize.

(tsutsui)

close uhidev only when it was successfully opened.

(mlelstv)

Add raid to the install kernel

(martin)

Add raid disk nodes for the install images

(martin)

Fix awk programs to use $$ so make doesn't eat each $

(matt)

remove useless flag mangling. It is done already.

(christos)

need to lock the process.

(christos)

It is wishful thinking that vn_readdir will return dirent12 structures.

(christos)

sigaction until 1.4 had an int sigmask, don't trash the stack.

(christos)

Fix locking botch, and ...
Yes, we want to be careful because the sizes are different.

(christos)

CID 1164169: integer overflow

(christos)

CID 1163170, 1164171, resource leak
CID 1161172 double free

(christos)

As requested by mrg@, since there is still a small window during which
the in-module ref-counting can fail, completely disable auto-unload.

(pgoyette)

Slightly bump ramdisk size, so the content fits even when compiled with
gcc 4.8 (usually binaries are smaller or same size, this increase is
strange)

(martin)

Support WSDISPLAYIO_GETCMAP/WSDISPLAYIO_PUTCMAP. Tested on evbmips/loongson

(bouyer)

Snapshot of work in progress:
Add support xorg wsfb driver. Tested on evbmips/loongson

(bouyer)

Add ATAIOCCOMMAND ioctl form COMPAT_NETBSD32

This enables SMART monitoring by a netbsd32 binary.

(manu)

Load the magic into the right register for the comparison in _longjmp

(skrll)

Be consistent about setting fpscr for Runfast.  No functional change.

(skrll)

Deal with writing bswapped com registers at COM_MULT register widths.

(matt)

Add a flag so the xputc will bswap registers.

(matt)

Replace copyright with TNF
Clear upper 4 bits of pc-relative offset for uboot_args since .start is
1GB below .text so we can access using the load address, not the KERNEL_BASE
address.

(matt)

After loading the pc-relative offset, clear the upper the 4 bits incase
.start starts at a different address than .text (bcm56340).

(matt)

use initarm_common()

(hkenken)

add missing library

(christos)

add debuglibs

(christos)

fix typo

(christos)

remove bogus comment, we are not working around an assembler issue.

(christos)

Checking _LP64 to determine the size of the built binary is just wrong. We
build n32 by default on mips64 now, we don't handle the compat build which
is n64. This is the same case for sparc64/sparc and amd64/i386 (lint does
not run in compat)

(christos)

+ szptvlfn.

(szptvlfn)

Remove time= keywords from METALOG.sanitised.

The times add unnecessary differences to the DESTDIR trees for MKREPRO
builds, and they are not needed by anything.  (maketars, the chief
consumer of METALOG.sanitised, explicitly ignores the time fields.)

Also factor out some duplicated code and update comments.

(apb)

Removed MAXBSIZE. Cleared with Anders Magnusson and tested myself.

(jklos)

Compile the armv4 cpufunc file with -mcpu=arm8 instead of arm9, to avoid
an armv4t marked object file which causes failure in the shark objcopy
step when creating the a.out version of the kernel.

(martin)

For armv7, CPU_CONTROL_UNAL_ENABLE is MBO/SBPO.
Make sure CPU_CONTROL_EX_BEND is set for big endian kernels.
Don't clear bits we aren't setting.

(matt)