Don't use #pragma weak to implicitly change argument types.

(joerg)

dmu_objset_rename is an alias for dsl_dataset_rename, but differs in
constness of the first argument. Newer clang checks for this and
prohibits it, so just be consistent by constifying the latter.

(joerg)

switch powerpc ports to new binutils

(macallan)

add cpu_lock

from freqlabs on irc

(pooka)

avoid re-def

(christos)

PR/50827: Richard Hansen: Fix default variable assignment with arithmetic,
from kre.

(christos)

undo accidental commit

(christos)

Add a test for PR/50827

(christos)

Initialize curFile->depending in Parse_SetInput.

(sjg)

Explicitly cast between char and unsigned char here.

(riastradh)

Need <stdbool.h> for true/false.

(riastradh)

Need <stdbool.h> for true/false.

(riastradh)

Various housekeeping.

- Include <ufs/lfs/lfs.h> for union lfs_dinode &c.
- Include <string.h> or <sys/systm.h> for memcpy.
- Avoid signedness mismatch in lfs dino accessor for `rdev'.
- Avoid shadowing global `index'.

(riastradh)

getBoolean:
We need :U to ensure we get an empty string if knob isn't set.

(sjg)

Delete :@ loop iterator when done with it

(sjg)

gcc does not pass floating point options to the assembler
by default, because it is afraid that the stricter tests
will break userland code. The new binutils is pickier about
this. Gcc 5.x fixes the issue so for now, set explicitly
the assembler soft-float flags when we build the kernel.
see: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64569

(macallan)

Adapt to changed Var_Subst()

(sjg)

Add .export-literal to avoid the need for $$ dance when trying to put
unexpanded variables into environment.

Reviewed by: christos

(sjg)

Collapse the 3 boolean parameter to 1 flags parameter. No functional change.

(christos)

Avoid reading one byte past end of array.

Fixes

  acpiout0: brightness levels: [20,25,30,35,40,45,50,55,60,65,70,75,80,85,90,uvm_fault(0xffffffff8133d200, 0xffff80000c1d8000, 1) -> e
  fatal page fault in supervisor mode
  trap type 6 code 0 rip ffffffff803b63d7 cs 8 rflags 10212 cr2 ffff80000c1d8000 ilevel 8 rsp ffffffff816a1b00

(riastradh)

It's 2016

(sjg)

Actually FALSE as last arg to Var_Subst makes more sense

(sjg)

Add support for .dinclude

Like .sinclude missing file will be ignored.
Like .depend stale dependencies will be ignored.
Allows better implementation of auto depend.

Reviewed by: christos
Requested by: Bryan Drewery at FreeBSD.

(sjg)

Adapt to new Var_Subst prototype

(sjg)

Adapt CI20 HWRNG to synchronous on-demand callback.
Omit needless softint/locking dance.
from riastradh@

(macallan)

PR/50815: David Binderman: Remove dup test

(christos)

PR/50818: David Binderman: Delete dead code (copied from if_et.c)

(christos)

PR/50817: David Binderman: Delete dead code (copy from if_et.c)

(christos)

PR/50819: David Binderman: Delete dead code.

(christos)

PR/50820: David Binderman: Fail when !(edge|pulse) instead of !edge && edge

(christos)

PR/50821: David Binderman: remove "i <" i <

(christos)

PR/50822: David Binderman: Fix copy length.

(christos)

PR/50823: David Binderman: Limit scanf width

(christos)

PR/50824: David Binderman: Limit scanf width

(christos)

whitespace and header sorting changes (Ingo Schwarze). No functional changes.

(christos)

Don't schedule a softint if we have nothing to do.

Some systems seem to have gotten stuck in a softint processing loop
doing nothing and then trying to do it again.  Might fix gson's
frozen qemu/anita sparc autobuilds -- tested on macallan's real sparc
hardware and confirmed to fix at least some freeze at boot.

(riastradh)

add a shadowfb update function which endian-swaps 32bit pixels, use if we
have hardware with BGRA pixel format ( like a wrong-endian ARGB framebuffer )
with this applications that use xrender work properly on things like mgx in
32bit colour

(macallan)

extend the match function, to match devices from a selection of
manufacturers that use modern Broadcom chips, and which represent
as a vendor-specific device class.

(plunky)

Need <sys/mutex.h> for mutex(9).

(riastradh)

Caller must have exclusive access to rndsource for rnd_add_data(_sync).

(riastradh)

Make hyperentropy rndsource work synchronously, again.

This time for real!  *crosses fingers*

(riastradh)

Move entropy_count into the scope where it is used.

Omit now-unused definition.

XXX This code probably shouldn't be under DIAGNOSTIC...  It is
significantly more than just an assertion.

(riastradh)

Use real atomics, not atomics faked with membars, for rnd_printing.

(riastradh)

Omit needless call to rnd_getmore in rnd_extract_data.

The only remaining caller -- rnd_extract -- already does it.

(riastradh)

Omit call to rnd_getmore from rnd_process_events.

There are three cases to consider here:

1. You have on-demand synchronous rndsources, e.g. RDRAND (and you're
willing to use it).  In that case, you'll just use those whenever you
need to extract entropy.  There's no benefit to requesting more from
them when we're entering data into the pool.

2. You have on-demand asynchronous rndsources.  These should continue
trying to feed data to the entropy pool as long as it needs more, so
there's no benefit to requesting more from them when we're entering
data into the pool.

3. You don't have any on-demand rndsources.  Then rnd_getmore does
nothing, so there's no benefit to calling it.

ok tls

(riastradh)

Match various Apple USB Bluetooth controllers.

From mlelstv.

(riastradh)

regen

(riastradh)

Add various USB Bluetooth controllers.

From mlelstv.

(riastradh)

Add rnd_add_data_sync for synchronous on-demand rndsource callbacks.

Avoids lock recursion in rndsinks:

rndsink user holds lock
calls rndsink_request
calls rnd_getmore
calls an rndsource callback
calls rnd_add_data
calls rnd_process_events
calls rndsinks_distribute
calls rndsink callback
acquires rndsink user lock.

This can happen only before the rnd_process_events softint is
established, but that's enough to cause a problem and is the main
reason why all our HWRNG drivers are so byzantine (soon to be
fixed!).

ok tls

(riastradh)

Cast uint8_t to uint32_t before shifting left by 24.

Otherwise this is undefined behaviour unless int is 32-bit and the
uint8_t value happens to be <=127.

(riastradh)

Need <sys/queue.h> for LIST_ENTRY and <sys/null.h> for NULL.

(riastradh)

More header cleanups from Ingo Schwarze.

(christos)

- don't set _GNU_SOURCE. We are not supposed to make decisions for others.
- don't special-case wcsdup()
From Ingo Schwarze.

(christos)

get rid of bool_t (Ingo Schwarze)

(christos)

more include file cleanup (Ingo Schwarze)

(christos)

include errno.h

(christos)

From Ingo Scharze:
Let "el.h" include everything needed for struct editline,
and don't include that stuff multiple times.  That also improves
consistency, also avoids circular inclusions, and also makes it
easier to follow what is going on, even though not quite as nice.
But it seems like the best we can do...

(christos)

cleanup chartype.h includes (Ingo Schwarze)

(christos)

one more

(christos)

cleanup inclusion of histedit.h (Ingo Schwarze)

(christos)

include explicitly errno.h since we use it.

(christos)

No need to include "sys.h" from here; it is included from config.h

(christos)

In xennet_xenbus_detach(), remove the event handler early (just after
xennet_stop()) so that we don't get events while slepping (e.g.
in softint_disestablish()) when some structures have already been
freed.
Problem reported and patch tested by Rohan Desai.

(bouyer)

Re-enable the TT support that existed previously.

(skrll)

Reduced INSTALL kernel size.

(nonaka)

Remove workaround for GATEWAY

The workaround was introduced because lltable/llentry uses rwlock
but it may be executed in hardware interrupt due to fast forward.
Now we don't run fast forward in hardware interrupt anymore, so
we can remove the workaround.

(ozaki-r)

attribute unused

(christos)

OpenBSD term.c rev. 1.7 2002/11/29 20:13:39 deraadt
spelling

(christos)

OpenBSD readline.c rev. 1.14 2015/02/06 23:21:58 millert
use SIZE_MAX

(christos)

cardbus autoconf uses cardbus_attach_args, not isapnp_attach_args.

(riastradh)

OpenBSD readline.c rev. 1.13 2015/01/13 08:33:12 reyk
rl_set_keyboard_input_timeout() for readline 4.2 compat

(christos)

OpenBSD eln.c rev. 1.3 2011/11/27 21:46:44 pascal
kill a C++-style comment

(christos)

Compile with WIDECHAR the same way the main Makefile does (Ingo Schwarze)

(christos)

Don't free getline memory (Ingo Schwarze).

(christos)

allllllexander the greeeaaaaat, his name struck fear into hearts of men!

...but let's conquer that fear and give ol' alex a consistent and
correct date of death.

(he died of fever in baaaaaabylooooooonnnn)

(snj)

Use KASSERTs supported by CTASSERTs, not __builtin_unreachable.

pcc has no __builtin_unreachable, and this is clearer anyway.

(riastradh)

fix english

(snj)

Print error code for failure to map PRAMIN BAR.

(riastradh)

Fix building of IPv4-Mapped IPv6 addresses.

As discussed on tech-net@ use in6_sin_2_v4mapsin6() to build mapped
addresses.

(rtr)

forgot one fgetln define

(christos)

change tests for fgetln.

(christos)

Fix function name, no functional change.
Found by Travis Paul, (see PR/50786), thanks for reporting!

(mbalmer)

Use getline for better portability.

(christos)

OpenBSD tokenizer.c rev. 1.8 2003/08/11 18:21:40 deraadt
don't increase amax on realloc failure

(christos)

OpenBSD term.c rev. 1.13 2009/12/11 18:58:59 jacekm
fix two memory leaks

(christos)

Change the test for the size of encoded buffer to include the NULL, from
OpenBSD; no functional change.

(christos)

OpenBSD sig.c rev. 1.6 2001/12/06 04:26:00 deraadt
save and restore errno in signal handler

(christos)

Use fparseln to avoid newline hacks.

(christos)

use fparseln() to avoid needing to deal with missing \n in the last line
and also to handle comments automatically.

(christos)

This needs _KERNTYPES on some platforms.

(christos)

Reduce code duplication.

Split creation of IPv4-Mapped IPv6 addresses into its own function
and use it.

No functional change intended.  As posted to tech-net@

(rtr)

Add missing closing ) and line continuation for mknative

(martin)

Avoid strict alias violation for VAX

(martin)

Run if_link_state_change in softint

if_link_state_change can execute the network stack that is expected to
not run in hardware interrupt (at least now), however network drivers
may call it in hardware interrupt. Avoid that by introducing a new
softint for if_link_state_change.

The original patch is provided by mlelstv@ and tweaked a bit by me.

Should fix PR kern/50602.

(ozaki-r)

Simplify bridge(4)

Thanks to introducing softint-based if_input, the entire bridge code now
never run in hardware interrupt context. So we can simplify the code.

- Remove spin mutexes
  - They were needed because some code of bridge could run in
    hardware interrupt context
  - We now need only an adaptive mutex for each shared object
    (a member list and a forwarding table)
- Remove pktqueue
  - bridge_input is already in softint, using another softint
    (for bridge_forward) is useless
  - Packet distribution should be down at device drivers

(ozaki-r)

remove duplicated #include of <netinet/in.h>

(rtr)

sun4v: add breakpoint trap entry to the sun4v trap table

(palle)

zero the i2c_attach_args structure before filling it in.
fixes occasional crashes in iic_attach().

(chs)

Remove extra stray break missed by Christos when committing PR 50783.

(dholland)

Restore the local change
  -gcc: internal compiler error: Killed (program cc1plus)
  +gcc: internal compiler error: Killed (program cc1plus received signal 9)
which was lost in an update somewhere after netbsd-6.

This clarification was found in the past to significantly reduce the
number of bug reports caused by people running out of swap, such as
(most likely) seen again today in PR 50802.

XXX: pullup-7

(dholland)

Make the vax target buildable

(martin)

Add missing va_end, from David Binderman in PR 50798.

(dholland)

Add missing va_end, from David Binderman in PR 50797.

(dholland)

Add missing va_list, from David Binderman in PR 50796.

XXX: this file and arch/arc/stand/boot/disk.c are cutpaste clones.

(dholland)

Add missing va_end; PR 50795 from David Binderman.

(dholland)

Fix wrong indent.

(dholland)

Add missing va_end(). PR 50794 from David Binderman.

(dholland)

Add missing va_end. PR 50793 from David Binderman.

(dholland)

From Ingo Schwarze:

el_getc() for the WIDECHAR case, that is, the version in eln.c.
For a UTF-8 locale, it is broken in four ways:

1. If the character read is outside the ASCII range, the function
    does an undefined cast from wchar_t to char.  Even if wchar_t
    is internally represented as UCS-4, that is wrong and dangerous
    because characters beyond codepoint U+0255 get their high bits
    truncated, meaning that perfectly valid printable Unicode
    characters get mapped to arbitrary bytes, even the ASCII escape
    character for some Unicode characters.  But wchar_t need not
    be implemented in terms of UCS-4, so the outcome of this function
    is undefined for any and all input.

2. If insufficient space is available for the result, the function
    fails to detect failure and returns garbage rather than -1 as
    specified in the documentation.

3. The documentation says that errno will be set on failure, but
    that doesn't happen either in the above case.

4. Even for ASCII characters, the results may be wrong if wchar_t
    is not using UCS-4.

(christos)

From Ingo Schwarze:

As we have seen before, "histedit.h" can never get rid of including
the <wchar.h> header because using the data types defined there is
deeply ingrained in the public interfaces of libedit.

Now POSIX unconditionally requires that <wchar.h> defines the type
wint_t.  Consequently, it can be used unconditionally, no matter
whether WIDECHAR is active or not.  Consequently, the #define Int
is pointless.

Note that removing it is not gratuitious churn.  Auditing for
integer signedness problems is already hard when only fundamental
types like "int" and "unsigned" are involved.  It gets very hard
when types come into the picture that have platform-dependent
signedness, like "char" and "wint_t".  Adding yet another layer
on top, changing both the signedness and the width in a platform-
dependent way, makes auditing yet harder, which IMHO is really
dangerous.  Note that while removing the #define, i already found
one bug caused by this excessive complication - in the function
re_putc() in refresh.c.  If WIDECHAR was defined, it printed an
Int = wint_t value with %c.  Fortunately, that bug only affects
debugging, not production.  The fix is contained in the patch.

With WIDECHAR, this doesn't change anything.  For the case without
WIDECHAR, i checked that none of the places wants to store values
that might not fit in wint_t.

This only changes internal interfaces; public ones remain unchanged.

(christos)

From Ingo Schwartze:

Next step:  Remove #ifdef'ing in read_char(), in the same style
as we did for setlocale(3) in el.c.

A few remarks are required to explain the choices made.

* On first sight, handling mbrtowc(3) seems a bit less trivial
  than handling setlocale(3) because its prototype uses the data
  type mbstate_t from <wchar.h>.  However, it turns out that
  "histedit.h" already includes <wchar.h> unconditionally (i don't
  like headers including other headers, but that ship has sailed,
  people are by now certainly used to the fact that including
  "histedit.h" doesn't require including <wchar.h> before), and
  "histedit.h" is of course included all over the place.  So from
  that perspective, there is no problem with using mbrtowc(3)
  unconditionally ever for !WIDECHAR.

* However, <wchar.h> also defines the mbrtowc(3) prototype,
  so we cannot just #define mbrtowc away, or including the header
  will break.  It would also be a bad idea to porovide a local
  implementation of mbrtowc() and hope that it overrides the one
  in libc.  Besides, the required prototype is subtly different:
  While mbrtowc(3) takes "wchar_t *" as its first argument, we
  need a function that takes "Char *".  So unfortunately, we have
  to keep a ct_mbrtowc #define, at least until we can maybe get
  rid of "Char *" in the more remote future.

* After getting rid of the #else clause in read_char(), we can
  pull "return 1;" into the default: clause.  After that, we can
  get rid of the ugly "goto again_lastbyte;" and just "break;".
  As a bonus, that also gets rid of the ugly CONSTCOND.

* While here, delete the unused ct_mbtowc() from chartype.h.

(christos)

Update for latest dwc2 dist

(skrll)

Merge conflicts

(skrll)

Fix DWC2_READ_4 in dwc2_sample_frrem which is unused

(skrll)

Map the same mmio size as Linux does.

Unlikely to have practical consequences -- just reduces differences
in behaviour from upstream, to rule them out.

(riastradh)

Fix bus_space_subregion error branch.

Return error code, not 0 (!), on bus_space_subregion failure.

In answer to `XXX error branch' comment: if nouveau_barobj_ctor
fails, then the caller will call nouveau_barobj_dtor too.  So there's
no leak here.

Unlikely to fix any observed bugs with nouveau -- there's no error
branch in the Linux side here.  But maybe it will catch some other
bug earlier.

(riastradh)

DNF

(dholland)

reduce ifdefs

(christos)

Add nouveau to x86 GENERIC kernels.

(riastradh)

Pass all unicast addresses if it is promisc mode.

(hikaru)

Make all packets are enqueued into queue 0.
queue 7 is not default, it is caused by the filter tables.
The fields are including queue number, not bitfields.
So MVXPE_DF_QUEUE_ALL (b111) means queue 7.

And also, pass all unicast addresses if it is promisc mode.
MVXPE_PXC_UPM is working in almost cases,
but this change is needed for some cases; bridging frames through inter units,
using products have consecutive MAC addresses.

(hikaru)

Fix typo in sysctls.

(hikaru)

Count up if_{opackets,oerrors,ierrors,collisions}.
And also "MAC Transmit Error" is tx error, not rx.

(hikaru)

Style.

(hikaru)

Preserve rx csum bad flags.

(hikaru)

Fix mitake, disable SERDES error interrupts again.

(hikaru)

Don't use RXTX intr that is excess and requires another IRQ handler,
and disable SERDES error interrupts.

(hikaru)

Correct bus_dmamap_sync operaion.

(hikaru)

Correct locks and missing breaks.

(hikaru)

Avoid c99 for now.

(christos)

GC IGNORE_EXTCHARS and simplify code (Ingo Schwarze)

(christos)

From Ingo Schwarze:

If CHARSET_IS_UTF8 is not set, read_char() is broken in a large
number of ways:

1. The isascii(3) check can yield false positives.  If a string in
    an arbitrary encoding contains a byte in the range 0..127,
    that does not at all imply that it forms a character all by
    itself, and even less that it represents the same character
    as in ASCII.  Consequently, read_char() may return characters
    the user never typed.
    Even if the encoding is not state dependent, the assumption that
    bytes in the range 0..127 represent ASCII characters is broken.
    Consider UTF-16, for example.

2. The reverse problem can also occur.  In an arbitrary encoding,
    there is no guarantee that a character that can be represented
    by ASCII is represented by a seven-bit byte, and even less by
    the same byte as in ASCII.
    Even for single-byte encodings, these assumptions are broken.
    Consider the ISO 646 national variants, for example.
    Consequently, the current code is insufficient to keep ASCII
    characters working even for single-byte encodings.

3. The condition "++cbp != 1" can never trigger (because initially,
    cbp is 0, and the code can only go back up via the final goto,
    which has another cbp = 0 right before it) and it has no effect
    (because cbp isn't used afterwards).

4. bytes = ct_mbtowc(cp, cbuf, cbp) is broken.  If this returns -1,
    the code assumes that is can just call mbtowc(3) again for later
    input bytes.  In some implementations, that may even be broken
    for state-independent encodings, but trying again after mbtowc(3)
    failure certainly produces completely erratic and meaningless
    results in state-dependent encodings.

5. The assignment "*cp = (Char)(unsigned char)cbuf[0]" is
    completely bogus.  Even if the byte cbuf[0] represents a
    character all by itself, which it usually will not, whether
    or not the cast produces the desired result depends on the
    internal representation of wchar_t in the C library, which
    the application program can know nothing about.  Even for ASCII
    in the C/POSIX locale, an ASCII character other than '\0' ==
    L'\0' == 0 need not have the same numeric value as a char and
    as a wchar_t.

To summarize, this code only works if all of the following
conditions hold:

- The encoding is a single-byte encoding.
- ASCII is a subset of the encoding.
- The implementation of mbtowc(3) in the C library does not
  require re-initialization after encoding errors.
- The implementation of wchar_t in the C library uses the
  same numerical values as ASCII.

Otherwise, it silently produces wrong results.

The simplest way to fix this is to just use the same code as for
UTF-8 (right above).  Of course, that causes functional changes
but that shouldn't matter since current behaviour is undefined.

The patch below provides the following improvements:

- It works for all stateless single-byte encodings, no matter
  whether they are somehow related to ASCII, no matter how
  mb[r]towc(3) are internally implemented, and no matter how
  wchar_t is internally represented.
- Instead of producing unpredictable and definitely wrong
  results for non-UTF-8 multibyte characters, it behaves in
  a well-defined way: It aborts input processing, sets errno,
  and returns failure.
  Note that short of providing full support for arbitrary locales,
  it is impossible to do better.  We cannot know whether a given
  unsupported locale is state-dependent, and for a state-dependent
  locale, it makes no sense to retry parsing after an encoding
  error, so the best we can do is abort processing for *any*
  unsupported multi-byte character.
- Note that single-byte characters in arbitrary state-independent
  locales still work, even in locales that may potentially also
  contain multibyte characters, as long as those don't occur in
  input.  I'm not sure whether any such locales exist in practice...

Tested with UTF-8 and C/POSIX on OpenBSD.  Also tested that in the
C/POSIX locale, non-ASCII bytes get through unmangled.  You may
wish to test with ISO-LATIN on NetBSD if NetBSD supports that.

----
Also use a constant for meta to avoid warnings.

(christos)

Fix the bitmask of MVXPE_PMACC0_FRAMESIZELIMIT. It did no harm.

(hikaru)

regen.

(mrg)

add several NVIDIA geforce and NVS cards.

(mrg)

Fix typo in comment.

(msaitoh)

support screen blanking and hw cursor

(macallan)

- Add some more Char casts
- reduce ifdefs by providing empty defs for nls functions (Ingo Schwarze)

(christos)

remove unused wrapper (Ingo Schwarze)

(christos)

Remove utf8_islead() mbrtowc() handles this just fine (Ingo Schwarze)

(christos)

texinfo-6.1 and grep-2.23 out.

(wiz)

Pass a sensible device state pointer to nouveau suspend/resume ops.

Gives nouveau half a chance of suspending and resuming -- not that it
works on my test laptop yet, but it's a start.

(riastradh)

Set nouveau to a more reasonable debug level.

Match the compile-time and default run-time settings in Linux.

(riastradh)

Try xf86-video-nouveau for nvidia devices by default.

Fall back to nv (and then vesa and then wsfb) if it doesn't work.

(riastradh)

switch to 32bit colour in WSDISPLAYIO_MODE_MAPPED, now X with wsfb will work
Not optimal though - for some reason the framebuffer's endianness in 32bit
colour is wrong and I have no idea (yet) how to change that, so many apps
using xrender will crash.

(macallan)

dtor gets called if ctor fails.  Make it work in that case.

(riastradh)

Don't share struct work, instead have one per softc

Pointed out by riastradh@

(ozaki-r)

use pcpp front end rather than libexec/cpp directly, since commandline
options may not be handled by the latter

(plunky)

add pcpp binary, now in pcc-20160208. Also, p++ in debug set

(plunky)

update build machinery for pcc-20160208

(plunky)

fix merge conflicts from import of pcc-20160208

(plunky)

Fix build

(ozaki-r)

Document file format better. From Travis Paul and Matthew Bauer.

Bump date.
Addresses PR 50787.

(wiz)

Add more #warning message 'Please use mvxpe instead of this'.

(kiyohara)

Clear ATAC_CAP_DATA32, if dev is atabus@wdc@pcmcia@aupcmcia.  It is 32-bit transfers are not supported.

(kiyohara)

Actually, descsz should not contain the padding. The note still needs to
be padded out.

(skrll)