indent: clean up code for appending to buffers

Use *e++ for appending and e[-1] for testing the previously appended
character, like in other places in the code.

No functional change.

(rillig)

indent: merge duplicate code for reading from input buffer

No functional change.

(rillig)

Drop unicast packets that are not for us
when lagg(4) is not in promisc

(yamaguchi)

Replace the list for vlan interfaces with the counter

The list had been used in vlan_ifdetach(), but it is no longer in
use as a linked list by introducing ether_ifdetach hook.

(yamaguchi)

Update agcre to version 20211001

20211001
========

+ move to a single source file, which can just be included as a single file
(the library method can still be used for this as well - sometimes, it's just easier to
drop in a single file to other projects)

+ add AGCRE_EMBED_CTAGS as a primary regular expresion type (same as basic regexps,
but where a '*' occurs as a non-column 0 element of the pattern, it is assumed
to be a standard character, as produced by ctags(1) output; basic regular expressions
only accept '*' as a standard character as the first element of the expression).
Yay.

+ make the version string in the program a real string, so it can be found using strings(1)

+ add an agcre_rev_regexec(3) function, which searches the text in an efficient, reverse
manner, a la strrchr(3) or memrchr(3) fashion.

+ add agcre_regnsub(3) and agcre_regasub(3) functions

+ bump the shared lib version to 1.0

+ cut version 20211001

(agc)

remove duplicate the article in comments.

(andvar)

follow fcambus steps by fixing typos for "unknown" in few more error messages.

(andvar)

s/userpsace/userspace/

(andvar)

riscv: fix build with r/o src tree

add 'compile' to the SUBDIR list so that 'make obj' creates the
kernel compile base objdir so <bsd.kernobj.mk> sets KERNOBJDIR
correctly.

(mrg)

Ticket #1354

(martin)

Pull up following revision(s) (requested by tsutsui in ticket #1354):

external/mit/xf86-input-keyboard/dist/src/kbd.c: revision 1.8

Handle DEVICE_ABORT to restore the original keyboard state after crash.

No particular comment in PR xsrc/56415.

(martin)

Tickets #1351 - #1353

(martin)

Pull up following revision(s) (requested by thorpej in ticket #1353):

sys/miscfs/fifofs/fifo_vnops.c: revision 1.90
tests/kernel/kqueue/write/t_fifo.c: revision 1.5

- Add a new EVFILT_WRITE test case for FIFOs that correctly validates
the writability thresholds.
- Fix a bug in fifo_kqfilter() exposed by the new test case; in the
EVFILT_WRITE case, we were attaching the wrong end of the socket
pair to the knote!
- In filt_fiforead(), use ">= so->so_rcv.sb_lowat" rather than "> 0"
for consistency with fifo_poll().  NFC.

(martin)

Pull up following revision(s) (requested by thorpej in ticket #1352):

tests/kernel/kqueue/read/t_fifo.c: revision 1.5

New EVFILT_READ test case for FIFOs; validates readability threshold and
EV_EOF behavior.

(martin)

remove duplicate the article in documentation.

(andvar)

Pull up following revision(s) (requested by thorpej in ticket #1351):

sys/miscfs/fifofs/fifo_vnops.c: revision 1.88
sys/kern/uipc_syscalls.c: revision 1.201
tests/lib/libc/sys/t_poll.c: revision 1.6
tests/lib/libc/sys/t_poll.c: revision 1.7
tests/lib/libc/sys/t_poll.c: revision 1.8

- Strenghen the poll(2) fifo_inout test to ensure that once the reader
has read enough that exactly PIPE_BUF space is available that the FIFO
becomes writable again.
- When creating a FIFO, ensure that the receive low water mark is 1
(a FIFO must be readable when at least 1 byte is available); this
was already the case implicitly, but this makes it explicit.
- Similarly, set the send low water mark to PIPE_BUF to ensure that
the pipe is writable when at least PIPE_BUF bytes of space are available
in the send buffer.  Without this change, the strengthened test case
above does not pass (the default send low water mark is larger than
PIPE_BUF; see soreserve()).
- Make the same low water mark changes to the PIPE_SOCKETPAIR case.

In the fifo_hup1 test, also ensure that POLLHUP is de-asserted when a
new writer appears.

Add a fifo_inout test case that validates the expected POLLIN / POLLOUT
behavior for FIFOs:
- A FIFO is readable so long as at least 1 byte is available.
- A FIFO is writable so long as at least PIPE_BUF (obtained with _PC_PIPE_BUF)
  space is avaiable.
This will be cloned for a forthcoming kevent test case.

(martin)

rasops_reconfig: Do not abort even if font stride is not supported by
rasops_make_box_chars_*() functions; Stop using ri_optfont instead.

XXX
Add rasops_make_box_chars_24().

(rin)

Add a KASSERT

(skrll)

Sorry.. missed the debug module build in the previous commit.

(brad)

Don't include bsd.own.mk here - none of the other etc.<foo>/Makefile.inc
files do.

(skrll)

disable GIC_SPLFUNCS until remaining issues are sorted out

(jmcneill)

indent: fix lint warning about signed '>>'

Lint couldn't infer that indent's list of type names will practically
never contain more that 2 billion entries and that the result of '>>'
would be the same in all cases.

(rillig)

indent: clean up load_profile

No functional change.

(rillig)

indent: reduce duplicate code in load_profiles

No functional change.

(rillig)

Fix typo when erroring out on unknown ELF machine type.

(fcambus)

indent: rename functions

There was no good reason for using the different verbs 'scan' and 'set'
for two functions that essentially do the same.

No functional change.

(rillig)

indent: fix content of profile_name

Previously, profile_name included the leading "-P", which was confusing.

(rillig)

New sentence, new line. Fix Dd. Use more macros.

(wiz)

A driver for the Sensirion SHT40/SHT41/SHT45 temperature and humidity
sensor.  An example of this chip is:

https://www.adafruit.com/product/4885

This is a lower cost chip that provides higher then usual precision
according to the data sheet.  This driver supports all of the published
functions that the chip has.

(brad)

avoid set-but-unused-variable warnings.

(mrg)

Trailing whitespace

(skrll)

regen.

(mrg)

add intel 660p ssd, and expand the 760p description.

(mrg)

default RK3399 PCIe to 1.x.

sometime in mid-2019, rockchip disabled PCIe 2.x referrencing a
non-public errata in linux and the dts, both of which were changed
to default to only PCIe 1.x (in code, and in DT.)  see:

  https://lore.kernel.org/all/1481881357-1793-1-git-send-email-shawn.lin@rock-chips.com/

we haven't observed any specific problems, but we've also tested
far fewer cards than linux.  there are other issues such as what
happens for a PCI abort depending on which CPU triggers it we do
not currently handle either.

tested on rockpro64 with nvme & ahcisata, and pinebook pro nvme.

still allows DT to set back to '2' if the user so desires.

(mrg)

- Add a new EVFILT_WRITE test case for FIFOs that correctly validates
  the writability thresholds.
- Fix a bug in fifo_kqfilter() exposed by the new test case; in the
  EVFILT_WRITE case, we were attaching the wrong end of the socket
  pair to the knote!
- In filt_fiforead(), use ">= so->so_rcv.sb_lowat" rather than "> 0"
  for consistency with fifo_poll().  NFC.

(thorpej)

New EVFILT_READ test case for FIFOs; validates readability threshold and
EV_EOF behavior.

(thorpej)

...and correct my terrible spelling.

(thorpej)

- Strenghen the poll(2) fifo_inout test to ensure that once the reader
  has read enough that exactly PIPE_BUF space is available that the FIFO
  becomes writable again.
- When creating a FIFO, ensure that the receive low water mark is 1
  (a FIFO must be readable when at least 1 byte is available); this
  was already the case implicitly, but this makes it explicit.
- Similarly, set the send low water mark to PIPE_BUF to ensure that
  the pipe is writable when at least PIPE_BUF bytes of space are available
  in the send buffer.  Without this change, the strengthened test case
  above does not pass (the default send low water mark is larger than
  PIPE_BUF; see soreserve()).
- Make the same low water mark changes to the PIPE_SOCKETPAIR case.

(thorpej)

Add a fifo_inout test case that validates the expected POLLIN / POLLOUT
behavior for FIFOs:
- A FIFO is readable so long as at least 1 byte is available.
- A FIFO is writable so long as at least PIPE_BUF (obtained with _PC_PIPE_BUF)
  space is avaiable.

This will be cloned for a forthcoming kevent test case.

(thorpej)

In the fifo_hup1 test, also ensure that POLLHUP is de-asserted when a
new writer appears.

(thorpej)

Pass the pmap in tlb_set_asid for the benefit of aarch64.

(skrll)

Use KERNEL_PID instead of 0. NFC.

(skrll)

Tickets #1349 and #1350

(martin)

Pull up following revision(s) (requested by thorpej in ticket #1350):

sys/kern/uipc_socket2.c: revision 1.140
tests/lib/libc/sys/t_poll.c: revision 1.5
sys/miscfs/fifofs/fifo_vnops.c: revision 1.87

- fifo_poll(): If the last writer has disappeared, detect this and return
  POLLHUP, per POSIX.
- fifo_close(): Use the new fifo_socantrcvmore(), which is like the
  garden-variety socantrcvmore(), except it specifies POLL_HUP rather
  than POLL_IN (so the correct code for SIGIO is sent).
- sowakeup(): Allow POLL_HUP as a code (notifies poll'ers with POLLHUP).
- Add test cases for correct POLLHUP behavior with FIFOs.

Fixes PR kern/56429.

(martin)

Apply patch, requested by manu in ticket #1349:

external/gpl3/gcc/dist/libgomp/oacc-init.c (apply patch)

Fix a pthread_key leak in libgomp. Cherry picked from -current where
it was solved by the GCC 8.4 import.

(martin)

We are now (partial) at tzdata2021c.  This involves no changes other than
here (this), as we had not adopted the changes to tzdata in 2021b that were
reverted in 2021c, and the other changes are all to tzcode (and not
very important to us, either 2021b or 2021c tzcode changes).

(kre)

Fix a deadlock where one thread writes to a pipe, has more data
and no space in the pipe and waits on "pipe_wcv" while the reader
is closing the pipe and waits on "pipe_draincv".

Swap the test for "PIPE_EOF" and the "cv_wait_sig()" in "pipe_write()".

PR bin/56422 "zgrep -l sometimes hangs"

(hannken)

Handle DEVICE_ABORT to restore the original keyboard state after crash.

No particular comment in PR xsrc/56415.

(tsutsui)

- fifo_poll(): If the last writer has disappeared, detect this and return
  POLLHUP, per POSIX.
- fifo_close(): Use the new fifo_socantrcvmore(), which is like the
  garden-variety socantrcvmore(), except it specifies POLL_HUP rather
  than POLL_IN (so the correct code for SIGIO is sent).
- sowakeup(): Allow POLL_HUP as a code (notifies poll'ers with POLLHUP).
- Add test cases for correct POLLHUP behavior with FIFOs.

Fixes PR kern/56429.

(thorpej)

tzdata partial update to 2021b

(kre)

Cherry pick the important changes from tzdata2031b:

Samoa (Pacific/Apia) has cancelled summer time this year
(change effective last Sunday).
Jordan (Asia/Amman) has moved the start of summer time from
the last Thu/Fri in March to the last Thu/Fri in Feb
to begin next year (2022).

Some more of the non-controversial (and less important, mostly some
corrections to historical data) changes may follow soon.

(kre)

use quirk bit for devices that need sc_alwayson instead of adding yet another
if (uha->uiaa->uiaa_vendor == USB_VENDOR_..., remove entries now covered
by UQ_ALWAYS_ON

(macallan)

add a quirk for devices that need ums::sc_alwayson

(macallan)

regen

(macallan)

add PixArt Raspberry Pi Mouse

(macallan)

remove parameter names from decls.

(christos)

t_simplehook requires RUMP

(martin)

Make this compilable as tool with MKCROSSGDB=yes

(martin)

PR/56424: Clay Mayers: recvfrom() is not a cancelation point as documented
in pthread_setcanceltype.3

(christos)

riscv: Include kernel and bootloader in releasedir.

This change adds a kern-GENERIC set and distributes an INSTALL kernel
along with bootriscv64.efi as part of a standard release build.

(jmcneill)

xf86-input-ws: fix building on xorg-server.old (vax, etc)

not entirely sure that the ABI version check is spot on, but the
relevant changes were made in 2011 so hopefully we only have to
support one old xinput API here.

(nia)

G/C unused over a decade.

(rin)

Comment out hpcarm/spl.S, which was superseded by arm32/spl.S.

No functional changes; We had two source files for spl.o, but correct
file was chosen by config(1).

(rin)

tests/indent: add more tests with ')(', found bug in '//'

(rillig)

make the test program  print the current version when invoked with -V

(agc)

Update the date for this man page

Also add the explanation for the string parameters to gap_exec_string()

(agc)

Update buffer gap routines to the latest API and functionality - 20210930.

The API has been consolidated around 4 functions:

1. gap_new() - returns a new (opaque) buffer gap structure
2. gap_dispose() - dispose of a previously allocated buffer gap structure
3. gap_exec() - perform operations on the buffer gap which expect an
  integer return value
4. gap_exec_string() - perform operations on the buffer gap which return a byte
  array (typically return buffer contents, or the buffer name)

All of the "verbs" are described in the libgap.3 manual page, found in the
dist directory. The main.c test program, again found in the dist directory,
gives 7 worked examples of how to use the functionality to achieve desired ends,
and there are 7 test input files, along with expected results, which are
provided as a regression test suite - accessible using "make t" from the top
level directory.

The buffer gap routines use a copy of the embedded agcre regexp library to
provide bounded (non-exponential time) results for regexp searching forwards
and backwards - for more information, see the agcre directory at the same level
as this code, or the source code in dist/agcre-embed.[ch].

Also move to a more usual reachover infrastructure

(agc)

indent: remove space between ')' and '(' in declarations

(rillig)

indent: untangle want_blank_before_lparen

No functional change.

(rillig)

indent: extract want_blank_before_lparen

No functional change.

(rillig)

Ensure TCR_EPD0 is set on entry to pmap_activate and ensure it is set as
early as possible for APs.

(skrll)

indent: add space between ',' and '[' in C99 initializations

(rillig)

efiboot: Build and install bootriscv64.efi for riscv64 builds.

(jmcneill)

gnu-efi: Import version 3.0.14.

(jmcneill)

gnu-efi: updated to 3.0.14

(jmcneill)

gnu-efi: riscv64 build fixes for NetBSD

(jmcneill)

merge conflicts

(jmcneill)

Make tlb_asid_t unsigned int as pmap_tlb.c expects tlb_asid_t to be able to
hold ASID_MAX + 1.

(skrll)

Print CPUID_PBE (Pending Break Enable) with "PBE".

(msaitoh)

kqueue(2) file descriptors are not inherited across fork(2). A
process that that calls getaddrinfo(3) will end up cacheing the
kqueue(2) file descriptor in its res_state structure. If that
process fork(2)s and calls getaddrinfo(3) again might end up closing
that cached file descriptor which can end up pointing to a different
file object than the kqueue(2) original one. To fix this, associate
the kqueue(2) file descriptor with the process id that created it,
and don't close(2) it if it is being closed from a different process.
An alternative fix would be to attach the resolver to a fork(2) hook
to cleanup the res_state, but handling it internally in the resolver
is less intrusive. This was discovered by Dima Veselov when using
the FreeRADIUS package.

(christos)

Drive-by KNF

(skrll)

More catching up with spl changes.

(skrll)

Some platforms need sys/cpu.h for curlwp so #include it

(skrll)

KNF

(skrll)

Bump the version for struct ifnet and ethercom changes

Welcome to 9.99.90

(yamaguchi)

lagg: Register lagg_ifdetach to ether_ifdetach hook

(yamaguchi)

Make a link-layer address of lagg(4) configurable by ifconfig(8)

lagg(4) uses a configured link-layer (MAC) address instead
of a random MAC address generated on creating.
The configured MAC address is copied to all child interface
and used for a system id of LACP.

(yamaguchi)

Fix to acquire LAGG_LOCK without psref
to remove possibility of deadlock

the deadlock maybe happened between lagg_ifdetach()
and lagg_delport()

1. lagg_ifdetach calls psref_target_acquire()
2. lagg_delport calls LAGG_LOCK()
3. lagg_ifdetach calls LAGG_LOCK()
  - wait for lagg_delport
4. lagg_delport calls psref_target_destroy()
  - wait for lagg_ifdetach

(yamaguchi)

vlan: Register vlan_ifdetach to ether_ifdetach hook

(yamaguchi)

Remove vlan_ifdetach() from ixgbe_detach()

The removed function is called in ether_ifdetach().
And ether_ifdetach is changed to be called before a device
is stopped, because vlan_ifdetach called from ether_ifdetach
configures VLAN settings.

(yamaguchi)

Remove vlan_ifdetach() from ixv_detach()

The removed function is called in ether_ifdetach().

(yamaguchi)

bridge: Register bridge_ifdetach to ether_ifdetach hook

(yamaguchi)

Provide a hook point called when ether_ifdetach is called

(yamaguchi)

net: obsolete ifnet::if_link_state_chenged
that was used for updating link-state of vlan I/F

The obsoleted function is replaced with
ifnet::if_linkstate_hooks

(yamaguchi)

vlan: Register the callback to update link-state of vlan I/F
to link-state change hook

The callback is registered in every vlan I/F even if the parent
interface is the same. Therefore it is not needed to search the
vlan I/F by the parent interface unlike the previous callback.

(yamaguchi)

carp: Register carp_carpdev_state to link-state change hook

(yamaguchi)

lagg: Register lagg_linkstate_changed to link-state change hook

(yamaguchi)

bridge: Register bridge_calc_link_state to link-state change hook

(yamaguchi)

Provide a hook point called at change of link state

(yamaguchi)

Replace ifnet::if_agriprivate with ifnet::if_lagg

agr(4) and lagg(4) can not be used on the same interface so that
if_agrprivate and if_lagg are not used at the same time.
For resolve this wasteful, if_lagg is used in not only lagg(4)
but also agr(4).

After this modification, if_lagg has 3 states:
1. if_lagg == NULL
  - Both agr(4) and lagg(4) are not running on the interface
2. if_lagg != NULL && ifp->if_type != IFT_IEEE8023ADLAG
  - agr(4) is running on the I/F
3. if_lagg != NULL && ifp->if_type == IFT_IEEE8023ADLAG
  - lagg(4) is running on the I/F

(yamaguchi)

Added tests for the linear hook APIs

(yamaguchi)

Introduce a generic linear hook list

(yamaguchi)

Make the info returned by kqueue_stat() a little less barren.

(thorpej)

In knote(), don't call kn->kn_fop->f_event() directly; use filter_event()
to get the correct KERNEL_LOCK handling for the filter attached to that
specific note.

(thorpej)

Confirm to the NOTE_SUBMIT protocol.  With this, btn_read_filtops is MPSAFE.

(thorpej)

The kq filterops that interact with sockets are MPSAFE.

(thorpej)

ttyread_filtops, ttywrite_filtops, ptcread_filtops, and ptcwrite_filtops
are MPSAFE.

(thorpej)

- Change selremove_knote() from returning void to bool, and return
  true if the last knote was removed and there are no more knotes
  on the selinfo.
- Use this new return value in filt_sordetach(), filt_sowdetach(),
  filt_fifordetach(), and filt_fifowdetach() to know when to clear
  SB_KOTE without having to know select/kqueue implementation details.

(thorpej)

Make sure the robust futex head is zeroed out, since this LWP
will live on with a different program image.  (Thanks ryo@ for
pointing out my mistake.)

(thorpej)

futex_release_all_lwp(): No need to pass the "tid" argument separately; that
is a vestige of an older version of the code.  Also, move a KASSERT() that
both futex_release_all_lwp() call sites had inside of futex_release_all_lwp()
itself.

(thorpej)

In the exec path, multi-LWP programs dispose of their robust futexes
by calling exit_lwps(), except for the last LWP.  So, dispose of that
LWP's robust futexes right before calling lwp_ctl_exit().

Fixes a "WARNING: ... : unmapped robust futex list head" message when
running bash under Linux emulation on aarch64.

Root caused and patch proposed by ryo@.  I have tweaked it slightly,
just to add a comment and a KASSERT().

(thorpej)

efiboot: Add efigop.c

(jmcneill)

efiboot: Add support for changing the video mode.

Add a new "gop" command that can query the list of available video modes.
With a mode number as argument (eg. "gop 16"), the new display mode will
be selected.

The "version" command prints the current display mode.

(jmcneill)

Switch to xf86-input-ws for mouse by default, to allow more advanced
input devices to Just Work.

(nia)

tests/indent: demonstrate missing space before designated initializer

Seen in usr.bin/indent/lexi.c, num_lex_row.

(rillig)

Remove duplicate tm_isdst.  Fixes PR misc/56419 for HEAD.

(kim)

wsconsctl: allow the new "precision scrolling" variables to be configured

(nia)

xf86-input-ws: implement support for WSCONS_EVENT_(H|V)SCROLL.

adapted from the OpenBSD code.

(nia)

synaptics: use the new WSCONS_EVENT_(H|V)SCROLL event types
to implement two-finger scrolling. remove non-functional code for
detecting reported finger width.

(nia)

wsmouse: add support for "precision scrolling" events and (GET|SET)PARAMS

WSCONS_EVENT_HSCROLL and WSCONS_EVENT_VSCROLL are two new wscons event
types that allow scrolling with a higher precision ("smoothness") than an
emulated scroll wheel, and are useful for touch input drivers.

WSMOUSEIO_GETPARAMS and WSMOUSEIO_SETPARAMS are two new ioctls that allow
the speed and direction of precision scrolling to be configured.

both features were originally implemented in OpenBSD.

(nia)

indent: let indent format the comments after previous refactoring

Before this refactoring, I had skipped this section of the code from
formatting since the 'default:' branch was enclosed in a block of its
own, and that block would have been indented one more level to the
right. Extracting that code into a separate function got rid of the
extra braces.

No functional change.

(rillig)

indent: split search_brace into smaller functions

No functional change.

(rillig)

tests/indent: document that label indentation is not configurable

For at least the past 27 years, labels have been indented 2 levels to
the left of the corresponding statement. See label_offset and
compute_label_indent (or compute_label_target in FreeBSD).

(rillig)

indent: use binary instead of linear search when adding types

No functional change.

(rillig)

remove unused code.
The syscall for 32bit uses aarch32_syscall.c, so there is no need to make syscall.c support it.

(ryo)

linux syscall should not break x1 register

(ryo)

In order to prevent uninitialized values from being reflected in the registers after syscall, rval[] must be initialized.

(ryo)

indent: extract is_typename from lexi

No functional change.

(rillig)

Improve variable sized TLB's output.

- Fix a bug that STLB is printed as DTLB.
- If a TLB is variable sized, print the max size instead of error message.
  XXX This is temporary solution.

(msaitoh)

new OpenSSH

(christos)

Merge conflicts between OpenSSH 8.7 and 8.8

(christos)

indent: rename rwcode to keyword_kind, various cleanup

No idea what the 'rw' in 'rwcode' meant, it had been imported that way
28 years ago. Since rwcode specifies the kind of a keyword, the prefix
'kw_' makes sense.

No functional change.

(rillig)

Add Load Only TLB and Store Only TLB.

(msaitoh)

Fix a bug that some TLB related lines were not printed.

(msaitoh)

Add ':' for readability.

(msaitoh)

openssh 8.8 is out

(wiz)

OpenSSH: Restore ChallengeResponseAuthentication as an alias

OpenSSH 8.7 has deprecated ChallengeResponseAuthentication, but not removed
it. It is now an alias for KbdInteractiveAuthentication (as are the prior
aliases of ChallengeResponseAuthentication).

I think this chunk was accidentally dropped in the OpenSSH 8.7 merge.

(kim)

No longer need to call arm_pmu_init on all CPUs.

(jmcneill)

Use unsigned to avoid undefined behavior in pppasyncstart().

Reported-by: syzbot+7c8c7977e2756ac13f0a@syzkaller.appspotmail.com

(msaitoh)

Tweak filt_piperead() and filt_pipewrite() so that:
- There is only a single return from the function (and thus a single
  place where the pipe lock must be released).
- kn->kn_data is referenced only inside the lock perimeter.

(thorpej)

Consistently reference kn->kn_data only within the lock perimeter in
the filtops f_event() callback.

(thorpej)

In kqueue_kqfilter(), return EINVAL instead of 1 if something other than
EVFILT_READ was requested.

(thorpej)

- Rename kqueue_misc_lock -> kqueue_timer_lock, since EVFILT_TIMER is
  now its only user.  Also initialize it as IPL_SOFTCLOCK; there is no
  practical difference in how it operates (it is still an adaptive lock),
  but this serves as a visual reminder that we are interlocking against
  a callout.
- Add some comments that describe why we don't need to hold kqueue_timer_lock
  when detaching an EVFILT_TIMER due to guarantees made by callout_halt().
- Mark timer_filtops as MPSAFE.

(thorpej)

indent: fix documentation of opt.case_indent

See io.c, compute_label_indent.

(rillig)

indent: fix definition of -cli in manual page

See io.c, compute_label_indent.

(rillig)

Fix the locking around EVFILT_FS.  Previously, the code would walk the
fs_klist and take the kqueue_misc_lock inside the event callback.
However, that list can be modified by the attach and detach callbacks,
which could result in the walker stepping right off a cliff.

Instead, we give the fs_klist it's own lock, and hold it while we
call knote(), using the NOTE_SUBMIT protocol.  Also, fs_filtops
into vfs_syscalls.c so all of the locking logic is contained in one
file (there is precedence with sig_filtops).  fs_filtops is now marked
MPSAFE.

(thorpej)

indent: unexport global variables

The variable match_state was write-only and was thus removed.

No functional change.

(rillig)

indent: unexport keyword table, clean up

No functional change.

(rillig)