Apply patch, requested by mrg in ticket #1752:

xsrc-9/external/mit/xorg-server/dist/xkb/xkb.c apply patch

Cherry-pick upstream fixes for
* CVE-2022-2319/ZDI-CAN-16062: X.Org Server ProcXkbSetGeometry Out-Of-Bounds
  Access
* CVE-2022-2320/ZDI-CAN-16070: X.Org Server ProcXkbSetDeviceInfo Out-Of-Bounds
  Access

(martin)

Ticket #1478

(martin)

Apply patch, requested by mrg in ticket #1478:

xsrc-9/external/mit/xorg-server/dist/xkb/xkb.c apply patch

Cherry-pick upstream fixes for
* CVE-2022-2319/ZDI-CAN-16062: X.Org Server ProcXkbSetGeometry Out-Of-Bounds
  Access
* CVE-2022-2320/ZDI-CAN-16070: X.Org Server ProcXkbSetDeviceInfo Out-Of-Bounds
  Access

(martin)

properly mark glint_drv has obsolete.

(mrg)

fix some builds (no longer built drivers.)

(mrg)

update for xorg-server 21.1.4.

(mrg)

update for xorg-server 21.1.4.

(mrg)

merge xorg-server 21.1.4.

(mrg)

include the return value of rv770_restrict_performance_levels_before_switch()
in error messages.

(mrg)

convert "ksyms: checking <symbol>" into KSYMS_DEBUG over normal DEBUG.

(mrg)

add pool usage to the output, sharing the line with swap data.

looks like, eg:

Swap: 64G Total, 135M Used, 64G Free / Pools: 16G Used

future work:  figure out the utilisation percentage like vmstat -m

(mrg)

make this actually build again on arm/mips.

(mrg)

add a note about xorg-server possibly breaking builds.

(mrg)

properly mark an obsolete file obsolete.

(mrg)

use "1ULL << 32" for a 64 bit member, not 1UL.

(mrg)

adjust the list of drivers per platform / xorg-server version for 21.1.3.

while here, remove x86 support for xorg-server 1.10.  the set lists were
out of date for ages, and have also been cleaned up.

the tga, trident, and s3 drivers are disabled for xorg-server 21.1.3, on
top of the various previously disabled for newer server drivers.

(mrg)

update for xorg-server 21.1.3.

(mrg)

updates for xorg-server 21.1.3.

(mrg)

more xf86RamDac.h vs xf86Cursor.h.

(mrg)

various updates to build with xorg-server 21.1.3.

- the s3, tga, and trident drivers use old functionality and are disabled
- dixmods/fb is removed, main fb has all these now (we retain our revert
  of an upstream change here.)
- adjust source lists as necessary
- remove vbe module, now part of int10
- use -lxcvt
- misc clean up
- ramdac loses IBM, BT and TI support (this is one of the problems for
  the disabled drivers, but also some structures in RamDac.h are gone
  entirely, not copied anywhere.)
- old apm code is gone, all bsd uses pm_noop. now

this probably require a clean build in at least xorg-server and the
drivers, and perhaps the DESTDIR as well.  will update src/UPDATING with
details when they are fully known.

(mrg)

various updates to build against xorg-server 21.1.3.

- avoid a const issue with an expanded __UNCONST().
- cope with renamed member names
- deal with "xf86RamDac.h" going away (now use "xf86Cursor.h")

(mrg)

update these for xorg-server 21.1.3.

(mrg)

fix merge botch.

(mrg)

merge xorg-serer 21.1.3.

(mrg)

also move cvt manual parts into the xbase set.

(mrg)

fix sets after making cvt(1) for everyone

remove duplicate entries for Xorg and gtf

(mrg)

Misc KNF and cosmetics.

(tsutsui)

Fix pasto.

(tsutsui)

build cvt here.

(mrg)

default to apple layout for both adb and usb keyboards

(macallan)

doin't build cvt(1) here any more.

(mrg)

build and install libxcvt, and it's cvt(1).

(mrg)

enable thread safety constructor.

(mrg)

add libxvct include subdir.

(mrg)

add libxcvt.

(mrg)

build framework for libxcvt (needed for xorg-server 21.x).

doesn't work quite yet.  lib/libxcvt can't find sources for
some odd reason.

(mrg)

merge libX11 1.8.1.

(mrg)

merge xkbutils 1.0.5, xmag 1.0.7, xmessage 1.0.6, and xrefresh 1.0.7.

(mrg)

Check for right sysctl node to make live image work OOTB on OpenStack:

  machdep.dmi.system-product = OpenStack Compute

(hgutch)

Move klist_{init,fini,insert,remove}() into kern_event.c.  NFC.

(thorpej)

Funnel knote alloc/free into a single pair of functions.  NFCI.

(thorpej)

x86: Move lfence into _bus_dmamap_sync and comment it.

No functional change intended.  This just keeps the bus_dma_* and
_bus_dma_* functions organized more consistently.

(riastradh)

make.1: fix markup for variable modifiers, add italic correction

(rillig)

bump the krb_drv.so version (oops forgot to commit this yesterday.)

should fix most of the 46 builds broken currently..

(mrg)

make.1: fix markup for options, they are not arguments

(rillig)

make.1: fix ugly line break in PostScript output

(rillig)

make.1: fix markup

Not all square brackets are options, some are wildcards.

(rillig)

make.1: replace redundant .Ql with .Sq

(rillig)

make: describe variable assignment and evaluation more precisely

Variables in make are not "much like variables in the shell", they only
look that way.  Their behavior is more similar to C preprocessor macros.

The spaces around a variable value are removed when parsing the variable
assignment, not after evaluating it.

Before describing the variable assignment operators, introduce the
general syntax of a variable assignment.

Add a separate subsection "Expansion of variables", since assignment and
expansion are unrelated.  Illustrate how variable expansions look
instead of describing it in many words.

Describe how variables are expanded in conditionals, since that is
different from everywhere else.

(rillig)

don't build newly obsolete xf86vidmode.h.

(mrg)

update for libxcb 1.15.

(mrg)

updates for xcb-proto 1.15.2.

(mrg)

updates for generated files.

(mrg)

updates for libxcb 1.15.

(try to make xcbgen work here still, but still doesn't quite work.)

(mrg)

build the combined cirrus driver.

(mrg)

merge libxcb 1.15.

(mrg)

Remove unneeded bus_space_barrier() calls.

(thorpej)

Remove unneeded bus_space_barrier() calls.

(thorpej)

Remove uneeded bus_space_barrier() calls.

(thorpej)

Remove unneeded bus_space_barrier() calls.

(thorpej)

updates for the merged cirrus driver (apline/laguna sub-drivers are gone.)

(mrg)

more updates for xfs 1.2.1, xdpyinfo 1.3.3, and libXcursor 1.2.1.

(mrg)

merge libXvMC 1.0.13.

(mrg)

updates for xfs 1.2.1, and xdpyinfo 1.3.3

(mrg)

xf86-input-keyboard 2.0.0, xf86-video-cirrus 1.6.0, xf86-video-mach64 6.9.7,
and xf86-video-mga 2.0.1.

(mrg)

merge xsm 1.0.5.

(mrg)

updates for libdrm 2.4.112.

(mrg)

updates for libdrm 2.4.112.

(mrg)

update for xterm 372.

(mrg)

merge libdrm 2.4.112, and xterm-372.

(mrg)

p2k(3): drop reference to rump_smbfs(8).

(andvar)

i915: Defer destroying waitqueue until after callback is removed.

Candidate fix for PR kern/56561.

(riastradh)

Avoid C++-isms (stupid muscle memory) to fix the build.

(martin)

Ticket #1751

(martin)

Pull up the following revisions, requested by msaitoh in ticket #1751:

sys/dev/pci/if_wmreg.h 1.122-1.125 via patch
sys/dev/pci/if_wmvar.h 1.48
sys/dev/pci/if_wm.c 1.719-1.720,1.722-1.725,
1.727-1.740 via patch

- wm_tick: Add missing splx(s) when not WM_MPSAFE.
- Print DMA range info if the system is booting in the verbose mode.
- Micro optimization:
  - Call m_freem(m) only if m0 == NULL.
  - Call wm_xxeof() only when limit > 0.
  - Don't set the more flag when there is no packet to process.
- No functional changes:
  - Call txeof first, then rxeof for the consistency.
  - Remove duplicated break.
  - Remove stray semicolons from struct declaration.
  - Fix value return from void function.
  - Use macros.
  - Modify comment.
  - KNF.

(martin)

Ticket #1477

(martin)

Pull up the following revisions, requested by msaitoh in ticket #1477:

sys/dev/pci/if_wmreg.h 1.122-1.125
sys/dev/pci/if_wmvar.h 1.48
sys/dev/pci/if_wm.c 1.719-1.720,
1.722-1.725,
1.727-1.740 via patch

- wm_tick: Add missing splx(s) when not WM_MPSAFE.
- Print DMA range info if the system is booting in the verbose mode.
- Micro optimization:
  - Call m_freem(m) only if m0 == NULL.
  - Call wm_xxeof() only when limit > 0.
  - Don't set the more flag when there is no packet to process.
- No functional changes:
  - Call txeof first, then rxeof for the consistency.
  - Remove duplicated break.
  - Remove stray semicolons from struct declaration.
  - Fix value return from void function.
  - Use macros.
  - Modify comment.
  - KNF.

(martin)

PR 56922: disabel fast softints for powerpc for now.

(martin)

updates for twm 1.0.12.

(mrg)

merge mkfontscale 1.2.2, twm 1.0.12, xauth 1.1.2, and xdm 1.1.13.

(mrg)

link in xrandr.

(mrg)

merge xcmsdb 1.0.6, xconsole 1.0.8, xmodmap 1.0.11, and xwud 1.0.6.

(mrg)

merge xdpyinfo 1.3.3, xedit 1.2.3, xfontsel 1.1.0, xfs 1.2.1, xload 1.1.4,
and xlsfonts 1.0.7.

(mrg)

find new bitmaps location.

(mrg)

merge bitmap 1.1.0, iceauth 1.0.9, xclipboard 1.1.4, and xclock 1.1.1.

(mrg)

KNF two comments.

(skrll)

Grammar in a comment.

(skrll)

merge appres 1.0.6, editres 1.0.8, and listres 1.0.5.

(mrg)

KNF. No functional change.

(msaitoh)

Consistently use semicolon. No functional change.

(msaitoh)

readv(2), writev(2): Avoid arithmetic overflow in bounds check.

(riastradh)

physio(9): Avoid left shift of negative in alignment check.

Reported-by: syzbot+426b1db52f91ad2b994f@syzkaller.appspotmail.com
https://syzkaller.appspot.com/bug?id=f2456661792789db6d1c20e9d7e48db5c151d56d

(riastradh)

bsd.subdir.mk: replace obscure indirection with direct condition

Since 2008, when make generates the body of a .for loop, it does not
copy the items as raw strings but as expressions of the form ${:U...}.
This prevents syntactical changes from code injection in the body of the
.for loop.

Since then, using .for loop variables in conditions works.  Since the
same time, the variable assignment operator '+=' doesn't need to be
protected by a leading space anymore.  Even if the item from the .for
loop is 'libc++', the trailing '+' is not parsed as part of the variable
assignment operator, and it never had been parsed this way.  The space
probably has been cargo cult programming.  The only ambiguous case had
been in a variable assignment, when the variable name ended with a
variable from the .for loop, and the assignment operator was '='.  Only
in this case could the parser be tricked into parsing the variable
assignment as 'libc+ +=' instead of the intended 'libc++ ='.

(rillig)

bsd.subdir.mk: remove practically dead code

The condition '${dir:H} != ""' is almost always satisfied, since 2001 at
least.  The only input for the modifier ':H' that leads to an empty
string is "/".  It doesn't make sense though to specify the root
directory as a subdirectory.

While here, remove quotes around the left-hand side of comparisons in
conditions.  The quotes have become redundant in 2008, when the .for
loop stopped inserting raw text into the .for loop body.

If the special case for "/" had actually been intended, it should have
been written as the equivalent '.if ${dir} == "/"' instead of hiding
this feature behind an edge case behavior of the modifier ':H'.

Based on the comments, the condition should have been '${dir:M/*}', but
it hadn't been written that way.

(rillig)

tests/make: extend tests for modifier ':H'

(rillig)

Add a manual page for udl(4)

(nia)

usr.sbin/installboot: Add a command line option (-u) to set U-Boot search paths

Installboot can install U-Boot boot blocks directly into a system
image.  Normally, the U-Boot files are searched for in
/usr/pkg/share/u-boot, under the expectation that most people will
build them with pkgsrc.  However, it is also possible to set an
environment variable (INSTALLBOOT_UBOOT_PATHS) to a colon-separated
search path to accommodate other situations.  This commit adds a
command line option (-u) to set the search path; if present, it
overrides the environment and default.

(brook)

drop reference to smbfs, which was removed in 2020.

while here, s/file-system/file system/

(snj)

ptrace(PT_LWPSTATUS): Fix lid=0 case.

Make ptrace_read_lwpstatus always overwrite the whole struct
ptrace_lwpstatus so the caller doesn't have to make sure pl_lwpid is
initialized correctly -- ptrace_lwpstatus was missing that
reinitialization after ptrace_update_lwp, but it's less work for all
the callers if we just make the assertion an assignment instead.

(riastradh)

Fix order of hi / lo halves in the ofw_address_cells == 2 case.

(thorpej)

kern/sys_ptrace.c: Nix trailing whitespace.

(riastradh)

tty_43: Do unsigned arithmetic to avoid shift into sign bits.

Omit input validation -- it's not a great idea for compatibility with
historical kernels, since they ignored the bits instead of rejecting
them if set.  With unsigned arithmetic, we get the same semantics as
was previously assumed (discarding bits that get shifted into the
sign bit or off into oblivion) without the formal undefined
behaviour.

Reported-by: syzbot+e408764cdd8c0c0ff535@syzkaller.appspotmail.com
https://syzkaller.appspot.com/bug?id=8318f0039e68187cd8d27b2c520816fff3f7266a

(riastradh)

linux/hdmi: Fix size of product id.

(riastradh)

We may try to extract the "last mounted on" field from EFI partitions
too.

(martin)

linux/hdmi: Fix return value in unpack.

(riastradh)

Preparations to make the extended partitioning during normal installation
more usefull - pass an optional install_description to partman().

(martin)

cpuio.h: Use uint8_t, not bool.

In principle bool can only store two values, 0 and 1, but occupies
eight bits of storage (on all NetBSD platforms to my knowledge),
meaning there are 254 other values which might be trap
representations which would be undefined behaviour to use but which
userland can control the bits of.

To placate ubsan, just use uint8_t here instead of bool.  No bump or
anything because the ABI is the same: one-byte field with one-byte
alignment.

Reported-by: syzbot+3bc3d6837da151448e87@syzkaller.appspotmail.com

(riastradh)

pci(4): mention gtp(4) in the man page

(nia)

include bsd.own.mk to ensure that USE_XZ_SETS is set properly before using it.

this should fix PR#56919.

(mrg)

csh(1): fix punctuation at the end of an .Xr line

(uwe)

s/desciption/description/

(andvar)

linux/hdmi: Fix hdmi_infoframe_unpack to record the unpacked header.

(riastradh)

s/wokstations/workstations/ in rs6000 description.

(andvar)

linux/hdmi: Handle audio infoframes in hdmi_infoframe_pack.

Sort cases by type number.

(riastradh)

linux/hdmi: Convert failure branch to kassert for internal invariant.

(riastradh)

linux/hdmi: Use ssize_t for all hdmi_*_infoframe_pack functions.

(riastradh)

linux/hdmi: Zero entire union hdmi_infoframe on unpack.

Don't just zero the part that we're about to initialize -- i915 does
memcmp on the whole structure.

(riastradh)

linux/hdmi: Handle unpacking audio frames too.  Sort by number.

(riastradh)

linux/hdmi: Make some private functions static.

(riastradh)

linux/hdmi: Fix missing hdmi_avi_infoframe_init namespacing.

(riastradh)

linux/hdmi.h: Split out logic into .c file.

Shoulda done this ages ago.

(riastradh)

linux/hdmi.h: Implement hdmi_infoframe_log.

(riastradh)

linux/hdmi.h: Fix spd infoframe header version.

Possibly relevant to PR kern/56740.

(riastradh)

Fix GENFB_GLYPHCACHE with shadow fb.

(rin)

Add preliminary support for Aero.

- Aero supports 32bit descriptor.
- Even if the Aero card supports 32bit descriptor, 64bit descriptor
  access is required for MFI_CMD_INIT.
- Add quirk for Aero to retry register read.
- The PCI ID table for Aero is disabled. Even if it's enabled, any logical
  volumes are not found. Perhaps the reason is that our mfii driver is
  currently not supports more than 64 logical volumes and MR_DCMD_LD_GET_LIST
  returns incorrect value. Tested by Mark Daies' H750.

(msaitoh)

Use bus_space_write_8() if available.

(msaitoh)

Modify strange bus_space_barrier().

(msaitoh)

devsw(9): Forbid extant linked autoconf instances in devsw_detach.

(riastradh)

If /rescue is at the head of $PATH (which is how it starts out in /rescue/sh
unless changed by something - and has been since NetBSD 3) don't blindly
simply change it to a PATH that doesn't have /rescue in it at all.

This doesn't solve the "/rescue/tar execs /usr/bin/gzip" problem completely,
as if PATH is in the environment when /rescue/sh is started, that one will
override the shell's built in PATH, but this is better than nothing.

(kre)

mk.conf(5): we default to USE_SSP=yes these days.

(nia)

Remove redundant ``devsw_ok = false;''

Thanks mrg@

(pgoyette)

Prevent crashes as shown in PR 56918.

(mlelstv)

lint: fix edge cases in the query for redundant cast before assignment

Casting from and to _Bool is only allowed outside strict bool mode.
Outside strict bool mode, _Bool is an integer type, therefore return
early if any of the operands has type _Bool.  In strict bool mode, even
casting from _Bool to _Bool is not allowed, as it is not needed in
practice.

Handle _Complex types before real floating-point types.  Return early
for _Complex types, as these are floating-point types as well.

For pointer casts, not only flag casts to or from 'pointer to void', but
also casts between the same types.

In debug mode, when constructing the type '_Complex float', the type
name of '_Complex' occurs in the debug log.  Outside of debug mode,
printing this type name is an error since this type keyword only occurs
internally, when constructing a type.  At that point, it is not supposed
to occur in any user-visible message.

(rillig)

lint: fix query for redundant cast before assignment

Previously, 'i = (int)dbl' was marked as redundant, even though it
performs a value conversion.

(rillig)

s/maek/make/ in comment.

(andvar)

devsw_ok needs to survive across invocations of nvme_modcmd() so
allocate it statically.

Should address remaining issues with kern/56914

(pgoyette)