The NCQ support added a private request queue to the wd driver. This
makes the regular buffer queue ineffective, it also allowed to queue
an unlimited number of requests.

Fix this by limiting the number of requests queued to the driver to
the possible number of concurrent NCQ transactions.

(mlelstv)

- add ddb command "machine ttbr" to dump MMU tables.
- tidy up descriptions, usages and messages.

(ryo)

- free L1-L3 pages that has been emptied by pmap_remove().
- if no memories, pmap_enter will return correctly ENOMEM if PMAP_CANFAIL, or wait until available any memories if !PMAP_CANFAIL.

These changes improves the stability when we use a huge virtual memory spaces with mmap.

(ryo)

Ticket #1218

(martin)

Pull up following revision(s) (requested by ozaki-r in ticket #1218):

sys/netinet/ip_carp.c: revision 1.102

carp: don't skip pserialize_read_enter and ifa_release

(martin)

Fix typos in comment:
- s/paylaod/payload/
- s/dstination/destination/

(msaitoh)

New sentence, new line. Fix Dd argument.

(wiz)

Whitespace nits.

(wiz)

- dwc2 need calling linux_workqueue_init() to avoid panic.
- use INIT_ONCE/FINI_ONCE to linux_workqueue_{init,fini}() for being called from dwc2.

TODO: dwc2 should be written as kernel module depenging on a linux module.

(ryo)

add INIT_ONCE(9), FINI_ONCE(9) with changing once_t.

Welcome to 8.99.36

(ryo)

Provide CID data (manufacturer/product id/product name) as disk type.

(mlelstv)

Let backend provide a disk type. The information can be queried through
drvctl(4).

(mlelstv)

Set disk product name as disk type. The information can be queried through
drvctl(4).

(mlelstv)

Set disk model name as disk type. The information can be queried through
drvctl(4).

(mlelstv)

Move standby on detach after wedges deletion in case wedges trigger
I/O on the parent disk. Add debug messages.

(mlelstv)

libvdpau depends upon libXext and libX11, so make sure they're built first.

fixes x86 and evbarm builds.

(mrg)

enable the workarounds for cortex a53 errata 835769 and 843419 by default.

(mrg)

pax.1 & tar.1: add a minor clarification about "-s"

As a somewhat pedantic clarification, "-s" does not accept backslashes
as delimiters. (While here, also make the macro use of an expression
shared between pax.1 and tar.1 consistent.)

(gutteridge)

pax.1: document the "s" flag of the "s" option

Note the "s" option has an "s" flag that "prevents substitutions from
being performed on symbolic link destinations". Carry over r. 1.25 from
christos@ and part of r. 1.26 from wiz@ from tar.1, since this
functionality is available in pax as well as tar.

(gutteridge)

Report the users score _after_ the game has ended and the screen has
been restored to pre-game state.  Otherwise, the user gets a message
that their score did (or more likely, did not) improve over their
previous score, yet the user has no indication of what the current
score actually was!

(pgoyette)

Revert previous.

The lines look like misapplied patches but I'm told they are correct.

(wiz)

Kernel Heap Hardening: manage freed items with bitmaps rather than linked
lists when we're on-page and the page header is naturally big enough to
contain a bitmap.

This comes with no increase in memory consumption, and similar CPU cost
(maybe it's a little faster actually).

We want to favor bitmaps over linked lists, because linked lists install
kernel pointers inside the items, and this can be too easily exploitable
in use-after-free or double-free conditions, or in item buffer overflows
occurring within a pool page.

(maxv)

add libXext and libX11 to LIBDPLIBS.  fixes eg, minidlna build.

(mrg)

s/pakcet/packet/ in comment.

(msaitoh)

Make ZFS exportable by NFS, implement zfsctl_vptofh(),
zfs_netbsd_vptofh() and zfs_netbsd_fhtovp().

Undo changes to now completely unused zfs_fhtovp().

(hannken)

Implement zfsctl_lookup_objset() to look up "zfsvfs" of an mounted snapshot.

(hannken)

Set "mnt_stat.f_fsid" and "mnt_stat.f_fsidx" from the objset guid
like FreeBSD and Illumos do.

Use "f_fsid" for "va_fsid" and cheat NFSD to export snapshots under
".zfs" by setting these snaphots "f_fsidx" to the parents "f_fsidx".

(hannken)

As zfs_netbsd_getpages() ignores "a_count" and processes exactly one page
return an error if "a_count != 1".

(hannken)

Change zfs_readdir() to use malloc() instead of kmem_alloc() to
allocate cookies.  Required by protocol, caller uses free().

(hannken)

Remove weird lines from set lists.

(wiz)

merge imake 1.0.8.

(mrg)

merge makedepend 1.0.6 and x11perf 1.6.1.

(mrg)

merge libXvMC 1.0.11, libXxf86dga 1.1.5, libxkbfile 1.1.0, and xtrans 1.4.0.

(mrg)

revert previous - was wrong.

(mrg)

Add Davicom DM9161,DM9161A,DM9161B and DM9601 support.

(msaitoh)

remove delete sources in libXft 2.3.3.

(mrg)

merge libXdmcp 1.1.3, libXext 1.3.4, libXft 2.3.3, libXmu 1.1.3 and
libXrandr 1.5.2.

(mrg)

restore the removed "function does something horrible" description,
but leave the real description as well.

(mrg)

merge changes from xorg-server 1.20 list.h into this older version.

fixes sandy bridge crashing problems for me.

(mrg)

call destroy on a mutex before freeing it.
fixes LOCKDEBUG assert on a GT 6xx card.

(mrg)

Introduce a new flag, PR_USEBMAP, that indicates whether the pool uses a
bitmap to manage freed items. It dissociates PR_NOTOUCH from bitmaps, but
for now is set only when PR_NOTOUCH is set, which reproduces the current
behavior. Therefore, no functional change. Also clarify the code.

(maxv)

fix maintainer of nsd/unbound, update version

(christos)

Kernel Heap Hardening: put the pool header at the beginning of the backing
page, not at the end of it.

This makes it harder to exploit buffer overflows, because it eliminates the
certainty that sensitive kernel data is located after the item space and is
therefore overwritable.

The pr_itemoffset field is recycled, and holds the (aligned) offset of the
item space. The pr_phoffset field becomes unused. We align 'itemspace' for
clarity, but it's not strictly necessary.

This comes with no performance cost or increase in memory usage, in
particular the potential padding consumed by roundup(PHSIZE, align) was
already implicitly consumed before, because of the (necessary) truncations
in the divisions. Now it's just more explicit, but not bigger.

(maxv)

Move some code into a separate function, and explain a bit. Also define
PHSIZE. No functional change.

(maxv)

fix the build: include the needed osf1 files for compat_linux on alpha

(christos)

Remove unused install kernels.

(jmcneill)

Remove Pp that do nothing. Break line after macro. Fix typo.

(wiz)

With TRYEMULROOT namei_getstartdir() gets used twice so have to
vrele() "ni_rootdir" and "ni_erootdir" on entry.

(hannken)

Define missing HAVE_FONT for newcomers.

(rin)

Trailing whitespace

(skrll)

Whitespace

(skrll)

Wrap arm_cpu_marker if #ifdef MPDEBUG.

One day a9_mpsubr.S will die

(skrll)

Trailing whitespace

(skrll)

Fixup a comment

(skrll)

G/C

(skrll)

cosmetic

(maxv)

Prepare the removal of the 'ioff' argument: add a KASSERT to ensure it is
zero, and remove the internal logic. The pool code is simpler now.

(maxv)

Hard-align the fields of the structures with __aligned(32), and pass ioff=0
in the pool cache.

(maxv)

update SYNOPSIS; suggest 5x8 font instead of 8x8

(tnn)

import the Spleen wscons fonts from OpenBSD

Spleen is a 2-clause BSD licensed font set designed by Frederic Cambus.
The 5x8 font is designed for low resolution displays, but the font
comes in a variety of sizes suitable for larger display too.

(tnn)

ensure the device's UVM pager object is present before using uvm_pageratop

(tnn)

delete unused variable

(tnn)

formatting

(tnn)

note addition of ssdfb(4)

(tnn)

config glue for ssdfb

(tnn)

add i2c attachment for ssdfb

(tnn)

add ssdfb

The ssdfb driver provides wsdisplay(4) support for OLED/PLED framebuffer
modules based on one of the following controller chips:

- Solomon Systech Ltd SSD1306
- Sino Wealth Electronic Ltd SH1106

It supports xf86-video-wsfb and can optionally attach as the console.

Some products that should work with this driver include:
- Adafruit 0.96" 128x64 OLED graphic display
- Adafruit 1.3" 128x64 OLED graphic display
- Adafruit 128x32 OLED graphic display
- Adafruit PiOLED 128x32 Monochome OLED for Raspberry Pi
- "GM009605" commonly distributed with Arduino starter kits
- display modules made by Chengdu Heltec Automation technology co. LTD

Note: I used the name ssdfb(4) because that's what OpenBSD calls their
driver but the implementation is entirely unique to NetBSD.

(tnn)

add ssdfb(4) manual page

(tnn)

allocate wsdisplay type 66 for ssdfb(4)

(tnn)

we have MAP_ALIGNED, so use it (although it does not do anything by default)

(christos)

PR/54061: Bump memory for jemalloc (a lot).

(christos)

PR/54062: Using random garbage in the mmsghdr (in this case 0xa5's from
jemalloc debug) makes the syscall unhappy. Zero out the memory instead...

(christos)

PR/54062: Fix the "snprintf_float" test failing after the jemalloc import
issue: jemalloc uses a lot more memory (if it can find it) by default and
this test limited the amount of memory it could get, causing it to get to
an error path with a missing mutex_unlock...

(christos)

Misc changes:

- Turn two KASSERTs to real panics, they are useful and not expensive.
- Rename a few variables for clarity.
- Add a new panic, to make sure a freed item is in the item space.

(maxv)

Disable multi-segment DMA as a workaround for init crashes on ODROID-C2 when using root on eMMC

(jmcneill)

White space.

(isaki)

Use C99 style struct initializer to audio_hw_if.

(isaki)

Bump STACK_ALIGNBYTES to (16 - 1) to satisfy requirement by AMD64
System V ABI in kernel level. This is because

(1) for LLDB, we want to bypass libc/csu (and therefore manual stack
    alignment in _start), and

(2) rtld in glibc >= 2.23 for Linux/x86_64 requires it.

Fix SEGV for Linux/x86_64 binaries with glibc >= 2.23, reported as
PR port-amd64/54052.

(rin)

Make it compilable (remove unused variables).

(isaki)

Spell common correctly in comment. Spotted by jmcneill

(skrll)

Style and whitespace. NFC.

(skrll)

Flush the BP cache just before starting APs so that cache coherency starts
with a correct view.

(skrll)

Disable COMPAT_OSF1, will be removed.

(maxv)

Fix (probably) typo.  SLINEAR_LE -> ULINEAR_LE.

(isaki)

Swap wrong ordered halt_input and halt_output (since rev1.1).
But I guess that it didn't lead a big impact.  Because hw_if->close
is called immediately after that in most cases.

(isaki)

Remove duplicated option.

(isaki)

Support more ALPS V2 devices

* The V2 devices found in Toshiba dynabook satellite B551/D and
  dynabook SS RX1/T8E.

(ryoon)

moar registers

(macallan)

Tickets #1214 - #1217

(martin)

Pull up following revision(s) (requested by kardel in ticket #1217):

sys/dev/pci/mpii.c: revision 1.22

PR/54045

fix mpii to adhere to physio diagnostic invariant that
fully processed data must not post an error:

1) verify expected scspi state via KASSERT() instead of just
    setting the variables.
2) set xs->resid only in known good conditions
3) insure setting errors in all error paths and refrain
    from clearing xs->resid in error paths.

While there do some cosmectic clean up:
1) extend and relocate some debug output
2) mpii HBAs can also manage non-disk devices like tapes etc,
    so log that physical "devices" instead of physical "disks" are
    attached or detached.

Tested with NEOSeries FlexStor II and luckily a broken tape drive 8-(

mpii0 at pci1 dev 0 function 0: vendor 1000 product 00ab (rev. 0x01)
mpii0: interrupting at irq 11
mpii0: HBA 9400-8i8e, firmware 3.0.4.0, MPI 2.6
mpii0: physical device inserted in slot 9
mpii0: physical device inserted in slot 13
mpii0: physical device inserted in slot 16
st0 at scsibus0 target 9 lun 0: <IBM, ULTRIUM-HH7, J4D1> tape removable
st0: density code 92, variable blocks, write-enabled
ch0 at scsibus0 target 9 lun 1: <BDT, FlexStor II, 5.50> changer removable
ch0: 23 slots, 2 drives, 1 picker, 1 portal
st0: tagged queueing
ch0: tagged queueing
st1 at scsibus0 target 13 lun 0: <IBM, ULTRIUM-HH7, J4D1> tape removable
st1: density code 92, variable blocks, write-enabled
st1: tagged queueing
ses0 at scsibus0 target 16 lun 0: <LSI, VirtualSES, 01> enclosure
services fixed

Note: pullup-8

(martin)

Pull up following revision(s) (requested by knakahara in ticket #1216):

sys/net/if_ipsec.c: revision 1.21

Fix ipsecif(4) memory leak in some ioctl cases.

(martin)

Pull up following revision(s) (requested by ozaki-r in ticket #1215):

sys/net/route.c: revision 1.217

Add missing ifa_release on error paths

(martin)

Ticket #1686

(martin)

Pull up following revision(s) (requested by rin in ticket #1686):

lib/libcurses/get_wch.c: revision 1.21
lib/libcurses/getch.c: revision 1.71

Rename global variable "state" to "_cursesi_state".

Until now, if application happens to have a global variable of the same
name, it was overridden by curses routines. This is the scenario in
which aspell crashes when linked to our curses, reported in pkg/44005.

We need to wipe out global/static variables like "_cursesi_state" or
"wstate" for thread safety. But it would be a future task...

XXX pullup to netbsd-8 and netbsd-7

(martin)

Pull up following revision(s) (requested by rin in ticket #1214):

lib/libcurses/get_wch.c: revision 1.21
lib/libcurses/getch.c: revision 1.71

Rename global variable "state" to "_cursesi_state".

Until now, if application happens to have a global variable of the same
name, it was overridden by curses routines. This is the scenario in
which aspell crashes when linked to our curses, reported in pkg/44005.

We need to wipe out global/static variables like "_cursesi_state" or
"wstate" for thread safety. But it would be a future task...

XXX pullup to netbsd-8 and netbsd-7

(martin)

Fix SCF_RSP_R1B test

(jmcneill)

spell it defined(__NetBSD__) so it actually matches and we skip this
pragma warning during the build:

Warning: Per application configuration won't work with your OS version.

(mrg)

properly mark libmesa_dri.so obsolete in mesa18 x86 builds.

(mrg)

libmesa_dri.so is obsolete in mesa18 for x86 as well.

(mrg)

Regen.

(msaitoh)

Add Davicom DM9161, DM9161[ABC] and DM9601.

(msaitoh)

remove INCLUDES.all setting and it's processing.  already done elsewhere.

(mrg)

- Simplily. Suggested by knakahara.
- Modify comment. Per queue VLAN enable flags is on 82599 and later.
- Fix typo in comment.

(msaitoh)

Use copyin(), not fusword().

(thorpej)

revert previous sshd hack (from Rin Okuyama)

(christos)

Grr, disable the "time goes backwards" test here since this breaks too
frequently. We need a real fix.

(christos)

unify rounding and range checking.

(christos)

Optimize NVMM-Intel: keep the VMCS active on the host CPU, and lazy-switch
it on demand only when needed. This allows the CPU to use the cached
version of the guest state, rather than the in-memory copy of it. This is
much more performant.

A VMCS must be active on only one CPU, but one CPU can have several active
VMCSs at the same time.

We keep track of which CPU each VMCS is active on. When we want to execute
a VCPU, we determine whether its VMCS is loaded on another CPU, and if so
send an IPI to ask it to unbusy that VMCS. In most cases the VMCS is
already active on the current CPU, so we don't have to do anything and can
proceed with a fast VMRESUME.

We send IPIs with kpreemption enabled but with a bound LWP, because we
don't want to get context-switched to the CPU we just sent an IPI to.

Overall, with this in place, I see a ~15% performance increase in the
guests on NVMM-Intel.

(maxv)

syscall debug - fix build when SYSCALL_DEBUG option is present in kernel config file

(palle)

Move a KASSERT, applies to all branches.

(maxv)

Reduce the mask of the VTPR, only the first four bits matter.

(maxv)

Fail early if we're beyond the guest max ram.

(maxv)

Avoid a panic from the sequence

mlock(buf, 0);
munlock(buf, 0);
mlock(buf, page);
munlock(buf, page);

where buf is page aligned, and page is actually anything > 0
(but not too big) which will get rounded up to the next multiple
of the page size.

In that sequence, it is possible that the 1st munlock() is optional.

Add a KASSERT() (or two) to detect the first effects of the problem
(without that, or in !DIAGNOSTIC kernels) the problem eventually
causes some kind of problem or other (most often still a panic.)

After this, mlock(anything, 0) (or munlock) validates "anything"
but is otherwise a no-op (regardless of the alignment of anything).

Also, don't treat mlock(buf, verybig) as equivalent to mlock(buf, 0)
which is (more or less) what we had been doing.

XXX pullup -8 (maybe -7 as well, need to check).

(kre)

Sync with installed versions.

(christos)

Remove __clang___ hacks for noreturn, reduce footprint when we don't compile
with JEMALLOC_PROF. More to do here.

(christos)

G/C "MAPPECOPY".  It is the deadest of wood, and never actually worked in
NetBSD.

(thorpej)

print little-endian header fields correctly.

(mlelstv)

U-Boot fails to boot a compressed kernel_noload image, so stop emitting
them.

(jmcneill)

carp: don't skip pserialize_read_enter and ifa_release

(ozaki-r)

Fix ipsecif(4) memory leak in some ioctl cases.

(knakahara)

Rename global variable "state" to "_cursesi_state".

Until now, if application happens to have a global variable of the same
name, it was overridden by curses routines. This is the scenario in
which aspell crashes when linked to our curses, reported in pkg/44005.

We need to wipe out global/static variables like "_cursesi_state" or
"wstate" for thread safety. But it would be a future task...

XXX pullup to netbsd-8 and netbsd-7

(rin)

regen

(christos)

In zx_cursor_color(), don't throw away the second half of the color
map data.

(thorpej)

Fix an issue with FIOCSCURSOR FB_CUR_SETCMAP on zx described here:

    http://mail-index.netbsd.org/port-sparc/2019/02/11/msg002134.html

by replacing the fubyte() calls with copyin() into temporary buffers.

The other issue in zx_cursor_color() mentioned in that message will
be addressed separately.

(thorpej)

style

(maxv)

Ticket #1684.

(msaitoh)

Ticket #1684.

(msaitoh)

Ticket 1683-1685.

(msaitoh)

Add support for Arasan SDHCI 5.1

(jmcneill)

Add vendor callback for post-bus clock ops and add SDHC_FLAG_STOP_WITH_TC flag

(jmcneill)

Pull up following revision(s) (requested by martin in ticket #1685):
sys/arch/mvme68k/dev/pcctwo_68k.c: revision 1.11
Fix small (but fatal) oversight in device/softc split.

(msaitoh)

Pull up following revision(s) (requested by martin in ticket #1684):
external/bsd/bzip2/dist/bzip2recover.c: revision 1.4
Apply fix for CVE-2016-3189 bzip2: heap use after free in bzip2recover

(msaitoh)

Pull up following revision(s) (requested by martin in ticket #1684):
external/bsd/bzip2/dist/bzip2recover.c: revision 1.4
Apply fix for CVE-2016-3189 bzip2: heap use after free in bzip2recover

(msaitoh)

Pull up following revision(s) (requested by martin in ticket #1684):
external/bsd/bzip2/dist/bzip2recover.c: revision 1.4
Apply fix for CVE-2016-3189 bzip2: heap use after free in bzip2recover

(msaitoh)

Pull up following revision(s) (requested by martin in ticket #1683):
sys/compat/netbsd32/netbsd32_compat_60.c: revision 1.3
don't forget to convert the timespec.
XXX: pullup-8

(msaitoh)

Ticket #1213.

(msaitoh)

Pull up following revision(s) (requested by martin in ticket #1213):
sys/arch/mvme68k/dev/pcctwo_68k.c: revision 1.11
Fix small (but fatal) oversight in device/softc split.

(msaitoh)

Fix aclk_emmc register offset, set RK_COMPOSITE_ROUND_DOWN for SD/EMMC clocks, and add a few more emmc clock nodes

(jmcneill)

Set drive strength and output tap delay

(jmcneill)

Fix a bug that the VLAN HW filter function is not correctly disabled when
all vlan is detached. Part of ixgbe.c rev. 1.177.

(msaitoh)

- Fix a bug that the VLAN HW filter function is not correctly disabled
  when all vlan is detached.
- Fix a bug that VLAN HW filter function is not correctly controlled on 82598.
- Control VLAN HW filter function correctly.
- Don't clear IXGBE_VLNCTRL_CFIEN bit When ETHERCAP_VLAN_HWFILTER is set.
  I think it's not required (and Linux doesn't do it). This change has no
  effect to NetBSD because ETHERCAP_VLAN_HWFILTER is not supported yet.

(msaitoh)

fix merge botch in previous.  i *think* this is now right.  it sure builds..

(mrg)

regen for openssl 1.1.1b.

not 100% sure this is right but my build works now.  here's what i did:

  % cd crypto/external/bsd/openssl/lib/libcrypto/arch/powerpc
  % make regen

after checking that the openssl generator calling code had not change
(but the relevant .pl file had.)

(mrg)

Add a bunch of fprintf(stderr,...) diagnostics to try and help determine
what is going wrong which is causing this test to crash the kernel.

(kre)

Fix small (but fatal) oversight in device/softc split.

(martin)

Add TSX_FORCE_ABORT related definitions.

(msaitoh)

catch up the default __XCURSORPATH__ from libXcursor 1.2.0.

(mrg)

lint library for jemalloc

(christos)

lint library for vdpau

(christos)

Fix typo.

(wiz)

Fix paste-o in an error path diagnostic message.

(jakllsch)

121[12]

(snj)

Pull up following revision(s) (requested by martin in ticket #1212):
external/bsd/bzip2/dist/bzip2recover.c: revision 1.4
Apply fix for CVE-2016-3189 bzip2: heap use after free in bzip2recover

(snj)

Pull up following revision(s) (requested by martin in ticket #1211):
sys/compat/netbsd32/netbsd32_compat_60.c: 1.3
don't forget to convert the timespec.

(snj)

new openssl

(christos)

merge conflicts

(christos)

fix pasto (gnumalloc vs jemalloc)

(christos)

add libjemalloc for the non-standard jemalloc symbols.

(christos)

- descend to build libjemalloc
- namespace protect all functions __je
- provide getters and setters for malloc_conf and malloc_message

(christos)

remove global pointers, add get/seters.

(christos)

Add libjemalloc.so

(christos)

Take a reference on ndp->ni_rootdir and ndp->ni_erootdir.

A multithreaded process may chroot during namei() and we end up with
vn_under() trying to reference the now unreferenced ni_rootdir.

Ok: David Holland <dholland@netbsd.org>

Reported-by: syzbot+889319cdf91a3d0373a9@syzkaller.appspotmail.com

(hannken)

it is CPPFLAGS for -D etc otherwise lint breaks!

(christos)

Add libvdpau_g.a here as well

(martin)

remove the recently added files or added entries for what is
./usr/include/$MACHINE and automatically add it.

add sun68k for sun2/sun3.  tested on about 12 platforms.

there may be more failures to figure out but this should make
them all special cases rather than always expected cases.

fortunately, most builds don't see to hit these issues for
reasons i haven't determined yet.

(mrg)

Add missing space in "wiring for pmap .. did not change" message

(gson)

KNF

(skrll)

move the mesa 18 khrplatform.h header into xcomp/mi.

(mrg)

Use pmf(9).

(msaitoh)

Correct copy-paste error. From joerg, thanks.

(maya)

This had a similar problem to that reported in PR lib/54053
for the userland (libc) version of humanize_number(),
except in a much more limited, and less harmful, way ...

If the value to be printed was > ~0/10 (eg ~0 itself) and
the buffer given was big enough to hold it, "umax" would overflow,
and the expected result was not produced (here at least the effect was
simply to scale the result for huge values by one unnecessary step,
rather than maximally scale all values usually generating 0E, but
incorrect nonetheless.)

(kre)