No need to use I2C_F_POLL here.

(thorpej)

No need to use I2C_F_POLL here.

(thorpej)

Also force I2C_F_POLL if shutting_down.

(thorpej)

Welcome to NetBSD 9.99.33 -- boottime global is gone.

(thorpej)

- Eliminate the global "boottime" variable, which was being accessed
  without any synchronization against changes by e.g. clock_settime().
- Replace with new getbinboottime() / getnanoboottime() / getmicroboottime()
  functions (naming mirrors that of other time access functions in kern_tc.c).
  It returns the (maybe-converted) value of timebasebin, which also tracks
  our estimate of when the system was booted (i.e. the legacy "boottime" was
  redundant).

XXX There needs to be a lockless synchronization mechanism for reading
timebasebin, but this is a problem in kern_tc.c that pre-existed these
"boottime" changes.  At least now the problem is centralized in one location.

(thorpej)

gcc 8 needs -mapcs-frame when DDB is specified for backtrace to work

(skrll)

Tickets #592 - #599

(martin)

Pull up following revision(s) (requested by skrll in ticket #599):

sys/arch/arm/broadcom/bcm283x_platform.c: revision 1.34 (patch)

Provide a single generic bcm283x_platform_uart_freq which checks for
"brcm,bcm2835-aux-uart" and returns the appropriate core clock based
results; otherwise it returns the uart clock from firmware.

Should fix early console output on the rpi0w.

Based on a diff from mlelstv

(martin)

Ticket #1719

(martin)

Pull up following revision(s) (requested by mlelstv in ticket #1719):

sys/conf/copyright: revision 1.18

Welcome to 2020.

(martin)

Ticket #1719

(martin)

Pull up following revision(s) (requested by mlelstv in ticket #1719):

sys/conf/copyright: revision 1.18

Welcome to 2020.

(martin)

Ticket #1719

(martin)

Pull up following revision(s) (requested by mlelstv in ticket #1719):

sys/conf/copyright: revision 1.18

Welcome to 2020.

(martin)

Ticets #1480 and #1481

(martin)

Pull up following revision(s) (requested by mlelstv in ticket #1481):

sys/conf/copyright: revision 1.18

Welcome to 2020.

(martin)

Pull up following revision(s) (requested by mlelstv in ticket #598):

sys/conf/copyright: revision 1.18

Welcome to 2020.

(martin)

Pull up following revision(s) (requested by maxv in ticket #597):

sys/compat/common/kern_sig_43.c: revision 1.36
sys/compat/netbsd32/netbsd32_compat_20.c: revision 1.39
sys/compat/netbsd32/netbsd32_compat_43.c: revision 1.59
sys/compat/netbsd32/netbsd32_compat_50.c: revision 1.44

Fix sizeof mismatch in copyin. This leads to a user-triggerable stack
overflow. On my test build at least, by luck, the compiler orders the
variables in a way that the overflow hits only local structures which
haven't yet been initialized and used, so the overflow is harmless.

Very easily seeable with kASan - just invoke the syscall from a 32bit
binary.

Fix three stack info leaks, found by kMSan when just invoking all syscalls
with a zero page as argument.

MSan: Uninitialized Stack Memory In copyout() At Offset 0, Variable 'sb32' From compat_20_netbsd32_getfsstat()
MSan: Uninitialized Stack Memory In copyout() At Offset 12, Variable 'oss' From compat_43_sys_sigstack()
MSan: Uninitialized Stack Memory In copyout() At Offset 0, Variable 'sb' From compat_50_netbsd32___fhstat40()

(martin)

Pull up following revision(s) (requested by skrll in ticket #596):

usr.bin/mkubootimage/mkubootimage.c: revision 1.28
usr.bin/mkubootimage/mkubootimage.c: revision 1.29

Adjust image size appropriately when using update_image
OK jmcneill

Avoid mixing signed/unsigned arguments to the ? operator.

(martin)

Pull up following revision(s) (requested by maxv in ticket #1480):

sys/dev/usb/uthum.c: revision 1.18
sys/dev/usb/ucycom.c: revision 1.49
sys/dev/usb/uhid.c: revision 1.111

Fix buffer overflows. sc_{o,f}len are controlled by the USB device. By
crafting the former the device can leak stack data. By crafting the latter
the device can overwrite the stack. The combination of the two means the
device can ROP the kernel and obtain code execution (demonstrated with an
actual exploit over vHCI).

Truncate the lengths to the size of the buffers, and also drop sc_ilen
since it is unused. Patch tested with vHCI+kASan.

-

Fix buffer overflows. Also add missing mutex_exit.

-

Fix buffer overflows: validate the lengths at attach time, given that they
are apparently not supposed to be variable. Drop sc_ilen since it is
unused.

(martin)

Pull up following revision(s) (requested by maxv in ticket #595):

sys/dev/usb/uthum.c: revision 1.18
sys/dev/usb/ucycom.c: revision 1.49
sys/dev/usb/uhid.c: revision 1.111

Fix buffer overflows. sc_{o,f}len are controlled by the USB device. By
crafting the former the device can leak stack data. By crafting the latter
the device can overwrite the stack. The combination of the two means the
device can ROP the kernel and obtain code execution (demonstrated with an
actual exploit over vHCI).

Truncate the lengths to the size of the buffers, and also drop sc_ilen
since it is unused. Patch tested with vHCI+kASan.

-

Fix buffer overflows. Also add missing mutex_exit.

-

Fix buffer overflows: validate the lengths at attach time, given that they
are apparently not supposed to be variable. Drop sc_ilen since it is
unused.

(martin)

Pull up following revision(s) (requested by isaki in ticket #594):

sys/arch/x68k/include/param.h: revision 1.29

Bump MSGBUFSIZE (2 * NBPG).

Since timestamp was introduced, it was too small to store full dmesg
of one boot.

And putting this behind of #include <m68k/param.h> had no effect.

This was fixed on many arch in 1997 but x68k was not...
http://mail-index.netbsd.org/source-changes/1997/09/20/0021.html

(martin)

Pull up following revision(s) (requested by isaki in ticket #593):

sys/dev/audio/audio.c: revision 1.34
sys/dev/audio/audio.c: revision 1.35

Use M_WAITOK instead of M_NOWAIT.
These allocations don't require NOWAIT constraints.

Will fix PR kern/54796.

-

Improve and simplify around audio_realloc().

(martin)

Pull up following revision(s) (requested by tsutsui in ticket #592):

usr.sbin/postinstall/postinstall.in: revision 1.9
usr.sbin/postinstall/postinstall.in: revision 1.10
usr.sbin/postinstall/postinstall.in: revision 1.11
usr.sbin/postinstall/postinstall.in: revision 1.12

remove debugging.

PR/54730: Izumi Tsutsui: Use /var/db/obsolete/<set> to remove obsolete rc.d
files when not in $SOURCEMODE.

PR/54730: Izumi Tsutsui: obsolete etc files are not being cleaned up on
an upgrade build.

Make sure rc, rc.subr, and rc.shutdown are properly updated.

Currently there is no info which rc* files should be updated
or not on upgrade (at least rc.conf and rc.local shouldn't),
so put back an explicit list in the postinstall script.
"Go for it" by christos@ in PR/54741.

Should be pulled up to netbsd-9.

(martin)

fstyp: Fix "exfat: iconv_open UCS-2LE: Invalid argument" failure

Fix below error for the time being.
The removed code is only relevant to Capsicum.

$ fstyp -l /dev/wd1
fstyp: exfat: iconv_open UCS-2LE: Invalid argument

(tkusumi)

Add shutting_down variable for rump.

(martin)

Remove the call to KERNEL_UNLOCK_ONE(), it was forgotten when the biglock
was dropped in rev1.63. Found via vHCI.

(maxv)

dm: Add dm-flakey target

Ported from DragonFlyBSD, but this target had originally existed in
Linux kernel. See below for details.
https://www.kernel.org/doc/Documentation/device-mapper/dm-flakey.txt

Due to two technical issues, this hasn't been hooked to dm.kmod yet.
1) "tick" in hz(9) not working.
2) Unable to use ->b_private in nestiobuf callback when it's already
used for mbp (see HAS_BUF_PRIV2).

taken-from: DragonFlyBSD

(tkusumi)

Sync the e_ident[EI_OSABI] magic values with the ELF specification

source: http://www.sco.com/developers/gabi/latest/ch4.eheader.html

(kamil)

Update e_machine magic values with the ELF specification

source: http://www.sco.com/developers/gabi/latest/ch4.eheader.html

(kamil)

Fix the EM_TILEGX number in the ELF header

source: http://www.sco.com/developers/gabi/latest/ch4.eheader.html

191 is confirmed in LLVM, GNU toolchain, FreeBSD.

(kamil)

Back out the amap allocation  changes from earlier today - have seen a panic
with them.  Retain the lock changes.

(ad)

mi_cpu_init: set ci_smt_primary early.

(ad)

Add driver for simple-audio-amplifier binding

(jmcneill)

Add support for single power writes, and enable for Ricoh 5u823.

(mlelstv)

Now that we know if we're shutting down, enable locking in resettodr().
In the shutting_down case, we acquire the todr_mutex as a trylock.  Under
most circumstances, this lock will be uncontended.  But if we happen to
panic while accessing the TOD clock (via clock_settime() or whatever),
then trylock will prevent us from getting stuck (and a warning will be
displayed on the console).

(thorpej)

Pull in <sys/stdbool.h>.

(thorpej)

- Introduce a new global kernel variable "shutting_down" to indicate that
  the system is shutting down or rebooting.
- Set this global in a new function called kern_reboot(), which is currently
  just a basic wrapper around cpu_reboot().
- Call kern_reboot() instead of cpu_reboot() almost everywhere; a few
  places remain where it's still called directly, but those are in early
  pre-main() machdep locations.

Eventually, all of the various cpu_reboot() functions should be re-factored
and common functionality moved to kern_reboot(), but that's for another day.

(thorpej)

- Start trying to reduce the high cache miss rate observed around vm_amap.
  On _LP64, pad struct vm_amap to 128 bytes and use the additional space to
  hold the arrays for tiny amaps which are common.  Carefully size the array
  allocations to avoid false sharing, and for smaller amaps try to share
  allocated cache lines.

- Eliminate most contention due to amap_list: maintain the list in the pool
  cache constructor / destructor like we do for struct file.  Cache the
  mutexes we allocate here.

- Don't do PR_WAITOK mutex allocations when NOWAIT has been specified.

(ad)

Add some new functions for lock objects:

mutex_obj_refcnt(), mutex_obj_tryalloc()
rw_obj_refcnt(), rw_obj_tryalloc()

(ad)

First steps towards properly serializing access to the TOD clock.
- Add a mutex around the TODR, and provide lock/unlock/lock-owned
  functions to manipulate it.
- Rename inittodr() to todr_set_systime() and resettodr() to
  todr_save_systime() to better reflect what they do.  These functions
  are intended to be called with the TODR lock held, which will allow
  for a pattern like:
-> todr_lock()
-> todr_save_systime()
-> [do machine-dependent stuff to sleep/suspend]
-> [magically awaken]
-> todr_set_systime(...)
-> todr_unlock()
- Provide historically-named wrappers inittodr() and resettodr() that
  do the dance of acquiring / releasing the lock around the actual
  substance.

NOTE: resettodr()'s use of the TODR lock is currently disabled (and
todr_save_systime() does not assert it's held) until such time as
issues around shutdown / reboot under duress can be addressed.

(thorpej)

Break line after macro ends; add article

(wiz)

Try fixing a sentence.

(wiz)

One more vestige.

(pgoyette)

Emove vestigial remains of the monolithic compat module.

(pgoyette)

Fix some issues around todr_wenable():
- As previously defined, it was not possible to return an error from the
  back-end RTC driver.  Make it a real function so that it can do so.
- Only the mc146818 and mk48txx drivers used it (for historical reasons).
  Centralize the logic for how it's used in kern_todr.c (and make it private
  to that file) for consistency.

(thorpej)

tmux-3.0a out.

(wiz)

No need to use I2C_F_POLL here.

(thorpej)

todr_gettime() and todr_settime() are not referenced outside of kern_todr.c,
so make them private to that file.

(thorpej)

Remove superfluous splclock()/splx() pair around tc_setclock().

(thorpej)

Revert previous (include of sys/param.h) - the headers requiring this
have been fixed.

(martin)

Fix three stack info leaks, found by kMSan when just invoking all syscalls
with a zero page as argument.

MSan: Uninitialized Stack Memory In copyout() At Offset 0, Variable 'sb32' From compat_20_netbsd32_getfsstat()
MSan: Uninitialized Stack Memory In copyout() At Offset 12, Variable 'oss' From compat_43_sys_sigstack()
MSan: Uninitialized Stack Memory In copyout() At Offset 0, Variable 'sb' From compat_50_netbsd32___fhstat40()

(maxv)

Fix a comment.

(ad)

upstream_kernel=1 is required with new firmware/kernel.

(skrll)

Disable SPI for now (rkspi driver hangs at boot)

(jmcneill)

Provide a single generic bcm283x_platform_uart_freq which checks for
"brcm,bcm2835-aux-uart" and returns the appropriate core clock based
results; otherwise it returns the uart clock from firmware.

Should fix early console output on the rpi0w.

Based on a diff from mlelstv

(skrll)

PR kern/54821: 9.99.32 assertion in uvm_pageactivate

Looks like I forgot to commit this file yesterday.

(ad)

No need to print all supported levels at attach, print the range and total number of steps

(jmcneill)

fstyp: Consider '@' syntax in device file path for HAMMER2

though devpath is unsupported in NetBSD atm.
taken-from: DragonFlyBSD

(tkusumi)

Quiet chatty printfs

(jmcneill)

Fix performance regression with previous

(jmcneill)

Use correct firmware for BCM43456

(jmcneill)

regen

(jmcneill)

Add product ID for Broadcom BCM43455

(jmcneill)

Avoid mixing signed/unsigned arguments to the ? operator.

(martin)

fstyp: Fix build failure on i386

http://releng.netbsd.org/b5reports/i386/commits-2020.01.html#2020.01.01.10.13.16
Explicitly cast to size_t.

(tkusumi)

dwcmmc improvements:
- Use mmcpwrseq resources if available
- Only set 4- or 8-bit mode if specified in the dt properties
- Add quirk for implementations with inverted power enable logic
- Support switching signal voltage between 1.8V and 3.3V
- Fix a clock divider issue on Rockchip SoCs

(jmcneill)

Attach mmcpwrseq resource earlier

(jmcneill)

+ rk3399-pinebook-pro.dts

(jmcneill)

Add Pinebook Pro dts, from Manjaro Linux.

https://gitlab.manjaro.org/tsys/linux-pinebook-pro/blob/877ca0e7283596f37845de50dc36bff5b88b91e1/arch/arm64/boot/dts/rockchip/rk3399-pinebook-pro.dts

(jmcneill)

add aq(4)

(ryo)

Adjust image size appropriately when using update_image

OK jmcneill

(skrll)

note aq(4)

(ryo)

Add module build for aq(4)

(ryo)

add support Aquantia AQC seriese 10G network adapters.

this driver is based on the FreeBSD version https://github.com/Aquantia/aqtion-freebsd ,
but drastically rewritten for NetBSD.

(ryo)

regen

(ryo)

add Aquantia AQC 10G network adapters

(ryo)

Add the ETHERTYPE_QINQ for 802.1ad VLAN stacking

(ryo)

Fix small read overflows when parsing HID tables. Noticed by kASan the
other day while I was playing with vHCI.

(maxv)

The DMA range doesn't cover all of memory on the bcm2711 (rpi4) - it
covers (most of) the first 1GB.

(skrll)

Fix build for non-__HAVE_UCAS_FULL platforms where the symbol dance for
assembler functions isn't needed.

Suggestion from maxv.  Thanks.

(skrll)

Fix sizeof mismatch in copyin. This leads to a user-triggerable stack
overflow. On my test build at least, by luck, the compiler orders the
variables in a way that the overflow hits only local structures which
haven't yet been initialized and used, so the overflow is harmless.

Very easily seeable with kASan - just invoke the syscall from a 32bit
binary.

(maxv)

fstyp: Add HAMMER1 multi-volume support (missed in "fstyp: Add HAMMER1/2 support")

taken-from: DragonFlyBSD

(tkusumi)

fstyp: Cleanup and minor sync up with FreeBSD/DragonFlyBSD

(tkusumi)

Fix buffer overflows: validate the lengths at attach time, given that they
are apparently not supposed to be variable. Drop sc_ilen since it is
unused.

(maxv)

Fix buffer overflows. Also add missing mutex_exit.

(maxv)

Fix buffer overflows. sc_{o,f}len are controlled by the USB device. By
crafting the former the device can leak stack data. By crafting the latter
the device can overwrite the stack. The combination of the two means the
device can ROP the kernel and obtain code execution (demonstrated with an
actual exploit over vHCI).

Truncate the lengths to the size of the buffers, and also drop sc_ilen
since it is unused. Patch tested with vHCI+kASan.

(maxv)

fstyp: Add HAMMER1/2 support

FreeBSD has recently imported HAMMER1/2 support from DragonFlyBSD,
so why not in NetBSD as well.

taken-from: DragonFlyBSD

(tkusumi)

Welcome to 2020.

(mlelstv)

explicitely include sys/atomic.h for atomic operations.

(mlelstv)

Add clk provider

(jmcneill)

Make recent change build for SMALL builds as well as normal ones.

(kre)

Fix rump.

(ad)

- Add and use wrapper functions that take and acquire page interlocks, and pairs
  of page interlocks.  Require that the page interlock be held over calls to
  uvm_pageactivate(), uvm_pagewire() and similar.

- Solve the concurrency problem with page replacement state.  Rather than
  updating the global state synchronously, set an intended state on
  individual pages (active, inactive, enqueued, dequeued) while holding the
  page interlock.  After the interlock is released put the pages on a 128
  entry per-CPU queue for their state changes to be made real in batch.
  This results in in a ~400 fold decrease in contention on my test system.
  Proposed on tech-kern but modified to use the page interlock rather than
  atomics to synchronise as it's much easier to maintain that way, and
  cheaper.

(ad)

NetBSD 9.99.32 - struct vm_page changed

(ad)

Remove a stupid printf

(jmcneill)

Fix dmesg(8) to preserve leading whitespaces of kernel messages.

Closes PR/54729.

(tsutsui)

Remove a stray '

(skrll)

Improve a comment

(skrll)

struct vm_page: cluster fields most heavily used by the page allocator and
uvmpdpol at the start of the structure, so that while under global lock we
need only touch one cache line for each vm_page.  There is still the problem
of vm_page not being aligned, but this seems to drop lock wait time for
(a modified) uvmpdpol and the allocator by 20-30% in a quick test.

(ad)

Rely on 32/64-bit overflow to calculate translation offsets. Store this
as a separate ar_xbase field in acpi_mem instead of having separate
ar_offset and ar_decode fields.

(jmcneill)

Update to not include COMPAT_90

(pgoyette)

Another rename from uvm_free() --> uvm_availmem()

(pgoyette)

Fix mis-placed parentheses.  PR kern/54813.

(thorpej)

Fix mis-placed parentheses.  PR kern/54812.

(thorpej)

Create bus_dma tags for each device node based on _CCA and _DMA properties
found by walking up the device node tree. These tags encode range
restrictions, address translations, and whether or not the device is
cache coherent.

(jmcneill)

Include device decoding type in acpi_mem resources

(jmcneill)

Rename uvm_free() -> uvm_availmem().

(ad)

Rename uvm_page_locked_p() -> uvm_page_owner_locked_p()

(ad)

Fetch bus_dma tags when acpi devnodes are created. They do not change
and this allows MD code to create more complex tags without being
concerned with the tag being destroyed later. While here, capture
translations offsets for address32/address64 resources.

(jmcneill)

sys_fchdir: use LK_SHARED.

(ad)

The DMA restrictions may not be defined in the direct parent of a device,
so search up the tree for a module device.

(jmcneill)

dm: Remove unnecessary inlining

These two don't really need to be inlined.

(tkusumi)

Free buffers only when allocated.

(mlelstv)

Add another cache invalidation before reading memory for firmware query
results.

(skrll)

Done giving syspkg names without xorg_server_ver=110

(uki)

Give syspkg name to macppc machine dependency files

(uki)

Give syspkg name to x68k machine dependency files

(uki)

Split vchiq into separate fdt and acpi attachments.

(jmcneill)

acpi glue is conditional on acpinodebus

(jmcneill)

Fix a problem with intr_unmask() that can cause a forever-loop:
- When handling the source-is-masked case in the interrupt vector, set the
  interrupt bit in a new ci_imasked field and ensure the bit is cleared
  from ci_ipending.
- In intr_unmask(), transfer the bit from ci_imasked to ci_ipending for
  non-level-sensitive interrupts (the PIC does the work for us in the
  level-sensitive case), and only force pending interrupts to be processed
  in this case.  (In all cases, make sure the now-unmasked bit is cleared
  from ci_imasked.)

Before, the bit was left in ci_ipending so as not to use edge-triggered
interrupts while the source is masked, but Xspllower() relies on the
pending bits getting cleared.

Tested by forcing all wm(4) interrupts on my test system though an
intr_mask() / softint / intr_unmask() cycle and exercising the network
heavily.

(thorpej)

PR port-sparc/54789: Sparc boot fails with "Trap 0x21 while interrupts disabled"

Remove assumption that cpu_info is smaller than 1kB.

(ad)

Hide ACPI0004 devices on aarch64

(jmcneill)

If a device node is a child of a module device, and the module device declares DMA range restrictions, use them.

(jmcneill)

MD code may allocate a new dma tag for attach args, so destroy it if the device node is not claimed

(jmcneill)

Add bcmmbox at acpi

(jmcneill)

Add BCM2835 EMMC acpi glue

(jmcneill)

Split bcm2835 mbox driver into separate fdt and acpi frontends.

(jmcneill)

pg->phys_addr -> VM_PAGE_TO_PHYS().

(ad)

Add missing newlines in log messages.

(tsutsui)

Whitespace.

(ad)

pagedaemon:

- Use marker pages to keep place in the queue when scanning, rather than
  relying on assumptions.

- In uvmpdpol_balancequeue(), lock the object once instead of twice.

- When draining pools, the situation is getting desperate, but try to avoid
  saturating the system with xcall, lock and interrupt activity by sleeping
  for 1 clock tick if being continually awoken and all pools have been
  cycled through at least once.

- Pause & resume the freelist cache during pool draining.

PR kern/54209: NetBSD 8 large memory performance extremely low
PR kern/54210: NetBSD-8 processes presumably not exiting
PR kern/54727: writing a large file causes unreasonable system behaviour

(ad)

Freelist cache: drain using a high-priority xcall and re-enable now that
the pagedaemon starvation problem should be fixed.

(ad)

uvm_pagealloc_pgb(): don't fill cache if we're into the reserves.

uvm_pagereplace(): use radix_tree_replace_node() to avoid alloc/free.

(ad)

Oops, don't match brcm,bcm2711-emmc2 here.

Thanks to jmcneill for pointing out my mistake

(skrll)

Follow the Linux driver an use the FDT "compatible" property to build a
filename for the nvram config file, fall back to the standard filename.

E.g.

# ofctl -p /
[Caching 123 nodes and 1093 properties]
#address-cells          00000001 ........ ........ ........    1
#size-cells            00000001 ........ ........ ........    1
compatible              73696e6f 766f6970 2c627069 2d6d322d  "sinovoip,bpi-m2-
            0010:      7a65726f 00...... ........ ........  zero"
            0015:      616c6c77 696e6e65 722c7375 6e38692d  "allwinner,sun8i-
            0025:      68322d70 6c757300 ........ ........  h2-plus"
interrupt-parent        00000001 ........ ........ ........  ....
model                  42616e61 6e612050 69204250 492d4d32  "Banana Pi BPI-M2
            0010:      2d5a6572 6f00.... ........ ........  -Zero"
name                    00...... ........ ........ ........  ""
serial-number          30326330 30303432 65636431 36376566  02c00042ecd167ef
            0010:      00...... ........ ........ ........  .

-rw-r--r--  1 root  wheel    875 Nov  2 12:06 brcmfmac43430-sdio.AP6212.txt
lrwxr-xr-x  1 root  wheel      29 Dec 30 16:19 brcmfmac43430-sdio.sinovoip,bpi-m2-zero.txt -> brcmfmac43430-sdio.AP6212.txt
-rw-r--r--  1 root  wheel    874 Jun 30  2019 brcmfmac43430-sdio.raspberrypi,3-model-b.txt
-rw-r--r--  1 root  wheel    1864 Jun 30  2019 brcmfmac43455-sdio.raspberrypi,3-model-b-plus.txt
lrwxr-xr-x  1 root  wheel      29 Dec 30 11:24 brcmfmac43455-sdio.raspberrypi,4-model-b-plus.txt -> brcmfmac43455-sdio.raspberrypi,3-model-b-plus.txt

(mlelstv)

Fix up ARM_LOCAL for 2711 so that it works now.

(skrll)

Invalidate the cache just before trying to read the results from the
firmware as RPI4 A72s can fill the cache with speculation

(skrll)

Drop DEVMAP_{TRUNK_ADDR,ROUND_SIZE} to 4KB pages now that pmap_map_chunk
allows this.

(skrll)

Update pmap_map_chunk to allow L[12] block mappings and L3 page mappings

(skrll)

Add emmc2 clock goop from mlelstv@

(missing commit)

(skrll)

Flush the cache and disable TTBR0 translations once we're done with
them in cpu_kernel_vm_init

The Cortex A72s in RPI4 need the cache flush for some reason.

(skrll)

Remove unnecessary brackets and unwrap a conditional.  Same code before
and after.

(skrll)

Add emmc2 clock goop from mlelstv@

(skrll)

Match brcm,bcm2711-emmc2

(skrll)

Ticket #591

(martin)

Pull up following revision(s) (requested by maya in ticket #591):

external/mpl/bind/dist/lib/isc/unix/app.c: revision 1.5

Only ignore signals if we are bind (not dhcpd).

(martin)

add highly experimental NUMA support for US-IIIi machines

(macallan)

Fix incorrect obsolete package name

(uki)

Give syspkg names to sparc machine dependency files

(uki)

Give syspkg name to sparc64 machine dependency files

(uki)

Use -dash list instead of -tag list without content.

(uwe)

Allow MD code to provide custom bus_dma tags on a per-node basis. On Arm
this is required to return non-coherent bus_dma tags for device nodes with
_CCA=0

(jmcneill)

Add missing '-width' argument to macro.

(wiz)

Make sure rc, rc.subr, and rc.shutdown are properly updated.

Currently there is no info which rc* files should be updated
or not on upgrade (at least rc.conf and rc.local shouldn't),
so put back an explicit list in the postinstall script.
"Go for it" by christos@ in PR/54741.

Should be pulled up to netbsd-9.

(tsutsui)

Simplify some macros: if we only ever use them on the local cpu,
there is no need to do another pointer derefernence to go via the global
address of our cpu info.
Pointed out by ad.

(martin)

urndis.4: add LGE Nexus 5 as a knonw working device.

(triaxx)