Account for the -1 hack to fit 0x10000 in a short in hash_page.c

(christos)

Delete #ifdef __HAVE_PCI_MSI_MSIX and use pci_intr_allo()/pci_intr_release()
on any systems. The third argument of pci_intr_alloc() is left because we will
use the code to support MSI-X in future.

(msaitoh)

Stop passing llinfo_nd6 to nd6_ns_output

This is a restructuring for coming changes to nd6 (replacing
llinfo_nd6 with llentry). Once we have a lock of llinfo_nd6,
we need to pass it to nd6_ns_output with holding the lock.
However, in a function subsequent to nd6_ns_output, the llinfo_nd6
may be looked up, i.e., its lock would be acquired again.
To avoid such a situation, pass only required data (in6_addr) to
nd6_ns_output instead of passing whole llinfo_nd6.

Inspired by FreeBSD

(ozaki-r)

- Add the Auto Slot Power Limit Disable bit in Slot Control register and
  the Completion Timeout Prefix/Header Log Capable bit in the AER capability
  and control register (ECN: Downstream Port Containment (DPC)).
- Add the Poisoned TLP Egress Block bit (ECN: Enhanced DPC).
- Update Link Capabilities 2 register and Link Control 3 register (ECN:
  Separate Refclk Independent SSC Architecture (SRIS))
- ECN: Readiness Notifications (RN)
- Add the Retimer Presence Detect Supported bit in the Link Capabilities 2
  register and the Retimer Presence Detected bit in the Link Status 2 register
  (ECN: Extension Devices)

(msaitoh)

Don't assign unused IP address

It sometimes creates an unexpected NDP cache.

(ozaki-r)

Unify nd6_ns_output calls in nd6_llinfo_timer

Inspired by FreeBSD

(ozaki-r)

If MAX_BSIZE == hashp->BSIZE (65536) then it does not fit in a short, and
we end up storing 0... This means that every entry needs a page. We store
MAX_BSIZE - 1 here, but it would be better to always store (avail - 1) here
so that we don't waste a byte and be consistent.

(christos)

Add PCI Extended Configuration support for tegrapcie(4).

Similar to the acpimcfg code, this only maps the extended configuration
space into KVA for known busses.

(jakllsch)

PR/50441: Manuel Bouyer: hash seq enumeration skips keys on big data.
XXX: pullup-7

(christos)

- ARI's function group is not bit 32-24 but 22-20.
- Add the Structure Length field in AF capability register.
- Add Enhanced Allocation extended capability ID (ECN: Enhanced Allocation (EA)
  for Memory and I/O Resources).
- Add LN System CLS (ECN: Lightweight Notification (LN) Protocol).
- Add ST Upper and Lower bit definitions (ECN: TLP Processiong Hints).
- Add the Global Invalidate bit in the ATS capability register and the PRG
  Response PASID Required bit in the Page Request status register (ECN: PASID
  Translation)
- Decode ASPM support bit more (ECN: ASPM Optionally)
- Use __BITS()

(msaitoh)

No functional change:
- Add comments.
- Remove obsolete comment.
- Move definitions to better location.
- Rename bit definition.
- KNF.
- Indent.

(msaitoh)

Switch PCI athn(4) attachment from pci_intr_map() to
pci_intr_alloc()/pci_intr_release(), this enables MSI where available.

(jakllsch)

Long overdue suggestions from Taylor Campbell and a few syntax/style
tweaks from myself.

From Michael McConville

(macallan)

Revert changes of athn.c r1.8.

- aprint_normal is correct for the attach message
- KNF adjustments in code we sync with upstream are a nuisance
- aprint_error{,_dev}() is inappropriate for use in runtime code

(jakllsch)

Replace SIMPLEQ_FOREACH with SIMPLEQ_FOREACH_SAFE to prevent use-after-free.

(hannken)

Add tests for GC of neighbor caches

(ozaki-r)

Fix argument to empty()

Guess none of the fast-running tests exercised the
component being initialized ...

(pooka)

Do not clear the PCIe interrupt until we've allowed its cause to become
deasserted.  This halves the PCIe interrupt rate.

(jakllsch)

Use IOCONF #define to determine component name.

Avoids having to manually retype info already in .ioconf.

Also, COMPONENT_SIMPLE -> RUMP_COMPONENT=simple.  That way, we can add
other attributes where necessary.  As "future directions", the "ioconf"
attribute for RUMP_COMPONENT can just be removed when a driver is
converted to modular and has a proper modcmd.

(pooka)

initialize allocated gem object memory to 0

(jmcneill)

wscons and Xorg support for CirrusLogic GD542x based boards (grf3).

(phx)

Fixed bug in wsdisplay copyrows textop.
Fixed WSDISPLAYIO_SVIDEO, which had inverted functionality.
Improved FBINFO setup. 15 and 16 bits depth both have 16bpp now.

(phx)

commit window changes when updating base

(jmcneill)

use XRGB8888 for framebuffer

(jmcneill)

PR/50434: Louis Guillaume: Bump the in and out formats to 511 (NAME_MAX)

(christos)

- update NAME_MAX to match the kernel.
- add a comment about where the constant is used.

(christos)

Add Braswell PCU SMBus.

(msaitoh)

Regen.

(msaitoh)

Change macro name of Braswell PCU SMBus.

(msaitoh)

Regen.

(msaitoh)

Add Braswell devices from "Table 23-4 PCI Devices and Functions,
N-Series Intel Pentium Processorts and Intel Celeron Processors Datasheet
(Document Number: 332093-001)"

(msaitoh)

Remove USB HID entries, they're already in usbdevices.config

(bouyer)

correctly print the gpt header revision as a "float"

(jnemeth)

Define PCIE_XCAP_{VER,TYPE}(x) and use them.

(msaitoh)

Add missing rtfree

(ozaki-r)

note cpuctl and multiple-cpu support.

(mrg)

allow most commands to specify more than one cpu.  now you can online or
offline (or identify, or intr/nointr) a list of cpus all together.

(mrg)

convert getcpuid() to take char* not char**

(mrg)

use stdbool.h

(mrg)

disable data island packet transmission on Tegra124 HDMI codec

(jmcneill)

Avoid NULL dereference in linux_worker_intr() in WORK_DELAYED_CANCELLED case.

Potentially fixes PR#49560.

(jakllsch)

Add genfb1, awindebe1, and awintcon1. These will attach if the fex
script request them.
While there add USB HID, keyboard, mouse and touchscreen.
If you have a display you'll most proabbly want input devices too.

(bouyer)

Add support for LVDS output on tcon0. Tested with a full HD LVDS display.
Other LCD interfaces not supported yet, mostly by lack of hardware.
The LVDS output and timings are set from the FEX script.

(bouyer)

PR 32999: adjust VM parameters for 64 bit kernel/userlands.
Limit old binaries (i.e. processes not using topdown VA layout) to 2 GB
data size.
Use the same limits as native sparc for netbsd32 emulation - we need to
revisit this, but currently automatic test runs in this environment
fail big time, so there is no easy way to verify any changes.

(martin)

Antti is Kantee

(plunky)

Always descend into component directories for rumpdescribe.

(pooka)

Fix typo which caused the kenrel thread to be created with a 0 priority.
This would cause the thread to be almost never scheduled when a userland
process could use all CPU.
Should fix the problem reported by Torbj�rn Granlund on port-xen@

(bouyer)

If we are going to be doing read/modify/write in triple-buffered mode, make sure we don't read back the buffered values and not the active ones

(jmcneill)

describe => rumpdescribe

(pooka)

Correct the x68k's Backspace key behavior.
With this fix, the Backspace key on x68k keyboard erases
previous one character, instead of outputting '^?' character.
Reported by Yasushi Oshima (two months ago..)

(isaki)

Add the newly-created wapbl module to the sets list.

(pgoyette)

Actually create the wapbl module, since it works.

(pgoyette)

Enable the module's MODULE_CMD_FINI action.  It actually works as
intended.

(pgoyette)

If file system ffs is built with WAPBL defined, make sure that the
module depends on the wapbl module.

No impact to users of built-in ffs file system code, as the WAPBL
#define will cause inclusion of the code in the kernel.

A standard build of the modular ffs file system code will #define
WAPBL, so the module will only work on a kernel which was also
built with WAPBL defined (or, once I commit it, with a dynamically-
loaded wapbl module).

(pgoyette)

set vbus-gpio property on tegrausbphy, not ehci

(jmcneill)

tegracec can get physical address from tegradrm now

(jmcneill)

expose HDMI from HDMI VSDB as physical-address device property when available

(jmcneill)

parse_hdmi_vsdb: read the HDMI physical address from the HDMI VSDB and store it on the connector

(jmcneill)

sleep before asking the PMIC to power off

(jmcneill)

Enable installation of amd64 on an lfsv2 partition.

XXX Other ports will need similar changes to arch/.../md.h after testing!

XXX The mount_lfs warning about lfs being experimental may confuse or
XXX corrupt the sysinst display window.

(pgoyette)

Remove historic references to wapbl.

(pgoyette)

hdmi_audio_infoframe_pack: Initialize the entire output buffer.
hdmi_infoframe_checksum: Fix HDMI infoframe checksum calculations.

(jmcneill)

refresh

(christos)

KNF, and fix some comments

(maxv)

PR/50428: Rin Okuyama: support SA_RESTORER flag for rt_sigaction(2) in the
ports that support it.

(christos)

support video=<mode> kernel cmdline for setting HDMI-A-1 preferred video mode

(jmcneill)

Support command-line modes by picking up connector modes from the device
properties. The connector name is the key name in the device properties
dictionary.

(jmcneill)

Merge upstream commit 91571cce03adc58a10b5d76307629efc7c994c3e

Subject: modesetting: Don't pretend to support rotation

It only ends in tears

Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

(jmcneill)

Hardware cursor support.

(jmcneill)

Wait 20ms for the PLL to stabilize after enabling it.
Fixes occasional bogus size report for my SD card.

(bouyer)

Always include the "vfs" attribute (module).  Although all of the
ufs/xxx file systems depend on the vfs attribute, it is not required
that any file system actually be built-in to the kernel.  (At least
on some architectures, file system modules can be loaded at boot
time.)

(pgoyette)

SNAPSHOT: DOES NOT BOOT

First crack at configuration files for the Hardkernel ODROID XU4.

This kernel hangs attempting to write to the console early in boot.

(marty)

Jetson TK1 u-boot sets up PCI IO space in an impossible-to-use
configuration.  As we're already allocating resources on the PCI
bus, set up our own mapping of PCI address spaces into the ARM
address space.  We rely on a potential overlap of address space
windows to allow us to use the same bus_space_tag for PCI Memory
and IO spaces.

The PCI attachment of the onboard re(4) uses PCI IO space in
preference to PCI Memory space for register accessses.  As IO space
was impossible to use, we had to avoid IO space.  This is now no
longer the case, so set up and enable IO space for PCI devices.

Also, map ROM BARs.

(jakllsch)

Fix obvious typo - even though it is inside a #ifdef notyet ... #endif

(pgoyette)

Correct TEGRA_PCIE_A[123] window definitions.  Replace existing
usages thereof (and related bus space handles, etc.) with more
appropriate names.

(jakllsch)

Increment TEGRA_PCIE_SIZE

It's a size, not a last-valid-offset.

(jakllsch)

add more AFI BAR register offsets

(jakllsch)

We don't need to use any particular value, but use the actual interrupt
number for the PCI_INTERRUPT_LINE.

(jakllsch)

PR/50426: Dave Tyson: Fix ntp reported version.

(christos)

redo previous, not needed.

(christos)

Limit configuration space access to the non-emulated busses to legacy
PCI_CONF_SIZE.  Extended configuration access is possible, we just have
to implement it without wasting up to 256MiB of KVA.

(jakllsch)

validate emulated bus 0 configuration space access more carefully

(jakllsch)

drop white space on empty line

(jakllsch)

Handle 0b locally since anyway this is not portable.

(christos)

Make sure the graphics device which has the console gets wsdisplay0,
otherwise neither virtual terminals nor Xorg's wsfb driver will work.

(phx)

comment out 0{b,B} handling.

(christos)

revert 0b documentation.

(christos)

Generalize the test, compare on the protocol semantics instead of the name
since it could be udp6...

(christos)

Do proper accounting for the extra -1 slot. Perhaps this is too confusing
and it would be better to just access the array with [fd + 1] instead?

(christos)

Fix bug in readdir loop condition.

Reading all dirents using a small buffer and multiple calls now works.

Bug found by "Shamar" on #rumpkernel

(pooka)

Remove now unnecessary cast.

(tron)

Don't try to use listen(2) on a UDP socket which will always fail.
Previously this was not a problem because the return value of listen(2)
was ignored. With this fix amd(8) no longer fails to start with the
error message "cannot create rpc/udp service".

TL;DR: Make amd(8) work again

(tron)

Avoid broken state if realloc(3) fails.

(tron)

Family is singular, so use singular form of verb.

(wiz)

Bump date for previous, remove empty line.

(wiz)

Remove trailing whitespace.

(wiz)

Tweak the KASSERTs in pmap_update in the ARM_MMU_EXTENDED case - Lazy
updating is not done on the kernel pmap.

PR port-arm/50420: curcpu()->ci_pmap_cur != pm || pm->pm_pai[0].pai_asid == curcpu()->ci_pmap_asid_cur" failed

(skrll)

Wrap long lines.

(skrll)

Fix register offset to print HyperTransport registers correctly.

(msaitoh)

remove extra semicolons.

(christos)

regen

(christos)

Declare variables under the same conditions as their use.

(riz)

update version numbers too

(agc)

Update sid to version 20151014

+ move to one output format, just punctuate the "human" output with
newline characters.  Note that jq can parse the "human" output just
fine.

+ escape the pattern to ignore when printing (this produces valid
JSON should it be wanted)

+ add ftype to the RegFile definition in the example config file

+ remove the runtime timestamp from the JSON output

(agc)

put back the old ones until the rest is generated

(christos)

Regen; it's been a *long* while.

(christos)

PR/50422: Robert Elz: df -G prints the wrong value for fragsize (+FIX)
For df -G, print the block and fragment size instead of the iosize
and the blocksize. If we need to print the iosize, it should be done
in a different field. Nevertheless printing the blocksize in the fragment
size field is just wrong.
XXX: pullup-6, pullup-7

(christos)

consistency fixes.

(christos)

Add sandpoint ddb-entry method.

(phx)

Recognize binary constants

(christos)

Fix capitalization

(christos)

Now that our source tree has binary constants, recognize them.

(christos)

Recognize 0[bB] as binary (base 2)

(christos)

Add usage function, various cosmetics; no functional change.

(christos)

Use the new DK_DEV_BSIZE_OK() macro.

(christos)

Add DK_DEV_BSIZE_OK()

(christos)

Amiga Xorg support.

(phx)

Ensure that the callout of the multicast address is valid before
hooking it up.

(joerg)

In ioconfname mode, #define IOCONF as the ioconf token.

Avoids having to retype the name to call config_init/fini_component().

(pooka)

Use the MD_BTOP() macro as replacement for m68k_btop() in all device
drivers which are shared with amigappc.

(phx)

- Restore pci_subr.c rev. 1.135's change in pci_conf_print_caplist().
  As wrote in the comment, HyperTransport capability appears multiple times.
  pci_conf_cap() reruns only the first entry, so it can't be used here.
- Try to decode HyperTransport capability. Currently, the capability type
  of each HyperTransport capability is printed and only the MSI Mapping
  capability is decoded.
- Style change.

(msaitoh)

PCI_HT_CAP() is right shifted value, so PCI_HT_CAP_* should not use
__SHIFTIN(). This change fixes a bug that HyperTransport system misunderstand
whether MSI/MSI-X can be used or not.

(msaitoh)

Scancode translation table for wskbd on Amiga keyboards.

(phx)

Xorg for amiga via wsfb. Currently available on CV64 and CV64/3D. More
graphics cards will follow, when adding wscons support to them.

I didn't manage to make the "wskbd" protocol the default in the X server,
so you have to provide a small xorg.conf with Option "Protocol" "wskbd".
The standard protocol will not work as the Amiga wskbd cannot be switched
into raw mode and has no AT-scancode translation in the kernel.

On the positive side, wskbd also works with national keymaps under X.

The graphics modes can still be defined via grfconfig(8). Xorg's wsfb
driver uses the current grf video mode, which is usually mode #1.
All resolutions and depths seem to work (tested on CV64).

(phx)

Take the vnode lock before the vnode is marked VI_CHANGING and fed
to vclean().  Prevents a deadlock with two null mounts on the same
physical mount where one thread tries to vclean() a layer node and
another thread tries to vget() a layer node pointing to the same
physical node.

Fixes PR kern/50375 layerfs (nullfs) locking problem leading to livelock

(hannken)

change some register dumps from aprint_verbose to aprint_debug

(jmcneill)

enable NFSSERVER (mostly to silence module warning)

(jmcneill)

Drop to spl0 after the pmap_activate call.  Should address

PR port-arm/50420: curcpu()->ci_pmap_cur != pm || pm->pm_pai[0].pai_asid == curcpu()->ci_pmap_asid_cur" failed

Discussed with matt@ and rmind@

(skrll)

Standard mode for I2C needs a source divider of 20, not 2.

(jmcneill)

Add tests of IPv6 link local address

From s-yamaguchi@IIJ

(ozaki-r)

Fix up the header

Remove unnecessary shebang and add missing keyword expansion,
copyright and license.

(ozaki-r)

fix incorrect memset.

(christos)

Use GEM for memory management. Fixes a couple issues while here:
- No longer needs to allocate 35MB (!) for framebuffer console.
- Allows xrandr to switch to modes larger than the framebuffer console.
- Removes hack that redirected mmap calls to wsdisplay0

(jmcneill)

Make it easier to create rump kernel components, part 1.

Reduce copypasteware for the component constructors.  If a constructor
calls only config_init_component(), handle it from a common source file
instead of copying the same(ish) file around to every component.

(pooka)

Support pmap_pv_track and friends

(skrll)

not used by tegradrm

(jmcneill)

Add tty2 tty3 devices to evbarm MAKEDEV; JETSONTK1 has console on com3.

(jakllsch)

Update loadbsd to 2.17, which can load ELF kernels.
Remove obsolete loadbsd binaries from installation/misc as they
have become useless (and can only load a.out kernels).

(phx)

Build instructions and Makefile for loadbsd 3.0.
Submitted by Gunther Nikl.

(phx)

fix RB_POWERDOWN test

(jmcneill)

enable cd(4) in JETSONTK1

(jakllsch)

make VDD_CPU programming a bit easier to understand, and while here, actually program it to 1.4V as intended instead of 1.39V

(jmcneill)

support RB_POWERDOWN using the AMS AS3722 PMIC when available

(jmcneill)

add support for optional RB_POWERDOWN handler

(jmcneill)

enable as3722pmic

(jmcneill)

Add basic driver for AMS AS3722 power management IC

(jmcneill)

only send repeat start if we are doing write-then-read

(jmcneill)

fix i2c periph clock, send repeat start for write-then-read xfers

(jmcneill)

Fix XEN build - does XEN really want/need pmap_pv_track?

(skrll)

Remove #if 0 / #endif includes

(skrll)

Split out the pmap_pv_track stuff for use by others.

Discussed with riastradh@

(skrll)

Add tests for RA

From s-yamaguchi@IIJ (with some tweaks by me)

(ozaki-r)

Introduce rump.rtadvd

It is used to write ATF tests for RA.

From s-yamaguchi@IIJ.

(ozaki-r)

Introduce rump.rtadvd

It is used to write ATF tests for RA.

From s-yamaguchi@IIJ.

(ozaki-r)

add an entry for netbsd-7-0

(snj)

fix CID 980463

(knakahara)

Note that the netbsd-5{,-1,-2} branches have been terminated.

Goodbye, dear friends.

(snj)

fix panic after "ifconfig gifX tunnel src dst" failed for the reason of address pair duplication.

e.g.
    ====================
    # ifconfig gif0 create
    # ifconfig gif0 tunnel 192.168.0.1 192.168.0.2
    # ifconfig gif0 inet 172.16.0.1/24 172.16.0.2
    # route add 10.1.0.0/24 172.16.0.1

    # ifconfig gif1 create
    # ifconfig gif1 tunnel 192.168.0.1 192.168.0.3

    # ifconfig gif0 tunnel 192.168.0.1 192.168.0.3
    ifconfig: SIOCSLIFPHYADDR: Can't assign requested address # expected
    # ping 10.1.0.1
    (panic)
    ====================

(knakahara)

Only sync and unload dma map in ahci_atapi_complete() if appropriate.

Fixes ahcisata atapi cd(4) on arm (specifically, JETSONTK1), which insists
that a zero-length dmamap can not be synced.

(jakllsch)

Fix typo that prevented CONADDR from being defined for XU4

(marty)

use NetBSD-inv.xpm for CTWM_WELCOME_FILE

(jmcneill)

if CTWM_WELCOME_FILE is defined, try that before falling back to welcome.xwd and .xpm

(jmcneill)

small cleanup of odroid_start.S

removed attempt to 'rescue bootargs' that was causing a hang and isn't needed
anyway.

changed calls to XPUTC to use text constants rather than ascii decimal
equivalents.

adopted some of the techniques used in awin_start.s for reducing code and
improving performance.

polluted the namespace with _XPUTC so that xputc can be used from C code
before early console is enabled.  (Due to tracking a bug in early console
initialization.)

replaced a hardwired constant with the proper mannifest in the definition
of TEMP_L1_TABLE

This still needs MP support added but it should continue to work for XU3
while starting to work for XU4

(marty)

fix default pixmap search directory

(jmcneill)

Add vblank support

(jmcneill)

fix compilation/lint

(christos)

Apply patch by Jouni Malinen. We don't have CONFIG_EAP_PWD enabled so we are
not affected:

EAP-pwd peer error path failure on unexpected Confirm message

Published: November 10, 2015
Identifier: CVE-2015-5316
Latest version available from: http://w1.fi/security/2015-8/

Vulnerability

A vulnerability was found in EAP-pwd peer implementation used in
wpa_supplicant. If an EAP-pwd Confirm message is received unexpectedly
before the Identity exchange, the error path processing ended up
dereferencing a NULL pointer and terminating the process.

For wpa_supplicant with EAP-pwd enabled in a network configuration
profile, this could allow a denial of service attack by an attacker
within radio range.

Vulnerable versions/configurations

wpa_supplicant v2.3-v2.5 with CONFIG_EAP_PWD=y in the build
configuration (wpa_supplicant/.config) and EAP-pwd enabled in a network
profile at runtime.

Possible mitigation steps

- Merge the following commits and rebuild wpa_supplicant:

  EAP-pwd peer: Fix error path for unexpected Confirm message

  This patch is available from http://w1.fi/security/2015-8/

- Update to wpa_supplicant v2.6 or newer, once available

- Remove CONFIG_EAP_PWD=y from build configuration

- Disable EAP-pwd in runtime configuration

(christos)

correct mistake in previous

(christos)

CID 1338516: Handle svc_run() returning.

(christos)

CID 1338520: Check NULL
CID 1338521: Fix error (realloc returns different pointer)

(christos)

CID 1338517: Check negative returns

(christos)

CID 1338515: Make it clear that the pfd variable can't be NULL

(christos)

CID 1338514: Check NULL returns

(christos)

CID 1338513: Check listen(2) return

(christos)

CID 980463: Provide common error path for rollback. Remove extra check for
success.

(christos)

Add "pseudo-device iscsi" to ALL and as a commented out hint in XEN3_*

(tnn)

fix iSCSI RFC number

(tnn)

support disabling the HDMI output

(jmcneill)

support disabling the HDMI output

(jmcneill)

allow either crtc to be used with hdmi encoder

(jmcneill)

If the "is_console" property is already specified, do not overwrite it.

(jmcneill)

Port the Tegra (2D) display drivers to the DRM framework.

tegradrm0 at tegraio0
tegrafb0 at tegradrm0
tegrafb0: framebuffer at 0x9b000000, size 1280x720, depth 32, stride 5120
wsdisplay0 at tegrafb0 kbdmux 1
wsmux1: connecting to wsdisplay0
wsdisplay0: screen 0-3 added (default, vt100 emulation)
tegradrm0: info: registered panic notifier
tegradrm0: initialized tegra 0.1.0 20151108 on minor 0

Same features as before (fb console, X wsfb driver works) with the addition
of being able to use xf86-video-modesetting and xrandr to switch video
modes at runtime.

(jmcneill)

only clear agp hook on last close when agp is in use, from riastradh@

(jmcneill)

Use PRIx64 and %# requested by phone@

(christos)

explain why the int cast works (suggested by kre)

(christos)

Return ENXIO if the get ioctl exceeds the number of configured devices.
XXX: pullup-7

(christos)