openwrt/staging/jogo.git
7 years agoar71xx: Detect USB port on Mikrotik RB750UP
João Chaínho [Fri, 13 Jan 2017 16:23:09 +0000 (16:23 +0000)]
ar71xx: Detect USB port on Mikrotik RB750UP

The USB port on Mikrotik RB750UP isn’t detected. This patch provides a fix. Tested and working.

Signed-off-by: João Chaínho <joaochainho@gmail.com>
7 years agoramips: Add I2C driver to the default kernel config
Rosen Penev [Fri, 13 Jan 2017 15:14:38 +0000 (07:14 -0800)]
ramips: Add I2C driver to the default kernel config

I made a commit that added the RTC driver to the kernel config with
the intent that it would fix hctosys. Unfortunately while the RTC
driver is in there, it's connected through I2C, the driver for which
comes in module form and is thus loaded late. After this commit, it
works fine.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 years agoimx6: kernel: Backport serial port fixes
Petr Štetiar [Fri, 13 Jan 2017 10:42:08 +0000 (11:42 +0100)]
imx6: kernel: Backport serial port fixes

Signed-off-by: Petr Štetiar <ynezz@true.cz>
7 years agokernel: backport MIPS changes introducing a separate IRQ stack
Felix Fietkau [Wed, 11 Jan 2017 09:57:49 +0000 (10:57 +0100)]
kernel: backport MIPS changes introducing a separate IRQ stack

Prevents crashes when IRQs arrive when the current kernel stack context
already contains deeply nested function calls, e.g. when stacking lots
of network devices on top of each other

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agogitignore: add /overlay
Matthias Schiffer [Sun, 15 Jan 2017 17:16:29 +0000 (18:16 +0100)]
gitignore: add /overlay

Similar to /files, /overlay is only useful for modifications not tracked in
the LEDE repository itself.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoath10k-firmware: Update QCA988X firmware to latest version
Timo Sigurdsson [Wed, 11 Jan 2017 21:49:29 +0000 (22:49 +0100)]
ath10k-firmware: Update QCA988X firmware to latest version

This patch updates the QCA988X firmware to the latest revision
  firmware-5.bin_10.2.4-1.0-00016
found in the official ath10k-firmware repository.

Tested on TP-Link Archer C7 v2.

Signed-off-by: Timo Sigurdsson <public_timo.s@silentcreek.de>
7 years agohostapd: fix passing jobserver to hostapd/supplicant build processes
Felix Fietkau [Sun, 15 Jan 2017 13:56:34 +0000 (14:56 +0100)]
hostapd: fix passing jobserver to hostapd/supplicant build processes

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agohostapd: backport a few upstream fixes
Felix Fietkau [Sun, 15 Jan 2017 13:19:43 +0000 (14:19 +0100)]
hostapd: backport a few upstream fixes

Fixes reassoc issues with WDS mode
Fixes reassoc issues in AP mode
Fixes IBSS reauthentication issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomt76: update to the latest version, adds support for 802.11w
Felix Fietkau [Sat, 14 Jan 2017 22:09:34 +0000 (23:09 +0100)]
mt76: update to the latest version, adds support for 802.11w

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoncurses: revert $(STAGING_DIR_HOSTPKG) to $(STAGING_DIR)/host where appropriate
Matthias Schiffer [Sat, 14 Jan 2017 17:11:26 +0000 (18:11 +0100)]
ncurses: revert $(STAGING_DIR_HOSTPKG) to $(STAGING_DIR)/host where appropriate

Host files installed in Build/InstallDev are target-specific and will stay
in $(STAGING_DIR)/host after the STAGING_DIR_HOSTPKG unification.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoinclude/autotools.mk: use STAGING_DIR_HOSTPKG where appropriate
Matthias Schiffer [Sat, 14 Jan 2017 17:09:33 +0000 (18:09 +0100)]
include/autotools.mk: use STAGING_DIR_HOSTPKG where appropriate

Make sure binaries install to STAGING_DIR_HOSTPKG are still found when
this variable is eventually modified.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agomac80211: backport a fix for a tx related race condition
Felix Fietkau [Sat, 14 Jan 2017 17:02:54 +0000 (18:02 +0100)]
mac80211: backport a fix for a tx related race condition

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: refresh patches
Felix Fietkau [Fri, 13 Jan 2017 09:42:35 +0000 (10:42 +0100)]
mac80211: refresh patches

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: backport some upstream fixes
Koen Vandeputte [Fri, 13 Jan 2017 09:34:52 +0000 (10:34 +0100)]
mac80211: backport some upstream fixes

Backports the following upstream fixes:

mac80211: initialize fast-xmit 'info' later
mac80211: fix legacy and invalid rx-rate report
mac80211: fix tid_agg_rx NULL dereference

Compiled and tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
7 years agokernel: update 4.4 kernel to 4.4.42
Hauke Mehrtens [Fri, 13 Jan 2017 21:35:45 +0000 (22:35 +0100)]
kernel: update 4.4 kernel to 4.4.42

Refresh patches on all 4.4 supported platforms.
Compile & run tested: lantiq/xrx200

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agoimagebuilder: remove existing debug kernel image
Felix Fietkau [Fri, 13 Jan 2017 18:22:09 +0000 (19:22 +0100)]
imagebuilder: remove existing debug kernel image

Reduces tarball size and improves build time

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoimagebuilder: remove existing root filesystem images
Felix Fietkau [Fri, 13 Jan 2017 18:21:31 +0000 (19:21 +0100)]
imagebuilder: remove existing root filesystem images

Reduces tarball size and improves build time

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoxburst: remove hack to determine entry point
Felix Fietkau [Fri, 13 Jan 2017 18:20:57 +0000 (19:20 +0100)]
xburst: remove hack to determine entry point

xburst has CONFIG_BOOT_RAW set, so jumping to the load address is valid

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomalta: remove hack to determine entry point
Felix Fietkau [Fri, 13 Jan 2017 18:19:34 +0000 (19:19 +0100)]
malta: remove hack to determine entry point

Malta has CONFIG_BOOT_RAW set, so jumping to the load address is valid

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agosdk: exclude locale files to save some space
Felix Fietkau [Fri, 13 Jan 2017 17:00:18 +0000 (18:00 +0100)]
sdk: exclude locale files to save some space

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agofirewall: fix forwarding local subnet traffic
Jo-Philipp Wich [Fri, 13 Jan 2017 17:31:08 +0000 (18:31 +0100)]
firewall: fix forwarding local subnet traffic

Packets which are merely forwarded by the router and which are neither
involved in any DNAT/SNAT nor originate locally, are considered INVALID
from a conntrack point of view, causing them to get dropped in the
zone_*_dest_ACCEPT chains, since those only allow stream with state NEW
or UNTRACKED.

Remove the ctstate restriction on dest accept chains to properly pass-
through unrelated 3rd party traffic.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agomvebu: simplify etc/board.d/02_network
Paul Wassi [Fri, 28 Oct 2016 18:22:22 +0000 (20:22 +0200)]
mvebu: simplify etc/board.d/02_network

Unify switch configuration on Linksys WRTxx00AC series.
LAN = eth0, WAN = eth1

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
[Álvaro]: also change WAN LEDs
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
7 years agomvsw61xx: add support for MV88E6352
Imre Kaloz [Thu, 12 Jan 2017 21:50:06 +0000 (22:50 +0100)]
mvsw61xx: add support for MV88E6352

MV88E6352 is used on Linksys WRT3200ACM

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
7 years agomvebu: several fixes for Linksys WRT3200ACM
Álvaro Fernández Rojas [Wed, 11 Jan 2017 17:43:38 +0000 (18:43 +0100)]
mvebu: several fixes for Linksys WRT3200ACM

- Fix WLAN LEDs definition.
- Fix USB pinctrl definition.
- Remove SDHCI definition.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
7 years agomxs: gzip ext4 images
Felix Fietkau [Fri, 13 Jan 2017 14:23:36 +0000 (15:23 +0100)]
mxs: gzip ext4 images

Significantly reduces disk usage

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarc770: gzip rootfs image to save some space
Felix Fietkau [Fri, 13 Jan 2017 15:22:13 +0000 (16:22 +0100)]
arc770: gzip rootfs image to save some space

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarchs38: gzip rootfs image to save some space
Felix Fietkau [Fri, 13 Jan 2017 15:20:56 +0000 (16:20 +0100)]
archs38: gzip rootfs image to save some space

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoimage.mk: add generic function for gzipping images if enabled
Felix Fietkau [Fri, 13 Jan 2017 15:19:00 +0000 (16:19 +0100)]
image.mk: add generic function for gzipping images if enabled

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomvebu: remove the clearfog-bundle
Felix Fietkau [Fri, 13 Jan 2017 15:02:52 +0000 (16:02 +0100)]
mvebu: remove the clearfog-bundle

It is obsoleted by proper SD card image support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoimage: when using the new image build code, gzip ext4 images by default
Felix Fietkau [Fri, 13 Jan 2017 14:58:37 +0000 (15:58 +0100)]
image: when using the new image build code, gzip ext4 images by default

This reduces the amount of hacks in the makefile code.

Remove the apm821xx code to do the same - it was broken and left both
compressed and uncompressed images in $(BIN_DIR)

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodnsmasq: use ubus signalling in ntp hotplug script
Kevin Darbyshire-Bryant [Thu, 12 Jan 2017 19:28:53 +0000 (19:28 +0000)]
dnsmasq: use ubus signalling in ntp hotplug script

Use ubus process signalling instead of 'kill pidof dnsmasq' for
SIGHUP signalling to dnsmasq when ntp says time is valid.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agox86: disable crashlog
Felix Fietkau [Fri, 13 Jan 2017 13:48:25 +0000 (14:48 +0100)]
x86: disable crashlog

It could cause crashes with some forms of virtualization, and it is
unlikely to work properly with most systems.
It's safer to just disable it.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86: disable a workaround for a buggy glibc version
Felix Fietkau [Fri, 13 Jan 2017 13:47:40 +0000 (14:47 +0100)]
x86: disable a workaround for a buggy glibc version

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoocteon: disable ext4 images
Felix Fietkau [Fri, 13 Jan 2017 13:03:01 +0000 (14:03 +0100)]
octeon: disable ext4 images

They're pretty large and squashfs images tend to be more useful anyway

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agonetifd: update to the latest version
Felix Fietkau [Fri, 13 Jan 2017 12:07:33 +0000 (13:07 +0100)]
netifd: update to the latest version

This disables IGMP snooping by default, which was causing various issues
over time, like FS#95

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoimage.mk: use LINUX_KARCH rather than ARCH for mkits
Ian Pozella [Mon, 26 Sep 2016 14:06:10 +0000 (15:06 +0100)]
image.mk: use LINUX_KARCH rather than ARCH for mkits

The generated 'its' is passed to mkimage which expects linux arch
strings rather than the full arch (e.g. mips not mipsel).

It currently works in some cases where LINUX_KARCH == ARCH but
otherwise you get an unknown arch build error.

Signed-off-by: Ian Pozella <Ian.Pozella@imgtec.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agouqmi: mark as nonshared because of the usb dependencies
Felix Fietkau [Fri, 13 Jan 2017 11:07:53 +0000 (12:07 +0100)]
uqmi: mark as nonshared because of the usb dependencies

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoumbim: mark as nonshared because of the usb dependencies
Felix Fietkau [Fri, 13 Jan 2017 11:07:32 +0000 (12:07 +0100)]
umbim: mark as nonshared because of the usb dependencies

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agocomgt: mark as nonshared because of the usb dependencies
Felix Fietkau [Fri, 13 Jan 2017 11:07:06 +0000 (12:07 +0100)]
comgt: mark as nonshared because of the usb dependencies

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agopolarssl: remove package
Felix Fietkau [Fri, 13 Jan 2017 10:39:56 +0000 (11:39 +0100)]
polarssl: remove package

The mbedTLS 1.3 branch has been EOL since end of 2016 and now all
remaining users have been converted.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarmvirt: add kernel config change missing from 0d44f0cb
Felix Fietkau [Fri, 13 Jan 2017 09:56:29 +0000 (10:56 +0100)]
armvirt: add kernel config change missing from 0d44f0cb

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: default to external USB power on RB-912UAG
Felix Fietkau [Thu, 12 Jan 2017 14:16:02 +0000 (15:16 +0100)]
ar71xx: default to external USB power on RB-912UAG

There's a GPIO that can switch between power to USB pins for the
internal MiniPCIe slot or the external USB port.
People are more likely to use the external one so enable it by default.
Existing configurations should be unaffected

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: let make check warn about use of legacy PKG_MD5SUM variable in feeds
Felix Fietkau [Thu, 12 Jan 2017 14:18:11 +0000 (15:18 +0100)]
build: let make check warn about use of legacy PKG_MD5SUM variable in feeds

The variable rename change has been merged in OpenWrt now, so it's
possible to convert the feeds as well.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agocmake: properly pass host cflags/ldflags to the build
Felix Fietkau [Thu, 12 Jan 2017 13:41:01 +0000 (14:41 +0100)]
cmake: properly pass host cflags/ldflags to the build

This should hopefully fix build issues where libraries that we ship in
tools/ were accidentally picked up from other places on the system, e.g.
/usr/local

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agocmake: support verbose build that shows compiler commands
Felix Fietkau [Thu, 12 Jan 2017 13:32:02 +0000 (14:32 +0100)]
cmake: support verbose build that shows compiler commands

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agocmake: restore parallel build support for bootstrap
Felix Fietkau [Thu, 12 Jan 2017 13:31:40 +0000 (14:31 +0100)]
cmake: restore parallel build support for bootstrap

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86/64: enable the fusion scsi driver
Felix Fietkau [Wed, 11 Jan 2017 22:35:37 +0000 (23:35 +0100)]
x86/64: enable the fusion scsi driver

Is it used by VMware Fusion by default. This allows images to boot
without further config changes in VMware.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: fix broken spatial multiplexing defaults
Felix Fietkau [Wed, 11 Jan 2017 22:35:19 +0000 (23:35 +0100)]
mac80211: fix broken spatial multiplexing defaults

Most mac80211 drivers leave the SMPS field in the HT capabilities
uninitialized (unfortunately defaults to static SMPS), which leads to
some devices limiting themselves to single-stream rates in some modes
(mostly mesh and IBSS).

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k-fw: Update to latest CT firmware
Ben Greear [Tue, 10 Jan 2017 23:49:42 +0000 (15:49 -0800)]
ath10k-fw: Update to latest CT firmware

For 988X, 9980, 9984 CT firmware.

This should allow IBSS + RSN on at least the 988X firmware,
and includes recent stability fixes for all firmware.

Signed-off-by: Ben Greear <greearb@candelatech.com>
7 years agoath10k-ct: Update to latest CT 4.7 ath10k driver.
Ben Greear [Tue, 10 Jan 2017 23:49:41 +0000 (15:49 -0800)]
ath10k-ct: Update to latest CT 4.7 ath10k driver.

This at least makes it harder to hit some txq related
crashes on firmware restart, a potential memory leak,
and some other fixes.

Signed-off-by: Ben Greear <greearb@candelatech.com>
7 years agoarmvirt: enable the USB feature flag
Felix Fietkau [Fri, 13 Jan 2017 08:15:03 +0000 (09:15 +0100)]
armvirt: enable the USB feature flag

USB devices can be passed through from the host to the guest

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomountd: drop USB related dependencies
Felix Fietkau [Fri, 13 Jan 2017 08:10:47 +0000 (09:10 +0100)]
mountd: drop USB related dependencies

That way we can avoid making it nonshared

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoramips: fix WLI-TX4-AG300N boot and network
Yo Abe [Wed, 11 Jan 2017 18:07:14 +0000 (03:07 +0900)]
ramips: fix WLI-TX4-AG300N boot and network

Partition label "linux" prevents the root file system to be mounted at
boot time leading to a kernel panic. After changing it to "firmware",
the 2 uimage partitions "kernel", "rootfs" and squashfs "rootfs_data"
are correctly recognized.

The attached IP175C 10/100 MBit switch cannot connect to a link with
fixed 1000Mbit speed. The correct link speed is 100MBit. The switch
is detected and can be configured via mdio bus and should allow two
separable VLANs to be configured for the 4 available ports.

Signed-off-by: Yo Abe <abe.geel@gmail.com>
[picked from openwrt/PR#330]
Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoramips: add support for VoCore2
Andrew Yong [Sun, 8 Jan 2017 06:30:10 +0000 (14:30 +0800)]
ramips: add support for VoCore2

This patch adds support for the VoCore VoCore2 and its complementary
"ultimate" dock.

Specifications:
- SoC: MediaTek MT7628AN (580MHz, ramips)
- RAM: 128MB DDR2 166MHz
- Storage: 16MB NOR SPI flash onboard + microSD slot on dock
- Wireless: Built into MT7628AN (mt76) with 1T1R firmware on VoCore2
            boards with onboard 1x chip antenna
- Ethernet: 1x100M (port0) on dock, 1x100M (port2) on PCB header
- Dock hardware:
  - USB 2.0 socket
  - MicroSD socket
  - 100Mbps Ethernet x1
  - 3.5mm headphone jack (TRRS) connected to Everest Semi ES8388 I2S
    DAC/ADC (support WIP)
  - Micro USB for power and console (UART2)

Initial installation:
- VoCore2 comes preinstalled with a fork of OpenWrt CC and AP on
  SSID "VoCore2"
- Connect to VoCore2 by Ethernet or Wi-Fi
- `ssh root@192.168.1.1` (password is "vocore")
- scp/wget/etc. LEDE sysupgrade.bin to VoCore2
- `sysupgrade -n <your image>.bin` (don't keep old config, as the
  original firmware uses Ralink SDK Wi-Fi drivers and not
  mt76+mac80211)
- after sysupgrade completes, Wi-Fi will be disabled by default so use
  Ethernet or the micro USB console to configure Wi-Fi again

Signed-off-by: Andrew Yong <me@ndoo.sg>
7 years agoar71xx: enable serial console on Mikrotik RB411/RB433
David Varga [Tue, 10 Jan 2017 18:12:34 +0000 (19:12 +0100)]
ar71xx: enable serial console on Mikrotik RB411/RB433

This patch adds support for serial console on Mikrotik RB411 and RB433
series devices.

Signed-off-by: David Varga <duvi@duvinet.hu>
[picked from FS#377]
Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoramips: rt3883: fix typo in pinctrl lna_g_func
Daniel Golle [Fri, 13 Jan 2017 04:41:18 +0000 (05:41 +0100)]
ramips: rt3883: fix typo in pinctrl lna_g_func

Spotted by forum users
https://forum.openwrt.org/viewtopic.php?pid=347439#p347439

Reported-by: Tom Psyborg <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
7 years agomap: take over maintainership
Hans Dedecker [Thu, 12 Jan 2017 11:11:09 +0000 (12:11 +0100)]
map: take over maintainership

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: Jo-Philipp Wich <jo@mein.io>
7 years agoodhcp6c: take over maintainership
Hans Dedecker [Thu, 12 Jan 2017 11:10:18 +0000 (12:10 +0100)]
odhcp6c: take over maintainership

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: Jo-Philipp Wich <jo@mein.io>
7 years agoodhcpd: take over maintainership
Hans Dedecker [Thu, 12 Jan 2017 11:06:11 +0000 (12:06 +0100)]
odhcpd: take over maintainership

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: Jo-Philipp Wich <jo@mein.io>
7 years agoRevert "dnsmasq: change 'add_local_hostname' to use dnsmasq '--interface-name'"
Hans Dedecker [Thu, 12 Jan 2017 09:17:48 +0000 (10:17 +0100)]
Revert "dnsmasq: change 'add_local_hostname' to use dnsmasq '--interface-name'"

This causes problem when a FQDN is configured in /etc/config/system. The
domain name will appear twice in reverse DNS.

Next to that, there seems to be a bug in dnsmasq. From the manual page:

--interface-name=<name>,<interface>[/4|/6]
Return  a  DNS  record  associating  the  name  with  the primary address
on the given interface. This flag specifies an A or AAAA record for the
given name in the same way as an /etc/hosts line, except that the address
is not constant, but taken from the given interface. The interface may be
followed by "/4" or "/6" to specify  that  only  IPv4  or  IPv6 addresses
of the interface should be used. If the interface is down, not configured
or non-existent, an empty record is returned. The matching PTR record is
also created, mapping the interface address to the name. More than one name
may be associated with an interface address by repeating the flag; in that
case the first instance is used for  the  reverse address-to-name mapping.

It does not just create an A/AAAA record for the primary address, it creates
one for all addresses. And what is worse, it seems to actually resolve to the
non-primary address first. This is quite annoying when you use floating IP
addresses (e.g. VRRP), because when the floating IP is on the other device,
SSH failes due to incorrect entry in the known hosts file.

I know that this is not a common setup, but it would be nice if there was an
option to restore the previous behaviour, rather than just forcing this new
feature on everybody.

Reported-by: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agomap: delete map-t device when tearing down map interface
Hans Dedecker [Thu, 12 Jan 2017 08:53:32 +0000 (09:53 +0100)]
map: delete map-t device when tearing down map interface

Delete the map-t device when tearing down the map-t interface; as such
there's no conflict when the map-t interface comes up again when trying
to add the map-t device as the map-t device was still present
(Can not add: device 'map-wan6_4' already exists!).

Only call ifdown in teardown for map-e and lw6o4 map interfaces types
in order to suppress the trace "wan6_4 (6652): Interface wan6_4_ not found"

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoRevert the recent dependency and metadata scanning rework
Felix Fietkau [Wed, 11 Jan 2017 18:47:33 +0000 (19:47 +0100)]
Revert the recent dependency and metadata scanning rework

This reverts the following commits:
fbe522d1204149b6c128d55e360bfc15fa8258e1
278ad007ee03c4455c3507322e34b0e3e6ec050d
863888e44f7a1fb1675a2e7a2eaabfec3561d2a2
96daf6352f6e04bc22789466cb5409b93fbec191
cfd83555fc4f0bab18a26f6812da18e64df46ff3

This seems to trigger some mconf bugs when built with all feeds
packages, so I will try to find a less intrusive solution before the
release.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agocomgt: allow build without USB_SUPPORT
Felix Fietkau [Wed, 11 Jan 2017 17:06:06 +0000 (18:06 +0100)]
comgt: allow build without USB_SUPPORT

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoumbim: allow build without USB_SUPPORT
Felix Fietkau [Wed, 11 Jan 2017 17:05:38 +0000 (18:05 +0100)]
umbim: allow build without USB_SUPPORT

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agouqmi: allow build without USB_SUPPORT
Felix Fietkau [Wed, 11 Jan 2017 17:05:16 +0000 (18:05 +0100)]
uqmi: allow build without USB_SUPPORT

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomountd: allow build without USB_SUPPORT
Felix Fietkau [Wed, 11 Jan 2017 17:04:18 +0000 (18:04 +0100)]
mountd: allow build without USB_SUPPORT

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoscripts/package-metadata.pl: fix overriding conditional dependencies with conditional...
Felix Fietkau [Wed, 11 Jan 2017 17:22:56 +0000 (18:22 +0100)]
scripts/package-metadata.pl: fix overriding conditional dependencies with conditional select

Conditional dependencies use the '(!cond) || dep' syntax, whereas
conditional select uses 'dep if cond'.
Add an extra check to suppress emitting a conditional if an equal
conditional select already exists.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarc770: build dtb files in Image/Prepare so that they are available for Device/*
Felix Fietkau [Wed, 11 Jan 2017 15:25:32 +0000 (16:25 +0100)]
arc770: build dtb files in Image/Prepare so that they are available for Device/*

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarchs38: build dtb files in Image/Prepare so that they are available for Device/*
Felix Fietkau [Wed, 11 Jan 2017 15:25:48 +0000 (16:25 +0100)]
archs38: build dtb files in Image/Prepare so that they are available for Device/*

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobcm53xx: backport upstream bcm53xx spi driver changes
Rafał Miłecki [Wed, 11 Jan 2017 14:21:13 +0000 (15:21 +0100)]
bcm53xx: backport upstream bcm53xx spi driver changes

One replaces our local patch and another is just for keeping driver in
sync (license change).

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agomvebu: work around an ethernet tx scheduling fairness issue
Felix Fietkau [Wed, 11 Jan 2017 13:55:45 +0000 (14:55 +0100)]
mvebu: work around an ethernet tx scheduling fairness issue

The hardware queue scheduling is apparently configured with fixed
priorities, which creates a nasty fairness issue where traffic from one
CPU can starve traffic from all other CPUs.

Work around this issue by forcing all tx packets to go through one CPU,
until this issue is fixed properly.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokirkwood: only add UBI EOF markers for devices that need it
Felix Fietkau [Wed, 11 Jan 2017 13:03:46 +0000 (14:03 +0100)]
kirkwood: only add UBI EOF markers for devices that need it

Fixes issues where the second-stage u-boot already uses UBI.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoprocd: add support for overriding the tar sysupgrade board name
Felix Fietkau [Sun, 8 Jan 2017 13:26:57 +0000 (14:26 +0100)]
procd: add support for overriding the tar sysupgrade board name

Useful for providing images that work for multiple devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: disable sub-page writes on routerboard nand drivers
Felix Fietkau [Sun, 8 Jan 2017 12:45:02 +0000 (13:45 +0100)]
ar71xx: disable sub-page writes on routerboard nand drivers

They seem to fail in tests using UBI, and are not used by yaffs2

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: remove linux 4.1 support
Felix Fietkau [Tue, 10 Jan 2017 17:15:39 +0000 (18:15 +0100)]
kernel: remove linux 4.1 support

The only target still referencing it is omap24xx, and it is marked as
broken.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: fix legacy image build error
Felix Fietkau [Sat, 7 Jan 2017 20:00:47 +0000 (21:00 +0100)]
ar71xx: fix legacy image build error

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarc770: fix parallel build issue
Felix Fietkau [Wed, 11 Jan 2017 10:20:25 +0000 (11:20 +0100)]
arc770: fix parallel build issue

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoarchs38: fix parallel build issue
Felix Fietkau [Wed, 11 Jan 2017 10:17:13 +0000 (11:17 +0100)]
archs38: fix parallel build issue

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: fix RB4xx CPLD SPI device mode setup
Felix Fietkau [Wed, 11 Jan 2017 09:42:04 +0000 (10:42 +0100)]
ar71xx: fix RB4xx CPLD SPI device mode setup

Commit af79fdbe4af3 changed the code to use tx_nbits for dual SPI tx
transfers, however the SPI stack only allows this when the device mode
includes the relevant bit as well

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: fix RB4xx SPI driver mode bits
Felix Fietkau [Wed, 11 Jan 2017 11:26:27 +0000 (12:26 +0100)]
ar71xx: fix RB4xx SPI driver mode bits

Accept SPI_TX_DUAL in device mode to fix the CPLD driver

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomvebu: update mwlwifi driver to version 10.3.2.0-20170110
Kabuli Chana [Tue, 10 Jan 2017 16:46:24 +0000 (09:46 -0700)]
mvebu: update mwlwifi driver to version 10.3.2.0-20170110

Improves automatic channel section support.

Tested on AC (mamba) and ACM (rango).

Signed-off-by: Kabuli Chana <newtownBuild@gmail.com>
[Jo-Philipp Wich: reword commit message]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agokernel: backport a MIPS SMP icache flush fix
Felix Fietkau [Tue, 27 Dec 2016 23:23:00 +0000 (00:23 +0100)]
kernel: backport a MIPS SMP icache flush fix

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: update bcm47xxpart failsafe partition patches
Rafał Miłecki [Wed, 11 Jan 2017 08:53:44 +0000 (09:53 +0100)]
kernel: update bcm47xxpart failsafe partition patches

V2 using helper function was sent for upstream kernel

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agowpa_supplicant: Fix mesh encryption config
Sujith Manoharan [Wed, 11 Jan 2017 02:09:22 +0000 (07:39 +0530)]
wpa_supplicant: Fix mesh encryption config

wpa_supplicant allows only SAE as the key management
type for mesh mode. The recent key_mgmt rework unconditionally
added WPA-PSK - this breaks interface bringup and wpa_s
throws this error message:

Line 10: key_mgmt for mesh network should be open or SAE
Line 10: failed to parse network block.
Failed to read or parse configuration '/var/run/wpa_supplicant-wlan0.conf

Fix this by making sure that only SAE is used for mesh.

Signed-off-by: Sujith Manoharan <m.sujith@gmail.com>
7 years agogettext-full: avoid using iconv for host builds
Jo-Philipp Wich [Wed, 11 Jan 2017 02:44:51 +0000 (03:44 +0100)]
gettext-full: avoid using iconv for host builds

The gettext-full host build might pick up iconv-stub host build  headers
during the build, leading to stray linker errors with unresolved references
to libiconv_open(), libiconv() and libiconv_close().

Since we're not needing iconv support on the host, pass the appropriate
cache variables to configure to prevent detection and linking of iconv.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agotools: cmake: link librt if needed (FS#381)
Jo-Philipp Wich [Wed, 11 Jan 2017 02:41:10 +0000 (03:41 +0100)]
tools: cmake: link librt if needed (FS#381)

Patch cmakes cmcurl component to unconditionally link librt when we discover
a clock_gettime() symbol in -lrt.

This is required to sucessfully link LibreSSL's libcrypto which might or
might not reqauire librt.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agotools: mkimage: pass crypto libraries through HOST_LOADLIBES (FS#381)
Jo-Philipp Wich [Wed, 11 Jan 2017 01:43:04 +0000 (02:43 +0100)]
tools: mkimage: pass crypto libraries through HOST_LOADLIBES (FS#381)

Since linker argument order is significant on modern host compilers, pass
the required libcrypto libraries through the HOST_LOADLIBES variable in
order to ensure proper order of the final linker command line.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agotools: libressl: always build as PIC
Matthias Schiffer [Tue, 10 Jan 2017 21:10:59 +0000 (22:10 +0100)]
tools: libressl: always build as PIC

Fixes link errors for host packages like ruby like the following:

/usr/bin/ld: .../staging_dir/host/lib/libcrypto.a(libcrypto_la-md5_dgst.o):
relocation R_X86_64_PC32 against symbol `memcpy@@GLIBC_2.14' can not be used when making a shared object; recompile with -fPIC

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agopackage: replace $(STAGING_DIR)/host with $(STAGING_DIR_HOSTPKG)
Matthias Schiffer [Tue, 10 Jan 2017 17:43:57 +0000 (18:43 +0100)]
package: replace $(STAGING_DIR)/host with $(STAGING_DIR_HOSTPKG)

Cleanup to prepare for changing STAGING_DIR_HOSTPKG. The actual change of
STAGING_DIR_HOSTPKG (i.e., moving the host packages back into a common, not
target-specific directory) will be done after the first LEDE release, but
the cleanup will also be useful for projects like Gluon.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agokernel: add missing config symbols
Jo-Philipp Wich [Tue, 10 Jan 2017 21:02:22 +0000 (22:02 +0100)]
kernel: add missing config symbols

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agokernel: add missing config symbol
Felix Fietkau [Tue, 10 Jan 2017 16:40:19 +0000 (17:40 +0100)]
kernel: add missing config symbol

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: pending brcmfmac patches cleaning channels management
Rafał Miłecki [Tue, 10 Jan 2017 16:18:01 +0000 (17:18 +0100)]
mac80211: pending brcmfmac patches cleaning channels management

They prepare brcmfmac for using wiphy_read_of_freq_limits.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agomac80211: rename b43 patches to make more space
Rafał Miłecki [Tue, 10 Jan 2017 14:51:51 +0000 (15:51 +0100)]
mac80211: rename b43 patches to make more space

Just 6 patches were using 80*, 81*, 82* and 84* prefixes.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agobuild: scan.mk: consider KernelPackage pattern as well
Jo-Philipp Wich [Tue, 10 Jan 2017 15:25:10 +0000 (16:25 +0100)]
build: scan.mk: consider KernelPackage pattern as well

The removal of the ".+Package" pattern in scan.mk also caused the build
system to skip over Makefiles defining only kmods. Adjust the grep pattern
to consider packages with "call KernelPackage" signatures as well.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoprocd: update mirror hash
John Crispin [Tue, 10 Jan 2017 12:38:53 +0000 (13:38 +0100)]
procd: update mirror hash

Signed-off-by: John Crispin <john@phrozen.org>
7 years agokernel: remove gpiommc patches / driver
Felix Fietkau [Tue, 10 Jan 2017 14:44:43 +0000 (15:44 +0100)]
kernel: remove gpiommc patches / driver

This code was marked as incompatible to Linux 4.4 well over a year ago
and nobody cared, and now it's breaking builds.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: add pending bcm47xxpart support for failsafe TRX partition
Rafał Miłecki [Tue, 10 Jan 2017 14:12:04 +0000 (15:12 +0100)]
kernel: add pending bcm47xxpart support for failsafe TRX partition

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agokernel: update spi-nor.h include fix with upstream accepted version
Rafał Miłecki [Tue, 10 Jan 2017 13:50:44 +0000 (14:50 +0100)]
kernel: update spi-nor.h include fix with upstream accepted version

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agokernel: rename bcm47xxpart patches to fit more of them
Rafał Miłecki [Tue, 10 Jan 2017 13:48:44 +0000 (14:48 +0100)]
kernel: rename bcm47xxpart patches to fit more of them

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agobuild: scan.mk: remove overlay broad grep pattern
Jo-Philipp Wich [Tue, 10 Jan 2017 14:12:21 +0000 (15:12 +0100)]
build: scan.mk: remove overlay broad grep pattern

Commit af0b91c "allow scan.mk to find python packages introduced in [8639]"
added some special casing to scan.mk to accomodate some nonstandard python
packages.

Nowadays this pattern is not needed anymore and produces false positives
when using the LEDE source repository as feed within the SDK since the
metadata scanning wrongly picks up target/imagebuilder/Makefile as package,
leading to an  "ERROR: please fix feeds/base/target/imagebuilder/Makefile"
message.

Remove the now uneeded pattern to fix such stray errors during metadata
scanning.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>