openwrt/staging/xback.git
7 years agoiproute2: cake: update cake support
Kevin Darbyshire-Bryant [Sat, 28 Jan 2017 16:44:30 +0000 (16:44 +0000)]
iproute2: cake: update cake support

Updated cake's tc patch to match the official cake repository
formatting.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agokmod-sched-cake: Bump to latest version
Kevin Darbyshire-Bryant [Sat, 28 Jan 2017 16:38:59 +0000 (16:38 +0000)]
kmod-sched-cake: Bump to latest version

wash, mpu & some memory optimisation have now made it to the official
cake repository.

Point LEDE to the official repository.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agolibtool: don't clobber host libtool infrastructure
Jo-Philipp Wich [Sat, 21 Jan 2017 23:28:54 +0000 (00:28 +0100)]
libtool: don't clobber host libtool infrastructure

The libtool target package stages its files into the host staging directory
and moves the libltdl library parts from there into the target staging
directory afterwards.

By doing so, the package essentially renders the host libtool infrastructure
unusable, leading to the below error in subsequent package builds:

    libtoolize: $pkgltdldir is not a directory: `.../hostpkg/share/libtool`

Prevent this problem by using a dedicated libltdl install prefix in order to
avoid overwriting and moving away preexisting files belonging to tools/libtool.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobuild: properly pass CPP and CXX flags in HOST_MAKE_VARS
Jo-Philipp Wich [Tue, 17 Jan 2017 08:29:50 +0000 (09:29 +0100)]
build: properly pass CPP and CXX flags in HOST_MAKE_VARS

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobuild: introduce default HOST_MAKE_VARS for host-builds
Alexandru Ardelean [Mon, 9 Jan 2017 12:59:02 +0000 (14:59 +0200)]
build: introduce default HOST_MAKE_VARS for host-builds

Inspired/adapted from `package-defaults.mk` MAKE_VARS.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
7 years agotools/cmake: remove HOST_CONFIGURE_CMD and re-distribute the args & vars
Alexandru Ardelean [Thu, 12 Jan 2017 13:31:13 +0000 (15:31 +0200)]
tools/cmake: remove HOST_CONFIGURE_CMD and re-distribute the args & vars

The final semantic is the same, but this is a bit more correct.

Build tested on Windows 10 (yes, there is some
Ubuntu mode for Windows 10, and I've been also building LEDE
on it for a few weeks).

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
7 years agoodhcp6c: use LEDE_GIT in package source url
Hans Dedecker [Mon, 30 Jan 2017 14:15:18 +0000 (15:15 +0100)]
odhcp6c: use LEDE_GIT in package source url

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoodhcp6c: update to git HEAD version
Hans Dedecker [Mon, 30 Jan 2017 14:03:29 +0000 (15:03 +0100)]
odhcp6c: update to git HEAD version

c13b6a0 dhcpv6: fix white space error
e9d80cc dhcpv6: trigger restart of DHCPv6 state machine when not
receiving statefull options
c7122ec update README
419fb63 dhcpv6: server unicast option support

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoodhcpd: use LEDE_GIT in package source url
Hans Dedecker [Mon, 30 Jan 2017 21:15:48 +0000 (22:15 +0100)]
odhcpd: use LEDE_GIT in package source url

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoodhcpd: update to git HEAD version
Hans Dedecker [Thu, 26 Jan 2017 20:28:28 +0000 (21:28 +0100)]
odhcpd: update to git HEAD version

c4f9ace odhcpd: decrease default log level to LOG_INFO
a6eadd7 odhcpd: rework IPv6 interface address dump
44965f1 odhcpd: extra syslog tracing

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoodhcpd: update to git HEAD version
Hans Dedecker [Tue, 24 Jan 2017 09:01:00 +0000 (10:01 +0100)]
odhcpd: update to git HEAD version

e447ff9 router: fix compile issue on 64 bit systems

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoodhcpd: update to git HEAD version
Hans Dedecker [Mon, 23 Jan 2017 17:36:25 +0000 (18:36 +0100)]
odhcpd: update to git HEAD version

237f1f4 router: convert syslog lifetime traces into LOG_INFO prio
da660c7 treewide: rework prio of syslog messages
0485580 ndp: code cleanup
c5040fe router: add syslog debug tracing for trouble shooting
df023ad treewide: use RELAYD_MAX_ADDRS as address array size
c8ac572 ndp: don't scan netlink attributes in case of netlink route
event

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agonetifd: update to git HEAD version
Hans Dedecker [Thu, 26 Jan 2017 20:22:18 +0000 (21:22 +0100)]
netifd: update to git HEAD version

650758b interface-ip: route proto config support (FS#170)

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agobase-files: add /etc/iproute2/rt_protos
Hans Dedecker [Tue, 17 Jan 2017 21:50:40 +0000 (22:50 +0100)]
base-files: add /etc/iproute2/rt_protos

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agobase-files: uppercase default hostname: LEDE
Yousong Zhou [Wed, 11 Jan 2017 03:21:16 +0000 (11:21 +0800)]
base-files: uppercase default hostname: LEDE

The name will appear in shell prompt and LuCI page title.  Uppercase
letters seem to be more vigorous

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
7 years agoprocd: update procd.sh to disallow signal-numbers, enforce signal-names
Bastian Bittorf [Fri, 13 Jan 2017 19:22:51 +0000 (20:22 +0100)]
procd: update procd.sh to disallow signal-numbers, enforce signal-names

A given signal-name is now converted to the corresonding number. In general
it's good style to use names (readability) and it's more portable: signal
numbers can be architecture-dependent, so we are more safe giving names.

A real world example is signal 10, which is BUS on ramips and USR1 on PPC.

All users of 'procd_send_signal' must change their code to reflect this.

Signed-off-by: Bastian Bittorf <bb@npl.de>
7 years agoopenvpn: ssl-enabled variants also provide a virtual openvpn-crypto package
Sven Roederer [Sun, 22 Jan 2017 18:00:51 +0000 (19:00 +0100)]
openvpn: ssl-enabled variants also provide a virtual openvpn-crypto package

When relying on x.509 certs for auth and / or encryption of traffic you can't
use package openvpn-nossl.
Just have your package depend on openvpn-crypto to have SSL-encryption and
X.509-support enabled in OpenVPN. If encryption / X.509 is not a must, use
virtual packge openvpn, which is provided by all OpenVPN-variants.

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
7 years agoiproute2: cake: add 'mpu' minimum packet length support
Kevin Darbyshire-Bryant [Sat, 31 Dec 2016 17:15:56 +0000 (17:15 +0000)]
iproute2: cake: add 'mpu' minimum packet length support

Add 'mpu' minimum length packet size parameter for scheduling/bandwidth
accounting.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agokmod-sched-cake: add 'mpu' minimum packet length support
Kevin Darbyshire-Bryant [Fri, 23 Dec 2016 16:46:45 +0000 (16:46 +0000)]
kmod-sched-cake: add 'mpu' minimum packet length support

Add 'mpu' minimum packet length for scheduling/bandwidth accounting
purposes.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agohostapd: default to wps_independent 1
Steven Honson [Sat, 14 Jan 2017 11:22:24 +0000 (22:22 +1100)]
hostapd: default to wps_independent 1

Signed-off-by: Steven Honson <steven@honson.id.au>
7 years agohostapd: expose wps_independent and ap_setup_locked as uci options
Steven Honson [Tue, 27 Dec 2016 05:16:23 +0000 (16:16 +1100)]
hostapd: expose wps_independent and ap_setup_locked as uci options

ap_setup_locked is named wps_ap_setup_locked in uci for consistency with other
wps related uci options.

Signed-off-by: Steven Honson <steven@honson.id.au>
7 years agomwlwifi: Fixes rewritten history hash and latest version
Gabe Rodriguez [Wed, 25 Jan 2017 04:48:53 +0000 (20:48 -0800)]
mwlwifi: Fixes rewritten history hash and latest version

The author of the upstream mwlwifi edited the history of the previous commit.
This commit not only fixes the updated hash but also sends in the latest
commits he made to the code which are mainly testing.

Signed-off-by: Gabe Rodriguez <lifehacksback@gmail.com>
7 years agoopkg: clarify messages and errors related to downloads
Hannu Nyman [Mon, 23 Jan 2017 09:40:29 +0000 (11:40 +0200)]
opkg: clarify messages and errors related to downloads

Clarify opkg's messages related to downloads:

* more visible error message for package list download failure
* separate error message for signature file download error
* if wget returns 4, signal the network error more clearly
* remove '.' from end of filenames and URLs

* try signature check only if the package list was downloaded ok.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
7 years agoubus: update to the latest version
Felix Fietkau [Sun, 22 Jan 2017 13:37:24 +0000 (14:37 +0100)]
ubus: update to the latest version

Adds uloop related libubus fixes

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoubus: update to the latest version
Felix Fietkau [Fri, 20 Jan 2017 10:34:26 +0000 (11:34 +0100)]
ubus: update to the latest version

- Adds support for passing file descriptors in ubus invoke requests
- Fixes clearing pending timers on ubus_shutdown()
- Fixes checking the amount of written data in ubusd
- Fixes an ubusd crash when trying to subscribe to system objects

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: ar8327/ar8337: disable ARL access code to avoid lockups (FS#384)
Jo-Philipp Wich [Fri, 13 Jan 2017 15:38:22 +0000 (16:38 +0100)]
kernel: ar8327/ar8337: disable ARL access code to avoid lockups (FS#384)

Running ar8327_get_arl_entry() early after boot leads to MDIO related system
lockups on several devices using this driver.

Since dumping the ARL table contens is an optional, uncritical feature, simply
disable the code for now.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobcm53xx: fix LAN MAC address for devices that use eth2 originally
Rafał Miłecki [Tue, 31 Jan 2017 12:53:46 +0000 (13:53 +0100)]
bcm53xx: fix LAN MAC address for devices that use eth2 originally

We override default Ethernet interface with eth0 which often uses random
MAC due to missing proper NVRAM entry. Fix this by manually assigning
MAC in the config.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agoar71xx: add missing DEVICE_TITLE for mikrotik devices
Felix Fietkau [Sun, 29 Jan 2017 16:32:40 +0000 (17:32 +0100)]
ar71xx: add missing DEVICE_TITLE for mikrotik devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: Fix mikrotik subtarget default profile for device profile selection
Alex Samorukov [Sun, 22 Jan 2017 17:56:37 +0000 (17:56 +0000)]
ar71xx: Fix mikrotik subtarget default profile for device profile selection

Signed-off-by: Alex Samorukov <samm@os2.kiev.ua>
7 years agoar71xx: improve Mikrotik hAP Lite device support
Alex Samorukov [Thu, 19 Jan 2017 14:38:49 +0000 (15:38 +0100)]
ar71xx: improve Mikrotik hAP Lite device support

- remove CONFIG_MTD_SPI_NOR_USE_4K_SECTORS
- enable jffs2 support
- fix sysupgrade

Signed-off-by: Alex Samorukov <samm@os2.kiev.ua>
7 years agoar71xx: mark soft_config mtd part as writeable for RB-941-2nD
Thibaut VARENE [Wed, 18 Jan 2017 09:44:19 +0000 (10:44 +0100)]
ar71xx: mark soft_config mtd part as writeable for RB-941-2nD

The soft_config partition must be writeable for rbcfg to be able to
enact changes to the routerboot configuration.
The read-only flag was a mistake in the initial patch. Removing it
brings mach-rb941.c in line with all other RB platforms.

Signed-off-by: Thibaut VARENE <hacks@slashdirt.org>
7 years agoar71xx: fix up the kernel config for the mikrotik subtarget
Felix Fietkau [Tue, 10 Jan 2017 17:09:22 +0000 (18:09 +0100)]
ar71xx: fix up the kernel config for the mikrotik subtarget

Limit enabled devices to routerboards, disable all other devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: fix network config for Mikrotik RB411U
Cezary Jackiewicz [Mon, 9 Jan 2017 21:06:12 +0000 (22:06 +0100)]
ar71xx: fix network config for Mikrotik RB411U

Mikrotik RB411U has only one ethernet port - eth0. This patch allows to create
correct config with one lan section.

Signed-off-by: Cezary Jackiewicz <cezary@eko.one.pl>
7 years agoar71xx: add support for RB-941-2nD
Thibaut VARENE [Wed, 4 Jan 2017 21:44:54 +0000 (22:44 +0100)]
ar71xx: add support for RB-941-2nD

Add full support for Mikrotik RB-941-2nD (hAP lite)

Original patch by Sergey Sergeev <adron@yapic.net> and
more information is available here:
https://wiki.openwrt.org/toh/mikrotik/rb941_2nd
I updated and adapted the patch to apply cleanly to LEDE trunk
and added proper numbering for the switch ports (matching case
labels).
Tested working on actual hardware with the information
provided in the above webpage. Sysupgrade works.

Signed-off-by: Thibaut VARENE <hacks@slashdirt.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: convert mikrotik routerboard support to UBI
Felix Fietkau [Sun, 8 Jan 2017 13:27:38 +0000 (14:27 +0100)]
ar71xx: convert mikrotik routerboard support to UBI

Remove the wget2nand script, drop the need for manual installation,
use sysupgrade instead.

There are now two different NAND images, one for 64 MiB flashes, the
other for >= 128 MiB

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: create a proper default profile for the mikrotik subtarget, drop other profiles
Felix Fietkau [Sat, 7 Jan 2017 19:54:42 +0000 (20:54 +0100)]
ar71xx: create a proper default profile for the mikrotik subtarget, drop other profiles

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agotools: update kernel2minor to 0.24 version
Sergey Sergeev [Tue, 24 Jan 2017 13:04:18 +0000 (16:04 +0300)]
tools: update kernel2minor to 0.24 version

Signed-off-by: Sergey Sergeev <adron@yapic.net>
7 years agobase-files: don't overwrite model name set by target
Mathias Kresin [Sat, 28 Jan 2017 15:19:35 +0000 (16:19 +0100)]
base-files: don't overwrite model name set by target

The condition is always true due to the literal string followed the
-n test parameter. A model name set by target scripts always gets
overwritten this way.

Change the condition to check for an already existing destination file
as it was before 5e85ae9 ("base-files: fix error message during boot").

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agomt76: update to the latest version
Felix Fietkau [Tue, 31 Jan 2017 20:28:28 +0000 (21:28 +0100)]
mt76: update to the latest version

Fixes DFS detection false positive issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: update bcma to fix devm memory leaks
Rafał Miłecki [Tue, 31 Jan 2017 13:21:53 +0000 (14:21 +0100)]
kernel: update bcma to fix devm memory leaks

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agoath9k: remove the deaf rx path state check patch
Felix Fietkau [Tue, 31 Jan 2017 10:38:25 +0000 (11:38 +0100)]
ath9k: remove the deaf rx path state check patch

This needs to be refined and reworked before we can safely leave it
enabled by default

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomdns: update to the latest version
Felix Fietkau [Tue, 31 Jan 2017 10:23:02 +0000 (11:23 +0100)]
mdns: update to the latest version

- fixes unaligned acccesses, causing DNS parsing issues on ARMv5
- fixes service timeout handling

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodownload.pl: fix detecting download errors with curl
Felix Fietkau [Tue, 31 Jan 2017 10:22:21 +0000 (11:22 +0100)]
download.pl: fix detecting download errors with curl

Avoid treating error pages as successfully downloaded content

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoramips: add missing DTS pcie node for WSR-600
Felix Fietkau [Tue, 31 Jan 2017 09:27:25 +0000 (10:27 +0100)]
ramips: add missing DTS pcie node for WSR-600

Enables PCI bus and support for the MT7603 and RT5592 chips

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoqemu: rename internal crypto/aes symbols
Stijn Tintel [Mon, 30 Jan 2017 16:47:11 +0000 (17:47 +0100)]
qemu: rename internal crypto/aes symbols

Qemu's local AES code defines symbols that conflict with
LibreSSL/OpenSSL's libcrypto. Rename them to avoid build problems.

See upstream commit c8d70e59738e672021926c7747af8ef9dea15c82.

Fixes FS#444.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
7 years agotools/qemu: use default host configure rule ; set appropriate vars & args
Alexandru Ardelean [Mon, 9 Jan 2017 12:46:18 +0000 (14:46 +0200)]
tools/qemu: use default host configure rule ; set appropriate vars & args

Admittedly, this is my own OCD wanting to get rid of this.

Because I tried (a while back to upgrade QEMU to a newer version),
and (during that attempt) I tried to get rid of this.

Tested on Linux & Mac.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
7 years agolantiq: fix unaligned access in xrx200_poll_rx()
Alexander Couzens [Sat, 28 Jan 2017 16:23:16 +0000 (17:23 +0100)]
lantiq: fix unaligned access in xrx200_poll_rx()

Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
7 years agoledtrig-netdev: don't cancel work on events for different interfaces
Sergey Sergeev [Sun, 29 Jan 2017 16:01:40 +0000 (19:01 +0300)]
ledtrig-netdev: don't cancel work on events for different interfaces

This fixes logic bug(in function netdev_trig_notify) introduced in
0b2991a8ed32b5d9ba0a7c3441a72e1c1a1c4fde commit.
Events triggered by different interfaces were stopping work queue so it
wasn't working for tx/rx mode.

Signed-off-by: Sergey Sergeev <adron@yapic.net>
7 years agokmod-sched-core: Add HTB and TBF traffic shapers
Toke Høiland-Jørgensen [Fri, 20 Jan 2017 12:13:35 +0000 (13:13 +0100)]
kmod-sched-core: Add HTB and TBF traffic shapers

HTB and TBF are the basic traffic shapers used by sqm-scripts. Moving
these into kmod-sched-core enables sqm-scripts to downgrade its
dependency from kmod-sched to kmod-sched-core, potentially making it
useful on devices with smaller flash sizes.

This adds around 30k to the size of kmod-sched-core (20k for sch_htb.ko
and 10k for sch_tbf.ko).

Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
7 years agoar71xx: prepare jffs2 partition properly in factory.bin for BHR-4GRV2
FUKAUMI Naoki [Fri, 9 Dec 2016 08:53:15 +0000 (17:53 +0900)]
ar71xx: prepare jffs2 partition properly in factory.bin for BHR-4GRV2

"pad-rootfs" is required just after "append-rootfs".

Signed-off-by: FUKAUMI Naoki <naobsd@gmail.com>
7 years agomxs: remove stale references to obsolete kernel module packages
Michael Heimpold [Sat, 28 Jan 2017 15:32:34 +0000 (16:32 +0100)]
mxs: remove stale references to obsolete kernel module packages

Commit 29443e2 (mxs: remove modules.mk, select drivers in the kernel config)
missed to remove these references, so cleanup it now.

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
7 years agoopenvpn: add support for various new 2.4 configuration options
Magnus Kroken [Mon, 16 Jan 2017 21:56:03 +0000 (22:56 +0100)]
openvpn: add support for various new 2.4 configuration options

Updates to openvpn.init were included in early OpenVPN 2.4 patch
series, but got lost along the way and were never merged.

Signed-off-by: Magnus Kroken <mkroken@gmail.com>
7 years agokernel: backport support for BCM54210E PHY
Rafał Miłecki [Sat, 28 Jan 2017 17:51:57 +0000 (18:51 +0100)]
kernel: backport support for BCM54210E PHY

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agokernel: backport support for BCM54810 PHY
Rafał Miłecki [Sat, 28 Jan 2017 17:44:23 +0000 (18:44 +0100)]
kernel: backport support for BCM54810 PHY

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agokernel: backport support for BCM54612E PHY
Rafał Miłecki [Sat, 28 Jan 2017 16:33:56 +0000 (17:33 +0100)]
kernel: backport support for BCM54612E PHY

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agoopenssl: update to version 1.0.2k
Hauke Mehrtens [Fri, 27 Jan 2017 22:36:45 +0000 (23:36 +0100)]
openssl: update to version 1.0.2k

This fixes the following security problems:
CVE-2017-3731: Truncated packet could crash via OOB read
CVE-2017-3732: BN_mod_exp may produce incorrect results on x86_64
CVE-2016-7055: Montgomery multiplication may produce incorrect results

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agolantiq: fix brnImage signature for the VGV7510KW22BRN images
Mathias Kresin [Sat, 28 Jan 2017 11:36:15 +0000 (12:36 +0100)]
lantiq: fix brnImage signature for the VGV7510KW22BRN images

The VGV7510KW22BRN and VGV7519BRN do not have the same brnImage
signature. It was accidentally changed with ba42c1d ("lantiq: un-macro
the image building code").

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoath9k: fix up a refcount imbalance error in the IRQ related fix
Felix Fietkau [Fri, 27 Jan 2017 14:43:12 +0000 (15:43 +0100)]
ath9k: fix up a refcount imbalance error in the IRQ related fix

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoprocd: update to latest git HEAD
Matthias Schiffer [Thu, 26 Jan 2017 15:18:41 +0000 (16:18 +0100)]
procd: update to latest git HEAD

0f58977 init: fix /tmp permissions on zram

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agobase-files: fix user creation on sysupgrade with few opkg control files
Matthias Schiffer [Tue, 24 Jan 2017 17:55:13 +0000 (18:55 +0100)]
base-files: fix user creation on sysupgrade with few opkg control files

If only a single opkg control file exists (which can happen with
CONFIG_CLEAN_IPKG), grep would not print the file name by default. Instead
of forcing it using -H, we just switch to -l (print only file names) and
get rid of the cut.

Add -s to suppress an error message when no control files exist.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoinclude/rootfs.mk: keep Require-User lines with CONFIG_CLEAN_IPKG
Matthias Schiffer [Tue, 24 Jan 2017 16:00:43 +0000 (17:00 +0100)]
include/rootfs.mk: keep Require-User lines with CONFIG_CLEAN_IPKG

Require-User is handled by /etc/uci-defaults/13_fix_group_user on first
boot, so we need to keep these when removing all opkg data with
CONFIG_CLEAN_IPKG.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoimagebuilder: properly escape single quotes in device titles
Jo-Philipp Wich [Thu, 26 Jan 2017 09:13:23 +0000 (10:13 +0100)]
imagebuilder: properly escape single quotes in device titles

The name "Plat'Home OpenBlocks AX3" causes the imagebuilders "make info"
command to fail with:

    bash: -c: line 0: syntax error near unexpected token `('
    bash: -c: line 0: `echo;  [...]'
    Makefile:99: recipe for target '_call_info' failed

Properly escape single quotes to avoid breaking the echo commands.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoath9k: add stability fixes for long standing hang issues (FS#13, #34, #373, #383)
Felix Fietkau [Wed, 25 Jan 2017 14:32:18 +0000 (15:32 +0100)]
ath9k: add stability fixes for long standing hang issues (FS#13, #34, #373, #383)

The radio would stop communicating completely. This issue was easiest to
trigger on AR913x devices, e.g. the TP-Link TL-WR1043ND, but other
hardware was occasionally affected as well.

The most critical issue was a race condition in disabling/enabling IRQs
between the IRQ handler and the IRQ processing tasklet

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: refresh patch
Felix Fietkau [Wed, 25 Jan 2017 14:31:51 +0000 (15:31 +0100)]
mac80211: refresh patch

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoAdd back the commit "ath9k: Add airtime fairness scheduler"
Felix Fietkau [Tue, 24 Jan 2017 14:23:30 +0000 (15:23 +0100)]
Add back the commit "ath9k: Add airtime fairness scheduler"

This reverts commit c296ba834db4ce8c71e0ad7030aab188fe60b27b.
According to several reports, the issues with the airtime fairness
changes are gone in current versions.
It's time to re-apply the patch now.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobcm53xx: disable building Linksys EA6300 V1 image
Rafał Miłecki [Wed, 25 Jan 2017 09:48:51 +0000 (10:48 +0100)]
bcm53xx: disable building Linksys EA6300 V1 image

This device has 2 TRX partitions (main one and failsafe one) and Linux
may not detect them properly failing to run userspace.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agox86: fix sysupgrades on disks with 4k block size
Felix Fietkau [Wed, 25 Jan 2017 07:13:11 +0000 (08:13 +0100)]
x86: fix sysupgrades on disks with 4k block size

Even when the disk uses 4k blocks, the partition table still uses units
of 512 byte sectors. Always use ibs=512 for the offsets

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: brcmfmac: don't use uninitialize mem for country codes
Rafał Miłecki [Tue, 24 Jan 2017 15:54:09 +0000 (16:54 +0100)]
mac80211: brcmfmac: don't use uninitialize mem for country codes

There was a bug in brcmfmac patch that could result in treating random
memory as source of country codes.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agomac80211: move (& update) upstream accepted brcmfmac patches
Rafał Miłecki [Tue, 17 Jan 2017 21:41:18 +0000 (22:41 +0100)]
mac80211: move (& update) upstream accepted brcmfmac patches

These 3 patches are now in wireless-drivers-next tree.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agobrcm47xx: backport arch patch with Luxul devices support
Rafał Miłecki [Tue, 24 Jan 2017 19:22:14 +0000 (20:22 +0100)]
brcm47xx: backport arch patch with Luxul devices support

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agopackages: mark packages depending on a target as nonshared
Mathias Kresin [Thu, 19 Jan 2017 08:18:57 +0000 (09:18 +0100)]
packages: mark packages depending on a target as nonshared

The packages can't be build as shared packages due to the unmet
dependencies.

Fixes FS#418.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agosdk: explicitely remove ccache directories when packing SDK
Jo-Philipp Wich [Tue, 24 Jan 2017 14:29:20 +0000 (15:29 +0100)]
sdk: explicitely remove ccache directories when packing SDK

Upon first invocation, the ccache program will create the required directory
hierarchy so there is no point in shipping these empty directories.

Removing those paths also avoids shipping dangling symlinks in case the
directories got linked elsewhere, e.g. into a shared global cache.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agolantiq: remove CPU_TYPE:=mips32r2, it gets overwritten anyway
Felix Fietkau [Tue, 24 Jan 2017 12:33:45 +0000 (13:33 +0100)]
lantiq: remove CPU_TYPE:=mips32r2, it gets overwritten anyway

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86: unify CPU_TYPE for legacy and geode
Felix Fietkau [Tue, 24 Jan 2017 12:11:43 +0000 (13:11 +0100)]
x86: unify CPU_TYPE for legacy and geode

According to some reports, -march=pentium-mmx is a better choice for
older Geode CPUs than -march=geode anyway.

Bump the minimum architecture of the legacy target from i486 to
pentium-mmx. Anything older is not worth supporting anyway.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agouml: mark as source-only
Felix Fietkau [Tue, 24 Jan 2017 11:44:59 +0000 (12:44 +0100)]
uml: mark as source-only

Get rid of a special case in the buildbot script

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomalta: mark as source-only to avoid wasting build resources
Felix Fietkau [Tue, 24 Jan 2017 11:41:40 +0000 (12:41 +0100)]
malta: mark as source-only to avoid wasting build resources

The le64 and be64 subtargets do not share a package architecture with
any other targets, so they are pretty wasteful for a development-only
target.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomalta: move FEATURES to the target makefile
Felix Fietkau [Tue, 24 Jan 2017 11:41:19 +0000 (12:41 +0100)]
malta: move FEATURES to the target makefile

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: remove mips16 feature flag from target makefiles
Felix Fietkau [Tue, 24 Jan 2017 11:40:00 +0000 (12:40 +0100)]
build: remove mips16 feature flag from target makefiles

It can be implicitly derived from the MIPS32 revision support in the
kernel configuration

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: remove obsolete mips32r2 CPU_TYPE
Felix Fietkau [Tue, 24 Jan 2017 11:30:26 +0000 (12:30 +0100)]
build: remove obsolete mips32r2 CPU_TYPE

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoramips/rt288x: switch CPU_TYPE to 24kc
Felix Fietkau [Tue, 24 Jan 2017 11:25:05 +0000 (12:25 +0100)]
ramips/rt288x: switch CPU_TYPE to 24kc

While rt288x only has a MIPS 4KEc processor, it implements the MIPS32r2
architecture just like the 24Kc, so the instruction set should be 100%
compatible.
Switching it to 24kc allows it to share the package architecture with a
lot of other targets instead of creating a special case, saving
buildbot resources.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86: remove the xen_domu subtarget
Felix Fietkau [Fri, 13 Jan 2017 14:04:17 +0000 (15:04 +0100)]
x86: remove the xen_domu subtarget

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86/generic: add xen DomU support
Felix Fietkau [Fri, 13 Jan 2017 14:03:24 +0000 (15:03 +0100)]
x86/generic: add xen DomU support

Copy the relevant config options from the xen_domu subtarget

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agohostapd: fix stray "out of range" shell errors in hostapd.sh
Jo-Philipp Wich [Mon, 23 Jan 2017 13:52:27 +0000 (14:52 +0100)]
hostapd: fix stray "out of range" shell errors in hostapd.sh

The hostapd_append_wpa_key_mgmt() procedure uses the possibly uninitialized
$ieee80211r and $ieee80211w variables in a numerical comparisation, leading
to stray "netifd: radio0 (0000): sh: out of range" errors in logread when
WPA-PSK security is enabled.

Ensure that those variables are substituted with a default value in order to
avoid emitting this (harmless) shell error.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoopenvpn: let all openvpn variants provide a virtual openvpn package
Jo-Philipp Wich [Sat, 21 Jan 2017 22:03:36 +0000 (23:03 +0100)]
openvpn: let all openvpn variants provide a virtual openvpn package

Add PROVIDES:=openvpn to the default recipe in order to let all build variants
provide a virtual openvpn package.

The advantage of this approach is that downstream packages can depend on just
"openvpn" without having to require a specific flavor.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agokmod-sched-cake: fix parameter passing kernel/user space
Kevin Darbyshire-Bryant [Fri, 23 Dec 2016 16:46:45 +0000 (16:46 +0000)]
kmod-sched-cake: fix parameter passing kernel/user space

The last two parameters passed between user space tc and kernel space
sched-cake were transposed due to a merge mistake in a parameter header
file.

As such, using a packet overhead figure was likely to set cake to wash
packet DSCP values.  Similarly, the DSCP wash flag was used as an offset
to the displayed packet overhead value.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agozlib: Update to 1.2.11
Daniel Engberg [Sun, 15 Jan 2017 20:46:27 +0000 (21:46 +0100)]
zlib: Update to 1.2.11

Update to 1.2.11 as suggested by upstream
Also add SF as primary source and main site as fallback

Note: SF doesn't carry the 1.2.11 update yet.

Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
7 years agolede-keyring: bundle latest usign certificates
Jo-Philipp Wich [Fri, 20 Jan 2017 09:09:30 +0000 (10:09 +0100)]
lede-keyring: bundle latest usign certificates

Includes the public usign certificates used by the 17.01.* release builds.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agosdk: fix Git URL detection
Jo-Philipp Wich [Fri, 20 Jan 2017 08:51:38 +0000 (09:51 +0100)]
sdk: fix Git URL detection

Instead of relying on complex sed patterns that trip up make syntax rules, use
GNU Makes builtin filter function to match desired URLs.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agosdk: avoid using private repository clone urls as base repo entry
Jo-Philipp Wich [Thu, 19 Jan 2017 19:41:46 +0000 (20:41 +0100)]
sdk: avoid using private repository clone urls as base repo entry

Only consider the repository origin url as valid base feed entry if it is a
git://, http:// or https:// location.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agolantiq: fix an ethernet stability issue triggered by receving packets during boot
Felix Fietkau [Thu, 19 Jan 2017 13:25:09 +0000 (14:25 +0100)]
lantiq: fix an ethernet stability issue triggered by receving packets during boot

Disabling ethernet during reboot (only to enable it again when the
ethernet driver attaches) can put the chip into a faulty state where it
corrupts the header of all incoming packets.

This happens if packets arrive during the time window where the core is
disabled, and it can be easily reproduced by rebooting while sending a
flood ping to the broadcast address.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agolantiq: fix spurious irq storm
Felix Fietkau [Thu, 19 Jan 2017 11:31:34 +0000 (12:31 +0100)]
lantiq: fix spurious irq storm

Since the MIPS IRQ stack patches, lantiq devices were emitting a storm
of messages like this:

[  567.872172] Spurious IRQ: CAUSE=0x1100c300

Fix this by reworking the IRQ dispatch code

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agogit-kernel: $(SUBDIR) should always be $(LINUX_VERSION)
Mirko Vogt [Wed, 18 Jan 2017 13:36:45 +0000 (14:36 +0100)]
git-kernel: $(SUBDIR) should always be $(LINUX_VERSION)

Before SUBDIR was set to $(PATCHVER) which may
or may not include the minor version number of
the linux kernel version. Usually it doesn't.

So the git-clone'd linux kernel was packed without
the minor version number taken into account, which
broke further processing, as it expected the
extracted dir being named linux-$(LINUX_VERSION)
(=with minor version) rather than linux-$(PATCHVER)
(=without minor version).

Changing SUBDIR to $(LINUX_VERSION) creates
consistent behaviour here.

Signed-off-by: Mirko Vogt <mirko-openwrt@nanl.de>
7 years agosdk: do not strip static libraries
Jo-Philipp Wich [Wed, 18 Jan 2017 02:15:31 +0000 (03:15 +0100)]
sdk: do not strip static libraries

Do not strip static libraries shipped with the SDK in order to preserve the
archive index. If we strip the index of the shipped libraries, host programs
will fail to link these libraries with errors like:

    libssl.a: error adding symbols: Archive has no index; run ranlib to add one

The error was found while investigating a Python host build failure within
the SDK environment.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agonetifd: update to git HEAD version
Hans Dedecker [Tue, 17 Jan 2017 13:00:02 +0000 (14:00 +0100)]
netifd: update to git HEAD version

a057f6e device: fix DEV_OPT_SENDREDIRECTS definition

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agocurl: fix HTTPS network timeouts with OpenSSL
Stijn Segers [Tue, 10 Jan 2017 18:57:47 +0000 (19:57 +0100)]
curl: fix HTTPS network timeouts with OpenSSL

Backport an upstream change to fix HTTPS timeouts with OpenSSL.
Upstream curl bug #1174.

Signed-off-by: Stijn Segers <francesco.borromini@inventati.org>
[Jo-Philipp Wich: reword commit message, rename patch to 001-*]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoLEDE v17.01: set branch defaults
Jo-Philipp Wich [Mon, 16 Jan 2017 17:56:07 +0000 (18:56 +0100)]
LEDE v17.01: set branch defaults

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobase-files: add ARCH_PACKAGES to openwrt_release and os-release
Matthias Schiffer [Mon, 16 Jan 2017 12:25:14 +0000 (13:25 +0100)]
base-files: add ARCH_PACKAGES to openwrt_release and os-release

Knowing the package architecture at runtime can be useful, e.g. to
configure opkg repository URLs. The value of ARCH_PACKAGES ("%A" in
VERSION_SED) as added to openwrt_release (as DISTRIB_ARCH) and os-release
(as LEDE_ARCH).

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoubox: turn logd into a separate package
Andrej Vlasic [Sun, 15 Jan 2017 18:30:53 +0000 (19:30 +0100)]
ubox: turn logd into a separate package

Currently system log is always included as a part of ubox. Add logd as a
seperate package and add it to default packages list.

Signed-off-by: Andrej Vlasic <andrej.vlasic@sartura.hr>
Signed-off-by: Luka Perkov <luka.perkov@sartura.hr>
7 years agombedtls: add static files in staging_dir
Domagoj Pintaric [Sun, 1 Jan 2017 18:10:38 +0000 (19:10 +0100)]
mbedtls: add static files in staging_dir

Signed-off-by: Domagoj Pintaric <domagoj.pintaric@sartura.hr>
Signed-off-by: Luka Perkov <luka.perkov@sartura.hr>
7 years agokernel: make ledtrig-netdev use a work queue for updates
Felix Fietkau [Sat, 14 Jan 2017 21:29:05 +0000 (22:29 +0100)]
kernel: make ledtrig-netdev use a work queue for updates

This fixes hangs in igb that happen if the update call interrupts an
already existing dev_get_stats call. In that case the calling CPU
deadlocks because it's trying to acquire the same spinlock recursively.

Signed-off-by: Felix Fietkau <nbd@nbd.name>