openwrt/staging/nbd.git
8 years agomt76: update to the latest version
Felix Fietkau [Thu, 25 Aug 2016 10:59:44 +0000 (12:59 +0200)]
mt76: update to the latest version

Adds client + ad-hoc mode fixes and some initial TPC preparation work

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoar71xx: disable flow control to the built-in switch on AR934x
Felix Fietkau [Thu, 25 Aug 2016 09:53:50 +0000 (11:53 +0200)]
ar71xx: disable flow control to the built-in switch on AR934x

It apparently causes a regression on some devices if the ethernet cable
is plugged in while booting.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agolantiq: fix ath5k EEPROM loading
Mathias Kresin [Tue, 23 Aug 2016 19:14:36 +0000 (21:14 +0200)]
lantiq: fix ath5k EEPROM loading

With 12fe4b579801ea812b64fc7e689716cd39c895ec I switched the ath5k
eeprom extraction to an alternate code path. Unfortunately this code
seams to be broken since ages and broke the ath5k EEPROM extraction.

Reported-by: Mohammed Berdai <mohammed.berdai@gmail.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
Acked-by: John Crispin <john@phrozen.org>
8 years agolinux/arc*: disable MAC frame filter in DW GMAC
Alexey Brodkin [Wed, 24 Aug 2016 12:44:04 +0000 (15:44 +0300)]
linux/arc*: disable MAC frame filter in DW GMAC

DW GMAC on ARC SDP boards doesn't enter promiscuous mode if
Ethernet PHY haven't got established link. Good examples are
auto-negotiation in progress or disconnected cable.

We do see Linux kernel sets GMAC's MAC filter register
properly but GMAC's hardware doesn't accept new settings.

We believe it is a hardware issue, most probably problem of
integration of DW GMAC and PHY on the board.

As a work-around we completely disable frame filtering
in GMAC hardware (once and for good) which forces GMAC to enter
promiscuous mode with the first write to MAC filter register.
That gives us working bridge that consists of eth0 and wlan0
(USB Wi-Fi dongle). I.e. we finally have working "Dumb AP" setup made
of ARC AXS10x boards.

Given that hack is quite dirty (in loaded wired networks this will
effectively load CPU with junk packets even if user doesn't need promisc
mode) and there's no indication any other boards with DW GMAC suffer
from the same issue we're patching only kernel for ARC boards.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: John Crispin <john@phrozen.org>,
Cc: Felix Fietkau <nbd@nbd.name>
8 years agouqmi: update to the latest version, adds QMI-in-MBIM support
Felix Fietkau [Tue, 23 Aug 2016 11:39:29 +0000 (13:39 +0200)]
uqmi: update to the latest version, adds QMI-in-MBIM support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoipq806x: fix ipq8065 s2a/s2b regulator frequency (fixes ethernet performance)
Roman Yeryomin [Fri, 19 Aug 2016 10:56:26 +0000 (13:56 +0300)]
ipq806x: fix ipq8065 s2a/s2b regulator frequency (fixes ethernet performance)

Signed-off-by: Roman Yeryomin <roman@advem.lv>
8 years agoipq806x: Archer C2600: Renaming LED accordance with the standard
Cezary Jackiewicz [Tue, 23 Aug 2016 17:26:16 +0000 (19:26 +0200)]
ipq806x: Archer C2600: Renaming LED accordance with the standard

Signed-off-by: Cezary Jackiewicz <cezary.jackiewicz@gmail.com>
8 years agoipq806x: enable ondemand governor
John Crispin [Tue, 23 Aug 2016 22:18:54 +0000 (00:18 +0200)]
ipq806x: enable ondemand governor

Signed-off-by: John Crispin <john@phrozen.org>
8 years agokernel: backport patch for specifying USB ports/devices in DT
Rafał Miłecki [Wed, 24 Aug 2016 06:42:10 +0000 (08:42 +0200)]
kernel: backport patch for specifying USB ports/devices in DT

This allows referencing USB ports/devices that are always present in a
device. This applies to some internal devices, root hub ports and
internal USB hubs.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agoipq806x: rename R7800 device tree file, merge R7500v2 dts addition
Jo-Philipp Wich [Tue, 23 Aug 2016 23:35:22 +0000 (01:35 +0200)]
ipq806x: rename R7800 device tree file, merge R7500v2 dts addition

The R7800 is an IPQ8065, so rename its dts file to reflect that fact.
Also fold the R7500v2 dts addition into the existing 800-devicetree.patch.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agorealview: rename config-default to config-4.4
Stijn Tintel [Tue, 23 Aug 2016 22:43:43 +0000 (01:43 +0300)]
realview: rename config-default to config-4.4

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agorb532: rename config-default to config 4.1
Stijn Tintel [Tue, 23 Aug 2016 22:43:19 +0000 (01:43 +0300)]
rb532: rename config-default to config 4.1

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agoorion: rename config-default to config-3.18
Stijn Tintel [Tue, 23 Aug 2016 22:42:54 +0000 (01:42 +0300)]
orion: rename config-default to config-3.18

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agogemini: rename config-default to config-4.4
Stijn Tintel [Tue, 23 Aug 2016 22:42:12 +0000 (01:42 +0300)]
gemini: rename config-default to config-4.4

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agoarm64: rename config-default to config-4.4
Stijn Tintel [Tue, 23 Aug 2016 22:41:38 +0000 (01:41 +0300)]
arm64: rename config-default to config-4.4

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agoopenvpn: update to 2.3.12
Magnus Kroken [Tue, 23 Aug 2016 22:14:46 +0000 (00:14 +0200)]
openvpn: update to 2.3.12

300-upstream-fix-polarssl-mbedtls-builds.patch has been applied upstream.
Replaced 101-remove_polarssl_debug_call.patch with upstream backport.

Changelog: https://community.openvpn.net/openvpn/wiki/ChangesInOpenvpn23#OpenVPN2.3.12

Signed-off-by: Magnus Kroken <mkroken@gmail.com>
8 years agogemini: rename patches directory to patches-4.4
Stijn Tintel [Tue, 23 Aug 2016 22:08:32 +0000 (01:08 +0300)]
gemini: rename patches directory to patches-4.4

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agoperf: drop sched_getcpu wrapper
Ralph Sennhauser [Tue, 23 Aug 2016 18:35:10 +0000 (20:35 +0200)]
perf: drop sched_getcpu wrapper

Current musl already provides sched_getcpu

Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
8 years agostrace: bump to 4.13
Ralph Sennhauser [Tue, 23 Aug 2016 18:36:15 +0000 (20:36 +0200)]
strace: bump to 4.13

Fixes broken btrfs support in 4.12

Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
8 years agogemini: drop Linux 4.1 support
Jo-Philipp Wich [Mon, 22 Aug 2016 15:35:47 +0000 (17:35 +0200)]
gemini: drop Linux 4.1 support

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agogemini: add Linux 4.4 support
Roman Yeryomin [Sun, 21 Aug 2016 23:36:50 +0000 (02:36 +0300)]
gemini: add Linux 4.4 support

Signed-off-by: Roman Yeryomin <roman@advem.lv>
[Jo-Philipp Wich: change commit title]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agoath9k: Set ATH9K_STATION_STATISTICS when enabling debugging
Toke Høiland-Jørgensen [Tue, 23 Aug 2016 10:36:12 +0000 (12:36 +0200)]
ath9k: Set ATH9K_STATION_STATISTICS when enabling debugging

The ATH9K_STATION_STATISTICS kernel config variable enables some extra
statistics that are useful for debugging (in particular with the airtime
fairness patches enabled). This adds that kernel config when selecting
ath9k debugging.

Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
8 years agoubus: update to the latest version, adds object remove fixes
Felix Fietkau [Tue, 23 Aug 2016 11:29:36 +0000 (13:29 +0200)]
ubus: update to the latest version, adds object remove fixes

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoubox: move logd into ubox package
Felix Fietkau [Fri, 5 Aug 2016 11:06:13 +0000 (13:06 +0200)]
ubox: move logd into ubox package

Preparation for further build rework

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agohostapd: use printf to improve portability.
Ash Benz [Tue, 23 Aug 2016 10:07:01 +0000 (18:07 +0800)]
hostapd: use printf to improve portability.

Signed-off-by: Ash Benz <ash.benz@bk.ru>
8 years agobuild: use perl instead of GNU date for KBUILD_BUILD_TIMESTAMP
Felix Fietkau [Tue, 23 Aug 2016 09:05:56 +0000 (11:05 +0200)]
build: use perl instead of GNU date for KBUILD_BUILD_TIMESTAMP

Fixes warning on non-Linux systems

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agotools: build GNU date from coreutils on non-Linux systems
Felix Fietkau [Tue, 23 Aug 2016 08:58:34 +0000 (10:58 +0200)]
tools: build GNU date from coreutils on non-Linux systems

Required for reproducible builds in the kernel build system

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agobuild: pass $(STAGING_DIR_HOST) to Host/Install
Felix Fietkau [Tue, 23 Aug 2016 08:58:00 +0000 (10:58 +0200)]
build: pass $(STAGING_DIR_HOST) to Host/Install

makes it more consistent with package builds

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoath10k-ct: Update to latest ath10k-ct driver.
Ben Greear [Thu, 11 Aug 2016 18:20:46 +0000 (11:20 -0700)]
ath10k-ct: Update to latest ath10k-ct driver.

This uses GFP_DMA32 for firmware swap.  Fixes issue on x86-64 with
QCA 9984 chipset when host system does not have vt-d enabled.

Also tested on linksys ea8500 with 9980 chipset.

Signed-off-by: Ben Greear <greearb@candelatech.com>
8 years agoramips: mt7621: switch to 24kc
Jonas Gorski [Sun, 21 Aug 2016 17:23:13 +0000 (19:23 +0200)]
ramips: mt7621: switch to 24kc

1004kc is just a SMP capable 34kc, and GCC treats 24kc and 34kc exactly
the same and will generate identical code, so there is no need to tune
to 1004kc instead of 24kc.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agoinclude: remove 34k distinction
Jonas Gorski [Sun, 21 Aug 2016 17:23:12 +0000 (19:23 +0200)]
include: remove 34k distinction

Now that there are no users anymore, drop the 34kc CPU_FLAGS.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agolantiq: switch from 34k to 24k
Jonas Gorski [Sun, 21 Aug 2016 17:23:11 +0000 (19:23 +0200)]
lantiq: switch from 34k to 24k

GCC treats 24kc and 34kc exactly the same and will generate identical
code, so there is no need to tune to 34kc instead of 24kc.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agoar71xx: switch to 24kc
Jonas Gorski [Sun, 21 Aug 2016 17:23:10 +0000 (19:23 +0200)]
ar71xx: switch to 24kc

GCC treats 24kc and 34kc exactly the same and will generate identical
code, so there is no need to tune to 34kc instead of 24kc.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agonetifd: update to the latest version
Felix Fietkau [Tue, 23 Aug 2016 08:16:29 +0000 (10:16 +0200)]
netifd: update to the latest version

Adds fixes for wireless device error handling
Adds link state fixes for shell proto handlers

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agokernel: update kernel 4.4 to version 4.4.19
Stijn Tintel [Mon, 22 Aug 2016 17:05:45 +0000 (19:05 +0200)]
kernel: update kernel 4.4 to version 4.4.19

Refresh patches for all targets that support kernel 4.4.
Compile-tested on all targets that use kernel 4.4 and aren't marked broken.
Runtime-tested on ar71xx, octeon and x86/64.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
8 years agobrcm47xx: add missing config symbol
Stijn Tintel [Mon, 22 Aug 2016 18:57:09 +0000 (21:57 +0300)]
brcm47xx: add missing config symbol

8 years agobcm53xx: backport patch adding ARCH_BCM_53573 symbol
Rafał Miłecki [Tue, 23 Aug 2016 05:10:05 +0000 (07:10 +0200)]
bcm53xx: backport patch adding ARCH_BCM_53573 symbol

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agocyassl: make CyaSSL/WolfSSL more configurable
Andreas Schultz [Sat, 20 Aug 2016 10:52:42 +0000 (12:52 +0200)]
cyassl: make CyaSSL/WolfSSL more configurable

The default configuration might not be suitable for
every use case. Add options to enable/disable additional
options.

Signed-off-by: Andreas Schultz <aschultz@tpip.net>
8 years agoimx6: Fix pointing to msata/mmc bootdevs
Koen Vandeputte [Mon, 22 Aug 2016 15:05:11 +0000 (17:05 +0200)]
imx6: Fix pointing to msata/mmc bootdevs

Point to correct devices when booting from msata or MMC
Changes fetched from Gateworks git

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
[Jo-Philipp Wich: rebased to apply against current LEDE version]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agoipq806x: Build image for Netgear Nighthawk X4 R7500v2
Alexis Green [Fri, 19 Aug 2016 20:56:24 +0000 (13:56 -0700)]
ipq806x: Build image for Netgear Nighthawk X4 R7500v2

Signed-off-by: Alexis Green <alexis@cessp.it>
[Jo-Philipp Wich: add missing DEVICE_TITLE, fix model name in commit title]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agokernel: add fake users for udptunnel and iptunnel modules
Daniel Golle [Mon, 22 Aug 2016 15:14:03 +0000 (17:14 +0200)]
kernel: add fake users for udptunnel and iptunnel modules

Without any in-tree users enabled the Kernel's build process doesn't
actually build those modules. Enable some potential in-tree users
during Kernel build, so out-of-tree modules can depend on them.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 years agooxnas: drop explicit pre-init iface
Daniel Golle [Fri, 19 Aug 2016 16:22:20 +0000 (18:22 +0200)]
oxnas: drop explicit pre-init iface

According to Jonas Gorski this is no longer needed, the problem
originally addressed by first adding 07_preinit_iface and subsequently
merging that into 02_network was probably a bug during the transition
to the new board.d pre-init approach.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 years agokernel: bgmac: use upstream accepted patches
Rafał Miłecki [Mon, 22 Aug 2016 07:47:34 +0000 (09:47 +0200)]
kernel: bgmac: use upstream accepted patches

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agolantiq: fix building AVM/EVA sysupgrade images on NOR devices
Martin Blumenstingl [Sun, 21 Aug 2016 15:01:33 +0000 (17:01 +0200)]
lantiq: fix building AVM/EVA sysupgrade images on NOR devices

Device/AVM unconditionally inherits Device/NAND, which changes the
sysupgrade image to sysupgrade-tar (instead of a "raw" sysupgrade image
as used in the Device/Default template).
This moves the Device/NAND call to the FRITZ3370 device definition, as
this is the only AVM/EVA device that uses NAND.
Additionally we have to define the IMAGE_SIZE for the AVM/EVA NOR
devices to ensure that the images are not discarded.

Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
8 years agoipq806x: add usb pins to r7800 dts file
John Crispin [Mon, 15 Aug 2016 20:54:23 +0000 (22:54 +0200)]
ipq806x: add usb pins to r7800 dts file

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoimx6: fix generating bootfs in imagebuilder (FS#102)
Jo-Philipp Wich [Sat, 20 Aug 2016 18:17:51 +0000 (20:17 +0200)]
imx6: fix generating bootfs in imagebuilder (FS#102)

The image builder does not bundle sub-directories within $(KERNEL_BUILD_DIR),
therfore the intermediate "ventana-uImage.boot" directory is not shipped,
leading to failures with "make image" later on.

To circumvent the issue, store the intermediate boot files as tar archive
instead of putting them into a directory to ensure that they end up in the
final image builder tarball. Fixes FS#102

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agoramips: fix build of TRENDnet TEW-69xGR images
Mathias Kresin [Fri, 19 Aug 2016 23:18:51 +0000 (01:18 +0200)]
ramips: fix build of TRENDnet TEW-69xGR images

The image generation for TEW-691GR and TEW-692GR was broken since
79d02229 due to the move of the UMedia recipe to another Makefile.

Signed-off-by: Mathias Kresin <dev@kresin.me>
8 years agoramips: move umedia build step to Makefile
Mathias Kresin [Fri, 19 Aug 2016 23:17:16 +0000 (01:17 +0200)]
ramips: move umedia build step to Makefile

Some rt3883 based boards need this one as well.

Signed-off-by: Mathias Kresin <dev@kresin.me>
8 years agoramips: TEW-691GR: fix switch and wireless
Mathias Kresin [Fri, 19 Aug 2016 18:46:50 +0000 (20:46 +0200)]
ramips: TEW-691GR: fix switch and wireless

Due to the missing phy-mode setting, the switch wasn't initialised.

The wireless requires an eeprom to work. Use the same mac addresses as
the stock firmware.

Signed-off-by: Mathias Kresin <dev@kresin.me>
8 years agoramips: TEW-692GR: fix switch and wireless
Mathias Kresin [Sat, 13 Aug 2016 22:47:14 +0000 (00:47 +0200)]
ramips: TEW-692GR: fix switch and wireless

The AR8327 initvals were not copied to the DTS during the switch from
mach file to device tree and broke the switch on the device.

The former used PORT6 related initvals were wrong and have been
corrected.

The phy mode setting for the switch was missing in the DTS as well.

The wireless requires an eeprom to work. The dual band wireless chips
have both bands enabled by default but only one band per chip is
working.

The stock firmware uses the following mac addresses:

LAN:    AA:BB:CC:DD:EE:E4
WAN:    AA:BB:CC:DD:EE:E4
2.4GHz: AA:BB:CC:DD:EE:E4
5 GHz:  AA:BB:CC:DD:EE:E8

Assuming the mac address range :E4 to :E8 is reserved for this device,
the MAC addresses were reorder to have a unique MAC address for each
interface:

LAN:    AA:BB:CC:DD:EE:E4
2.4GHz: AA:BB:CC:DD:EE:E4
WAN:    AA:BB:CC:DD:EE:E5
5 GHz:  AA:BB:CC:DD:EE:E8

Signed-off-by: Mathias Kresin <dev@kresin.me>
8 years agoramips: rt3883: fix dtc compiler warning
Mathias Kresin [Tue, 16 Aug 2016 05:39:28 +0000 (07:39 +0200)]
ramips: rt3883: fix dtc compiler warning

Fixes the following compiler warning:

Warning (reg_format): "reg" property in /ethernet@10100000/port@0 has invalid length (4 bytes) (#address-cells == 2, #size-cells == 1)
Warning (avoid_default_addr_size): Relying on default #address-cells value for /ethernet@10100000/port@0
Warning (avoid_default_addr_size): Relying on default #size-cells value for /ethernet@10100000/port@0

Signed-off-by: Mathias Kresin <dev@kresin.me>
8 years agolantiq: generate switch configuration for BTHOMEHUBV2B.
Sigurd Hogsbro [Wed, 17 Aug 2016 23:43:16 +0000 (07:43 +0800)]
lantiq: generate switch configuration for BTHOMEHUBV2B.

Split from HH3A as it appears to have a different switch.
https://en.wikipedia.org/wiki/BT_Home_Hub#Models_and_technical_specifications

Signed-off-by: Sigurd Hogsbro <sigurd@hogsbro.org>
8 years agoarchs38: Merge sd and ramfs subtargets in generic again
Alexey Brodkin [Fri, 19 Aug 2016 10:58:36 +0000 (13:58 +0300)]
archs38: Merge sd and ramfs subtargets in generic again

As it turned out we may actually build both initramfs and
SD-based images in one run with just a small tweak in
existing image/Makefile and so we do here.

We're back with one "generic" target for all ARC HS38-based bords.

Kudos to John for his patience and very helpful tips!

Signed-off-by: John Crispin <john@phrozen.org>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Jo-Philipp Wich <jo@mein.io>
8 years agoltq-deu: fix cra_priority
Martin Schiller [Thu, 18 Aug 2016 10:03:17 +0000 (12:03 +0200)]
ltq-deu: fix cra_priority

With the default priority of 0, the DEU algos would be overlapped
by the generic algos (if available).

To fix this, set the cra_priority of the hardware algos to the
recommended value of 300/400.

Signed-off-by: Martin Schiller <mschiller@tdt.de>
8 years agoltq-deu: fix handling of data blocks with sizes != AES/DES block size
Martin Schiller [Thu, 18 Aug 2016 10:03:16 +0000 (12:03 +0200)]
ltq-deu: fix handling of data blocks with sizes != AES/DES block size

This fix is a backport from the lantiq UGW-6.1.1-MR1

Signed-off-by: Martin Schiller <mschiller@tdt.de>
8 years agoltq-deu: fix aes initialization vector handling
Martin Schiller [Thu, 18 Aug 2016 10:03:15 +0000 (12:03 +0200)]
ltq-deu: fix aes initialization vector handling

This fix is a backport from the lantiq UGW-6.1.1-MR1

Signed-off-by: Martin Schiller <mschiller@tdt.de>
8 years agotoolchain: include yasm in x86 toolchain
Daniel Golle [Fri, 19 Aug 2016 15:49:09 +0000 (17:49 +0200)]
toolchain: include yasm in x86 toolchain

Some libraries require yasm to build with in-line assembly for x86
targets.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 years agooxnas: set preinit interface in /etc/board.d/02_network
Daniel Golle [Mon, 15 Aug 2016 00:47:39 +0000 (02:47 +0200)]
oxnas: set preinit interface in /etc/board.d/02_network

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 years agokernel: replace cosmetic UBIFS patches with what went upstream
Daniel Golle [Thu, 11 Aug 2016 22:54:41 +0000 (00:54 +0200)]
kernel: replace cosmetic UBIFS patches with what went upstream

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 years agobcm53xx: update copy of ASM entry flushing whole D-cache
Rafał Miłecki [Fri, 19 Aug 2016 10:44:44 +0000 (12:44 +0200)]
bcm53xx: update copy of ASM entry flushing whole D-cache

Previous version was based on code from kernel 2.6.22 with Broadcom two
trivial modifications. This updates the copy to the version from current
kernel and refreshes the patch.

This was tested for regressions on Netgear R6250 (BCM4708A0), D-Link
DIR-885L (BCM4709C0) and Tenda AC9 (BCM47189B0).

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agobcm53xx: build Tenda AC9 image
Rafał Miłecki [Fri, 19 Aug 2016 09:36:31 +0000 (11:36 +0200)]
bcm53xx: build Tenda AC9 image

We finally got kernel booting on this device and LAN ports are working,
so it won't "brick" the device. One more big missing thing is WAN port
support. It requires fix for BCM53125 to get switch port 5 working.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agobcm53xx: switch back to standalone ASM entry flushing cache
Rafał Miłecki [Fri, 22 Jul 2016 15:26:33 +0000 (17:26 +0200)]
bcm53xx: switch back to standalone ASM entry flushing cache

Over a year ago in a commit ac96a1665ac7 ("bcm53xx: update Disable MMU
and Dcache during decompression") we switched to Florian's patch for
workarounding CFE bug. The main difference was using a part of existing
__armv7_mmu_cache_flush instead of implementing flushing separately.

This worked well for Northstar devices but doesn't work for BCM53573 as
these devices simply don't start booting with Florian's patch. It's
because of the ldmfd ASM instruction in the __armv7_mmu_cache_flush.

So this commit switches back to using standalone implementation. This
time instead of copying Broadcom's copy of cache-v7.S, we just make a
copy of the original file on our own. Unfortunately we can't cross-dir
compile cache-v7.S from ../../mm/ as that one also adds __INITDATA with
define_cache_functions v7 which would just trigger
> Error: unrecognized/unsupported machine ID (r1 = 0x0000007f).

The only real change Broadcom did in copied .S file was modifying mcr
instruction to use c6 instead of c14. It isn't clear to me if we really
need it, but let's use it for now.

By the way we also update commit message of the
[PATCH] ARM: BCM5301X: Disable MMU and Dcache during decompression

This makes kernel booting on BCM53573 successfully.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agobcm53xx: package drivers for Northstar USB PHYs
Rafał Miłecki [Sun, 14 Aug 2016 14:00:10 +0000 (16:00 +0200)]
bcm53xx: package drivers for Northstar USB PHYs

We don't use them yet as we still init PHY in controller driver but they
will be needed soon.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: check the right directories for rebuild
Jonas Gorski [Fri, 19 Aug 2016 08:57:30 +0000 (10:57 +0200)]
kernel: check the right directories for rebuild

Pass KERNEL_FILE_DEPENDS to rdep instead of PKG_FILE_DEPENDS, which is
empty. Also don't pass $(CURDIR) as the directory to timestamp, as it
would also pick up non kernel related changes like image building code.

Should fix kernel being rebuild for unrelated changes, as well as not
being rebuild for changes in target/linux/generic.

Fixes: 22ef1c83b35c ("kernel: make the kernel build auto-clean the build dir like package build")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agotools: Select dosfstools for archs38
Alexey Brodkin [Fri, 19 Aug 2016 04:58:11 +0000 (07:58 +0300)]
tools: Select dosfstools for archs38

In commit df4f41261cf9 ("archs38: Introduce images for SD-cards")
we introduced building of SD-card images for ARC HS38-based boards.

While building images mkdosfs utility is used.
On machines I used for testing mentioned change this utility was
already installed so I didn't figure-out that requirement.

But thanks to Lede's autobuilder this missing bit was highlighted,
see failed build job here:
http://phase1.builds.lede-project.org/builders/archs38%2Fsd/builds/0/steps/images/logs/stdio

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: John Crispin <john@phrozen.org>
Cc: Jo-Philipp Wich <jo@mein.io>
8 years agobusybox: enable sha256sum by default
Jo-Philipp Wich [Thu, 18 Aug 2016 13:06:56 +0000 (15:06 +0200)]
busybox: enable sha256sum by default

Now that snapshot builds are only publishing SHA-256 checksums, it makes
sense to ship an appropriate utility for verification.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agobcm53xx: generate proper network config for Tenda AC9
Rafał Miłecki [Thu, 18 Aug 2016 08:39:34 +0000 (10:39 +0200)]
bcm53xx: generate proper network config for Tenda AC9

It's a more complex device with 2 interfaces on the SoC, one we can use
for LAN and another for WAN.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agomvebu: fix boot stalls on WRT1900AC V1
John Crispin [Thu, 18 Aug 2016 07:38:26 +0000 (09:38 +0200)]
mvebu: fix boot stalls on WRT1900AC V1

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoppp: Extend uci datamodel with persistency sypport
Hans Dedecker [Thu, 14 Jul 2016 08:35:40 +0000 (10:35 +0200)]
ppp: Extend uci datamodel with persistency sypport

PPP daemon can be put into persist mode meaning the
daemon will not exit after a connection gets terminated
but will instead try to reopen the connection.
The re-initiation after the link has been terminated
can be controlled via holdoff; this is helpfull in
scenarios where a BRAS is in denial of service mode
due to link setup requests after a BRAS has gone down

Following uci parameters have been added :
persist (boolean) : Puts the ppp daemon in persist mode
maxfail (integer) : Number of consecutive fail attempts which
puts the PPP daemon in exit mode
holdoff (interget) : Specifies how many seconds to wait
before re-initiating link setup after it has been terminated

Signed-off-by: Alin Nastac <alin.nastac@gmail.com>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
8 years agoar71xx: add support for Zbtlink ZBT-WE1526
Piotr Dymacz [Wed, 17 Aug 2016 22:05:03 +0000 (00:05 +0200)]
ar71xx: add support for Zbtlink ZBT-WE1526

Zbtlink ZBT-WE1526 is based on Qualcomm Atheros QCA9531 v2.
Short specification:

- 650/400/200 MHz (CPU/DDR/AHB)
- 5x 10/100 Mbps Ethernet
- 1x USB 2.0
- 128 MB of RAM (DDR2)
- 16 MB of FLASH
- 2T2R 2.4 GHz, up to 22 dBm
- two external, non-detachable antennas
- 8x LED, 1x button
- UART header (pinout: VCC, RX, TX, GND)

Flash instruction:

Use sysupgrade in vendor firmare which is based on OpenWrt.

Signed-off-by: Piotr Dymacz <pepe2k@gmail.com>
8 years agokernel: prevent adding custom string to localversion
Josua Mayer [Wed, 17 Aug 2016 14:18:04 +0000 (16:18 +0200)]
kernel: prevent adding custom string to localversion

When the kernel build picks up a localversion file in the source tree,
that string is unconditionally appended to LOCALVERSION and affects the
uname string.
Make sure to delete any such file.

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
8 years agokernel: prevent addition of scm marker to localversion
Josua Mayer [Wed, 17 Aug 2016 14:09:46 +0000 (16:09 +0200)]
kernel: prevent addition of scm marker to localversion

When building the kernel from a git repository, the kernel build appends
either a + or a short commit hash to localversion.
This behaviour can be prevented by passing the empty LOCALVERSION variable
to make.

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
8 years agomtd: fix building with glibc
Josua Mayer [Wed, 17 Aug 2016 08:27:12 +0000 (10:27 +0200)]
mtd: fix building with glibc

src/linksys_bootcount.c misses to include stdint.h.
Apparently musl doesn't mind and includes this header by default,
but glibc does not and causes the build to fail.

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
8 years agou-boot-envtools: fix building with glibc
Josua Mayer [Fri, 24 Jun 2016 10:46:49 +0000 (12:46 +0200)]
u-boot-envtools: fix building with glibc

tools/env/fw_env.c misses to include stdint.h.
Apparently musl doesn't mind and includes this header by default,
but glibc does not and causes the build to fail.

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
8 years agoipq806x: sync with latest patches sent by QCA
John Crispin [Thu, 11 Aug 2016 14:34:08 +0000 (16:34 +0200)]
ipq806x: sync with latest patches sent by QCA

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoubox: fixes segfault inside kmodloader
John Crispin [Thu, 18 Aug 2016 07:00:49 +0000 (09:00 +0200)]
ubox: fixes segfault inside kmodloader

null pointer deref when no modules folder was present

Signed-off-by: John Crispin <john@phrozen.org>
8 years agokernel: bgmac: add support for BCM53573
Rafał Miłecki [Thu, 18 Aug 2016 04:59:09 +0000 (06:59 +0200)]
kernel: bgmac: add support for BCM53573

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: bgmac: fix regression in MII registration
Rafał Miłecki [Thu, 18 Aug 2016 04:52:04 +0000 (06:52 +0200)]
kernel: bgmac: fix regression in MII registration

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: bgmac: backport 3 remaining small fixes
Rafał Miłecki [Wed, 17 Aug 2016 10:54:49 +0000 (12:54 +0200)]
kernel: bgmac: backport 3 remaining small fixes

This makes bgmac in sync with upstream verison.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: bgmac: backport patch adding platform support
Rafał Miłecki [Wed, 17 Aug 2016 07:30:44 +0000 (09:30 +0200)]
kernel: bgmac: backport patch adding platform support

It also reworks bcma support by moving specific code to separated file.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: bgmac: backport patch switching to feature_flags
Rafał Miłecki [Wed, 17 Aug 2016 05:45:48 +0000 (07:45 +0200)]
kernel: bgmac: backport patch switching to feature_flags

It was needed upstream to avoid bcma references in the main code. To
match this new code our patch adding SRAB was also updated.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agoarchs38: Introduce images for SD-cards
Alexey Brodkin [Tue, 16 Aug 2016 11:49:20 +0000 (14:49 +0300)]
archs38: Introduce images for SD-cards

Historically on ARC we started from initramfs-based images because:
 a) It was much easier to debug especially when toolchain and other
    components were changing quite dynamically
 b) It was our usual approach for embedded Linux

But now with ARC port of Lede/OpenWRT getting more stable and mature
we're ready for more real-life scenarios with FS permanently stored
on SD-card. This essentially benefits from ability to setup devices
that survive reboots with all settings and extra packages kept in place.

Still we keep an ability to build images with initramfs.
This allows us to use storage-less simulators for testing still.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: John Crispin <john@phrozen.org>
8 years agoarchs38: Remove MMC kernel modules, they are built-in now
Alexey Brodkin [Tue, 16 Aug 2016 11:49:19 +0000 (14:49 +0300)]
archs38: Remove MMC kernel modules, they are built-in now

If we want to boot from SD-card we need to have corresponding
drivers already built-in so there's no point in having these
modules.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agoarchs38: Update kernel configuration
Alexey Brodkin [Tue, 16 Aug 2016 11:49:18 +0000 (14:49 +0300)]
archs38: Update kernel configuration

Now when we're switching to FS on SD-card it's necessary to have
full stack of MMC block & FC drivers built-in otherwise kernel won't
be able to mount FS with needed modules.

Also we enable parsing of input parameters passed to the kernel by
U-Boot. Otherwise kernel won't know where to look for command line and
what's worse device tree blob (we had to disable this by default for
cases when kernel is loaded by JTAG and core registers may have
undefined state lading to kernel going bonkers).

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agoarchs38: Reduce generalization
Alexey Brodkin [Tue, 16 Aug 2016 11:49:17 +0000 (14:49 +0300)]
archs38: Reduce generalization

As support of ARC HS38 in OpenWRT/Lede matures we don't need
debug-only output binaries any longer, so purging vmlinux for
AXS10x boards.

As for uImage for nSIM it makes completely no sense because there's no
way to run U-Boot on nSIM.

So we remove add_archs38_XXX scripts making code more compact and
cleaner.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agoarc: use patched .dts from sources
Alexey Brodkin [Tue, 16 Aug 2016 11:49:16 +0000 (14:49 +0300)]
arc: use patched .dts from sources

Instead of using off-the-tree .dts files for ARC boards we're
switching to use in-tree ones. And for that to work properly
we apply upstream patch that adds currently missing "model"
property.

Upstream patch and discussion could be found here:
http://lists.infradead.org/pipermail/linux-snps-arc/2016-August/001394.html

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Jonas Gorski <jonas.gorski@gmail.com>
Cc: John Crispin <john@phrozen.org>
8 years agotools/tar: Bump to 1.29
Alexey Brodkin [Tue, 16 Aug 2016 10:44:05 +0000 (13:44 +0300)]
tools/tar: Bump to 1.29

This change updates tar from 1.28 to 1.29.
Among other changes following commit
http://git.savannah.gnu.org/cgit/tar.git/commit/lib/xattr-at.c?h=9c2b57232e3bc2e5ba85387560bcdd851849a128
substitutes previously used off-the-tree patch 100-fix_xattr_disable.patch

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agoath9k: switch to using mac80211 intermediate software queues
Felix Fietkau [Fri, 8 Jul 2016 16:08:01 +0000 (18:08 +0200)]
ath9k: switch to using mac80211 intermediate software queues

Provides a nice latency reduction under load, due to mac80211's fq_codel
support.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agokernel: backport bgmac patch moving MDIO code into separated file
Rafał Miłecki [Tue, 16 Aug 2016 22:06:14 +0000 (00:06 +0200)]
kernel: backport bgmac patch moving MDIO code into separated file

This prepares bgmac for the biggest change: dropping bcma dependency.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: backport simple bgmac cleanup & fix from Jon
Rafał Miłecki [Tue, 16 Aug 2016 21:24:26 +0000 (23:24 +0200)]
kernel: backport simple bgmac cleanup & fix from Jon

These 2 trivial patches will be followed by a bigger bgmac rework so
they are worth backporting in a separated patch.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: add BCM53573 flash fix sent upstream
Rafał Miłecki [Tue, 16 Aug 2016 20:33:21 +0000 (22:33 +0200)]
kernel: add BCM53573 flash fix sent upstream

This fixes flash corruptions. It touches bcm47xxsflash driver shared
between brcm47xx and bcm53xx so put it in generic.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: update bgmac by adding Florian's upstream changes
Rafał Miłecki [Tue, 16 Aug 2016 15:40:32 +0000 (17:40 +0200)]
kernel: update bgmac by adding Florian's upstream changes

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agoswconfig: revert the portmapping patches, they seem to cause a segfault
John Crispin [Tue, 16 Aug 2016 08:19:13 +0000 (10:19 +0200)]
swconfig: revert the portmapping patches, they seem to cause a segfault

Revert "kernel/swconfig: remove obsolete portmapping feature from swconfig"

This reverts commit 675407baa44a8700de20b6b2857009a552a807ba.

Revert "swconfig: remove obsolete portmapping feature"

This reverts commit fca1eb349ef31b133a62880cbd562d6bf17500aa.

Signed-off-by: John Crispin <john@phrozen.org>
8 years agobcm53xx: fix warning caused by m25p80 patch
Rafał Miłecki [Tue, 16 Aug 2016 15:11:09 +0000 (17:11 +0200)]
bcm53xx: fix warning caused by m25p80 patch

It was adding unused variable:
drivers/mtd/devices/m25p80.c: In function 'm25p80_write':
drivers/mtd/devices/m25p80.c:85:6: warning: unused variable 'i' [-Wunused-variable]
  int i;
      ^

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agokernel: support watchdog (and so reboot) on BCM53573
Rafał Miłecki [Tue, 16 Aug 2016 14:53:11 +0000 (16:53 +0200)]
kernel: support watchdog (and so reboot) on BCM53573

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 years agolantiq/xrx200-net: Add support for eth0 as WAN interface
Martin Schiller [Tue, 16 Aug 2016 04:53:58 +0000 (06:53 +0200)]
lantiq/xrx200-net: Add support for eth0 as WAN interface

Use priv->wan instead of priv->id as indicator if packets should go to the
Ethernet WAN group (DPID=1) or not (DPID=0). This way, it's independant of
interface names or indexes.

Signed-off-by: Martin Schiller <mschiller@tdt.de>
8 years agokernel: add kmod-squashfs
Matteo Croce [Thu, 11 Aug 2016 14:11:39 +0000 (16:11 +0200)]
kernel: add kmod-squashfs

add kernel package to build squashfs as module when it's not the root filesystem

Signed-off-by: Matteo Croce <matteo.croce@canonical.com>
8 years agolantiq: add kmod-ledtrig-usbdev to device packages which have usb led
Matti Laakso [Sat, 13 Aug 2016 12:07:48 +0000 (15:07 +0300)]
lantiq: add kmod-ledtrig-usbdev to device packages which have usb led

By default USB LEDs are associated with usbdev trigger. However,
not all devices which have USB LEDs have kmod-ledtrig-usbdev in their
list of default packages. Add it to the relevant devices.

Signed-off-by: Matti Laakso <matti.laakso@outlook.com>
8 years agolibs/gmp: update to 6.1.1
Hannu Nyman [Mon, 15 Aug 2016 08:26:55 +0000 (11:26 +0300)]
libs/gmp: update to 6.1.1

Update libgmp to 6.1.1

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
8 years agotools/gmp: update to 6.1.1
Hannu Nyman [Mon, 15 Aug 2016 08:26:13 +0000 (11:26 +0300)]
tools/gmp: update to 6.1.1

Update gmp to 6.1.1

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>