openwrt/staging/aparcar.git
6 years agozlib: only enable NEON optimizations on eligible targets
Jo-Philipp Wich [Thu, 4 Jan 2018 13:43:34 +0000 (14:43 +0100)]
zlib: only enable NEON optimizations on eligible targets

Instead of inferring the availability of NEON support from the target
optimization flags, use a preprocessor test to decide whether to enable
ARMv8 NEON optimizations.

Fixes the following build error spotted by the mediatek/32 buildbot:

    [ 26%] Building C object CMakeFiles/zlib.dir/contrib/arm/inflate.o
    In file included from .../zlib-1.2.11/contrib/arm/chunkcopy.h:10:0,
                     from .../zlib-1.2.11/contrib/arm/inflate.c:87:
    .../arm_neon.h:31:2: error: #error You must enable NEON instructions (e.g. -mfloat-abi=softfp -mfpu=neon) to use arm_neon.h
     #error You must enable NEON instructions (e.g. -mfloat-abi=softfp -mfpu=neon) to use arm_neon.h
      ^
    In file included from .../zlib-1.2.11/contrib/arm/inflate.c:87:0:
    .../zlib-1.2.11/contrib/arm/chunkcopy.h:18:9: error: unknown type name 'uint8x16_t'
     typedef uint8x16_t chunkcopy_chunk_t;
             ^
    [...]
    CMakeFiles/zlib.dir/build.make:302: recipe for target 'CMakeFiles/zlib.dir/contrib/arm/inflate.o' failed

Fixes: 3acecba520 "package/libs/zlib: Add ARM and NEON optimizations"
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agohostapd: set group_mgmt_cipher when ieee80211w is enabled
Jo-Philipp Wich [Fri, 5 Jan 2018 12:15:01 +0000 (13:15 +0100)]
hostapd: set group_mgmt_cipher when ieee80211w is enabled

In order to properly support 802.11w, hostapd needs to advertise a group
management cipher when negotiating associations.

Introduce a new per-wifi-iface option "ieee80211w_mgmt_cipher" which
defaults to the standard AES-128-CMAC cipher and always emit a
"group_mgmt_cipher" setting in native hostapd config when 802.11w is
enabled.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agoargp-standalone: clean up patch fuzz
Kevin Darbyshire-Bryant [Thu, 4 Jan 2018 13:50:10 +0000 (13:50 +0000)]
argp-standalone: clean up patch fuzz

Refresh patches to tidy up fuzz.  No functional changes

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agolua: clean up patch fuzz
Kevin Darbyshire-Bryant [Thu, 4 Jan 2018 20:07:59 +0000 (20:07 +0000)]
lua: clean up patch fuzz

Refresh patches to tidy up fuzz.  No functional changes

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agobusybox: clean up patch fuzz
Kevin Darbyshire-Bryant [Thu, 4 Jan 2018 21:22:38 +0000 (21:22 +0000)]
busybox: clean up patch fuzz

Refresh patches to tidy up fuzz.  No functional changes

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agolantiq: fix wbmr-hp-g300h image metadata
Mathias Kresin [Thu, 4 Jan 2018 20:18:28 +0000 (21:18 +0100)]
lantiq: fix wbmr-hp-g300h image metadata

The Annex A and Annex B version are using the same (old) userspace
boardname. Update the SUPPORTED_DEVICES to allow an update from lede-17.01.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoscripts/package-metadata.pl: remove broken detection of self-dependencies
Matthias Schiffer [Thu, 4 Jan 2018 16:18:16 +0000 (17:18 +0100)]
scripts/package-metadata.pl: remove broken detection of self-dependencies

A self-dependency is not an error worth a warning; rather, it is very
common: whenever there are dependencies between different binary packages
originating from the same source package, such dependencies occur. Not
actually generating dependency rules is correct, but already handled a few
lines below.

A typo prevented this redundant rule from working, which is the reason the
warning was not actually printed.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agoopkg: switch source url to git.openwrt.org
Jo-Philipp Wich [Thu, 4 Jan 2018 10:15:45 +0000 (11:15 +0100)]
opkg: switch source url to git.openwrt.org

As LEDE is rebranding to OpenWrt now, adjust the Git source references
accordingly.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agosdk: switch base feed url to git.openwrt.org
Jo-Philipp Wich [Thu, 4 Jan 2018 10:02:01 +0000 (11:02 +0100)]
sdk: switch base feed url to git.openwrt.org

As LEDE is rebranding to OpenWrt now, adjust the base feed Git reference
accordingly.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agox86: preinit: use board_name on Sm Super Servers
Philip Prindeville [Mon, 30 Oct 2017 18:32:46 +0000 (12:32 -0600)]
x86: preinit: use board_name on Sm Super Servers

Supermicro puts "Super Server" into their product_name DMI value
for a whole slew of products, making this value about as useful
as not having been filled in at all.  Instead, fall back on the
board_name instead.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agox86: preinit: make name rewrite into reusable function
Philip Prindeville [Mon, 30 Oct 2017 18:31:04 +0000 (12:31 -0600)]
x86: preinit: make name rewrite into reusable function

There might be other places (such as vendor-supplied preinit scripts)
where we wish to take a DMI name and clean it up in a consistent way,
so make the sed command into a function.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agonetifd: update to latest git HEAD
Hans Dedecker [Thu, 4 Jan 2018 08:47:07 +0000 (09:47 +0100)]
netifd: update to latest git HEAD

fd5c399 proto: allow dumping protocol handlers without config_params

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agomerge: github: use OpenWrt in issue/pr templates
Stijn Tintel [Wed, 3 Jan 2018 19:33:53 +0000 (20:33 +0100)]
merge: github: use OpenWrt in issue/pr templates

Keep the link to submitting patches for now.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agokernel: bump 4.14 to 4.14.11
Stijn Tintel [Tue, 2 Jan 2018 21:12:28 +0000 (22:12 +0100)]
kernel: bump 4.14 to 4.14.11

Rename unwinder config symbols to match upstream changes.
Refresh patches.
Update patch that no longer applies: 202-reduce_module_size.patch

Also enable CONFIG_PAGE_TABLE_ISOLATION. This feature was backported
from 4.15 to the 4.14 stable series. It is enabled by default, so enable
it in OpenWrt as well.

Compile-tested on x86/64.
Runtime-tested on x86/64.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agokernel: move CONFIG_KASAN to generic config
Stijn Tintel [Tue, 2 Jan 2018 17:53:04 +0000 (18:53 +0100)]
kernel: move CONFIG_KASAN to generic config

While bumping 4.14, the kernel build failed due to missing CONFIG_KASAN
symbol. Move it to generic config instead of defining it for all arm64
and x86/64 targets.

It was only added in 4.0, so not needed in config-3.18.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agouci: add missing 'option' support to uci_rename()
Tony Ambardar [Tue, 25 Apr 2017 09:16:56 +0000 (02:16 -0700)]
uci: add missing 'option' support to uci_rename()

When using the uci.sh wrapper, allow parameters to match those supported
by the uci binary i.e. "uci rename <config>.<section>[.<option>]=<name>".

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
6 years agopackage/libs/zlib: Add host build
Daniel Engberg [Tue, 12 Dec 2017 14:31:59 +0000 (15:31 +0100)]
package/libs/zlib: Add host build

Some packages such as Python/Python3 (host pip/pip3) needs this
to compile.

More detailed explanation provided by Alexandru:

"i need the zlib/host for Python/Python3 ; because, it seems the
host pip/pip3 needs this to work ; i suspect in older versions
this worked, because some of the host's build env would be used
in the build, and then the zlib-dev from the host distro would
be used ; now, the host-build does not seem to have any
-I/usr/include stuff, which is good

and it also seems that Python/Python3 does not like it if the
zlib-dev package is too old, so using this zlib/host would be
good for this as well"

Source:
https://github.com/lede-project/source/pull/1329#issuecomment-351055861

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
6 years agopackage/libs/zlib: Add option for O3 optimization
Daniel Engberg [Tue, 5 Sep 2017 17:31:10 +0000 (19:31 +0200)]
package/libs/zlib: Add option for O3 optimization

Add option to use O3 optimization as not all devices have
space constraints. This option is default using GCC in upstream
but isn't in the CMake makefile for some reason.

Source: https://github.com/madler/zlib/blob/master/configure#L170

Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
6 years agopackage/libs/zlib: Add ARM and NEON optimizations
Daniel Engberg [Tue, 5 Sep 2017 18:38:07 +0000 (20:38 +0200)]
package/libs/zlib: Add ARM and NEON optimizations

This adds two optimizations for ARM:
NEON optimized Adler(-)32 checksum algorithm (ARMv7 and newer NEON CPUs)
ARM(v7+) specific optimization for inflate
I've also connected inflate optimization to the build using the following
source as template.
https://github.com/mirror/chromium/commit/0397489124ce7e6aced020f8b85f5034c7d5f49b#diff-a62ad2db6c83dbc205d34bb9a8884f16

Additional info:
https://codereview.chromium.org/2676493007/
https://codereview.chromium.org/2722063002/

Sources:
https://github.com/madler/zlib/pull/251 (only the first commit)
https://github.com/madler/zlib/pull/256

Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
6 years agopackage/libs/zlib: Use toolchain build logic
Daniel Engberg [Tue, 5 Sep 2017 17:31:10 +0000 (19:31 +0200)]
package/libs/zlib: Use toolchain build logic

Use build logic provided by toolchain instead of doing it manually.

Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
6 years agoprocd: procd_send_signal use signal names
Kevin Darbyshire-Bryant [Tue, 2 Jan 2018 14:34:51 +0000 (14:34 +0000)]
procd: procd_send_signal use signal names

Usage documentation for 'procd_send_signal' states "The signal is SIGHUP
by default, and must be specified by NAME."  Make actual behaviour match
the stated documented behaviour.

https://wiki.openwrt.org/inbox/procd-init-scripts

Suggested-by: Jo-Philip Wich <jow@mein.io>
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agoprocd: update to latest git HEAD
John Crispin [Tue, 2 Jan 2018 12:04:06 +0000 (13:04 +0100)]
procd: update to latest git HEAD

a5954cf procd: Add %m to several functions that return errno.
810d7a5 procd: Remove redundant errno variable in several printf functions.
fa5ce1c procd: Replace strerror(errno) with %m.

Signed-off-by: John Crispin <john@phrozen.org>
6 years agoumdns: update to latest git HEAD
John Crispin [Tue, 2 Jan 2018 11:58:05 +0000 (12:58 +0100)]
umdns: update to latest git HEAD

7897441 umdnsd: Replace strerror(errno) with %m.

Signed-off-by: John Crispin <john@phrozen.org>
6 years agokernel: make uas work during early boot
Daniel Albers [Thu, 28 Dec 2017 17:36:24 +0000 (18:36 +0100)]
kernel: make uas work during early boot

Add the uas(p) module to the modules loaded early on the boot process.

The uas(p) is an modern alternative, which is used by the modern USB3
storage cases, compared to the bot protocol. To be able to use uas(p)
storage cases for extroot, the kernel module has to be loaded before the
search for extroot has been called. This patch changes the load order to
support uas(p) storage cases for extroot.

Signed-off-by: Daniel Albers <daniel.albers@public-files.de>
6 years agofstools: update to latest git HEAD
John Crispin [Tue, 2 Jan 2018 06:55:41 +0000 (07:55 +0100)]
fstools: update to latest git HEAD

11efbf3 overlay: fix race condition when switching to jffs2
bdeb95a libblkid-tiny: add support for NTFS superblock
ef2cc03 fstools: Replace strerror(errno) with %m format.
98fd5b4 libblkid-tiny: add support for UBI superblock

Signed-off-by: John Crispin <john@phrozen.org>
6 years agobase-files: protect stop and reload actions with procd_lock
Roman Yeryomin [Thu, 14 Dec 2017 21:55:08 +0000 (23:55 +0200)]
base-files: protect stop and reload actions with procd_lock

Signed-off-by: Roman Yeryomin <roman@advem.lv>
Signed-off-by: Andrejs Hanins <ahanins@gmail.com>
6 years agoprocd: introduce procd_lock for init script protection
Roman Yeryomin [Thu, 14 Dec 2017 21:54:26 +0000 (23:54 +0200)]
procd: introduce procd_lock for init script protection

Use flock to protect init script from concurrent execution
(of the same script).
Important for services which generate native config files.

Signed-off-by: Roman Yeryomin <roman@advem.lv>
Signed-off-by: Andrejs Hanins <ahanins@gmail.com>
6 years agobusybox: enable flock by default
Roman Yeryomin [Sun, 17 Dec 2017 18:30:42 +0000 (20:30 +0200)]
busybox: enable flock by default

This is needed for procd init script protection to work.
flock adds 4248 bytes to stripped busybox binary.

Signed-off-by: Roman Yeryomin <roman@advem.lv>
6 years agolayerscape: activate fpu feature
Hauke Mehrtens [Mon, 1 Jan 2018 15:58:33 +0000 (16:58 +0100)]
layerscape: activate fpu feature

The CPU sub type was set to a CPU version with FPU, but the FPU feature
was not activated before, so a soft float toolchain was created.
Activate also the FPU feature to create the correct toolchain.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoat91: sama5: activate fpu
Hauke Mehrtens [Sun, 31 Dec 2017 19:35:40 +0000 (20:35 +0100)]
at91: sama5: activate fpu

This activates neon and VFPv4 support for this target. The CPU support
these feature so also use them.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agofirmware-tools/ptgen: use portable exit codes
Michael Heimpold [Thu, 21 Dec 2017 22:59:30 +0000 (23:59 +0100)]
firmware-tools/ptgen: use portable exit codes

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
6 years agofirmware-tools/ptgen: fix minor coding style issues
Michael Heimpold [Thu, 21 Dec 2017 22:59:28 +0000 (23:59 +0100)]
firmware-tools/ptgen: fix minor coding style issues

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
6 years agokernel: bump 4.9 to 4.9.73
Kevin Darbyshire-Bryant [Sat, 30 Dec 2017 14:22:36 +0000 (14:22 +0000)]
kernel: bump 4.9 to 4.9.73

Refresh patches.

Runtime tested: ar71xx - Archer C7 v2

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agobusybox: backport cve-2017-16544 fix
John Crispin [Mon, 1 Jan 2018 10:46:03 +0000 (11:46 +0100)]
busybox: backport cve-2017-16544 fix

In the add_match function in libbb/lineedit.c in BusyBox through 1.27.2,
the tab autocomplete feature of the shell, used to get a list of filenames
in a directory, does not sanitize filenames and results in executing any
escape sequence in the terminal. This could potentially result in code
execution, arbitrary file writes, or other attacks.

Fixes: FS#1181 - CVE-2017-16544:
Backport the patch from:
https://git.busybox.net/busybox/commit/?id=c3797d40a1c57352192c6106cc0f435e7d9c11e8
https://nvd.nist.gov/vuln/detail/CVE-2017-16544

Signed-off-by: Derek Werthmuller <thewerthfam@gmail.com>
Signed-off-by: John Crispin <john@phrozen.org>
6 years agobase-files: rc.common: fix enable() return code and logic
Roman Yeryomin [Fri, 22 Dec 2017 11:04:46 +0000 (13:04 +0200)]
base-files: rc.common: fix enable() return code and logic

In current state, if there is START but no STOP, enbale()
will return 1 (failure), which is wrong.
Moreover there is no need to check for START/STOP twice.
Instead, add err variable to save success state and
and return it's value.
Also eliminate the need to disable() by using 'ln -sf',
which will first delete the old symlink if one exists.

Changes from v1:
- fixed description

Signed-off-by: Roman Yeryomin <roman@advem.lv>
6 years agoprocd: update to latest git HEAD
John Crispin [Fri, 29 Dec 2017 23:15:29 +0000 (00:15 +0100)]
procd: update to latest git HEAD

7aad940 procd: initd: fix path allocation in early_insmod
006c19c procd: Remove unnecessary memset calls.

Signed-off-by: John Crispin <john@phrozen.org>
6 years agouci: bump to git HEAD version
Jo-Philipp Wich [Mon, 1 Jan 2018 15:40:15 +0000 (16:40 +0100)]
uci: bump to git HEAD version

5beb95d lua: additionally return name when looking up sections
ff33bb2 lua: support extended section notation

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agonftables: fix sha256sum
Hauke Mehrtens [Tue, 26 Dec 2017 22:29:55 +0000 (23:29 +0100)]
nftables: fix sha256sum

The mirror was delivering a file with a different hash.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agomwlwifi: update to version 10.3.4.0 / 2017-12-14
Kabuli Chana [Mon, 18 Dec 2017 01:51:40 +0000 (18:51 -0700)]
mwlwifi: update to version 10.3.4.0 / 2017-12-14

Lastest stamped version

Signed-off-by: Kabuli Chana <newtownBuild@gmail.com>
6 years agokmod-sched-cake: bump to latest bake
Kevin Darbyshire-Bryant [Fri, 22 Dec 2017 13:50:08 +0000 (13:50 +0000)]
kmod-sched-cake: bump to latest bake

Fix overhead accounting error introduced by f33c4d6 refactor
cake_advance_shaper and ack_filter

Symptoms were links running under rate.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agobcm53xx: use iflag=skip_bytes for dd command during sysupgrade
Rafał Miłecki [Sat, 23 Dec 2017 22:14:25 +0000 (23:14 +0100)]
bcm53xx: use iflag=skip_bytes for dd command during sysupgrade

Since BusyBox 1.25.0 dd command supports iflag=skip_bytes which allows
skipping requested amount of bytes without reducing blocksize. Thanks to
this we can leave default blocksize and let dd work more efficiently.

On Netgear R6250 "dd skip=58 iflag=skip_bytes" can be 5 times faster
than "dd bs=58 skip=1" when extracting TRX out of CHK.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agobrcm47xx: image: build firmware for Asus WL-500g Deluxe
Rafał Miłecki [Wed, 27 Dec 2017 07:55:28 +0000 (08:55 +0100)]
brcm47xx: image: build firmware for Asus WL-500g Deluxe

It's a device based on BCM5365P (0x5365 package 0x00). This SoC has
USB 1.1 controller but device has two USB 2.0 parts. They are handled by
PCI-based controllers: 1106:3038 UHCI and 1106:3104 EHCI.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agobcm53xx: add early support for kernel 4.14
Rafał Miłecki [Fri, 29 Dec 2017 22:09:32 +0000 (23:09 +0100)]
bcm53xx: add early support for kernel 4.14

Don't switch to it by default yet as it requires more testing.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agoramips: fix Omega2 compatible string
Mathias Kresin [Wed, 27 Dec 2017 20:56:07 +0000 (21:56 +0100)]
ramips: fix Omega2 compatible string

Change the compatibel from Omega to Omega2 and add the compatible string
where missing.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: fix widora neo diag led
Mathias Kresin [Wed, 27 Dec 2017 21:09:52 +0000 (22:09 +0100)]
ramips: fix widora neo diag led

The diag LED is named widora:orange:wifi and can't be derived from the
boardname.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agolantiq: fix dts compatible string
Thomas Nixon [Fri, 29 Dec 2017 20:04:08 +0000 (20:04 +0000)]
lantiq: fix dts compatible string

This was broken in 7bab49fd ("lantiq: add compatible strings to dts
files"), causing for the dm200 garbled serial output during boot, and
likely other issues.

Signed-off-by: Thomas Nixon <tom@tomn.co.uk>
[fix the compat string of the P2601HNFX as well]
Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoiptables: fix nftables compile issue (FS#711)
rektide de la faye [Thu, 14 Dec 2017 06:29:11 +0000 (01:29 -0500)]
iptables: fix nftables compile issue (FS#711)

Enabling IPTABLES_NFTABLES resulted in an error during build:#
*** No rule to make target '../extensions/libext.a',
needed by 'xtables-compat-multi'."

Comments from Alexander Lochmann and Fedor Konstantinov in FS#711
provided fixes for this build error, allowing iptables to compile.
https://bugs.lede-project.org/index.php?do=details&task_id=711.

This commit updates the Makefile.am xtables_compat_multi_LDFLAGS
and _LDADD, moving linking of extensions to LDFLAGS.

Signed-off-by: rektide de la faye <rektide@voodoowarez.com>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agobcm53xx: make USB packages compatible with kernels 4.13+
Rafał Miłecki [Fri, 29 Dec 2017 16:15:32 +0000 (17:15 +0100)]
bcm53xx: make USB packages compatible with kernels 4.13+

In kernels 4.13+ modules are located in an extra "broadcom" subdir.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agobase-files: fix logic when to show failsafe banner
Matthias Schiffer [Fri, 29 Dec 2017 14:56:01 +0000 (15:56 +0100)]
base-files: fix logic when to show failsafe banner

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Fixes: 8170f280c4 ("base-files: set FAILSAFE in /etc/profile when
/tmp/.failsafe exists")

6 years agobcm53xx: add upstream patch fixing SPI controller driver
Rafał Miłecki [Fri, 29 Dec 2017 14:31:06 +0000 (15:31 +0100)]
bcm53xx: add upstream patch fixing SPI controller driver

That patch fixes handling SPI messages with two writing transfers. It's
important when using e.g. by m25p80 driver which uses one transfer for
opcode and another one for data.

Thanks to that fix we can now drop m25p80 workaround patch. It means one
less hack and also a better flash writing performance as there is no
more data buf copying.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agoag71xx: Fix rx ring buffer stall on small packets flood on qca956x and qca953x.
Vittorio Gambaletta [Sat, 25 Mar 2017 17:08:02 +0000 (18:08 +0100)]
ag71xx: Fix rx ring buffer stall on small packets flood on qca956x and qca953x.

Backported from Code Aurora QSDK

Signed-off-by: Vittorio Gambaletta <openwrt@vittgam.net>
6 years agobase-files: set FAILSAFE in /etc/profile when /tmp/.failsafe exists
Matthias Schiffer [Fri, 29 Dec 2017 13:34:03 +0000 (14:34 +0100)]
base-files: set FAILSAFE in /etc/profile when /tmp/.failsafe exists

Since dropbear clears the environment, FAILSAFE was not set as intended in
failsafe mode. This also broke sysupgrade from failsafe mode over SSH.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agobusybox: add missing TARGET_CPPFLAGS and TARGET_LDFLAGS
Matthias Schiffer [Sun, 10 Dec 2017 17:04:53 +0000 (18:04 +0100)]
busybox: add missing TARGET_CPPFLAGS and TARGET_LDFLAGS

Unconditionally pass TARGET_CPPFLAGS (not passed at all before) and
TARGET_LDFLAGS (passed only in certain non-default configuration before the
Makefile streamlining). Without these flags, hardening options
(PKG_FORTIFY_SOURCE and PKG_RELRO) were not actually applied to busybox.

The addition of these flags increases the size of the stripped busybox
binary by about 6KB (~4KB with fortify headers, ~2KB with "-znow -zrelro")
with the default hardening options PKG_FORTIFY_SOURCE_1 and PKG_RELRO_FULL.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agobusybox: streamline Makefile
Matthias Schiffer [Sun, 10 Dec 2017 17:01:42 +0000 (18:01 +0100)]
busybox: streamline Makefile

Use default Build/Install steps where possible. No binary change in default
configuration, so PKG_RELEASE is not incremented.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agoinclude/package.mk: remove old configured stamps before attempting configuration
Matthias Schiffer [Sun, 10 Dec 2017 14:30:30 +0000 (15:30 +0100)]
include/package.mk: remove old configured stamps before attempting configuration

Some packages, e.g. busybox, explicitly remove old .configured stamps
before attempting configuration, rather than after the actual configuration
step. This seems like a good idea, as there will be no stamp left if
configuration fails. Change generic rules to work like this, so package-
specific rules can be dropped.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agodnsmasq: send procd signal on service reload
Florian Eckert [Tue, 19 Dec 2017 15:11:58 +0000 (16:11 +0100)]
dnsmasq: send procd signal on service reload

Send a SIGHUP signal via procd to the dnsmasq service so the instance(s)
re-read(s) the /tmp/hosts/dhcp config.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
6 years agodnsmasq: rewrite config on host name modification
Florian Eckert [Tue, 19 Dec 2017 14:08:48 +0000 (15:08 +0100)]
dnsmasq: rewrite config on host name modification

If the hostname in /etc/config/system is modified the dnsmasq should also
get triggered to rewrite/reload the config.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
6 years agox86: add NVMe support for motherboards w/ M.2 slot
Philip Prindeville [Fri, 22 Dec 2017 04:10:03 +0000 (21:10 -0700)]
x86: add NVMe support for motherboards w/ M.2 slot

A lot of 1U x86_64 servers have NVMe support, which is lower-power
and higher speed than SSD or CFast drives, etc.  The drivers
required to make this work are trivial.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agokernel: bump 4.4 to 4.4.108
Kevin Darbyshire-Bryant [Tue, 26 Dec 2017 17:58:08 +0000 (17:58 +0000)]
kernel: bump 4.4 to 4.4.108

Refresh patches.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agokernel: bump 4.9 to 4.9.72
Kevin Darbyshire-Bryant [Mon, 25 Dec 2017 17:33:03 +0000 (17:33 +0000)]
kernel: bump 4.9 to 4.9.72

Refresh patches.

Runtime tested: ar71xx - Archer C7 v2

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agolayerscape: add byte_swap.py script for ls-rcw package
Yangbo Lu [Mon, 25 Dec 2017 06:38:11 +0000 (14:38 +0800)]
layerscape: add byte_swap.py script for ls-rcw package

This patch is to add a byte_swap.py script for rcw bin
byte swapping instead of using tclsh.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: fix PKG_BUILD_DIR for packages
Yangbo Lu [Mon, 25 Dec 2017 04:21:26 +0000 (12:21 +0800)]
layerscape: fix PKG_BUILD_DIR for packages

PKG_BUILD_DIR was defined with quoting PKG_VERSION in
layerscape package makefiles. Now PKG_VERSION has been
removed from these makefiles. When PKG_BUILD_DIR quotes
PKG_VERSION, '=' should be used instead ':=' to make
sure PKG_VERSION has been defined in common makefile.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agokernel: add i2c-smbus module package
Philip Prindeville [Tue, 17 Oct 2017 04:11:24 +0000 (22:11 -0600)]
kernel: add i2c-smbus module package

Intel motherboards (as well as the Cavium ThunderX SoC) use a
superset of the I2C protocol called SMBus.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agolinux: unbreak host tools build for 4.14
Jonas Gorski [Mon, 25 Dec 2017 18:04:35 +0000 (19:04 +0100)]
linux: unbreak host tools build for 4.14

Remove a stray -Wp left in host_c_flags causing build failures for newer
4.14 versions.

Reported-by: Michael Marley <michael@michaelmarley.com>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm2708: use x86's upgrade scripts for all rpi targets
Christian Lamparter [Sun, 17 Dec 2017 14:18:49 +0000 (15:18 +0100)]
brcm2708: use x86's upgrade scripts for all rpi targets

Advantages:
 - preserves existing partition layout on the sd-card.
   Only the boot and rootfs partition will be overwritten.

Please note that sysupgrade will refuse to upgrade, if the existing
installation  has an incompatible partition layout. Future changes
to the bootfs and/or rootfs partition size will likely cause breakage
to the sysupgrade procedure. In these cases, the ext4-sdcard.img.gz
will have to be written to the sdcard manually.
Please don't forget to backup your configuration in this case.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agobrcm2708: convert to dt-based board-detection
Christian Lamparter [Sun, 17 Dec 2017 14:18:48 +0000 (15:18 +0100)]
brcm2708: convert to dt-based board-detection

Use the values populated by the generic board detect function. The
first compatible from the device tree source file will be the board
name in userspace. The model property from the device tree source file
will be the model name.

Change the board name where used in the userspace and drop the target
specific board detect, to use the generic one.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agobrcm2708: add compatible strings
Christian Lamparter [Sun, 17 Dec 2017 14:18:47 +0000 (15:18 +0100)]
brcm2708: add compatible strings

This patch adds the compatible string for the various RPIs from
4.14 upstream.

Note: The 4.14 upstream does not include the compute modules.
If the CM* would just house the SoC, it could in theory use the
"raw" chip compatible string. However, these CM boards also come
with RAM and eMMC. So they have to have a proper comaptible.

For now, "raspberrypi,compute-module-{1|3}" will be good enough.

Note2: The original CM was renamed to CM1 when CM3 was released.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agobrcm2708: convert to metadata
Christian Lamparter [Sun, 17 Dec 2017 14:18:46 +0000 (15:18 +0100)]
brcm2708: convert to metadata

This patch converts all the raspberrypi images to utilize
the common metadata-based image verification.

Note: the CM1 and CM3 currently use the same "rpi-cm"
boardname.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agonghttp2: bump to 1.29.0
Hans Dedecker [Sun, 24 Dec 2017 16:01:35 +0000 (17:01 +0100)]
nghttp2: bump to 1.29.0

439b9b6c (tag: v1.29.0) Update manual pages
48498452 Bump up version number to v1.29.0, LT revision to 29:1:15
d30f3816 Update manual pages
4d1139f6 Remove SPDY
48f57407 nghttpx: Update doc
c1f14d73 Update manual pages
216f4dad nghttpx: Remove redundant check
a4e27d76 Revert "nghttpx: Use an existing h2 backend connection as much as possible"
2365f12e Fix CMAKE_MODULE_PATH
03f7ec0f nghttpx: Write API request body in temporary file
2056e812 nghttpx: Increase api-max-request-body
1ebb6810 nghttpx: Faster configuration loading with lots of backends
a3ebeeaf nghttpx: Fix crash with --backend-http-proxy-uri option
422ad1be Use NGHTTP2_REFUSED_STREAM for streams which are closed by GOAWAY
97f1735c Bump up version number to 1.29.0

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agousbmode: update to latest git HEAD
John Crispin [Tue, 19 Dec 2017 19:49:06 +0000 (20:49 +0100)]
usbmode: update to latest git HEAD

f40f84c support PantechMode
d8dc335 support Quanta and Blackberry modes
333e486 fix support for Option modems

Signed-off-by: John Crispin <john@phrozen.org>
6 years agoar71xx: spi-rb4xx: balance clk_prepare_enable() call
Gabor Juhos [Tue, 19 Dec 2017 07:24:38 +0000 (08:24 +0100)]
ar71xx: spi-rb4xx: balance clk_prepare_enable() call

Commit c312cef223dd ("ar71xx: spi-rb4xx fix.") replaced the clk_enable() call
with clk_prepare_enable() to meet the common clock framework requirements.
However it did not change the clk_disable() call in the error patch which
thus leads to imbalance.

Fix the code by using the correct counterpart of clk_prepare_enable() in both
places.

Fixes: c312cef223dd ("ar71xx: spi-rb4xx fix.")
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
6 years agokernel/4.14: add missing newline character to UBI messages
Gabor Juhos [Tue, 19 Dec 2017 08:11:57 +0000 (09:11 +0100)]
kernel/4.14: add missing newline character to UBI messages

A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:

  [    6.649159] UBI error: no valid UBI magic found inside mtd6[    6.667751] Freeing unused kernel memory: 2196K

Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:

  [    6.609182] UBI error: no valid UBI magic found inside mtd6
  [    6.627599] Freeing unused kernel memory: 2132K

Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
6 years agokernel/4.9: add missing newline character to UBI messages
Gabor Juhos [Tue, 19 Dec 2017 08:11:48 +0000 (09:11 +0100)]
kernel/4.9: add missing newline character to UBI messages

A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:

  [    6.649159] UBI error: no valid UBI magic found inside mtd6[    6.667751] Freeing unused kernel memory: 2196K

Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:

  [    6.609182] UBI error: no valid UBI magic found inside mtd6
  [    6.627599] Freeing unused kernel memory: 2132K

Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
6 years agokernel/4.4: add missing newline character to UBI messages
Gabor Juhos [Tue, 19 Dec 2017 08:11:24 +0000 (09:11 +0100)]
kernel/4.4: add missing newline character to UBI messages

A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:

  [    6.649159] UBI error: no valid UBI magic found inside mtd6[    6.667751] Freeing unused kernel memory: 2196K

Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:

  [    6.609182] UBI error: no valid UBI magic found inside mtd6
  [    6.627599] Freeing unused kernel memory: 2132K

Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
6 years agowireguard: bump to 20171221
Kevin Darbyshire-Bryant [Sat, 23 Dec 2017 19:26:52 +0000 (19:26 +0000)]
wireguard: bump to 20171221

7e945a8 version: bump snapshot
f2168aa compat: kernels < 3.13 modified genl_ops
52004fd crypto: compile on UML
6b69b65 wg-quick: dumber matching for default routes
aa35d9d wg-quick: add the "Table" config option
037c389 keygen-html: remove prebuilt file

No patch refresh required.

Compile-test-for: ar71xx
Run-tested-on: ar71xx Archer C7 v2

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agokernel: bump 4.4 to 4.4.107
Kevin Darbyshire-Bryant [Thu, 21 Dec 2017 19:12:27 +0000 (19:12 +0000)]
kernel: bump 4.4 to 4.4.107

Refresh patches.

Update patch that no longer applied:
oxnas/0072-mtd-backport-v4.7-0day-patches-from-Boris.patch

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Tested-by: Rosen Penev <rosenp@gmail.com>
6 years agolantiq: remove DSL_ChipSetHWVersion from status information
Martin Schiller [Fri, 22 Dec 2017 07:42:18 +0000 (08:42 +0100)]
lantiq: remove DSL_ChipSetHWVersion from status information

The value DSL_ChipSetHWVersion is fetched from the dsl frontend
via the dsl_control service, but not really provided by the dsl
frontend firmware and for now always "UNKNOWN".

The lantiq support told us that this information wouldn't be
provided in the foreseeable future, so let's remove this
useless "UNKNOWN" information.

Signed-off-by: Martin Schiller <ms@dev.tdt.de>
6 years agoramips: fix Linkit smart 7688 support
Mathias Kresin [Thu, 21 Dec 2017 18:55:25 +0000 (19:55 +0100)]
ramips: fix Linkit smart 7688 support

Use a more appropriate compatible string. Fix the wireless led GPIO and
add the default wireless trigger. Use the wireless LED for boot state
indication as well.

Remove the GPIO pinmux for pins not exposed on the board.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: add support for TP-Link TL-MR3420 v5
Henryk Heisig [Fri, 15 Dec 2017 14:57:49 +0000 (15:57 +0100)]
ramips: add support for TP-Link TL-MR3420 v5

TP-Link TL-MR3420 v5 are simple N300 router with
5-port FE switch and non-detachable antennas.
Its very similar to TP-Link TL-WR841N V13.

Specification:

- MT7628N/N (580 MHz)
- 64 MB of RAM (DDR2)
- 8 MB of FLASH
- 2T2R 2.4 GHz
- 5x 10/100 Mbps Ethernet
- 2x external, non-detachable antennas
- USB 2.0 Port
- UART (J1) header on PCB (115200 8n1)
- 8x LED, 2x button, power input switch

Flash instruction:

The only way to flash LEDE image in mr3420v5 is to use
tftp recovery mode in U-Boot:

1. Configure PC with static IP 192.168.0.225/24 and tftp server.
2. Rename "lede-ramips-mt7628-tplink_tl-mr3420-v5-squashfs-tftp-recovery.bin"
   to "tp_recovery.bin" and place it in tftp server directory.
3. Connect PC with one of LAN ports, press the reset button, power up
   the router and keep button pressed for around 6-7 seconds, until
   device starts downloading the file.
4. Router will download file from server, write it to flash and reboot.

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
6 years agoramips: move common tplink image build code into own recipe
Mathias Kresin [Sun, 17 Dec 2017 09:44:41 +0000 (10:44 +0100)]
ramips: move common tplink image build code into own recipe

Move common tplink image build code into own recipe. Include the common
parts instead of including a full build recipe and overwriting former set
varaibles.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agokernel: move console loglevel to generic
Tomasz Maciej Nowak [Wed, 20 Dec 2017 18:55:16 +0000 (19:55 +0100)]
kernel: move console loglevel to generic

Move CONFIG_CONSOLE_LOGLEVEL_DEFAULT=7 to generic, to make it consistent
across all targets.

Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
6 years agouci: update to latest git HEAD (FS#1182)
Hans Dedecker [Fri, 22 Dec 2017 17:16:41 +0000 (18:16 +0100)]
uci: update to latest git HEAD (FS#1182)

25fd142 list: remove unncessary increment of n_section (FS#1182)
021fd65 file: remove unnecessary uci_fixup_section calls

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agoodhcpd: update to latest git HEAD
Hans Dedecker [Fri, 22 Dec 2017 17:11:20 +0000 (18:11 +0100)]
odhcpd: update to latest git HEAD

7aa2594 odhcpd: Replace strerror(errno) with %m format
750e457 Support muliple RAs on single interface

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agokernel: add missing config symbols for 4.14
Stijn Tintel [Thu, 21 Dec 2017 22:54:41 +0000 (23:54 +0100)]
kernel: add missing config symbols for 4.14

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agoar71xx: add support for MikroTik RouterBOARD 921GS-5HPacD r2 (mANTBox 15s)
Roger Pueyo Centelles [Thu, 21 Dec 2017 14:53:05 +0000 (15:53 +0100)]
ar71xx: add support for MikroTik RouterBOARD 921GS-5HPacD r2 (mANTBox 15s)

This patch adds support for the MikroTik RouterBOARD 921GS-5HPacD r2
(mANTBox 15s), an outdoor sector antenna with a built-in 802.11ac
wireless router. Additionally, it adds a new profile for devices with
>= 128 MB NAND flash and 802.11ac to the ar71xx/mikrotik subtarget.

See https://mikrotik.com/product/RB921GS-5HPacD-15S for more info.

Specifications:
 - SoC: Qualcomm Atheros QCA9558 (720 MHz)
 - RAM: 128 MB
 - Storage: 128 MB NAND
 - Wireless: external QCA9822 802.11a/ac 2x2:2
 - Ethernet: 1x 1000/100/10 Mbps, integrated, via AR8031 PHY, passive PoE in
 - SFP: 1x host

Working:
 - Board/system detection
 - NAND storage detection
 - Wireless
 - Ethernet
 - 1x user LED
 - Reset button
 - Sysupgrade

Untested:
 - SFP cage (probably not working)

Installation:
 - Boot initramfs image via TFTP and then flash sysupgrade image

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
6 years agoar71xx: fix spelling in mach-rb922.c
Roger Pueyo Centelles [Thu, 21 Dec 2017 14:52:38 +0000 (15:52 +0100)]
ar71xx: fix spelling in mach-rb922.c

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
6 years agoar71xx: fix rb922gs_nand_scan_fixup()
Roger Pueyo Centelles [Thu, 21 Dec 2017 14:51:10 +0000 (15:51 +0100)]
ar71xx: fix rb922gs_nand_scan_fixup()

Apply code from commits 9e1bc27e6d67fa29791eb0fac5e5f0066d12a761
(ar71xx: Fix UBIFS work on Mikrotik RB95x devices) and
665bb2749906aff4d4ec607a308353fa30cb1b3a (ar71xx: fix invalid pointer
dereference in rb95x_nand_scan_fixup()) to RB92x devices too.

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
6 years agolayerscape: fix package download
Hauke Mehrtens [Thu, 21 Dec 2017 08:32:51 +0000 (09:32 +0100)]
layerscape: fix package download

The git hash was changed for multiple layerscape packages without
changing the version number. The LEDE build system will not download the
packages again if the old version is already there and so some people
and the build bots are using wrong version of some packages. Use
PKG_SOURCE_DATE instead of PKG_VERSION to generate packages with the
date and the first charterers of the git hash. This will change the file
name and make the build system download them again, also if in future
the git hash is changed the file name will change and trigger a new
download.

This should fix a problem spotted by build bot.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoapm821xx: convert to device-tree board detection
Christian Lamparter [Sun, 17 Dec 2017 14:00:01 +0000 (15:00 +0100)]
apm821xx: convert to device-tree board detection

This patch converts all apm821xx devices to the device-tree
board-detection method. All instances of the legacy
boardnames (mbl,mr24,...) are converted to "vendor,device"
identifier.

The custom board-detection code in apm821xx.sh is removed as
it no longer serves any purpose.

Signed-off-by: Mathias Kresin <dev@kresin.me>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: enable metadata for packaging
Christian Lamparter [Sun, 17 Dec 2017 14:00:00 +0000 (15:00 +0100)]
apm821xx: enable metadata for packaging

This patch enables metadata-supported image verification
for all apm821xx supported devices. Since this method comes
with a built-in image verification tool (fwtool), the previous
image checks can be removed.

Signed-off-by: Mathias Kresin <dev@kresin.me>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: align device names with vendor_device format
Christian Lamparter [Sun, 17 Dec 2017 13:59:59 +0000 (14:59 +0100)]
apm821xx: align device names with vendor_device format

Currently, the device name handle does not include the
manufacturer. This can make it hard do differentiate
between products from different vendors that have the
same product name. As the handle is used to derive
the image name.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: dts: rename devices dts files to include the manufacturer
Christian Lamparter [Sun, 17 Dec 2017 13:59:58 +0000 (14:59 +0100)]
apm821xx: dts: rename devices dts files to include the manufacturer

This patch rename all the DT source files in order to
match upstream's "manufacturer-device.dts" format.

Please note that the DEVICE_DTB isn't changed. This is
because the u-boot of the MyBook Live defines the
fdt_file variable to be "apollo3g/apollo3g.dtb".

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: add product names to the dt compatible for Meraki
Christian Lamparter [Sun, 17 Dec 2017 13:59:57 +0000 (14:59 +0100)]
apm821xx: add product names to the dt compatible for Meraki

Meraki choose to use their product's codename as the main
compatible string. Mathias Kresin commented that this is
a poor choice as this will confuse the users and devs once
the device-tree compatible is used for board-detection and
possible the image name.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: dts: append SoC compatible to DTS
Christian Lamparter [Sun, 17 Dec 2017 13:59:56 +0000 (14:59 +0100)]
apm821xx: dts: append SoC compatible to DTS

This patch appends the "apm,bluestone" or "amcc,apollo3g"
machine compatible string to the current device tree source.

Please note that unlike other archs the PPC DT code does
not regard the machine's compatible string as a priority
list. This is explained in the kernel's usage-model.txt as follows:
"PowerPC uses a slightly different scheme where it calls the .probe()
hook from each machine_desc, and the first one returning TRUE is used.
However, this approach does not take into account the priority of the
compatible list, and probably should be avoided for new architecture
support."

For this reason, the "apm,bluestone" compatible string can't be
added to the WNDR4700. As otherwise the target specific pci
fix-up code will get ignored and this causes the ath9k WIFI
to not get initialized.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: replace DEVICE_{PROFILE|NAME} with BOARD_NAME
Christian Lamparter [Sun, 17 Dec 2017 13:59:55 +0000 (14:59 +0100)]
apm821xx: replace DEVICE_{PROFILE|NAME} with BOARD_NAME

This patch sets the BOARD_NAME variable on each affected
apm821xx device. The existing DEVICE_PROFILE and
DEVICE_NAME assignments are deprecated as they no longer
serve any purpose.

The BOARD_NAME variable is used by the sysupgrade-tar
method to specifiy a directory overwrite for the
sysupgrade-$dir directory in the generated tar file.
Keeping the original boardname in this context will be
necessary for targets that utilize the sysupgrade-tar
method. Otherwise, sysupgrade on an previous installation
will not recognize the newly generated images.

This step is necessary since an upcoming patch realigns
the existing shortname for a device with a proper
"manufacturer_device" identifier.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: replace whitespace with tabs
Christian Lamparter [Sun, 17 Dec 2017 13:59:54 +0000 (14:59 +0100)]
apm821xx: replace whitespace with tabs

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: explicitly build the rootfs.img.gz target
Christian Lamparter [Sun, 17 Dec 2017 13:59:53 +0000 (14:59 +0100)]
apm821xx: explicitly build the rootfs.img.gz target

The commit 87b668765e1
("image: when using the new image build code, gzip ext4 images by default")

forced that all targets that select the ext4 as the root filesystem
to always compress the generated rootfs. This is fine, but this method
doesn't not allow to append the metadata on a per-target base.

Therefore this patch changes the rootfs image production rule to generate
the gzip step manually. This way the metadata can be appended at a later
date.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoramips: fix devicetree compat strings
Mathias Kresin [Tue, 19 Dec 2017 20:55:42 +0000 (21:55 +0100)]
ramips: fix devicetree compat strings

"PandoraBox" is not the name of the manufacturer, it's a firmware made by
the manufacturer actually. Their official English name is "D-Team".

PBR-M1 is the only one they use "PandoraBox" as a brand name. Their other
products are using "Newifi" as their trademark (including Y1 and Y1S which
used to be OEM products for Lenovo).

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: add support for newifi d2
Jackson Ming Hu [Fri, 15 Dec 2017 03:51:33 +0000 (11:51 +0800)]
ramips: add support for newifi d2

Previously Newifi D2 could only use PandoraBox M1's firmware.
It works fine, but LED GPIO is different.
As a result, a separated DTS file for this device should be implemented.

Hardware spec:

* CPU: MTK MT7621A
* RAM: 512MB
* ROM: 32MB SPI Flash
* WiFi: MTK MT7603+MT7612
* Button: 2 buttons (reset, wps)
* LED: 3 single-color LEDs (USB, WiFi 2.4GHz, WiFi 5GHz) &
       2 dual-color LEDs (Power, Internet)
* Ethernet: 5 ports, 4 LAN + 1 WAN

Installation method:

Same as Newifi D1, users may need to request unlock code from the device
manufacturer. Otherwise, a SPI flash programmer may be necessary to get
the firmware flashed. After the device is unlocked, press and hold reset
button before power cable plugs in. Then go to http://192.168.1.1 to
upload and flash the firmware package.

Signed-off-by: Jackson Ming Hu <huming2207@gmail.com>
6 years agoramips: add support for Vonets VAR11N-300
Andrew Crawley [Thu, 23 Nov 2017 23:52:13 +0000 (23:52 +0000)]
ramips: add support for Vonets VAR11N-300

The VAR11N-300 is a tiny wireless-N device with a hardwired Ethernet
cable, one extra Ethernet port, and an internal antenna, based on the
MediaTek MT7620n chipset.

Specs:
- MT7620n WiSoC @ 600MHz
- 32 MB SDRAM
- 4 MB SPI flash
- 2T2R 2.4GHz WiFi-N
- 1 attached 10/100 Ethernet cable (LAN)
- 1 10/100 Ethernet port (WAN)
- 1 attached USB / barrel 5vdc power cable
- 5 LEDs (see notes below)
- 1 reset button
- 1 UART (3 pads on board)

Installation:

The stock firmware does not support uploading new firmware directly,
only checking the manufacturer's site for updates.  This process may be
possible to spoof, but the update check uses some kind of homebrew
encryption that I didn't investigate.  Instead, you can install via a
backdoor:

1. Set up a TFTP server to serve the firmware binary
(lede-ramips-mt7620-var11n-300-squashfs-sysupgrade.bin)
2. Factory reset the device by holding the reset button for a few
seconds.
3. Open the web interface (default IP: 192.168.253.254)
4. Log in with the "super admin" credentials: username `vonets`,
password `vonets26642519`.
5. On the "Operative Status" page, click the text "System Uptime", then
quickly click the uptime value.
6. If successful, an alert dialog will appear reading "Ated start", and
the device will now accept telnet connections.  If the alert does not
appear, repeat step 5 until it works (the timing is a bit tricky).
7. Telnet to the device using credentials "admin / admin"
8. Retrieve the firmware binary from the tftp server: `tftp -l lede.bin
-r lede-ramips-mt7620-var11n-300-squashfs-sysupgrade.bin -g
<tftp-server-ip>`
9. Write the firmware to flash: `mtd_write write lede.bin /dev/mtd4`
10. Reboot

Tested:
- LAN / WAN ethernet
- WiFi
- LAN / WAN / status LED GPIOs (see notes below)
- Reset button
- Sysupgrade

Notes:

LEDs:

The board has 5 LEDs - two green LEDs for LAN / WAN activity, one blue
LED for WiFi, and a pair of "status" LEDs connected to the same GPIO
(the blue LED lights when the GPIO is low, and the green when it's
high).  I was unable to determine how to operate the WiFi LED, as it
does not appear to be controlled by a GPIO directly.

Recovery:

The default U-boot installation will only boot from flash due to a
missing environment block.  I generated a valid 4KB env block using
U-boot's `fw_setenv` tool and wrote it to flash at 0x30000 using an
external programmer.  After this, it was possible to enter the U-boot
commandline interface and download a new image via TFTP (`tftpboot
81b00000 <image-filename>`), but while I could boot this image
sucessfully (`bootm`), writing it to flash (`cp.linux`) just corrupted
the flash chip.  The sysupgrade file can be written to flash at 0x50000
using an external programmer.

Signed-off-by: Andrew Crawley <acrawley@gmail.com>