lede.git
8 years agoscripts/getver.sh: fix revision number on BSD/MacOS
Felix Fietkau [Sat, 9 Jul 2016 18:00:50 +0000 (20:00 +0200)]
scripts/getver.sh: fix revision number on BSD/MacOS

Strip leading whitespace from wc -l output

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoath10k: fix a compiler warning
Felix Fietkau [Sat, 9 Jul 2016 13:00:30 +0000 (15:00 +0200)]
ath10k: fix a compiler warning

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoath10k: fix #if vs #ifdef in led trigger patch
Felix Fietkau [Fri, 8 Jul 2016 16:51:37 +0000 (18:51 +0200)]
ath10k: fix #if vs #ifdef in led trigger patch

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomac80211: make package ath9k-common hidden
Felix Fietkau [Sat, 9 Jul 2016 17:37:36 +0000 (19:37 +0200)]
mac80211: make package ath9k-common hidden

It cannot be selected independently of ath9k and ath9k-htc

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agobrcm2708-gpu-fw: update to latest version
Álvaro Fernández Rojas [Fri, 8 Jul 2016 07:02:27 +0000 (09:02 +0200)]
brcm2708-gpu-fw: update to latest version

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
8 years agobrcm2708: update linux 4.4 patches to latest version
Álvaro Fernández Rojas [Thu, 7 Jul 2016 07:22:07 +0000 (09:22 +0200)]
brcm2708: update linux 4.4 patches to latest version

As usual these patches were extracted and rebased from the raspberry pi repo:
https://github.com/raspberrypi/linux/tree/rpi-4.4.y

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
8 years agoscripts/getver.sh: avoid use of git rev-list --count
Jonas Gorski [Sat, 9 Jul 2016 13:01:58 +0000 (15:01 +0200)]
scripts/getver.sh: avoid use of git rev-list --count

This is not a valid option in older git version, used in e.g. RHEL6.

Reported-by: Steven Haigh <netwiz@crc.id.au>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agobrcm63xx: fix image generation with offsets/blocksizes != 64K
Jonas Gorski [Sat, 9 Jul 2016 11:10:08 +0000 (13:10 +0200)]
brcm63xx: fix image generation with offsets/blocksizes != 64K

Defer the CFE_EXTRAs evaluation to allow overriden BLOCK_SIZE and
IMAGE_OFFSET take effect, and replace the unused IMAGE_SIZE with
IMAGE_OFFSET in DEVICE_VARS.

This fixes image generation for targets using different offsets or
blocksizes.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agobrcm63xx: fix CVG834G compatible string
Jonas Gorski [Wed, 29 Jun 2016 11:05:40 +0000 (13:05 +0200)]
brcm63xx: fix CVG834G compatible string

The CVG834G claimed to be a BCM6348, but is actually BCM3368. Since we
don't test against it, this was harmless.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agoscripts/getver.sh: improve revision output
Jonas Gorski [Sat, 25 Jun 2016 12:23:30 +0000 (14:23 +0200)]
scripts/getver.sh: improve revision output

Change the revision output to r<upstream-revision>+<local commits> so
it is easier to get the base revision (and see if there are local
commits).

Example:
$ ./scripts/getver.sh
r794+3
$

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agoscripts/getver.sh: allow conversion between git hash and revision
Jonas Gorski [Sun, 5 Jun 2016 18:45:25 +0000 (20:45 +0200)]
scripts/getver.sh: allow conversion between git hash and revision

Add code allowing easy conversion between git commit ids and revisions.

Example:

$ ./scripts/getver.sh
r792
$ ./scripts/getver.sh r123
b7fc892eb56b52d25bc00578af9d1ee058dcf6b3
$ ./scrpts/getver.sh b7fc892eb56b52d25bc00578af9d1ee058dcf6b3
r123

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agoscripts/getver.sh: simplify revision calculation
Jonas Gorski [Sat, 4 Jun 2016 16:35:21 +0000 (18:35 +0200)]
scripts/getver.sh: simplify revision calculation

Use git rev-list --count to get the revision number.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
8 years agomac80211: backport brcmfmac changes from 2016-07-08
Rafał Miłecki [Fri, 8 Jul 2016 16:17:53 +0000 (18:17 +0200)]
mac80211: backport brcmfmac changes from 2016-07-08

To work correctly hostapd requires wireless driver to allow interfaces
removal. It was working with brcmfmac only partially. Firmware for
BCM43602 got some special hack (feature?) that allowed removing all
interfaces by disabling mbss mode. It wasn't working with BCM4366
firmware and remaining interfaces were preventing hostapd from starting
again.

Those patches add support for "interface_remove" firmware method which
works with BCM4366 firmware and they make it finally possible to use
BCM4366 & brcmfmac & multiple interfaces.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agobcm53xx: fix kernel config after USB2 PHY driver backport
Felix Fietkau [Fri, 8 Jul 2016 10:03:15 +0000 (12:03 +0200)]
bcm53xx: fix kernel config after USB2 PHY driver backport

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomt76: update to the latest version, fixes powersave issues
Felix Fietkau [Fri, 8 Jul 2016 08:41:45 +0000 (10:41 +0200)]
mt76: update to the latest version, fixes powersave issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agotoolchain: skip gcc/minimal for musl
Felix Fietkau [Thu, 7 Jul 2016 12:12:30 +0000 (14:12 +0200)]
toolchain: skip gcc/minimal for musl

No extra libc header build step is done, so no extra toolchain is needed
for preparing it.
This saves a significant amount of build time and disk space

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agobcm53xx: use upstream fix simplifying USB power GPIO usage
Rafał Miłecki [Thu, 7 Jul 2016 13:06:34 +0000 (15:06 +0200)]
bcm53xx: use upstream fix simplifying USB power GPIO usage

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agobcm53xx: backport driver for Northstar's USB 2.0 PHY
Rafał Miłecki [Thu, 7 Jul 2016 12:49:35 +0000 (14:49 +0200)]
bcm53xx: backport driver for Northstar's USB 2.0 PHY

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agobcm53xx: backport BCM5301x patches from 2016-07-06
Rafał Miłecki [Thu, 7 Jul 2016 11:54:11 +0000 (13:54 +0200)]
bcm53xx: backport BCM5301x patches from 2016-07-06

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agoscripts/package-metadata.pl: fix kmod pakage dependencies within the SDK
Felix Fietkau [Thu, 7 Jul 2016 10:39:27 +0000 (12:39 +0200)]
scripts/package-metadata.pl: fix kmod pakage dependencies within the SDK

Instead of ignoring all metadata for package/kernel/linux, process it
and only suppress emitting config data to tmp/.config-package.in

This ensures that packages that select kmod-* packages can inherit their
depdendencies.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoscripts/diffconfig.sh: fix handing of CONFIG_TARGET_MULTI_PROFILE
Felix Fietkau [Thu, 7 Jul 2016 09:23:33 +0000 (11:23 +0200)]
scripts/diffconfig.sh: fix handing of CONFIG_TARGET_MULTI_PROFILE

Fixes FS#43

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomwlwifi: Update to latest version
Gabe Rodriguez [Sun, 3 Jul 2016 09:15:14 +0000 (02:15 -0700)]
mwlwifi: Update to latest version

Signed-off-by: Gabe Rodriguez <lifehacksback@gmail.com>
8 years agotarget.mk: fix ARM architecture feature flag detection
Felix Fietkau [Tue, 5 Jul 2016 07:49:12 +0000 (09:49 +0200)]
target.mk: fix ARM architecture feature flag detection

Move feature flag checks below CFLAGS handling, use CPU_TYPE CFLAGS for
architecture detection.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agovalgrind: update to the latest version, fix build issues on ARM
Felix Fietkau [Tue, 5 Jul 2016 08:08:59 +0000 (10:08 +0200)]
valgrind: update to the latest version, fix build issues on ARM

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agotools: build b43 tools if the SDK was enabled
Felix Fietkau [Thu, 7 Jul 2016 08:52:51 +0000 (10:52 +0200)]
tools: build b43 tools if the SDK was enabled

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoprism54-firmware: change prism54/p54-firmware package versioning
Christian Lamparter [Thu, 30 Jun 2016 22:17:54 +0000 (00:17 +0200)]
prism54-firmware: change prism54/p54-firmware package versioning

There are multiple prism54/p54 firmware versions for different
drivers and devices. Therefore, assigning the package version
of all the different firmware packages on the old prism54
firmware could break if any of the p54 firmwares are updated
and we need to roll out new packages.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
[removed PKG_VERSION]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
8 years agoinclude/toplevel.mk: fix defconfig when ~/.openwrt/defconfig exists - take 2
John Crispin [Wed, 6 Jul 2016 08:34:44 +0000 (10:34 +0200)]
include/toplevel.mk: fix defconfig when ~/.openwrt/defconfig exists - take 2

commit 5b728074160b ("include/toplevel.mk: fix defconfig when
~/.openwrt/defconfig exists") was missing a !

Signed-off-by: John Crispin <john@phrozen.org>
8 years agomac80211: backport brcmfmac changes from 2016-06-29
Rafał Miłecki [Wed, 6 Jul 2016 08:31:39 +0000 (10:31 +0200)]
mac80211: backport brcmfmac changes from 2016-06-29

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agoramips: fix build error in ubnt-erx initramfs factory image
Felix Fietkau [Wed, 6 Jul 2016 08:11:29 +0000 (10:11 +0200)]
ramips: fix build error in ubnt-erx initramfs factory image

Build initramfs kernel image, but skip factory image if initramfs size
is too big.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agohostapd: add a WDS AP fix for reconnecting clients
Felix Fietkau [Wed, 6 Jul 2016 06:06:46 +0000 (08:06 +0200)]
hostapd: add a WDS AP fix for reconnecting clients

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoinclude/toplevel.mk: fix defconfig when ~/.openwrt/defconfig exists
John Crispin [Mon, 4 Jul 2016 09:08:58 +0000 (11:08 +0200)]
include/toplevel.mk: fix defconfig when ~/.openwrt/defconfig exists

./scripts/feeds update will reset the .config file if ~/.openwrt/defconfig
exists, thus resetting the target to ar71xx.

Signed-off-by: John Crispin <john@phrozen.org>
8 years agousb: Remove annoying warning about bogus URB
Alexey Brodkin [Tue, 5 Jul 2016 12:42:21 +0000 (15:42 +0300)]
usb: Remove annoying warning about bogus URB

When USB Wi-Fi dongle based on Atheros AR9271 is connected to OHCI
(USB 1.1) controller following warnings flood debug console:
------------------------>8---------------------------
usb 1-1: new full-speed USB device number 2 using ohci-platform
usb 1-1: ath9k_htc: Firmware ath9k_htc/htc_9271-1.4.0.fw requested
usb 1-1: ath9k_htc: Transferred FW: ath9k_htc/htc_9271-1.4.0.fw, size: 51008
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450
usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Not tainted 4.6.3 #10
Workqueue: events request_firmware_work_func

Stack Trace:
  arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d1 ]---
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450 usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Tainted: G        W       4.6.3 #10
Workqueue: events request_firmware_work_func

Stack Trace:
  arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d2 ]---
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450 usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Tainted: G        W       4.6.3 #10
Workqueue: events request_firmware_work_func

Stack Trace:
  arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d3 ]---

...
------------------------>8---------------------------

With removed warning Wi-Fi dongle works properly.

Even though this is not the best solution it gets us a working Wireless
AP. Anyways new discussion was started in linux-usb mailing list to find
a proper solution instead of that hack.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agolinux/archs38: Disable USB 2.0
Alexey Brodkin [Tue, 5 Jul 2016 12:41:58 +0000 (15:41 +0300)]
linux/archs38: Disable USB 2.0

As of today USB 2.0 (AKA EHCI) doesn't work on axs103 board
and so there's no reason to build corresponding software for it.

Once USB 2.0 gets fixed on axs103 thi patch might be reverted.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
8 years agolantiq: mac address setting on BTHOMEHUBV3A
Ben Mulvihill [Tue, 5 Jul 2016 17:35:05 +0000 (19:35 +0200)]
lantiq: mac address setting on BTHOMEHUBV3A

Rename uboot environment partition on BT Home Hub 3A so that mac address
setting works correctly.

Also, the mac address field in the ath9k calibration data is not used,
and should not be referenced in the dts.

Signed-off-by: Ben Mulvihill <ben.mulvihill@gmail.com>
8 years agotools: add kernel2minor utility for Mikrotik devices
Sergey Sergeev [Tue, 28 Jun 2016 11:55:12 +0000 (14:55 +0300)]
tools: add kernel2minor utility for Mikrotik devices

This patch adds support of Mikrotik yaffs2 filesystem image for kernel file
and tools/kernel2minor package.
We neede this to boot kernel through RouterBoot on new Mikrotik NOR flash devices.

Signed-off-by: Sergey Sergeev <adron@yapic.net>
8 years agosamba: Update smb.conf.template
neheb [Sat, 2 Jul 2016 03:02:17 +0000 (20:02 -0700)]
samba: Update smb.conf.template

Removed some options which are default anyway and added bind interfaces
only which causes the interfaces line to actually have an effect. Can be
verified with netstat.

Signed-off by: Rosen Penev <rosenp@gmail.com>

8 years agokernel: Move POSIX ACL and attr support options into submenu
Daniel Dickinson [Thu, 17 Dec 2015 03:58:11 +0000 (22:58 -0500)]
kernel: Move POSIX ACL and attr support options into submenu

Make global options menuconfig cleaner by moving POSIX ACL
and attr support options into a submenu.

Signed-off-by: Daniel Dickinson <openwrt@daniel.thecshore.com>
8 years agoramips: fix buffalo wmr-300 lan port
John Crispin [Tue, 5 Jul 2016 07:07:33 +0000 (09:07 +0200)]
ramips: fix buffalo wmr-300 lan port

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoar71xx: fix nand device profile
John Crispin [Tue, 5 Jul 2016 07:07:16 +0000 (09:07 +0200)]
ar71xx: fix nand device profile

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoprocd: update to latest git HEAD
John Crispin [Tue, 5 Jul 2016 08:24:51 +0000 (10:24 +0200)]
procd: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agouci: update to latest git HEAD
John Crispin [Mon, 4 Jul 2016 14:37:55 +0000 (16:37 +0200)]
uci: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agousign: update to latest git HEAD
John Crispin [Mon, 4 Jul 2016 14:36:47 +0000 (16:36 +0200)]
usign: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoumbim: update to latest git HEAD
John Crispin [Mon, 4 Jul 2016 14:35:39 +0000 (16:35 +0200)]
umbim: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agolibubox: update to latest git HEAD
John Crispin [Mon, 4 Jul 2016 14:32:27 +0000 (16:32 +0200)]
libubox: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoramips: fix MT7621 gsw handling
John Crispin [Mon, 4 Jul 2016 13:20:25 +0000 (15:20 +0200)]
ramips: fix MT7621 gsw handling

based on https://forum.openwrt.org/viewtopic.php?pid=323671#p323671

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoramips: fix mt7621 SoC detection
John Crispin [Mon, 4 Jul 2016 13:20:05 +0000 (15:20 +0200)]
ramips: fix mt7621 SoC detection

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoscripts/feeds: Prevent .config autocreation
Daniel Dickinson [Mon, 4 Jul 2016 08:43:02 +0000 (04:43 -0400)]
scripts/feeds: Prevent .config autocreation

When using scripts/feeds upgrade the .config needs to be
updated but the code to do so was also autocreating a
.config if one didn't exist.  This is counter-productive
when you have not yet used menuconfig (or other config targets)
because things like selecting or deselecting CONFIG_ALL
(to build all package by default) only works if the
package selection has not already been done via an
existing .config selection.

Signed-off-by: Daniel Dickinson <lede@daniel.thecshore.com>
8 years agopx5g: Create mbedtls variant
Daniel Dickinson [Thu, 23 Jun 2016 01:12:16 +0000 (21:12 -0400)]
px5g: Create mbedtls variant

px5g has been listed as a blocker for switching to new mbedtls
as the default, therefore make and mbedtls variant of px5g so
that an new mbedtls-only image can be created.

Signed-off-by: Daniel Dickinson <lede@daniel.thecshore.com>
8 years agopackages: fix bmp085-spi typo
Zoltan HERPAI [Mon, 4 Jul 2016 17:34:48 +0000 (19:34 +0200)]
packages: fix bmp085-spi typo

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
8 years agouclient: update to the latest version, fixes HTTP redirect support
Felix Fietkau [Tue, 5 Jul 2016 10:40:15 +0000 (12:40 +0200)]
uclient: update to the latest version, fixes HTTP redirect support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoath9k: add beacon related stability fixes
Felix Fietkau [Tue, 5 Jul 2016 08:45:17 +0000 (10:45 +0200)]
ath9k: add beacon related stability fixes

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoRevert "ustream-ssl: Fix recursive dependency"
Felix Fietkau [Mon, 4 Jul 2016 10:59:03 +0000 (12:59 +0200)]
Revert "ustream-ssl: Fix recursive dependency"

This reverts commit abf0768131db659c6819de9e7149624dd044c345.
The description is wrong, there is no recursive dependency here. The
conditions were added intentionally to avoid bogus build dependencies.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoiwinfo: fix nl80211 phy lookup without platform prefix
Jo-Philipp Wich [Mon, 4 Jul 2016 14:19:43 +0000 (16:19 +0200)]
iwinfo: fix nl80211 phy lookup without platform prefix

Commit d9b20a6f351c50d206fb6d2f965a16318c0cecf7 (SVN r48426) changed the
mac80211 phy lookup logic to strip the platform/ directory component from
the phy path specification.

Fix iwinfo to follow that logic by trying to lookup phys both with and
without "platform/" prefix.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agoe2fsprogs: fix build on OS X systems
Jo-Philipp Wich [Fri, 1 Jul 2016 14:41:46 +0000 (16:41 +0200)]
e2fsprogs: fix build on OS X systems

On OS X systems, the compilation of e2fsprogs fails at subst.c due to a
missing sys/stat.h include:

    subst.c:333:14: error: variable has incomplete type 'struct stat'
            struct stat stbuf;
                        ^
    subst.c:333:9: note: forward declaration of 'struct stat'
            struct stat stbuf;
                   ^
    subst.c:392:8: warning: implicit declaration of function 'fstat' is invalid in C99
          [-Wimplicit-function-declaration]
                            if (fstat(fd, &stbuf) == 0) {
                                ^
    subst.c:438:12: warning: implicit declaration of function 'fchmod' is invalid in C99
          [-Wimplicit-function-declaration]
                                    (void) fchmod(ofd, 0444);
                                           ^
    2 warnings and 1 error generated.
    make[3]: *** [subst.o] Error 1

Declare the nescessary HAVE_SYS_STAT_H macro to include the required header in
order to avoid the undeclared stat structure.

Tested-By: David Thornley <david.thornley@touchstargroup.com>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
8 years agobcm53xx: calculate Seama MD5 using content of kernel partition
Rafał Miłecki [Mon, 4 Jul 2016 10:20:54 +0000 (12:20 +0200)]
bcm53xx: calculate Seama MD5 using content of kernel partition

We don't want checksum to cover any part of UBI as even its part with
SquashFS may be changed due to e.g. flash wearing.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agoRevert "bcm53xx: use uncompressed zImages"
Rafał Miłecki [Mon, 4 Jul 2016 09:48:10 +0000 (11:48 +0200)]
Revert "bcm53xx: use uncompressed zImages"

This reverts commit b3be33f135f447d56ef9a5461e7e675d5cc35730.

CFE is known to fail in some non-standard cases, e.g. when using kernel
or format different that what was tested by Broadcom. This kernel change
triggered some problem with booting OpenWrt kernel stored in Seama.

As long as Seama checksum was covering enough data, it was working fine.
We need to change it however, because calculating MD5 over part of UBI
containing SquashFS is unsafe. UBI may move PEBs depending on flash
wearing level which would break CFE booting the image.

For some reason this kernel change was breaking CFE. Calculating MD5
over 0x47ffc0 B data or less (there is 0x40 B long header) was stopping
booting process at:
Starting program at 0x00008000

As our kernel is usually 0x3fffc0 B it was affected by this problem.

Reverting this change fixes sysupgrade (which already uses kernel size
for MD5 calculation) and will allow us to adjust "fixseama" command call
on the first boot.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
8 years agodosfstools: update to v4.0
Álvaro Fernández Rojas [Tue, 21 Jun 2016 06:33:50 +0000 (08:33 +0200)]
dosfstools: update to v4.0

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
8 years agoipq806x: enable ath10k firmware for 988x, 99x0 and 9984 by default
Felix Fietkau [Mon, 4 Jul 2016 09:39:24 +0000 (11:39 +0200)]
ipq806x: enable ath10k firmware for 988x, 99x0 and 9984 by default

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoustream-ssl: Fix recursive dependency
Daniel Dickinson [Mon, 4 Jul 2016 08:49:18 +0000 (04:49 -0400)]
ustream-ssl: Fix recursive dependency

Two variants incorrectly include themselves in
conditional depends on ssl libraries, which results
in a recursive dependency.

Signed-off-by: Daniel Dickinson <lede@daniel.thecshore.com>
8 years agoramips: fix up switch settings for Sitecom WL-351 v1 002
John Crispin [Wed, 29 Jun 2016 13:09:09 +0000 (15:09 +0200)]
ramips: fix up switch settings for Sitecom WL-351 v1 002

Pinmux for rgmii needs to be set to rgmii, not gpio.
Hide the ESW switch on boot (using new rgmii esw devicetree attribute).
Also add a Sitecom-specific profile, since the image needs to include
the rtl8366 kernel driver.

Signed-off-by: Tobias Diedrich <ranma+openwrt@tdiedrich.de>
8 years agox86: enable kmod-igb on 64-bit by default
Felix Fietkau [Sun, 3 Jul 2016 17:43:16 +0000 (19:43 +0200)]
x86: enable kmod-igb on 64-bit by default

It is used on APU2 devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agokernel: fix duplicate drivers for the PC speaker in one package
Felix Fietkau [Sun, 3 Jul 2016 17:35:42 +0000 (19:35 +0200)]
kernel: fix duplicate drivers for the PC speaker in one package

Only use the ALSA capable one, it supports regular beeps as well

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agox86: stop relying on hexdump for image build signature
Felix Fietkau [Sun, 3 Jul 2016 16:39:08 +0000 (18:39 +0200)]
x86: stop relying on hexdump for image build signature

Use perl to generate random number instead.
See https://github.com/freifunk-gluon/gluon/issues/811

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomac80211: move include statements for skb_get_hash_perturb() to prevent issues with...
Felix Fietkau [Sun, 3 Jul 2016 16:13:11 +0000 (18:13 +0200)]
mac80211: move include statements for skb_get_hash_perturb() to prevent issues with newer kernels

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomac80211: backport skb_get_hash_perturb() for 4.1 and older
Felix Fietkau [Sun, 3 Jul 2016 15:56:53 +0000 (17:56 +0200)]
mac80211: backport skb_get_hash_perturb() for 4.1 and older

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agokernel: mac80211: set the parent of the ath9k gpio_chip
Martin Blumenstingl [Fri, 1 Jul 2016 13:41:19 +0000 (15:41 +0200)]
kernel: mac80211: set the parent of the ath9k gpio_chip

This allows gpiolib to re-use ath9k's devicetree node as GPIO
controller.
Example:

ath9k: ath9k@0 {
#gpio-cells = <2>;
gpio-controller;
}

Now the ath9k node can be used just like any other GPIO controller:
gpios = <&ath9k 1 GPIO_ACTIVE_HIGH>;

Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
8 years agokernel: mac80211: enable the gpio controller for all ath9k devices
Martin Blumenstingl [Fri, 1 Jul 2016 13:41:18 +0000 (15:41 +0200)]
kernel: mac80211: enable the gpio controller for all ath9k devices

This enables ath9k's built-in GPIO controller for all chip versions
(instead of an explicit whitelist). This also allows us to get rid of
some duplicate code between hw.c and gpio.c because hw.c already
determines the number of GPIOs.

Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
8 years agokernel: mac80211: fold the AR9280 GPIO patch into the ath9k GPIO patch
Martin Blumenstingl [Fri, 1 Jul 2016 13:41:17 +0000 (15:41 +0200)]
kernel: mac80211: fold the AR9280 GPIO patch into the ath9k GPIO patch

This folds 550-ath9k_add_ar9280_gpio_chip.patch into
548-ath9k_enable_gpio_chip.patch because the former patch only extends
code which is introduced in the latter.

Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
8 years agomac80211: refresh patches
Mathias Kresin [Fri, 1 Jul 2016 13:41:16 +0000 (15:41 +0200)]
mac80211: refresh patches

Signed-off-by: Mathias Kresin <dev@kresin.me>
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
8 years agoath10k: fix tx performance regression on older chipsets
Felix Fietkau [Wed, 29 Jun 2016 09:01:16 +0000 (11:01 +0200)]
ath10k: fix tx performance regression on older chipsets

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomac80211: enable STBC and LDPC for VHT rates
Felix Fietkau [Wed, 29 Jun 2016 08:44:39 +0000 (10:44 +0200)]
mac80211: enable STBC and LDPC for VHT rates

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoath10k-firmware: add symlink for QCA9984 board.bin
Felix Fietkau [Sun, 26 Jun 2016 19:47:51 +0000 (21:47 +0200)]
ath10k-firmware: add symlink for QCA9984 board.bin

Use first caldata for devices without OTP. The driver uses the caldata
instead of the board.bin data anyway

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agomac80211: update to wireless-testing 2016-06-20
Felix Fietkau [Sun, 26 Jun 2016 17:00:01 +0000 (19:00 +0200)]
mac80211: update to wireless-testing 2016-06-20

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoimx6: fix redundant PROFILES overrides in image makefile
Felix Fietkau [Sat, 2 Jul 2016 17:33:52 +0000 (19:33 +0200)]
imx6: fix redundant PROFILES overrides in image makefile

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoimx6: fix build breakage introduced in 9e41c3f54e3c
Felix Fietkau [Sat, 2 Jul 2016 17:33:10 +0000 (19:33 +0200)]
imx6: fix build breakage introduced in 9e41c3f54e3c

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agoimx6: fix image makefile indentation for Device/ sections
Felix Fietkau [Sat, 2 Jul 2016 17:12:09 +0000 (19:12 +0200)]
imx6: fix image makefile indentation for Device/ sections

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agokernel: fix build of kmod-udptunnel4 and kmod-udptunnel6
Hauke Mehrtens [Sat, 2 Jul 2016 16:03:56 +0000 (18:03 +0200)]
kernel: fix build of kmod-udptunnel4 and kmod-udptunnel6

The kernel config option CONFIG_NET_UDP_TUNNEL is not visible and can
not directly be activated. When kmod-udptunnel4 or kmod-udptunnel6 are
build these packages could be empty when no other kernel module selects
CONFIG_NET_UDP_TUNNEL.

Reported-by: Baptiste Jonglez <baptiste@bitsofnetworks.org>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
8 years agojsonfilter: update to latest git HEAD
John Crispin [Sat, 2 Jul 2016 08:15:20 +0000 (10:15 +0200)]
jsonfilter: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoprocd: Set /dev/kmsg to 600
neheb [Sat, 2 Jul 2016 01:32:27 +0000 (18:32 -0700)]
procd: Set /dev/kmsg to 600

Small cleanup. I initially though /dev/kmsg was used for dmsg(and journald
on desktops) but this seems not to be the case. dmsg is still accessible
as non-root(gives output) which begs the question what does this do? Some
googling reveals that permissions are set to 600 for some embedded systems
while 644 for others. I can't find any justification for the latter. Might
as well err on the side of caution.

Signed-off by: Rosen Penev <rosenp@gmail.com>

8 years agofstools: update to latest git HEAD
John Crispin [Fri, 1 Jul 2016 13:27:04 +0000 (15:27 +0200)]
fstools: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoprocd: update to latest git HEAD
John Crispin [Fri, 1 Jul 2016 13:23:44 +0000 (15:23 +0200)]
procd: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoubox: update to latest git HEAD
John Crispin [Fri, 1 Jul 2016 13:20:20 +0000 (15:20 +0200)]
ubox: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoustream-ssl: update to latest git HEAD
John Crispin [Fri, 1 Jul 2016 13:18:55 +0000 (15:18 +0200)]
ustream-ssl: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoubus: update to latest git HEAD
John Crispin [Fri, 1 Jul 2016 13:13:12 +0000 (15:13 +0200)]
ubus: update to latest git HEAD

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoar71xx: spi-rb4xx fix.
Sergey Sergeev [Thu, 30 Jun 2016 10:23:15 +0000 (13:23 +0300)]
ar71xx: spi-rb4xx fix.

In new kernels we should use clk_prepare_enable instead of clk_enable
since clk_enable does not make proper initialization that leads
to rise WARN_ON messages and not working spi bus on the device.

Signed-off-by: Sergey Sergeev <adron@yapic.net>
8 years agonand-utile: add package
Sergey Sergeev [Tue, 28 Jun 2016 12:23:24 +0000 (15:23 +0300)]
nand-utile: add package

This package is a custom build(like ubi-utils) of mtd-utils from infradead.org
It is required to work with Mikrotik NAND based devices

Signed-off-by: Sergey Sergeev <adron@yapic.net>
8 years agoar71xx: prevent spurious ethernet resets from dma hang check false positives
Felix Fietkau [Fri, 17 Jun 2016 21:58:11 +0000 (23:58 +0200)]
ar71xx: prevent spurious ethernet resets from dma hang check false positives

Signed-off-by: Felix Fietkau <nbd@nbd.name>
8 years agolibnl-tiny: Generic Netlink multicast groups support
Hauke Mehrtens [Wed, 29 Jun 2016 21:13:58 +0000 (23:13 +0200)]
libnl-tiny: Generic Netlink multicast groups support

This adds this commit from normal libnl to libnl-tiny:
https://github.com/tgraf/libnl/commit/2dbc1ca76c5b82c40749e609eb83877418abb006

commit 2dbc1ca76c5b82c40749e609eb83877418abb006
Author: dima <dima.ky@gmail.com>
Date:   Wed Oct 13 17:53:34 2010 +0300

    Generic Netlink multicast groups support

    I have a patch against commit d378220c96c3c8b6f27dca33e7d8ba03318f9c2d
    extending libnl with a facility to receive generic netlink messages sent
    to multicast groups.

    Essentially it add one new function genl_ctrl_resolve_grp which
    prototype looks like this
    int genl_ctrl_resolve_grp(struct nl_sock *sk, const char *family_name,
            const char *grp_name)
    It resolves  the family name and the group name to group id. Then
    the returned id can be used in nl_socket_add_membership to subscribe
    to multicast messages.

    Besides that it adds two more functions

    uint32_t nl_socket_get_peer_groups(struct nl_sock *sk)
    void nl_socket_set_peer_groups(struct nl_sock *sk, uint32_t groups)

    allowing to modify the socket peer groups field. So it's possible to
    multicast messages from the user space using the legacy interface.
    Looks like there is no way (or I was not able to find one?) to modify
    the netlink socket destination group from the user space, when the
    group id is greater then 32.

Signed-off-by: Hauke Mehrtens <hauke.mehrtens@intel.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name> [cosmetic style fix]
8 years agokernel: Fix ipv6 mc snooping if bridge has no ipv6 address
daniel [Thu, 30 Jun 2016 21:49:52 +0000 (23:49 +0200)]
kernel: Fix ipv6 mc snooping if bridge has no ipv6 address

This patch is a backport for current LEDE 4.4 Kernels.
It is already upstream, for linux-next and stable.
The initial commit message is below:

The bridge is falsly dropping ipv6 mulitcast packets if there is:
 1. No ipv6 address assigned on the brigde.
 2. No external mld querier present.
 3. The internal querier enabled.

When the bridge fails to build mld queries, because it has no
ipv6 address, it slilently returns, but keeps the local querier enabled.
This specific case causes confusing packet loss.

Ipv6 multicast snooping can only work if:
 a) An external querier is present
 OR
 b) The bridge has an ipv6 address an is capable of sending own queries

Otherwise it has to forward/flood the ipv6 multicast traffic,
because snooping cannot work.

This patch fixes the issue by adding a flag to the bridge struct that
indicates that there is currently no ipv6 address assinged to the bridge
and returns a false state for the local querier in
__br_multicast_querier_exists().

Special thanks to Linus Lüssing.

Signed-off-by: Daniel Danzberger <daniel@dd-wrt.com>
8 years agospidev_test: copy the source code into the package folder
John Crispin [Thu, 30 Jun 2016 11:05:22 +0000 (13:05 +0200)]
spidev_test: copy the source code into the package folder

Signed-off-by: John Crispin <john@phrozen.org>
8 years agokernel: other.mk: add pps-ldisc support
Stephen Walker [Wed, 29 Jun 2016 22:34:50 +0000 (18:34 -0400)]
kernel: other.mk: add pps-ldisc support

The pps-ldisc kernel module supports Pulse-Per-Second connected with the CD (Carrier Detect) pin.

Signed-off-by: Stephen Walker <stephendwalker+github@gmail.com>
8 years agoar71xx: seama: fix making factory images
Yousong Zhou [Thu, 30 Jun 2016 01:54:21 +0000 (09:54 +0800)]
ar71xx: seama: fix making factory images

rootfs part needs to be aligned to erase block size which is passed as
the 6th argument to Image/Build/Seama  and is now 65536 since commit
commit 5119ee9 "ar71xx: fix bogus hardcoded kernel image size for Seama
images (fixes #20585)", but $(($(6) - 64)) still assumes that the
argument is a limit on kernel partition size, i.e. 1310720, so the
generated factory image is wrong in that the kernel will fail to find
the rootfs (FlySpray link at [1])

This patch will workaround it with the following steps

1. Calculate the required space for seama header and META data in step 5
2. Pre-padding 64 bytes to lzma-compressed loader
3. Generate correctly padded image-$(2).tmp
4. Strip out the padding
5. Seal it with seama utility

While at it convert seama to new build method

[1] FS#35 - mynet-n750 factory images don't find root partition,
    https://bugs.lede-project.org/index.php?do=details&task_id=35

Reported-by: Steven Haigh <netwiz@crc.id.au>
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
8 years agoar71xx: image: unify indentation with 2 whitespaces
Yousong Zhou [Thu, 30 Jun 2016 01:54:20 +0000 (09:54 +0800)]
ar71xx: image: unify indentation with 2 whitespaces

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
8 years agoar71xx: image: remove duplicate IMAGES definition
Yousong Zhou [Thu, 30 Jun 2016 01:54:19 +0000 (09:54 +0800)]
ar71xx: image: remove duplicate IMAGES definition

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
8 years agoar71xx: hiwifi-hc6361: move packages selection to generic.mk
Yousong Zhou [Thu, 30 Jun 2016 01:54:18 +0000 (09:54 +0800)]
ar71xx: hiwifi-hc6361: move packages selection to generic.mk

Remove kmod-crypto-deflate and kmod-ledtrig-gpio as the device works the
same by default without them

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
8 years agoprocd: change /dev/{gpio,hvc*} perms to 0600
John Crispin [Wed, 29 Jun 2016 13:44:30 +0000 (15:44 +0200)]
procd: change /dev/{gpio,hvc*} perms to 0600

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoprocd: adjust /dev entries to desktop distro defaults
neheb [Sat, 25 Jun 2016 05:39:25 +0000 (22:39 -0700)]
procd: adjust /dev entries to desktop distro defaults

This changes the default permissions for /dev entries to be more similar to
a desktop distro. Taken from the defaults of Arch Linux and Ubuntu. Also
changed some that were nonsensical. For example, all 660 permissions on
desktop distros were of the form root:x where x is something other than
root. As such, 660 is useless for LEDE where the specific group is missing.
audio seems to be the only group that isn't.

Signed-off by: Rosen Penev <rosenp@gmail.com>

8 years agoramips: disable the WP pin on the SDK mmc driver
John Crispin [Tue, 28 Jun 2016 09:15:31 +0000 (11:15 +0200)]
ramips: disable the WP pin on the SDK mmc driver

Signed-off-by: John Crispin <john@phrozen.org>
8 years agoramips: fix pinmux typo
John Crispin [Tue, 28 Jun 2016 08:59:56 +0000 (10:59 +0200)]
ramips: fix pinmux typo

Reported-by: Joy Kicer <joykicer@gmail.com>
Signed-off-by: John Crispin <john@phrozen.org>
8 years agobase-files: seed /dev/urandom
Etienne CHAMPETIER [Mon, 27 Jun 2016 15:53:30 +0000 (15:53 +0000)]
base-files: seed /dev/urandom

This commit:
1) seed /dev/urandom with the saved seeds as early as possible
   (see /lib/preinit/81_urandom_seed)
2) save a seed at /etc/urandom.seed if it doesn't exists
3) save a new seed each boot at "system.@system[0].urandom_seed"
   (see /etc/init.d/urandom_seed)

We use getrandom() so we are sure /dev/urandom pool is initialized

Seed size is 512 bytes (ie /proc/sys/kernel/random/poolsize / 8)
it's the same size as in ubuntu 14.04 and all systemd systems

Seeding /dev/urandom doesn't change entropy estimation, so we still have
"random: ubus urandom read with 4 bits of entropy available"
messages in the logs, but we can now ignore them if
after "urandom-seed: Seeding with ..." message

Saving a new seed on each boot is disabled by default to avoid too much
writes without user consent

v2: log preinit messages to /dev/kmsg
v3: use non generic function name for logging, as /lib/preinit/ files
    are all sourced together in /etc/preinit
v4: after a lot of discussion on the ML, use a uci config param
v5: config param is now the path of the seed

Signed-off-by: Etienne CHAMPETIER <champetier.etienne@gmail.com>
Acked-by: Jo-Philipp Wich <jo@mein.io>