openwrt/staging/981213.git
7 years agouboot-envtools: drop executable file attribute from files/ipq
Piotr Dymacz [Tue, 13 Dec 2016 21:54:15 +0000 (22:54 +0100)]
uboot-envtools: drop executable file attribute from files/ipq

Signed-off-by: Piotr Dymacz <pepe2k@gmail.com>
7 years agouboot-envtools: fix Edimax BR-6425 board name
Piotr Dymacz [Tue, 13 Dec 2016 21:44:35 +0000 (22:44 +0100)]
uboot-envtools: fix Edimax BR-6425 board name

Signed-off-by: Piotr Dymacz <pepe2k@gmail.com>
7 years agouboot-envtools: fix code formatting style in uci-defaults files
Piotr Dymacz [Tue, 13 Dec 2016 21:30:11 +0000 (22:30 +0100)]
uboot-envtools: fix code formatting style in uci-defaults files

Few minor code formatting style fixes, including:

- keep one board per line
- always use "|\" (for consistency)
- remove redundant double quotes and empty lines

Signed-off-by: Piotr Dymacz <pepe2k@gmail.com>
7 years agouboot-envtools: keep boards in alphabetical order
Piotr Dymacz [Tue, 13 Dec 2016 21:09:07 +0000 (22:09 +0100)]
uboot-envtools: keep boards in alphabetical order

Signed-off-by: Piotr Dymacz <pepe2k@gmail.com>
7 years agouqmi: Prevent 'POLICY MISMATH' error.
Nickolay Ledovskikh [Fri, 16 Dec 2016 12:41:06 +0000 (15:41 +0300)]
uqmi: Prevent 'POLICY MISMATH' error.

Add uqmi 'sync' command call to release stalled cid when preparing to
setup new connection. As a result it prevents 'POLICY MISMATCH' errors.

Signed-off-by: Nickolay Ledovskikh <nledovskikh@gmail.com>
7 years agouqmi: bump to latest git HEAD
John Crispin [Mon, 19 Dec 2016 08:34:20 +0000 (09:34 +0100)]
uqmi: bump to latest git HEAD

8ceeab6 uqmi: Change returned value to QMI_CMD_REQUEST for 'sync' command.
1dc7be1 uqmi: Add sync command to release all cids.

Signed-off-by: John Crispin <john@phrozen.org>
7 years agoramips: WNDR3700v5 fix mtd partitions and radios
L. D. Pinney [Sat, 17 Dec 2016 00:59:53 +0000 (18:59 -0600)]
ramips: WNDR3700v5 fix mtd partitions and radios

Fix the mtd partition layout and enable both radios in the dts

Tested-by Andrius Štikonas <andrius@stikonas.eu>
Signed-off-by: L. D. Pinney <ldpinney@gmail.com>
7 years agouqmi: add support of using device symlinks.
Nickolay Ledovskikh [Fri, 16 Dec 2016 11:14:05 +0000 (14:14 +0300)]
uqmi: add support of using device symlinks.

It's useful when using multiple usb devices that should be bound to
certain usb ports. Symlinks are created by hotplug handlers.

Signed-off-by: Nickolay Ledovskikh <nledovskikh@gmail.com>
7 years agocomgt: add support of using device symlinks.
Nickolay Ledovskikh [Fri, 16 Dec 2016 11:14:05 +0000 (14:14 +0300)]
comgt: add support of using device symlinks.

It's useful when using multiple usb devices that should be bound to
certain usb ports. Symlinks are created by hotplug handlers.

Signed-off-by: Nickolay Ledovskikh <nledovskikh@gmail.com>
7 years agohostapd: remove never-used Package/<name>/Description
Yousong Zhou [Sat, 17 Dec 2016 09:06:01 +0000 (17:06 +0800)]
hostapd: remove never-used Package/<name>/Description

The build system only accepts Package/<name>/description and since the
typoed version virtually has the same content as the TITLE field, remove
them altogether

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
7 years agomdns: bump to latest git HEAD
John Crispin [Mon, 19 Dec 2016 07:51:55 +0000 (08:51 +0100)]
mdns: bump to latest git HEAD

be8ae8d cmake: Search for libjson-c
1fa9077 Fix IPv6 read
846369c Revert "mdnsd: interface: enable looped back messages"

Signed-off-by: John Crispin <john@phrozen.org>
7 years agomountd: update to latest git HEAD
John Crispin [Mon, 19 Dec 2016 07:40:04 +0000 (08:40 +0100)]
mountd: update to latest git HEAD

dd5799c mountd: Add static qualifier to missing variable declarations.
bc4a6dd mountd: Fix fgets check.
3dd115b cmake: Really look for uci.h

Signed-off-by: John Crispin <john@phrozen.org>
7 years agotar: fix reproducibility issue
Felix Fietkau [Mon, 19 Dec 2016 20:06:07 +0000 (21:06 +0100)]
tar: fix reproducibility issue

Force root/root as names for uid0/gid0 instead of using the system
names. This helps make packed download tarballs more reproducible

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agousbutils: fix download filename
Felix Fietkau [Fri, 16 Dec 2016 16:17:02 +0000 (17:17 +0100)]
usbutils: fix download filename

The name needs to include the revision, otherwise it is not
automatically re-downloaded on updates.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoRevert "ath9k: Add airtime fairness scheduler"
Felix Fietkau [Mon, 19 Dec 2016 13:07:58 +0000 (14:07 +0100)]
Revert "ath9k: Add airtime fairness scheduler"

Accidentally pushed to the wrong branch

This reverts commit 47bc081e76f6b04c0e9db687c8abb78eb437021f.

7 years agoath9k: Add airtime fairness scheduler
Toke Høiland-Jørgensen [Mon, 19 Dec 2016 12:59:15 +0000 (13:59 +0100)]
ath9k: Add airtime fairness scheduler

This adds a patch that introduces airtime fairness scheduling to ath9k,
which can significantly improve network efficiency in mixed-rate
environments.

Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: backport "cfg80211: limit scan results cache size"
Christian Lamparter [Tue, 13 Dec 2016 17:35:27 +0000 (18:35 +0100)]
mac80211: backport "cfg80211: limit scan results cache size"

The patch commit states:
"It's possible to make scanning consume almost arbitrary amounts
of memory, e.g. by sending beacon frames with random BSSIDs at
high rates while somebody is scanning.

Limit the number of BSS table entries we're willing to cache to
1000, limiting maximum memory usage to maybe 4-5MB, but lower
in practice - that would be the case for having both full-sized
beacon and probe response frames for each entry; this seems not
possible in practice, so a limit of 1000 entries will likely be
closer to 0.5 MB."

Signed-off-by: Johannes Berg <johannes.berg@intel.com>"
This patch was added in 4.4.36. But because LEDE backports
cfg80211, mac80211 and the wifi drivers separately, it needs
to be added manually for now. It can be dropped later as it
will be part of the next mac80211 refresh.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
7 years agomac80211: merge a number of minstrel/minstrel_ht performance and memory usage improve...
Felix Fietkau [Thu, 15 Dec 2016 10:48:12 +0000 (11:48 +0100)]
mac80211: merge a number of minstrel/minstrel_ht performance and memory usage improvements

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoodhcpd: Bump to git HEAD version (various fixes)
Hans Dedecker [Sat, 17 Dec 2016 21:16:25 +0000 (22:16 +0100)]
odhcpd: Bump to git HEAD version (various fixes)

e055530 Don't print non bound assignments in the state file
3af23ad config: Fix RA interface config being overwritten
41b5268 dhcpv6-ia : Fix static DHCPv6 assignments becoming non static
be6c515 dhcpv6-ia: Fix assignment of static DHCPv6 leases
374dc3f cmake: Find libubox/uloop.h
01c919c odhcpd: Display infinite valid lifetime as -1

7 years agoscripts/update-package-md5sum: remove file, it is obsoleted by make check FIXUP=1
Felix Fietkau [Fri, 16 Dec 2016 16:19:20 +0000 (17:19 +0100)]
scripts/update-package-md5sum: remove file, it is obsoleted by make check FIXUP=1

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: add CHECK_ALL variable to allow make download/check to include not selected...
Felix Fietkau [Fri, 16 Dec 2016 15:08:44 +0000 (16:08 +0100)]
build: add CHECK_ALL variable to allow make download/check to include not selected packages

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: add FIXUP option for make check
Felix Fietkau [Fri, 16 Dec 2016 12:34:37 +0000 (13:34 +0100)]
build: add FIXUP option for make check

This will attempt to automatically fix common mistakes like using MD5
instead of SHA256, using the MD5SUM variable instead of HASH, or even a
missing mirror file hash.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: implement make check and make package/X/check
Felix Fietkau [Wed, 14 Dec 2016 14:36:39 +0000 (15:36 +0100)]
build: implement make check and make package/X/check

This is intended to be used for a wide array of package sanity checks.

The first check that is implemented is for the hash of downloaded files.
It checks:
  - Missing hash
  - Use of SHA256 instead of MD5
  - dl/<file> hash not matching hash in makefile
  - deprecated MD5SUM variable

The deprecated MD5SUM variable check is skipped for feeds/ until OpenWrt
is updated as well

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agotreewide: clean up download hashes
Felix Fietkau [Fri, 16 Dec 2016 16:32:10 +0000 (17:32 +0100)]
treewide: clean up download hashes

Replace *MD5SUM with *HASH, replace MD5 hashes with SHA256

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k-ct: Update to latest CT ath10k driver.
Ben Greear [Fri, 9 Dec 2016 20:13:40 +0000 (12:13 -0800)]
ath10k-ct: Update to latest CT ath10k driver.

This works around deadlock and/or memory corruption during
firmware crash and improves ability to configure number of
tids in firmware.

Signed-off-by: Ben Greear <greearb@candelatech.com>
7 years agoodhcp6c: Pass parameters to user dhcpv6 script
Hans Dedecker [Fri, 16 Dec 2016 17:35:03 +0000 (18:35 +0100)]
odhcp6c: Pass parameters to user dhcpv6 script

Pass all the parameters like device, dhcpv6 state to user script

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agomap: Have cmake find libubus.h
Florian Fainelli [Tue, 13 Dec 2016 22:55:38 +0000 (14:55 -0800)]
map: Have cmake find libubus.h

Update CMakeList.txt to look for libubus.h since we depend on it.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
7 years agoinclude/package.mk: sync default value for hash fallback with mirror hash
Felix Fietkau [Fri, 16 Dec 2016 14:29:36 +0000 (15:29 +0100)]
include/package.mk: sync default value for hash fallback with mirror hash

Simplifies dealing with PKG_HASH vs PKG_MD5SUM

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: remove duplicate Download/default definition from include/host-build.mk
Felix Fietkau [Fri, 16 Dec 2016 14:26:22 +0000 (15:26 +0100)]
build: remove duplicate Download/default definition from include/host-build.mk

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodownload.pl: check for existing file before the first download attempt
Felix Fietkau [Fri, 16 Dec 2016 14:15:48 +0000 (15:15 +0100)]
download.pl: check for existing file before the first download attempt

In the build system, flock will prevent multiple concurrent downloads
for the same file. However, if one download request for the same file is
waiting for another one to finish, it will result in downloading the
same file twice consecutively.

Prevent this issue by exiting immediately if the file has already been
downloaded

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k-firmware: fix typo
Felix Fietkau [Fri, 16 Dec 2016 14:08:57 +0000 (15:08 +0100)]
ath10k-firmware: fix typo

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agopackage/kernel/linux: only access kernel config if DUMP is unset
Felix Fietkau [Wed, 14 Dec 2016 14:35:16 +0000 (15:35 +0100)]
package/kernel/linux: only access kernel config if DUMP is unset

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k-firmware: untangle CT firmware filenames, fix conflicts
Felix Fietkau [Fri, 16 Dec 2016 13:36:53 +0000 (14:36 +0100)]
ath10k-firmware: untangle CT firmware filenames, fix conflicts

CT firmware files use the same filename with a different URL for
different chips. Since all files end up in dl/, filenames need to be
unique as well.
Add a chip prefix to the output filename to fix this issue.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobuild: replace MD5SUM variables with HASH
Felix Fietkau [Wed, 14 Dec 2016 16:11:03 +0000 (17:11 +0100)]
build: replace MD5SUM variables with HASH

Since we've switched to preferring SHA256 over MD5, the old variable
name is misleading. Packages using the old name remain compatible.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: add support for TP-LINK TL-WR940N v4
David Lutz [Mon, 12 Dec 2016 14:59:03 +0000 (15:59 +0100)]
ar71xx: add support for TP-LINK TL-WR940N v4

Signed-off-by: David Lutz <kpanic@hirnduenger.de>
7 years agodnsmasq: fix service reload
John Crispin [Fri, 16 Dec 2016 09:34:52 +0000 (10:34 +0100)]
dnsmasq: fix service reload

The SIGHUP also got sent to the reload script making it bail out
with an error

Revert "dnsmasq: reload config if host name is modified"
This reverts commit 854459a2f923376e0e509ebc0fb8ff90e9f13c02.

Reported-by: Hans Dedecker <dedeckeh@gmail.com>
Signed-off-by: John Crispin <john@phrozen.org>
7 years agorealview: drop the target
Yousong Zhou [Thu, 8 Dec 2016 15:21:31 +0000 (23:21 +0800)]
realview: drop the target

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
7 years agoarmvirt: new target
Yousong Zhou [Thu, 8 Dec 2016 15:21:30 +0000 (23:21 +0800)]
armvirt: new target

An ARM Cortex-A15 machine provided by QEMU.

Kernel drivers enabled:

 - pl011, uart
 - pl031, rtc
 - pl061, gpio
 - pci-host-generic
 - virtio_{mmio,pci,net,blk,scsi,9p,console,balloon}
 - smp with NR_CPUS=4
 - cpu-hotplug
 - ext4
 - neon, vfp extensions support (otherwise userland will fail with
   illegal instruction signal (code 0x00000004))
 - DEBUG_BUGVERBOSE for debug purposes

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
7 years agomt76: update to the latest version, fixes a build error on some platforms
Felix Fietkau [Thu, 15 Dec 2016 12:02:31 +0000 (13:02 +0100)]
mt76: update to the latest version, fixes a build error on some platforms

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agox86: revert default root size back to 256 MB
Felix Fietkau [Thu, 15 Dec 2016 10:43:34 +0000 (11:43 +0100)]
x86: revert default root size back to 256 MB

2 GB is overkill and was only added to allow unlimited ext4 resizing,
which is a pretty rare use case. 256 MB allows resizing up to 256 GB,
which should be good enough for almost all users.

A lot of this is mostly irrelevant anyway, since you can just use
squashfs + ext4 overlay.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoodhcpd: update sha256sum
Hauke Mehrtens [Wed, 14 Dec 2016 20:43:58 +0000 (21:43 +0100)]
odhcpd: update sha256sum

The sha256sum was not updated in the last commit.

Fixes: a7c231027 [odhcpd: Fix dnsmasq re-reading hostfile]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agobase-files: fix CONFIG_VERSION_DIST default
Jo-Philipp Wich [Wed, 14 Dec 2016 18:54:51 +0000 (19:54 +0100)]
base-files: fix CONFIG_VERSION_DIST default

Fix the default value of CONFIG_VERSION_DIST to use uppercase "LEDE".

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobuild: use SNAPSHOT instead of CURRENT to designate untagged branch builds
Jo-Philipp Wich [Fri, 2 Dec 2016 21:11:40 +0000 (22:11 +0100)]
build: use SNAPSHOT instead of CURRENT to designate untagged branch builds

Calling a build ##.##-CURRENT might mislead users into thinking that this
build is the most current release of a branch.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoar71xx: fix TL-WR842N v2 switch port order
Jo-Philipp Wich [Tue, 6 Dec 2016 17:20:07 +0000 (18:20 +0100)]
ar71xx: fix TL-WR842N v2 switch port order

The TL-WR842N v2 has the following switch port layout with swconfig port 0
being connected to eth1:

            Power  LAN4    LAN3    LAN2    LAN1    WAN     USB
             (*) | [...] | [...] | [...] | [...] | [...] | [.]
    swconfig port:   1       4       3       2      n/a
    netdev:        eth1    eth1    eth1    eth1    eth0

Adjust the board defaults to match this model specific layout.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoar71xx: Added missing support for Linksys E2100L
Gad Krumholz [Sun, 27 Nov 2016 06:52:53 +0000 (00:52 -0600)]
ar71xx: Added missing support for Linksys E2100L

It's based on the WRT160NL according to
https://wiki.openwrt.org/toh/linksys/e2100l

Based on research done here: https://forum.openwrt.org/viewtopic.php?id=24244
and here: https://forum.openwrt.org/viewtopic.php?pid=120791#p120791 this patch
was conceived.

Signed-off-by: Gad Krumholz <gad.krumholz@gmail.com>
[Jo-Philipp Wich: fix ordering, move addpattern change into separate commit]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agofirmware-utils: add E2100L support to addpattern.c
Jo-Philipp Wich [Fri, 2 Dec 2016 10:39:42 +0000 (11:39 +0100)]
firmware-utils: add E2100L support to addpattern.c

Based on OpenWrt PR#250

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agolinux: do not autoload sdhci.ko as sdhci-pltfm.ko already depends on it
Jo-Philipp Wich [Tue, 6 Dec 2016 16:09:34 +0000 (17:09 +0100)]
linux: do not autoload sdhci.ko as sdhci-pltfm.ko already depends on it

Currently, installing kmod-sdhci fails with "sdhci is already loaded" since
"sdhci.ko" is inserted explicitely first, the implicitely loaded again when
"sdhci-pltfm.ko" is inserted as the latter depends on the former.

Remove the explicit autoprobe of "sdhci.ko" to fix the postinstall script.

Acked-by: John Crispin <john@phrozen.org>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobrcm63xx: image: trim revision code used for --rsa-signature
Jo-Philipp Wich [Wed, 14 Dec 2016 16:10:52 +0000 (17:10 +0100)]
brcm63xx: image: trim revision code used for --rsa-signature

The imagetag RSA signature field may not exceed 20 characters, so trim the
Git hash portion from the REVISION variable in order to ensure that the
resulting string fits within this field.

The current revision values may be longer than 20 chars, especially when
building within a dirty tree. In this case, the revision will look like
"r2435+75-b4aa3c8" which, combined with the "LEDE-" prefix, is 21 chars long.

Fixes the following error spotted by the buildbots:

    Error: RSA Signature (rsa_signature,r) too long.
    make[4]: *** [.../lede-brcm63xx-generic-NEUFBOX6-squashfs-cfe.bin] Error 1

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoipkg-make-index.sh: drop a few non-essential fields
Felix Fietkau [Tue, 13 Dec 2016 14:20:03 +0000 (15:20 +0100)]
ipkg-make-index.sh: drop a few non-essential fields

Saves some more space on the package index

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoipkg-make-index.sh: drop md5sum from package index
Felix Fietkau [Tue, 13 Dec 2016 14:08:54 +0000 (15:08 +0100)]
ipkg-make-index.sh: drop md5sum from package index

We have switched opkg to sha256 a long time ago, and shrinking package
lists is useful for systems that are running low on RAM

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: netfilter: split out iptable_raw into a separate package
Felix Fietkau [Tue, 13 Dec 2016 13:54:55 +0000 (14:54 +0100)]
kernel: netfilter: split out iptable_raw into a separate package

This will avoid loading it in the default configuration, which reduces
image size a bit, and (more importantly) improves performance by
avoiding some unnecessary netfilter hooks

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agogcc: rip out transactional memory related bloat from crtbegin
Felix Fietkau [Tue, 13 Dec 2016 13:20:49 +0000 (14:20 +0100)]
gcc: rip out transactional memory related bloat from crtbegin

Slightly improves compression for each executable, saving about 4k from
the default ar71xx rootfs

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoswconfig: replace the shared library with a static one
Felix Fietkau [Tue, 13 Dec 2016 12:15:05 +0000 (13:15 +0100)]
swconfig: replace the shared library with a static one

Reduces binary size

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoppp: use --gc-sections to save a tiny bit of space
Felix Fietkau [Tue, 13 Dec 2016 12:09:58 +0000 (13:09 +0100)]
ppp: use --gc-sections to save a tiny bit of space

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agolibs/libpcap: update to 1.8.1
p-wassi [Sat, 10 Dec 2016 09:56:22 +0000 (10:56 +0100)]
libs/libpcap: update to 1.8.1

Update libpcap to upstream release 1.8.1
Change the name from libpcap.so.1.3 to libpcap.so.1
Remove parts of patch 201 which moved code among src files.
Import patch 204 from Debian to update the USB path.

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
Signed-off-by: Felix Fietkau <nbd@nbd.name> [fix parallel build bug]
7 years agoath10k: fix a soft-lockup on firmware restart
Felix Fietkau [Tue, 13 Dec 2016 11:52:16 +0000 (12:52 +0100)]
ath10k: fix a soft-lockup on firmware restart

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k: fix a bug on sending null-func frames
Felix Fietkau [Tue, 13 Dec 2016 11:52:01 +0000 (12:52 +0100)]
ath10k: fix a bug on sending null-func frames

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath10k-firmware: removed broken submenu
Alberto Bursi [Sat, 10 Dec 2016 16:02:58 +0000 (17:02 +0100)]
ath10k-firmware: removed broken submenu

this package references an undefined variable for its submenu.
Remove this NOP variable assignment.

Signed-off-by: Alberto Bursi <alberto.bursi@outlook.it>
7 years agokernel: add a missing submenu
Alberto Bursi [Sun, 11 Dec 2016 18:51:04 +0000 (19:51 +0100)]
kernel: add a missing submenu

this kernel module currently does not set submenu.
Fix this by adding it to the "Others" submenu

Signed-off-by: Alberto Bursi <alberto.bursi@outlook.it>
7 years agocomgt-ncm: Add support for specifying profile index
Matti Laakso [Sun, 11 Dec 2016 15:28:52 +0000 (17:28 +0200)]
comgt-ncm: Add support for specifying profile index

Add support for specifying a call profile index instead of APN. A
specific index different from 1 must be used for some service
provider and modem combinations.

In addition, change the manufacturer detection to use the standard
AT+CGMI command, which produces more predictable output than ATI,
remove the redundant ipv6 option, since it is less ambiguous to
directly specify the PDP context type with mobile connections, and
fix missing device during teardown when using ncm through the wwan
proto.

Signed-off-by: Matti Laakso <malaakso@elisanet.fi>
7 years agouqmi: Add support for specifying profile index
Matti Laakso [Sun, 11 Dec 2016 15:23:38 +0000 (17:23 +0200)]
uqmi: Add support for specifying profile index

Update uqmi to latest version, which brings about support for
specifying a call profile index instead of APN. A specific index
different from 1 must be used for some service provider and modem
combinations.

Also change option dhcp to dhcpv6, since IPv4 now always uses DHCP,
replace option ipv6 with pdptype, which is less ambiguous, and
make autoconnect optional and default it to off for IPv6 due to it
not working with statically configured IPv6.

Signed-off-by: Matti Laakso <malaakso@elisanet.fi>
7 years agodropbear: clean up default PATH handling in makefile
Dario Ernst [Tue, 29 Nov 2016 15:02:48 +0000 (16:02 +0100)]
dropbear: clean up default PATH handling in makefile

Harmonise handling of DEFAULT_PATH by removing the patch introducing #ifndef
guards around the path, and only using one means to set the path in the
makefile.

Signed-off-by: Dario Ernst <Dario.Ernst@riverbed.com>
7 years agoramips: adding DWM-158 3g Modem
Giuseppe Lippolis [Fri, 9 Dec 2016 18:25:00 +0000 (19:25 +0100)]
ramips: adding DWM-158 3g Modem

This patch add the 3G modem DWM-158 to the usb-serial option driver.
The DWM-158 is a pcie 3G modem. It is embedded in the DWR-512 modem
supportet by lede. The patch has been submitted and merged in the upstream
linux-next repository.

Signed-off-by: Giuseppe Lippolis <giu.lippolis@gmail.com>
7 years agoar71xx: Add support for Netgear WNR2000v1
Huan Truong [Mon, 12 Dec 2016 21:25:01 +0000 (15:25 -0600)]
ar71xx: Add support for Netgear WNR2000v1

This patch adds supports for the WNR2000v1 board with 4MB flash, and
produces device-specific factory, rootfs, and sysupgrade files for the
WNR2000v1. This board is errorneously claimed as supported on the OpenWRT
wiki as AP81, but AP81 image would not work because of APT81 image
requiring having 8MB of flash, while WNR2000v1 has only 4MB.

The image requires the u-boot bootloader to be modified to fuhry's
bootloader first.

Short specification:

- CPU: Atheros AR9132
- 4x 10/100 Mbps Ethernet, 1x WAN 10/100 Mbps
- 4 MB of Flash
- 32 MB of RAM
- UART header (J1) on board
- 1x button

Factory/Initial flash instructions:

- Set up a TFTP server on your local machine.
- Download the uImage for ar71xx-generic and the rootfs image for
  ar71xx-generic-wnr2000 and save in the tftp server root.
- Gain serial access to the router via the UART port (telnetenable over
  the network only won't work!).
- Upgrade the u-boot bootloader to fuhry's version by running the
  script: http://fuhry.com/b/wnr2000/install-repart.sh
- When the router restarts, interrupt u-boot and gain access to u-boot command line.
- Repartititon the board and flash initial uImage and rootfs as follow.

Commands to type in u-boot:

# tells u-boot that we have a tftp server on 192.168.1.10
setenv serverip 192.168.1.10

# tells u-boot that the router should take the address 192.168.1.1
setenv ipaddr 192.168.1.1

# erase the region from 0x050000-0x3f0000
erase 0xbf050000 +0x3A0000

# loads sqfs.bin on TFTP server, and put it to memory address 0x81000000
tftpboot 0x81000000 sqfs.bin
# it will tell you the length of sqfs.bin in hex, let's say ZZZZZZ
# copy bit by bit 0xZZZZZZ bytes from offset 0x050000
cp.b 0x81000000 0xbf050000 0xZZZZZZ

# same to the uImage.bin, write it right next to sqfs.bin
# again, 0xYYYYYY is the length that tftpboot reports
tftpboot 0x81000000 uImage.bin
cp.b 0x81000000 0xbf2a0000 0xYYYYYY

# We need to tell the kernel what board it is booting into, and where to find the partitions
setenv bootargs "board=WNR2000 console=ttyS0,115200 mtdparts=spi0.0:256k(u-boot)ro,64k(u-boot-env)ro,3712k(firmware),64k(art)ro rootfstype=squashfs,jffs2 noinitrd"

# Tell u-boot where to find the uImage
setenv bootcmd "bootm 0xbf2a0000"

# Tell u-boot to save parameters to the u-boot-env partitions
saveenv

# Reset the board
reset

Tested on:

- WNR2000v1 board.
- Initial flash works.

Known bugs:

- I don't know why factory image doesn't work on initial flash on stock
  firmware in u-boot recovery mode while it should.
- Sysupgrade does not yet work, if you do -f it will mess up your
  installation (requiring a reinstall of sqfs and uImage).

Signed-off-by: Huan Truong <htruong@tnhh.net>
7 years agoar71xx: Add ath10k-firmware-qca988x for DomyWifi DW33D
Jing Lin [Mon, 12 Dec 2016 07:10:22 +0000 (15:10 +0800)]
ar71xx: Add ath10k-firmware-qca988x for DomyWifi DW33D

The default package selection was missing the correct FW for ath10k to
work.

Signed-off-by: Jing Lin <mumuqz@163.com>
7 years agoprocd: add support for service signals
Jo-Philipp Wich [Tue, 13 Dec 2016 17:02:56 +0000 (18:02 +0100)]
procd: add support for service signals

Update procd to latest HEAD in order to introduce support for services signals:

- Adds a new service.signal ubus call to send a kill() signal to one or all
  running instances of a given service

- Adds a new "reload_signal" property which allows service init scripts to
  request procd to send a specific kill() signal on reload, instead of
  stopping and restarting running processes

Also fixes some potential memory leaks reported by cppcheck and an environment
variable corruption in the trace command.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agonetfilter: drop proprietary xt_id match
Jo-Philipp Wich [Tue, 13 Dec 2016 23:36:28 +0000 (00:36 +0100)]
netfilter: drop proprietary xt_id match

The xt_id match was used by the firewall3 package to track its own rules but
the approach has been changed to use xt_comment instead now, so we can drop
this nonstandard extension.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agofirewall3: drop support for automatic NOTRACK rules
Jo-Philipp Wich [Tue, 13 Dec 2016 23:44:22 +0000 (00:44 +0100)]
firewall3: drop support for automatic NOTRACK rules

Update to current HEAD in order to drop automatic generation of per-zone
NOTRACK rules.

The NOTRACK rules used to provide a little performance improvement but the
later introduction of the netfilter conntrack cache made those rules largely
unnecessary. Additionally, those rules caused various issues which broke
stateful firewalling in some scenarios.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agokernel: kmod-hwmon-tmp102: add dependency to kmod-thermal
Hauke Mehrtens [Tue, 13 Dec 2016 23:15:33 +0000 (00:15 +0100)]
kernel: kmod-hwmon-tmp102: add dependency to kmod-thermal

Depend on the kmod-thermal package when it is activated, this fixes
compile problems on some targets.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agoodhcpd: Fix dnsmasq re-reading hostfile
Hans Dedecker [Thu, 17 Nov 2016 12:34:34 +0000 (13:34 +0100)]
odhcpd: Fix dnsmasq re-reading hostfile

Depending on the dhcp uci config pidof dnsmasq can return
multiple pids. Fix re-reading of the hostfile by dnsmasq in
such case by sending SIGHUP signal to each of the returned
pids.

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agodnsmasq: Specify directory /tmp/hosts as argument for --addn-hosts
Hans Dedecker [Tue, 13 Dec 2016 13:18:31 +0000 (14:18 +0100)]
dnsmasq: Specify directory /tmp/hosts as argument for --addn-hosts

Let dnsmasq read all hosts files in /tmp/hosts directory by specifying
/tmp/hosts as argument of --addn-host

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoath10k: fix DMA allocation issues
Felix Fietkau [Tue, 13 Dec 2016 11:44:52 +0000 (12:44 +0100)]
ath10k: fix DMA allocation issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomac80211: refresh all patches
Felix Fietkau [Tue, 13 Dec 2016 11:33:42 +0000 (12:33 +0100)]
mac80211: refresh all patches

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoath9k: add a RCU related bugfix
Felix Fietkau [Tue, 13 Dec 2016 11:32:13 +0000 (12:32 +0100)]
ath9k: add a RCU related bugfix

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodownload.mk: improve download tarball reproducibility
Felix Fietkau [Mon, 12 Dec 2016 12:12:43 +0000 (13:12 +0100)]
download.mk: improve download tarball reproducibility

Store only numeric user/group id. Group 0 is 'root' on linux and 'wheel' on
BSD and Mac OS X.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodownload.mk: remove code duplication in $(TAR) call
Felix Fietkau [Mon, 12 Dec 2016 10:42:14 +0000 (11:42 +0100)]
download.mk: remove code duplication in $(TAR) call

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agodownload.mk: use $(error) instead of a regular shell error
Felix Fietkau [Mon, 12 Dec 2016 10:37:23 +0000 (11:37 +0100)]
download.mk: use $(error) instead of a regular shell error

Useful for further rework

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agokernel: backport ubifs support for dirty_writeback_interval
Rafał Miłecki [Tue, 13 Dec 2016 08:39:33 +0000 (09:39 +0100)]
kernel: backport ubifs support for dirty_writeback_interval

These patches were queued for 4.10. For possible use cases see added:
[PATCH] ubifs: Use dirty_writeback_interval value for wbuf timer

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agobcm53xx: backport missed BCM53573 ILP patch from 4.10
Rafał Miłecki [Tue, 13 Dec 2016 08:13:40 +0000 (09:13 +0100)]
bcm53xx: backport missed BCM53573 ILP patch from 4.10

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agokernel: bump to 4.4.38
Kevin Darbyshire-Bryant [Sun, 11 Dec 2016 08:41:34 +0000 (08:41 +0000)]
kernel: bump to 4.4.38

Bump & refresh patches for all 4.4 supported targets.

Compile & run tested: ar71xx - Archer C7 v2

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agobcm53xx: backport DTS patches accepted for 4.11
Rafał Miłecki [Tue, 13 Dec 2016 06:37:40 +0000 (07:37 +0100)]
bcm53xx: backport DTS patches accepted for 4.11

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agoiproute2: support latest cake & restore DSCP washing
Kevin Darbyshire-Bryant [Wed, 12 Oct 2016 09:40:05 +0000 (10:40 +0100)]
iproute2: support latest cake & restore DSCP washing

Support new packet overhead passing paradigm in cake qdisc, also restore
DSCP wash/nowash keywords.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agokmod-sched-cake: update & restore DSCP washing
Kevin Darbyshire-Bryant [Fri, 28 Oct 2016 14:43:53 +0000 (15:43 +0100)]
kmod-sched-cake: update & restore DSCP washing

Track upstream changes, incl changes in packet overhead accounting
(automatically taking care of linux' packet sizing knowledge),
improvements to triple isolated DRR handling (new flow dominance),
statistics tweak & allow more packet drops in stressed conditions.

Under tests this has significantly improved latency control under
'many flows to one' scenarious as is typical of bittorrent and MS
Windows update.

I also restored 'DSCP washing' functionality in my repo which follows
upstream closely (like a hawk!) with tc keywords 'wash/nowash'.  This
allows cake to limit/control packets in bands determined by a packet's
DSCP but to clear those DSCP bits on qdisc egress.  This functionality
was originally removed as part of an attempt to push cake into the
kernel, which hasn't actually happened as yet.

A matching commit is required to iproute2/tc to support the new overhead
handling, keyword changes as well as the 'wash/nowash' tweak.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
7 years agofirmware-utils: tplink-safeloader: update support lists for CPE210/510/...
Matthias Schiffer [Sun, 11 Dec 2016 17:00:13 +0000 (18:00 +0100)]
firmware-utils: tplink-safeloader: update support lists for CPE210/510/...

Adds support for new EU and US variants; removes a few strings that were
never actually used and have been removed from the stock firmwares.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agoar71xx: simplify model detection for TP-Link Pharos devices
Matthias Schiffer [Sun, 11 Dec 2016 16:54:49 +0000 (17:54 +0100)]
ar71xx: simplify model detection for TP-Link Pharos devices

This also makes the detection more flexible, as it doesn't need to check
for each model explicitly.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agomt76: update to the latest version, fixes dfs issues
Felix Fietkau [Mon, 12 Dec 2016 09:54:47 +0000 (10:54 +0100)]
mt76: update to the latest version, fixes dfs issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agotcpdump: reduce size of -mini by removing more infrequently used protocols
Felix Fietkau [Sat, 10 Dec 2016 13:02:15 +0000 (14:02 +0100)]
tcpdump: reduce size of -mini by removing more infrequently used protocols

This removes:
- BGP
- CDP
- SCTP

MIPS binary .ipk size is reduced from ~150k to ~130k

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agonet/utils/tcpdump: update to 4.8.1
p-wassi [Sat, 10 Dec 2016 10:10:03 +0000 (11:10 +0100)]
net/utils/tcpdump: update to 4.8.1

Update tcpdump to upstream release 4.8.1

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
7 years agombedtls: tune config to reduce size and improve performance
Felix Fietkau [Sat, 10 Dec 2016 11:27:23 +0000 (12:27 +0100)]
mbedtls: tune config to reduce size and improve performance

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agombedtls: sync with polarssl config
Felix Fietkau [Sat, 10 Dec 2016 10:22:58 +0000 (11:22 +0100)]
mbedtls: sync with polarssl config

One of those changes is re-enabling blowfish support to make
openvpn-mbedtls compatible with common configurations

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoopenvpn: quote parameters to --push in openvpn config file
Magnus Kroken [Sat, 10 Dec 2016 11:02:03 +0000 (12:02 +0100)]
openvpn: quote parameters to --push in openvpn config file

OpenVPN requires arguments to --push to be enclosed in double quotes.
One set of quotes is stripped when the UCI config is parsed.
Change append_params() of openvpn.init to enclose push parameters in
double quotes.

Unquoted push parameters do not cause errors in OpenVPN 2.3,
but OpenVPN 2.4 fails to start with unquoted push parameters.

Fixes: FS#290.
Signed-off-by: Magnus Kroken <mkroken@gmail.com>
7 years agombedtls: enable MBEDTLS_DHM_C
Magnus Kroken [Fri, 9 Dec 2016 20:07:37 +0000 (21:07 +0100)]
mbedtls: enable MBEDTLS_DHM_C

This option is required by OpenVPN, and OpenVPN 2.4 uses mbedTLS 2.x.
DHM_C is also already enabled in the PolarSSL 1.3.x config.h.

Signed-off-by: Magnus Kroken <mkroken@gmail.com>
7 years agoar71xx: remove obsolete flash chip locking code
Felix Fietkau [Fri, 9 Dec 2016 20:07:33 +0000 (21:07 +0100)]
ar71xx: remove obsolete flash chip locking code

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: remove PB92 reference design board support
Felix Fietkau [Fri, 9 Dec 2016 13:29:21 +0000 (14:29 +0100)]
ar71xx: remove PB92 reference design board support

Due to flash size limitations, support for this board has not been
functional for years

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: remove AP113 reference design board support
Felix Fietkau [Fri, 9 Dec 2016 13:26:12 +0000 (14:26 +0100)]
ar71xx: remove AP113 reference design board support

Due to flash size limitations, support for this board has not been
functional for years

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: remove AP81 reference design board support
Felix Fietkau [Fri, 9 Dec 2016 13:23:39 +0000 (14:23 +0100)]
ar71xx: remove AP81 reference design board support

This board is very old and unlikely to still be relevant today.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: remove obsolete duplicate driver source file
Felix Fietkau [Fri, 9 Dec 2016 09:07:14 +0000 (10:07 +0100)]
ar71xx: remove obsolete duplicate driver source file

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: remove AP83 reference design board support
Felix Fietkau [Thu, 8 Dec 2016 20:46:15 +0000 (21:46 +0100)]
ar71xx: remove AP83 reference design board support

This board is very old and unlikely to still be relevant today. Support
for it contains a significant amount of device specific baggage which is
worth getting rid of.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoar71xx: clean up spi controller related patches
Felix Fietkau [Thu, 8 Dec 2016 20:10:50 +0000 (21:10 +0100)]
ar71xx: clean up spi controller related patches

Remove various hacks for fast read, un-break device tree support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agobcm53xx: disable CONFIG_FW_LOADER_USER_HELPER_FALLBACK
Rafał Miłecki [Sun, 11 Dec 2016 12:05:08 +0000 (13:05 +0100)]
bcm53xx: disable CONFIG_FW_LOADER_USER_HELPER_FALLBACK

There are many targets using user space scripts to generate firmware but
bcm53xx doesn't need this, so let's disable that kernel option.

This lets us avoid some scary-looking kernel warnings like:
brcmfmac 0001:04:00.0: Falling back to user helper
firmware brcm!brcmfmac43602-pcie.txt: firmware_loading_store: map pages failed

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>