openwrt/staging/jogo.git
7 years agoramips: fix polarity in gpio-export node
Mathias Kresin [Sat, 16 Dec 2017 23:02:06 +0000 (00:02 +0100)]
ramips: fix polarity in gpio-export node

If we need to set the initial output value to GPIOF_OUT_INIT_HIGH (1) to
enable something, the pin is ACTIVE_HIGH. The same applies to
GPIOF_OUT_INIT_LOW (0) and ACTIVE_LOW.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoar71xx: fix lan ports on ens202ext
Marty E. Plummer [Thu, 23 Nov 2017 20:36:28 +0000 (14:36 -0600)]
ar71xx: fix lan ports on ens202ext

In the stock firmware both eth0 and eth1 are set to br-lan,
add this behavior to our images.

Fixes: FS#1084
Signed-off-by: Marty E. Plummer <hanetzer@protonmail.com>
7 years agobrcm63xx: fix Sercomm AD1018 status led
Jonas Gorski [Sun, 17 Dec 2017 10:52:52 +0000 (11:52 +0100)]
brcm63xx: fix Sercomm AD1018 status led

The status led part was missed when changing the board name to *-nor.

Fixes: e12c72bb52eb ("brcm63xx: Add Sercomm AD1018 support")
Reported-by: Daniel Gonzalez Cabanelas <dgcbueu@gmail.com>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agomt76: update to the latest version, reverts MAC address changes causing performance...
Felix Fietkau [Sun, 17 Dec 2017 11:41:58 +0000 (12:41 +0100)]
mt76: update to the latest version, reverts MAC address changes causing performance regressions

cf4a5ce mt76: fix memcpy to potential null pointer on failed allocation
ca5ca8c mt76x2: fix possible NULL pointer dereferencing in mt76x2_ampdu_action()
2d4b8f5 mt76x2: mac: fix possible NULL pointer dereferencing in mt76x2_mac_write_txwi()
4ca8bba mt76x2: eeprom: fix typo in mt76x2_get_power_info_5g()
00f7e60 mt7603: fix configuration for HT40
31e9bdb mt7603: main: fix possible NULL pointer dereferencing in mt7603_ampdu_action()
bf90dd2 mt7603: mcu: fix potential NULL pointer dereferencing on failed allocation
860b185 Revert "mt76x2: init: disable APCLI by default"
bcd576e Revert "mt76x2: drop wiphy->addresses"
bdc6f2f Revert "mt76x2: clean up MAC/BSSID address initialization"
d242650 Revert "mt76x2: remove MAC address limitation for multi-vif setups"
68b0cf1 mt76x2: add functions for setting extended MAC address registers

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agolantiq: sync image filename with boardname
Mathias Kresin [Fri, 8 Dec 2017 07:42:11 +0000 (08:42 +0100)]
lantiq: sync image filename with boardname

Use <manufacturer>_<modelname> as image name.

Use the BOARD_NAME variable to ensure that the former used boardname is
still used as the subdirectory name for the sysupgrade-tar image, to
not break sysupgrade from earlier versions.

While at it, normalise the image filenames by using only lower case
characters and bin as file extension for sysupgrade images.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: drop unused variable from image build code
Mathias Kresin [Fri, 8 Dec 2017 07:07:50 +0000 (08:07 +0100)]
lantiq: drop unused variable from image build code

Remove the DEVICE_PROFILE variable. The variable isn't
used at all.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: use the generic board detect
Mathias Kresin [Sat, 8 Apr 2017 09:59:24 +0000 (11:59 +0200)]
lantiq: use the generic board detect

Drop the target specific detection function in favour of the generic
one provided by base-files.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: add missing model names
Mathias Kresin [Sat, 8 Apr 2017 09:47:44 +0000 (11:47 +0200)]
lantiq: add missing model names

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: remove the former board name from device tree model
Mathias Kresin [Sat, 8 Apr 2017 09:40:07 +0000 (11:40 +0200)]
lantiq: remove the former board name from device tree model

Remove the former used board name from the device model property and
use the model name as it is.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: use the compatible string as board name
Mathias Kresin [Sat, 8 Apr 2017 09:16:21 +0000 (11:16 +0200)]
lantiq: use the compatible string as board name

Use the first compatible string as board name in userspace. Add the new
board name as well as the former used board name to the image metadata
to keep compatibilty with already deployed installations.

Don't add the former used boardname for boards which exists only in
master or evaluation boards.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: drop SoC name from eval boards compatible string
Mathias Kresin [Sat, 8 Apr 2017 08:59:32 +0000 (10:59 +0200)]
lantiq: drop SoC name from eval boards compatible string

Keep it in the same format as used for the other eval boards.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agolantiq: add compatible strings to dts files
Mathias Kresin [Sat, 8 Apr 2017 08:33:53 +0000 (10:33 +0200)]
lantiq: add compatible strings to dts files

The compatible string is mandetory for devicetree source file.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agox86: Add support for kernel 4.14
Hauke Mehrtens [Sun, 29 Oct 2017 18:34:13 +0000 (19:34 +0100)]
x86: Add support for kernel 4.14

This adds basic support for kernel 4.14, this was tested in qemu only.
The subtarget configuration was refresh with kernel 4.14 and the
options needed to make it compile on kernel 4.9 were added manually.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: include: remove last .0 from kernel versions again
Hauke Mehrtens [Fri, 17 Nov 2017 21:55:03 +0000 (22:55 +0100)]
kernel: include: remove last .0 from kernel versions again

Kernel 4.14 has the version number 4.14 and not 4.14.0. This was
different in some older Linux kernel versions, This change makes it
possible to use kernel 4.14 without any minor version.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add missing dependencies to kmod-crypto-acompress
Hauke Mehrtens [Sat, 25 Nov 2017 23:24:28 +0000 (00:24 +0100)]
kernel: add missing dependencies to kmod-crypto-acompress

kmod-lib-lzo and kmod-lib-lz4 depend in kernel 4.14 on
kmod-crypto-acompress, add this missing dependency.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-rtc-ds1307: add dependency to regmap
Hauke Mehrtens [Sat, 25 Nov 2017 21:05:21 +0000 (22:05 +0100)]
kernel: kmod-rtc-ds1307: add dependency to regmap

In kernel 4.14 kmod-rtc-ds1307 depends on regmap-i2c, add this missing
dependency.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-iio-bmp280: allow compile with kernel 4.14
Hauke Mehrtens [Mon, 13 Nov 2017 21:19:43 +0000 (22:19 +0100)]
kernel: kmod-iio-bmp280: allow compile with kernel 4.14

This package also compiles with kernel 4.14.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-zram: deactivate CONFIG_ZRAM_WRITEBACK for kernel 4.14
Hauke Mehrtens [Sun, 12 Nov 2017 21:22:07 +0000 (22:22 +0100)]
kernel: kmod-zram: deactivate CONFIG_ZRAM_WRITEBACK for kernel 4.14

This new option was introduced in kernel 4.14 and should be deactivated
by default.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agoxtables-addons: fix compile with kernel 4.14
Hauke Mehrtens [Sun, 12 Nov 2017 21:20:22 +0000 (22:20 +0100)]
xtables-addons: fix compile with kernel 4.14

This fixes a compile problems seen with kernel 4.14.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-fbcon: deactivate for kernel 4.14
Hauke Mehrtens [Wed, 29 Nov 2017 21:51:55 +0000 (22:51 +0100)]
kernel: kmod-fbcon: deactivate for kernel 4.14

CONFIG_FRAMEBUFFER_CONSOLE does not activate new modules any more in
kernel 4.14, but CONFIG_FRAMEBUFFER_CONSOLE is now a boolean option
which change the kmod-fb package. kmod-fbcon should be split up.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-sound-core: handle moved snd-seq-device.ko
Hauke Mehrtens [Sun, 12 Nov 2017 21:17:30 +0000 (22:17 +0100)]
kernel: kmod-sound-core: handle moved snd-seq-device.ko

The snd-seq-device.ko kernel module was moved with kernel 4.13.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-mmc: handle moved mmc_block.ko
Hauke Mehrtens [Sun, 12 Nov 2017 21:16:46 +0000 (22:16 +0100)]
kernel: kmod-mmc: handle moved mmc_block.ko

mmc_block.ko was moved with kernel 4.10.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: kmod-rxrpc: handle renamed kernel module
Hauke Mehrtens [Sun, 12 Nov 2017 21:15:14 +0000 (22:15 +0100)]
kernel: kmod-rxrpc: handle renamed kernel module

The kernel module was renamed with kernel 4.11, handle that in the
package definition.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: deactivate some new network features
Hauke Mehrtens [Sun, 12 Nov 2017 21:13:23 +0000 (22:13 +0100)]
kernel: deactivate some new network features

This deactivates the following options which were introduced between
kernel 4.9 and 4.14 in some kernel packages:
CONFIG_INET_ESP_OFFLOAD
CONFIG_INET6_ESP_OFFLOAD
CONFIG_LWTUNNEL_BPF
CONFIG_NET_9P_XEN

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: unset CONFIG_CIFS_SMB311 in kmod-fs-cifs
Hauke Mehrtens [Sun, 12 Nov 2017 21:08:28 +0000 (22:08 +0100)]
kernel: unset CONFIG_CIFS_SMB311 in kmod-fs-cifs

Do not activate the experimental feature CONFIG_CIFS_SMB311.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add kmod-crypto-ecdh
Hauke Mehrtens [Sun, 12 Nov 2017 21:06:38 +0000 (22:06 +0100)]
kernel: add kmod-crypto-ecdh

In kernel 4.14 kmod-bluetooth depends on kmod-crypto-ecdh, add
kmod-crypto-ecdh to LEDE.
Both packages also depend on the kmod-crypto-kpp package. To build this
we have to fix the dependency of CRYPTO_ECDH which has a typo.
This patch is already accepted upstream.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add kmod-crypto-rsa
Hauke Mehrtens [Sun, 12 Nov 2017 21:04:17 +0000 (22:04 +0100)]
kernel: add kmod-crypto-rsa

In kernel 4.14 kmod-crypto-hw-ccp depends on kmod-crypto-rsa, add it.
kmod-crypto-rsa also packages the ASN1 parser and some other code which
is currently only used by this module.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add kmod-crypto-acompress
Hauke Mehrtens [Sun, 12 Nov 2017 21:02:27 +0000 (22:02 +0100)]
kernel: add kmod-crypto-acompress

In kernel 4.14 kmod-crypto-deflate depends on kmod-crypto-acompress.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add kmod-lib-zstd
Hauke Mehrtens [Sun, 12 Nov 2017 21:00:20 +0000 (22:00 +0100)]
kernel: add kmod-lib-zstd

In kernel 4.14 kmod-fs-btrfs depends on the zstd compression libraries,
add it to the package system.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: add kmod-dax
Hauke Mehrtens [Sun, 12 Nov 2017 20:58:19 +0000 (21:58 +0100)]
kernel: add kmod-dax

In kernel 4.14 kmod-dm depends on kmod-dax.
Add DAX: "Direct access to differentiated memory" to LEDE.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: tg3: deactivate hwmon for kernel 4.14
Hauke Mehrtens [Sun, 12 Nov 2017 14:34:05 +0000 (15:34 +0100)]
kernel: tg3: deactivate hwmon for kernel 4.14

In kernel 4.14 hwmon support can be deactivated for the tg3 driver,
deactivate it by default to save some space.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: generic: Add kernel 4.14 support
Hauke Mehrtens [Sun, 29 Oct 2017 18:32:10 +0000 (19:32 +0100)]
kernel: generic: Add kernel 4.14 support

This adds initial support for kernel 4.14 based on the patches for
kernel 4.9.

In the configuration I deactivated some of the new possible security
features like:
CONFIG_REFCOUNT_FULL
CONFIG_SLAB_FREELIST_HARDENED
CONFIG_SOFTLOCKUP_DETECTOR
CONFIG_WARN_ALL_UNSEEDED_RANDOM

And these overlay FS options are also deactivated:
CONFIG_OVERLAY_FS_INDEX
CONFIG_OVERLAY_FS_REDIRECT_DIR

I activated this:
CONFIG_FORTIFY_SOURCE
CONFIG_POSIX_TIMERS
CONFIG_SLAB_MERGE_DEFAULT
CONFIG_WATCHDOG_HANDLE_BOOT_ENABLED

I am not sure if I did the porting correct for the following patches:
target/linux/generic/backport-4.14/020-backport_netfilter_rtcache.patch
target/linux/generic/hack-4.14/220-gc_sections.patch
target/linux/generic/hack-4.14/321-powerpc_crtsavres_prereq.patch
target/linux/generic/pending-4.14/305-mips_module_reloc.patch
target/linux/generic/pending-4.14/611-netfilter_match_bypass_default_table.patch
target/linux/generic/pending-4.14/680-NET-skip-GRO-for-foreign-MAC-addresses.patch

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agox86: refresh configuration
Hauke Mehrtens [Sun, 26 Nov 2017 10:24:11 +0000 (11:24 +0100)]
x86: refresh configuration

Just refresh the kernel configuration, some options are removed because
they are now in the generic kernel configuration.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agokernel: generic: add some more 4.9 configure options
Hauke Mehrtens [Sun, 26 Nov 2017 00:28:37 +0000 (01:28 +0100)]
kernel: generic: add some more 4.9 configure options

These are taken from the x86 target and should make support kernel 4.9
and 4.14 in the x86 target easier.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agoxtables-addons: update to version 2.14
Hauke Mehrtens [Sun, 26 Nov 2017 22:54:03 +0000 (23:54 +0100)]
xtables-addons: update to version 2.14

This includes a compile fix needed for kernel 4.14.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 years agonetdevices.mk: add Sun Neptune 10Gbit Ethernet
Lucian Cristian [Thu, 2 Nov 2017 15:44:02 +0000 (17:44 +0200)]
netdevices.mk: add Sun Neptune 10Gbit Ethernet

Signed-off-by: Lucian Cristian <lucian.cristian@gmail.com>
7 years agokernel: tune e1000e module parameters
Jo-Philipp Wich [Wed, 23 Aug 2017 22:17:56 +0000 (00:17 +0200)]
kernel: tune e1000e module parameters

The default e1000e parameters (interrupt throttling rate, MSI/MSI-X
mode) are optimized for desktop and server computers to optimize
user-space execution (i.e. what's typically referred to as "useful"
work).  This assumption breaks on a router under load where most of
the "useful" work actually takes place either in hardware interrupt
handlers (IRQ) or at software IRQ (swirq) modes, so we try to reflect
that by overriding these parameters with more appropriate values.

Patch-by: Philip Prindeville <philipp@redfish-solutions.com>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agox86: add on-chip thermal sensors and cpuid support
Philip Prindeville [Fri, 3 Nov 2017 23:22:29 +0000 (17:22 -0600)]
x86: add on-chip thermal sensors and cpuid support

Cover temperature sensors for all mainstream 64-bit processors, including
AMD 10h and 15h families, Intel iCore, Xeon, Atom, and Via Nano.  Also
add CPUID support for user-space applications to detect CPU type.

Include the on-chip sensors for 64-bit CPU's in the generic profile
in case someone builds a 32-bit kernel to run on a Xeon SoC, etc.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
7 years agox86: refresh settings for 4.9.59
Philip Prindeville [Fri, 3 Nov 2017 23:01:55 +0000 (17:01 -0600)]
x86: refresh settings for 4.9.59

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
7 years agotools/expat: Update to 2.2.5
Daniel Engberg [Mon, 4 Dec 2017 18:23:15 +0000 (19:23 +0100)]
tools/expat: Update to 2.2.5

Update (lib)expat to 2.2.5

Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
7 years agobrcm63xx: backport mdio-bus reset gpio support
Jonas Gorski [Wed, 6 Dec 2017 13:12:06 +0000 (14:12 +0100)]
brcm63xx: backport mdio-bus reset gpio support

Backport the mdio-bus reset gpio support from 4.12 and use it instead
of toggling the reset ourself.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobrcm63xx: backport enet cleanup patches
Jonas Gorski [Mon, 2 Oct 2017 09:41:43 +0000 (11:41 +0200)]
brcm63xx: backport enet cleanup patches

Align with upstream version, mostly non-fixes and small clean ups.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobrcm63xx: register serial through device tree
Jonas Gorski [Mon, 31 Jul 2017 19:03:37 +0000 (21:03 +0200)]
brcm63xx: register serial through device tree

Register serial consoles through device tree instead of through board
data.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobrcm63xx: add uart nodes
Jonas Gorski [Mon, 31 Jul 2017 09:08:20 +0000 (11:08 +0200)]
brcm63xx: add uart nodes

Now that we can register uarts through device-tree, add them to the dtsi
files.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobrcm63xx: add clkdev lookup support
Jonas Gorski [Mon, 31 Jul 2017 08:52:48 +0000 (10:52 +0200)]
brcm63xx: add clkdev lookup support

Add clockdev lookup support for easier providing of clocks for
devices.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobrcm63xx: Add Sercomm AD1018 support
Daniel Gonzalez Cabanelas [Thu, 31 Aug 2017 20:22:37 +0000 (22:22 +0200)]
brcm63xx: Add Sercomm AD1018 support

Add support for the Sercomm AD1018 router

This a BCM6328 based board, 128 MB RAM, 128 MiB NAND flash,
with an onboard BCM43217 wifi, 4 ethernet ports and 1 USB
host port (not soldered). The board also has an FXS chip (Si32177)
connected via SPI (SS2#), without support in LEDE.

Since NAND flash chips aren't still supported in brcm63xx, the
support is for now added to work only with SPI flash chips. Therefore
hardware modding, soldering a new SPI flash chip, is required
to make the board work with LEDE (tested and working OK).
The flash at dts is intentionally left without partitioning to let
the user choose a NOR chip of any size (8, 16 or 32 MB).

Signed-off-by: Daniel Gonzalez Cabanelas <dgcbueu@gmail.com>
[jonas.gorski: renamed ad1018 to ad1018-nor to signify the modification]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agoat91: create sdcard image for sama5
Sandeep Sheriker Mallikarjun [Tue, 12 Dec 2017 22:08:26 +0000 (15:08 -0700)]
at91: create sdcard image for sama5

create sdcard image using gen_at91_sdcard_img.sh for sama5 platform
and sdcard image partition layout is:
  P0: Boot (fat32) - contains(at91bootstrap,u-boot,zImage & dtb)
  p1: Rootfs (ext4)

Signed-off-by: Sandeep Sheriker Mallikarjun <sandeepsheriker.mallikarjun@microchip.com>
7 years agoar71xx: fix invalid pointer dereference in rb95x_nand_scan_fixup()
Gabor Juhos [Tue, 12 Dec 2017 20:49:50 +0000 (21:49 +0100)]
ar71xx: fix invalid pointer dereference in rb95x_nand_scan_fixup()

Since Linux 4.6, mtd->priv no longer points to the NAND specific
structure. Under 4.9 it contains NULL, thus using it to access
chip->options causes an invalid pointer dereference (FS#1200).

Update the code to use the mtd_to_nand() helper under 4.9 to obtain
the address of the chip specific data.

Fixes: 7bbf4117c6fe ("ar71xx: Add kernel 4.9 support")
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
7 years agoar71xx: fix invalid pointer dereference in c60_nand_scan_fixup()
Gabor Juhos [Wed, 13 Dec 2017 19:31:50 +0000 (20:31 +0100)]
ar71xx: fix invalid pointer dereference in c60_nand_scan_fixup()

Since Linux 4.6, mtd->priv no longer points to the NAND specific
structure. Under 4.9 it contains NULL, thus using it to access
the fields of the nand_chip structure causes an invalid pointer
dereference.

Update the code to use the mtd_to_nand() helper under 4.9 to obtain
the address of the chip specific data.

Compile tested only.

Fixes: 7bbf4117c6fe ("ar71xx: Add kernel 4.9 support")
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
Tested-by: Christian Lamparter <chunkeey@gmail.com>
7 years agokernel: fix compiler warning in mtdsplit_minor.c under 4.4
Gabor Juhos [Wed, 13 Dec 2017 20:56:03 +0000 (21:56 +0100)]
kernel: fix compiler warning in mtdsplit_minor.c under 4.4

When mtdsplit_minor.c is compiled under Linux 4.4, the compiler
drops the following warning:

    CC      drivers/mtd/mtdsplit/mtdsplit_minor.o
  drivers/mtd/mtdsplit/mtdsplit_minor.c:106:14: warning: initialization from incompatible pointer type [-Wincompatible-pointer-types]
    .parse_fn = mtdsplit_parse_minor,
                ^
  drivers/mtd/mtdsplit/mtdsplit_minor.c:106:14: note: (near initialization for 'mtdsplit_minor_parser.parse_fn')

The second parameter of the parser function must not have a 'const'
qualifier in 4.4. The 001-mtdsplit_backport.patch removes the qualifier
from other partition parsers. Update it to handle mtdsplit_minor.c as
well.

Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
7 years agoiproute2: cake: support new operating modes
Kevin Darbyshire-Bryant [Fri, 15 Dec 2017 16:19:22 +0000 (16:19 +0000)]
iproute2: cake: support new operating modes

There has been recent significant activity with the cake qdisc of late
Some of that effort is related to upstreaming to kernel & iproute2
mainline but we're not quite there yet.  This commit teaches tc how to
activate and interprete the latest cake operating modes, namely:

ingress mode: Instead of only counting packets that make it past the
shaper, include packets we've decided to drop as well, since they did
arrive with us on the link and took link capacity.
This mode is more suitable for shaping the ingress of a link
(e.g. from ISP) rather than the more normal egress.

ack-filter/ack-filter-aggressive: Filter excessive TCP ACKS.  Useful in
highly assymetric links (downstream v upstream capacity) where the
majority of upstream link capacity is occupied with ACKS for downstream
traffic.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
7 years agokmod-sched-cake: bump to latest bake of cake
Kevin Darbyshire-Bryant [Fri, 15 Dec 2017 16:19:21 +0000 (16:19 +0000)]
kmod-sched-cake: bump to latest bake of cake

There has been recent significant activity with the cake qdisc of late
but in the cobalt branch.  Some of that effort is related to upstreaming
to kernel & iproute2 mainline but we're not quite there yet.  Relevant
feature changes:

ingress mode: Instead of only counting packets that make it past the
shaper, include packets we've decided to drop as well, since they did
arrive with us on the link and took link capacity.
This mode is more suitable for shaping the ingress of a link
(e.g. from ISP) rather than the more normal egress.

ptm mode: Minor optimisation in packet overhead calculation.

dual-src/dsthost/triple-isolate: Optimise only calculating src or dst
host hashes only if required.

ack-filter/ack-filter-aggressive: Filter excessive TCP ACKS.  Useful in
highly assymetric links (downstream v upstream capacity) where the
majority of upstream link capacity is occupied with ACKS for downstream
traffic.

A separate iproute2 patch to teach it about Cake's new features will
follow.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
7 years agoar71xx: fix LED name typo for GL-AR300M
kyson Lok [Fri, 15 Dec 2017 08:21:27 +0000 (16:21 +0800)]
ar71xx: fix LED name typo for GL-AR300M

It shouldn't use double colon characters.

Signed-off-by: kyson Lok <kysonlok@gmail.com>
7 years agonetifd: update to latest git HEAD
Hans Dedecker [Fri, 15 Dec 2017 16:40:00 +0000 (17:40 +0100)]
netifd: update to latest git HEAD

4268193 interface-ip: harden eui64 IPv6 prefix address generation
81ff6d1 interface-ip: fix race condition in IPv6 prefix address generation
d3a5df0 handler: replace is_error() helper with NULL check

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agouhttpd: fix PKG_CONFIG_DEPENDS (FS#1189)
Hans Dedecker [Fri, 15 Dec 2017 16:13:16 +0000 (17:13 +0100)]
uhttpd: fix PKG_CONFIG_DEPENDS (FS#1189)

Remove PACKAGE_uhttpd_debug config as this is an unused leftover
Add CONFIG_uhttpd_lua to PKG_CONFIG_DEPENDS

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoRevert "ar71xx: TL-WR1043N v4/v5: add orange WAN LED support"
John Crispin [Thu, 14 Dec 2017 09:58:15 +0000 (10:58 +0100)]
Revert "ar71xx: TL-WR1043N v4/v5: add orange WAN LED support"

This reverts commit 256990cbc0157feb8a9f37efeebf0330ab73727f.

this commit caused a compile error

"TL_WR1043_V5_GPIO_LED_WANORANGE" is undefined.

Signed-off-by: John Crispin <john@phrozen.org>
7 years agoopkg: fix PKG_CONFIG_DEPENDS to include version.mk entries
Rafał Miłecki [Thu, 14 Dec 2017 08:39:18 +0000 (09:39 +0100)]
opkg: fix PKG_CONFIG_DEPENDS to include version.mk entries

Including version.mk sets PKG_CONFIG_DEPENDS to config entries used for
VERSION_SED command. We should keep these configs to make sure package
gets refreshed when needed.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 years agobase-files: allow skipping of hash verification
Jo-Philipp Wich [Tue, 12 Dec 2017 13:55:11 +0000 (14:55 +0100)]
base-files: allow skipping of hash verification

When calling a download target, hash verification is now completely
skipped if we set PKG_HASH=skip.

This allows to easily bump package version:

$ make package/<mypackage>/download PKG_HASH=skip V=s
$ make package/<mypackage>/check FIXUP=1 V=s

This will download the new version of the package, and then automatically
update PKG_HASH with the hash of the new version.  Of course, it is still
the responsibility of the packager to ensure that the new tarball is
legitimate, because it is downloaded from a possibly untrusted source.

Fixes: b30ba14e ("scripts/download.pl: fail loudly if provided hash is unsupported")
Signed-off-by: Baptiste Jonglez <git@bitsofnetworks.org>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Acked-by: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: John Crispin <john@phrozen.org>
7 years agoar71xx: TL-WR1043N v4/v5: add orange WAN LED support
Tim Thorpe [Thu, 30 Nov 2017 08:32:04 +0000 (02:32 -0600)]
ar71xx: TL-WR1043N v4/v5: add orange WAN LED support

Add missing definitions for the orange WAN LED on the TL-WR1043N(D) v4 and
v5. Minor format correction on a constant for consistency.

Signed-off-by: Tim Thorpe <tim@tfthorpe.net>
7 years agoar71xx: generate BR region-code factory image for TP-Link TL-WR940N
Henryk Heisig [Thu, 7 Dec 2017 20:30:24 +0000 (21:30 +0100)]
ar71xx: generate BR region-code factory image for TP-Link TL-WR940N

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
7 years agofirmware-utils: mktplinkfw: add support for BR region code
Henryk Heisig [Thu, 7 Dec 2017 20:28:28 +0000 (21:28 +0100)]
firmware-utils: mktplinkfw: add support for BR region code

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
7 years agobusybox: enable find -newer needed for shorewall firewall, no size increase on binary
Lucian Cristian [Sat, 9 Dec 2017 17:11:28 +0000 (19:11 +0200)]
busybox: enable find -newer needed for shorewall firewall, no size increase on binary

Signed-off-by: Lucian Cristian <lucian.cristian@gmail.com>
7 years agofirmware-utils: mktplinkfw: fix JFFS2 EOF markers
Sergey Ryazanov [Tue, 10 Oct 2017 22:07:08 +0000 (01:07 +0300)]
firmware-utils: mktplinkfw: fix JFFS2 EOF markers

mktplinkfw/mktplinkfw2 utilities put JFFS2 EOF market only at 64KB
boundary, this could lead to current device configuration lost during
the sysupgrade on a device, which is equpped with flash with the 4KB
erase block size (e.g. TP-Link Archer C20).

This happens when 64KB and 4KB alignments do not match, so the JFFS2
data is written not exactly at the partition beginnig and startup
scripts can not find the JFFS2 during the first boot just after the
sysupgrade.

Fix this by placing additional JFFS2 EOF marker at a 4KB boundary. Also
keep the marker at 64KB intact, so the utilities will produce images
suitable for devices with both 4KB and 64KB erase blocks.

Fixes: 29a2c2ea80441895a2ffe100d854d2b26d5fa606 (add ability to put
jffs2 eof marker into the image)

Signed-off-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
7 years agoar71xx: Fix switch port numbering on MIkrotik RB951Ui-2nD and RB493G
João Chaínho [Sat, 14 Oct 2017 19:34:05 +0000 (20:34 +0100)]
ar71xx: Fix switch port numbering on MIkrotik RB951Ui-2nD and RB493G

This patch fixes the switch port numbering on Mikrotik RB951Ui-2nD (hAP).
Also fixes the switch port numbering shown on LuCI for Mikrotik RB493G.

Signed-off-by: João Chaínho <joaochainho@gmail.com>
7 years agobase-files: fix sysupgrade -b/-l when -c is used
Luiz Angelo Daros de Luca [Tue, 21 Nov 2017 03:45:05 +0000 (01:45 -0200)]
base-files: fix sysupgrade -b/-l when -c is used

Since /overlay/upper appeared, -b ignored -c silently (cause it was
still checking for /overlay/etc). Now, if /overlay/upper is absent,
sysupgrade -c will fail and exit verbosely.

Fix -l to consider -c (it never did).

Clean up to always use /overlay/upper/xxx instead of still checking
for /overlay/xxx.

Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
7 years agoapm821xx: use x86's upgrade scripts for MyBook Live
Christian Lamparter [Fri, 1 Dec 2017 21:41:11 +0000 (22:41 +0100)]
apm821xx: use x86's upgrade scripts for MyBook Live

Advantages:
 - preserves existing partition layout. On the hard-drive.
   Only the boot and rootfs partition will be overwritten.

Disadvantages:
 - The upgrade process takes much longer to run.
   from 2-3 seconds to 15-25 seconds.

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-rootfs.img.gz
has to be written manually onto the disk. Please don't forget to backup
your configuration in this cases.

Note2: This patch requires
"base-files: upgrade: make get_partitions() endian agnostic"

Note3: If your current installation does not host the two
changes, sysupgrading will wipe the existing partition
layout. Don't forget to backup your data!

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
7 years agoe2fsprogs: Update to 1.43.7
Rosen Penev [Sat, 2 Dec 2017 22:09:48 +0000 (14:09 -0800)]
e2fsprogs: Update to 1.43.7

Compiled and tested on ramips with no noticeable problems.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 years agoapm821xx: MyBook Live convert to DT PHY
Christian Lamparter [Tue, 28 Nov 2017 21:51:09 +0000 (22:51 +0100)]
apm821xx: MyBook Live convert to DT PHY

Changes MyBook Live to use DT PHY probing and the broadcom phy driver.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
7 years agoapm821xx: convert to dt based diag LED script
Christian Lamparter [Tue, 28 Nov 2017 21:51:07 +0000 (22:51 +0100)]
apm821xx: convert to dt based diag LED script

Please note that users with a Netgear WNDR4700
will need to update the device-tree partition
manually.

For instructions, please refere to commit 49856a4bb581
("apm821xx: make it possible to update the dtb partition on the WNDR4700")

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
7 years agobase-files: unify get_dt_led helper function
Christian Lamparter [Tue, 28 Nov 2017 21:51:06 +0000 (22:51 +0100)]
base-files: unify get_dt_led helper function

Lantiq and IPQ806X (which includes IPQ40XX) both define the
same custom function {ipq806x|lantiq}_get_dt_led.

This patch moves the function into the base-file package at
lib/functions/leds.sh to make it more accessible for other
targets as well.

Cc: Mathias Kresin <dev@kresin.me>
Cc: John Crispin <john@phrozen.org>
Cc: Hannu Nyman <hannu.nyman@iki.fi>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
7 years agobrcm63xx: fix an OOPS when accessing the basemode register on 6368
Jonas Gorski [Wed, 13 Dec 2017 15:36:20 +0000 (16:36 +0100)]
brcm63xx: fix an OOPS when accessing the basemode register on 6368

The bcm6368 pinctrl driver passed the wrong variable to
devm_regmap_field_alloc, causing it to blow up when later trying to
access the field.

Fixes #1211.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobase-files: make including distfeeds.conf optional base-files-distfeeds-fixes
Jonas Gorski [Wed, 13 Dec 2017 13:41:28 +0000 (14:41 +0100)]
base-files: make including distfeeds.conf optional

To not clutter the system when building an opkg free image, generate the
distfeeds.conf only if CLEAN_IPKG is unset.

Since opkg is now a shared package, we can't rely on PACKAGE_opkg, but
since opkg is not reasonably usable without the status information, we
can tie the distfeeds.conf to it.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agobase-files: create /etc/opkg before generating distfeeds.conf
Jonas Gorski [Wed, 13 Dec 2017 13:30:30 +0000 (14:30 +0100)]
base-files: create /etc/opkg before generating distfeeds.conf

Ensure /etc/opkg exists before trying to write there. This fixes a build
failure if SIGNED_PACKAGES is disabled.

Reported-by: Matthias Schiffer <mschiffer@universe-factory.net>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agopackage: move distfeeds.conf from opkg to base-files
Jonas Gorski [Tue, 12 Dec 2017 10:19:07 +0000 (11:19 +0100)]
package: move distfeeds.conf from opkg to base-files

All the relevant options used for distfeeds.conf are part of base-files,
so it makes more sense to move the file there as well.

This has the added benefit that the we can share the opkg package again,
reducing the amount of target specific packages.

Acked-by: Jo-Philipp Wich <jo@mein.io>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agoar71xx: fix lzma-loader build with glibc
Jonas Gorski [Wed, 13 Dec 2017 10:12:11 +0000 (11:12 +0100)]
ar71xx: fix lzma-loader build with glibc

For an unknown reason gcc tries to link in crti.o when building with a
glibc toolchain (this does not happen with other targets). Prevent this
by telling gcc explicitly to not do that.

Fixes the following build error:

/home/jonas/git/lede/staging_dir/toolchain-mips_24kc_gcc-5.5.0_glibc/lib/gcc/mips-openwrt-linux-gnu/5.5.0/../../../../mips-openwrt-linux-gnu/lib/crti.o: In function `_init':
(.init+0x18): relocation truncated to fit: R_MIPS_GOT16 against `__gmon_start__'
/home/jonas/git/lede/staging_dir/toolchain-mips_24kc_gcc-5.5.0_glibc/lib/gcc/mips-openwrt-linux-gnu/5.5.0/../../../../mips-openwrt-linux-gnu/lib/crti.o: In function `_init':
(.init+0x28): relocation truncated to fit: R_MIPS_CALL16 against `__gmon_start__'
collect2: error: ld returned 1 exit status

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
7 years agowireguard: bump to 20171211
Kevin Darbyshire-Bryant [Tue, 12 Dec 2017 19:34:17 +0000 (19:34 +0000)]
wireguard: bump to 20171211

Bump to latest WireGuard snapshot release:

44f8e4d version: bump snapshot
bbe2f94 chacha20poly1305: wire up avx512vl for skylake-x
679e53a chacha20: avx512vl implementation
10b1232 poly1305: fix avx512f alignment bug
5fce163 chacha20poly1305: cleaner generic code
63a0031 blake2s-x86_64: fix spacing
d2e13a8 global: add SPDX tags to all files
d94f3dc chacha20-arm: fix with clang -fno-integrated-as.
3004f6b poly1305: update x86-64 kernel to AVX512F only
d452d86 tools: no need to put this on the stack
0ff098f tools: remove undocumented unused syntax
b1aa43c contrib: keygen-html for generating keys in the browser
e35e45a kernel-tree: jury rig is the more common spelling
210845c netlink: rename symbol to avoid clashes
fcf568e device: clear last handshake timer on ifdown
d698467 compat: fix 3.10 backport
5342867 device: do not clear keys during sleep on Android
88624d4 curve25519: explictly depend on AS_AVX
c45ed55 compat: support RAP in assembly
7f29cf9 curve25519: modularize dispatch

Refresh patches.

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

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
7 years agodropbear: disable MD5 HMAC and switch to sha1 fingerprints
Martin Schiller [Wed, 22 Nov 2017 12:39:51 +0000 (13:39 +0100)]
dropbear: disable MD5 HMAC and switch to sha1 fingerprints

As MD5 is known weak for many years and more and more
penetration test tools complain about enabled MD5 HMAC
I think it's time to drop it.

By disabling the MD5 HMAC support dropbear  will also
automatically use SHA1 for fingerprints.
This shouldn't be a problem too.

Signed-off-by: Martin Schiller <ms@dev.tdt.de>
7 years agotreewide: add only one device when appending to TARGET_DEVICES
Luis Araneda [Sun, 10 Dec 2017 22:45:20 +0000 (19:45 -0300)]
treewide: add only one device when appending to TARGET_DEVICES

This will avoid some conflicts when doing a git rebase or merge,
specially when adding support to a new device.

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
[drop brcm47xx changes which rename the images]
Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agorules.mk: export TMPDIR
Jo-Philipp Wich [Tue, 12 Dec 2017 15:28:04 +0000 (16:28 +0100)]
rules.mk: export TMPDIR

Set TMPDIR to the same value as the existing TMP_DIR variable in order to
let gcc and various other utilities use the local temporary directory
instead of the system-wide one.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agowolfssl: update to 3.12.2 (1 CVE)
Jo-Philipp Wich [Tue, 12 Dec 2017 16:30:34 +0000 (17:30 +0100)]
wolfssl: update to 3.12.2 (1 CVE)

Update wolfssl to the latest release v3.12.2 and backport an upstream
pending fix for CVE-2017-13099 ("ROBOT vulnerability").

Ref: https://github.com/wolfSSL/wolfssl/pull/1229
Ref: https://robotattack.org/

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobuild: allow PKG_PREPARED_DEPENDS and PKG_CONFIG_DEPENDS to be changed after includin...
Felix Fietkau [Tue, 12 Dec 2017 11:20:56 +0000 (12:20 +0100)]
build: allow PKG_PREPARED_DEPENDS and PKG_CONFIG_DEPENDS to be changed after including package.mk

Reverts commit a9c96ef0ac7ac99e4928f5312f3d0d1252c98328 and replaces it
with a different approach

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agomwlwifi: update to version 10.3.4.0 / 2017-11-29
Kabuli Chana [Wed, 29 Nov 2017 02:46:32 +0000 (19:46 -0700)]
mwlwifi: update to version 10.3.4.0 / 2017-11-29

Improves stability on WRT3200ACM

Signed-off-by: Kabuli Chana <newtownBuild@gmail.com>
7 years agoar71xx: fix board detection with newer RouterBOOT versions
Gabor Juhos [Fri, 8 Dec 2017 22:31:55 +0000 (23:31 +0100)]
ar71xx: fix board detection with newer RouterBOOT versions

Recent RouterBOOT version (at least version 3.41 on RB911G-5HPacD)
use "Board=" kernel parameter instead of "board=" to pass the board
name to the kernel. Due to this change the board detection code is
not working on the devices shipped with the new RouterBOOT version.
Because the kernel is unable to identify these boards they become
unusable despite that they are supported by the current code.

Update the prom_init code to convert the 'Board' kernel parameter to
'board'. After this change, the board detection works also with the
new RouterBOOT versions.

Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
7 years agobrcm47xx: remove versions from linksys-e1000 target
Moritz Warning [Sat, 9 Dec 2017 01:57:51 +0000 (02:57 +0100)]
brcm47xx: remove versions from linksys-e1000 target

The target name does not need to included a revision
if all revisions are supported.
This target supports all revisions (v1, v2, v2.1).

Signed-off-by: Moritz Warning <moritzwarning@web.de>
[Keep the version numbers in the device title, it doesn't harm]
Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agobrcm47xx: use proper region code in image name
Moritz Warning [Sat, 9 Dec 2017 01:44:43 +0000 (02:44 +0100)]
brcm47xx: use proper region code in image name

Replace 'north-america' by 'na' and remove 'other-regions' in image
files for Netgear WGR614 v10.

Signed-off-by: Moritz Warning <moritzwarning@web.de>
7 years agolantiq: dgn3500 drop worldwide suffix
Mathias Kresin [Mon, 11 Dec 2017 09:27:17 +0000 (10:27 +0100)]
lantiq: dgn3500 drop worldwide suffix

Remove the WW suffix, everything without a region suffix is world wide
anyway.

While at it, normalise the image filenames by using only lower case
characters.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agokernel: fix spelling in CONFIG_DEVTMPFS help text
Sascha Paunovic [Wed, 29 Nov 2017 18:53:33 +0000 (19:53 +0100)]
kernel: fix spelling in CONFIG_DEVTMPFS help text

Change "ti" to "to", as that's the correct spelling.

Signed-off-by: Sascha Paunovic <azarus@posteo.net>
7 years agolantiq: nand: drop ubifs images
Mathias Kresin [Thu, 19 Oct 2017 21:11:59 +0000 (23:11 +0200)]
lantiq: nand: drop ubifs images

Users are confused which image type they should use and there are more
drawbacks than adavantages in using a r/w ubifs rootfs in constrast to
a read-only squashfs rootfs like:

 - less available free flash space due to better compression of squashfs
   images
 - no support for factory reset due to r/w filesystem
 - possibility to break failsafe due to r/w filesystem

Therefore, drop support for r/w ubifs rootfs images.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agodnsmasq: add DHCP build switch support in full variant
Hans Dedecker [Sun, 10 Dec 2017 15:25:18 +0000 (16:25 +0100)]
dnsmasq: add DHCP build switch support in full variant

Add config option which allows to enable/disable DHCP support at compile
time. Make DHCPv6 support dependant on DHCP support as DHCPv6 support
implies having DHCP support.

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agobusybox: fix glibc libresolv dependency for LEDE nslook applet
Jo-Philipp Wich [Sun, 10 Dec 2017 13:26:19 +0000 (14:26 +0100)]
busybox: fix glibc libresolv dependency for LEDE nslook applet

Fixes d1ba483472 merge: busybox: update CONFIG_NSLOOKUP in busybox config.
Fixes FS#1212.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agobase-files: remove remaining uses of %N, and drop VERSION_NICK config symbol
Matthias Schiffer [Sat, 9 Dec 2017 22:25:21 +0000 (23:25 +0100)]
base-files: remove remaining uses of %N, and drop VERSION_NICK config symbol

Fixes d23e1e1e1a "merge: properly remove %n / %N references"

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
7 years agomerge: properly remove %n / %N references
Jo-Philipp Wich [Sat, 9 Dec 2017 15:01:14 +0000 (16:01 +0100)]
merge: properly remove %n / %N references

- use %d instead of %n for opkg feed identifiers
- remove %n / %N references from version files

Fixes bf5cef47b3 merge: release/banner: drop release name and update banner.
Fixes FS#1213.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
7 years agoramips: fix a typo in 02_network
Pavel Kubelun [Fri, 8 Dec 2017 19:36:06 +0000 (22:36 +0300)]
ramips: fix a typo in 02_network

The typo in network defaults script in ramips target that prevents
defaults to initialize.

Signed-off-by: Pavel Kubelun <be.dissent@gmail.com>
7 years agonetifd: always send DHCPv4 hostname
Mathias Kresin [Fri, 8 Dec 2017 08:35:26 +0000 (09:35 +0100)]
netifd: always send DHCPv4 hostname

udhcpc doesn't send a hostname by default. Use the system hostname if
nothing else is specified, to always send a hostname.

It syncs the behaviour to odhcpc, which always sends a hostname.

Signed-off-by: Mathias Kresin <dev@kresin.me>
Acked-by: Stijn Tintel <stijn@linux-ipv6.be>
Acked-by: Hans Dedecker <dedeckeh@gmail.com>
7 years agoprocd: nand: remove nand_board_name platform override
Mathias Kresin [Fri, 7 Apr 2017 16:06:17 +0000 (18:06 +0200)]
procd: nand: remove nand_board_name platform override

It isn't uses anymore by any target.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoar71xx: remove nand_board_name platform override
Mathias Kresin [Fri, 7 Apr 2017 16:03:50 +0000 (18:03 +0200)]
ar71xx: remove nand_board_name platform override

The boardname isn't used any longer to find the subdirectory in the
sysupgrade tar archive, which makes this override useless.

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agoprocd: nand: dont rely on boardname in nand_upgrade_tar
Mathias Kresin [Fri, 7 Apr 2017 06:39:59 +0000 (08:39 +0200)]
procd: nand: dont rely on boardname in nand_upgrade_tar

Kernel and rootfs in a subdirectory matching the userspace boardname,
was intended to use a single sysupgrade-tar archive for multiple boards
with different kernel/rootfs images. This feature was never used.

Use the first found directory in the tar archive instead of relying on
a directory named according to the userspace boardname.

It allows to change the boardname without adding another compatibility
layer - using the nand_board_name() function - for (sub)targets using
the metadata based image validation in favour to
nand_do_platform_check().

Signed-off-by: Mathias Kresin <dev@kresin.me>
7 years agokernel: MIPS compile out no-op DMA mapping ops where possible
Felix Fietkau [Tue, 5 Dec 2017 12:01:40 +0000 (13:01 +0100)]
kernel: MIPS compile out no-op DMA mapping ops where possible

Slightly improves networking throughput on some devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 years agoag71xx: Reduce NAPI weight to 32.
Rosen Penev [Mon, 4 Dec 2017 19:40:22 +0000 (11:40 -0800)]
ag71xx: Reduce NAPI weight to 32.

Qualcomm claims this reduces cache misses. Original commit message below:

From: Ben Menchaca <ben.menchaca@qca.qualcomm.com>
Date: Tue, 11 Jun 2013 12:18:46 -0500
Subject: [ag71xx] reduce NAPI weight

In an attempt to increase our cache warmth, we are decreasing NAPI.
This increases the warmth of the reused SKBs.

Signed-off-by: Ben Menchaca <ben.menchaca@qca.qualcomm.com>
Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 years agoRevert "ag71xx: Switch from driver to kernel macro for NAPI_WEIGHT."
Rosen Penev [Mon, 4 Dec 2017 19:40:21 +0000 (11:40 -0800)]
Revert "ag71xx: Switch from driver to kernel macro for NAPI_WEIGHT."

The motivation for this was misguided. It turns out tuning the NAPI weight could be useful for testing purposes. Therefore reverting.

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