openwrt/staging/linusw.git
6 years agolibtool: Update to 2.4.6
Rosen Penev [Tue, 20 Mar 2018 00:41:05 +0000 (17:41 -0700)]
libtool: Update to 2.4.6

Compile tested on mvebu.

old size:
12947 bin/packages/mips_24kc/base/libltdl_2.4-2_mips_24kc.ipk

new size:
13002 bin/packages/mips_24kc/base/libltdl_2.4.6-1_mips_24kc.ipk

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agopistachio: remove kernel 4.9 support
Matti Laakso [Sun, 18 Mar 2018 15:26:24 +0000 (17:26 +0200)]
pistachio: remove kernel 4.9 support

Signed-off-by: Matti Laakso <matti.laakso@outlook.com>
6 years agopistachio: Switch to kernel 4.14
Matti Laakso [Sun, 18 Mar 2018 15:25:43 +0000 (17:25 +0200)]
pistachio: Switch to kernel 4.14

Signed-off-by: Matti Laakso <matti.laakso@outlook.com>
6 years agopistachio: add kernel 4.14 support
Matti Laakso [Sun, 18 Mar 2018 15:23:08 +0000 (17:23 +0200)]
pistachio: add kernel 4.14 support

Add patches and config for 4.14, refreshed from 4.9.

Signed-off-by: Matti Laakso <matti.laakso@outlook.com>
6 years agomvebu: add support for MACCHIATObin (cortex-a72)
Damir Samardzic [Mon, 26 Mar 2018 09:37:58 +0000 (09:37 +0000)]
mvebu: add support for MACCHIATObin (cortex-a72)

Add initial support for Marvell MACCHIATObin, cortex-a72 based Marvell
ARMADA 8040 Community board. Comes in two forms: Single Shot and Double
Shot.

Specifications:
- Quad core Cortex-A72 (up to 2GHz)
- DDR4 DIMM slot with optional ECC and single/dual chip select support
- Dual 10GbE (1/2.5/10GbE) via copper or SFP
  2.5GbE (1/2.5GbE) via SFP
  1GbE via copper
- SPI Flash
- 3 X SATA 3.0 connectors
- MicroSD connector
- eMMC
- PCI x4 3.0 slot
- USB 2.0 Headers (Internal)
- USB 3.0 connector
- Console port (UART) over microUSB connector
- 20-pin Connector for CPU JTAG debugger
- 2 X UART Headers
- 12V input via DC Jack
- ATX type power connector
- Form Factor: Mini-ITX (170 mm x 170 mm)

More details at http://macchiatobin.net

Booting from micro SD card:
 1. reset U-Boot environment:
      env default -a
      saveenv

 2. prepare U-Boot with boot script:
      setenv bootcmd "load mmc 1:1 0x4d00000 boot.scr; source 0x4d00000"
      saveenv

   or manually:
      setenv fdt_name armada-8040-mcbin.dtb
      setenv image_name Image
      setenv bootcmd 'mmc dev 1; ext4load mmc 1:1 $kernel_addr $image_name;ext4load mmc 1:1 $fdt_addr $fdt_name;setenv   bootargs $console root=/dev/mmcblk1p2 rw rootwait; booti $kernel_addr - $fdt_addr'
      saveenv

Signed-off-by: Damir Samardzic <damir.samardzic@sartura.hr>
6 years agou-boot-mvebu: update to 2018.03
Josua Mayer [Wed, 21 Mar 2018 00:13:02 +0000 (01:13 +0100)]
u-boot-mvebu: update to 2018.03

This release brings various improvements to clearfog support, such as distro-boot.
Obsoletes:
0002-clearfog-reset-usom-onboard-1512-phy.patch
0003-clearfog-enable-distro-boot-code.patch

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
6 years agomvebu: clearfog-pro: set new DTB name in boot-script
Josua Mayer [Wed, 21 Mar 2018 00:32:06 +0000 (01:32 +0100)]
mvebu: clearfog-pro: set new DTB name in boot-script

The DTB for Clearfog Pro has been renamed in mainline. However U-Boot
hasn't picked up that change yet :(, so we need to hardcode it for now.

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
6 years agou-boot-mvebu: set configuration options in Makefile
Josua Mayer [Tue, 20 Mar 2018 23:10:17 +0000 (00:10 +0100)]
u-boot-mvebu: set configuration options in Makefile

CONFIG_* variables can easily be set by overriding Build/Configure.
so set NET_RANDOM_ETHADDR=y and CMD_SETEXPR=y here.

This replaces the following patches:
0001-clearfog-generate-random-MAC-address.patch
0004-clearfog-enable-setexpr-command-by-default.patch

Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
6 years agomvebu: Add support for WRT3200ACM with new NAND flash
Imre Kaloz [Thu, 15 Feb 2018 18:57:05 +0000 (19:57 +0100)]
mvebu: Add support for WRT3200ACM with new NAND flash

Newer Linksys boards might come with a Winbond W29N02GV which can be
configured in different ways. Make sure we configure it the same way as
the older chips so everything keeps working.

Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
6 years agomvebu: Get rid of RTC hack for Turris Omnia.
Rosen Penev [Mon, 26 Mar 2018 02:14:20 +0000 (19:14 -0700)]
mvebu: Get rid of RTC hack for Turris Omnia.

As Solidrun's RTC patch got merged, this hack is no longer needed.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agoath10k-ct: Update firmware to latest.
Ben Greear [Mon, 26 Mar 2018 22:28:00 +0000 (15:28 -0700)]
ath10k-ct: Update firmware to latest.

Wave-1 firmware has a fix for 'addba' not finding the peer.  Thanks to Hauke
for finding and reporting this.

Wave-2 firmware has a fix for leaking a peer multicast key when a monitor device
is created.

And I re-ordered the '4019' firmware images in the Makefile to match the order
of the others.  No functional change for that reorder.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Tested-by: Rosen Penev <rosenp@gmail.com>
6 years agoopenssl: update to 1.0.2o
Paul Wassi [Fri, 30 Mar 2018 06:15:00 +0000 (08:15 +0200)]
openssl: update to 1.0.2o

Fixes CVE-2018-0739

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
6 years agoRevert "iproute2: fix hidden uint to uin64_t promotion in json_print"
Kevin Darbyshire-Bryant [Fri, 30 Mar 2018 15:46:58 +0000 (15:46 +0000)]
Revert "iproute2: fix hidden uint to uin64_t promotion in json_print"

This reverts commit 745d0e7f4b6e8659cc967291acd33889035127f0.

It looks like upstream don't want the patch so let's revert it here too.

I hope a fix from upstream is forthcoming.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agobuild: filter out kmod-ipt-offload from the default selection on targets that do...
Felix Fietkau [Thu, 29 Mar 2018 14:12:43 +0000 (16:12 +0200)]
build: filter out kmod-ipt-offload from the default selection on targets that do not support it

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: add missing config option
Hauke Mehrtens [Thu, 29 Mar 2018 23:06:29 +0000 (01:06 +0200)]
kernel: add missing config option

CONFIG_NVMEM_BCM_OCOTP was added in kernel 4.10 and it is possible to
activate it on the bcm53xx target. Deactivate it by default to fix the
build of the bcm53xx target.
This was found by build bot.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoapm821xx: increase WNDR4700's dtb+kernel partition to 3.5 MiB
Christian Lamparter [Mon, 26 Mar 2018 15:25:32 +0000 (17:25 +0200)]
apm821xx: increase WNDR4700's dtb+kernel partition to 3.5 MiB

This patch cleans and reworks the WNDR4700 dts to increase the
now combined dtb+kernel partition to 3.5 MiB. This has become
necessary due to the switch to GCC 7.3 and the ever increasing
kernel binary size.

The dtb+kernel partition was combined in order to finally
fix the problem with out-of-sync device-trees. From now
on, the kernel and device-tree will always be updated together.

Upgrade Note:
Existing installations will have to use the TFTP firmware
recovery option in order to install the update. Affected users
are advised to make a backup of their existing configuration
prior to running sysupgrade:
<https://openwrt.org/docs/guide-user/installation/generic.backup#backup_openwrt_configuration>
Due to the repartitioning of the NAND, the generated backup
should be placed on either the internal HDD, an attached
USB-Stick or on another PC (externally).

To manually trigger the firmware recovery, the reset button has
to be pressed (and hold) during boot. U-boot will enter the "Upgrade
Mode" and starts a tftpserver listening on 192.168.1.1 for a
tftp client from one of the four LAN/Ethernet ports to connect and
upload the new system: (enable tftp binary mode!).
openwrt-apm821xx-nand-netgear_wndr4700-squashfs-factory.img

Cc: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
6 years agoapm821xx: switch MR24's initramfs to multi-image method
Christian Lamparter [Sun, 18 Mar 2018 21:51:07 +0000 (22:51 +0100)]
apm821xx: switch MR24's initramfs to multi-image method

The recent change to switch to gcc 7.3 broke the image
generation code, as the kernel would no longer fit into
KERNEL_SIZE.

This patch fixes the issue by reworking the initramfs
creation and packaging, which will get rid of the
KERNEL_SIZE check in the process.

This new initramfs can be loaded through the MR24 U-boot
in the following way:

=> setenv ipaddr 192.168.1.1
=> setenv bootargs console=ttyS0,$baudrate
=> tftpboot c00000 192.168.1.2:meraki_mr24-initramfs-kernel.bin
[...]
Load address: 0xc00000
Loading: ################################################ [...]
done
    Bytes transferred = 5952544 (5ad420 hex)
    => bootm $fileaddr
    \## Booting kernel from Legacy Image at 00c00000 ...
    ...

For more information and the latest flashing guide:
please visit the OpenWrt Wiki Page for the MR24:
<https://openwrt.org/toh/meraki/mr24#flashing>

Cc: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Signed-off-by: Chris Blake <chrisrblake93@gmail.com>
6 years agomap: fix psidlen becoming negative (FS#1430)
Hans Dedecker [Thu, 29 Mar 2018 14:25:14 +0000 (16:25 +0200)]
map: fix psidlen becoming negative (FS#1430)

Fix psidlen becomes negative in case embedded address bit lenght is smaller than
IPv4 suffix length.
While at it improve parameter checking making the code more logical and
easier to read.

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agobcm53xx: switch to kernel 4.14
Rafał Miłecki [Thu, 29 Mar 2018 13:31:17 +0000 (15:31 +0200)]
bcm53xx: switch to kernel 4.14

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agobcm53xx: update kernel 4.14 config
Rafał Miłecki [Thu, 29 Mar 2018 13:29:23 +0000 (15:29 +0200)]
bcm53xx: update kernel 4.14 config

There are few symbol changes but only 3 symbols were possible to set:
THERMAL_EMERGENCY_POWEROFF_DELAY_MS
CLK_BCM_SR
PHY_NS2_USB_DRD

Both new drivers don't seem to apply to Northstar so they are disabled.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agobcm53xx: fix earlycon regression in kernel 4.14
Rafał Miłecki [Thu, 29 Mar 2018 13:10:01 +0000 (15:10 +0200)]
bcm53xx: fix earlycon regression in kernel 4.14

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agoRevert "ppp: make ppp-multilink provide ppp"
Felix Fietkau [Thu, 29 Mar 2018 09:57:08 +0000 (11:57 +0200)]
Revert "ppp: make ppp-multilink provide ppp"

opkg currently has some issues with Provides and this change makes the
image builder fail because of that. Revert the change for now until opkg
is fixed

This reverts commit 092d75aa3e86db8331fffdbd0a99987df9dc438b.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoar71xx: fix ar934x usb controller resource conflict
Felix Fietkau [Tue, 27 Mar 2018 22:58:58 +0000 (00:58 +0200)]
ar71xx: fix ar934x usb controller resource conflict

Use the right size for the EHCI block

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoprocd: update to the latest version
Yousong Zhou [Wed, 28 Mar 2018 09:36:57 +0000 (17:36 +0800)]
procd: update to the latest version

Changes since last version

    dfb68f8 service: initialize supplementary group ids
    3db4e6d service: add func for string config change check
    c3faabe procd: get rid of putenv usage.

The supplementary group id change fixes FS#988

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
6 years agohostapd: update to git snapshot of 2018-03-26
Daniel Golle [Tue, 27 Mar 2018 17:24:27 +0000 (19:24 +0200)]
hostapd: update to git snapshot of 2018-03-26

The following patches were merged upstream:
000-hostapd-Avoid-key-reinstallation-in-FT-handshake.patch
 replaced by commit 0e3bd7ac6
001-Prevent-reinstallation-of-an-already-in-use-group-ke.patch
 replaced by commit cb5132bb3
002-Extend-protection-of-GTK-IGTK-reinstallation-of-WNM-.patch
 replaced by commit 87e2db16b
003-Prevent-installation-of-an-all-zero-TK.patch
 replaced by commit 53bb18cc8
004-Fix-PTK-rekeying-to-generate-a-new-ANonce.patch
 replaced by commit 0adc9b28b
005-TDLS-Reject-TPK-TK-reconfiguration.patch
 replaced by commit ff89af96e
006-WNM-Ignore-WNM-Sleep-Mode-Response-without-pending-r.patch
 replaced by commit adae51f8b
007-FT-Do-not-allow-multiple-Reassociation-Response-fram.patch
 replaced by commit 2a9c5217b
008-WPA-Extra-defense-against-PTK-reinstalls-in-4-way-ha.patch
 replaced by commit a00e946c1
009-Clear-PMK-length-and-check-for-this-when-deriving-PT.patch
 replaced by commit b488a1294
010-Optional-AP-side-workaround-for-key-reinstallation-a.patch
 replaced by commit 6f234c1e2
011-Additional-consistentcy-checks-for-PTK-component-len.patch
 replaced by commit a6ea66530
012-Clear-BSSID-information-in-supplicant-state-machine-.patch
 replaced by commit c0fe5f125
013-WNM-Ignore-WNM-Sleep-Mode-Request-in-wnm_sleep_mode-.patch
 replaced by commit 114f2830d

Some patches had to be modified to work with changed upstream source:
380-disable_ctrl_iface_mib.patch (adding more ifdef'ery)
plus some minor knits needed for other patches to apply which are not
worth being explicitely listed here.

For SAE key management in mesh mode, use the newly introduce
sae_password parameter instead of the psk parameter to also support
SAE keys which would fail the checks applied on the psk field (ie.
length and such). This fixes compatibility issues for users migrating
from authsae.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
6 years agouci: update to latest git HEAD
Hans Dedecker [Tue, 27 Mar 2018 09:37:34 +0000 (11:37 +0200)]
uci: update to latest git HEAD

5d2bf09 uci: fix a potential use-after-free in uci_set()
3b3d63e list: only record ordering deltas if element position changed
4c4d343 cmake: Fix cli shared linking against ubox

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agonetifd: update to the latest version (fixes FS#1452)
Felix Fietkau [Tue, 27 Mar 2018 09:32:54 +0000 (11:32 +0200)]
netifd: update to the latest version (fixes FS#1452)

9c8d781 netifd: return the interface for locally addressable host dependencies

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agobuild: include kmod-ipt-offload in default images
Felix Fietkau [Tue, 27 Mar 2018 06:24:40 +0000 (08:24 +0200)]
build: include kmod-ipt-offload in default images

Netfilter flow offload has now started to become useful and suitable for
a wider testing audience. Configuring it via UCI is also integrated in
firewall3 by adding 'option flow_offloading 1' to the 'defaults'
section in /etc/config/firewall

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: add kmod-sound-ens1371
Felix Fietkau [Wed, 21 Mar 2018 17:49:29 +0000 (18:49 +0100)]
kernel: add kmod-sound-ens1371

This audio chip is provided as a virtual audio device by VMware

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agodnsmasq: improve init script portability (FS#1446)
Hans Dedecker [Sat, 24 Mar 2018 17:28:01 +0000 (18:28 +0100)]
dnsmasq: improve init script portability (FS#1446)

Improve portability of init script by declaring resolvfile as local
in dnsmasq_stop function.
Fixes resolvfile being set for older busybox versions in dnsmasq_start
in a multi dnsmasq instance config when doing restart; this happens when
the last instance has a resolvfile configured while the first instance
being started has noresolv set to 1.

Base on a patch by "Phil"

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agosunxi: add build for sopine
Fan Fan [Mon, 19 Mar 2018 09:52:46 +0000 (17:52 +0800)]
sunxi: add build for sopine

This will generate image for Pine64 Sopine board.

Signed-off-by: Fan Fan <fkpwolf@gmail.com>
6 years agoethtool: Update to 4.15.
Rosen Penev [Mon, 19 Mar 2018 00:07:28 +0000 (17:07 -0700)]
ethtool: Update to 4.15.

Contains kernel 4.14 updates. Compile tested on mvebu.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agoKernel: bump 4.14 to 4.14.29
Stijn Segers [Thu, 22 Mar 2018 17:34:08 +0000 (18:34 +0100)]
Kernel: bump 4.14 to 4.14.29

Right patch version this time, sorry!

* Patch 180-usb-xhci-add-support-for-performing-fake-doorbell.patch had to be adjusted slightly because of upstream adapted code.
* Refreshed patches.

Compile-tested: ramips/mt7621, x86/64
Run-tested: ramips/mt7621, x86/64
Signed-off-by: Stijn Segers <foss@volatilesystems.org>
Tested-by: Rosen Penev <rosenp@gmail.com>
6 years agoiproute2: cake: support new overhead reporting & stats structures
Kevin Darbyshire-Bryant [Thu, 22 Mar 2018 11:43:05 +0000 (11:43 +0000)]
iproute2: cake: support new overhead reporting & stats structures

Cake in kernel space now splits stats structure handling across netlink
messages to reduce stack usage issue flagged by upstream kernel checks.
Update user space (tc) qdisc handling to understand this new regime.

Cake also reports packet overheads & compensation in a different way so
add display code for this. e.g.

'tc -s qdisc show dev eth0' reports this extra detail:

 min/max transport layer size:         28 /    1500
 min/max overhead-adjusted size:       65 /    1550
 average transport hdr offset:         14

Cake also supports output in JSON format.

Patch is bulkier than before because a (slightly out of date - see above
stats) man page is included for reference. Better than nothing!

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agokmod-sched-cake: split stats structures, add overhead stats
Kevin Darbyshire-Bryant [Thu, 22 Mar 2018 11:43:04 +0000 (11:43 +0000)]
kmod-sched-cake: split stats structures, add overhead stats

Relevant changes:

0afc1be Fixes for kernel 4.16
d2d6780 Reinitialise overhead compensation stats when reconfiguring.
a3bab9d Export overhead compensation stats to userspace.
9cd2fa8 Split tin stats to its own structure to decrease size of tc_cake_xstats
71c7b44 Gather more statistics about packet length transformations.
0517357 Rework overhead compensation to use dynamic transport header offset instead of (inaccurate) static one.
c1a0c8e Refactor length handling code to better centralise overhead calculations

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agoath10k-ct: fix module depends
Hauke Mehrtens [Thu, 22 Mar 2018 22:24:42 +0000 (23:24 +0100)]
ath10k-ct: fix module depends

PKG_EXTMOD_SUBDIRS should be set to the sub directory where the kernel
module gets build in, for the ath10k-ct driver this changed in commit
3888e77c1c68 from ath10k to ath10k-4.13. Without this fix the depends
line of the ath10*.ko modules is empty and the kernel module load system
will not automatically load the depended modules like mac80211.

Fixes: 3888e77c1c68 ("ath10k-ct driver: use dma_alloc_coherent, 4.13 based driver")
Fixes: 23a388fe41ec ("ath10k-ct: Force loading mac80211 and ath modules.")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoath10k-ct: Add htt-mgt variants of ath10k-ct firmware.
Ben Greear [Wed, 21 Mar 2018 17:29:03 +0000 (10:29 -0700)]
ath10k-ct: Add htt-mgt variants of ath10k-ct firmware.

The HTT-MGT variants transport management frames over the
normal HTT tx path, just like data frames.  This saves
limitted WMI buffers which can become depleted if lots of
management frames become stuck in TX queues due to peer
that went away.

In addition, at least for the wave-1 firmware, htt-mgt is
required in order for 802.11r (fast roaming) authentication
to function properly.

The htt-mgt firmware requires the use of the ath10k-ct
driver.  Normal non-htt-mgt ath10k-ct firmware should work
with stock drivers.

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agoath10k-ct: Update wave-1 and wave-2 firmware to latest.
Ben Greear [Wed, 21 Mar 2018 17:29:02 +0000 (10:29 -0700)]
ath10k-ct: Update wave-1 and wave-2 firmware to latest.

This updates to latest ath10k-ct firmware.  Hopefully we are
at the end of the development cycle for this firmware release,
so these should be stable.

wave-1 changes since last release:

Release 20

  *  Allow flushing peer when deleting.  Hopefully this will allow the
     peer delete command to happen in a reasonable amount of time even
     if the RF environment is busy (or peer has died).
     To enable this, set the high flag in the mac-addr second word in the
     ath10k driver near end of the ath10k_wmi_op_gen_peer_delete method:

     cmd->peer_macaddr.word1 |= __cpu_to_le32(0x80000000);

  *  Attempt to fix crash seen in resmgr-ocs, appearantly due to list corruption.
     Use a temporary list instead of trying to rely on for-each-safe.

  *  Add flag to tx-descriptor to allow driver to request no-ack on data
     frames.  This is bit 15 on the flag1 field (previously un-used).

  *  Add option to support specifying the tx-rate-code and retry count on
     a per-packet basis.  Only a single series is supported at this time.
     Useful mainly for radiotap monitor-tx type testing at this point.

  *  Fix crash on startup when chip is at -40 deg C and calibration fails.  Instead
     of asserting, just keep retrying calibration, which appears to start working
     after a few minutes (when the chip warms up).

  *  Allow reporting per-chain rssi for management frames.  We pack the values into
     empty space in the mgt-frame wmi header.  This will only be enabled if the driver
     requests it, since otherwise the driver is assumed to not understand the new API.
     ath10k-ct drivers that support this feature will automatically enable it.

  *  A customer reports a case that appears to be the hardware not properly detecting
     end of AMPDU, so frames were being mis-delivered to the wrong peer.  Attempt to
     work around this, and in doing so, clean up a bunch of void* abuse in the block-ack
     reordering code (could not ever confirm there was a problem in this area).

  *  Re-work the rx-mem logic to be less complicated and to use less memory.

  *  Attempt to fix crash that appearanty happens because the driver can sometimes
     delete a vdev in 'up' state.

  *  Attempt to fix hung scan state machine issues.

  *  Fix crash in tx path due to un-initialized memory.

wave-2 changes since last release:

Release 10

  *  Fix an assert related to tx scheduling.  This hopefully fixes
     what appears to be a regression that I added some time back.

  *  Enable CSI reporting for 9984, and maybe 9888/9886.  Only in
     non-trimmed builds.

  *  Other stability improvements, including regression fixes from
     some tricky bugs introduced in earlier releases.

  *  Allow compiling for IPQ4019 chipset.

  *  Firmware will now send txbf frames to the host (driver) if the
     TXBF (0xF00000001) set-special feature is enabled, or when the radio
     is in monitor mode.  But, if the frame is consumed by the txbf_cv
     logic, then the pkt cannot be delivered to the host in this manner.  Instead,
     a WMI event will be sent and host can find the txbf_cv data in shared
     memory.  See ath10k_wmi_event_txbf_cv_mesg() in ath10k-ct driver.

  *  Support rx-all-mgt option.  When enabled, the firmware will deliver all
     management frames that it can to the host.  No RX filters are changed
     when this option is enabled.

  *  Fix at least some problems with sending tx-beamforming frames to SU-MIMO
     peers.  Looks like this was a regression in my code.

  *  Fix a crash in rate-ctrl due to nss mismatch.  This was something I introduced
     while trying to fix other bugs in rate-ctrl some time back.

  *  Attempt to fix a sw-peer-key object leak in IBSS mode.  The peer key code
     is very complex, and shares some pointers as union members.  I think I fixed
     at least some of the issues, but would not be surprised if more exist.

  *  Improve ath10k user guide to document CT firmware features:
     https://www.candelatech.com/ath10k-ug.php

  *  Add ct-special option to configure the txbf sounding time.  See ath10k-ug.php

  *  Fix and allow the driver to tell the firmware to send sounding frames.  See ath10k-ug.php
     In further testing, this seems to fail much of the time, and I am not sure why.
     Disabling this in diet (trimmed) builds.

  *  Fix crashes related to deleting peers while they are in power-save mode.  Reported
     by LEDE user on r7800 with 9984 NIC.
  *  Make rate-ctrl txbf probe work better.  If enabled, the rate-ctrl logic will periodically
     send out probes at an NSS that can to txbf.  Previously, txbf probes would not reliably happen
     if both AP and peer had the same nss (ie, 2x2 talking to 2x2).  To enable this feature, you
     need to enable the fwtest-cmdid number 20.

  *  Report rx-timeout error counters.  These were previously un-reported, though the
     field existed in the wmi struct already.

  *  txbf:  Ignore frames not destined for us.  If NIC is in promisc mode, it
     could acquire and process NDPA frames that were not destined for it.  Check
     the dest-MAC and ignore frames not for us (pass them up the stack for monitor
     mode instead of save them in the peer's rate-ctrl logic.)

  *  Port ping-pong crash handling and othe related features to IPQ4019 target.  It should
     now act similar to 9984 in this regard.

  *  Fix a few asserts related to txbf and tx-seq logic.

  *  Add custom-stats support, for rx-reorder-stats.  Similar to what I did for wave-1.

  *  Disable AMSDU for IBSS.  This now matches what I did for peregrine.  It seems to
     work better this way, though I did not debug it in detail.

  *  Enable the set-special command to re-enable AMSDU for IBSS if user wants to experiment.

  *  Fix bug where dbglog did not disable IRQs, so if you made dbglog messages from the IRQ
     handler, it could cause corruption that could crash the firmware and/or corrupt the log
     message buffers.

  *  Don't assert if there are no buffer descriptors for RX of non-data frame.

  *  Retry any stuck block-ack sessions every 20 seconds instead of just disabling BA for
     ever when we get too many failures.

  *  Fix SGI flag when reporting tx-rate info.  The flag moved since wave-1 days, and
     I did not notice that when I ported my changes forward to wave-2.

  *  Allow disabling special CCA handling for IBSS txqs.  Earlier testing indicated this
     might improve throughput in some testing on 9984 chips in IBSS mode, but subsequent
     testing looks about the same without it.  Since I do not really understand what this
     setting exists for, leave it at upstream defaults.  A new set-special API command (0x12)
     can be used to enable this hack for testing.  Setting 0x1 bit disables special CCA handling
     for non-beacon IBSS txqs, setting 0x2 bit disables it for beacon queues as well.

  *  Add MCAST-BCAST feature flag.  This tells driver we do not need a monitor interface
     to do MESH.

  *  When calculating the rx-address filter (affects ACK & BLOCK-ACK, among other things),
     to not add in monitor interfaces if other interfaces are up.  There is no need for
     a monitor device to ACK frames.

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agoath10k-ct: Update driver to latest.
Ben Greear [Wed, 21 Mar 2018 17:29:01 +0000 (10:29 -0700)]
ath10k-ct: Update driver to latest.

Among other things, this will check for an htt-mgt variant of
ath10k-ct firmware before loading 'normal' firmware, and it disables
verbose printing of firmware DBGLOG messages by default.

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agoath10k-ct: Update DEPENDS and PROVIDES
Ben Greear [Wed, 21 Mar 2018 17:29:00 +0000 (10:29 -0700)]
ath10k-ct: Update DEPENDS and PROVIDES

Update DEPENDS and PROVIDES so that ath10k-ct firmware
and drivers can be used to replace stock firmware
and drivers.  The -htt firmware variant, which requires
ath10k-ct driver now selects ath10k-ct driver when the
firmware is selected.

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agolibubox: update to the latest version
Felix Fietkau [Wed, 21 Mar 2018 16:57:14 +0000 (17:57 +0100)]
libubox: update to the latest version

3aad294 libubox: Plug a small memory leak.
eebe3fc utils: use constant byte-order conversion

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoramips: remove dead (and potentially crashy) code in mt7621 gsw init
Felix Fietkau [Fri, 23 Mar 2018 12:40:41 +0000 (13:40 +0100)]
ramips: remove dead (and potentially crashy) code in mt7621 gsw init

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoramips: rename ethernet priv->device field to dev to match upstream
Felix Fietkau [Fri, 23 Mar 2018 12:37:00 +0000 (13:37 +0100)]
ramips: rename ethernet priv->device field to dev to match upstream

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: netfilter: fix dst entries in flowtable offload
Felix Fietkau [Thu, 15 Mar 2018 17:03:22 +0000 (18:03 +0100)]
kernel: netfilter: fix dst entries in flowtable offload

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: fix crash in flow offload when removing net devices
Felix Fietkau [Fri, 23 Mar 2018 19:09:23 +0000 (20:09 +0100)]
kernel: fix crash in flow offload when removing net devices

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: flow-offload: only offload connections that have been fully established
Felix Fietkau [Tue, 13 Mar 2018 08:16:20 +0000 (09:16 +0100)]
kernel: flow-offload: only offload connections that have been fully established

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agolantiq: intel-xway: add vr9 v1.1 phy support
Mathias Kresin [Thu, 22 Mar 2018 19:00:13 +0000 (20:00 +0100)]
lantiq: intel-xway: add vr9 v1.1 phy support

During upstreaming the intel phy driver, support for the vr9 v1.1
embedded phys got lost. Backport the upstream send patch adding support
for the vr9 v1.1 embbeded phys to the driver.

Signed-off-by: Mathias Kresin <dev@kresin.me>
cosmetic fixes

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: fix switch and MAC address for WHR-G300N
INAGAKI Hiroshi [Fri, 23 Mar 2018 05:59:39 +0000 (14:59 +0900)]
ramips: fix switch and MAC address for WHR-G300N

WHR-G300N has 5 ethernet ports (lan: 4, wan: 1), but there was no
correct configuration in 02_network script and 6 ports was configured
on the switch.
Also, since the MAC address was not acquired from factory partition,
incorrect values was set to LAN and WAN interfaces.

This commit fixes these issues.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
6 years agoipq40xx: Add reserved memory for WPJ428
Sven Eckelmann [Tue, 20 Mar 2018 08:04:54 +0000 (09:04 +0100)]
ipq40xx: Add reserved memory for WPJ428

The APPSBL and anything after that it not allowed to touch some of the
memory regions which are used by other components. Still trying to write to
the memory can lead to sudden device restarts

  (IPQ40xx) # mw 87e80000 0
  data abort
  pc : [<873149f8>]          lr : [<87308578>]
  sp : 86edfc28  ip : 86ef4412     fp : 00000000
  r10: 00000000  r9 : 00000000     r8 : 86edff68
  r7 : 00000003  r6 : 8737e624     r5 : 86ef4420  r4 : 8736c154
  r3 : 00000000  r2 : 00000010     r1 : 00000000  r0 : 00000000
  Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
  Resetting CPU ...

The device manufacturer only provided a very rough list of regions:

* rsvd1:     0x87000000 0x500000
* wifi_dump: 0x87500000 0x600000
* rsvd2:     0x87b00000 0x500000

A more detailed list for devices using the AP.DK reference design memory
maps was provided by Roman Yeryomin <roman@advem.lv> in commit 56f2df879fd
("ipq806x: ipq4019: add ap-dk01.1-c1 board support"):

* apps_bl:          0x87000000 0x400000
* sbl:              0x87400000 0x100000
* cnss_debug:       0x87500000 0x600000
* cpu_context_dump: 0x87b00000 0x080000
* tz_apps:          0x87b80000 0x280000
* smem:             0x87e00000 0x080000
* tz:               0x87e80000 0x180000

The u-boot function ipq_fdt_mem_rsvd_fixup seems to suggest that only the
rsvd2 (tz_apps, smem, tz) should be protected. All other regions would have
been removed by it when CONFIG_QCA_APPSBL_DLOAD is not enabled. This allows
to reduce the 16MB reserved memory region to only 4.5MB.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agoipq40xx: Use detailed reserved memory for A42
Sven Eckelmann [Tue, 20 Mar 2018 08:04:54 +0000 (09:04 +0100)]
ipq40xx: Use detailed reserved memory for A42

The APPSBL and anything after that it not allowed to touch some of the
memory regions which are used by other components. Still trying to write to
the memory can lead to sudden device restarts

  (IPQ40xx) # mw 87e80000 0
  data abort
  pc : [<873149f8>]          lr : [<87308578>]
  sp : 86edfc28  ip : 86ef4412     fp : 00000000
  r10: 00000000  r9 : 00000000     r8 : 86edff68
  r7 : 00000003  r6 : 8737e624     r5 : 86ef4420  r4 : 8736c154
  r3 : 00000000  r2 : 00000010     r1 : 00000000  r0 : 00000000
  Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
  Resetting CPU ...

The device manufacturer only provided a very rough list of regions:

* rsvd1:     0x87000000 0x500000
* wifi_dump: 0x87500000 0x600000
* rsvd2:     0x87b00000 0x500000

A more detailed list for devices using the AP.DK reference design memory
maps was provided by Roman Yeryomin <roman@advem.lv> in commit 56f2df879fd
("ipq806x: ipq4019: add ap-dk01.1-c1 board support"):

* apps_bl:          0x87000000 0x400000
* sbl:              0x87400000 0x100000
* cnss_debug:       0x87500000 0x600000
* cpu_context_dump: 0x87b00000 0x080000
* tz_apps:          0x87b80000 0x280000
* smem:             0x87e00000 0x080000
* tz:               0x87e80000 0x180000

The u-boot function ipq_fdt_mem_rsvd_fixup seems to suggest that only the
rsvd2 (tz_apps, smem, tz) should be protected. All other regions would have
been removed by it when CONFIG_QCA_APPSBL_DLOAD is not enabled. This allows
to reduce the 16MB reserved memory region to only 4.5MB.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agoipq40xx: Remove phy reset gpio from Cisco Meraki MR33
Sven Eckelmann [Mon, 19 Mar 2018 20:10:14 +0000 (21:10 +0100)]
ipq40xx: Remove phy reset gpio from Cisco Meraki MR33

There is currently no code to read the phy reset gpios for the ethernet
PHY. It would also have been better to use the more common name
"phy-reset-gpios" for this property.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
6 years agoipq40xx: Adjust SoC name of AVM Fritz!Box 4040
Sven Eckelmann [Mon, 19 Mar 2018 19:53:09 +0000 (20:53 +0100)]
ipq40xx: Adjust SoC name of AVM Fritz!Box 4040

The AVM Fritz!Box 4040 uses an IPQ4018 as SoC and not an IPQ4019. The DTS
must be adjusted to reflect this.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
6 years agoipq40xx: Use constant to set gpio active low/high
Sven Eckelmann [Mon, 19 Mar 2018 13:52:38 +0000 (14:52 +0100)]
ipq40xx: Use constant to set gpio active low/high

The GPIO configuration in the DTS have as third parameter the active
low/high configuration. This parameter is not easy to parse by humans when
it is only set to 0/1. It is better to use the predefined constants
GPIO_ACTIVE_HIGH and GPIO_ACTIVE_LOW.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agoipq40xx: Fix DTS status parameter values
Sven Eckelmann [Mon, 19 Mar 2018 11:23:00 +0000 (12:23 +0100)]
ipq40xx: Fix DTS status parameter values

The "Devicetree Specification, Release v0.2 - 2.3.4 status" [1] only allows
the "okay" value for an operational device. The "ok" value will be accepted
by the kernel but should be avoided.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agoipq40xx: Provide prefered DTS config name for A42
Sven Eckelmann [Mon, 19 Mar 2018 11:36:13 +0000 (12:36 +0100)]
ipq40xx: Provide prefered DTS config name for A42

The OpenMesh A42 will use the default config entry in the FIT when no other
on is found but prefers the config@om.a42. This only becomes relevant when
a Multi-FIT image is prepared for this device.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agoipq40xx: Remove SUPPORTED_DEVICES for A42
Sven Eckelmann [Mon, 19 Mar 2018 11:21:36 +0000 (12:21 +0100)]
ipq40xx: Remove SUPPORTED_DEVICES for A42

The SUPPORTED_DEVICES will be defined automatically via Device/Default in
ipq40xx.

Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
6 years agokernel: fix offloading connections with SNAT + DNAT
Felix Fietkau [Fri, 23 Mar 2018 18:15:58 +0000 (19:15 +0100)]
kernel: fix offloading connections with SNAT + DNAT

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: fix flow offload UDP handling issue
Felix Fietkau [Fri, 23 Mar 2018 16:18:00 +0000 (17:18 +0100)]
kernel: fix flow offload UDP handling issue

Only run the TCP state check for TCP connections

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agobusybox: drop providing virtual package ip
Hans Dedecker [Thu, 22 Mar 2018 20:43:04 +0000 (21:43 +0100)]
busybox: drop providing virtual package ip

Drop providing the virtual package ip by busybox which was added in commit
1cec4d4ef0.

Letting busybox provide the virtual package ip is not optimal for the
following reasons :

- Applications depending on ip expect either the ip-full or
  ip-tiny package to be enabled.
- Busybox ip applet cannot be added or removed at runtime

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agofirewall: update to latest git HEAD
Hans Dedecker [Wed, 21 Mar 2018 20:45:56 +0000 (21:45 +0100)]
firewall: update to latest git HEAD

5cdf15e helpers.conf: add CT rtsp helper
d5923f1 Reword rule comments
c1a295a defaults: add support for xt_FLOWOFFLOAD rule
41c2ab5 ipsets: add support for specifying entries

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agoath10k-firmware: Support CT IPQ4019 firmware.
Ben Greear [Sat, 20 Jan 2018 00:27:04 +0000 (16:27 -0800)]
ath10k-firmware: Support CT IPQ4019 firmware.

Initial beta release of the CT IPQ4019 firmware.  Features are
similar to the CT 9984 firmware

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agoath10k-ct: Force loading mac80211 and ath modules.
Ben Greear [Sat, 20 Jan 2018 00:27:03 +0000 (16:27 -0800)]
ath10k-ct: Force loading mac80211 and ath modules.

They are not automatically loaded on IPQ4019 (at least) machines
for some reason.

Signed-off-by: Ben Greear <greearb@candelatech.com>
6 years agoath10k-ct: update to latest version, enable AHB.
Ben Greear [Sat, 20 Jan 2018 00:27:02 +0000 (16:27 -0800)]
ath10k-ct: update to latest version, enable AHB.

The driver updates include:

ath10k driver backport to fix WPA 'pn' related security bugs
(4.13 based driver only currently),
a fix for off-channel TX for CT wave-1 firmware, a likely
fix for napi related crashes, and a backport of the firmware fetch
patch.

AHB is needed for the IPQ4019 platform radios.

Signed-off-by: Ben Greear <greearb@candelatech.com>
[use common subject format]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agointel-microcode: update to 20180312
Zoltan HERPAI [Tue, 20 Mar 2018 13:02:20 +0000 (14:02 +0100)]
intel-microcode: update to 20180312

- Update microcode for 24 CPU types
- Implements IBRS/IBPB/STIPB support, Spectre-v2 mitigation for:
  Sandybridge, Ivy Bridge, Haswell, Broadwell, Skylake, Kaby Lake,
  Coffee Lake
- Missing production updates:
   - Broadwell-E/EX Xeons (sig 0x406f1)
   - Anniedale/Morefield, Apollo Lake, Avoton, Cherry Trail, Braswell,
     Gemini Lake, Denverton
- New Microcodes:
   - sig 0x00050653, pf_mask 0x97, 2018-01-29, rev 0x1000140
   - sig 0x00050665, pf_mask 0x10, 2018-01-22, rev 0xe000009

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agouboot-sunxi: refresh patches
Zoltan HERPAI [Wed, 21 Mar 2018 15:59:46 +0000 (16:59 +0100)]
uboot-sunxi: refresh patches

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agouboot-sunxi: fix build by adding comparabilities for old dtc
Hauke Mehrtens [Wed, 21 Mar 2018 16:08:19 +0000 (17:08 +0100)]
uboot-sunxi: fix build by adding comparabilities for old dtc

We use the dtc from the kernel and that does not have all the options
which u-boot would like to use now. make these parameters optional.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
[wigyori@uid0.hu: renamed to 221-compatible-old-dtc.patch from 220-]
Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agouboot-sunxi: bump again to 2017.11
Zoltan HERPAI [Mon, 5 Feb 2018 16:02:49 +0000 (17:02 +0100)]
uboot-sunxi: bump again to 2017.11

Runtime-tested on:
 - Pine64 (A64)
 - Orange Pi 2 (H3)
 - Bananapro (A20)
 - Olimex A20-Micro (A20)
 - Pcduino v3 (A20)
 - Pcduino v2 (A10)

Compile-tested on:
 - all A8/A7/A53 boards

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agokernel: b53: fix compilation on BCM47XX
Rafał Miłecki [Wed, 21 Mar 2018 09:54:14 +0000 (10:54 +0100)]
kernel: b53: fix compilation on BCM47XX

This fixes:
drivers/net/phy/b53/b53_priv.h:325:2: error: enumeration value '<board>' not handled in switch [-Werror=switch]
errors.

Fixes: 0de2213eeade7 ("kernel: b53: look for NVRAM's "robo_reset" entry on every platform")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agomac80211: backport brcmfmac fixes from 4.16
Rafał Miłecki [Wed, 21 Mar 2018 07:35:58 +0000 (08:35 +0100)]
mac80211: backport brcmfmac fixes from 4.16

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agokernel: b53: look for NVRAM's "robo_reset" entry on every platform
Rafał Miłecki [Tue, 13 Mar 2018 11:09:20 +0000 (12:09 +0100)]
kernel: b53: look for NVRAM's "robo_reset" entry on every platform

Since kernel 4.1 bcm47xx_nvram_gpio_pin() is now defined in a global
header and can be safely called even on non-Broadcom platforms.

This change makes b53 look for "robo_reset" on ARCH_BCM_5301X and
slightly simplifies the code.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
6 years agokirkwood: fix initramfs boot warning on iconnect
Paul Wassi [Fri, 16 Mar 2018 08:29:40 +0000 (09:29 +0100)]
kirkwood: fix initramfs boot warning on iconnect

The chosen dts configuration linux,initrd-* gives an error message
on bootup of kirkwood-iconnect. Since initramfs/initrd is not used
remove these options from the dts.

Reported-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Tested-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Signed-off-by: Paul Wassi <p.wassi@gmx.at>
6 years agokirkwood: switch to kernel 4.14
Paul Wassi [Mon, 12 Mar 2018 22:36:19 +0000 (23:36 +0100)]
kirkwood: switch to kernel 4.14

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
6 years agokirkwood: add kernel 4.14 support
Paul Wassi [Mon, 12 Mar 2018 19:50:48 +0000 (20:50 +0100)]
kirkwood: add kernel 4.14 support

Add patches for 4.14, undoing upstream changes for Linksys devices
regarding DSA. Instead, the switchdev driver marvell,88e6171 is used.

Tested-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Tested-by: Alberto Bursi <alberto.bursi@outlook.it>
Signed-off-by: Paul Wassi <p.wassi@gmx.at>
6 years agouboot-kirkwood: update to 2018.03
Paul Wassi [Tue, 13 Mar 2018 16:25:20 +0000 (17:25 +0100)]
uboot-kirkwood: update to 2018.03

U-Boot now requires GCC > 5

Catch up with upstream and move some configuration options from
the header files to the corresponding defconfig files.
Also move some options of patch 010 affecting the whole platform
to 010's device only.

Signed-off-by: Paul Wassi <p.wassi@gmx.at>
6 years agoopenvpn: remove deprecated config options
Hans Dedecker [Sat, 17 Mar 2018 15:56:30 +0000 (16:56 +0100)]
openvpn: remove deprecated config options

Remove deprecated config options in 2.5 as described in [0]

[0] https://community.openvpn.net/openvpn/wiki/DeprecatedOptions

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agomvebu: disable cpu idle on WRT1900ACv1
Felix Fietkau [Mon, 19 Mar 2018 20:58:12 +0000 (21:58 +0100)]
mvebu: disable cpu idle on WRT1900ACv1

A regression seriously affecting the Linksys WRT1900ACv1 (Mamba) was
introduced some time between the OpenWrt/LEDE v4.4 and v4.9 kernels.
The root cause has not yet been identified, despite many attempts for
more than a year. Disabling the SoC specific CPU idle support should
mitigate this issue.

The symptoms on an affected system are unwanted reboots at a variable
frequency. In many cases almost immediately after boot, causing a
bootloop.  This effectively disables support for Mamba on OpenWrt
with kernels > v4.4.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agomac80211: avoid changing skb truesize in A-MSDU aggregation
Felix Fietkau [Mon, 19 Mar 2018 20:34:25 +0000 (21:34 +0100)]
mac80211: avoid changing skb truesize in A-MSDU aggregation

Should fix recently reported data corruption issues

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoiproute2: fix hidden uint to uin64_t promotion in json_print
Kevin Darbyshire-Bryant [Sun, 18 Mar 2018 15:50:04 +0000 (15:50 +0000)]
iproute2: fix hidden uint to uin64_t promotion in json_print

print_int used 'int' type internally, whereas print_uint used 'uint64_t'

These helper functions eventually call vfprintf(fp, fmt, args) which is
a variable argument list function and is dependent upon 'fmt' containing
correct information about the length of the passed arguments.

Unfortunately print_int v print_uint offered no clue to the programmer
that internally passed ints to print_uint were being promoted to 64bits,
thus the format passed in 'fmt' string vs the actual passed integer
could be different lengths.  This is even more interesting on big endian
architectures where 'vfprintf' would be looking in the middle of an
int64 type.  Symptoms of this included tc qdisc showing bizarre values
for a variety of fields across a variety of qdiscs (e.g. refcnt, flows,
quantum)

print_u/int now stick with native int size.

A similar patch has been sent upstream.

Fixes FS#1425

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agodnsmasq: bump to 2.79 release
Kevin Darbyshire-Bryant [Sun, 18 Mar 2018 22:25:42 +0000 (22:25 +0000)]
dnsmasq: bump to 2.79 release

94b6878 Tidy crypto.c of old library compat. Now need libnettle 3.
8b96552 Fix compiler warning.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
6 years agocurl: Update to 7.59
Rosen Penev [Mon, 19 Mar 2018 00:29:16 +0000 (17:29 -0700)]
curl: Update to 7.59

Compile tested on ar71xx.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agofirmware-utils: mkdlinkfw: compile as gnu99
Mathias Kresin [Mon, 19 Mar 2018 07:10:14 +0000 (08:10 +0100)]
firmware-utils: mkdlinkfw: compile as gnu99

Compile as gnu99 to fix the build.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agolantiq: fix DM200 boot with fake uImage headers
Thomas Nixon [Sat, 3 Mar 2018 15:06:53 +0000 (15:06 +0000)]
lantiq: fix DM200 boot with fake uImage headers

The latest bootloader versions load the firmware into memory and call
`chk_dniimg` (defined in Netgear GPL release), which expects to find
three consecutive block-aligned uImages. Add two fake uImage headers
after the kernel to fool this check.

This wastes up to 128k of space for alignment. The alternative would be
to put the rootfs in a second uImage, but this would limit the firmware
size to 0x710000 (the number of bytes loaded and verified by the
bootloader) instead of 0x7b0000 (the size of the firmware partition).

Signed-off-by: Thomas Nixon <tom@tomn.co.uk>
6 years agolantiq: fix AVM initramfs image
Valentin Spreckels [Sun, 18 Mar 2018 19:27:50 +0000 (20:27 +0100)]
lantiq: fix AVM initramfs image

Create an initramfs that can be used with the EVA bootloader.

Signed-off-by: Valentin Spreckels <Valentin.Spreckels@Informatik.Uni-Oldenburg.DE>
6 years agoar71xx: fix Fritz 300e initramfs image
Mathias Kresin [Sun, 18 Mar 2018 19:25:21 +0000 (20:25 +0100)]
ar71xx: fix Fritz 300e initramfs image

Create an initramfs that can be used with the EVA bootloader.

Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoscripts: add EVA ramboot script
Valentin Spreckels [Fri, 9 Mar 2018 19:17:16 +0000 (20:17 +0100)]
scripts: add EVA ramboot script

Use the EVA bootloader to load a small linux system into the ram and boot
it from there:

  ./scripts/flashing/eva_ramboot.py 192.168.178.1 path/to/initramfs-kernel.bin

Signed-off-by: Valentin Spreckels <Valentin.Spreckels@Informatik.Uni-Oldenburg.DE>
Acked-by: John Crispin <john@phrozen.org>
[reworded commit message]
Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: add support for TP-Link TL-WR902AC v3
Peter Lundkvist [Sat, 17 Mar 2018 17:47:51 +0000 (18:47 +0100)]
ramips: add support for TP-Link TL-WR902AC v3

TP-Link TL-WR902AC v3 is a pocket-size dual-band (AC750) router
based on MediaTek MT7628N + MT7650E.

Specification:
- MediaTek MT7628N/N (580 Mhz)
- 64 MB of RAM
- 8 MB of FLASH
- 2T2R 2.4 GHz and 1T1R 5 GHz
- 1x 10/100 Mbps Ethernet

* MT7650 ac chip isn't not supported by LEDE/OpenWrt at the moment.
  Therefore 5Ghz won' work.

Flash instruction:

The only way to flash LEDE image in TL-WR902AC v3 is to use
tftp recovery mode in U-Boot:

1. Configure PC with static IP 192.168.0.66/24 and tftp server.
2. Rename "openwrt-ramips-mt76x8-tplink_tl-wr902ac-v3-squashfs-tftp-recovery.bin"
   to "tp_recovery.bin" and place it in tftp server directory.
3. Connect PC with the LAN port, press the reset button, power up
   the router and keep button pressed for around 6-7 seconds, until
   device starts downloading the file.
4. Router will download file from server, write it to flash and reboot.

Signed-off-by: Peter Lundkvist <peter.lundkvist@gmail.com>
[drop p2led_an pinmux, this pin isn't used as gpio, fix whitespace issues]
Signed-off-by: Mathias Kresin <dev@kresin.me>
6 years agoramips: add support for D-Link DWR-116-A1/2
Pawel Dembicki [Sat, 20 Jan 2018 08:27:03 +0000 (09:27 +0100)]
ramips: add support for D-Link DWR-116-A1/2

The DWR-116-A1/2 Wireless Router is based on the MT7620N SoC.

Specification:

  MediaTek MT7620N (580 Mhz)
  32 MB of RAM
  8 MB of FLASH
  802.11bgn radio
  5x 10/100 Mbps Ethernet (1 WAN and 4 LAN)
  2x external, non-detachable antennas
  UART (J1 in A1, JP1 in A2) header on PCB (57600 8n1)
  6x LED (GPIO-controlled), 2x button
  JBOOT bootloader

Known issues:
WAN LED is drived by uartl tx pin. I decide to use this pin as
uartlite tx pin.

Installation:
Apply factory image via http web-gui.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
6 years agofirmware: add JBOOT based devices config extractor
Pawel Dembicki [Thu, 1 Mar 2018 19:11:01 +0000 (20:11 +0100)]
firmware: add JBOOT based devices config extractor

Adds tool to extract MAC and pre-calibration data required for JBOOT
based D-Link routers.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
6 years agofirmware-utils: add JBOOT bootloader image support
Pawel Dembicki [Sat, 3 Feb 2018 11:59:50 +0000 (12:59 +0100)]
firmware-utils: add JBOOT bootloader image support

Tested on D-Link DWR-116.
Based on mktplinkfw.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
6 years agoltq-vdsl-mei: Fix section mismatch
Hauke Mehrtens [Sun, 18 Mar 2018 00:33:07 +0000 (01:33 +0100)]
ltq-vdsl-mei: Fix section mismatch

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoltq-deu: Fix section mismatches
Hauke Mehrtens [Sun, 18 Mar 2018 00:32:47 +0000 (01:32 +0100)]
ltq-deu: Fix section mismatches

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoe2fsprogs: fix InstallDev recipe
Jo-Philipp Wich [Sun, 18 Mar 2018 13:04:05 +0000 (14:04 +0100)]
e2fsprogs: fix InstallDev recipe

Create the correct bin directory before staging the host utilities.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
6 years agomtd: fix compile warnings
Hauke Mehrtens [Sat, 17 Mar 2018 23:53:32 +0000 (00:53 +0100)]
mtd: fix compile warnings

This callback should have one parameter less, this parameter is not used
so this was not a so big problem.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agouboot-mxs: fix compile problems related to OpenSSL
Hauke Mehrtens [Sat, 17 Feb 2018 14:38:26 +0000 (15:38 +0100)]
uboot-mxs: fix compile problems related to OpenSSL

Use the UBOOT_MAKE_FLAGS defined in include/u-boot.mk and do not
overwrite them to compile the host tools against the shipped LibreSSL.
In addition add a patch to fix a compile problem when compiling the
tools against LibreSSL caused by differences in the API between OpenSSL
1.1 and LibreSSL.

This should fix the compile problems seen in build bot from time to time
by not depending on the host libssl-dev package any more but using the
LibreSSL version from OpenWrt.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agouboot-mxs: refresh patches
Hauke Mehrtens [Sat, 17 Mar 2018 14:19:10 +0000 (15:19 +0100)]
uboot-mxs: refresh patches

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agomvebu: Disable internal RTC on Linksys devices
Hauke Mehrtens [Sat, 17 Mar 2018 14:52:47 +0000 (15:52 +0100)]
mvebu: Disable internal RTC on Linksys devices

The internal RTC does not work correctly on these Linksys boards based
on Marvell SoCs. It is off by 3 minutes in 10 minutes running, this
was reported by multiple users. On the Linksys Mamba device the device
tree comment says that no crystal is connected to the internal RTC, this
is probably also true for the other devices.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agomvebu: Add Solidrun RTC init patch.
Rosen Penev [Fri, 16 Mar 2018 00:53:27 +0000 (17:53 -0700)]
mvebu: Add Solidrun RTC init patch.

Some boards like the Turris Omnia have an RTC chip that does not get
initialized. Initializing the RTC at the driver level helps get rid of
bootloader hacks that write special register values.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agomvebu: Backport RTC trimming support.
Rosen Penev [Fri, 16 Mar 2018 00:53:26 +0000 (17:53 -0700)]
mvebu: Backport RTC trimming support.

This enables a higher precision mode for the RTC.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agosunxi: update Xunlong Orange Pi Zero Plus dts file
Hauke Mehrtens [Fri, 16 Mar 2018 22:53:52 +0000 (23:53 +0100)]
sunxi: update Xunlong Orange Pi Zero Plus dts file

This updates the DTS file to the version send upstream

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agokernel: bump 4.14 to 4.14.27
Stijn Segers [Thu, 15 Mar 2018 21:00:08 +0000 (22:00 +0100)]
kernel: bump 4.14 to 4.14.27

* Refreshed patches.
* Deleted 812-pci-dwc-fix-enumeration.patch (was accepted upstream)

Compile-tested: ramips/mt7621, x86/64
Run-tested: ramips/mt7621, x86/64

Signed-off-by: Stijn Segers <foss@volatilesystems.org>