openwrt/staging/linusw.git
6 years agomediatek: remove the CPU_SUBTYPE
John Crispin [Mon, 12 Feb 2018 20:02:01 +0000 (21:02 +0100)]
mediatek: remove the CPU_SUBTYPE

Signed-off-by: John Crispin <john@phrozen.org>
6 years agocns3xxx: remove linux 4.9 support
Koen Vandeputte [Thu, 1 Feb 2018 16:50:27 +0000 (17:50 +0100)]
cns3xxx: remove linux 4.9 support

- Remove kernel 4.9 support
- Apply specific 4.14 changes directly to source
- Refreshed all

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
6 years agotarget: Remove nomips16 option.
Rosen Penev [Fri, 2 Feb 2018 22:06:14 +0000 (14:06 -0800)]
target: Remove nomips16 option.

There is no target with a CPU subtype of nomips16.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agotools/mkimage: copy mkenvimage to bin directory
Evgeniy Didin [Mon, 12 Feb 2018 16:42:01 +0000 (19:42 +0300)]
tools/mkimage: copy mkenvimage to bin directory

For some boards u-boot binary environment file is required,
which is generated by mkenvimage utility. But in OpenWrt there
is no separate support for mkenvimage, which is a part of u-boot tools.
mkenvimage gets built in u-boot/tools as well as mkimage anyways.

So lets just copy mkenvimage to the $(STAGING_DIR_HOST)/bin/ directory.

Signed-off-by: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
CC: Alexey Brodkin <abrodkin@synopsys.com>
CC: Hauke Mehrtens <hauke@hauke-m.de>
CC: John Crispin <john@phrozen.org>
6 years agokernel: add missing symbol for v4.14
John Crispin [Mon, 12 Feb 2018 16:38:58 +0000 (17:38 +0100)]
kernel: add missing symbol for v4.14

Signed-off-by: John Crispin <john@phrozen.org>
6 years agoocteontx: add support for OCTEON TX target
Tim Harvey [Tue, 23 Jan 2018 23:15:44 +0000 (15:15 -0800)]
octeontx: add support for OCTEON TX target

The Cavium OCTEON TX is an ARM 64-bit SoC leveraging CPU cores and
periperhals from the Cavium ThunderX SoC.

This initial support provides a 4.14 kernel and kernel+initramfs that is
bootable on the Gateworks Newport GW630x as well as the Cavium sff8104
reference board.

Signed-off-by: Tim Harvey <tharvey@gateworks.com>
6 years agopackages: uboot-mxs: override instead of appending u-boot make flags
Michael Heimpold [Sat, 10 Feb 2018 13:05:17 +0000 (14:05 +0100)]
packages: uboot-mxs: override instead of appending u-boot make flags

This prevents passing down the HOSTCC stuff set in u-boot.mk
which results in linking errors against openssl:

tools/mxsimage.o: In function `sb_aes_reinit':
mxsimage.c:(.text+0x202): undefined reference to `EVP_CIPHER_CTX_reset'
tools/mxsimage.o: In function `mxsimage_generate':
mxsimage.c:(.text+0x110d): undefined reference to `EVP_MD_CTX_new'
mxsimage.c:(.text+0x114f): undefined reference to `EVP_MD_CTX_free'
mxsimage.c:(.text+0x11c3): undefined reference to `EVP_MD_CTX_new'
mxsimage.c:(.text+0x1323): undefined reference to `EVP_MD_CTX_free'
mxsimage.c:(.text+0x134a): undefined reference to `EVP_CIPHER_CTX_reset'
tools/mxsimage.o: In function `mxsimage_verify_print_header':
mxsimage.c:(.text+0x23ce): undefined reference to `EVP_MD_CTX_new'
mxsimage.c:(.text+0x242c): undefined reference to `EVP_MD_CTX_new'
mxsimage.c:(.text+0x246b): undefined reference to `EVP_MD_CTX_free'
mxsimage.c:(.text+0x24ef): undefined reference to `EVP_CIPHER_CTX_reset'
mxsimage.c:(.text+0x2e52): undefined reference to `EVP_MD_CTX_free'
collect2: error: ld returned 1 exit status

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
6 years agopackages: uboot-mxs: bump to 2017.11
Michael Heimpold [Sat, 10 Feb 2018 13:05:10 +0000 (14:05 +0100)]
packages: uboot-mxs: bump to 2017.11

Also update the U-Boot BSP patch for I2SE Duckbill devices.

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
6 years agomxs: rename files to files-4.9
Michael Heimpold [Sat, 10 Feb 2018 13:04:34 +0000 (14:04 +0100)]
mxs: rename files to files-4.9

The Device Tree file for I2SE Duckbill boards was updated upstream.
Let's use the upstream version for upcoming kernel 4.14 by keeping
our current version for v4.9 still around.

Signed-off-by: Michael Heimpold <mhei@heimpold.de>
6 years agoramips: Add support for the GnuBee Personal Cloud Two
Rosen Penev [Mon, 5 Feb 2018 00:24:15 +0000 (16:24 -0800)]
ramips: Add support for the GnuBee Personal Cloud Two

The GnuBee Personal Cloud Two crowdfunded on https://www.crowdsupply.com
It is a low-cost, low-power, network-attached storage device.

Specifications:

- SoC: MediaTek MT7621AT
- RAM: DDR3 512 MB
- Flash: 32 MB
- Six SATA ports for 3.5" Drives
- One SDcard
- One USB 3.0
- Two USB 2.0
- Gigabit Ethernet: Three Ports
- UART 3.5mm Audio Jack or 3 pin header - 57600 8N1
- Three GPIOs available on a pin header

Flash instructions:

The GnuBee Personal Cloud Two ships with libreCMC installed.
libreCMC is a Free Software Foundation approved fork of LEDE/OpenWrt.
As such one can upgrade using the webinterface or sysupgrade.

Das U-Boot has multiple options for recovery or updates including :

- USB
- http
- tftp

Errata:
- While there are three ethernet ports, the third requires support for
  the second GMAC. This will come in kernel 4.14.
- The first hard drive slot has a clearance issue with the two fan
  headers. Workaround is to pull the headers out and connect the pins to
  jumper wires.
- Using this device as a NAS is problematic with the 4.9 kernel as many
  /dev/sdX reads throw silent errors. The current theory behind this is
  some kind of unhandled DMA mapping error in the kernel. This is not an
  issue with kernel 4.4.

Signed-off-by: L. D. Pinney <ldpinney@gmail.com>
Signed-off-by: Rosen Penev <rosenp@gmail.com>
6 years agoiptables: Support building connlabel module
Kristian Evensen [Mon, 22 Jan 2018 17:52:28 +0000 (18:52 +0100)]
iptables: Support building connlabel module

It is currently possible to enable connlabel-support in iptables.
However, in order for connlabel to work properly, the kernel module must
also be present. This patch adds support for building the
connlabel-module, and selects it by default when connlabel-support is
enabled.

Signed-off-by: Kristian Evensen <kristian.evensen@gmail.com>
6 years agolayerscape: fix call trace of pfe module removing
Yangbo Lu [Tue, 30 Jan 2018 12:45:25 +0000 (20:45 +0800)]
layerscape: fix call trace of pfe module removing

Updated pfe kernel patch to clean up iounmap(pfe->ddr_baseaddr).
pfe->ddr_baseaddr was got through phys_to_virt() not ioremap(),
so iounmap() for pfe->ddr_baseaddr should be removed.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: fix DPAA2 QDMA issue
Yangbo Lu [Mon, 29 Jan 2018 10:30:44 +0000 (18:30 +0800)]
layerscape: fix DPAA2 QDMA issue

This patch is to reverse a upstream dmatest patch for now
which is causing DPAA2 QDMA test issue.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: add a make menuconfig known issue in README
Yangbo Lu [Fri, 26 Jan 2018 08:46:17 +0000 (16:46 +0800)]
layerscape: add a make menuconfig known issue in README

This patch is to add a known issue of make menuconfig
in README file. This is also a common issue for OpenWrt.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: fix type for kernel 4.9 dpaa2_mac_get_stats() function
Yangbo Lu [Thu, 25 Jan 2018 08:52:38 +0000 (16:52 +0800)]
layerscape: fix type for kernel 4.9 dpaa2_mac_get_stats() function

This patch changes the declarations of ndo_get_stats64 handlers
to the previous struct rtnl_link_stats64 * return type instead of
the mainline void return.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update README
Yangbo Lu [Mon, 22 Jan 2018 08:42:05 +0000 (16:42 +0800)]
layerscape: update README

Tis patch is to fix up some descriptions in README.
Per-device rootfs had been supported, and a known issue
had been found. These also should be updated accordingly.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: support ubifs rootfs in u-boot env
Yangbo Lu [Mon, 22 Jan 2018 08:33:29 +0000 (16:33 +0800)]
layerscape: support ubifs rootfs in u-boot env

ls1012ardb/ls1012afrdm/ls1046ardb/ls1088ardb firmwares now use ubifs
rootfs. So u-boot env should be set accordingly.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: support ubifs rootfs
Yangbo Lu [Fri, 19 Jan 2018 10:38:03 +0000 (18:38 +0800)]
layerscape: support ubifs rootfs

Some layerscape devices (ls1012ardb/ls1012afrdm/ls1046ardb/ls1088ardb)
were using ext4 rootfs because there were issues using squashfs.
This patch is to drop using ext4 rootfs and use ubifs rootfs instead
which is more proper for SPI-NOR flash.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update u-boot to LSDK1712
Yangbo Lu [Fri, 19 Jan 2018 03:04:25 +0000 (11:04 +0800)]
layerscape: update u-boot to LSDK1712

This patch is to update layerscape u-boot to
NXP LSDK1712 release.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update ppfe firmware to LSDK1712
Yangbo Lu [Thu, 18 Jan 2018 08:36:49 +0000 (16:36 +0800)]
layerscape: update ppfe firmware to LSDK1712

Updated ppfe firmware to NXP LSDK1712 release. Used
ppfe firmware git tree on NXP github since it was
migrated here from qoriq-open-source github.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: add ls-rcw-bin package
Yangbo Lu [Thu, 18 Jan 2018 08:24:24 +0000 (16:24 +0800)]
layerscape: add ls-rcw-bin package

NXP LSDK1712 release used two rcw git trees. The
original rcw git tree was still source code but
dropping ls1012a/ls1088a/ls2088a boards in LSDK1712.
Instead another new rcw git tree was used to just
provided rcw binaries for these boards dropped. So
this patch is to update ls-rcw to LSDK1712 release
and add a new ls-rcw-bin package.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update PPA firmware to LSDK1712
Yangbo Lu [Thu, 18 Jan 2018 07:24:07 +0000 (15:24 +0800)]
layerscape: update PPA firmware to LSDK1712

This patch is to update PPA firmware to NXP LSDK1712
release.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: migrate fman-ucode to NXP github
Yangbo Lu [Thu, 18 Jan 2018 07:08:25 +0000 (15:08 +0800)]
layerscape: migrate fman-ucode to NXP github

fman-ucode had been migrated from qoriq-open-source
github to NXP github. So the Makefile should be fixed
accordingly.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update MC firmware to LSDK1712
Yangbo Lu [Thu, 18 Jan 2018 06:55:19 +0000 (14:55 +0800)]
layerscape: update MC firmware to LSDK1712

Updated MC firmware to NXP LSDK1712 release. Used
MC firmware git tree on NXP github since it was
migrated here from qoriq-open-source github.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: remove USB configs for 64-bit
Yangbo Lu [Thu, 18 Jan 2018 06:14:16 +0000 (14:14 +0800)]
layerscape: remove USB configs for 64-bit

Removed USB configs for 64-bit device since they're
already in kmod package.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: remove USB and TI specific configs for 32-bit
Yangbo Lu [Thu, 18 Jan 2018 06:02:58 +0000 (14:02 +0800)]
layerscape: remove USB and TI specific configs for 32-bit

This patch is to remove USB configs for 32-bit device since
they're already in kmod package. Also removed TI platform
specific configs.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update patches-4.9 to LSDK1712
Yangbo Lu [Tue, 6 Feb 2018 04:39:05 +0000 (12:39 +0800)]
layerscape: update patches-4.9 to LSDK1712

Patches changes
- Updated patches-4.9 to NXP LSDK1712 linux-4.9.
- Merged changes of patch 303 into integrated patch 201.
- Split changes of patch 706 into dpaa part and dpaa2
  part, and merged these changes into integrated patches
  701 and 705.
- Removed patch 819 since ehci-fsl driver could be compiled now.
- Refreshed these patches.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update restool to 2017-12-03
Yangbo Lu [Thu, 11 Jan 2018 09:23:00 +0000 (17:23 +0800)]
layerscape: update restool to 2017-12-03

Updated restool to 2017-12-03 and removed patches
since the new version had involved them.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agolayerscape: update build guide in README
Yangbo Lu [Mon, 8 Jan 2018 04:12:16 +0000 (12:12 +0800)]
layerscape: update build guide in README

Added package feeds updating and installing in build guide
of README.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
6 years agoar71xx: add support for TP-Link RE355
Henryk Heisig [Tue, 6 Feb 2018 21:08:28 +0000 (22:08 +0100)]
ar71xx: add support for TP-Link RE355

This device is identical as TP-Link RE450
RE355 is a dual-band AC1200 router, based on Qualcomm/Atheros
QCA9558+QCA9880.

Specification:

720/600/200 MHz (CPU/DDR/AHB)
64/128 MB of RAM (DDR2)
8 MB of FLASH (SPI NOR)
3T3R 2.4 GHz
3T3R 5 GHz
1x 10/100/1000 Mbps Ethernet
7x LED, 3x button
UART header on PCB

Flash instruction:
Web:
Download lede-ar71xx-generic-archer-c60-v2-squashfs-factory.bin
and use OEM System Tools - Firmware Upgrade site.

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
6 years agokmod-sched: add sch_multiq to extra schedulers
Philip Prindeville [Fri, 26 Jan 2018 21:28:02 +0000 (14:28 -0700)]
kmod-sched: add sch_multiq to extra schedulers

For hardware that supports multiple h/w output queues, add
a compatible scheduler (NET_SCH_MULTIQ).

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agobase-files: don't evaluate block-device uevent
Daniel Golle [Mon, 12 Feb 2018 22:36:54 +0000 (23:36 +0100)]
base-files: don't evaluate block-device uevent

Current code and also before commit da52dd0c83 was vulnerable to shell
injection using volume lables in the GPT partition table of block
devices. Given that partition names can be freely defined in GPT tables
we really shouldn't evaluate a string which is potentially crafted with
evil intentions. Hence rather use `export -n` to absorb the uevent's
variables into the environment.

Fixes commit da52dd0c83 (base-files: quote values when evaluating uevent)
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[mschiffer@universe-factory.net: suggested export -n usage]

6 years agokernel: add IEEE-1284 parallel port support
Daniel Gimpelevich [Thu, 18 Jan 2018 11:52:12 +0000 (03:52 -0800)]
kernel: add IEEE-1284 parallel port support

The kmod-lp package included both lp.ko and ppdev.ko, but ECP device
drivers may or may not require lp NOT to be loaded, needing only ppdev.
Additionally, There were no packages for any parport interface modules,
such as uss720 or parport_pc, provided here. It has not been otherwise
possible to use PC-style parport hardware for kmod-lp.

Signed-off-by: Daniel Gimpelevich <daniel@gimpelevich.san-francisco.ca.us>
6 years agobrcm63xx: switch to kernel 4.9
Jonas Gorski [Mon, 12 Feb 2018 10:44:12 +0000 (11:44 +0100)]
brcm63xx: switch to kernel 4.9

Now that we have support for it, let's switch to kernel 4.9.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agoar7: switch to kernel 4.9
Jonas Gorski [Mon, 12 Feb 2018 10:42:48 +0000 (11:42 +0100)]
ar7: switch to kernel 4.9

With generic run tested and ac49x compile tested, and no complaints so
far, let's switch to 4.9.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agoar7: fix ac49x compilation with 4.9
Jonas Gorski [Mon, 12 Feb 2018 10:40:38 +0000 (11:40 +0100)]
ar7: fix ac49x compilation with 4.9

This will introduce a warning on 3.18, but fixes an error when
compiling for 4.9.

Fixes: cf9e0a59aabd ("ar7: add kernel 4.9 support")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agoRevert "mediatek: activate fpu feature flag"
John Crispin [Mon, 12 Feb 2018 10:35:47 +0000 (11:35 +0100)]
Revert "mediatek: activate fpu feature flag"

This reverts commit 3594447a7d0398c55ea78f658b52b7d8084ae9f9.

This causes the userland to not come up properly

Signed-off-by: John Crispin <john@phrozen.org>
6 years agotools: update patch to v2.7.6
Russell Senior [Mon, 12 Feb 2018 07:40:59 +0000 (23:40 -0800)]
tools: update patch to v2.7.6

Removes a redundant patch with a fix already upstream.  Build tested on
apm821xx (PPC).

Signed-off-by: Russell Senior <russell@personaltelco.net>
6 years agomvebu: fix sysupgrade from 17.04 for clearfog pro
Jonas Gorski [Sun, 11 Feb 2018 21:31:36 +0000 (22:31 +0100)]
mvebu: fix sysupgrade from 17.04 for clearfog pro

When clearfog was renamed to clearfog pro, it broke sysupgrade from
17.04 as the new images now get rejected as incompatible. Fix this by
adding the legacy boardname to the compatible devices.

Fixes: ec4a8c6dee81 ("mvebu: ClearFog renamed upstream to ClearFog Pro")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: add kernel 4.14 support
Jonas Gorski [Sat, 13 Jan 2018 17:43:32 +0000 (18:43 +0100)]
brcm63xx: add kernel 4.14 support

Add support for kernel 4.14. Increases compressed kernel size by ~64k
compared to 4.9.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobroadcom-wl: fix compilation with 4.14
Jonas Gorski [Sat, 13 Jan 2018 17:44:04 +0000 (18:44 +0100)]
broadcom-wl: fix compilation with 4.14

The last_rx field was removed from net_device. Since the field wasn't
used by the generic subsystem, and the driver only writes to it, just
remove the usage.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: add kernel 4.9 support
Jonas Gorski [Sat, 13 Jan 2018 12:18:10 +0000 (13:18 +0100)]
brcm63xx: add kernel 4.9 support

Add support for kernel 4.9 based on the more upstream comformant
partition defintions. Increases compressed kernel size by ~95k
compared to 4.4.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: switch to new partition layout specification
Jonas Gorski [Wed, 28 Jun 2017 08:44:38 +0000 (10:44 +0200)]
brcm63xx: switch to new partition layout specification

Now that we support parsing the expected upstream definition, update
the dts files to it.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: enhance dt partitions support to match upstream more closely
Jonas Gorski [Mon, 26 Jun 2017 17:01:20 +0000 (19:01 +0200)]
brcm63xx: enhance dt partitions support to match upstream more closely

Partitions are supposed to be enclosed in an extra partitions node. Allow
the bcm63xxpart parser to cope with that.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: backport latest bcm63xx_enet changes from upstream
Jonas Gorski [Sun, 17 Dec 2017 13:15:29 +0000 (14:15 +0100)]
brcm63xx: backport latest bcm63xx_enet changes from upstream

No functional changes, just simple cleanups.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agobrcm63xx: fix ephy reset gpio lookup
Jonas Gorski [Sat, 13 Jan 2018 17:41:14 +0000 (18:41 +0100)]
brcm63xx: fix ephy reset gpio lookup

The lookup needs to be called just "reset" for the common code.

Fixes: 054c0d4e3193 ("brcm63xx: backport mdio-bus reset gpio support")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
6 years agoar71xx: add missing config symbol ATH79_MACH_CF_E375AC
Matthias Schiffer [Sun, 11 Feb 2018 17:41:25 +0000 (18:41 +0100)]
ar71xx: add missing config symbol ATH79_MACH_CF_E375AC

Fixes: bf66bb8c96 "ar71xx:add support for COMFAST CF-E375AC"
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
6 years agowrt55agv2-spidevs: remove obsolete package
Felix Fietkau [Sun, 11 Feb 2018 17:05:14 +0000 (18:05 +0100)]
wrt55agv2-spidevs: remove obsolete package

It hasn't compiled in a long time and needs to be updated before it can
be restored anyway.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agomediatek: activate fpu feature flag
Hauke Mehrtens [Fri, 9 Feb 2018 23:52:20 +0000 (00:52 +0100)]
mediatek: activate fpu feature flag

The arm CPUs uses in the supported Mediatket SoCs have a FPU accordingly
to the datasheet, activate it also. The CPU subtype "neon-vfpv4" is
selected, but the toolcahin generated for this SoC will still be
compiled with soft float and not with the hard float ABI as we haven't
the fpu feature flag set. If this toolchain is reused by other targets
this will even affect other targets.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agobase-files: quote values when evaluating uevent
Daniel Golle [Fri, 2 Feb 2018 00:57:46 +0000 (01:57 +0100)]
base-files: quote values when evaluating uevent

When sourcing /sys/class/block/*/uevent values have to be quoted as
they may contain spaces (e.g. in PARTNAME).
Fix this by pre-processing with sed before sourcing.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
6 years agoipq-wifi: drop OpenMesh A42 board-2.bin
Sven Eckelmann [Mon, 29 Jan 2018 10:52:51 +0000 (11:52 +0100)]
ipq-wifi: drop OpenMesh A42 board-2.bin

The BDFs for OpenMesh A42 were upstreamed [1] to the ath10k-firmware
repository and are now part of ath10k-firmware 2018-01-26. The
ipq-wifi-openmesh_a42 package can now be dropped because OpenWrt already
ships the QCA4019 board-2.bin from this version.

[1] https://wireless.wiki.kernel.org/en/users/drivers/ath10k/boardfiles

Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
6 years agofirmware: ath10k-firmware: update to 2018-01-26
Sven Eckelmann [Mon, 29 Jan 2018 10:46:35 +0000 (11:46 +0100)]
firmware: ath10k-firmware: update to 2018-01-26

* introduces the BDFs for the OpenMesh A42 in
  /lib/firmware/ath10k/QCA4019/hw1.0/board-2.bin.
* adds new firmware firmware-6.bin_RM.4.4.1.c1-00037-QCARMSWP-1 for
  QCA6174 hw3.0

Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
6 years agoar71xx:add support for COMFAST CF-E375AC
Ding Tengfei [Wed, 7 Feb 2018 03:47:52 +0000 (11:47 +0800)]
ar71xx:add support for COMFAST CF-E375AC

COMFAST CF-E375AC is a ceiling mount AP with PoE support, based on
Qualcomm/Atheros QCA9563 + QCA9886 + QCA8337.

Short specification:

    2x 10/100/1000 Mbps Ethernet, with PoE support
    128MB of RAM (DDR2)
    16 MB of FLASH
    3T3R 2.4 GHz, 802.11b/g/n
    2T2R 5 GHz, 802.11ac/n/a, wave 2
    built-in 5x 3 dBi antennas
    output power (max): 500 mW (27 dBm)
    1x RGB LED, 1x button
    built-in watchdog chipset

Flash instruction:

Original firmware is based on OpenWrt.
Use sysupgrade image directly in vendor GUI.

Signed-off-by: Ding Tengfei <dtf@comfast.cn>
6 years agoar71xx: remove bs-partition ro-flag for UniFi AC
David Bauer [Tue, 6 Feb 2018 18:44:36 +0000 (19:44 +0100)]
ar71xx: remove bs-partition ro-flag for UniFi AC

This removes the read-only flag from the bs (bootselect) partition
on UniFi AC devices. This allows to correct the indicator from which
partition the device is booting its kernel from.

See also:
 - https://github.com/freifunk-gluon/gluon/issues/1301
 - https://bugs.lede-project.org/index.php?do=details&task_id=662

Signed-off-by: David Bauer <mail@david-bauer.net>
6 years agokexec-tools: bump version to 2.0.16
Philip Prindeville [Sat, 3 Feb 2018 01:50:24 +0000 (18:50 -0700)]
kexec-tools: bump version to 2.0.16

All patches have been integrated upstream.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agokexec-tools: issue warning when dd'ing vmcore
Philip Prindeville [Sat, 16 Sep 2017 16:16:29 +0000 (10:16 -0600)]
kexec-tools: issue warning when dd'ing vmcore

With no warning, it just looks like the box has hung during boot.

We don't want users resetting it without having captured a crashdump.

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
6 years agoramips: add support for TP-Link Archer C50 v3
Henryk Heisig [Sun, 28 Jan 2018 18:49:38 +0000 (19:49 +0100)]
ramips: add support for TP-Link Archer C50 v3

TP-Link Archer C50 v3 is a router with 5-port FE switch and
non-detachable antennas. It's based on MediaTek MT7628N+MT7612E.

Specification:
- MediaTek MT7628N/N (580 Mhz)
- 64 MB of RAM
- 8 MB of FLASH
- 2T2R 2.4 GHz and 2T2R 5 GHz
- 5x 10/100 Mbps Ethernet
- 4x external, non-detachable antennas
- UART (J1) header on PCB (115200 8n1)
- 7x LED (GPIO-controlled*), 2x button, power switch

* WAN LED in this devices is a dual-color, dual-leads type which isn't
  (fully) supported by gpio-leds driver. This type of LED requires both
  GPIOs state change at the same time to select color or turn it off.
  For now, we support/use only the green part of the LED.

Flash instruction:

The only way to flash LEDE image in ArcherC50v3 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-mt7628-ArcherC50v3-squashfs-tftp-recovery.bin"
   to "tp_recovery.bin" and place it in tftp server directory.
3. Connect PC with one of LAN ports, press the reset button, power up
   the router and keep button pressed for around 6-7 seconds, until
   device starts downloading the file.
4. Router will download file from server, write it to flash and reboot.

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
6 years agoramips: ArcherC20v4: use common TP-Link dtsi file
Henryk Heisig [Sun, 28 Jan 2018 18:35:17 +0000 (19:35 +0100)]
ramips: ArcherC20v4: use common TP-Link dtsi file

and remove common part from dts file.

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
6 years agoramips: change the TP-Link dtsi filename
Henryk Heisig [Sun, 28 Jan 2018 18:09:51 +0000 (19:09 +0100)]
ramips: change the TP-Link dtsi filename

Signed-off-by: Henryk Heisig <hyniu@o2.pl>
6 years agokernel: mtdsplit: Add support for D-link JBOOT
Pawel Dembicki [Sun, 4 Feb 2018 12:11:16 +0000 (13:11 +0100)]
kernel: mtdsplit: Add support for D-link JBOOT

The D-Link devices with JBOOT bootloader use their own kernel
image header (stag + sch2 headers).

This driver find jImage header and set rootfs start after kernel file.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
6 years agoipq806x: mark AVM FB4040 switchport as untagged
David Bauer [Tue, 6 Feb 2018 20:00:53 +0000 (21:00 +0100)]
ipq806x: mark AVM FB4040 switchport as untagged

This commit marks the CPUs switchport explicit as untagged.
Otherwise, an eth0.1 interface is created and the devices
LAN-ports are not working.

Signed-off-by: David Bauer <mail@david-bauer.net>
6 years agox86: add preinit hook to reload microcode
Zoltan HERPAI [Sat, 10 Feb 2018 18:16:37 +0000 (19:16 +0100)]
x86: add preinit hook to reload microcode

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agofirmware: add microcode package for Intel
Zoltan HERPAI [Sat, 10 Feb 2018 20:19:41 +0000 (21:19 +0100)]
firmware: add microcode package for Intel

Compiling the Intel microcode package results in a
microcode.bin and a microcode-64.bin. As we can
decide based on the subtarget which should be used,
we'll only split the required .bin file with
iucode-tool.

x64 will get the intel-microcode-64.bin
All other variants will get intel-microcode.bin

The microcodes will be updated from preinit via a common
script - that's the earliest place where we can do it.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agofirmware: add microcode package for AMD
Zoltan HERPAI [Sat, 10 Feb 2018 18:06:56 +0000 (19:06 +0100)]
firmware: add microcode package for AMD

Use the Debian repository for sourcing the ucode files.

Current (20171205) includes support for fam17h CPUs already.

The microcodes will be updated from preinit via a common
script - that's the earliest place where we can do it.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agotools: add iucode-tool
Zoltan HERPAI [Sun, 11 Feb 2018 11:30:07 +0000 (12:30 +0100)]
tools: add iucode-tool

Add tool to "compile" Intel microcode files. The tool will be
compiled for host (to split the microcode.dat) and for target
(to forcibly reload the microcode if required).

Instead of using the large microcode.bin/microcode-64.bin, the
splitted ucode files (separate for CPU families) will be
installed.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agox86: enable microcode loading for Intel and AMD
Zoltan HERPAI [Wed, 17 Jan 2018 08:58:26 +0000 (09:58 +0100)]
x86: enable microcode loading for Intel and AMD

Enable for 4.9 and 4.14.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agofstools: update to latest git HEAD
John Crispin [Sun, 11 Feb 2018 13:32:04 +0000 (14:32 +0100)]
fstools: update to latest git HEAD

3d23981 strip trailing spaces from vfat labels
c4a3c97 fix vfat volume label
5010710 block: support /dev/nvme* nodes

Signed-off-by: John Crispin <john@phrozen.org>
6 years agouboot-fritz4040: Fix build with HOSTCFLAGS
Hauke Mehrtens [Sat, 10 Feb 2018 23:52:08 +0000 (00:52 +0100)]
uboot-fritz4040: Fix build with HOSTCFLAGS

When we provide the HOSTCFLAGS to the U-Boot build it will fail because
it can not find the u-boot provided header files any more.
Just overwrite and not append the package specific configuration on top
of the configuration provided by u-boot.mk.
uboot-fritz4040 is based on U-Boot 2012.07 and this problem is probably
similar to the problem seen with the lantiq and ar71xx u-boot build.

Fixes: df9781a4205 ("u-boot,at91bootstrap: fix incorrect HOSTCPPFLAGS variable")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoar71xx: Orders the names of the devices alphabetically.
Arne Zachlod [Sat, 10 Feb 2018 22:39:50 +0000 (23:39 +0100)]
ar71xx: Orders the names of the devices alphabetically.

Signed-off-by: Arne Zachlod <arne@nerdkeller.org>
6 years agoimx6: use DTS_DIR at image build code
Adrià Llaudet [Thu, 8 Feb 2018 11:02:24 +0000 (12:02 +0100)]
imx6: use DTS_DIR at image build code

Use "$(DTS_DIR)", defined at include/image.mk, instead of
"$(LINUX_DIR)/arch/$(LINUX_KARCH)/boot/dts" in order to generalize and
allow a better Device/* device-tree parameterization (i.e. DEVICE_DTS_DIR
and DTS_DIR).

Signed-off-by: Adrià Llaudet <adria.llaudet@gmail.com>
6 years agoarchs38: bump kernel to 4.14.
Evgeniy Didin [Tue, 23 Jan 2018 16:48:48 +0000 (19:48 +0300)]
archs38: bump kernel to 4.14.

Update Linux kernel version from 4.9 to 4.14 for archs38.
config-4.14 was simply regenerated with "make kernel_menuconfig".

Signed-off-by: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: John Crispin <john@phrozen.org>
6 years agokernel: backport fix undefined abort
Evgeniy Didin [Tue, 30 Jan 2018 12:36:51 +0000 (15:36 +0300)]
kernel: backport fix undefined abort

While building mpi.ko module with stable Linux v4.14.14 an error occured:
>ERROR: "abort" [lib/mpi/mpi.ko] undefined!
In upstream Linux 4.15 this issue is fixed:
Commit 7c2c11b208be ("arch: define weak abort()")
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7c2c11b208be09c156573fc0076b7b3646e05219

Commit dc8635b78cd8 ("kernel/exit.c: export abort() to modules")
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dc8635b78cd8669c37e230058d18c33af7451ab1

So lets add backport patches until these fixes
are not applied in stable version.

Signed-off-by: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
CC: John Crispin <john@phrozen.org>
CC: Hauke Mehrtens <hauke@hauke-m.de>
6 years agokernel: add kmod-fou
Filip Moc [Thu, 1 Feb 2018 19:42:30 +0000 (20:42 +0100)]
kernel: add kmod-fou

Once installed fou kernel module allows you to use FOU (Foo over UDP)
and GUE (Generic UDP encapsulation) tunnel protocols.

To get ip fou command working you also need to install ip-full.

Signed-off-by: Filip Moc <lede@moc6.cz>
6 years agouboot-mvebu: fix build ; use the build's tools/libressl
Alexandru Ardelean [Tue, 23 Jan 2018 14:42:09 +0000 (16:42 +0200)]
uboot-mvebu: fix build ; use the build's tools/libressl

Since I have no openssl-dev on my machine, I first
get this error:

```
tools/kwbimage.c:21:10: fatal error: openssl/bn.h: No such file or directory
 #include <openssl/bn.h>
```

After removing the UBOOT_MAKE_FLAGS the next error is:
```
tools/kwbimage.c:40:6: error: conflicting types for ‘EVP_MD_CTX_cleanup’
 void EVP_MD_CTX_cleanup(EVP_MD_CTX *ctx)
```

After removing the OpenSSL patches the next error is:
```
  HOSTLD  tools/dumpimage
/usr/bin/ld: cannot find -lssl
/usr/bin/ld: cannot find -lcrypto
collect2: error: ld returned 1 exit status
scripts/Makefile.host:108: recipe for target 'tools/dumpimage' failed
make[5]: *** [tools/dumpimage] Error 1

```

So, the final part is to add the build system's
HOST_LDFLAGS to the UBOOT_MAKE_FLAGS.
(which was done in the previous commit)

Then the image builds.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
6 years agou-boot.mk: add HOST_LDFLAGS to UBOOT_MAKE_FLAGS
Alexandru Ardelean [Tue, 23 Jan 2018 14:51:52 +0000 (16:51 +0200)]
u-boot.mk: add HOST_LDFLAGS to UBOOT_MAKE_FLAGS

This will make sure that the build system's
paths for linking are available.
This is needed mostly for linking with tools/libressl.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
6 years agou-boot,at91bootstrap: fix incorrect HOSTCPPFLAGS variable
Alexandru Ardelean [Mon, 22 Jan 2018 15:30:06 +0000 (17:30 +0200)]
u-boot,at91bootstrap: fix incorrect HOSTCPPFLAGS variable

This would should up as `$$(HOSTCPPFLAGS)` in the host CFLAGS.
```
make --jobserver-fds=3,4 -j -C <openwrt>/build_dir/target-arm_cortex-a8+vfpv3_musl_eabi/u-boot-A10-OLinuXino-Lime/u-boot-2017.07 CROSS_COMPILE=arm-openwrt-linux-muslgnueabi- DTC="<openwrt>/build_dir/target-arm_cortex-a8+vfpv3_musl_eabi/linux-sunxi_cortexa8/linux-4.9.76/scripts/dtc/dtc" HOSTCC="gcc" HOSTCFLAGS='-O2 -I<openwrt>/staging_dir/host/include -I<openwrt>/staging_dir/host/usr/include -I<openwrt>/staging_dir/hostpkg/include -I<openwrt>/staging_dir/target-arm_cortex-a8+vfpv3_musl_eabi/host/include $$(HOSTCPPFLAGS)' HOSTLDFLAGS="" BL31=<openwrt>/staging_dir/target-arm_cortex-a8+vfpv3_musl_eabi/image/bl31.bin
```

And then it would complain with:
```
 /bin/sh: 1: HOSTCPPFLAGS: not found
```

Also, HOSTCPPFLAGS does not exist.
The correct var is HOST_CPPFLAGS.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
6 years agovalgrind: Fix compile on ARM64
Hauke Mehrtens [Sat, 10 Feb 2018 19:11:22 +0000 (20:11 +0100)]
valgrind: Fix compile on ARM64

Activate the support for 64 bit on all 64 bit CPUs and not only x86_64.
ARM64 does not provide an xml file, so do not pack any.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agobunwind: build for ARM64
Hauke Mehrtens [Mon, 5 Feb 2018 21:43:02 +0000 (22:43 +0100)]
bunwind: build for ARM64

ARM64 is supported by libunwind since some versions, allow building it
for aarch64.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agoconfig: fix ARM64 dependency check
Hauke Mehrtens [Mon, 5 Feb 2018 21:40:56 +0000 (22:40 +0100)]
config: fix ARM64 dependency check

The ARM64 CPUs use aarch64 config symbol, fix the depends lines.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agotoolchain: gdb: activate on ARM64
Hauke Mehrtens [Mon, 5 Feb 2018 21:39:58 +0000 (22:39 +0100)]
toolchain: gdb: activate on ARM64

gdb supports ARM64 since quite some time now, activate it there like on
other CPU architectures.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agokernel: rename CONFIG_TRACE_ENUM_MAP_FILE to CONFIG_TRACE_EVAL_MAP_FILE
Hauke Mehrtens [Fri, 9 Feb 2018 23:41:56 +0000 (00:41 +0100)]
kernel: rename CONFIG_TRACE_ENUM_MAP_FILE to CONFIG_TRACE_EVAL_MAP_FILE

This config option was renamed in upstream Linux commit 681bec0367
("tracing: Rename update the enum_map file")

Reported-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agouboot-sunxi: fix build by adding comparabilities for old dtc
Hauke Mehrtens [Fri, 9 Feb 2018 22:53:47 +0000 (23:53 +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>
6 years agouboot-sunxi: refresh patches
Hauke Mehrtens [Fri, 9 Feb 2018 22:58:06 +0000 (23:58 +0100)]
uboot-sunxi: refresh patches

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
6 years agomt76: update to the latest version, fixes mt7603 stability issues
Felix Fietkau [Fri, 9 Feb 2018 15:04:30 +0000 (16:04 +0100)]
mt76: update to the latest version, fixes mt7603 stability issues

3413961 mt7603: avoid reordering qos-null data packets
c60e6db mt76: toggle driver station powersave bit before notifying mac80211
246d548 mt76: stop tx queues from the driver callback instead of common code

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agokernel: remove kmod-spi-gpio-old
Felix Fietkau [Wed, 7 Feb 2018 13:28:47 +0000 (14:28 +0100)]
kernel: remove kmod-spi-gpio-old

It is unused and has been deprecated for a long time

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agowrt55agv2-spidevs: mark as broken
Felix Fietkau [Wed, 7 Feb 2018 13:27:52 +0000 (14:27 +0100)]
wrt55agv2-spidevs: mark as broken

The target it was meant for was updated to a version that this does not
compile with. It probably also hasn't been used in years

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoixp4xx: remove linux 4.4 support
Felix Fietkau [Wed, 7 Feb 2018 13:26:03 +0000 (14:26 +0100)]
ixp4xx: remove linux 4.4 support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
6 years agoocteon: switch to kernel 4.14
Stijn Tintel [Thu, 8 Feb 2018 18:04:38 +0000 (19:04 +0100)]
octeon: switch to kernel 4.14

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agokernel: bump 4.14 to 4.14.18
Stijn Tintel [Thu, 8 Feb 2018 08:36:38 +0000 (09:36 +0100)]
kernel: bump 4.14 to 4.14.18

Refresh patches.

Remove upstreamed patches:
- apm821xx/010-crypto-gcm-add-GCM-IV-size-constant.patch
- backport/040-crypto-fix-typo-in-KPP-dependency-of-CRYPTO_ECDH.patch
Remove pending-4.14/650-pppoe_header_pad.patch, it is superseded by
upstream commit d32e5740001972c1bb193dd60af02721d047a17e.
Update patch that no longer applies: hack/204-module_strip.patch

Compile-tested: octeon, x86/64.
Runtime-tested: octeon, x86/64.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
6 years agoodhcp6c: fix appending of emtpy sendopt value (FS#1336)
Hans Dedecker [Thu, 8 Feb 2018 17:21:12 +0000 (18:21 +0100)]
odhcp6c: fix appending of emtpy sendopt value (FS#1336)

Don't append an empty sendopts value as odhcp6c bails out
immediately on an empty -x option triggering an infinite start
loop of odhcp6c

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agolibubox: bump to git HEAD version
Hans Dedecker [Thu, 8 Feb 2018 12:12:22 +0000 (13:12 +0100)]
libubox: bump to git HEAD version

b0c830 sh/jshn.sh: add json_for_each_item()

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agoodhcp6c: change sendopts option into list
Hans Dedecker [Mon, 5 Feb 2018 08:57:48 +0000 (09:57 +0100)]
odhcp6c: change sendopts option into list

Commit a26045049b added support for sendopts as a string; since multiple
sendopts values can be specified it makes more sense to model it as a
list of strings.

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agonetifd: update to latest git HEAD
Hans Dedecker [Mon, 5 Feb 2018 20:56:16 +0000 (21:56 +0100)]
netifd: update to latest git HEAD

1be329c netifd-proto: add proto_config_add_array wrapper

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
6 years agotools/ccache: update to 3.3.6
Hannu Nyman [Sun, 4 Feb 2018 18:56:59 +0000 (20:56 +0200)]
tools/ccache: update to 3.3.6

Update ccache to 3.3.6

Release notes:
https://ccache.samba.org/releasenotes.html#_ccache_3_3_6

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
6 years agouboot-sunxi: bump to 2017.11
Zoltan HERPAI [Mon, 5 Feb 2018 16:02:49 +0000 (17:02 +0100)]
uboot-sunxi: bump 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 agoramips: various fixes for zbt-we1226
Daniel Golle [Wed, 24 Jan 2018 00:27:51 +0000 (01:27 +0100)]
ramips: various fixes for zbt-we1226

Convert userspace code to use generic device-tree compatible board
detection method.  Users of the existing code will have to use
sysupgrade -F once to switch to the new generic board naming.
Properly setup pinctrl fixing the switch port LEDs.

Fixes commit 9c4fe103cb (ramips: add support for ZBT-WE1226)
Reported-by: Mathias Kresin <dev@kresin.me>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
6 years agoltq-atm: cleanup unused variables and functions
Martin Schiller [Mon, 8 Jan 2018 08:54:59 +0000 (09:54 +0100)]
ltq-atm: cleanup unused variables and functions

Signed-off-by: Martin Schiller <ms@dev.tdt.de>
6 years agoltq-atm: rewrite tx path to use IRQs
Alexander Couzens [Mon, 8 Jan 2018 08:54:58 +0000 (09:54 +0100)]
ltq-atm: rewrite tx path to use IRQs

The ATM subsystem is different from the generic ethernet NICs. The ATM
subsystem requires a callback when a packet has been sent. It means a
tx skb_buff need to be used after it has sent. While the generic NIC
can fill up the TX ring and free skb_buffs if it encounter a ring buffer slot
with an already sent skbuff.
The ATM drivers need call the pop() function after it has send a
single ATM package. The ATM subsystem controls via this ways the queuing.

The ppe engine use DMA channels for read and write. Every atm_vcc has it's
own TX DMA channel and each TX DMA channel has it's own ring buffer.

The old driver had multiple issues:
- Call the subsystem callback at the beginning of tx function (ppe_send).
  Didn't allowed the ATM subsystem to control the enqueued package
  amount.
- Filled up the TX ring until full and fail futher
- copy or decouple the skb from all other subsystem before giving it
  over to TX ring

The new tx path uses interupts.
- call the subsystem callback _after_ it was sent by hardware
- no need to copy our decouple the skb any more
- gives back control to the atm subsystem over the enqueued packages
- use an interupt for every sent atm package

Using interupts shouldn't be a problem because of the slow uplink bandwidth of
ADSL.
The speed _through_ the DSL router was always as high as it should
be, only traffic generated on the router itself were affected.

After changing to new tx path, the speed of iperf's run on the
router itself reached the same speed. The master/trunk wasn't as much
affected because of TCP optimisations (reboot-5022-gb2ea46fe236a).
The following results are taken on the remote server, which receives
the stream over the internet and the DSL line.

The sync moves between every sync a litte bit, but is so far stable
Latency / Interleave Delay:               Down: Fast (0.25 ms) / Up: Fast (0.50 ms)
Data Rate:                                Down: 13.287 Mb/s / Up: 1.151 Mb/s

reboot-5521-g9f8d28285d without patch
[ ID] Interval           Transfer     Bandwidth       Retr
[  5]   0.00-10.04  sec   947 KBytes   773 Kbits/sec    0             sender
[  5]   0.00-10.04  sec   928 KBytes   757 Kbits/sec                  receiver

reboot-5521-g9f8d28285d with patch
[  5]   0.00-10.06  sec  1.16 MBytes   970 Kbits/sec    0             sender
[  5]   0.00-10.06  sec  1.15 MBytes   959 Kbits/sec                  receiver

v17.01.4-239-g55c23e44f4 without patch
[ ID] Interval           Transfer     Bandwidth       Retr
[  5]   0.00-10.04  sec  87.4 KBytes  71.3 Kbits/sec    0             sender
[  5]   0.00-10.04  sec  59.6 KBytes  48.7 Kbits/sec                  receiver

v17.01.4-239-g55c23e44f4 with patch
[ ID] Interval           Transfer     Bandwidth       Retr
[  5]   0.00-10.05  sec  1.18 MBytes   983 Kbits/sec    1             sender
[  5]   0.00-10.05  sec  1.15 MBytes   959 Kbits/sec                  receiver

Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
6 years agoarm-trusted-firmware-sunxi: use release build and bump version
Zoltan HERPAI [Mon, 5 Feb 2018 11:50:52 +0000 (12:50 +0100)]
arm-trusted-firmware-sunxi: use release build and bump version

 - use release build instead of debug to reduce size
 - bump to use latest commit in allwinner tree

Tested on Pine64/1G.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
6 years agomt76: update to the latest version
Felix Fietkau [Thu, 25 Jan 2018 15:49:14 +0000 (16:49 +0100)]
mt76: update to the latest version

2b7fae4 mt76: fix returnvar.cocci warnings
939e3e0 mt76x2: dfs: avoid tasklet scheduling during mt76x2_dfs_init_params()
cf59170 mt76x2: dfs: add set_domain handler
5e4d60e mt76x2: dfs: take into account dfs region in mt76x2_dfs_init_params()
f76e25f mt76x2: fix WMM parameter configuration
34d612d mt76: retry rx polling as long as there is budget left
0f8327a mt76x2: fix TSF value in probe responses
ad3f8e9 mt76: add an intermediate struct for rx status information
58a41f1 mt76: get station pointer by wcid and pass it to mac80211
b0508d3 mt76: implement A-MPDU rx reordering in the driver code
cf3cfc4 mt76: split mt76_rx_complete
461cdf9 mt76: pass the per-vif wcid to the core for multicast rx
9b2c778 mt76: validate rx CCMP PN
302af90 mt76x2: init: disable all pending tasklets during device removal
9f685fe mt7603: init: disable tbtt tasklet during device removal
c6f8cac mt76: let mac80211 validate CCMP PN for fragmented frames
3968dae mt7603: fix 40 mhz channel bandwidth reporting
9c2e03d mt7603: fix rx LDPC reporting
f515dfc mt76: implement AP_LINK_PS
974142c mt76: implement processing of BlockAckReq frames
c5209db mt76: avoid re-queueing A-MPDU rx reorder work if no frames are pending
e67e7a5 mt76x2: do not set status->aggr for NULL data frames
8693864 mt76: check qos ack policy before reordering packets

Signed-off-by: Felix Fietkau <nbd@nbd.name>