* socket: mark skbs as not on list when receiving via gro
Certain drivers will pass gro skbs to udp, at which point the udp driver
simply iterates through them and passes them off to encap_rcv, which is
where we pick up. At the moment, we're not attempting to coalesce these
into bundles, but we also don't want to wind up having cascaded lists of
skbs treated separately. The right behavior here, then, is to just mark
each incoming one as not on a list. This can be seen in practice, for
example, with Qualcomm's rmnet_perf driver. This lead to crashes on
OnePlus devices and possibly other Qualcomm 4.14 devices. But I fear
that it could lead to issues on other drivers on weird OpenWRT routers.
This commit is upstream in net-next as:
https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=736775d06bac60d7a353e405398b48b2bd8b1e54
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
TP-Link Archer C20 v5 is a router with 5-port FE switch and
non-detachable antennas. It's based on MediaTek MT7628N+MT7610EN.
Specification:
- MediaTek MT7628N/N (580 Mhz)
- 64 MB of RAM
- 8 MB of FLASH
- 2T2R 2.4 GHz and 1T1R 5 GHz
- 5x 10/100 Mbps Ethernet
- 3x external, non-detachable antennas
- UART (J1) header on PCB (115200 8n1)
- 7x LED (GPIO-controlled*), 2x button, power input 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.
Create Factory image
--------------------
As all installation methods require a U-Boot to be integrated into the
Image (and we do not ship one with the image) we are not able to create
an image in the OpenWRT build-process.
Download a TP-Link image from their Website and a OpenWRT sysupgrade
image for the device and build yourself a factory image like following:
TP-Link image: tpl.bin
OpenWRT sysupgrade image: owrt.bin
> dd if=tpl.bin of=boot.bin bs=131584 count=1
> cat owrt.bin >> boot.bin
Installing via Web-UI
---------------------
Upload the boot.bin via TP-Links firmware upgrade tool in the
web-interface.
Installing via Recovery
-----------------------
Activate Web-Recovery by beginning the upgrade Process with a
Firmware-Image from TP-Link. After starting the Firmware Upgrade,
wait ~3 seconds (When update status is switching to 0%), then
disconnect the power supply from the device. Upgrade flag (which
activates Web-Recovery) is written before the OS-image is touched and
removed after write is succesfull, so this procedure should be safe.
Plug the power back in. It will come up in Recovery-Mode on 192.168.0.1.
When active, all LEDs but the WPS LED are off.
Remeber to assign yourself a static IP-address as DHCP is not active in
this mode.
The boot.bin can now be uploaded and flashed using the web-recovery.
Installing via TFTP
-------------------
Prepare an image like following (Filenames from factory image steps
apply here)
> dd if=/dev/zero of=tp_recovery.bin bs=196608 count=1
> dd if=tpl.bin of=tmp.bin bs=131584 count=1
> dd if=tmp.bin of=boot.bin bs=512 skip=1
> cat boot.bin >> tp_recovery.bin
> cat owrt.bin >> tp_recovery.bin
Place tp_recovery.bin in root directory of TFTP server and listen on
192.168.0.66/24.
Connect router LAN ports with your computer and power up the router
while pressing the reset button. The router will download the image via
tftp and after ~1 Minute reboot into OpenWRT.
U-Boot CLI
----------
U-Boot CLI can be activated by holding down '4' on bootup.
Dual U-Boot
-----------
This is TP-Link MediaTek device with a split-uboot feature design like
a TP-Link Archer C50 v4. The first (factory-uboot) provides recovery via
TFTP and HTTP, jumping straight into the second (firmware-uboot) if no
recovery needs to be performed. The firmware-uboot unpacks and executed
the kernel.
Web-Recovery
------------
TP-Link integrated a new Web-Recovery like the one on the Archer C7v4 /
TL-WR1043v5 / Archer C50v4. Stock-firmware sets a flag in the "romfile"
partition before beginning to write and removes it afterwards. If the
router boots with this flag set, bootloader will automatically start
Web-recovery and listens on 192.168.0.1. This way, the vendor-firmware
or an OpenWRT factory image can be written.
By doing the same while performing sysupgrade, we can take advantage of
the Web-recovery in OpenWRT.
It is important to note that Web-Recovery is only based on this flag. It
can't detect e.g. a crashing kernel or other means. Once activated it
won't boot the OS before a recovery action (either via TFTP or HTTP) is
performed. This recovery-mode is indicated by an illuminated WPS-LED on
boot.
Signed-off-by: Maxim Anisimov <maxim.anisimov.ua@gmail.com>
[adjust some node names for LEDs in DTS]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
- add "gpio" group for wan_orange led
- use tpt triggers for wifi led indication
- add wifi 5 GHz led support
Signed-off-by: Maxim Anisimov <maxim.anisimov.ua@gmail.com>
[slight commit message adjustment]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
When running sysupgrade from an existing configuration, UCI option
dhcp.@dnsmasq[0].resolvfile needs to be modified in case it has not
been changed from it's original value.
Accomplish that using a uci-defaults script.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Image builds for the ramips-mt7621 target currently fail with:
> WARNING: Image file ./hiwifi_hc5962-kernel.bin is too big
Disable this board for now. It can still be built using the SDK.
Signed-off-by: David Bauer <mail@david-bauer.net>
Without this commit, ft_psk_generate_local is enabled for non-PSK
networks by default. This breaks 802.11r for EAP networks.
Disable ft_psk_generate_local by default for non-PSK networks resolves
this misbehavior.
Reported-by: Martin Weinelt <martin@darmstadt.freifunk.net>
Signed-off-by: David Bauer <mail@david-bauer.net>
Tested-by: Martin Weinelt <martin@darmstadt.freifunk.net>
This target is still on kernel 4.9, and it looks like there is no
active maintainer for this target anymore.
Remove the code and all the packages which are only used by this target.
To add this target to OpenWrt again port it to a recent and supported
kernel version.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This target seems to have been unmaintained for quite a while, and not a
single tester for the (now outdated) kernel 4.14 patches has been found.
Remove the code and all the packages which are only used by this target.
To add this target to OpenWrt again port it to a recent and supported
kernel version.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This target seems to have been unmaintained for quite a while, and not a
single tester for the (now outdated) kernel 4.14 patches has been found.
Remove the code and all the packages which are only used by this target.
To add this target to OpenWrt again port it to a recent and supported
kernel version.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This adds support for kernel 4.14 to the target and directly make it the
default kernel version to use.
This patch is build-tested only, but has never been device-tested. It is
only added to preserve the changes in Git history prior to removing this
target. Use it with care.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
[rebased and extended commit message, refreshed patches for 4.14.162]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The configuration of the generic subtarget was used as the default
configuration and then the subtarget configurations were adapted.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This adds support for kernel 4.14 to the target and directly make it the
default kernel version to use.
This patch is build-tested only, but has never been device-tested. It is
only added to preserve the changes in Git history prior to removing this
target. Use it with care.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
[rebased and extended commit message, refreshed patches for 4.14.162]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The configuration of the generic subtarget was used as the default
configuration and then the subtarget configurations were adapted.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
The device label shows the address currently assigned to the OpenWrt
LAN interface.
Current setup is:
LAN *:b8 factory 0xe006 label
WAN *:b7 factory 0xe000
For vendor FW bootlog we get (manually removed parts of the address):
[ 7.520000] set LAN/WAN LWLLL
[ 7.530000] GMAC1_MAC_ADRH -- : 0x00004031
[ 7.530000] GMAC1_MAC_ADRL -- : 0x3c****b7
[ 7.530000] GDMA2_MAC_ADRH -- : 0x00004031
[ 7.540000] GDMA2_MAC_ADRL -- : 0x3c****b8
[ 7.540000] eth1: ===> VirtualIF_open
Without further information, this does not allow verification of
the currently unexpected LAN/WAN assignment (we would expect 0xe000
to be LAN).
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This move the slightly different target-specific implementations of
mktplinkfw from the targets to include/image-commands.mk and renames
it to tplink-v1-image. Having a common version will increase
consistency between implementation and will complete the
tplink build command already present in the new location.
Due to the slight differences of the original implementations, this
also does some adjustments to the device build commands/variables.
This also moves rootfs_align as this is required as dependency.
Tested on:
- TL-WDR4300 v1 (ath79, factory)
- TL-WDR4900 v1 (mpc85xx, sysupgrade)
- RE210 v1 (ramips, see Tested-by)
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Tested-by: Christoph Krapp <achterin@googlemail.com>
Refactor l2 freq scaling patch to support voltage
scaling and add support to base cache scaling on
cpu freq scaling. Update the dtsi files with the new
definition used in the new code.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
It's wrong set the mux to bias-disable. The best way to
do this is by creating a separate group and disable the
specific pins.
By documentation, any subgroup with no bias definition
is ignored so the mux definition is useless.
Rework the definition by sremoving the mux subgroup and
set the remaining subgroup with the mux function and
drive-strength
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
The opp_notifier introduced with patch 0054-Handle-OPP-voltage adjust
for some reason missed the actual registration in the opp struct,
resulting in never being registred.
This was present in kernel 4.9 patchset but dropped in the transition
to 4.14. Reintroduce this and fix patch 0055 about L2 cache scaling.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Previous commit should have bumped PKG_RELEASE, but git add was
forgotten... Add it now.
Fixes: cd48d8d342 ("dnsmasq: switch to /tmp/resolv.conf.d/resolv.conf.auto")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Mount-bind directory instead of resolv.conf.auto file in jail to
avoid problems when the file is deleted/replaced.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This is a backport from the dnsmasq master which should fix a bug which
could cause a crash in dnsmasq.
I saw the following crashes in my log:
[522413.117215] do_page_fault(): sending SIGSEGV to dnsmasq for invalid read access from 2a001450
[522413.124464] epc = 004197f1 in dnsmasq[400000+23000]
[522413.129459] ra = 004197ef in dnsmasq[400000+23000]
This is happening in blockdata_write() when block->next is
dereferenced, but I am not sure if this is related to this problem or if
this is a different problem. I am unable to reproduce this problem.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
ipTIME A8004T is a 2.4/5GHz band AC2600 router, based on Mediatek
MT7621A.
Specifications:
- SoC: MT7621A
- RAM: DDR3 256M
- Flash: SPI NOR 16MB
- WiFi:
- 2.4GHz: MT7615E
- 5GHz: MT7615E
- Ethernet: 5x 10/100/1000Mbps
- Switch: SoC internal
- USB: 1 * USB3.0 port
- UART:
- J4: 3.3V, TX, RX, GND (3.3V is the square pad) / 57600 8N1
- Other info:
- J9: Unknown unpopulated header.
Installation via web interface:
1. Flash **initramfs** image through the stock web interface.
2. Boot into OpenWrt and perform sysupgrade with sysupgrade image.
Revert to stock firmware:
1. Perform sysupgrade with stock image.
Signed-off-by: Yong-hyu Ban <perillamint@quendi.moe>
[do not enable xhci node in DTS which is already enabled in DTSI]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The MAC address on the label of this device corresponds to the
2.4 GHz and ethernet MAC address.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Add call to fclose for file pointer fp in function t_openpw.
The resource leak could happen during an error handling.
Signed-off-by: Andrea Dalla Costa <andrea@dallacosta.me>
Currently this device fails to boot with the OpenWrt snapshot images
(release images are unaffected). The error message is:
"LZMA ERROR 1 - must RESET board to recover".
This happens because the kernel image is too big for the bootloader
to boot. This commit works around this by decreasing the lzma dictionary
size option from the default 23 to 10.
Before this change the current OpenWrt snapshot image (uncompressed
kernel size 4875139 bytes) failed to boot, while now an even bigger
image (kernel 4.19 with snapshot default config; uncompressed kernel
size 5162833 bytes) boots just fine.
The highest lzma dictionary size option this image booted with was 11.
10 was chosen to have a bit more room for growth.
An unavoidable side-effect of this change is that the compressed kernel
image will take up more space.
Total image size with different dictionary size options:
D23 - 3973903 bytes (base)
D16 - 4113167 bytes (+3.5% - +139264 bytes)
D12 - 4317967 bytes (+8.7% - +344064 bytes)
D11 - 4383503 bytes (+10.3% - +409600 bytes)
D10 - 4461327 bytes (+12.3% - +487424 bytes)
Fixes: FS#1484
Signed-off-by: Mason Clarke <mclarke2355@gmail.com>
78cd384 Update README and patchlevel.h for 2.4.8 release
5d03403 pppd: Avoid use of strnlen (and strlen) in vslprintf
a1e950a pppd: Fix IPv6 default route code for Solaris
ca5e61b plugins/rp-pppoe: Make tag parsing loop condition more accurate
c10c3c7 pppd: Make sure word read from options file is null-terminated
b311e98 pppd: Limit memory accessed by string formats with max length specified
3ea9de9 pppd: Eliminate some more compiler warnings
57edb1a pppd: Include time.h header before using time_t
09f695f pppd: Don't free static string
03104ba pppd.h: Add missing headers
388597e pppd: Add defaultroute6 and related options
66ce4ba pppd: Avoid declarations within statements in main.c
5637180 pppd: Fix `ifname` option in case of multilink (#105)
d00f8a0 pppd: Fix variable reference syntax in Makefile.linux
b6b4d28 pppd: Check tdb pointer before closing
Signed-off-by: DENG Qingfang <dengqf6@mail2.sysu.edu.cn>
If CONFIG_DROPBEAR_ZLIB is set, building fails at the packaging stage
due to an undeclared dependency on libz.so.1.
As is already done for the main dropbear package, conditionally add a
dependency on zlib.
Signed-off-by: Matt Merhar <mattmerhar@protonmail.com>
Both devices are available in 64M and 128M RAM configurations but there
is no visial indication which configuration one might get.
So just to be sure we properly support both configurations switch to
kmod-atk10k-ct-smallbuffers.
Signed-off-by: Christoph Krapp <achterin@googlemail.com>
It seems the buildbots can't handle it.
Added a cmake option to find the cxxabi files as they are part of the
toolchain and not in the normal path. It doesn't seem to make a
difference, just gets rid of cmake warnings.
Added another small GCC warning fix. It's fairly minor.
This has no change in compiled size, and most likely no change in
behavior. Bumped the PKG_RELEASE anyway.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
