Without patch unloading the dwc3-of-simple module went stuck after
successfully removing hcd.1 during the hcd.0 removal:
root@LEDE:/# rmmod dwc3-of-simple
[ 21.391846] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 21.391931] usb usb4: USB disconnect, device number 1
[ 21.397038] xhci-hcd xhci-hcd.1.auto: USB bus 4 deregistered
[ 21.401111] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 21.406685] usb usb3: USB disconnect, device number 1
[ 21.412848] xhci-hcd xhci-hcd.1.auto: USB bus 3 deregistered
[ 21.417248] xhci-hcd xhci-hcd.0.auto: remove, state 1
[ 21.422521] usb usb2: USB disconnect, device number 1
followed by nothing.
Sometimes a stall CPU was detected, or a kernel panic,
or a reboot occurred after a couple of minutes.
At the same time unloading the dwc3 module followed by dwc3-of-simple
module was working repeatedly.
root@LEDE:/# rmmod dwc3
[ 53.827328] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 53.827412] usb usb4: USB disconnect, device number 1
[ 53.832630] xhci-hcd xhci-hcd.1.auto: USB bus 4 deregistered
[ 53.836452] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 53.842314] usb usb3: USB disconnect, device number 1
[ 53.848412] xhci-hcd xhci-hcd.1.auto: USB bus 3 deregistered
[ 53.852542] xhci-hcd xhci-hcd.0.auto: remove, state 1
[ 53.857882] usb usb2: USB disconnect, device number 1
[ 53.863956] xhci-hcd xhci-hcd.0.auto: USB bus 2 deregistered
[ 53.867875] xhci-hcd xhci-hcd.0.auto: remove, state 1
[ 53.873696] usb usb1: USB disconnect, device number 1
[ 53.879742] xhci-hcd xhci-hcd.0.auto: USB bus 1 deregistered
root@LEDE:/# rmmod dwc3-of-simple
root@LEDE:/#
For the non-working case, the code was stuck in a readl() in
http://lxr.free-electrons.com/source/drivers/usb/host/xhci.c#L91
because
http://lxr.free-electrons.com/source/drivers/usb/dwc3/dwc3-of-simple.c#L126
was disabling the wrong clocks when removing hcd.1 (it was disabling
the clock of hcd.0). That's why the readl() went stuck when removing
hcd.0
The patch however addresses the clock assignment from the Netgear R7500
dts file and backs off the previous attempt.
Now unloading and repeated module loading is working just fine.
root@LEDE:/# rmmod dwc3-of-simple
[ 24.089679] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 24.089765] usb usb4: USB disconnect, device number 1
[ 24.094856] xhci-hcd xhci-hcd.1.auto: USB bus 4 deregistered
[ 24.098963] xhci-hcd xhci-hcd.1.auto: remove, state 1
[ 24.104522] usb usb3: USB disconnect, device number 1
[ 24.111194] xhci-hcd xhci-hcd.1.auto: USB bus 3 deregistered
[ 24.115086] xhci-hcd xhci-hcd.0.auto: remove, state 1
[ 24.120396] usb usb2: USB disconnect, device number 1
[ 24.126503] xhci-hcd xhci-hcd.0.auto: USB bus 2 deregistered
[ 24.130347] xhci-hcd xhci-hcd.0.auto: remove, state 1
[ 24.135948] usb usb1: USB disconnect, device number 1
[ 24.142085] xhci-hcd xhci-hcd.0.auto: USB bus 1 deregistered
root@LEDE:/#
Fixes: dwc3-of-simple module unloading for Netgear R7500
Signed-off-by: Thomas Reifferscheid <thomas@reifferscheid.org>
The patch follows the qualcomm code comments setting
SSUSB_CTRL_TEST_POWERDOWN to 0x1 and is testing and clearing the
bit during USB superspeed PHY init. According to Andy Gross it
needs to be BIT(26).
Signed-off-by: Thomas Reifferscheid <thomas@reifferscheid.org>
Acked-by: Andy Gross <andy.gross@linaro.org>
We need to ensure there is enough headroom to push extra header,
but we also need to check if we are allowed to change headers.
skb_cow_head() is the proper helper to deal with this.
Fixes Ethernet<->WiFi bridge for Raspberry Pi and probably other devices.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
The script can be used to quickly spin up QEMU virtual machines with lan
and wan network. Please read the initial part of the script for
instructions about how to configure host machine
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Original armvirt target is now subtarget 32. Other than that the built
result should remain the same as before
Besides old features already available with arm64, the new armvirt/64
subtarget will also have those features originally enabled for
armvirt/32 with commit 44ecfc2 ("armvirt: new target")
- pl011, uart
- pl031, rtc
- pl061, gpio
- pci-host-generic
- virtio_{mmio,pci,net,blk,scsi,9p,console,balloon}
- smp with NR_CPUS=4
- cpu-hotplug
- ext4
- DEBUG_BUGVERBOSE for debug purposes
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Not sure since when the issue emerged, but according to the current doc of gcc
and as, armv8-a is intended as argument of -march
The change will affect at the moment arm64 and layerscape/64b
Below is the relevant error messages when building toolchain
Assembler messages:
Error: unknown cpu `armv8-a'
Error: unrecognized option -mcpu=armv8-a
/home/yousong/git-repo/lede-project/lede/build_dir/toolchain-aarch64_armv8-a_gcc-5.4.0_musl/gcc-5.4.0/libgcc/libgcc2.c:1:0: error: unknown value 'armv8-a' for -mcpu
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
The dep for the nftables support was wrong, if someone actually enable
that option gain a compilation error. This fix this problem.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Fix a '==' that should be a '=' in a test condition. Busybox fortunately
doesn't care.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Protect any IPv6 related with appropriate guards to fix compilation with
disabled IPv6 support in Busybox.
Fixes#728.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Add red:internet led on gpio 30 previously claimed as missing.
Wifi led was claimed as blue however there are no blue leds on the
board at all. Actually there are two wifi leds, green & amber so add
definitions for those. Make the newly discovered green wifi led gpio
14 the default.
There is no amber power led on this board. The mention of the amber
power led in the manual means power red and power green at the same
time.
Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
[removed unnecessary aliases, add migration script]
Signed-off-by: Mathias Kresin <dev@kresin.me>
Use fixed led names and add each board variant instead of manipulating
the board name.
It makes the ramips board name function less different to the one used
in other targets and allows to merge them with a common function.
Signed-off-by: Mathias Kresin <dev@kresin.me>
We need to keep the former used (unmodified) boardname in the metadata.
Otherwise an upgrade from an board using the old boardname will be
refused.
Fixes: a75ce960ac ("ramips: use different board names for variants")
Signed-off-by: Mathias Kresin <dev@kresin.me>
PSG1218 got only 4 Ethernet ports and WAN on port 3 while
PSG1218K2C got 5 Ethernet ports and WAN on port 4
Switch to use kmod-kt76x2 instead of kmod-mt76 for both devices while
at it.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Use the stateless 4-byte op codes for this flash chip to fix reboot
hangs on SoCs expecting the flash chip in 3-byte mode.
Fixes: FS#179
Signed-off-by: Mathias Kresin <dev@kresin.me>
update RF register 47 and 54 values according to vendor driver
Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[daniel@makrotopia.org: moved changes into a separate patch]
Use register values from init LNA function instead of the ones from
restore LNA function. Apply register values based on rx path
configuration.
Signed-off-by: Tomislav Požega <pozega.tomislav@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[daniel@makrotopia.org: moved changes into a separate patch]
Make brcmfmac depend on !TARGET_uml.
Technically, brcmfmac could be built for uml because only SDIO support
won't work on that target. However, selectively avoiding the dependency
propagation of !TARGET_uml from kmod-mmc to avoid including a reference
to BRCMFMAC_SDIO doesn't work.
In practice, brcmfmac is completely useless on uml, so let's just
disable it there.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
minor/cosmetic: fixes the following misleading message:
root@box:~ /etc/init.d/dnsmasq restart
sh: out of range
Signed-off-by: Bastian Bittorf <bb@npl.de>
This reverts commit 51397d7d95.
There are some unresolved random crashes on WRT1900AC v1 that still need
to be sorted out
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This fixes:
Thermal sensors on bcm2835 SoC (BCM2835_THERMAL) [N/m/y/?] (NEW) aborted!
This symbol depends on ARCH_BCM2835 so it should be OK to disable it at
brcm2708 target only.
Fixes: 7d32caa61fb0 ("brcm2708: backport upstream sdhost controller driver")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Since 192f0a3db8 ("ath9k: unset the default LED pin if used by
platform leds") the default ath9k wireless LED is not set as soon as
any pin of the ath ath9k gpio controller is used.
All touched boards have leds defined which are using the gpio pins
exposed by the ath9k driver but rely on a default set wireless led
trigger.
Add the wireless leds were missing and setup the wireless phy trigger
in userspace.
Signed-off-by: Russell Senior <russell@personaltelco.net>
Signed-off-by: Mathias Kresin <dev@kresin.me>
The Netgear R6220 requires the kmod-usb3 package and misses
kmod-usb-ledtrig-usbport package to setup the configured usb led
trigger.
Drop the already target selected kmod-mt76.
Fixes: FS#686
Fixes: 38bee61dab ("ramips: add support for Netgear R6220")
Signed-off-by: Mathias Kresin <dev@kresin.me>
Fix the PCIe 5GHz wireless by using the on flash eeprom/caldata.
Disable the 2.4GHz band as this band has no antennas attached but is
enabled in the eeprom/caldata.
Fixes: FS#691
Signed-off-by: Mathias Kresin <dev@kresin.me>
Fix the PCIe 5GHz wireless by using the ralink mtd-eeprom property as
this board have a RT5592 and uses the rt2x00 driver. The mediathek
device tree bindings do not work here.
Fixes: FS#691
Fixes: d8dd207ea6 ("ramips: use the ralink,mtd-eeprom device tree property")
Signed-off-by: Mathias Kresin <dev@kresin.me>
The factory image has an uImage header covering the entire image and
not only the kernel. The original uImage header which covers only the
kernel is appended to the end of the image.
During LEDE boot the uImage rootfs splitter skips the whole filesystem,
can't find a valid filesystem magic and panics.
The last known working version was OpenWrt 14.07, which simply kept on
searching for an uImage header if the first found didn't resulted into
a working rootfs. This behaviour is kind of error prone since it could
produce false positives.
Since the sysupgrade image works fine in combination with the tftp
recovery for doing the initial installation of LEDE, simply drop the
factory image.
Related: FS#462
Signed-off-by: Mathias Kresin <dev@kresin.me>
Since the LEDE nslookup applet is already specific to LEDE, move the
libresolv detection into the busybox Makefile that LEDE uses.
This fixes builds with external toolchains that don't automatically
search for headers and/or libraries without being told so.
Fixes: de5b8e5d2f ("busybox: add musl compatible nslookup replacement")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Apply the !UML dependency to both the symbol and the DEPENDS so there is
no recursive dependency anymore.
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
The original implementation loaded the count register with (wrong) semi-
random values due to its implemenation nature.
If the wrongly calulated value was below the kickrate,
the WD was triggered and rebooted the system.
Rework this, partly based on upstream patches, to dynamically fetch the
current clockrate and calculate the proper offset for the WD countdown
register.
Before:
[ 143.800000] count val: 27219720
[ 148.820000] count val: 50623201
[ 153.830000] count val: 96425250
[ 158.830000] count val: 89735401
[ 163.840000] count val: 4756110
After:
[ 0.700000] MPCore WD init. clockrate: 299984500 prescaler: 256
countrate: 1171814 timeout: 60s
[ 358.530000] count val: 35154751
[ 363.540000] count val: 35154750
[ 368.540000] count val: 35154751
[ 373.550000] count val: 35154750
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
binutils 2.27 checks if the target compiler supports -Wstack-
usage=262144, and also uses this setting for the host compiler. If the
host compiler is gcc < 4.9 binutils build will fail. This backports 2
commits which are fixing this problem for binutils 2.28.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This was also working before, with a slightly
different semantic.
[ Original semantic ]
If no reload hooks was implemented, the default one would
kick in, it would return fail, and restart would happen.
This would happen also in the case where a reload hook
would be implemented, it would fail, and it would restart
the service.
[ New semantic ]
The default reload hook calls restart.
Services can implement their own reload.
If reload fails, then the '/etc/init.d/<service> reload'
would return a non-zero code, and the caller can choose
a way to handle this.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
John Crispin
Thomas Reifferscheid
Álvaro Fernández Rojas
Yousong Zhou
Ansuel Smith
Matthias Schiffer
Daniel Golle
Jo-Philipp Wich
Kevin Darbyshire-Bryant
Mathias Kresin
Rafał Miłecki
Tomislav Požega
Felix Fietkau
Bastian Bittorf
Florian Fainelli
Jonas Gorski
Koen Vandeputte
Alif M. Ahmad
Matthias Fritzsche
Hauke Mehrtens
Alexandru Ardelean
Pavel Kubelun