In the xrx200_close() function we call napi_disable(), that could
sleep, with priv->hw->chan[i].lock held. This could lead to deadlock
and causes the kernel to complain.
Look at the code I couldn't convince myself about why we
need to protect that specific code part with the lock. IMHO there
seems no reason to protect the refcount variables, because AFAIK
ndo_close() and ndo_open() callbacks are already called with a
semaphore held. Neither I could figure out why napi_disable() have to
be called with that lock held. The only remaining code part for
which I could guess the lock is useful for is ltq_dma_close()
function call.
This patch reduces the lock to the said function call, avoiding the
sleep-with-spinlock-held situation
Signed-off-by: Andrea Merello <andrea.merello@gmail.com>
[fold into 0025-NET-MIPS-lantiq-adds-xrx200-net.patch, backport to
kernel 4.4]
Signed-off-by: Mathias Kresin <dev@kresin.me>
The driver is used for boards outside the lantiq target as well. Move
it to generic to make it available for more targets.
The phy driver is included in kernel 4.8 as INTEL_XWAY_PHY.
Signed-off-by: Mathias Kresin <dev@kresin.me>
The patch 0008 removes the vanilla kernel gptu.c of lantiq.
But the replacement file timer.c does not work Amazon-SE and
is also protected therefore with an CONFIG_SOC_AMAZON_SE :-)
This commit will re-activate the default vanilla kernel code
in gptu.c for Amazon-SE.
Signed-off-by: Tino Reichardt <milky-lede@mcmilk.de>
[refresh patches and add same change for kernel 4.9]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Refresh patches. A number of patches have landed upstream & hence are no
longer required locally:
062-[1-6]-MIPS-* series
042-0004-mtd-bcm47xxpart-fix-parsing-first-block
Reintroduced lantiq/patches-4.4/0050-MIPS-Lantiq-Fix-cascaded-IRQ-setup
as it was incorrectly included upstream thus dropped from LEDE.
As it has now been reverted upstream it needs to be included again for
LEDE.
Run tested ar71xx Archer C7 v2 and lantiq.
Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
[update from 4.4.68 to 4.4.69]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
With the fritz-tools a userspace parser for the Tiny Flash File System
is available and makes the tffs2 kernel patches obsolete.
Signed-off-by: Mathias Kresin <dev@kresin.me>
modify 0152-lantiq-VPE.patch from
commit e0229a16b0 ("lantiq: added support for VPE1")
- remove wdog function declarations from header (asm/vpe.h)
- remove wdog command line init functions (vpe.c)
Signed-off-by: Stefan Koch <stefan.koch10@gmail.com>
- remove patched header (vpe.h) because kernel
already provides one (asm/vpe.h)
- remove module softdog_vpe since watchdog functionality is unused
- remove module softdog_vpe from Makefile
Signed-off-by: Stefan Koch <stefan.koch10@gmail.com>
SMP state is generally affected by
- CONFIG_SMP build-time kernel configuration option and
- 'nosmp' runtime kernel commandline option
The SMP state within vpe-mt.c is determined by CONFIG_SMP option.
A runtime check is needed if VPE functionality
should be used with a kernel image that supports SMP.
This fix introduces a check for 'nosmp' command line option
if CONFIG_SMP kernel configuration option is enabled.
Note: This patch is needed to use lantiq FXS if CONFIG_MIPS_MT_SMP
(that activates CONFIG_SMP) is enabled within kernel configuration
and the 'nosmp' command line argument is given to disable SMP at runtime.
Without this patch CONFIG_MIPS_MT_SMP must be disabled before using FXS.
With this patch setting the 'nosmp' parameter is enough.
In general, concurrent usage of FXS and SMP
is incompatible and will cause kernel panics.
Signed-off-by: Stefan Koch <stefan.koch10@gmail.com>
The two phy operation mode where one phy is assigned to an interface
without lantiq,* device tree property and the other phy is assigned to
an interface with the lantiq,wan device property was broken with the
multicast package leaks between vlans fixes.
Move the multicast packages relevant portmap settings to the condition
which handles multicast packages for better readability.
Replace the priv->port_map based port_map only for the interface which
has the lantiq,switch device tree property set, to allow tagged
multicast packages in two phy mode where the lantiq,switch device tree
property isn't used.
Signed-off-by: Mathias Kresin <dev@kresin.me>
The ethernet driver uses a port map override via special tag to control
the ports on which multicast packets are sent. This was added to work
around an issue in the switch that was occasionally leaking packets onto
the wrong vlan.
Unfortunately the change had some leftover lines that were overwrting
the port map with a list of all ports, thus always leaking packets onto
the wront vlan.
Fix this by only enabling the override with the VLAN port map and only
if a matching VLAN port map was actually found
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Do not put the probe function into the __init section, but use the normal
section. This fixes a section mismatch warning.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Disabling ethernet during reboot (only to enable it again when the
ethernet driver attaches) can put the chip into a faulty state where it
corrupts the header of all incoming packets.
This happens if packets arrive during the time window where the core is
disabled, and it can be easily reproduced by rebooting while sending a
flood ping to the broadcast address.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Since the MIPS IRQ stack patches, lantiq devices were emitting a storm
of messages like this:
[ 567.872172] Spurious IRQ: CAUSE=0x1100c300
Fix this by reworking the IRQ dispatch code
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Prevents crashes when IRQs arrive when the current kernel stack context
already contains deeply nested function calls, e.g. when stacking lots
of network devices on top of each other
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This adds USB initialization fixes for Danube, Amazon SE and xrx300 and
should fix the clock on at least Danube which hopefully closes FS#351.
The xrx200 usb driver now uses more memory for the dwc2 fifos, this was
increased in the chip compared to ar9.
This is based in part on the vendor documentation and the vendor code
base as a reference.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
The DMA controller used only local locks, which will not lock the
section against other CPUs, fix this by using spin locks.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Refresh patches on all 4.4 supported platforms.
077-0005-bgmac-stop-clearing-DMA-receive-control-register-rig.patch
removed as now upstream.
Compile & run tested: ar71xx - Archer C7 v2
Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
ifxhcd never had roothub support but since kernel 3.x it was expected
that a roothub always exists.
The patched fixed a null pointer deref in the usb subsystem because of
the missing roothub.
Since ifxhcd is gone, this whole patch isn't required any longer.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Backport upstream commit a61ae81a1907af1987ad4c77300508327bc48b23.
The actually purpose of the patch was to do some cleanup. As a side
effect of this cleanup, the device node is now passed to nand_dt_init.
This allows to use the common nand device tree properties
- nand-bus-width
- nand-on-flash-bbt
- nand-ecc-mode
- nand-ecc-step-size
- nand-ecc-strength
for the plat_nand driver.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Due to the addition of the bindings to the ath9k driver, this code
isn't used any longer. The fixup is now done by the owloader.
Rename the ath_eep.c file to ath5k_eep.c to indicate that this files
includes ath5k related code only.
Signed-off-by: Mathias Kresin <dev@kresin.me>
This allows specifying PCI devices as children of the PCIe controller
node to pass configuration data to them.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Parameters for dwc2 on lantiq.
A separate dwc2_core_params structure is defined for danube because danube
fifo sizes are large enough to be autodetected. This is not the case on
arx and vrx.
Signed-off-by: Ben Mulvihill <ben.mulvihill@gmail.com>
Refresh patches for all targets that support kernel 4.4.
Compile-tested on all targets that use kernel 4.4 and aren't marked broken.
Runtime-tested on ar71xx, octeon.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
As specified by e0229a16b0 if the VR9
based router provides FXS ports and they should enabled then the
following must added to the kernel command line:
mem=[TOTALMEMSIZE-2M] vpe1_load_addr=ADDRESS vpe1_mem=2M
By adding mem= parameter a pci device stop working correctly. The
pci-lantiq.c module use get_num_physpages() to compute dynamically the
memory amount of the board.
The mem= make the module to compute in the wrong way the BAR11MASK, so
in this situation the mask is misaligned with the dma area that the
hardware expects.
This patch is a port of what legacy ifxmips_pci.c does.
Signed-off-by: Eddi De Pieri <eddi@depieri.net>
This patch bumps the 4.4 kernel from .28 to .30 and refreshes the patches.
Compile-tested on ar71xx, x86/64, ramips/mt7621, brcm47xx and kirkwood.
Run-tested on ar71xx & ramips/mt7621, brcm47xx and kirkwood (last two confirmed
by P. Wassi).
Signed-off-by: Stijn Segers <francesco.borromini@inventati.org>
(required not-distributable firmware blob - dump it by yourself from original firmware)
Signed-off-by: Eddi De Pieri <eddi@depieri.net>
(cherry picked from commit eb0ce57270d0b5b81b224b9336cf54707497eede)
Modified after cherry-pick:
obj in Makefile
Signed-off-by: Stefan Koch <stefan.koch10@gmail.com>
Created minimal patchset based on BB rev 43158 by Eddi De Pieri
14.07/openwrt.git 79472c025449efae9310defad0d3a73cff14d756
If the VR9 based router provides FXS ports and they shoud enabled then
the following must added to the kernel command line:
mem=[TOTALMEMSIZE-2M] vpe1_load_addr=ADDRESS vpe1_mem=2M maxvpes=1
maxtcs=1
To use FXS 2M of RAM are needed for the VPE firmware. The size is set
by vpe1_mem.
The available RAM must be reduced by this size using the mem argument.
A correct load address (example 0x83e00000) for the firmware must be given,
too.
Signed-off-by: Stefan Koch <stefan.koch10@gmail.com>
Read the temperature including the decimale place from the CGU_GPHY1_CR
register.
Decrement the temperature read from the register by 38.0 degree celsius.
The temperature range of the sensor is -38.0 to +154 °C and the register
value 0 is equal to -38.0 °C. This fixes the report of unrealistic
temperatures as seen on all tested boards.
Give the SoC a few milliseconds to get the first temperature value. On
some rare occasions there is no temperature value in the register when
read the first time after activation. This leads to a reported
temperature of -38.0 °C on boot.
Only version 1.2 of the vr9 SoC has a temperature sensor. Add a check
to make sure the driver doesn't load on v1.1 vr9 SoCs.
Signed-off-by: Mathias Kresin <dev@kresin.me>
It missed some changes needed for kernel 4.4. This is only used by the
Falcon SoC and not for the xRX SoCs.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Refresh patches for all targets that support kernel 4.4.
compile/run-tested on brcm2708/bcm2710 only.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Backport patches from upstream Linux kernel which are making the
kernel stores the appended dtb not in the same resisters as defined in
the UHI specification, use a separate variable on MIPS.
Signed-off-by: Daniel Gimpelevich <daniel@gimpelevich.san-francisco.ca.us>
[some modifications]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
With 12fe4b5798 I switched the ath5k
eeprom extraction to an alternate code path. Unfortunately this code
seams to be broken since ages and broke the ath5k EEPROM extraction.
Reported-by: Mohammed Berdai <mohammed.berdai@gmail.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
Acked-by: John Crispin <john@phrozen.org>
Use priv->wan instead of priv->id as indicator if packets should go to the
Ethernet WAN group (DPID=1) or not (DPID=0). This way, it's independant of
interface names or indexes.
Signed-off-by: Martin Schiller <mschiller@tdt.de>
With an own DMA TX channel for each network device (eth0 + eth1) there
won't be any "tx ring full" errors any more.
This patch also move the spinlocks to the channel level instead of locking
the whole xrx200_hw structure.
Signed-off-by: Martin Schiller <mschiller@tdt.de>
The device tree binding and the associated code duplicates functionality
already patched into the etop driver. The compatible string isn't used
any more. Therefore the whole code can be dropped.
The "mac-increment" property allowed to increment a mac address received
via kernel cmdline. This functionality isn't used by any device and
should be added as etop driver device tree property if required again.
Signed-off-by: Mathias Kresin <dev@kresin.me>
All device tree nodes are using the named properties now and the code
path handling the reg property isn't required any more.
The code related to the ath,eep-flash property has been reformatted to
be better readable.
Signed-off-by: Mathias Kresin <dev@kresin.me>
This patch was introduced in commit r16412 for the brcm47xx target only
and then moved to generic in commit r32395. It was initially added
because of ticket #5186 and should fix some problems with fuse file
systems and MIPS caches. The commit comment in r32395 says that this a
generic problem in MIPS CPUs, but does not name any specifics about
that. There was a fix added to kernel 2.6.21 in commit commit
7575a49f20 "[MIPS] Implement flush_anon_page()." that should fix this
problem, but that was already available before both commits were done
to OpenWrt.
I just tested fuse with ntfs.3g without this patch on a BCM4704
(BMIPS3300 V0.6) SoC and haven't seen any problems. Someone reported
that removing this patch improves some fuse operations by 5 times on
some modern MIPS cores.
My test was only a simple "dd if=/dev/zero of=/mnt/zero bs=5000" to an
USB stick.
This patch removes the patch to OpenWrt, because I assume that it is
not needed any more and Felix, the orginal author, also thinks so.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
