The IRQ init structs are marked as __initconst which
means this memory can be free after init.
On this platform, the PCI IRQ init happens very late _after_ the
kernel already freed the memory allocated for these structs.
During IRQ allocation, the allocation function is passed
with invalid data at this point leading to following error:
[ 0.000000] SoC: Qualcomm Atheros QCA9533 ver 2 rev 0
[ 2.382828] Freeing unused kernel memory: 264K
[ 34.414816] pci 0000:00:00.0: no irq found for pin 1
and
[ 0.000000] SoC: Qualcomm Atheros QCA956X ver 1 rev 0
[ 2.125401] Freeing unused kernel memory: 284K
[ 9.526479] pci 0000:00:00.0: no irq found for pin 1
After this patch:
[ 14.960814] pci 0000:00:00.0: using irq 40 for pin 1
Commit 318e19ba67 ("ar71xx: add v4.14 support") fixed this for the
default targets already present in the source by default but forgot
to remove the __initconst attribute for targets QCA953x and QCA956x
which are only added later through platform patches.
Fixes: 318e19ba67 ("ar71xx: add v4.14 support")
Reported-by: Sven Schönhoff <sven.schoenhoff@gmail.com>
Reported-by: Dirk Brenken <dev@brenken.org>
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Tested-by: Dirk Brenken <dev@brenken.org>
CONFIG_SUN4I_A10_CCU controls both the A10 and the A20 enabling of the
CCU (LCCF) driver, this will be necessary once we move beyond kernel
4.14 because 4.15 has commit f18698e1c66338b902de386e4ad97b8b1b9d999d
("ARM: dts: sun7i: Convert to CCU") which requires this driver.
Fixes: ad2b3bf310 ("sunxi: Add support for kernel 4.14")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Revert "Add workaround for wrong skb->mac_len values after splitting GSO"
Remove our local patch which did the same thing.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Kernels 4.14.73 & 4.9.140 include the gso fixup fix, so cake
doesn't need to do it. Let's not waste cpu cycles by doing it in
cake which could be really important on cpu constrained devices.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
This configuration option is not set when building the
layerscape/armv8_64b target.
Fixes: 92aa21497b ("kernel: build support for NFSv4 in nfsd")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Sometimes, the CMakeLists.txt file is not in the root directory of a
repo. In those cases, the CMAKE_SOURCE_SUBDIR variable can be specified
to use CMakeLists.txt from a subdirectory instead.
Signed-off-by: Amol Bhave <ambhave@fb.com>
Config option to limit maximum compression streams per zram dev for
multicore CPU's. This could be defined via 'zram_comp_streams' option in
the 'system' section of '/etc/config/system' file or via cli (for e.x.
with 'uci set system.@System[0].zram_comp_streams=2 && uci commit
system'). Default is number of logical CPU cores.
Signed-off-by: Emil Muratov <gpm@hotplug.ru>
Use only one zram swap device of the specified $size instead of
[N x $size] devices for multicore CPUs Now zram module uses multiple
compression streams for each dev by default, so we do not need to create
several zram devs to utilize multicore CPUs.
Signed-off-by: Emil Muratov <gpm@hotplug.ru>
* "zram stop" could reset up to $(num_of_cores) zram devices even if
some of those were not mounted as swap dev's. This fix tries to
enumerate mounted swap zram dev's before making a reset
* remove hot-added zram devs on stop (except zram0)
Signed-off-by: Emil Muratov <gpm@hotplug.ru>
Compression algorithms for zram are provided by kernel crypto API, could
be any of [lzo|zl4|deflate|<some_more>] depending on kernel modules.
Compress algo for zram-swap could be defined via 'zram_comp_algo' option
in 'system' section of '/etc/config/system' file, or via cli (for e.x.
with 'uci set system.@System[0].zram_comp_algo=lz4 && uci commit
system'). check available algo's via 'cat /sys/block/zram0
/comp_algorithm'
Signed-off-by: Emil Muratov <gpm@hotplug.ru>
Patch 303 is required for Traverse LS1043 targets when using the NXP DPAA1 driver.
The recent refresh of 4.9 patches on layerscape changed how FMan/BMan memory regions
were defined and meant Ethernet stopped working on these boards.
(Note that these definitions are only required for NXP's Ethernet driver, the new
upstream driver in >=4.15 works using the DTS provided in files/)
Signed-off-by: Mathew McBride <matt@traverse.com.au>
On the Netgear WNDAP620, the emac ethernet isn't receiving nor
xmitting any frames from/to the RTL8363SB (identifies itself
as a RTL8367RB).
This is caused by hardware not knowing the right forced link
settings (speed, duplex, pause, etc.) settings.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
The swconfig switch led driver has the ability to switch
between a "link, rx and/or tx" mode. However, this feature
was not implemented in uci, the led init script and
config_generate.
This patch adds a seventh parameter to the
ucidef_set_led_switch() function. The accepted values for
this parameter are: link, rx and tx.
Any permutations of these three values are supported, as
long as they are properly encased with quotes.
If the parameter is not specified it will default to "all"
(link rx tx).
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
This patch fixes some of the dtc warnings:
MR24:
"reg" property in [...]mdio/phy@1 has invalid length (4 bytes)
Node [...]button@1 has a unit name, but no reg property
Relying on default #address-cells value for [...]phy@1
Relying on default #size-cells value for [...]phy@1
MX60(W):
Node [..]nand/all has a reg or ranges property, but no unit name
Node [..]button@1 has a unit name, but no reg property
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
The unification of the My Book Live Single and duo image in
commit 9b47aa93c7 ("apm821xx: unify My Book Live Single + Duo images")
made the wd,mybooklive-duo obsolete in most places and can
be removed.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
