The skb is usually started by a padding which allows the protocols in the
network stack to add their headers in front of the payload. The skb can be
reallocated in case the preallocated padding is not large enough. This can for
example happen in the function __skb_cow which will check the requested extra
headroom and allocate more buffer when the requested headroom is bigger than
the available one. The extra buffer is aligned again to the multiple of the
NET_SKB_PAD of the target architecture.
The macro used to create the multiple of the NET_SKB_PAD is written in a way
which allows only values power two as alignment parameter. The currently used
value of 48 bytes can not be written as n ** 2 but as 2 ** 4 + 2 ** 5. The
extra buffer is therefore not always the multiple of 48 but can be 16, 64, 80,
128, 144 and so on. The generated values are also not monotonic (48 requested
bytes are mapped to 80 allocated bytes and 49 requested bytes are mapped to 64
allocated bytes).
These unexpected small values result in more reallocations of the buffer. This
was noticed prominently during tests between two QCA9558 720 MHz devices which
were connected via ethernet to PCs and had a HT40 802.11n 3x3 link between each
other. The throughput PC-to-PC during iperf TCP runs increased reliable from
186 Mibit/s to 214 Mibit/s in one direction and from 195 Mibit/s to 220 Mibit/s
in the other direction. This is a performance increase of ~14% just by reducing
the amount of reallocations.
Signed-off-by: Sven Eckelmann <sven@open-mesh.com>
SVN-Revision: 37948
Currently, the module causes an oops at least on rt5350. These patches
have been accepted upstream at:
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git
Signed-off-by: Daniel Gimpelevich <daniel@gimpelevich.san-francisco.ca.us>
SVN-Revision: 37931
The 'Warning: unable to open an initial console' message
indicates an error in the rootfs. Remove the patch which
hides the warning.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 37848
Linux expects that the /dev/console node is present
in the rootfs image. Create the node in initramfs,
in order to make std{in,out,err} usable even in early
init process.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 37846
vsnprintf returns the number of chars that would have been written, not
the actual number of chars written. This can lead to crashlog_buf->len
being too big which in turn can lead to get_maxlen() returning negative
numbers. The length argument of kmsg_dump_get_buffer will be casted to
a size_t which makes a negative input a big positive number allowing
kmsg_dump_get_buffer to write out of bounds.
Fix this by using vscnprintf which returns the actually written number
of chars.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
SVN-Revision: 37820
The BCM4331 supports a PCIe max request size of 512 bytes and uses
that, but the PCIe controller in the BCM4706 just supports 128 Bytes
and that causes a DMA error for packages bigger than 126 bytes. This
fixes the problem by setting the BCM4331 also to 128 Bytes.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 37709
There is no platform using the gpio-pwm driver, yet these patches break the
generic PWM framework that is in upstream. So just remove them.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
SVN-Revision: 37490
many boards have a disconnected TTL level serial which can generate
some garbage that can lead to spurious false sysrq detects.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 37210
This patch is a device tree enhancement that IMHO is worthy of mainline.
It allows the bootloader's commandline to be preserved even when the
device tree specifies one.
Signed-off-by: Daniel Gimpelevich <daniel@gimpelevich.san-francisco.ca.us>
SVN-Revision: 36780
zsmalloc was changed to a bool because of missing exports in the kernel,
but we already export the required symbol, so change it back to tristate.
Closes#13481.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 36587
Fixes the following warning:
CC [M] net/ipv6/addrconf.o
net/ipv6/addrconf.c: In function 'addrconf_init':
net/ipv6/addrconf.c:4944:2: warning: assignment from incompatible pointer type [enabled by default]
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 36364
Hauke Mehrtens