commit 8efa88540635 (sch_sfq: avoid giving spurious NET_XMIT_CN signals)
forgot to call qdisc_tree_decrease_qlen() to signal upper levels that a
packet (from another flow) was dropped, leading to various problems.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@29642 3c298f89-4303-0410-b956-a3cf2f4a3e73
This fixes the issues with the jffs2 images on various boards. Using
JFFS2 on devices with 4KiB erase sectors is not safe:
http://lists.infradead.org/pipermail/linux-mtd/2011-June/036591.html
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@29521 3c298f89-4303-0410-b956-a3cf2f4a3e73
To make the ioctl number "stable", use an arbitrary high number to prevent
conflicts with new mtd ioctls that would push MTDREFRESH's number.
Also make mtd use the in-kernel mtd headers.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@28893 3c298f89-4303-0410-b956-a3cf2f4a3e73
Add support for Ed Wildgoose's alix platform driver plus my
enhancement to add GPIO soft button support.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@28081 3c298f89-4303-0410-b956-a3cf2f4a3e73
on some platforms, erase suspend leads to data corruption and lockups when write
ops collide with erase ops. this has been observed on the buffalo wzr-hp-g300nh.
rather than play whack-a-mole with a hard to reproduce issue on a variety of devices,
simply disable erase suspend, as it will usually not produce any useful gain on
the small filesystems used on embedded hardware.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27984 3c298f89-4303-0410-b956-a3cf2f4a3e73
This is breaking some devices out there such as Winchiphead CH341 adapters (#9601)
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27885 3c298f89-4303-0410-b956-a3cf2f4a3e73
Fixes problem with TFM allocation in cryptosoft.c
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
Hauke:
* remove ubsec_ssb package and take it from ocf-linux
* use patches from ocf-linux package
* refresh all patches
* readd some build fixes for OpenWrt.
* readd CRYPTO_MANAGER dependency
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27753 3c298f89-4303-0410-b956-a3cf2f4a3e73
* add new patches for bcm4716 SoC
* add support for serial flash on bcma bus
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27723 3c298f89-4303-0410-b956-a3cf2f4a3e73
Since 2.6.38 the bridge module has a dependency to IPv6 if IPv6 is
enabled. Since the IPv6 module isn't exactly lightweight and bridge also
only needs a single function from IPv6, it's rather easy to create a
common "lib" module with a RCU pointer to the actual implementation, if
the IPv6 module is loaded (although slightly hackish).
The codepath seems to be only taken when using IPv6, so there should be
no negative side effects when IPv6 isn't loaded. I did not measure how
big the performance impact is.
Signed-off-by: Jonas Gorski <jonas.gorski+openwrt@gmail.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27237 3c298f89-4303-0410-b956-a3cf2f4a3e73
* backport 2.6.8 patches to .39 / .32.33
* remove lqtapi
* bump tapi/dsl to .39
* migrate to new ltq_ style api
* add amazon_se support
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27026 3c298f89-4303-0410-b956-a3cf2f4a3e73
It was originally tristate, but was changed to bool to prevent a build
breakage in an earlier patch revision (for a new syscall). The actual
commit[1] changed it in a way this isn't needed anymore, but
accidentially left the EXPORTFS change in. There's no reason for it
not being buildable as a module, so changing it back to tristate is
fine.
see: https://patchwork.kernel.org/patch/812832/
Thank you Jonas Gorski for the patch.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27011 3c298f89-4303-0410-b956-a3cf2f4a3e73
* add md5sum for kernel
* add some missing config symbols
* add some ssb patches to make compat-wireless compile
Thank you Peter Wagner for the patch
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27009 3c298f89-4303-0410-b956-a3cf2f4a3e73
This patch adds 802.1Q VLAN support for the ADM6996M chip.
The driver is loaded for both the FC and M model. It will detect which of the
two chips is connected. The FC model is initialised, but no further
functionality is offered.
The PHY driver will always report "100 Mbit/s, link up", for both the M and FC
models. This reflects the fact that the link between switch chip and Ethernet
MAC is always on[1].
Further documentation can be found in the kernel's
Documentation/networking/adm6996.txt
Signed-of-By: Peter Lebbing <peter@digitalbrains.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26865 3c298f89-4303-0410-b956-a3cf2f4a3e73
Use offset from struct mtd_part instead of struct mtd_partition for defining
the new partitions. This enables the usage of MTDPART_OFS_APPEND macro in
struct mtd_partition.
Signed-off-by: Jochen Friedrich <jochen@scram.de>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26705 3c298f89-4303-0410-b956-a3cf2f4a3e73
Add support for 2.6.39 based on rc3. Runtime tested on bcm63xx.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26615 3c298f89-4303-0410-b956-a3cf2f4a3e73
