Dual image capable CFEs store an image sequence at the same place as
currently OpenWrt stores the actual rootfs length, so it will get
overwritten when flashing through such a CFE.
To prevent this from happening, move the rootfs length field to the next
four bytes, thus completely using the reserved1 field.
Since the reserved1 field is now completely in use, it does not make sense
to allow it to be set from the imagetag utility, so remove the option.
Signed-off-by: Jonas Gorski <jonas.gorski+openwrt@gmail.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26680 3c298f89-4303-0410-b956-a3cf2f4a3e73
lars [Wed, 13 Apr 2011 13:37:40 +0000 (13:37 +0000)]
kernel: revert r24160 (remove the bogus ext2/ext3 dependency on linux 2.6.30/31)
For kernel versions newer then 2.6.31 the ext4 module can be used to mount
ext2/3 filesystems.
Building ext2/3 as modules on the other hand breaks using ext4 for mounting ext2
or ext3, which breaks booting from ext2/3 on machines where the ext4 module is
built into the kernel.
base-files: return success on lib-copying with external toolchain
when using an external toolchain the base-files package copies libc, libgcc and
others from the library directory.
The file list is given as following in the .config:
CONFIG_LIBC_FILE_SPEC="./lib/ld{-*.so,-linux*.so.*} ./lib/lib{anl,c,cidn,crypt,dl,m,nsl,nss_dns,nss_files,resolv,util}{-*.so,.so.*}"
Because the filenames are composed with different endings, not all files exist
and will be skipped. Currently, this works only if the last composed file
(util.so.*) really exists. At the moment this works - but only if you don't add
a new file like 'uClibc'.
Adding it at the end '...resolv,util,uClibc}{-*.so,.so.*}' will lead to this
message, because the combination 'libuClibc.so.*' doesn't exist and Make will
evaluate the last copy statement of the for loop.
--- Message Snippet ---
cp: cannot stat `/home/user/Desktop/code/meetwise/toolchain/staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/./lib/libnss_files.so.*': No such file or directory
cp: cannot stat `/home/user/Desktop/code/meetwise/toolchain/staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/./lib/libresolv-*.so': No such file or directory
cp: cannot stat `/home/user/Desktop/code/meetwise/toolchain/staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/./lib/libresolv.so.*': No such file or directory
cp: cannot stat `/home/user/Desktop/code/meetwise/toolchain/staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/./lib/libuClibc.so.*': No such file or directory
make[2]: *** [/home/user/Desktop/code/meetwise/openwrt/bin/at91/packages/libc_-68_at91.ipk] Error 1
make[2]: Leaving directory `/home/user/Desktop/code/meetwise/openwrt/package/base-files'
make[1]: *** [package/base-files/compile] Error 2
make[1]: Leaving directory `/home/user/Desktop/code/meetwise/openwrt'
make: *** [package/base-files/compile] Error 2
--- /Message Snippet/ ---
To fix this unwanted behaviour I added an extra 'exit 0' to each for-loop and
make ignores non-existing files as before.
Signed-off-by: Sven Bachmann <dev@mcbachmann.de>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26623 3c298f89-4303-0410-b956-a3cf2f4a3e73
This patch updates qos-scripts to support fair traffic sharing using the
SFQ with external classifiers method. It also corrects a bug in the
unsupported ESFQ method already used by qos-scripts. (ESFQ:
http://fatooh.org/esfq-2.6/ only updated to 2.6.24, it was switched to
an SFQ patch after that and not updated since 2008)
A class can be forced to use SFQ, and an external classifier added like
this:
config class "Normal"
option avgrate 10
option priority 30
option packetdelay 100
option limitrate 94
# option qdisc "sfq perturb 2"
config class "Normal_up"
# option filter "protocol all flow hash keys src divisor 1024"
config class "Normal_down"
# option filter "protocol all flow hash keys dst divisor 1024"
Using these options, the user needs to load cls_flow before qos-scripts
starts.
I've got more information here:
http://oneitguy.com/blogs/netprince/fair-traffic-sharing-esfq-broken-switching-sfqexternal-classifiers
This has been tested on r23914.
Signed-off-by: Ben Pfountz <netprince<>vt_edu>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26622 3c298f89-4303-0410-b956-a3cf2f4a3e73
config redirect
option src 'wan'
option dest 'lan'
option src_dport '22001'
option dest_port '22'
option proto 'tcp'
note the absence of the "dest_ip" field, meaning to terminate the connection on the firewall itself.
This patch makes three changes:
(1) moves the conntrack module into the conntrack package (but not any of the conntrack_* helpers).
(2) fixes a bug where the wrong table is used when the "dest_ip" field is absent.
(3) accepts incoming connections on the destination port on the input_ZONE table, but only for DNATted
connections.
In the above example,
ssh -p 22 root@myrouter
would fail from the outside, but:
ssh -p 22001 root@myrouter
would succeed. This is handy if:
(1) you want to avoid ssh probes on your router, or
(2) you want to redirect incoming connections on port 22 to some machine inside your firewall, but
still want to allow firewall access from outside.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26617 3c298f89-4303-0410-b956-a3cf2f4a3e73
mac80211: ath9k: register id table for platform device
Currently the device id in the platform driver is hardcoded to an
id which is specific to AR9130/AR9132 SOCs as it supports only wmac
(wireless mac) of these SOCs. But this needs to be dynamic when we
want to support different wmac of SOCs. So add id_table to driver to
make it extendable to more SOCs.
Add support for the Buffalo WZR-HP-G301NH. The only difference between it
and the WZR-HP-G00NH is that it has a RTL8366RB instead of a RTL8366S.
Since we don't do runtime detection of the switch, we need a separate
machine definition for it.
While we are at it, also rename the profile to reflect that it now is for
more than one device.
Signed-off-by: Jonas Gorski <jonas.gorski+openwrt@gmail.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26601 3c298f89-4303-0410-b956-a3cf2f4a3e73
Since the PHY driver is only used for the WAN port and there is virtually
no difference between it and the generic PHY driver, we can sefely remove
it.
Signed-off-by: Jonas Gorski <jonas.gorski+openwrt@gmail.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26600 3c298f89-4303-0410-b956-a3cf2f4a3e73
* Some module should be loaded later to load them after the modules they are depending on
* add some more missing config symbols
* make CS5535 build again
- Use physical addresses definition for AR2315 the same way as AR5312. Fixes ioremap
- Fix dma mapping for AHB bus (only use the PCI DMA offset for PCI devices)
uClibc: forced unwind for pthread_cancel handling is broken and triggers spurious abort() calls from libgcc. disable it and use the other method instead