In particular, it wants to run before the ntpclient script. Which may
block for a long time attempting to do DNS lookups for NTP servers. In
my case, that would have *worked* if the new device had been added to
teql first, rather than timing out.
This was effectively causing a huge delay between an interface coming
up, and routing actually starting to work.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 34442
This target is unused, does not support any off the shelf hardware and
has been a maintenance burden for quite some time now.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 34430
Patches about to go into net-next.git
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 34410
The genext2fs tool will 'reserve' 5% of the image size by default,
apparently by creating a huge lost+found directory. In fact it seemed to
be much more than 5% in practice — I saw an image with 8MiB used, and
recovered about 2MiB of it by deleting the 'empty' lost+found directory:
/dev/loop0 48377 8482 37438 19% /mnt/spare
# rmdir lost+found/
/dev/loop0 48377 6014 39906 14% /mnt/spare
This makes it configurable, but leaves it at the genext2fs default
of 5% for now. It should probably be changed to default to zero, but
that can be a separate patch.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 34408
The work has been backported from openwrt-dreambox with
some modifications & code cleanup.
* updated config-3.3
* updated config-3.6
* renamed rt-n13 to rt-n13u
* fixed mach-rt-n13u.c
[juhosg: move user-space support and image generation into separate
patches]
Signed-off-by: Amit Mendapara <mendapara.amit@gmail.com>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 34405
Currently we call mkfs.jffs2 --squash, which will change
both file permissions and owners.
If we have some file with suid bits, it will clear these bits.
Thus it's enough to just squash file owners, by replacing that
argument to --squash-uids.
Signed-off-by: Jonh Wendell <jonh.wendell@oiwifi.com.br>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 34397
Found in the AudioCodes Tulip AC494 ATA.
Signed-off-by: Daniel Golle <dgolle@allnet.de>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 34395
The general practice is to use the general SoC compatible string in both the
DTS and the board specific .c file.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 34393
* The device should get a unique name at the beginning and not wl%d.
* load the nvram just one time into the own buffer, also when there is
more than one device.
SVN-Revision: 34381
