I was never satisfied with the ugly hack, how kernel-modules
were loaded in openwrt, but it works since the beginning. here
is a proposal to do it clean and also fast with ash-builtins.
side-effect is a better logging. e.g.:
Jan 1 00:00:25 box daemon.info load_modules(): insmod: can't insert
It is always a bad idea to mix up different languages
in one environment. Another point is, to bury awk in
base-files. There is no reason for it, these are only
quick hacks from the old days of whiterussian 8-).
IMHO this a lot better and not that ugly, like with
awk. The function_name() says
It is always a bad idea to mix up different languages
in one environment. Another point is, to bury awk in
base-files. There is no reason for it, these are only
quick hacks from the old days of whiterussian 8-).
now we call the function + get a value, and do not rely
on a global set var_name,
It is always a bad idea to mix up different languages
in one environment. Another point is, to bury awk in
base-files. There is no reason for it, these are only
quick hacks from the old days of whiterussian 8-).
now we use ash-builtins and dont fork.
i have done some benchmarks, and the code is
it is always a bad idea to mix up different languages
in one environment. Another point is, to bury awk in
base-files. There is no reason for it, these are only
quick hacks from the old days of whiterussian 8-).
i have also removed all the /dev/null - statements,
which are IMHO not neccesary
it is always a bad idea to mix up different languages
in one environment. Another point is, to bury awk in
base-files. There is no reason for it, these are only
quick hacks from the old days of whiterussian 8-).
by using better varnames, the code is now readable.
Signed-off-by: Bastian Bittorf
Hello.
I am running Backfire 10.03.1-rc4 on an Asus WL-500gP v1 router
(original Broadcom card) and a Fonera 2100 client (builtin Atheros).
Basic AP - STA interaction is fine and reliable.
If I enable WDS, on both nodes, with or without STP, then the connection
is estabilished and terminated
these patches consolidate the former
301-kmod-fuse-dcache-bug-r4k.patch
302-kmod-fuse-dcache-bug-fuse.patch
of which 302 was missing in trunk's /target/linux/brcm47xx/patches-2.6.{36,37}
. this let to the reappearance of the DCACHE bug on brcm47xx platforms using
these kernels (listing a fuse
Hello!
Is wbd-111 (gemini) board GPIO outputs controllable using openwrt?
Janno
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
if i fetch a cgi-script and this script
also fetches something from localhost,
than the webserver hangs, until restart.
so:
cgi-bin-welcome.sh
has something like
wget http://127.0.0.1/cgi-bin-database.sh
then the server hangs.
bye, Bastian.
PS: the flag -A (tcp-keepalive) is not documented in
Hi all,
I am implementing a piggy-back style 2-hop Wifi deployment (2-hop is all we
need, so we don’t need mesh) with two TP-Link 741. What I did is the following:
clientA --- TPLinkA-TPLinkB --- clientB
The TPLinkA and TPLinkB are installed with backfire and are configured to run
in AP
At the rf level, your middle radios are still competing for the same space
if they are on the 2.4 spectrum. From the endpoints you are probably
creating a packet storm at the middle radios. If you change frequencies on
those radios, you will probably see better throughput. I.e. a 5.8 on one,
what is the best point to enforce an
compilation with -O3 instead of -Os ?
I tried to change
target/linux/ar71xx/Makefile - CFLAGS:
after menuconfig, but then my
image didn't work - or is there
a cleaner way?
bye, Bastian
___
openwrt-devel mailing
On Sun, Jan 16, 2011 at 10:06, Bastian Bittorf bitt...@bluebottle.com wrote:
what is the best point to enforce an
compilation with -O3 instead of -Os ?
Sorry to answer a question with another question, but why -O3? Are
you aware that on these size- and performance-restricted platforms -Os
is
Good to see this resonating in some form :)
I also like the idea of having it distributed.
I don't believe there has to be a lot of communication with the
build-bot, so buildbot and testbot can be two seperate animals. testbot
should probably have an interface accessible by buildbot to add
Hi Daniel!
Sorry for raining on your parade...
On 16/01/11 19:47, Daniel Golle wrote:
I don't believe there has to be a lot of communication with the
build-bot
Why not? I mean, it is not *much* communication bandwidth-wise, neither
does it have stringent latency requirements. But building on
==
Announcing the Wireless Battle Mesh v4
(16-20 March 2011, Sant Bartomeu del Grau, Spain)
==
The next 'Wireless Battle of the Mesh' will take place from
Wed 16th till Sun
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi.
This is expected. The uhttpd server is single threaded and will only
process one connection at a time, waits until the operation is finished,
and then switch to the next. By querying the server from within a
running cgi script you basically
Missing symbols for CONFIG_xxx ...
Index: target/linux/x86/config-default
===
--- target/linux/x86/config-default (revision 25026)
+++ target/linux/x86/config-default (working copy)
@@ -219,6 +220,7 @@
# CONFIG_M586TSC is
On Fri January 14 2011 22:52:08 Bastian Bittorf wrote:
this patch[1] seems trivial, but it is not in wireless-testing
yet - anybody knows why? (it seems to me, that mac80211 can
handle beacon-interval, but the iw commandline does not
handle it yet)
bye, Bastian
[1]
Building from scratch with -j6 fails because qemu is built with --enable-uuid
and libuuid is part of e2fsprogs.
Index: tools/Makefile
===
--- tools/Makefile (revision 25029)
+++ tools/Makefile (working copy)
@@ -56,7
Hi Kelly,
Thanks very much for the suggestion. I don't have a 5.8G router on hand, but
will try to get one asap to verify the bandwidth.
On the other hand, I am already using two non-overlapping channels (1 and 11),
the interference (or competition) among them should be minimal. I do expect
Hi Kelly,
Thanks very much for the suggestion. I don't have a 5.8G router on hand, but
will try to get one asap to verify the bandwidth.
On the other hand, I am already using two non-overlapping channels (1 and 11),
the interference (or competition) among them should be minimal. I do expect
The -O3 optimization adds a good bit of size that means more page
faults and cache misses, which translates to slower real-world
performance.
this is theory as must be checked.
our routers are so powerful (16mb RAM or more, these
where workstations some years ago), so we
have enough space and
24 matches
Mail list logo