A year of testing in the cerowrt project shows not using timestamps
to be a very bad idea in nearly any TCP at speeds above a few Mbit.
Lastly sack/dsack help on recovery from larger amounts of packet
loss.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@32513 3c298f89-4303-0410-b956-a3cf2f4a3e73
On slower devices wifi drivers might take too long for detecting
devices, resulting in the wifi detect call not seeing them.
This was observed on a bcm6348 with bcm4318 wifi. Adding a one second
pause was enough for b43 to expose the device.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@31639 3c298f89-4303-0410-b956-a3cf2f4a3e73
hi
Another version, in this one the enable_server option is in the timeserver ntp part of the "system" config file
You can patch trunk and bacfire (tested both)
You can put busybox ntpd in client mode (if you put server), in client & server (by putting enable_server to 1, ntpd listen to udp 123), and also in server mode only (if you didn't put any servers in the config and still put enable_server 1, ntpd will answer with the time of the router)
I've replaced "config_foreach getpeers timeserver" with "config_get peers ntp server" because we want ntp timeserver, not random ones (to pre-answer if someone want to say that it's intrusive ...)
Signed-off-by: Etienne CHAMPETIER <etienne.champetier@free.fr>
Le 27/03/2012 20:41, Etienne Champetier a écrit :
> I've now tested my trunk patch and it works fine
> But I still can't find were $PROG is defined (is this a mistake, or some sort of built in variable???)
> (I've made some grep and nothing)
>
> Le 23/03/2012 02:19, Philip Prindeville a écrit :
>> Maybe:
>>
>> [ -n "$PROG" -a -x "$PROG" ] || return 1
>>
>> instead?
>>
>>
>> On 3/22/12 4:34 PM, Etienne Champetier wrote:
>>> Hi
>>>
>>> The 2 attached patchs (trunk & bacfire) add busybox ntpd enable_server option, as busybox ntpd server is compiled by default.
>>> We only need 1 client/server daemon (olipro patch was launching 2 daemons)
>>> I've fully tested the bacfire patch, and as i don't have a running openwrt trunk i'm not sure for the trunk patch (i'm sure about my modifications, but i'm not sure about "[ -x $PROG ] || return 1", as "$PROG" isn't defined ?!)
>>>
>>> Signed-off-by: Etienne CHAMPETIER <etienne.champetier@free.fr>
>>>
>>>
>>> Le 16/01/2012 01:57, Philip Prindeville a écrit :
>>>> On 1/14/12 11:37 AM, Olipro wrote:
>>>>> On Saturday 14 Jan 2012 02:45:59 Philip Prindeville wrote:
>>>>>> Don't we already have a 'disabled' option? Now we're adding an
>>>>>> 'enable_server' option?
>>>>>>
>>>>>> That seems confusing for no useful reason.
>>>>>>
>>>>> have you bothered to read what I originally wrote? your response would make
>>>>> me inclined to believe that you didn't.
>>>>>
>>>>> currently the ntpd initscript only runs it as a CLIENT - this patch enables
>>>>> you to have one instance running as a client and another as a SERVER that
>>>>> other hosts can synchronise with.
>>>>>
>>>>> Or perhaps I'm misunderstanding, what would you propose for allowing the
>>>>> built-in busybox ntpd to be utilised as a server? a separate init script
>>>>> entirely perhaps?
>>>> Or separate config sections... instead of 'config ntp' have 'config ntp-server' and 'config ntp-client'.
>>>>
>>>> -Philip
>>>>
>>>>
>>>> _______________________________________________
>>>> openwrt-devel mailing list
>>>> openwrt-devel@lists.openwrt.org
>>>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>> _______________________________________________
>> openwrt-devel mailing list
>> openwrt-devel@lists.openwrt.org
>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
> _______________________________________________
> openwrt-devel mailing list
> openwrt-devel@lists.openwrt.org
> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@31374 3c298f89-4303-0410-b956-a3cf2f4a3e73
- Introduce new Kconfig symbols VERSION_DIST, VERSION_NICK, VERSION_NUMBER and VERSION_REPO to specify distribution, release name, version and repository for a given build
- Introduce include/version.mk to provide common helpers for packages dealing with versions
- Make opkg use version.mk to populate the opkg.conf template
- Make base-files use version.mk to populate /etc/openwrt_version, /etc/openwrt_release and /etc/banner
The available placeholders are:
%D .. Replace with $(CONFIG_VERSION_DIST), default to "OpenWrt"
%d .. Like %D, but all characters made lowercase and spaces substituted with "_" (e.g. "openwrt")
%N .. Replace with $(CONFIG_VERSION_NICK), default to the build tree release (e.g. "Attitude Adjustment")
%n .. Like %N, but all characters made lowercase and spaces substituted with "_" (e.g. "attitude_adjustment")
%V .. Replace with $(CONFIG_VERSION_NUMBER), default to the build tree release (e.g. "r31262")
%v .. Like %V, but all characters made lowercase and spaces substituted with "_"
%C .. Replace with $(CONFIG_VERSION_NUMBER), default to "Bleeding Edge"
%c .. Like %C, but all characters made lowercase and spaces substituted with "_" (e.g. "bleeding_edge")
%U .. Replace with $(CONFIG_VERSION_REPO), default to "http://downloads.openwrt.org/snapshots/trunk/%T/packages"
%R .. Replace with the current build tree revision (e.g. "r31262" or "75488c4a05b8033cf69e91874a61852db7ba9c6c")
%T .. Replace with the current target (e.g. "ar71xx")
%S .. Replace with the current target/subtarget combo (e.g. "adm5120/router_le")
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@31262 3c298f89-4303-0410-b956-a3cf2f4a3e73
The muninlite postinst fails to add munin to /etc/services.
This patch removes this broken postinst and add an entry for munin in
/etc/services. This should have been done when the rest of the
/etc/services postinsts were being removed (changesets 26142-26149).
Signed-off-by: Jonathan McCrohan <jmccrohan@gmail.com>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@30982 3c298f89-4303-0410-b956-a3cf2f4a3e73
Solves broken key based dropbear login with empty password after r28935.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@29130 3c298f89-4303-0410-b956-a3cf2f4a3e73
The rdate applet proved to be too unreliable to obtain the current time on boot:
- public time servers are rare and often unreachable or overloaded
- rdate does not daemonize, it needs a network connection the moment it is started, leading to race conditions
- the /etc/config/timeserver configuration is overly complex and there is no reliable way to disable rdate invocations
- the time protocol as specified in RFC 868 is considered obsolete
This commit adds an init script /etc/init.d/sysntpd which starts and stops the busybox ntpd accordingly.
The builtin ntpd can be disabled by either disabling the init script, removing the symlink to busybox or
by clearing the timeserver list in /etc/config/system.
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@28612 3c298f89-4303-0410-b956-a3cf2f4a3e73
eliminates unnecessary flash write cycles at every boot
patch by Peter Wagner (tripolar)
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@27127 3c298f89-4303-0410-b956-a3cf2f4a3e73
The script tests for the existance of /dev/root with test -e which fails if
/dev/root is a dangling symlink making the call to ln fail.
Signed-off-by: Justus Winter <4winter@informatik.uni-hamburg.de>
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@26483 3c298f89-4303-0410-b956-a3cf2f4a3e73
