Hi,
thank you for looking into this. Please see inline below.
On Tue, Oct 25, 2011 at 4:48 PM, Jo-Philipp Wich <x...@subsignal.org> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi.
>
>> +US_PATH="/usr/sbin/lisp"
>> +KMOD_PATH="/usr/sbin"
>> +CONF_PATH="/lib/network/lisp"
>> +IF_NAME="eth1"
>> +
>> +# RLOC used in ip rules for source routing
>> +#RLOC="128.107.165.94"
>> +RLOC=`/sbin/ifconfig eth1 | sed '/inet\ /!d;s/.*r://g;s/\ .*//g'`
>
> Here I'd suggest something like:
>
> IF_NAME=$(uci -q -P /var/state get network.wan.ifname)
> RLOC=$(uci -q -P /vat/state get network.wan.ipaddr)
Agreed on the IF_NAME.
Let's follow whatever you think will always work. If you think calling
uci will be more "universal" in the OpenWrt environment compared to
calling /sbin/ifconfig, that's totally fine with us.
>
>> +sleep 2
>> +insmod ${KMOD_PATH}/lisp_int.ko
>> +sleep 3
>> +insmod ${KMOD_PATH}/lisp.ko
>> +sleep 3
>
> Is there a particular reason for the delays or would loading those
> modules at startup be fine?
Though these delays are very conservative and most probably no delay
needs to be introduced between lisp_int.ko and lisp.ko, the last delay
(before the user space module is inserted) may be needed to make sure
that the virtual interface is already up so that the IP addresses in
the outer and inner header of the datagrams are properly set.
>
>> +/sbin/udhcpc -i $IF_NAME
>
> Why must DHCP be re-requested here? Is the WAN supposed to be DHCP or
> could it be static as well?
Either EID ('permanent' of inner header) or RLOC (routing locator,
globally routable) IP address can be fixed or DHCP assigned. We have
tried all of these scenarios on our LISP Beta network. Although
usually a DHCP re-request is not needed, we have found out that there
are occasions that if we do not issue a DHCP request right after we
start running LISP on a Linksys router, the inner header IP address
might not get correctly configured with the virtual interface's
('lmn0') address (which is the EID). So, for now we just reissue DHCP
request anytime we start running LISP.
>
>> Index: package/lisp/src/timerfd.h
>> ===================================================================
>
> You attached the complete sourcecode for the daemon here. That is not
> wrong per se, but I wonder what the future plans are. Do you plan to
> develop it within OpenWrt or will there be an external repository where
> the source can be pulled from?
I have personally come across this particular timerfd.h functionality
on the Internet. We have had issues with uClibc's 0.90.3
implementation of this and that's why it is slightly different. We are
fine with following either path that you guys suggest.
>
>> +# OpenWrt LISP Package
>> +# Vasileios Lakafosis (vasile...@gatech.edu)
>
> Maybe state a license here.
Good suggestion. Thanks!
>
>
> The rest looks fine so far.
>
> ~ Jow
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAk6nILIACgkQdputYINPTPOChwCfTsYYb8MfvKIu3bshPtK6wNna
> ziEAoI1bhQxyOOp2cWxNvG9BFNnysFkA
> =iDdW
> -----END PGP SIGNATURE-----
>
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
