[Olsr-dev] change to olsr_rt_flags() function in BSD but not Linux in stable branch

[Ferry Huberts]

On 06-04-12 09:06, Henning Rogge wrote:
> Kernel 2.4 does have netlink support for routing I think.
>
> See http://linux.about.com/library/cmd/blcmdl7_rtnetlink.htm
>
> VERSIONS
> rtnetlink is a new feature of Linux 2.2.
>

ok.
It's just that last time I worked on 2.4, there was no netlink for iptables.
That's why I said 2.4.
2.0 is even more ancient.
Go ahead and remove it :-)

> On Fri, Apr 6, 2012 at 09:03, Ferry Huberts<(spam-protected)>  wrote:
>>
>>
>> On 06-04-12 08:50, Markus Kittenberger wrote:
>>>
>>> thx for the reminder,..
>>> but my patch never was a build break, as it already did not compile
>>> without -D LINUX_NETLINK_ROUTING
>>>
>>> but i/we should really either remove kernel_routes_ioctl.c or fix it.
>>> (could push a fix, but will NOT do so,..)
>>> but i still think removing it makes more sense.
>>>
>>
>>
>> I agree.
>> 2.4 is _completely_ ancient and not worth the effort to maintain.
>>
>>
>>> as imho, the non netlink code likely did not work since a long time (and
>>> like the bsd code, it should have issues due to the wrong flags set in
>>> olsr_rt_flags())
>>>
>>> furthermore even with a fixed kernel_routes_ioctl.c olsrd did not
>>> compile without -D LINUX_NETLINK_ROUTING
>>>
>>> as e.g. the smartgateway code in not only within #ifdef
>>> LINUX_NETLINK_ROUTING
>>> therefore since olsrd 0.6.0 olsrd did not compile without
>>> LINUX_NETLINK_ROUTING
>>>
>>> while this was easy fixable too, and i could compile and run it, txtinfo
>>> did not work (or it crashes olsrd once he tries to load it) too.
>>> furthermore it would now make sense to test the ioctl routing code, for
>>> several issues we only fixed for netlink,..
>>>
>>> so i finally decided (with henning) NOT to fix it, as he will remove
>>> linux_ioctl stuff prior to olsrd 0.6.3 release,.. ( thx Henning! )
>>>
>>> Markus
>>>
>>> On Thu, Apr 5, 2012 at 7:56 PM, Hans-Christoph Steiner
>>> <(spam-protected)<mailto:(spam-protected)>>  wrote:
>>>
>>>
>>>     I just noticed something while rebasing my patches against the HEAD
>>>     of stable.  In the commit called "fix bsd RTM_GATEWAY routing flags
>>>     detection" by Markus Kittenberger
>>>     (66dcc5f8e3da8084cef0fcec3a4580d47b90f593), the function
>>>     olsr_rt_flags() was changed to have a second parameter:
>>>
>>>       olsr_rt_flags(const struct rt_entry *, int add);
>>>
>>>     This change was made in src/process_routes.h,  src/process_routes.c,
>>>     and src/bsd/kernel_routes.c but was not made in
>>>     src/linux/kernel_routes_ioctl.c.  Yet, olsrd seems to build and run
>>>     fine.  My guess is that src/linux/kernel_routes_ioctl.c is not used
>>>     at all, even though its being included it the build.
>>>
>>> it is only used if one want olsrd without netlink for linux,..
>>>
>>> and yes either i should change kernel_routes_ioctl.c (as i already
>>> promised)
>>> or remove it,..
>>>
>>>
>>>     If src/linux/kernel_routes_ioctl.c is not used, can it be removed?
>>>       If it is, then olsr_rt_flags() should be updated there.
>>>
>>>     On a related note, I setup an automated build on Debian/stable/amd64
>>>     and also runs olsrd using fakeroot after it builds:
>>>     https://build.safermobile.org/job/olsrd-stable/
>>>
>>>     .hc
>>>     --
>>>     Olsr-dev mailing list
>>>     (spam-protected)<mailto:(spam-protected)>
>>>     https://lists.olsr.org/mailman/listinfo/olsr-dev
>>>
>>>
>>>
>>>
>>
>> --
>> Ferry Huberts
>>
>>
>> --
>> Olsr-dev mailing list
>> (spam-protected)
>> https://lists.olsr.org/mailman/listinfo/olsr-dev
>
>
>

-- 
Ferry Huberts