On Wed, Dec 06, 2006 at 08:56:32AM +0100, Waldemar Brodkorb wrote:
> Hi,
> On Tue, 05 Dec 2006 at 21:45 +0100, Dirk Nehring wrote:
> > On Tue, Dec 05, 2006 at 09:02:11AM +0100, Markus Wigge wrote:
> > > Hi,
> > >
> > > >> Modified:
> > > >>    branches/freewrt_1_0/package/base-files-arch/Makefile
> > > >>    
> > > >> branches/freewrt_1_0/target/linux/brcm-2.4/files/etc/network/interfaces
> > > >> Log:
> > > >> add broadcast to interfaces template, fix switch config for non-switch 
> > > >> devices
> > > >   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> > > >
> > > > Personal interest: why is this necessary?
> > >
> > > This option is unnecessary. AFAIK you needed to explicitly give the
> > > broadcast address for 2.0 kernel series or before?
> > > So this option is deprecated and only available for historical reasons I
> > > think.
>
> It is not deprecated. If you have a more or less non-standard
> netmask for your network, you need to give the broadcast address,
> otherwise you get strange network problems. At least this is
> definitively needed in Debian GNU/Linux 3.1, because ifupdown does
> not compute the broadcast very well. (See bugs.debian.org for
> references) It has nothing todo with the kernel version. This
> happens to me on 2.6.x kernels.
>
> A correct broadcast address is always needed.
>
> The line does not harm, feel free to remove it on your router ;)

I don't thing we need the broadcast directive even if it does not
harm. busybox uses "ip" to set up the ip address and if you do not set
the broadcast, it's on the kernel to choose the correct one See
different outputs in "ip addr show":

    inet 192.168.213.1/24 brd 192.168.213.255 scope global eth1

vs.

    inet 192.168.213.1/24 scope global eth1

If the kernel is buggy in this case, you should give me a concrete
indication so we try to fix it. Perhaps you have a bug id or example to
reproduce?

Dirk
_______________________________________________
freewrt-developers mailing list
[email protected]
https://www.freewrt.org/lists/listinfo/freewrt-developers

Reply via email to