[lkml]   [1999]   [Feb]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Routing Table (Feature)
On Sun, 21 Feb 1999 wrote:

> Hello!
> > The user code doesn't care how a connection is established, nor
> > how a datagram gets to a destination. Machine-specific raw sockets
> > don't even care.
> Sure? Then try to start gated in your configration and
> look what will occur.

Well gated is one of the things that should have been fixed a long time
ago. Advertising that it has a route on the same physical interface
is broken. This makes cooperating machines send packets to a machine
advertising such a route, only to have that machine send these packets
out the same physical link, duplicating traffic.

> > on the physical network, by manipulating routing netmasks.
> ....
> [etc. explanation, when it can be useful skipped]
> I ABSOLUTELY agree. Then let's teach gated etc. not to rely
> on interface addresses for beginning. Then we can return
> to this question and delete from kernel useless SIOC[GS]IFNETMASK.
Correct on gated, we'll have to "agree to disagree" about the

> It is not joke. You may scan
> to look at apps, which DO use routing tables directly not relying
> on address information (pimd, rsvpd etc). If the program of sanitization
> of bsd heritage appeared realistic, I would sign under each your word.

Yes, BSD networking has a lot of baggage, but it became the de-facto

> For now, keep netmasks and routes in sync, please.

I can do that, but I have write code to reconfigure interfaces
during network routing changes (which I have done). Most would
like to rely on the de-facto standard tools of ifconfig and route.

---and I think you could help by deleting all instances of a route
when 'asked' by route.
> And if it is impossible, then try not to delete routes installed
> by kernel at least. (BTW you may just OVERRIDE interface routes
> without deleting route to wider network installed by kernel)

Yes, and presently, I can add a specific 'network' route to the device
itself.. without regard to its address. Nice, but most don't know
these tricks.

> Alexey

Well thanks, again. Please give my suggestion a good look.

Dick Johnson
Penguin : Linux version 2.2.1 on an i686 machine (400.59 BogoMips).
Warning : It's hard to remain at the trailing edge of technology.
Wisdom : It's not a Y2K problem. It's a Y2Day problem.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:50    [from the cache]
©2003-2014 Jasper Spaans. hosted at Digital Ocean