netdev - ip=dhcp woes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 28 Jul 2016 13:26:13 +0200
From:	Uwe Kleine-König 
	<u.kleine-koenig@...gutronix.de>
To:	netdev@...r.kernel.org
Cc:	kernel@...gutronix.de
Subject: ip=dhcp woes

Hello,

I have a machine with four network interfaces and I'm using ip=dhcp
during development on it.

in ic_dynamic the procedure is the following (assuming no successful
reply is received in time):

  timeout = 2s + random([0, 1]) s
loop:
  send bootp on 1st dev
  wait 1s
  send bootp on 2nd dev
  wait 1s
  send bootp on 3rd dev
  wait 1s
  send bootp on 4th dev
  wait timeout
  timeout = timeout * 4 / 7
  goto loop;

My problem now is: The dhcp server is reachable via the first device and
takes little more than 1s to respond. A reply must match the last sent
request to be accepted.

So the obvious questions are:

Why is only the last timeout increased for each loop? Why is there a
difference at all between the waits which results in a special casing of
the last device?

Alternatively, why not accept a reply on eth0 when eth1 has already sent
a request? Then the procedure could be:

  timeout = 2s + random([0, 1]) s
loop:
  send bootp on 1st dev
  send bootp on 2nd dev
  send bootp on 3rd dev
  send bootp on 4th dev
  wait timeout
  timeout = timeout * 4 / 7
  goto loop;

which looks more effective.

Is there anything I missed?

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |