replacing umlaute and fixing address in headers Haldeneggsteig -> Universitaetsstrasse

Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>

Fixed race condition when initializing benchmark using raw interface
Up until now the buffers were registered at the beginning for all
benchmarks. This caused problems with the lwip benchmark, because in
that case the buffers are registered before we have a connection to
netd, which in turn can lead to a screw up in the queue manager library,
where the wrong binding is used as a netd binding.
---
include/net_interfaces/net_interfaces.h | 2 +-
usr/bench/net_latency/elb/benchmark.c | 39 ++++++++++++++++++++--------
usr/bench/net_latency/elb/benchmark_tcp.c | 19 +++-----------
usr/bench/net_latency/elb/interface_lib.c | 9 +++++-
usr/bench/net_latency/elb/interface_raw.c | 12 +--------
5 files changed, 41 insertions(+), 40 deletions(-)

new interface is working with most of the applications.
There are some issues with memory allocation which triggers failure
in alloc_pool_pbuf for webserver when loading all files, but applications
like echo_server is running.

Applied patch 0002 by antoinek

modified loopback driver to work with ump based interface.
Also removed all the code from queue_manager which was using bulk_transport

Introduced raw flounder interface to replace bulk transport
Since we noted that bulk transport seems to cause a lot of overhead,
this seemed like an easy experiment to do. For low latency communication
benchmarks seem to indicate that we save quite some time. But note that
the interface is quite rudimentary and does not support everything bulk
transport does (esp. buffer chains for a single packet).
---
if/net_queue_manager.if | 6 +
lib/net_queue_manager/queue_manager.c | 112 +++++++++++++++-
usr/bench/net_latency/elb/Hakefile | 2 +-
usr/bench/net_latency/elb/interface_raw.c | 215 +++++++++++++++++++++++++++++
4 files changed, 333 insertions(+), 2 deletions(-)
create mode 100644 usr/bench/net_latency/elb/interface_raw.c