replacing umlaute and fixing address in headers Haldeneggsteig -> Universitaetsstrasse

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

nfs: increased max read size

Signed-off-by: Roni Häcki <roni.haecki@inf.ethz.ch>

tftp & vfs: properly return error when init of network stack fails

Signed-off-by: Roni Häcki <roni.haecki@inf.ethz.ch>

libvfs: nfs client: update write #define parameters to work with new network stack

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

Use min/max macros from bitmacros.h instead of including sys/param.h

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

net: converting lwip types to bsd types

Signed-off-by: Adam Turowski <adam.turowski@inf.ethz.ch>

net_sockets: converting nfs

Signed-off-by: Adam Turowski <adam.turowski@inf.ethz.ch>

libc: removing newlib, adding BSD libc
posixcompat: moving gettimeofday to libc
harness: adding a simple math test
lwip, lwip2: moving includes to /include, compiling with BSD headers

Signed-off-by: Adam Turowski <adam.turowski@inf.ethz.ch>

Move canonical definitions of MIN/MAX macros to bitmacros.h, remove duplicated definitions in lib and usr

Signed-off-by: Simon Gerber <simon.gerber@inf.ethz.ch>

NFS loads something now but its slow.

merge

this version of code works when application and driver are on different
cores. Which means, when they are using UMP it works, but not when they are
using LMP

unbreak nfs client for threadsafe lwip

This function "wait_for_condition_fair()" looks dubious at best, but
I'm just trying to make stuff work. Don't blame me for touching it
last!

Fixed minor issues in large reads over nfs, and also compilation
errors for scc architecture are fixed.

Removed the additional ring from packet send path. The driver code for e1000n
and e10k is modified but code for other drivers (eMAC, and rtl) is still to
be modified.

Made read over nfs faster by parallelizing the chunk fetching. Also
now using more fair wait_for_condition which does not block
waiting on events like timer without handling pending incoming packets.

Also, added test for measuring nfs throughput in harness. And re-arranged
few files related to net_tests.

--HG--
rename : usr/tests/net_throughput/Hakefile => usr/tests/net_tests/nfs_throughput/Hakefile
rename : usr/tests/net_throughput/main.c => usr/tests/net_tests/nfs_throughput/nfs_cat.c
rename : usr/tests/net_openport_test/Hakefile => usr/tests/net_tests/openport_test/Hakefile
rename : usr/tests/net_openport_test/net_openport_test.c => usr/tests/net_tests/openport_test/net_openport_test.c

Fixed the issues with nfs benchmarking code in net_throughput
Currently getting around 50MB/s speed for NFS read

Boots correctly for both freec and newlib

usr/replay: replace fwrite() with app buffers and write()

* Added compile trace replayer
* Disabled fixes to bcache locking
* Fixed hashtable to use const void instead of just void
* Made fopen buffer even bigger to preserve close-to-open semantics
for larger files.
* Worked around SCC bootup race (channel teardown NYI)

ICC compilation fixes

Applying the diff with my tree as patch

Applied the patch to rebase my changes on new barrelfish tree

Merge from ASPLOS tree.

fixes to enable compilation with ICC

The primary changes are avoiding pointer arithmetic on void * types, which is
evidently a GCCism: instead a byte-sized pointer type should be used.

Other changes include workarounds to avoid uninitialised variable warnings,
and an insistence that const values have an initialiser (being static and implicitly
initialised to zero isn't sufficient).

Initial version of public stable barrelfish repository.