Hi Pablo,
This series lays the groundwork for removing the socket references from
the GTP netdevice by removing duplicate code and simplifying the logic on
some code paths.
It slighly changes the GTP genl API by making the socket parameters optional
(though one of them is still required).
The removal of the socket references will break the 1:1 releation between
GTP netdevice and GTP socket that prevents us to support multiple VRFs with
overlaping IP addresse spaces attached to the same GTP socket (needed for
multi APN support).
Regards
Andreas
--
Andreas Schultz (6):
gtp: make GTP sockets in gtp_newlink optional
gtp: merge gtp_get_net and gtp_genl_find_dev
gtp: unify genl_find_pdp and prepare for per socket lookup
gtp: consolidate pdp context destruction into helper
gtp: add socket to pdp context
gtp: consolidate gtp socket rx path
drivers/net/gtp.c | 564 +++++++++++++++++++++++++++---------------------------
1 file changed, 280 insertions(+), 284 deletions(-)
--
2.10.2
This is the part of the previous "simple gtp improvements" series
that Pablo indicated should go into net-next.
The rcu_lock removal is small correctness changes. Passing invalid
to user space allows for more standards compliant handling of invalid
tunnels.
--
Andreas Schultz (2):
gtp: remove unnecessary rcu_read_lock
gtp: let userspace handle packets for invalid tunnels
drivers/net/gtp.c | 24 ++++--------------------
1 file changed, 4 insertions(+), 20 deletions(-)
--
2.10.2
I'm sorry for the compile error mess up in the last version.
It's no excuse for not test compiling, but the hunks got lost in
a rebase.
This is the part of the previous "simple gtp improvements" series
that Pablo indicated should go into net.
The addition of the module alias fixes genl family autoloading,
clearing the DF bit fixes a protocol violation in regard to the
specification and the netns comparison fixes a corner case of
cross netns recv.
Andreas
v2->v3: fix compiler error introduced in rebase
--
Andreas Schultz (3):
gtp: add genl family modules alias
gtp: clear DF bit on GTP packet tx
gtp: fix cross netns recv on gtp socket
drivers/net/gtp.c | 13 ++++++-------
1 file changed, 6 insertions(+), 7 deletions(-)
--
2.10.2
While I was waiting for jenkins just now, I noticed that it was getting
impossibly slow despite an idle CPU. An attempt to restart resulted in failure
to launch jenkins due to "Disk quota exceeded".
There are things happening that I don't understand (file system is shrinking as
I remove files). I need help from Holger (or Harald?), to understand and
probably fix with a simple command making more space for the jail...
Until then our jenkins will be DOWN :/
~N
--
- Neels Hofmeyr <nhofmeyr(a)sysmocom.de> http://www.sysmocom.de/
=======================================================================
* sysmocom - systems for mobile communications GmbH
* Alt-Moabit 93
* 10559 Berlin, Germany
* Sitz / Registered office: Berlin, HRB 134158 B
* Geschäftsführer / Managing Directors: Harald Welte
This is the part of the previous "simple gtp improvements" series
that Pablo indicated should go into net.
The addition of the module alias fixes genl family autoloading,
clearing the DF bit fixes a protocol violation in regard to the
specification and the netns comparison fixes a corner case of
cross netns recv.
I'm not sure if Pablos comments on the previous version qualify
as ACK, so I left that out.
--
Andreas Schultz (3):
gtp: add genl family modules alias
gtp: clear DF bit on GTP packet tx
gtp: fix cross netns recv on gtp socket
drivers/net/gtp.c | 9 ++++-----
1 file changed, 4 insertions(+), 5 deletions(-)
--
2.10.2
On jenkins and opensuse, we have builds where we cannot easily analyse the
testsuite.dir after a build failed. So we only see which tests failed and not
why.
On jenkins I have in most places added a script invocation that prints all
testsuite.dir/*/*.log in case of build failure, so we can see it in the jenkins
log right away.
On opensuse.org, IIUC, the 'make check' is run by dh or something, so it's not
trivial (to me) to make it output the test logs in case of failure.
I'm thinking now, don't we always want to see these failures anyway? We could
include in our Makefile.am a tweak to print the test logs in case of testsuite
failure.
Sometimes this output can become fairly large though; e.g. when a long test
fails early and prints "nothing", the test log shows a huge diff with all
expected output as minus-lines.
My main goal right now is to see why/how the libosmo-netif OBS build failed
(osmux test fails) without further local reproduction effort. This situation
will certainly repeat over and over, for all of the other projects as well.
What do you think? Can we get that easily?
~N
--
- Neels Hofmeyr <nhofmeyr(a)sysmocom.de> http://www.sysmocom.de/
=======================================================================
* sysmocom - systems for mobile communications GmbH
* Alt-Moabit 93
* 10559 Berlin, Germany
* Sitz / Registered office: Berlin, HRB 134158 B
* Geschäftsführer / Managing Directors: Harald Welte
FYI,
no "Broken Pipe" errors in a long time, so today I increased the OsmocomBuild1
nr of executors from 3 to 4 ... and immediately got two subsequent openbsc
builds hitting the "Broken Pipe" again.
https://jenkins.osmocom.org/jenkins/job/OpenBSC-gerrit/1716/IU=--enable-iu,…https://jenkins.osmocom.org/jenkins/job/OpenBSC-gerrit/1717/IU=--disable-iu…
This time I tried something else: lower the "make -j 9" to "make -j 3" and
leave the nr of executors at 4...
Just dabbling, would be nicer to pinpoint the real cause.
~N
--
- Neels Hofmeyr <nhofmeyr(a)sysmocom.de> http://www.sysmocom.de/
=======================================================================
* sysmocom - systems for mobile communications GmbH
* Alt-Moabit 93
* 10559 Berlin, Germany
* Sitz / Registered office: Berlin, HRB 134158 B
* Geschäftsführer / Managing Directors: Harald Welte