Re: OsmocomBB compile testing / Re: libosmocore embedded build

2017-05-29 Thread Harald Welte
Hi Craig,

On Thu, May 18, 2017 at 06:48:06PM +, Craig Comstock wrote:

> I was thinking about how to setup an automated real device test for
> the repo and/or PRs especially. I have devices and cables, just was
> thinking about how to automate the re-loading of firmware (via an
> interface to the power button I suppose).

I think there has been some work by Peter and/or Kevin (Cc) on setting
up OsmocomBB phones with a power supply (emulating a battery) and using
some serial handshake line (next to the UART) to simulate power button
presses in the past.  Peter?  Kevin?

It would be great if you'd be willing to set up such as system, document
it and then integate it.

I think you don't even need your own base setation.  Simple test cases
for OsmocomBB could include:

* running bcch_scan and check that certain (known strong) cells in the
  neigbhorhood are found

* trying to register to a given network using unknwon/expired IMSI and
  confirming that the channel is established and a LU REJECT is received

With such a test rig we could then automatically test every new
libosmocore and osmcoom-bb commit to verify that it still performs basic
receive and even basic rx+tx functionality.

> Any work ongoing on this front? 

I don't think there's much happening in OsmocomBB for many years, sadly.

-- 
- Harald Weltehttp://laforge.gnumonks.org/

"Privacy in residential applications is a desirable marketing option."
  (ETSI EN 300 175-7 Ch. A6)


Re: OsmocomBB compile testing / Re: libosmocore embedded build

2017-05-24 Thread André Boddenberg
Hi Harald,


> To resolve this, we should also start to have a jenkins compile testing
> job for OsmocomBB.  The "host" (PC) part of the code is built against
> regular libosmocore, just like e.g. openbsc or osmo-bts.  That should be
> possible even so far, and it might make sense to start with that.

Probably done [1][2], although I disabled following line to let
OsmocomBB build succeed:

libosmocore/contrib/verify_value_string_arrays_are_terminated.py
$(find . -name "*.[hc]")

Is it okay to disable this verification?


Regards,
  André

[1] https://gerrit.osmocom.org/#/c/2726/
[2] 
https://jenkins.osmocom.org/jenkins/view/Jenkins-Gerrit/job/osmocomBB-gerrit/


OsmocomBB compile testing / Re: libosmocore embedded build

2017-05-18 Thread Harald Welte
Hi Andre and others.

We currently have a series of patches from Vadim pending in gerrit for
OsmocomBB.  They cannot move ahead, as we have no compile testing /
jenkins job which would give this a +1.

To resolve this, we should also start to have a jenkins compile testing
job for OsmocomBB.  The "host" (PC) part of the code is built against
regular libosmocore, just like e.g. openbsc or osmo-bts.  That should be
possible even so far, and it might make sense to start with that.

Basically you need to:
* git clone osmocom-bb
* cd src/host/layer23
* regular 'autoreconf -fi && ./configure && make'

compile-testing the 'embedded' (firmware) part is not possible easily in
the current master.  However, as a second step, and after the
libosmocore embedded build has run (and it is installed to
/usr/local/arm-none-eabi), and if you use the laforge/remove-libosmocore
branch in OsmocomBB, you should also be able to compile-test the
firmware using

* git clone osmocom-bb
* cd src/target/firmware && make

There currently is no "make check" tests for either the host utilities
or the firmware, and of course we have no clue if the resulting binaries
will do anything useful on an actual pone [yet!], but I still think the
two steps above would be very useful to move ahead, and to unify the
patch submission/review/verification/approval/merge process in gerrit
with what we have established for the network-side projects like OsmoBTS
& co.

Regards,
Harald

-- 
- Harald Weltehttp://laforge.gnumonks.org/

"Privacy in residential applications is a desirable marketing option."
  (ETSI EN 300 175-7 Ch. A6)