Hi Oliver/Robert, Thanks very much for proposing ways to test CAN.
On Tue, 2008-10-28 at 09:42 +0100, Oliver Hartkopp wrote: > Robert Schwebel wrote: > > Hi Subrata, > > > > On Wed, Oct 22, 2008 at 05:02:06PM +0530, Subrata Modak wrote: > > > >>>> Controller area network (CAN) protocol support is recently added to > >>>> linux-2.6.25. At Linux Test Project (http://ltp.sourceforge.net/), we > >>>> aim to provide test cases that tests all features of Kernel. So, we > >>>> search for new test cases that can be added. Can you please share the > >>>> test cases for "Controller area network (CAN) protocol support" with us > >>>> under GPLv2 ? > >>>> > >>>> > >>> Hi Subrata, > >>> > >>> i have no objections. I created some test-apps in > >>> > >>> http://svn.berlios.de/svnroot/repos/socketcan/trunk/test/ > >>> > >>> for development and tutorial purposes. But at first glance this does not > >>> fit into the Linux Test Project. > >>> > >>> I would like to go a bit deeper into LTP before i can give you a > >>> detailed answer - but it should be no problem at all to share test cases > >>> (under GPLv2). > >>> > >> Did you find out how we can integrate the above test case infrastructure > >> for CAN into LTP ? In case you would also like to contribute these test > >> cases (above URL), you can reply to this mail with: > >> > >> Signed-Off-By: Oliver Hartkopp <[EMAIL PROTECTED]> > >> > >> which then gives me the right to include the same. > >> > > > > > > Hello Subrata, > > sorry for the delay. Of course you may use the mentioned test cases on > the public SocketCAN SVN on BerliOS under GPL v2: > > Signed-Off-By: Oliver Hartkopp <[EMAIL PROTECTED]> > > > But these test cases are only limited to some cases i wanted to check > during the development phase, so there are sometimes ... > > - too easy to have an evidence > - too special to help for a general test > > > > > It may be possible to add a virtual can interface to a box, then run > > tests against itself, but according to our experience (we have been > > involved into the design & implementation of SocketCAN), it is of > > limited help, because usual problems come up with real hardware. > > I do not mind testing it on a vitual can interface for the time being without having the real hardware. > > > > Using multiple instances of the virtual CAN driver (vcan) makes sense > and all of my current tests (for the network layer stuff) make use of > the virtual CAN driver. Of course some problems only emerge on real > hardware, but AFAICS testing HW drivers is not the main focus of LTP, > right? We need some test stuff for the CAN network layer functionality > that can be also performed on weird CPU architectures without real CAN > hardware. Yes, i agree. > > IMO the most interesting topic to test in the CAN subsystem are (with > priority): > 1. the filtersets of the CAN core > 2. the socket options of the can-raw sockets > 3. the filters and timeouts of the can-bcm > > Due to a sleepless night, i got an idea how to put all this (Point 1&2) > into one test program :-] > > To get this running i would appreciate if Robert could create some LTP > CAN test base framework which creates four virtual CAN interfaces (vcan0 > .. vcan3) with 'ip link add type vcan' & 'ifconfig vcanx up' and > probably a basic send/recv test. I recently added the missing > documentation for creating vcan devices in > Documentation/networking/can.txt . > Robert, We already have some infrastructure for Network tests in LTP, where you require 2 systems. Install the suite in same directory both in Client and Server. Setup some important services in the Server side. Export variable for somebody to login from Client to Server. And then fire the network tests from the Client. I feel the initial basic tests of CAN can be fitted in to this framework. If some services need to be started at the server, this can be done during the normal network tests server setup. And then CAN client scripts can run when other Network scripts run at the client side. Let me know what you think about this. Regards-- Subrata > On top of this work i would write and contribute a test that would check > the full CAN core filter mechanism and the can-raw sockopts. > > Would this be a reasonable approach for you? > > Best regards, > Oliver > ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Ltp-list mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ltp-list
