Hello Paul, thanks for your quick reply. I was not available a few days so I couldn't answer earlier.
> Hello Marco, > > Speaking as an OpenOCD "fan" here, so please excuse my bias and > writing off-list :) > > On Mon, Mar 28, 2016 at 08:00:50PM +0200, Marco Behr wrote: > > I am using a commercial tool (Goepel electronics) for boundary scan tests > > in the company. Then I began looking for a non-commercial / open source > > tool which can do the same (with an low-cost hardware platform) so > > everyone at home can use it also. Soon I found UrJtag and OpenOCD as two > > possible platforms but neither one has a working implementation of this > > kind of test. > > Yes, UrJTAG has BSDL parser integrated but I'm not sure what real > advantage that gives. When one is testing something manually he can > easily interpret BSDL himself and I wrote some additional OpenOCD > scripts to be able to do boundary scan, see here[1]. The BSDL parser is good for quickly collecting ALL the information in the BSDL file. For an manually written test, it works fine when you only look for a single ports BScan cells and then write test pattern to them. But what I want to do is to define attributes for all reachable ports on the chip so the software can automatically generate big test vectors (e.g. using the counting pattern algorithm described by Kenneth P. Parker in his patent EP 0543506 B1) to detect shorts / opens between many nets. > But when it comes to automating it, my guess is that the most > straightforward way would be to write an SVF file generator. With > OpenOCD you can playback SVF with any kind of adapter, be it cheap > J-Link clone, FTDI-based device or even just about any Single Board > Computer (and for RaspberryPi there's a dedicated fast driver). Yes, SVF might be an option and both OpenOCD and UrJtag are able to play them. (And I think a few more software platforms can do this). But neither one can play SVF files on multiple devices. And furthermore tests on multiple JTAG chains simultaneously are absolutely not possible. So I think SVF might only solve a part of the problem. > There's also an OpenOCD enthusiast that has just shared a Python > parser for BSDL[2]. I think if you take something like that and > generate an SVF file from Python, playing it back with OpenOCD would > be a nice way to test your hardware. I had a look at this Python based parser - unfortunately I am completely new to Python. Although this one looks interesting and I will consider using it too. > Please feel free to share your thoughts and needs on the openocd-devel > mailing list. As I already said, I am thinking also about implementing these functions into OpenOCD instead of UrJtag. + there is much more activity at the OpenOCD project than at UrJtag + the Jim-Tcl implementation is also pretty useful - the main object of OOCD is debugging of software - not testing / debugging of circuits - OpenOCD doesn't allow multiple devices in a chain to be active (not bypassed) > Good luck, > And happy hacking :) > > [1] https://sourceforge.net/p/openocd/mailman/message/31069985/ > [2] https://github.com/cyrozap/python-bsdl-parser Thank you for your ideas - I will share this also on the OpenOCD mailing list ------------------------------------------------------------------------------ _______________________________________________ UrJTAG-development mailing list UrJTAG-development@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/urjtag-development
