AW: Successfully built EPICS 7 with RTEMS 5 for mvme6100

Thu Mar 19 09:20:04 UTC 2020

> Von: devel [mailto:devel-bounces at rtems.org] Im Auftrag von Joel Sherrill
> Gesendet: Mittwoch, 18. März 2020 22:18
> An: Heinz Junkes
> Cc: rtems-devel at rtems.org
> Betreff: Re: Successfully built EPICS 7 with RTEMS 5 for mvme6100
> 
> 
> 
> On Wed, Mar 18, 2020 at 2:38 PM Heinz Junkes <junkes at fhi-berlin.mpg.de> wrote:
> Hello, Mritunjay,
> 
> I will contact you tomorrow during the day (European time).  We have here institute closings because of Corona Virus etc..  
> 
> It's great that you've come this far. I can also provide you with an MVME6100 remotely so you can get a feeling for target development. 
> Yes, I have also started development of libbsd and the MVME2500. More on this tomorrow.
> 
> But the first, important step in this proposal would be to get EPICS7 running with RTEMS5 on e.g. qemu, ideally on i386?
> I think I had some problems with the psim in the past. Unfortunately I don't know exactly what it is anymore. 
> More tomorrow.
> 
> We have had good experiences with RTEMS+libbsd with the Qemu ARM Zynq. 
> We have been able to telnet to the RTEMS target from Linux and use the RTEMS
> NFS client to mount Linux files plus dynamic loading works. Personally I would 
> start with that since the goal is EPICS not fixing a BSP. :)
> 
> My second would be Qemu i386 pc386 but I don't know the status of libbsd on 
> pc386 at the moment. This should work the same as the Zynq qemu but I 
> personally don't know. Heinz.. do you have recent experience on it?
> 

If the patches here (https://lists.rtems.org/pipermail/devel/2020-March/058119.html) make it through the review, rtems-libbsd should compile again for the pc386 BSP.
We also could successfully run network tests in qemu with tap devices.
Next step are real hardware tests. There were some issues during initialization, but we were interrupted with the preparations for Corona.

Best regards,

   Jan

> Gedare is correct though. Work on the proposal next. You are far enough along
> with what you have done to been into debugging and integrating on a real
> embedded target. 
> 
> 
> Many greetings and stay healthy,
> 
> Everything here is closed with teleworking. Two confirmed cases here 
> in town so far but one was at NASA Marshall and the other was a
> pharmacist at the Army clinic. I expect we will have more in the near future.
> 
> --joel
>  
> Heinz
> ------------------------------------------------------------------------------
> Fritz-Haber-Institut    | Phone:         (+49 30) 8413-4270
> Heinz Junkes             | Fax (G3+G4):   (+49 30) 8413-5900
> Faradayweg 4-6        | VC: 102220181216 at bjn.vc
> D - 14195 Berlin        | E-Mail:        junkes at fhi-berlin.mpg.de
> ------------------------------------------------------------------------------
> 
> > On 18. Mar 2020, at 19:56, Mritunjay Sharma <mritunjaysharma394 at gmail.com> wrote:
> > 
> > Thank you so much for the feedback. I will research more on the targets that can run with the simulator and would like to tell you that looking at https://github.com/hjunkes/epicsBaseOwnPlayground/tree/7.0, I feel that @Heinz Junkes is already working on libbsd for RTEMS-MVME2500 and I would like to hear more from him to get some clarity on what can be done. Being a beginner, it's really a great welcome to the world of cross-compiler development and testing. :) Thank you for always helping me out! 
> > 
> > On Wed, Mar 18, 2020 at 11:51 PM Gedare Bloom <gedare at rtems.org> wrote:
> > On Wed, Mar 18, 2020 at 11:40 AM Mritunjay Sharma
> > <mritunjaysharma394 at gmail.com> wrote:
> > >
> > > Hi everyone,
> > >
> > > As a part of working on ticket #3835 (BSP Buildset for EPICS), this is to inform the community that after a lot of research, help from RTEMS and EPICS community and https://epics.mpg.de/index.php?n=Software.EPICSRTEMS?userlang=en, I was able to build EPICS 7 for RTEMS 5 for mvme6100 (powerpc-rtems5).
> > > After this, my development environment is almost completely set up with RTEMS, rtems-source-builder and epics7 (taken from https://github.com/hjunkes/epicsBaseOwnPlayground/tree/7.0 ). While going through the entire process, I learnt about the changes to be made in epics7/configure/os/CONFIG_STATUS.Common.RTEMS and also encountered few errors while the 'make' process, however, it got fixed by commenting line 325 to 329 in '/home/mritunjay/development/rtems/5/powerpc-rtems5/beatnik/lib/include/rtems/confdefs/libio.h':
> > > //  #ifdef CONFIGURE_FILESYSTEM_ENTRY_DEVFS
> > > //    CONFIGURE_FILESYSTEM_ENTRY_DEVFS,
> > > //  #endif
> > > I would like to know more on why this behaviour occurred. Is it a bug? Or something else?
> > >
> > Include some of this detail in your proposal. I don't have time at the
> > moment to delve into details.
> > 
> > Next step would be to identify a target you can build EPICS7+RTEMS5
> > with simulator support. This one is challenging since AFAIK there
> > haven't been any successes in that regards. Find out what are the
> > blockers and maybe you can work on those.
> > 
> > > I wanted to test it but I am facing issues in connecting with tftp server when I run 'scp bin/RTEMS-beatnik/libComTestHarness.boot root at epics:/var/lib/tftp'
> > > The following error was encountered:
> > > /usr/bin/ssh: /lib/x86_64-linux-gnu/libcrypto.so.1.0.0: no version information available (required by /usr/bin/ssh)
> > > /usr/bin/ssh: /lib/x86_64-linux-gnu/libcrypto.so.1.0.0: no version information available (required by /usr/bin/ssh)
> > > ssh: Could not resolve hostname epics: Name or service not known
> > > lost connection
> > > I was wondering what can be the reason behind this, please help me fix this.
> > >
> > Test it how -- do you have a powerpc MVME6100 board?
> > 
> > I'm guessing whoever gave you that scp command line has a board with a
> > local /etc/hosts entry for 'epics' pointing to some local IP address
> > they connect to their board over.
> > 
> > Welcome to the world of cross-compiler development and testing. :)
> > https://docs.rtems.org/branches/master/user/hosts/index.html#development-host
> > 
> > > Also, what are the next steps? I am working on the GSoC proposal currently and since the current ticket #3835 is a shorter project, I have to add certain similar sub-tasks within the GSoC timeline as suggested by @Gedare Bloom :
> > > "* compile CFS by hand, create an RSB "recipe" to build it, then
> > > integrate it with vertical buildset configuration.
> > > * create vertical buildsets for PC, Raspberry Pi, and other boards.
> > > ** EPICS community uses many powerpc boards so adding something there
> > > would be good."
> > 
> > Identify what might be potential simulator targets for you to use, and
> > look at how easy they are to build with RSB. If any are missing
> > support, work on adding them. If any are broken, poke into why.
> > 
> > psim is a good powerpc simulator, although it may not have enough
> > support for running epics--I haven't looked.
> > 
> > For EPICS it is best to get something working in qemu so that you can
> > run some  networking. So I'd look at some qemu simulator targets and
> > check what is their status and whether you can build and run anything
> > for RTEMS on them. After that, can you build/run any RTEMS networking
> > applications?
> > 
> > Does the EPICS7+RTEMS5 stack still using cpukit/libnetworking, or is
> > it able to use libbsd? That is an interesting question to find out.
> > 
> > > I would like to have suggestions on resetting the timeline as I work on refining the proposal for GSoC based on Dr. @Gedare Bloom's suggestions.
> > >
> > For timeline the first step is to break down all the tasks to figure
> > out what you need to do to accomplish each of your goals. Then you
> > should try to estimate how long you (or mentors) think each task will
> > take, and use that create your plan. (Estimating time is notoriously
> > difficult, but do your best. If it comes in too heavy or light, that
> > is fine, you should put some activities in "Future Improvements" that
> > can be moved into your summer work in case you get through your
> > planned work faster than you expect; if you don't complete all your
> > planned work, we understand and will take everything into
> > consideration during evaluation. Do NOT think of your plan as a
> > contract: it is not, it is a guideline for your summer.)
> > 
> > > Thanks
> > > Mritunjay
> > >