TMS570LC4357 support in RTEMS

[Pavel Pisa]

Hello Juan and Gedare,

> On Sat, Feb 13, 2021 at 5:32 AM Juan Solano <jsm at jsolano.com> wrote:
> > I have seen an old thread of yours where you mention a BSP that works
> > with the TMS570LC4357. Is that available somewhere? Are there many
> > modifications needed to the standard tms570 bsp?
> >
> > I am working with a launchpad TI board and I would appreciate any
> > existing code to make it easier to use this.

I have invested considerable effort to TMS570 BSP to happen.
Unfortunately large amount of TMS570 HW/SW work which I have not
rip away and reimplemented again from scratch with my studnets
for RTEMS, rottens in the server and udder licenses claimed to
be owened by my former head, they probably continue to sell it
(in the fact I have found that they do not have right, because
no agreement about license transfer has been signed with our faculty).
But the work rottents in their new institute. 

As for the TMS570LC4357, I have generated PINMUX headers
and algorithms initially deigned for TMS570LS3137 are
extended to support TMS570LC4357 dual pinmux.

  https://git.rtems.org/rtems/tree/bsps/arm/tms570/include/bsp/tms570lc4357-pins.h

The IRI group of Johann Wolfgang Goethe-Universität Frankfurt
used RTEMS TMS570 BSP as base for their TMS570LC4357 projects.

  https://github.com/jalmito/rtems

Their local GitLab repo

  https://gitlab.iri.uni-frankfurt.de/jalm/rtems

But I have not registered their interrest to help with mainlining
of their work. I have analyzed diffs but only lightly.
It would be great if they can describe what works
and where are problems.

Generally, modifications to get BSP to state which runs on TMS570LC4357
is relatively small. RTEMS mainline is prepared even to use HalCoGen
generated files to startup the chip. Problem is that chips differs
sometimes even in the corresponding module registers between the
chips. And I think that it is unmaintainable (at least for small
group as RTEMS developres are) to maintain similar geerated and manually
tuned headers mess as HalCoGen provides (sometimes same bits labelled
different way between chips etc.). I think that for RTEMS it is necessary
to make clear abstractions and headers which would work on whole chips
families.

As for the actual headers versions and XML register description files,
they done big step forward and license is changed and (I believe)
compatible with RTEMS.

As for HW, I have TMDX570LC43HDK kit at home and lauchpad
kits are available at https://www.elektroline.cz/ where
I provide consultations. But they are using smaller TMS570
members for now and in SIL3 applications they try to keep
whole firmware in the form of single C loop and some DMAs
without OS.

There is other problem with TMS570 for me. I do not
like much CCS and my patches for OpenOCD support
are based on old OpenOCD branch and third party
base patches which did not found the way into mainline.

  https://cmp.felk.cvut.cz/~pisa/tms570/openocd-patches/

They support flashing only on TMS570LS3137.
Have somebody some experience with open tools
for TMS570LC4357 debugging and flashing.
By the way I have helped one of my students to
prepare nice XCP over Ethernet bootloader for
the TMS570 family. But it rottents forbidden
to share on the server of my frmer group as well.

Generally I have interrest to see RTEMS sunning
on TMDX570LC43HDK, it could work well with our port
of LwIP as well. But even its integration into RTEMS
mainline means considerable work. I am not sure if
I find time till summer, we have 300+ studnest in distance
teaching class again. I would try to help with consultations
as my time allows if there is interrest. If I do not reply
in week, please send reminder directly to me...

I plan to offer RTEMS GSoC to our students as each year,
so may it be somebody will have interrest in some RTEMS
porting, LwIP integration, etc...

Best wishes,

                Pavel Pisa
    phone:      +420 603531357
    e-mail:     pisa at cmp.felk.cvut.cz
    Department of Control Engineering FEE CVUT
    Karlovo namesti 13, 121 35, Prague 2
    university: http://dce.fel.cvut.cz/
    personal:   http://cmp.felk.cvut.cz/~pisa
    projects:   https://www.openhub.net/accounts/ppisa
    CAN related:http://canbus.pages.fel.cvut.cz/



On Sunday 21 of February 2021 08:40:01 Gedare Bloom wrote:
> Pavel had expressed some interest in this direction before, but I
> don't know that anyone has pushed this board far enough to commit a
> working BSP.
>
> I have this hardware available (Hercules Launchpad LAUNCHXL2-570LC43).
> But I haven't had any time to try to work with it. Unfortunately, when
> I looked at the halcogen stuff it was not license-compatible to RTEMS,
> but maybe that has changed. A lot of effort was put in the TMS570 port
> before to avoid using halcogen.
>
> I guess one reason this board may not be that widely used is because
> it is a pain to work with. The dual lockstep mode is great for
> production safety board, but quite terrible for development and
> debugging.
>
> Gedare
>