[OpenOCD-devel] OpenOCD with PPC architecture

[Lino - Andrea]

Hi,
I am a new OpenOCD developer and I am interested to extend the support in OpenOCD for PowerPC devices such as ST SPC56x. Is there an open project? Someone is already work on this porting?

Thanks in advance
Andrea

[Øyvind Harboe]

Post by Lino - Andrea
Hi,
I am a new OpenOCD developer and I am interested to extend the support in
OpenOCD for PowerPC devices such as ST SPC56x. Is there an open project?
Someone is already work on this porting?

The PowerPC CPU doesn't really support an open documented standard for
communicating with it's debug hardware does it?

I thought PowerPC's mostly just had a shift-register to access
internals and that you'd have to implement the entire debug logic
*outside* the PowerPC, right?

If so, then PowerPC debug support isn't suitable for being supported
by the opensource community.

That's PowerPC's loss, IMHO.

I haven't been following PowerPC. Are they used in mainstream embedded
MCUs still?

[David Riley]

Post by Øyvind Harboe
I haven't been following PowerPC. Are they used in mainstream embedded
MCUs still?

Sure are, especially when you need more muscle than an ARM (though they're catching up). They still employ VERY proprietary access methods over JTAG, though, and there's not much in common from generation to generation. My understanding is that you can't even get the COP specs from Freescale unless you're a big JTAG pod manufacturer; I don't know what the other PPC manufacturers' policies are. It's a shame, because I really like the PPC, but it's hard to justify spending $3k on a debug pod for projects my size.

- Dave

[CeDeROM]

Post by Øyvind Harboe
If so, then PowerPC debug support isn't suitable for being supported
by the opensource community.

How about this one https://www.power.org/ ? :-)

[Michael Schwingen]

Post by CeDeROM

Post by Øyvind Harboe
If so, then PowerPC debug support isn't suitable for being supported
by the opensource community.

How about this one https://www.power.org/ ? :-)

Where is the JTAG debug documentation on that site?

cu
Michael

[Lino - Andrea]

I would like to work on e300/e500 cores, but the debug interface seems =

to=20

be
non-documented for practical purposes.

I am interested to work on ST SPC560B and SPC560P devices. Assuming that =
I=20
have all the documentation, you think it's possible to extend the OpenOCD=
=20
compatibility?
Mainly, where to start?
The idea is to use a debugger based on the FT2232 chip with a new=20
development board based on PPC device. Moreover, I'm looking for a=20
collaboration, so I am ready to provide some board with this device.

Andrea=20

[Tomek CEDRO]

Post by Michael Schwingen
I would like to work on e300/e500 cores, but the debug interface seems to
be
non-documented for practical purposes.

I am interested to work on ST SPC560B and SPC560P devices. Assuming that I
have all the documentation, you think it's possible to extend the OpenOCD
compatibility?
Mainly, where to start?
The idea is to use a debugger based on the FT2232 chip with a new
development board based on PPC device. Moreover, I'm looking for a
collaboration, so I am ready to provide some board with this device.

Then you need to add new Target, name it PowerPC, you can start from
ARM or MIPS as reference.

Target is responsible for reading/writing target memory bus, access port, etc.

On top of your Target functions you can build perihperals access,
memory access, etc.

If you want to use FT2232 device, use new MPSSE interface driver in
OpenOCD, it already gives you the JTAG access.

Right now its a bit mixed, but looking on how ARM or MIPS is
implemented will be a good starting point.

If you have lots of spare boards I can take a look at one of them and
the documentation just for curiosity :-) However if you want to have
this implemented with a priority please consider hiring someone
developer from the group willing to do the job or assist you :-)

Best regards :-)
Tomek

[Lino - Andrea]

Post by Tomek CEDRO
If you have lots of spare boards I can take a look at one of them and
the documentation just for curiosity :-) However if you want to have
this implemented with a priority please consider hiring someone
developer from the group willing to do the job or assist you :-)

For now, I have not the possibility for hiring a developer, but I repeat, I
can only provide some PPC development board within 3 or 4 month. I'm trying
to obtain the debugger documentation and maybe I have good hopes. So, given
that I am a new OpenOCD developer, if someone can help me I would be
grateful.

[Michael Schwingen]

Post by Lino - Andrea

Post by Michael Schwingen
I would like to work on e300/e500 cores, but the debug interface
seems to be
non-documented for practical purposes.

I am interested to work on ST SPC560B and SPC560P devices. Assuming
that I have all the documentation, you think it's possible to extend
the OpenOCD compatibility?

Yes.

Post by Lino - Andrea
Mainly, where to start?

Look at the ARM and MIPS targets - MIPS is newer but a bit less mature than
ARM.

However, first work on getting the complete documentation for the debug
access: that might or might not be a problem.

when creating the new target, keep in mind that not all PowerPC chips are
the same: there are different cores, and SOCs using different debug
interfaces, so the structure should allow for these variations, even if the
code only handles one such variant.

Post by Lino - Andrea
The idea is to use a debugger based on the FT2232 chip with a new
development board based on PPC device. Moreover, I'm looking for a
collaboration, so I am ready to provide some board with this device.

The chip uses JTAG, so any JTAG adapter supported by OpenOCD *should* work.
What is interesting is that they claim NEXUS compatibility for the debug
interface, so there might be a chance to get documentation (the Freescale
devices I work with use COP, and the documentation is quite closed).

I work on MPC8314 and P1010 (Freescale), so I am afraid I won't be able to
help much.

cu
Michael

[Michael Schwingen]

I am not really interested in these devices, but the Power Project claims
to be open source, so if there is no debug documentation yet available
maybe it is a good idea to request one :-)

So you are just throwing us an unspecific URL without checking that it does
provide anything useful?

Your ample use of smilies does not make this any better - it feels like you
are making fun of me.

cu
Michael

[Tomek CEDRO]

Hey Michael. This may be the good starting point to obtain the
documentation as the project claims to be Open Source and there is some
community around it. Thats it, just a hint. You can give it a try, maybe
some nice Open Source PowerPC core is out there...
Best regards,
Tomek

--
CeDeROM, SQ7MHZ, http://www.tomek.cedro.info

Post by Michael Schwingen

I am not really interested in these devices, but the Power Project claims
to be open source, so if there is no debug documentation yet available
maybe it is a good idea to request one :-)

So you are just throwing us an unspecific URL without checking that it does
provide anything useful?
Your ample use of smilies does not make this any better - it feels like you
are making fun of me.
cu
Michael
------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
http://p.sf.net/sfu/learnnow-d2d
_______________________________________________
OpenOCD-devel mailing list
https://lists.sourceforge.net/lists/listinfo/openocd-devel

[Michael Schwingen]

Post by Tomek CEDRO
Hey Michael. This may be the good starting point to obtain the
documentation as the project claims to be Open Source and there is some
community around it. Thats it, just a hint. You can give it a try, maybe
some nice Open Source PowerPC core is out there...

You don't seem to understand.

power.org does not manufacture chips - they specify cores.

Manufacturers like Freescale, IBM etc. do make chips - and you need to ask
*them* for debugging specifications, since each implements it differently.

Even if I can get specifications about a free power.org core, that does not
help me in any way to debug real existing silicon that is sitting on my
desk.

cu
Michael

[CeDeROM]

On Fri, Jan 25, 2013 at 8:46 AM, Michael Schwingen

Post by Michael Schwingen
power.org does not manufacture chips - they specify cores.
Manufacturers like Freescale, IBM etc. do make chips - and you need to ask
*them* for debugging specifications, since each implements it differently.
Even if I can get specifications about a free power.org core, that does not
help me in any way to debug real existing silicon that is sitting on my
desk.

Hey Michael :-) Uhm, thats bad, I didnt know, Im not really familiar
with the PowerPC, but it seems nonsense that there is no common debug
standard like in MIPS or ARM :-( Isnt there ANY common part that could
be obtained from power.org or make them publish some parts of the real
silicon and then work out the difference? I know this is kind of alone
in the dark but I hoped power.org could help anything..

Best regards,
Tomek

[Lino - Andrea]

Hi,
I was able to obtain the documentation about the cores of the FreeScale P=
PC=20
microcontrollers. In them a section on the debug unit is reported.
PowerPC does not define a debug standard but apparently e200zX cores do, =
see=20
the following documents. See the =E2=80=9Cdebug=E2=80=9D section.
http://cache.freescale.com/files/32bit/doc/ref_manual/e200z0RM.pdf
http://cache.freescale.com/files/32bit/doc/ref_manual/e200z3RM.pdf
http://cache.freescale.com/files/32bit/doc/ref_manual/e200z4RM.pdf

The cores are increasingly popular and now there are also low cost device=
s=20
and boards. Still not mainstream but now there are two vendors supporting=
=20
them: Freescale and ST.

I'm trying to obtain some development kits (board plus a debug probe base=
d=20
on ft2232) to provide to someone that is interested to work on this proje=
ct.

Thanks=20

[Michael Schwingen]

Post by Øyvind Harboe
The PowerPC CPU doesn't really support an open documented standard for
communicating with it's debug hardware does it?

No, AFAIK not. There was free documentation for the Motorola MPC860 - at a
former employer, we implemented our own gdb backend for that, and there was
an open-source solution.

I would like to work on e300/e500 cores, but the debug interface seems to be
non-documented for practical purposes.

Post by Øyvind Harboe
I haven't been following PowerPC. Are they used in mainstream embedded
MCUs still?

Yes - we use them at work. The Freescale P1010 is a really nice device.

cu
Michael

[m***@tin.it]

Hi,
I was able to find some document on the PPC e200z core used in the SPC5 microcontrollers. PowerPC does not define a debug standard but apparently e200zX cores do, see the following documents.

http://cache.freescale.com/files/32bit/doc/ref_manual/e200z0RM.pdfhttp://cache.freescale.com/files/32bit/doc/ref_manual/e200z3RM.pdfhttp://cache.freescale.com/files/32bit/doc/ref_manual/e200z4RM.pdf
See the “debug” section. The cores are increasingly popular and now there are also low cost devices and boards. Still not mainstream but now there are two vendors supporting them: Freescale and ST. So, the idea is to integrate a standard FTDI debugger on a PPC development board. I would be happy, if someone can help me in this project.

Andrea

[CeDeROM]

[unknown]

Post by m***@tin.it
The cores are increasingly popular and now there are also low cost devices
and boards. Still not mainstream but now there are two vendors supporting
them: Freescale and ST. So, the idea is to integrate a standard FTDI
debugger on a PPC development board. I would be happy, if someone can help
me in this project.

Hey hey :-) A large and important discussion takes place right now on
the future directions of the OpenOCD project and internals redesign.
Please wait until consensus is acheved, then you should know exactly
how to implement new target and maybe there will be people willing to
help you out and test new architecture at the same time. At least you
won't get 99% blockers on your commits :-)

Best regards :-)
Tomek