[GSoC] Porting RTEMS to OpenRISC - Status Report 1

[Hesham Moustafa]

On Fri, May 2, 2014 at 11:40 PM, Joel Sherrill
<joel.sherrill at oarcorp.com> wrote:
>
> On 5/1/2014 10:17 PM, Hesham Moustafa wrote:
>> Hi,
>>
>> This is a status report of the project during last week.
>>
>> 1- newlib: is ported and gcc builds successfully. The only feedback I
>> got when I posted the newlib patch [1] is about licence of setjmp.S
>> file. I contacted Damjan Lampret, the author of the file, asking him
>> to change the licence as Ralf suggested but I got no reply yet. If
>> changing the licence has not been feasible, can I write a similar
>> setjmp.S one?
> As Chris points out, this needs to be discussed on the newlib mailing list.
>
> I suppose given a working simulator and or1k-elf toolset w/o setjmp,
> anyone could write one.
>
> But setjmp is not a requirement for building RTEMS. It can be deferred
> for a while while this is worked through.  [NOTE: there may be one
> or two tests which use setjmp. Can't remember.]
>
That's what I was thinking of. I will ask on the mailing list about
that. Meanwhile, I will move to another task.
> Damjan may have an old mailing list post for the code originally which
> makes the license clear.
>
> Also please cc me on all correspondence like that. It may not help much
> but I at least show up in Google and have some FLOSS credibility via RTEMS,
>  the GCC Steering Committee, and the Network Time Foundation.
>
Sure.
> If it has been more than a week, then it may be worth a second ping.
> Did you use damjanl at lampret.com?
Yes I used this e-email. It has been two days since I e-mailed him (April 30).
>>
>> 2- toolchain: is all built now providing or1k-rtems4.11-* tools. GDB
>> is built too but from or1k-src repo. Should I create a patch for it
>> against upsteam binutils-gdb cvs? If yes, there are two options: the
>> first is to add a patch for or1ksim library to be embedded into gdb
>> and used when typing "target sim" command, the other option is to
>> connect gdb remotely to or1ksim simulator (target remote :xxxxx).
>> Which one to implement?
> This is really the decision of the or1k tools folks plus the gdb folks.
> Whatever
> they think is the way they want it to work is what we will end up stuck with
> long term.
>
> Either way, the RSB must give you a GDB and simulator.
>
> Help or1k folks push to their solution here. Your effort is on the port.
>>
>> 3- RTEMS: I started investigating the old or32 port for RTEMS, that
>> led me to add few files and configurations for or1k cpu and or1ksim
>> BSP for testing purposes. I was trying to get a libbspstart.a and link
>> it with other standalone bare-metal main.c program, thus, I
>> familiarize my self with the process of adding a new architecture, and
>> have some code to debug.
> In the past when doing BSPs for simulators, I have used the linker
> script, start
> assembly and other helper bits and pieces from the bare metal examples.
> But if the license is unacceptable, then it can only be used as reference.
>
> As a first cut, look at the h8sim or shsim. Both as VERY simple and may
> be good starting points for polled console, simple clock tick BSP to get
> started with. For those, I can show you the code in libgloss they
> leveraged.
Time to move to RTEMS and porting task? No other tool-chain work
needed (other than newlib)? If so, we need to discuss about the old
or32 code, whether to re-use it and make it fit for RTEMS, or to begin
from scratch. What should I begin with, cpukit/score/cpu? or BSP code?
I would appreciate any code examples, documentations and/or details
about that.
>>
>> I would appreciate your feedback and guidance.
>>
>> [1] http://www.rtems.org/pipermail/rtems-devel/2014-April/006598.html
>>
>> Thanks,
>> Hesham
>
> --
> Joel Sherrill, Ph.D.             Director of Research & Development
> joel.sherrill at OARcorp.com        On-Line Applications Research
> Ask me about RTEMS: a free RTOS  Huntsville AL 35805
> Support Available                (256) 722-9985
>