Powerpc IRQ handling breaks strict EABI compliance

[Till Straumann]

OK, I fixed the motorola/shared BSP to not clobber R2/R13 anymore.
However, the question remains:

  - who is responsible for the setup (calling __eabi()) ?
    RTEMS or application code?

-- Till

Joel Sherrill wrote:
> I don't think his original email was offlist so try the index
> by author at http://www.oarcorp.com/rtems/maillistArchives/rtems-users/
> 
> And the search engine works again. :)
> 
> --joel
> 
> Till Straumann wrote:
> 
>>Valette Eric wrote:
>>
>>>Till ,
>>>
>>>First let me thank you for pushing to harden the IRQ API. I think the
>>>discussion was interessting and lead to good remarks from many people.
>>>
>>>I write this email to say that Sergei Organov <mailto:osv at javad.ru> is
>>>right to say that I sould not have used R2 as a scratch register for IRQ
>>>handling as EABI require this register to be *read only*. So as you are
>>>going to make some modifications, can you care to replace R2 with a RX
>>>of your choice in the low level execption and irq handling code.
>>
>>OK, that should be done - unfortunately the wrong usage of R2 has
>>already spread out to all of the PPC BSPs :-(. I can only test on
>>motorola/shared, though.
>>
>>Since I'm going to get other patches ready, I might as well integrate
>>the R2 issue...
>>
>>BTW: I lost (or never had) Sergei's orignal message - would you
>>mind sending it to me?
>>
>>Regards
>>-- Till
>>
>>
>>>I can do it but cannot test it :-(
>>>
>>>Have a nice day and thanks for all the good work you have done.
>>>
>>
>