Dear Joakim Tjernlund,
In message OFBC9C03BC.436C27C7-ONC1257877.00797190-C1257877.007A0FD7@transmode.se you wrote:
Yes, but you yorself pointed out that commit 337f5f5 missed a large number of boards, leaving the tree in a inconsistent state. Should we not revert that one as well?
It is not too bad, just not complete yet. Reverting all just makes it harder to get it all done. I still don't get why it broke really. hmm, did you by any chance include my patches to gcc too? if so I you should only have to fixup arch/powerpc/config.mk:
- PLATFORM_RELFLAGS += -fpic -mrelocatable -ffunction-sections -fdata-sections
- PLATFORM_RELFLAGS += $(call cc-option,-msingle-pic-base,)
- PLATFORM_RELFLAGS += -fPIC -mrelocatable -ffunction-sections -fdata-sections
Can you please have a look at this yourslef? You understand much better than me which of your patches are doing what, and how they might interact.
I don't have the time to dig into this any deeper. If we cannot find a quick solution, I suggest to back out all this stuff and redo once the problems have been sorted out. Sorry for applying the stuff to mainline without more thorough testing, I should have noticed these issues earlier and never checked in all that stuff.
Best regards,
Wolfgang Denk