Philip Blundell writes:
> Well, hmm, I'm not sure. The problem, as you say, is backwards compatibility.
> There's already an EF_ALIGN8 flag, so in theory we could make 8-bit alignment
> the default tomorrow and nobody would suffer any silent lossage. (I'm not sure
> whether the ARM tools generate or respect this.) But, on the other hand,
> everyone would have to rebuild all their objects. I guess the way ahead is to
> make it a multilib option for now. I'll think about whether the default could
> (should) be changed for gcc 3.0.
And when you do, how do you handle the kernel dependency between new/old.
As I said, making this change is an all or nothing situation; you can't
have half a system using the old and the other half using the new.
_____
|_____| ------------------------------------------------- ---+---+-
| | Russell King [EMAIL PROTECTED] --- ---
| | | | http://www.arm.linux.org.uk/ / / |
| +-+-+ --- -+-
/ | THE developer of ARM Linux |+| /|\
/ | | | --- |
+-+-+ ------------------------------------------------- /\\\ |
_______________________________________________
http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm
