On Thu, 15 Oct 2009, Benjamin Herrenschmidt wrote:

For some weird reason, our gcc until 4.3 (fixed in 4.3) had the weird
idea that the alignment attribute should not be allowed to force an
alignment greater than 32k. If attempted, it would warn -and- crop the
alignment to 32k.

[...]

This has a few issues for us:

	- The patch that converted bits of powerpc to the new macro break since
it now hits that bug

Hi Ben,

Just to make sure I understand the nature of the problem, is the current 
breakage that gcc < 4.3 will _warn_ on any compilation units on ppc64 that 
use __page_aligned data, or something worse?

The cropping is clearly a potential problem, but I read the rest of your 
email as saying that the cropping of the alignment isn't actually a 
problem with the current kernel because the kernel is currently only using 
the macro with things whose size is divisible by PAGE_SIZE.  However, I am 
not sure how to reconcile that with using the word "break" above...

	-Tim Abbott