On Wed, 2014-05-14 at 19:57 +0200, Takashi Iwai wrote:

Hm, so it actually regressed on BE?

It's strange because fb_math_be() should be true and the patch won't
change the values in that case...

Shouldn't the patch be based on foreign endian being set rather than
just "be" anyway ?

IE. If the fb is LE and the host is LE we *also* don't want to change
the ordering, which will be the case when we fix qemu...

Cheers,
Ben.