Dave Jones wrote: > Just one comment. You moved quite a few of the preempt_disable/enable > pairs outside of the CONFIG_SMP checks. The issue we're working against > here is to try and prevent preemption and ending up on a different CPU. > As this cannot happen if CONFIG_SMP=n, I don't see why you've done this. Andrew Morton wrote: > Just in two places. [snip] Yes, thanks for delivering this better patch! My approach wanted just to be the most simple possibility... ;-) Thomas