On Sat, Jul 30, 2016 at 8:27 PM, Michael Ellerman <[email protected]> wrote:
>
> Dropping NOFAIL means we need to handle allocation failures, which makes
> the patch a bit bigger, and less of a pure fix.
Hmm. If you get allocation failures for something like this at init
time, I think you're basically screwed anyway.
And I really only meant for the initial array allocation, and the
__GFP_ZERO thing.
Yes, __GFP_ZERO does work for kmalloc() too, but unlike the other GPF
flags, we do have special zalloc versions for zeroing that are
generally preferred.
Linus
