Re: [RFC PATCH] mm: introduce kv[mz]alloc helpers

From: Michal Hocko
Date: Wed Dec 21 2016 - 03:45:21 EST


It seems that this email didn't get delivered due to some stupid gmail
spam policy. Let me try to repost via a different relay. Sorry to those
who have seen the original message and get a duplicate now.

On Wed 21-12-16 08:03:53, Michal Hocko wrote:
> On Tue 20-12-16 14:13:41, Andrew Morton wrote:
> > On Tue, 20 Dec 2016 09:38:22 -0800 Joe Perches <joe@xxxxxxxxxxx> wrote:
> >
> > > > So what are we going to do about this patch?
> > >
> > > Well if Andrew doesn't object again, it should probably be applied.
> > > Unless his silence here acts like a pocket-veto.
> > >
> > > Andrew? Anything to add?
> >
> > I guess we should give in to reality and do this, or something like it.
> > But Al said he was going to dig out some patches for us to consider?
>
> Al wanted to cover vmalloc GFP_NOFS context _inside_ the vmalloc
> code. This is mostly orthogonal to this patch I believe. Besides
> that I _think_ that it would be better to convert those vmalloc(NOFS)
> users to the scope api rather than tweak the vmalloc. One reason to go
> that way is that those vmalloc(NOFS) users need to be checked anyway
> and something tells me that some of them can really be changed to
> GFP_KERNEL.
>
> This helper is clear about its gfp mask expectation and complain loudly
> if somebody wants something unexpected which is a good start I believe.
> --
> Michal Hocko
> SUSE Labs

--
Michal Hocko
SUSE Labs