Re: [PATCH] drm: use kvmalloc_array for drm_malloc*

[Michal Hocko]

On Tue 16-05-17 15:08:56, Daniel Vetter wrote:
> On Tue, May 16, 2017 at 11:52:55AM +0200, Michal Hocko wrote:
> > On Tue 16-05-17 11:22:30, Daniel Vetter wrote:
> > > On Tue, May 16, 2017 at 11:06:06AM +0200, Michal Hocko wrote:
> > > > From: Michal Hocko <mhocko@xxxxxxxx>
> > > > 
> > > > drm_malloc* has grown their own kmalloc with vmalloc fallback
> > > > implementations. MM has grown kvmalloc* helpers in the meantime. Let's
> > > > use those because it a) reduces the code and b) MM has a better idea
> > > > how to implement fallbacks (e.g. do not vmalloc before kmalloc is tried
> > > > with __GFP_NORETRY).
> > > > 
> > > > Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> > > 
> > > Shouldn't we go one step further and just remove these wrappers, maybe
> > > with cocci?
> > 
> > my cocci sucks...
> > 
> > > Especially drm_malloc_gfp is surpremely pointless after this
> > > patch (and drm_malloc_ab probably not that useful either).
> > 
> > So what about the following instead? It passes allyesconfig compilation.
> 
> Yeah, looks good, but perhaps rebased onto your first patch. That way we
> split the functional change from the refactor (not the first time innocent
> looking changes in i915 gem code resulted in surprises).

OK, I will split it.

> Your patch also seems to need some stuff from -rc1, and atm drm-misc is
> still pre-rc1, so I'll pull both patches in once that's sorted (I can do
> the rebase myself, since it's rather trivial). But pls remind me in case
> it falls through the cracks and isn't in linux-next by end of this week
> :-)

I have based it on top of the current linux next (next-20170516). Let me
know if other tree is more appropriate.
-- 
Michal Hocko
SUSE Labs