Re: [-mm PATCH v2 23/25] mm, x86: get_user_pages() for dax mappings

[Dan Williams <dan.j.williams@intel.com>]

On Tue, Dec 15, 2015 at 6:18 PM, Dan Williams <dan.j.williams@intel.com> wrote:
> On Tue, Dec 15, 2015 at 4:14 PM, Andrew Morton
> <akpm@linux-foundation.org> wrote:
[..]
>> And again, this is bloating up the kernel for not-widely-used stuff.
>
> I suspect the ability to compile it out is little comfort since we're
> looking to get CONFIG_ZONE_DEVICE enabled by default in major distros.
> If that's the case I'm wiling to entertain the coarse pinning route.
> We can always circle back for the finer grained option if a problem
> arises, but let me know if CONFIG_ZONE_DEVICE=n was all you were
> looking for...

I chatted with Dave Hansen a bit and we're thinking that just moving
the zone_device count updates out of line and marking the branch
unlikely should address this concern.

In fact my initial numbers are showing that moving the call to
"percpu_ref_get(page->pgmap->ref)" out of line saves nearly 24K of
text!

If that's not enough there's always jump labels, but that's likely to
show diminishing returns compared to the size reduction of moving the
percpu_refcount update out of line.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>