On 8/29/20 7:54 AM, Christoph Hellwig wrote:

On Sat, Aug 29, 2020 at 01:08:51AM -0700, John Hubbard wrote:

quoted

pin_user_page() is the FOLL_PIN equivalent of get_page().

This was always a missing piece of the pin/unpin API calls (early
reviewers of pin_user_pages() asked about it, in fact), but until now,
it just wasn't needed. Finally though, now that the Direct IO pieces in
block/bio are about to be converted to use FOLL_PIN, it turns out that
there are some cases in which get_page() and get_user_pages_fast() were
both used. Converting those sites requires a drop-in replacement for
get_page(), which this patch supplies.

I find the name really confusing vs pin_user_pages*, as it suggests as
single version of the same.  It also seems partially wrong, at least
in the direct I/O case as the only thing pinned here is the zero page.

So maybe pin_kernel_page is a better name together with an explanation?
Or just pin_page?

Yes. Both its behavior and use are closer to get_page() than it is to
get_user_pages(). So pin_page() is just right.


thanks,
-- 
John Hubbard
NVIDIA