On Wed, Apr 17, 2019 at 01:09:22AM +0300, Boaz Harrosh wrote: > On 16/04/19 22:57, Jerome Glisse wrote: > <> > > > > A very long thread on this: > > > > https://lkml.org/lkml/2018/12/3/1128 > > > > especialy all the reply to this first one > > > > There is also: > > > > https://lkml.org/lkml/2019/3/26/1395 > > https://lwn.net/Articles/753027/ > > > > OK I have re-read this patchset and a little bit of the threads above (not all) > > As I understand the long term plan is to keep two separate ref-counts one > for GUP-ref and one for the regular page-state/ownership ref. > Currently looking at page-ref we do not know if we have a GUP currently held. > With the new plan we can (Still not sure what's the full plan with this new info) > > But if you make it such as the first GUP-ref also takes a page_ref and the > last GUp-dec also does put_page. Then the all of these becomes a matter of > matching every call to get_user_pages or iov_iter_get_pages() with a new > put_user_pages or iov_iter_put_pages(). So sorry forgot to answer that part. So idea is to do: GUP() { ... - page_ref_inc(page); + page_ref_add(page, GUP_BIAS); ... } with GUP_BIAS = 1024 or something big but not too big to avoid risk of overflow by GUP. Then put_user_page() just ref_sub instead of ref_dec the same amount. We can have false GUP positive if a page is map so many time or reference so many time that its refcount reach the GUP_BIAS value but considering such page as GUPed should not be too harmful (not more harmful than what we do with GUPed page). So we want to call put_user_page() for GUPed page and only GUPed page so that we keep the reference count properly balance. Cheers, Jérôme
