Re: [PATCH v1 06/11] mm: support GUP-triggered unsharing via FAULT_FLAG_UNSHARE (!hugetlb)

From
David Hildenbrand <hidden>
Date
2021-12-17 20:55:42
Also in
linux-kselftest, linux-mm, lkml
Reply to
Linus Torvalds, 2021-12-17
On 17.12.21 21:51, Linus Torvalds wrote:
On Fri, Dec 17, 2021 at 12:45 PM David Hildenbrand [off-list ref] wrote:
quoted
If a page is mapped exactly once, page_mapcount(page) == 1 and there is
nothing to do.
Why?

You state that, but you stating that doesn't magically make it so.

What makes "mapcount==1" stable and special? Your "it's an
atomic_read()" argument is nonsense - it implies that the count can be
changing, but you will get _one_ answer.
And I explained how it can not increment. And the only way is via
fork(), which cannot run concurrently.
What makes that one answer of a changing count special?

What if there are other references to that same page, gotten with
vmsplice(), and just about to be mapped into another address space?
If we have a shared anonymous page we cannot have GUP references, not
even R/O ones. Because GUP would have unshared and copied the page,
resulting in a R/O mapped anonymous page.

What am I missing?


-- 
Thanks,

David / dhildenb
Thread (137 messages) 137 messages, 11 authors, last reply 2022-01-04
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help