Message-ID: <47c51c9a-140f-1ea1-b692-c4bae5d1fa58@google.com>
Date: Sun, 31 Aug 2025 02:05:56 -0700 (PDT)
From: Hugh Dickins <hughd@...gle.com>
To: Andrew Morton <akpm@...ux-foundation.org>
cc: Will Deacon <will@...nel.org>, David Hildenbrand <david@...hat.com>, 
    Shivank Garg <shivankg@....com>, Matthew Wilcox <willy@...radead.org>, 
    Christoph Hellwig <hch@...radead.org>, Keir Fraser <keirf@...gle.com>, 
    Jason Gunthorpe <jgg@...pe.ca>, John Hubbard <jhubbard@...dia.com>, 
    Frederick Mayle <fmayle@...gle.com>, Peter Xu <peterx@...hat.com>, 
    "Aneesh Kumar K.V" <aneesh.kumar@...nel.org>, 
    Johannes Weiner <hannes@...xchg.org>, Vlastimil Babka <vbabka@...e.cz>, 
    Alexander Krabler <Alexander.Krabler@...a.com>, 
    Ge Yang <yangge1116@....com>, Li Zhe <lizhe.67@...edance.com>, 
    Chris Li <chrisl@...nel.org>, Yu Zhao <yuzhao@...gle.com>, 
    Axel Rasmussen <axelrasmussen@...gle.com>, 
    Yuanchu Xie <yuanchu@...gle.com>, Wei Xu <weixugc@...gle.com>, 
    Konstantin Khlebnikov <koct9i@...il.com>, linux-kernel@...r.kernel.org, 
    linux-mm@...ck.org
Subject: [PATCH 2/7] mm/gup: check ref_count instead of lru before
 migration

Will Deacon reports:-

When taking a longterm GUP pin via pin_user_pages(),
__gup_longterm_locked() tries to migrate target folios that should not
be longterm pinned, for example because they reside in a CMA region or
movable zone. This is done by first pinning all of the target folios
anyway, collecting all of the longterm-unpinnable target folios into a
list, dropping the pins that were just taken and finally handing the
list off to migrate_pages() for the actual migration.

It is critically important that no unexpected references are held on the
folios being migrated, otherwise the migration will fail and
pin_user_pages() will return -ENOMEM to its caller. Unfortunately, it is
relatively easy to observe migration failures when running pKVM (which
uses pin_user_pages() on crosvm's virtual address space to resolve
stage-2 page faults from the guest) on a 6.15-based Pixel 6 device and
this results in the VM terminating prematurely.

In the failure case, 'crosvm' has called mlock(MLOCK_ONFAULT) on its
mapping of guest memory prior to the pinning. Subsequently, when
pin_user_pages() walks the page-table, the relevant 'pte' is not
present and so the faulting logic allocates a new folio, mlocks it
with mlock_folio() and maps it in the page-table.

Since commit 2fbb0c10d1e8 ("mm/munlock: mlock_page() munlock_page()
batch by pagevec"), mlock/munlock operations on a folio (formerly page),
are deferred. For example, mlock_folio() takes an additional reference
on the target folio before placing it into a per-cpu 'folio_batch' for
later processing by mlock_folio_batch(), which drops the refcount once
the operation is complete. Processing of the batches is coupled with
the LRU batch logic and can be forcefully drained with
lru_add_drain_all() but as long as a folio remains unprocessed on the
batch, its refcount will be elevated.

This deferred batching therefore interacts poorly with the pKVM pinning
scenario as we can find ourselves in a situation where the migration
code fails to migrate a folio due to the elevated refcount from the
pending mlock operation.

Hugh Dickins adds:-

!folio_test_lru() has never been a very reliable way to tell if an
lru_add_drain_all() is worth calling, to remove LRU cache references
to make the folio migratable: the LRU flag may be set even while the
folio is held with an extra reference in a per-CPU LRU cache.

5.18 commit 2fbb0c10d1e8 may have made it more unreliable.  Then 6.11
commit 33dfe9204f29 ("mm/gup: clear the LRU flag of a page before adding
to LRU batch") tried to make it reliable, by moving LRU flag clearing;
but missed the mlock/munlock batches, so still unreliable as reported.

And it turns out to be difficult to extend 33dfe9204f29's LRU flag
clearing to the mlock/munlock batches: if they do benefit from batching,
mlock/munlock cannot be so effective when easily suppressed while !LRU.

Instead, switch to an expected ref_count check, which was more reliable
all along: some more false positives (unhelpful drains) than before, and
never a guarantee that the folio will prove migratable, but better.

Note for stable backports: requires 6.16 commit 86ebd50224c0 ("mm:
add folio_expected_ref_count() for reference count calculation") and
6.17 commit ("mm: fix folio_expected_ref_count() when PG_private_2").

Reported-by: Will Deacon <will@...nel.org>
Link: https://lore.kernel.org/linux-mm/20250815101858.24352-1-will@kernel.org/
Fixes: 9a4e9f3b2d73 ("mm: update get_user_pages_longterm to migrate pages allocated from CMA region")
Signed-off-by: Hugh Dickins <hughd@...gle.com>
Cc: <stable@...r.kernel.org>
---
 mm/gup.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/gup.c b/mm/gup.c
index adffe663594d..82aec6443c0a 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2307,7 +2307,8 @@ static unsigned long collect_longterm_unpinnable_folios(
 			continue;
 		}
 
-		if (!folio_test_lru(folio) && drain_allow) {
+		if (drain_allow && folio_ref_count(folio) !=
+				   folio_expected_ref_count(folio) + 1) {
 			lru_add_drain_all();
 			drain_allow = false;
 		}
-- 
2.51.0

Powered by blists - more mailing lists