]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
arm64: mte: Do not warn if the page is already tagged in copy_highpage()
authorCatalin Marinas <catalin.marinas@arm.com>
Sun, 26 Oct 2025 18:05:45 +0000 (14:05 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 29 Oct 2025 13:09:02 +0000 (14:09 +0100)
[ Upstream commit b98c94eed4a975e0c80b7e90a649a46967376f58 ]

The arm64 copy_highpage() assumes that the destination page is newly
allocated and not MTE-tagged (PG_mte_tagged unset) and warns
accordingly. However, following commit 060913999d7a ("mm: migrate:
support poisoned recover from migrate folio"), folio_mc_copy() is called
before __folio_migrate_mapping(). If the latter fails (-EAGAIN), the
copy will be done again to the same destination page. Since
copy_highpage() already set the PG_mte_tagged flag, this second copy
will warn.

Replace the WARN_ON_ONCE(page already tagged) in the arm64
copy_highpage() with a comment.

Reported-by: syzbot+d1974fc28545a3e6218b@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/68dda1ae.a00a0220.102ee.0065.GAE@google.com
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Will Deacon <will@kernel.org>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: stable@vger.kernel.org # 6.12.x
Reviewed-by: Yang Shi <yang@os.amperecomputing.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
[ omitted hugetlb MTE changes ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/arm64/mm/copypage.c

index a7bb20055ce0948a6f586e29fd1a05bbbc1cc75f..9e734d6314e03e586f27cca7ccb64fb59a9b2fb7 100644 (file)
@@ -25,8 +25,13 @@ void copy_highpage(struct page *to, struct page *from)
                page_kasan_tag_reset(to);
 
        if (system_supports_mte() && page_mte_tagged(from)) {
-               /* It's a new page, shouldn't have been tagged yet */
-               WARN_ON_ONCE(!try_page_mte_tagging(to));
+               /*
+                * Most of the time it's a new page that shouldn't have been
+                * tagged yet. However, folio migration can end up reusing the
+                * same page without untagging it. Ignore the warning if the
+                * page is already tagged.
+                */
+               try_page_mte_tagging(to);
                mte_copy_page_tags(kto, kfrom);
                set_page_mte_tagged(to);
        }