kernel/linux.git/mm, branch v6.6.155

mm/huge_memory: fix huge_zero_pfn race

2026-08-23T12:20:05+00:00

commit 33192a26cddea7a7e4ca66e5c3eebd36fa8be2bb upstream. Patch series "mm/huge_memory: fix huge_zero_pfn race", v2. There is a subtle race in the reference-counted huge_zero_folio implementation. The fast path atomic logic fails to account for the fact that the shrinker (which drops the final huge_zero_refcount pin) can overwrite huge_zero_pfn with the ~0UL sentinel value in shrink_huge_zero_folio_scan() after a racing get_huge_zero_folio() installed a valid value there. This results in huge_zero_folio being correctly set but huge_zero_pfn being set incorrectly and thus is_huge_zero_pfn() and consequently is_huge_zero_pmd() will misidentify the huge zero folio as being an ordinary THP folio. This can result in the huge zero folio being split and otherwise treated incorrectly. The solution to this is very subtle as there is an atomic fast path, and thus ordering in weakly ordered architectures has to be treated very carefully. The first commit fixes the issue by introducing a spinlock around huge_zero_[pfn, folio, refcount] write, with careful consideration paid to load/store ordering in the fast path. It is placed first and kept as small as possible so that it can be backported on its own. The second commit is a pure cleanup which reworks the CONFIG_PERSISTENT_HUGE_ZERO_FOLIO logic to better separate the persistent logic from the dynamically allocated one. This patch (of 2): If !CONFIG_PERSISTENT_HUGE_ZERO_FOLIO, the huge_zero_folio is refcounted by huge_zero_refcount and returned by mm_get_huge_zero_folio(). When the caller is done with the huge zero page, its reference count is decremented. Only a shrinker can set the reference count to zero. A race can unfortunately occur between a shrinker decrementing the reference count to zero and a concurrent page fault. This is because shrink_huge_zero_folio_scan() might, if very unlucky, be preempted between setting huge_zero_refcount to zero and writing an invalid value. During this time get_huge_zero_folio() could write to huge_zero_pfn before shrink_huge_zero_folio_scan() resumes. In this event the huge zero folio will be persistently misidentified causing the THP code path to be entered inappropriately for the huge zero folio: CPU 0 CPU 1 =======================================|================================= shrink_huge_zero_folio_scan() | atomic_cmpxchg() sets refcount to 0 | xchg() sets huge_zero_folio to NULL | get_huge_zero_folio() | | atomic_inc_not_zero() -> zero preempted for a long time | Allocate new huge zero folio | | Write valid huge_zero_folio v | Write valid huge_zero_pfn Overwrite huge_zero_pfn with ~0UL <--- Invalid overwrite! This results in is_huge_zero_pfn() and is_huge_zero_pmd() incorrectly returning false for a huge zero page which could result in issues like the huge zero folio being incorrectly split. Note that the issue is with huge_zero_pfn not huge_zero_folio, as get_huge_zero_folio() uses cmpxchg() gated on huge_zero_folio being NULL with a retry loop and shrink_huge_zero_folio_scan() uses xchg() to set huge_zero_folio. Fix the issue by introducing a spinlock, huge_zero_lock, to prevent concurrent write of huge_zero_folio, huge_zero_pfn and huge_zero_refcount. There needs to be significant care taken here to ensure correctness: The fast path in get_huge_zero_folio() uses atomic_inc_not_zero(), which is outside of the critical section, and means huge zero allocation is gated on zero huge_zero_refcount. The fast path doesn't use huge_zero_lock, so the critical section is irrelevant to it. So invariants are required - huge_zero_refcount MUST: * Only be set in the huge_zero_lock critical section to ensure serialisation of huge_zero_pfn, huge_zero_folio and huge_zero_refcount writes. * Be set non-zero only AFTER huge_zero_[pfn, folio] are set to valid values so installation of the huge zero folio on read page fault ensures concurrent is_huge_zero_*() calls correctly identify the huge zero folio. * Be set zero only BEFORE huge_zero_[pfn, folio] are set to NULL and ~0UL respectively, and atomically. Establish these by: * Only setting huge_zero_refcount to zero or an absolute value in the huge_zero_lock critical section in get_huge_zero_folio() and shrink_huge_zero_folio_scan(), and always updating atomically there and elsewhere. * Using atomic_set_release(&huge_zero_refcount) in get_huge_zero_folio() after huge_zero_[pfn, folio] are set. This is paired with atomic_inc_not_zero() to ensure atomic_inc_not_zero() only observes a non-zero value if huge_zero_[pfn, folio] are set. * Using atomic_cmpxchg() in shrink_huge_zero_folio_scan() (as before) to ensure that it is set zero only when equal to 1 and set atomically. * atomic_cmpxchg() being fully ordered ensures this is done prior to huge_zero_[folio, pfn] being set to NULL and ~0UL respectively. Eliminate the retry loop in get_huge_zero_folio() as the atomic_cmpxchg() in shrink_huge_zero_folio_scan() is now performed under the lock, and replace with an equally locked atomic_inc() to set the reference count should the caller be raced on huge zero folio installation. folio_put() naturally implies a full memory barrier so its ordering is maintained correctly. The huge zero folio also cannot be released except when the shrinker does so as it is non-LRU and non-rmappable. Note that only the huge zero shrinker (via shrink_huge_zero_folio_scan()) can actually set huge_zero_refcount to zero, which is the count of mm's which have at least one huge zero folio installed plus one shrinker pin. Additionally convert a BUG_ON() to a VM_WARN_ON_ONCE(). Link: https://lore.kernel.org/20260730-fix-refcounted-huge-zero-v2-0-c5d8a41b317f@kernel.org Link: https://lore.kernel.org/20260730-fix-refcounted-huge-zero-v2-1-c5d8a41b317f@kernel.org Fixes: 3b77e8c8cde5 ("mm/thp: make is_huge_zero_pmd() safe and quicker") Signed-off-by: Lorenzo Stoakes (ARM) Reported-by: Hengbin Zhang Closes: https://lore.kernel.org/linux-mm/20260727154001.4102341-1-uqbarz@gmail.com/ Suggested-by: David Hildenbrand (Arm) Acked-by: David Hildenbrand (Arm) Cc: Baolin Wang Cc: Barry Song Cc: Dev Jain Cc: Hannes Reinecke Cc: Hugh Dickins Cc: Kiryl Shutsemau Cc: Lance Yang Cc: Liam R. Howlett Cc: Nico Pache Cc: Pankaj Raghav Cc: Ryan Roberts Cc: Yang Shi Cc: Zi Yan Cc: Signed-off-by: Andrew Morton [ adapted folio API (`huge_zero_folio`, `folio_pfn`, `folio_put`) to pre-6.11 page naming (`huge_zero_page`, `page_to_pfn`, `__free_pages`) ] Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

mm/ptdump: always stabilise against page table freeing using init_mm

2026-08-23T12:20:05+00:00

[ Upstream commit 27c32e5538344b13c1505a08861e04620c125d47 ] Previous commits have established the invariant that kernel page table freeing is performed while an mmap read lock on init_mm is held, which fixes races between ptdump and kernel page table freeing over init_mm. However, x86 and arm64 can perform a ptdump over an mm other than init_mm via ptdump_walk_pgd() and since kernel memory ranges are shared across non-kernel mm's, this means that the race still exists for these cases. Fix this by acquiring a nested mmap write lock for init_mm in ptdump_walk_pgd(). This is safe as we take this after mmap write locking the mm, and nothing acquires the init_mm lock first before locking an arbitrary mm, so no deadlock is possible. Also update walk_page_range_debug() to assert that init_mm is write locked, add a comment explaining why and remove some redundant code, and eliminate the unnecessary and confusing invocation of walk_kernel_page_table_range(). We can safely remove the non-NULL check for walk.mm, as the mmap lock asserts would NULL pointer deref if it was (and of course no callers do this). The first point at which ptdump can race kernel page table freeing is commit b6bdb7517c3d ("mm/vmalloc: add interfaces to free unmapped page table"), so we target this in the Fixes tag. Link: https://lore.kernel.org/20260723-series-vmap-race-fix-v6-4-8cc77dcc0018@kernel.org Fixes: b6bdb7517c3d ("mm/vmalloc: add interfaces to free unmapped page table") Signed-off-by: Lorenzo Stoakes (ARM) Reviewed-by: Mike Rapoport (Microsoft) Acked-by: David Hildenbrand (Arm) Reviewed-by: Kiryl Shutsemau Cc: Andy Lutomirski Cc: "Borah, Chaitanya Kumar" Cc: "Borislav Petkov (AMD)" Cc: Catalin Marinas Cc: Dave Hansen Cc: David Carlier Cc: Dev Jain Cc: "H. Peter Anvin" Cc: Ingo Molnar Cc: Liam R. Howlett Cc: Michal Hocko Cc: Peter Zijlstra Cc: Ryan Roberts Cc: Shakeel Butt Cc: Suren Baghdasaryan Cc: Toshi Kani Cc: "Uladzislau Rezki (Sony)" Cc: Vlastimil Babka Cc: Will Deacon Cc: Signed-off-by: Andrew Morton Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

mm/vmalloc: acquire init_mm lock on huge vmap to avoid ptdump UAF

2026-08-23T12:20:04+00:00

[ Upstream commit 26444eb71465c9934d9d418ef69c43f61185329b ] Patch series "mm: fix UAF caused by race between ptdump and vmap pgtable freeing", v6. Kernel page table walkers fall into two broad categories - those ranges where no exclusion is required via walk_kernel_page_table_range_lockless() and those where exclusion is required via walk_kernel_page_table_range() or walk_page_range_debug(). The former category is used only by arm64 arch code operating on ranges it both wholly owns and does not concurrently write. The latter category consists of kernel page table walkers operating on ranges that are wholly owned (but which need exclusion against concurrent writers). The lock used for exclusion is the mmap lock, and for kernel ranges this is the mmap lock on init_mm. ptdump is a special case being both the only user of walk_page_range_debug(), and the only case in which it walks ranges it does not own. This presents a problem, as page tables may be freed under ptdump. And indeed there is a use-after-free bug in the kernel as a result, which this series addresses. vmap promotes page tables to huge leaf entries where possible, freeing the lower page table when it does. It does this with no meaningful locks held against concurrent ptdump walks. As a result, use-after-free can currently occur. This series addresses the issue by having the vmap huge promotion logic acquire the mmap read lock while both setting the huge page table entry and freeing the prior leaf page table. The ptdump code already acquires the mmap write lock, so by doing so we ensure that the ptdump walker only ever observes either the huge page table entry or the existing page table entry, and nothing is freed underneath it. A mitigation for this issue was already applied for arm64 in commit fa93b45fd397 ("arm64: Enable vmalloc-huge with ptdump"), which this series has to deal with carefully. This mitigation resolves the issue by acquiring the mmap read lock on init_mm on vmap page table free if a ptdump is in progress. However the fix in this series would cause a deadlock if we were to simply apply it for arm64 without also reverting the change. This is because vmap may acquire the read lock before ptdump attempts to acquire the write lock, which then gets queued, and rwsem starvation rules mean that the (unacknowledged) nested mmap read lock in the arm64 code would also block, meaning the original read lock is never released and thus deadlock. This series works around this by #ifndef CONFIG_ARM64'ing the mmap read lock in vmap logic, then partially reverting commit fa93b45fd397 ("arm64: Enable vmalloc-huge with ptdump"), keeping the enablement of huge vmap support, and removing the ifdeffery with the partial revert patch. There are related issues that are also addressed in this series: * x86 page attribute logic, specifically Change Page Attributes (CPA), implements a feature whereby huge ranges can be collapsed into huge leaf entries. This can similarly cause a UAF when done in parallel with a ptdump walk, so similarly acquire the init_mm mmap lock to avoid this. * The CPA logic allows concurrent page table manipulation and CPA collapse, meaning the former risks accessing a page table the latter frees. Fix this by acquiring mmap write lock on init_mm across the whole CPA collapse operation and read lock on the page table manipulation. * x86 and arm64 permit walks of non-kernel mm's (both allowing efi mm walks, and in x86's case arbitrary mm's), so we ensure kernel mappings remain stable by locking the init_mm as well as the mm being walked. The ordering of patches is established for both strict dependencies (the arm64 partial revert in particular has to be done after the vmap changes) and logical ones (the non-kernel mm fix only makes sense once the vmap/CPA fixes are in place). This patch (of 3): Currently there is a nasty race between ptdump and vmap when attempting to map a huge P4D, PUD or PMD entry: * ptdump walks kernel page table ranges it doesn't own. * When vmap maps ranges it tries to promotes existing ones to huge page tables in vmap_try_huge_[p4d,pud,pmd]() at P4D, PUD and PMD level, freeing the lower page table in [p4d,pud,pmd]_free_[pud,pmd,pte]_page() when it succeeds. Both of these things can happen at the same time and as a result ptdump can access a freed page table, resulting in a use-after-free and memory corruption. This is possible because while ptdump_walk_pgd() holds both the mem hotplug lock and the mmap write lock before invoking walk_page_range_debug(), vmap takes no relevant locks at all. Fix this by holding the mmap read lock in vmap_try_huge_*() when freeing page tables. The read lock is sufficient: ptdump is the only walker that must be excluded and it holds the mmap write lock. Other holders of the read lock may run concurrently, but each exclusively owns the range it operates on and cannot reach the page tables freed here. We also hold the lock while assigning the huge page table entry, which means page table walkers observe only the huge or non-huge page table entry. We use a trylock to prevent ptdump from blocking vmap making forward progress. This is fine because it's an optimisation in any case, and thus the vmap can safely proceed regardless. All other kernel page table walkers that touch vmalloc ranges either exclusively own the memory walked or acquire the mmap lock, so this correctly excludes those walkers. One wrinkle here is commit fa93b45fd397 ("arm64: Enable vmalloc-huge with ptdump"), which addresses the issue for arm64 only by explicitly acquiring the mmap read lock on kernel page table freeing should a concurrent ptdump be in progress. This is problematic as vmap may acquire the mmap read lock prior to ptdump attempting to acquire an mmap write lock, leading to a deadlock when the mmap read lock is slept upon on page table freeing due to rwsem anti-starvation. We work around this by predicating the mmap lock being taken on !CONFIG_ARM64 for the time being. With this patch applied, a follow up will partially revert commit fa93b45fd397 ("arm64: Enable vmalloc-huge with ptdump") and at that stage remove the arm64 ifdeffery. We also update walk_page_range_debug() to assert the mmap write lock unconditionally and update the comment here to reflect this change. The issue has existed as long as ptdump was available and vmap freed page tables when promoting to a huge leaf entry, that is, since commit b6bdb7517c3d ("mm/vmalloc: add interfaces to free unmapped page table") for huge ioremap, and commit 121e6f3258fe ("mm/vmalloc: hugepage vmalloc mappings") for huge vmalloc. Since the former is the earlier of the two we choose that for our Fixes tag. We also define a guard class for mmap_read_trylock() so we can use cleanup.h to make the scope handling cleaner in the implementation. This patch is based on work by David Carlier (linked), with gratitude! Link: https://lore.kernel.org/20260723-series-vmap-race-fix-v6-0-8cc77dcc0018@kernel.org Link: https://lore.kernel.org/20260723-series-vmap-race-fix-v6-1-8cc77dcc0018@kernel.org Fixes: b6bdb7517c3d ("mm/vmalloc: add interfaces to free unmapped page table") Signed-off-by: Lorenzo Stoakes (ARM) Reported-by: syzbot+fd95a72470f5a44e464c@syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/6a287988.39669fcc.33b062.00a0.GAE@google.com/T/ Link: https://lore.kernel.org/linux-mm/20260706203128.162335-1-devnexen@gmail.com/ Reviewed-by: Mike Rapoport (Microsoft) Reviewed-by: Dev Jain Acked-by: David Hildenbrand (Arm) Reviewed-by: Kiryl Shutsemau Cc: Cc: Andy Lutomirski Cc: "Borah, Chaitanya Kumar" Cc: "Borislav Petkov (AMD)" Cc: Catalin Marinas Cc: Dave Hansen Cc: "H. Peter Anvin" Cc: Ingo Molnar Cc: Liam R. Howlett Cc: Michal Hocko Cc: Peter Zijlstra Cc: Ryan Roberts Cc: Shakeel Butt Cc: Suren Baghdasaryan Cc: Toshi Kani Cc: "Uladzislau Rezki (Sony)" Cc: Vlastimil Babka Cc: Will Deacon Signed-off-by: Andrew Morton [ added the missing `DEFINE_GUARD(mmap_read_lock, ...)` base class and dropped the `page_owner.h` include context absent from this tree ] Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

mm/pagewalk: split walk_page_range_novma() into kernel/user parts

2026-08-23T12:20:04+00:00

[ Upstream commit 96d81e4766f9e88b66a0502b5a7f34a4c20ac754 ] walk_page_range_novma() is rather confusing - it supports two modes, one used often, the other used only for debugging. The first mode is the common case of traversal of kernel page tables, which is what nearly all callers use this for. Secondly it provides an unusual debugging interface that allows for the traversal of page tables in a userland range of memory even for that memory which is not described by a VMA. It is far from certain that such page tables should even exist, but perhaps this is precisely why it is useful as a debugging mechanism. As a result, this is utilised by ptdump only. Historically, things were reversed - ptdump was the only user, and other parts of the kernel evolved to use the kernel page table walking here. Since we have some complicated and confusing locking rules for the novma case, it makes sense to separate the two usages into their own functions. Doing this also provide self-documentation as to the intent of the caller - are they doing something rather unusual or are they simply doing a standard kernel page table walk? We therefore establish two separate functions - walk_page_range_debug() for this single usage, and walk_kernel_page_table_range() for general kernel page table walking. The walk_page_range_debug() function is currently used to traverse both userland and kernel mappings, so we maintain this and in the case of kernel mappings being traversed, we have walk_page_range_debug() invoke walk_kernel_page_table_range() internally. We additionally make walk_page_range_debug() internal to mm. Link: https://lkml.kernel.org/r/20250605135104.90720-1-lorenzo.stoakes@oracle.com Signed-off-by: Lorenzo Stoakes Acked-by: Mike Rapoport (Microsoft) Acked-by: Qi Zheng Reviewed-by: Oscar Salvador Reviewed-by: Suren Baghdasaryan Reviewed-by: Vlastimil Babka Acked-by: David Hildenbrand Cc: Albert Ou Cc: Alexandre Ghiti Cc: Barry Song Cc: Huacai Chen Cc: Jann Horn Cc: Jonas Bonn Cc: Liam Howlett Cc: Michal Hocko Cc: Muchun Song Cc: Palmer Dabbelt Cc: Paul Walmsley Cc: Stafford Horne Cc: Stefan Kristiansson Cc: WANG Xuerui Signed-off-by: Andrew Morton Stable-dep-of: 26444eb71465 ("mm/vmalloc: acquire init_mm lock on huge vmap to avoid ptdump UAF") Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

mm: migrate_device: fix pte_pfn/pte_dirty called on non-present PTE

2026-08-23T12:20:03+00:00

[ Upstream commit 63867c82d0c0c2d182016a32b1cc0103116b0ea5 ] pte_pfn() and pte_dirty() have undefined behaviour when called on a non-present PTE. In migrate_vma_collect_pmd(), these functions may be invoked on non-present entries (e.g., device-private entries), leading to potential crashes from pte_pfn() or incorrect dirty folio accounting from pte_dirty(). Fix both by guarding with pte_present() checks. Link: https://lore.kernel.org/20260708003955.4024340-1-wangkefeng.wang@huawei.com Link: https://lore.kernel.org/20260706111958.3649651-1-wangkefeng.wang@huawei.com Fixes: fd35ca3d12cc ("mm/migrate_device.c: copy pte dirty bit to page") Fixes: 6c287605fd56 ("mm: remember exclusively mapped anonymous pages with PG_anon_exclusive") Signed-off-by: Kefeng Wang Reviewed-by: Balbir Singh Acked-by: Zi Yan Cc: Alistair Popple Cc: Byungchul Park Cc: David Hildenbrand Cc: Gregory Price Cc: "Huang, Ying" Cc: Joshua Hahn Cc: Matthew Brost Cc: Rakie Kim Cc: Ying Huang Cc: Signed-off-by: Andrew Morton [ adapted `folio_test_anon(folio)`/`folio_mark_dirty(folio)` context lines to 6.6's `PageAnon(page)`/`folio_mark_dirty(page_folio(page))` spelling since the `folio` local doesn't exist ] Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

userfaultfd: prevent registration of special VMAs

2026-08-23T12:20:01+00:00

[ Upstream commit 3c58f641e813c3c71039f8fd4d4e2a3aab713288 ] Vova Tokarev says: userfaultfd allows registration on shadow stack VMAs. With userfaultfd access, you can register on the shadow stack, discard a page ... and inject a page with chosen return addresses via UFFDIO_COPY. Update vma_can_userfault() to reject VM_SHADOW_STACK. While on it, also reject VM_SPECIAL so that if a driver would implement vm_uffd_ops, it wouldn't be possible to register special VMAs with userfaultfd. Since VM_SPECIAL includes VM_DONTEXPAND which is set but hugetlb, exclude hugetlb VMAs from the check for VM_SPECIAL. Link: https://lore.kernel.org/20260618095017.2553004-1-rppt@kernel.org Fixes: 54007f818206 ("mm: Introduce VM_SHADOW_STACK for shadow stack memory") Signed-off-by: Mike Rapoport (Microsoft) Reported-by: vova tokarev Acked-by: David Hildenbrand (Arm) Reviewed-by: Lorenzo Stoakes Cc: Al Viro Cc: Christian Brauner Cc: Jan Kara Cc: Linus Torvalds Cc: Mike Rapoport Cc: Oleg Nesterov Cc: Peter Xu Cc: Signed-off-by: Andrew Morton Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

userfaultfd: move vma_can_userfault out of line

2026-08-23T12:20:01+00:00

[ Upstream commit a5bb8669872b6b8463b8777a7a259a8305060016 ] vma_can_userfault() has grown pretty big and it's not called on performance critical path. Move it out of line. No functional changes. Link: https://lore.kernel.org/20260402041156.1377214-7-rppt@kernel.org Signed-off-by: Mike Rapoport (Microsoft) Reviewed-by: David Hildenbrand (Red Hat) Reviewed-by: Liam R. Howlett Cc: Andrea Arcangeli Cc: Andrei Vagin Cc: Axel Rasmussen Cc: Baolin Wang Cc: Harry Yoo Cc: Harry Yoo (Oracle) Cc: Hugh Dickins Cc: James Houghton Cc: Lorenzo Stoakes (Oracle) Cc: Matthew Wilcox (Oracle) Cc: Michal Hocko Cc: Muchun Song Cc: Nikita Kalyazin Cc: Oscar Salvador Cc: Paolo Bonzini Cc: Peter Xu Cc: Sean Christopherson Cc: Shuah Khan Cc: Suren Baghdasaryan Cc: Vlastimil Babka Cc: David Carlier Signed-off-by: Andrew Morton Stable-dep-of: 3c58f641e813 ("userfaultfd: prevent registration of special VMAs") Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

mm/huge_memory: unlock i_mmap_rwsem before releasing after-split folios

2026-08-09T18:22:00+00:00

[ Upstream commit e923bd21058ea02fd0dcd3549d151d143fd036e5 ] __folio_split() keeps dereferencing the mapping after the split: shmem_uncharge(mapping->host) and remap_page() while the folios are still frozen/locked, and i_mmap_unlock_read(mapping) at the very end, after the after-split folios have been unlocked and freed. Nothing holds an inode reference across that. The split relies on @folio -- which the beyond-EOF drop loop never removes, as it starts at folio_next(folio) -- staying locked and in the page cache to hold off eviction. But the unlock loop unlocks @folio before i_mmap_unlock_read() runs. If the caller's @lock_at is a tail beyond EOF, as memory_failure() passes when splitting a poisoned tail of a shmem THP that reaches past i_size during truncation, it too is gone from the page cache; so once @folio is unlocked no locked, in-cache folio pins the inode, and a concurrent final iput() can evict and RCU-free it before i_mmap_unlock_read() touches i_mmap_rwsem: BUG: KASAN: slab-use-after-free in __up_read+0x634/0x790 i_mmap_unlock_read include/linux/fs.h:537 [inline] __folio_split+0x732/0x1640 mm/huge_memory.c:4100 try_to_split_thp_page+0xab/0x390 mm/memory-failure.c:1675 memory_failure+0x1394/0x26e0 mm/memory-failure.c:2470 Freed by task 4601: shmem_free_in_core_inode+0x54/0xb0 mm/shmem.c:5177 evict+0x57f/0xac0 fs/inode.c:870 Do every mapping dereference while @folio still pins the inode: drop i_mmap_rwsem right after remap_page(), before the loop that unlocks and frees the after-split folios, and clear @mapping so the exit path does not unlock it again. shmem_uncharge() and remap_page() already run before that point, so after this nothing past the unlock loop touches the inode or the mapping. This is now a rule the split depends on, alongside keeping @folio frozen until the page cache is updated: no inode or mapping dereference once the after-split folios start being unlocked. Link: https://lore.kernel.org/20260716095424.471052-1-kirill@shutemov.name Fixes: baa355fd3314 ("thp: file pages support for split_huge_page()") Signed-off-by: Kiryl Shutsemau (Meta) Reported-by: Hao Zhang Closes: https://lore.kernel.org/linux-mm/20260710071344.GA106129@zh-pc Co-developed-by: Hao Zhang Signed-off-by: Hao Zhang Acked-by: David Hildenbrand (Arm) Reviewed-by: Zi Yan Reviewed-by: Baolin Wang Reviewed-by: Miaohe Lin Cc: Baolin Wang Cc: Barry Song Cc: Dev Jain Cc: Lance Yang Cc: Liam R. Howlett Cc: Lorenzo Stoakes Cc: Naoya Horiguchi Cc: Nico Pache Cc: Ryan Roberts Cc: Signed-off-by: Andrew Morton (cherry picked from commit e923bd21058ea02fd0dcd3549d151d143fd036e5) [ kas: adapt to the __split_huge_page()/split_huge_page_to_list() two-function split: pass @mapping into __split_huge_page() and drop it there, before the loop that frees the after-split subpages while the head is still locked; the caller then skips its own i_mmap unlock ] Signed-off-by: Kiryl Shutsemau (Meta) Signed-off-by: Sasha Levin

mm/hugetlb: fix swap entry corruption when clearing uffd-wp at fork()

2026-08-09T18:22:00+00:00

[ Upstream commit 83abe2fd5b3aeb3123b5408a5a91709c5538fb23 ] copy_hugetlb_page_range() clears the uffd-wp bit of migration and hwpoison entries with huge_pte_clear_uffd_wp(), which operates on the present-PTE bit position. Swap entries keep the uffd-wp state elsewhere -- the migration branch reads and sets it with pte_swp_uffd_wp() and pte_swp_mkuffd_wp() -- and the present-PTE position falls into the swap payload. On x86-64 it lands in the inverted swap offset, where a naturally-aligned hugetlb PFN always has the affected bit set, so the clear advances the encoded PFN by two pages. No userfaultfd needs to be involved: the clear is guarded only by the child VMA not being uffd-wp registered, so a plain fork() with an in-flight hugetlb migration entry (or a poisoned hugetlb page) corrupts the entry copied into the child. Instrumenting the clear and forking after MADV_HWPOISON on a 2MB anon hugetlb page shows: offset before=120e00 offset after =120e02 The fallout is mostly latent: rmap walks match migration entries by folio range and remove_migration_pte() rebuilds the PTE from the folio, so a within-folio PFN skew heals once migration completes. But any path that re-encodes the corrupted offset -- e.g. hugetlb_change_protection() rewriting a writable migration entry via make_readable_migration_entry(swp_offset(entry)) -- propagates it. Migration entries legitimately carry uffd-wp, so clear it with pte_swp_clear_uffd_wp(), matching copy_nonpresent_pte() and move_huge_pte(). A hwpoison entry, on the other hand, never carries the uffd-wp bit: it is installed fresh by make_hwpoison_entry() (try_to_unmap_one() does not preserve uffd-wp on the hwpoison path) and hugetlb_change_protection() leaves hwpoison entries untouched. There was nothing to clear there, only the corruption, so drop the clear entirely. Link: https://lore.kernel.org/20260708090110.136162-1-kirill@shutemov.name Fixes: bc70fbf269fd ("mm/hugetlb: handle uffd-wp during fork()") Signed-off-by: Kiryl Shutsemau Reported-by: Sashiko AI review Closes: https://lore.kernel.org/all/20260703140011.99E601F000E9@smtp.kernel.org/ Suggested-by: David Hildenbrand Acked-by: David Hildenbrand (Arm) Assisted-by: Claude:claude-fable-5 Cc: Muchun Song Cc: Oscar Salvador Cc: Peter Xu Cc: Signed-off-by: Andrew Morton (cherry picked from commit 83abe2fd5b3aeb3123b5408a5a91709c5538fb23) [ kas: adapt to the pre-softleaf idiom ] Signed-off-by: Kiryl Shutsemau (Meta) Signed-off-by: Sasha Levin

mm/page_reporting: use system_freezable_wq to fix UAF during suspend

2026-08-09T18:21:54+00:00

commit 0b45f6927a14914ff685fe0e6f9d11232a1e03df upstream. During PM freeze (e.g. S3 suspend or S4 hibernation), device drivers like virtio_balloon reset their underlying virtio devices and delete their virtqueues via vdev->config->del_vqs(). However, page reporting work (page_reporting_process) was scheduled on the global system_wq. Because system_wq lacks the WQ_FREEZABLE flag, the PM freezer skips it, leaving page_reporting_process active during suspend. If pages are freed into the buddy allocator while suspending (for example, when core MM invokes the balloon shrinker during S4 hibernation image saving), page reporting triggers virtballoon_free_page_report() on deleted virtqueues, resulting in a Use-After-Free / General Protection Fault: [ 196.795226] general protection fault, probably for non-canonical address 0xaa1436fe70dae6df: 0000 [#1] SMP NOPTI [ 196.825967] Workqueue: events page_reporting_process [ 196.831038] RIP: 0010:virtqueue_add_split+0x233/0x4c0 [virtio_ring] [ 196.927073] virtballoon_free_page_report+0x3a/0xe0 [virtio_balloon] [ 196.946943] page_reporting_process+0x370/0x4f0 Fix this by switching page reporting work to system_freezable_wq. This ensures that the PM freezer pauses page_reporting_process before device drivers destroy their reporting virtqueues. Because the reporting worker is frozen, memory reclamation/freeing (e.g. via shrinker execution) can safely return pages to MM during freeze without triggering unfrozen reporting work on deleted virtqueues. This aligns with the driver's existing design. The comment in virtballoon_freeze() states: /* * The workqueue is already frozen by the PM core before this * function is called. */ Testing: I have verified these fixes using Google’s virtualization infrastructure by running continuous suspend/resume iterations (40+ cycles) while churning memory using stress-ng (`stress-ng --vm 4 --vm-bytes 60% --timeout 1`) to constantly create free pages for the buddy allocator. We also set the `page_reporting_order` parameter to 0 to make the page reporting worker highly sensitive, forcing it to pick up any 4K free pages. This confirmed that the UAF crashes are no longer reproducible. Link: https://lore.kernel.org/20260721005603.1710551-1-linkl@google.com Fixes: 36e66c554b5c ("mm: introduce Reported pages") Signed-off-by: Link Lin Suggested-by: David Hildenbrand (Arm) Suggested-by: Michael S. Tsirkin Acked-by: David Rientjes Acked-by: David Hildenbrand (Arm) Acked-by: Michael S. Tsirkin Cc: Alexander Duyck Cc: Greg Thelen Cc: James Houghton Cc: Jason Wang Cc: Jiaqi Yan Cc: Vlastimil Babka Cc: Xuan Zhuo Cc: Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman