kernel/linux.git/include/asm-generic/tlb.h, branch v6.1.168

mm/hugetlb: fix excessive IPI broadcasts when unsharing PMD tables using mmu_gather

2026-03-25T10:03:14+00:00

commit 8ce720d5bd91e9dc16db3604aa4b1bf76770a9a1 upstream. As reported, ever since commit 1013af4f585f ("mm/hugetlb: fix huge_pmd_unshare() vs GUP-fast race") we can end up in some situations where we perform so many IPI broadcasts when unsharing hugetlb PMD page tables that it severely regresses some workloads. In particular, when we fork()+exit(), or when we munmap() a large area backed by many shared PMD tables, we perform one IPI broadcast per unshared PMD table. There are two optimizations to be had: (1) When we process (unshare) multiple such PMD tables, such as during exit(), it is sufficient to send a single IPI broadcast (as long as we respect locking rules) instead of one per PMD table. Locking prevents that any of these PMD tables could get reused before we drop the lock. (2) When we are not the last sharer (> 2 users including us), there is no need to send the IPI broadcast. The shared PMD tables cannot become exclusive (fully unshared) before an IPI will be broadcasted by the last sharer. Concurrent GUP-fast could walk into a PMD table just before we unshared it. It could then succeed in grabbing a page from the shared page table even after munmap() etc succeeded (and supressed an IPI). But there is not difference compared to GUP-fast just sleeping for a while after grabbing the page and re-enabling IRQs. Most importantly, GUP-fast will never walk into page tables that are no-longer shared, because the last sharer will issue an IPI broadcast. (if ever required, checking whether the PUD changed in GUP-fast after grabbing the page like we do in the PTE case could handle this) So let's rework PMD sharing TLB flushing + IPI sync to use the mmu_gather infrastructure so we can implement these optimizations and demystify the code at least a bit. Extend the mmu_gather infrastructure to be able to deal with our special hugetlb PMD table sharing implementation. To make initialization of the mmu_gather easier when working on a single VMA (in particular, when dealing with hugetlb), provide tlb_gather_mmu_vma(). We'll consolidate the handling for (full) unsharing of PMD tables in tlb_unshare_pmd_ptdesc() and tlb_flush_unshared_tables(), and track in "struct mmu_gather" whether we had (full) unsharing of PMD tables. Because locking is very special (concurrent unsharing+reuse must be prevented), we disallow deferring flushing to tlb_finish_mmu() and instead require an explicit earlier call to tlb_flush_unshared_tables(). From hugetlb code, we call huge_pmd_unshare_flush() where we make sure that the expected lock protecting us from concurrent unsharing+reuse is still held. Check with a VM_WARN_ON_ONCE() in tlb_finish_mmu() that tlb_flush_unshared_tables() was properly called earlier. Document it all properly. Notes about tlb_remove_table_sync_one() interaction with unsharing: There are two fairly tricky things: (1) tlb_remove_table_sync_one() is a NOP on architectures without CONFIG_MMU_GATHER_RCU_TABLE_FREE. Here, the assumption is that the previous TLB flush would send an IPI to all relevant CPUs. Careful: some architectures like x86 only send IPIs to all relevant CPUs when tlb->freed_tables is set. The relevant architectures should be selecting MMU_GATHER_RCU_TABLE_FREE, but x86 might not do that in stable kernels and it might have been problematic before this patch. Also, the arch flushing behavior (independent of IPIs) is different when tlb->freed_tables is set. Do we have to enlighten them to also take care of tlb->unshared_tables? So far we didn't care, so hopefully we are fine. Of course, we could be setting tlb->freed_tables as well, but that might then unnecessarily flush too much, because the semantics of tlb->freed_tables are a bit fuzzy. This patch changes nothing in this regard. (2) tlb_remove_table_sync_one() is not a NOP on architectures with CONFIG_MMU_GATHER_RCU_TABLE_FREE that actually don't need a sync. Take x86 as an example: in the common case (!pv, !X86_FEATURE_INVLPGB) we still issue IPIs during TLB flushes and don't actually need the second tlb_remove_table_sync_one(). This optimized can be implemented on top of this, by checking e.g., in tlb_remove_table_sync_one() whether we really need IPIs. But as described in (1), it really must honor tlb->freed_tables then to send IPIs to all relevant CPUs. Notes on TLB flushing changes: (1) Flushing for non-shared PMD tables We're converting from flush_hugetlb_tlb_range() to tlb_remove_huge_tlb_entry(). Given that we properly initialize the MMU gather in tlb_gather_mmu_vma() to be hugetlb aware, similar to __unmap_hugepage_range(), that should be fine. (2) Flushing for shared PMD tables We're converting from various things (flush_hugetlb_tlb_range(), tlb_flush_pmd_range(), flush_tlb_range()) to tlb_flush_pmd_range(). tlb_flush_pmd_range() achieves the same that tlb_remove_huge_tlb_entry() would achieve in these scenarios. Note that tlb_remove_huge_tlb_entry() also calls __tlb_remove_tlb_entry(), however that is only implemented on powerpc, which does not support PMD table sharing. Similar to (1), tlb_gather_mmu_vma() should make sure that TLB flushing keeps on working as expected. Further, note that the ptdesc_pmd_pts_dec() in huge_pmd_share() is not a concern, as we are holding the i_mmap_lock the whole time, preventing concurrent unsharing. That ptdesc_pmd_pts_dec() usage will be removed separately as a cleanup later. There are plenty more cleanups to be had, but they have to wait until this is fixed. [david@kernel.org: fix kerneldoc] Link: https://lkml.kernel.org/r/f223dd74-331c-412d-93fc-69e360a5006c@kernel.org Link: https://lkml.kernel.org/r/20251223214037.580860-5-david@kernel.org Fixes: 1013af4f585f ("mm/hugetlb: fix huge_pmd_unshare() vs GUP-fast race") Signed-off-by: David Hildenbrand (Red Hat) Reported-by: "Uschakow, Stanislav" Closes: https://lore.kernel.org/all/4d3878531c76479d9f8ca9789dc6485d@amazon.de/ Tested-by: Laurence Oberman Acked-by: Harry Yoo Reviewed-by: Lorenzo Stoakes Cc: Lance Yang Cc: Liu Shixin Cc: Oscar Salvador Cc: Rik van Riel Cc: Signed-off-by: Andrew Morton [ David: We don't have ptdesc and the wrappers, so work directly on page->pt_share_count and pass "struct page" instead of "struct ptdesc". CONFIG_HUGETLB_PMD_PAGE_TABLE_SHARING is still called CONFIG_ARCH_WANT_HUGE_PMD_SHARE and is set even without CONFIG_HUGETLB_PAGE. move_hugetlb_page_tables() still uses flush_tlb_range() instead of flush_hugetlb_tlb_range(). Some smaller contextual stuff. ] Signed-off-by: David Hildenbrand (Arm) Signed-off-by: Greg Kroah-Hartman

mm/khugepaged: fix GUP-fast interaction by sending IPI

2022-11-30T22:49:42+00:00

Since commit 70cbc3cc78a99 ("mm: gup: fix the fast GUP race against THP collapse"), the lockless_pages_from_mm() fastpath rechecks the pmd_t to ensure that the page table was not removed by khugepaged in between. However, lockless_pages_from_mm() still requires that the page table is not concurrently freed. Fix it by sending IPIs (if the architecture uses semi-RCU-style page table freeing) before freeing/reusing page tables. Link: https://lkml.kernel.org/r/20221129154730.2274278-2-jannh@google.com Link: https://lkml.kernel.org/r/20221128180252.1684965-2-jannh@google.com Link: https://lkml.kernel.org/r/20221125213714.4115729-2-jannh@google.com Fixes: ba76149f47d8 ("thp: khugepaged") Signed-off-by: Jann Horn Reviewed-by: Yang Shi Acked-by: David Hildenbrand Cc: John Hubbard Cc: Peter Xu Cc: Signed-off-by: Andrew Morton

mmu_gather: fix the CONFIG_MMU_GATHER_NO_RANGE case

2022-07-22T16:28:34+00:00

Sudip reports that alpha doesn't build properly, with errors like include/asm-generic/tlb.h:401:1: error: redefinition of 'tlb_update_vma_flags' 401 | tlb_update_vma_flags(struct mmu_gather *tlb, struct vm_area_struct *vma) | ^~~~~~~~~~~~~~~~~~~~ include/asm-generic/tlb.h:372:1: note: previous definition of 'tlb_update_vma_flags' with type 'void(struct mmu_gather *, struct vm_area_struct *)' 372 | tlb_update_vma_flags(struct mmu_gather *tlb, struct vm_area_struct *vma) { } the cause being that We have this odd situation where some architectures were never converted to the newer TLB flushing interfaces that have a range for the flush. Instead people left them alone, and we have them select the MMU_GATHER_NO_RANGE config option to make the tlb header files account for this. Peter Zijlstra cleaned some of these nasty header file games up in commits 1e9fdf21a433 ("mmu_gather: Remove per arch tlb_{start,end}_vma()") 18ba064e42df ("mmu_gather: Let there be one tlb_{start,end}_vma() implementation") but tlb_update_vma_flags() was left alone, and then commit b67fbebd4cf9 ("mmu_gather: Force tlb-flush VM_PFNMAP vmas") ended up removing only _one_ of the two stale duplicate dummy inline functions. This removes the other stale one. Somebody braver than me should try to remove MMU_GATHER_NO_RANGE entirely, but it requires fixing up the oddball architectures that use it: alpha, m68k, microblaze, nios2 and openrisc. The fixups should be fairly straightforward ("fix the build errors it exposes by adding the appropriate range arguments"), but the reason this wasn't done in the first place is that so few people end up working on those architectures. But it could be done one architecture at a time, hint, hint. Reported-by: Sudip Mukherjee (Codethink) Fixes: b67fbebd4cf9 ("mmu_gather: Force tlb-flush VM_PFNMAP vmas") Link: https://lore.kernel.org/all/YtpXh0QHWwaEWVAY@debian/ Cc: Peter Zijlstra Cc: Will Deacon Cc: "Aneesh Kumar K.V" Cc: Andrew Morton Cc: Nick Piggin Cc: Arnd Bergmann Signed-off-by: Linus Torvalds

mmu_gather: Force tlb-flush VM_PFNMAP vmas

2022-07-21T17:50:13+00:00

Jann reported a race between munmap() and unmap_mapping_range(), where unmap_mapping_range() will no-op once unmap_vmas() has unlinked the VMA; however munmap() will not yet have invalidated the TLBs. Therefore unmap_mapping_range() will complete while there are still (stale) TLB entries for the specified range. Mitigate this by force flushing TLBs for VM_PFNMAP ranges. Signed-off-by: Peter Zijlstra (Intel) Acked-by: Will Deacon Signed-off-by: Linus Torvalds

mmu_gather: Let there be one tlb_{start,end}_vma() implementation

2022-07-21T17:50:13+00:00

Now that architectures are no longer allowed to override tlb_{start,end}_vma() re-arrange code so that there is only one implementation for each of these functions. This much simplifies trying to figure out what they actually do. Signed-off-by: Peter Zijlstra (Intel) Acked-by: Will Deacon Signed-off-by: Linus Torvalds

mmu_gather: Remove per arch tlb_{start,end}_vma()

2022-07-21T17:50:13+00:00

Scattered across the archs are 3 basic forms of tlb_{start,end}_vma(). Provide two new MMU_GATHER_knobs to enumerate them and remove the per arch tlb_{start,end}_vma() implementations. - MMU_GATHER_NO_FLUSH_CACHE indicates the arch has flush_cache_range() but does *NOT* want to call it for each VMA. - MMU_GATHER_MERGE_VMAS indicates the arch wants to merge the invalidate across multiple VMAs if possible. With these it is possible to capture the three forms: 1) empty stubs; select MMU_GATHER_NO_FLUSH_CACHE and MMU_GATHER_MERGE_VMAS 2) start: flush_cache_range(), end: empty; select MMU_GATHER_MERGE_VMAS 3) start: flush_cache_range(), end: flush_tlb_range(); default Obviously, if the architecture does not have flush_cache_range() then it also doesn't need to select MMU_GATHER_NO_FLUSH_CACHE. Signed-off-by: Peter Zijlstra (Intel) Acked-by: Will Deacon Cc: David Miller Signed-off-by: Linus Torvalds

mm/mprotect: do not flush when not required architecturally

2022-05-13T14:20:05+00:00

Currently, using mprotect() to unprotect a memory region or uffd to unprotect a memory region causes a TLB flush. However, in such cases the PTE is often not modified (i.e., remain RO) and therefore not TLB flush is needed. Add an arch-specific pte_needs_flush() which tells whether a TLB flush is needed based on the old PTE and the new one. Implement an x86 pte_needs_flush(). Always flush the TLB when it is architecturally needed even when skipping a TLB flush might only result in a spurious page-faults by skipping the flush. Even with such conservative manner, we can in the future further refine the checks to test whether a PTE is present by only considering the architectural _PAGE_PRESENT flag instead of {pte|pmd}_preesnt(). For not be careful and use the latter. Link: https://lkml.kernel.org/r/20220401180821.1986781-3-namit@vmware.com Signed-off-by: Nadav Amit Cc: Andrea Arcangeli Cc: Andy Lutomirski Cc: Dave Hansen Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Will Deacon Cc: Yu Zhao Cc: Nick Piggin Cc: Andrew Cooper Cc: Peter Xu Signed-off-by: Andrew Morton

tlb: hugetlb: Add more sizes to tlb_remove_huge_tlb_entry

2022-04-06T11:41:48+00:00

tlb_remove_huge_tlb_entry only considers PMD_SIZE and PUD_SIZE when updating the mmu_gather structure. Unfortunately on arm64 there are two additional huge page sizes that need to be covered: CONT_PTE_SIZE and CONT_PMD_SIZE. Where an end-user attempts to employ contiguous huge pages, a VM_BUG_ON can be experienced due to the fact that the tlb structure hasn't been correctly updated by the relevant tlb_flush_p.._range() call from tlb_remove_huge_tlb_entry. This patch adds inequality logic to the generic implementation of tlb_remove_huge_tlb_entry s.t. CONT_PTE_SIZE and CONT_PMD_SIZE are effectively covered on arm64. Also, as well as ptes, pmds and puds; p4ds are now considered too. Reported-by: David Hildenbrand Suggested-by: Peter Zijlstra (Intel) Cc: Anshuman Khandual Cc: Catalin Marinas Cc: Will Deacon Link: https://lore.kernel.org/linux-mm/811c5c8e-b3a2-85d2-049c-717f17c3a03a@redhat.com/ Signed-off-by: Steve Capper Acked-by: David Hildenbrand Reviewed-by: Anshuman Khandual Reviewed-by: Catalin Marinas Acked-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20220330112543.863-1-steve.capper@arm.com Signed-off-by: Will Deacon

treewide: Replace zero-length arrays with flexible-array members

2022-02-17T13:00:39+00:00

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2]. This code was transformed with the help of Coccinelle: (next-20220214$ spatch --jobs $(getconf _NPROCESSORS_ONLN) --sp-file script.cocci --include-headers --dir . > output.patch) @@ identifier S, member, array; type T1, T2; @@ struct S { ... T1 member; T2 array[ - 0 ]; }; UAPI and wireless changes were intentionally excluded from this patch and will be sent out separately. [1] https://en.wikipedia.org/wiki/Flexible_array_member [2] https://www.kernel.org/doc/html/v5.16/process/deprecated.html#zero-length-and-one-element-arrays Link: https://github.com/KSPP/linux/issues/78 Reviewed-by: Kees Cook Signed-off-by: Gustavo A. R. Silva

tlb: mmu_gather: Introduce tlb_gather_mmu_fullmm()

2021-01-29T19:02:29+00:00

Passing the range '0, -1' to tlb_gather_mmu() sets the 'fullmm' flag, which indicates that the mm_struct being operated on is going away. In this case, some architectures (such as arm64) can elide TLB invalidation by ensuring that the TLB tag (ASID) associated with this mm is not immediately reclaimed. Although this behaviour is documented in asm-generic/tlb.h, it's subtle and easily missed. Introduce tlb_gather_mmu_fullmm() to make it clearer that this is for the entire mm and WARN() if tlb_gather_mmu() is called with the 'fullmm' address range. Signed-off-by: Will Deacon Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Yu Zhao Acked-by: Peter Zijlstra (Intel) Acked-by: Linus Torvalds Link: https://lkml.kernel.org/r/20210127235347.1402-4-will@kernel.org