Re: [PATCH v5 2/3] mm/khugepaged: Fix GUP-fast interaction by sending IPI

[Yang Shi]

On Tue, Nov 29, 2022 at 7:47 AM Jann Horn <jannh@xxxxxxxxxx> wrote:
>
> The khugepaged paths that remove page tables have to be careful to
> synchronize against the lockless_pages_from_mm() path, which traverses
> page tables while only being protected by disabled IRQs.
> lockless_pages_from_mm() must not:
>
>  1. interpret the contents of freed memory as page tables (and once a
>     page table has been deposited, it can be freed)
>  2. interpret the contents of deposited page tables as PTEs, since some
>     architectures will store non-PTE data inside deposited page tables
>     (see radix__pgtable_trans_huge_deposit())
>  3. create new page references from PTEs after the containing page
>     table has been detached and:
>     3a. __collapse_huge_page_isolate() has checked the page refcount
>     3b. the page table has been reused at another virtual address and
>         populated with new PTEs
>
> ("new page references" here refer to stable references returned to the
> caller; speculative references that are dropped on an error path are
> fine)
>
> commit 70cbc3cc78a99 ("mm: gup: fix the fast GUP race against THP
> collapse") addressed issue 3 by making the lockless_pages_from_mm()
> fastpath recheck the pmd_t to ensure that the page table was not
> removed by khugepaged in between (under the assumption that the page
> table is not repeatedly moving back and forth between two addresses,
> with one PTE repeatedly being populated with the same value).
>
> But to address issues 1 and 2, we need to send IPIs before
> freeing/reusing page tables. By doing that, issue 3 is also
> automatically addressed, so the fix from commit 70cbc3cc78a99 ("mm: gup:
> fix the fast GUP race against THP collapse") becomes redundant.
>
> We can ensure that the necessary IPI is sent by calling
> tlb_remove_table_sync_one() because, as noted in mm/gup.c, under
> configurations that define CONFIG_HAVE_FAST_GUP, there are two possible
> cases:
>
>  1. CONFIG_MMU_GATHER_RCU_TABLE_FREE is set, causing
>     tlb_remove_table_sync_one() to send an IPI to synchronize with
>     lockless_pages_from_mm().
>  2. CONFIG_MMU_GATHER_RCU_TABLE_FREE is unset, indicating that all
>     TLB flushes are already guaranteed to send IPIs.
>     tlb_remove_table_sync_one() will do nothing, but we've already
>     run pmdp_collapse_flush(), which did a TLB flush, which must have
>     involved IPIs.
>
> Cc: stable@xxxxxxxxxx
> Fixes: ba76149f47d8 ("thp: khugepaged")
> Reviewed-by: Yang Shi <shy828301@xxxxxxxxx>
> Acked-by: David Hildenbrand <david@xxxxxxxxxx>
> Signed-off-by: Jann Horn <jannh@xxxxxxxxxx>
> ---
>
> Notes:
>     v4:
>      - added ack from David Hildenbrand
>      - made commit message more verbose
>     v5:
>      - added reviewed-by from Yang Shi
>      - rewrote commit message based on feedback from Yang Shi

Thanks, Jann. Looks good to me.

>
>  include/asm-generic/tlb.h | 4 ++++
>  mm/khugepaged.c           | 2 ++
>  mm/mmu_gather.c           | 4 +---
>  3 files changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/include/asm-generic/tlb.h b/include/asm-generic/tlb.h
> index 492dce43236ea..cab7cfebf40bd 100644
> --- a/include/asm-generic/tlb.h
> +++ b/include/asm-generic/tlb.h
> @@ -222,12 +222,16 @@ extern void tlb_remove_table(struct mmu_gather *tlb, void *table);
>  #define tlb_needs_table_invalidate() (true)
>  #endif
>
> +void tlb_remove_table_sync_one(void);
> +
>  #else
>
>  #ifdef tlb_needs_table_invalidate
>  #error tlb_needs_table_invalidate() requires MMU_GATHER_RCU_TABLE_FREE
>  #endif
>
> +static inline void tlb_remove_table_sync_one(void) { }
> +
>  #endif /* CONFIG_MMU_GATHER_RCU_TABLE_FREE */
>
>
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 674b111a24fa7..c3d3ce596bff7 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -1057,6 +1057,7 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address,
>         _pmd = pmdp_collapse_flush(vma, address, pmd);
>         spin_unlock(pmd_ptl);
>         mmu_notifier_invalidate_range_end(&range);
> +       tlb_remove_table_sync_one();
>
>         spin_lock(pte_ptl);
>         result =  __collapse_huge_page_isolate(vma, address, pte, cc,
> @@ -1415,6 +1416,7 @@ static void collapse_and_free_pmd(struct mm_struct *mm, struct vm_area_struct *v
>                 lockdep_assert_held_write(&vma->anon_vma->root->rwsem);
>
>         pmd = pmdp_collapse_flush(vma, addr, pmdp);
> +       tlb_remove_table_sync_one();
>         mm_dec_nr_ptes(mm);
>         page_table_check_pte_clear_range(mm, addr, pmd);
>         pte_free(mm, pmd_pgtable(pmd));
> diff --git a/mm/mmu_gather.c b/mm/mmu_gather.c
> index add4244e5790d..3a2c3f8cad2fe 100644
> --- a/mm/mmu_gather.c
> +++ b/mm/mmu_gather.c
> @@ -153,7 +153,7 @@ static void tlb_remove_table_smp_sync(void *arg)
>         /* Simply deliver the interrupt */
>  }
>
> -static void tlb_remove_table_sync_one(void)
> +void tlb_remove_table_sync_one(void)
>  {
>         /*
>          * This isn't an RCU grace period and hence the page-tables cannot be
> @@ -177,8 +177,6 @@ static void tlb_remove_table_free(struct mmu_table_batch *batch)
>
>  #else /* !CONFIG_MMU_GATHER_RCU_TABLE_FREE */
>
> -static void tlb_remove_table_sync_one(void) { }
> -
>  static void tlb_remove_table_free(struct mmu_table_batch *batch)
>  {
>         __tlb_remove_table_free(batch);
> --
> 2.38.1.584.g0f3c55d4c2-goog
>