Re: [PATCH 2/3] userfaultfd: UFFDIO_REMAP uABI

[Jann Horn]

On Thu, Sep 14, 2023 at 5:26 PM Suren Baghdasaryan <surenb@xxxxxxxxxx> wrote:
>
> From: Andrea Arcangeli <aarcange@xxxxxxxxxx>
>
> This implements the uABI of UFFDIO_REMAP.
>
> Notably one mode bitflag is also forwarded (and in turn known) by the
> lowlevel remap_pages method.
>
> Signed-off-by: Andrea Arcangeli <aarcange@xxxxxxxxxx>
> Signed-off-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>
> ---
>  fs/userfaultfd.c                 |  49 +++
>  include/linux/rmap.h             |   5 +
>  include/linux/userfaultfd_k.h    |  17 +
>  include/uapi/linux/userfaultfd.h |  22 ++
>  mm/huge_memory.c                 | 118 +++++++
>  mm/khugepaged.c                  |   3 +
>  mm/userfaultfd.c                 | 586 +++++++++++++++++++++++++++++++
>  7 files changed, 800 insertions(+)
>
> diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
> index 56eaae9dac1a..7bf64e7541c1 100644
> --- a/fs/userfaultfd.c
> +++ b/fs/userfaultfd.c
> @@ -2027,6 +2027,52 @@ static inline unsigned int uffd_ctx_features(__u64 user_features)
>         return (unsigned int)user_features | UFFD_FEATURE_INITIALIZED;
>  }
>
> +static int userfaultfd_remap(struct userfaultfd_ctx *ctx,
> +                            unsigned long arg)
> +{
> +       __s64 ret;
> +       struct uffdio_remap uffdio_remap;
> +       struct uffdio_remap __user *user_uffdio_remap;
> +       struct userfaultfd_wake_range range;
> +
> +       user_uffdio_remap = (struct uffdio_remap __user *) arg;
> +
> +       ret = -EFAULT;
> +       if (copy_from_user(&uffdio_remap, user_uffdio_remap,
> +                          /* don't copy "remap" last field */
> +                          sizeof(uffdio_remap)-sizeof(__s64)))
> +               goto out;
> +
> +       ret = validate_range(ctx->mm, uffdio_remap.dst, uffdio_remap.len);
> +       if (ret)
> +               goto out;
> +       ret = validate_range(current->mm, uffdio_remap.src, uffdio_remap.len);
> +       if (ret)
> +               goto out;
> +       ret = -EINVAL;
> +       if (uffdio_remap.mode & ~(UFFDIO_REMAP_MODE_ALLOW_SRC_HOLES|
> +                                 UFFDIO_REMAP_MODE_DONTWAKE))
> +               goto out;

Do you not need mmget_not_zero(ctx->mm) to make sure the MM can't be
concurrently torn down while remap_pages() is running, similar to what
the other userfaultfd ioctl handlers do?

> +       ret = remap_pages(ctx->mm, current->mm,
> +                         uffdio_remap.dst, uffdio_remap.src,
> +                         uffdio_remap.len, uffdio_remap.mode);
> +       if (unlikely(put_user(ret, &user_uffdio_remap->remap)))
> +               return -EFAULT;
> +       if (ret < 0)
> +               goto out;
> +       /* len == 0 would wake all */
> +       BUG_ON(!ret);
> +       range.len = ret;
> +       if (!(uffdio_remap.mode & UFFDIO_REMAP_MODE_DONTWAKE)) {
> +               range.start = uffdio_remap.dst;
> +               wake_userfault(ctx, &range);
> +       }
> +       ret = range.len == uffdio_remap.len ? 0 : -EAGAIN;
> +out:
> +       return ret;
> +}
[...]
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 064fbd90822b..c7a9880a1f6a 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -1932,6 +1932,124 @@ int change_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
>         return ret;
>  }
>
> +#ifdef CONFIG_USERFAULTFD
> +/*
> + * The PT lock for src_pmd and the mmap_lock for reading are held by
> + * the caller, but it must return after releasing the
> + * page_table_lock. We're guaranteed the src_pmd is a pmd_trans_huge
> + * until the PT lock of the src_pmd is released. Just move the page
> + * from src_pmd to dst_pmd if possible. Return zero if succeeded in
> + * moving the page, -EAGAIN if it needs to be repeated by the caller,
> + * or other errors in case of failure.
> + */
> +int remap_pages_huge_pmd(struct mm_struct *dst_mm,
> +                        struct mm_struct *src_mm,
> +                        pmd_t *dst_pmd, pmd_t *src_pmd,
> +                        pmd_t dst_pmdval,
> +                        struct vm_area_struct *dst_vma,
> +                        struct vm_area_struct *src_vma,
> +                        unsigned long dst_addr,
> +                        unsigned long src_addr)
> +{
> +       pmd_t _dst_pmd, src_pmdval;
> +       struct page *src_page;
> +       struct anon_vma *src_anon_vma, *dst_anon_vma;
> +       spinlock_t *src_ptl, *dst_ptl;
> +       pgtable_t pgtable;
> +       struct mmu_notifier_range range;
> +
> +       src_pmdval = *src_pmd;
> +       src_ptl = pmd_lockptr(src_mm, src_pmd);
> +
> +       BUG_ON(!pmd_trans_huge(src_pmdval));
> +       BUG_ON(!pmd_none(dst_pmdval));

Why can we assert that pmd_none(dst_pmdval) is true here? Can we not
have concurrent faults (or userfaultfd operations) populating that
PMD?

> +       BUG_ON(!spin_is_locked(src_ptl));
> +       mmap_assert_locked(src_mm);
> +       mmap_assert_locked(dst_mm);
> +       BUG_ON(src_addr & ~HPAGE_PMD_MASK);
> +       BUG_ON(dst_addr & ~HPAGE_PMD_MASK);
> +
> +       src_page = pmd_page(src_pmdval);
> +       BUG_ON(!PageHead(src_page));
> +       BUG_ON(!PageAnon(src_page));
> +       if (unlikely(page_mapcount(src_page) != 1)) {
> +               spin_unlock(src_ptl);
> +               return -EBUSY;
> +       }
> +
> +       get_page(src_page);
> +       spin_unlock(src_ptl);
> +
> +       mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, src_mm, src_addr,
> +                               src_addr + HPAGE_PMD_SIZE);
> +       mmu_notifier_invalidate_range_start(&range);
> +
> +       /* block all concurrent rmap walks */
> +       lock_page(src_page);
> +
> +       /*
> +        * split_huge_page walks the anon_vma chain without the page
> +        * lock. Serialize against it with the anon_vma lock, the page
> +        * lock is not enough.
> +        */
> +       src_anon_vma = folio_get_anon_vma(page_folio(src_page));
> +       if (!src_anon_vma) {
> +               unlock_page(src_page);
> +               put_page(src_page);
> +               mmu_notifier_invalidate_range_end(&range);
> +               return -EAGAIN;
> +       }
> +       anon_vma_lock_write(src_anon_vma);
> +
> +       dst_ptl = pmd_lockptr(dst_mm, dst_pmd);
> +       double_pt_lock(src_ptl, dst_ptl);
> +       if (unlikely(!pmd_same(*src_pmd, src_pmdval) ||
> +                    !pmd_same(*dst_pmd, dst_pmdval) ||
> +                    page_mapcount(src_page) != 1)) {
> +               double_pt_unlock(src_ptl, dst_ptl);
> +               anon_vma_unlock_write(src_anon_vma);
> +               put_anon_vma(src_anon_vma);
> +               unlock_page(src_page);
> +               put_page(src_page);
> +               mmu_notifier_invalidate_range_end(&range);
> +               return -EAGAIN;
> +       }
> +
> +       BUG_ON(!PageHead(src_page));
> +       BUG_ON(!PageAnon(src_page));
> +       /* the PT lock is enough to keep the page pinned now */
> +       put_page(src_page);
> +
> +       dst_anon_vma = (void *) dst_vma->anon_vma + PAGE_MAPPING_ANON;
> +       WRITE_ONCE(src_page->mapping, (struct address_space *) dst_anon_vma);
> +       WRITE_ONCE(src_page->index, linear_page_index(dst_vma, dst_addr));
> +
> +       if (!pmd_same(pmdp_huge_clear_flush(src_vma, src_addr, src_pmd),
> +                     src_pmdval))
> +               BUG_ON(1);

I'm not sure we can assert that the PMDs are exactly equal; the CPU
might have changed the A/D bits under us?

> +       _dst_pmd = mk_huge_pmd(src_page, dst_vma->vm_page_prot);
> +       _dst_pmd = maybe_pmd_mkwrite(pmd_mkdirty(_dst_pmd), dst_vma);
> +       set_pmd_at(dst_mm, dst_addr, dst_pmd, _dst_pmd);
> +
> +       pgtable = pgtable_trans_huge_withdraw(src_mm, src_pmd);
> +       pgtable_trans_huge_deposit(dst_mm, dst_pmd, pgtable);

Are we allowed to move page tables between mm_structs on all
architectures? The first example I found that looks a bit dodgy,
looking through various architectures' pte_alloc_one(), is s390's
page_table_alloc() which looks like page tables are tied to per-MM
lists sometimes.
If that's not allowed, we might have to allocate a new deposit table
and free the old one or something like that.

> +       if (dst_mm != src_mm) {
> +               add_mm_counter(dst_mm, MM_ANONPAGES, HPAGE_PMD_NR);
> +               add_mm_counter(src_mm, MM_ANONPAGES, -HPAGE_PMD_NR);
> +       }
> +       double_pt_unlock(src_ptl, dst_ptl);
> +
> +       anon_vma_unlock_write(src_anon_vma);
> +       put_anon_vma(src_anon_vma);
> +
> +       /* unblock rmap walks */
> +       unlock_page(src_page);
> +
> +       mmu_notifier_invalidate_range_end(&range);
> +       return 0;
> +}
> +#endif /* CONFIG_USERFAULTFD */
> +
>  /*
>   * Returns page table lock pointer if a given pmd maps a thp, NULL otherwise.
>   *
[...]
> diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
> index 96d9eae5c7cc..0cca60dfa8f8 100644
> --- a/mm/userfaultfd.c
> +++ b/mm/userfaultfd.c
[...]
> +ssize_t remap_pages(struct mm_struct *dst_mm, struct mm_struct *src_mm,
> +                   unsigned long dst_start, unsigned long src_start,
> +                   unsigned long len, __u64 mode)
> +{
[...]
> +
> +       if (pgprot_val(src_vma->vm_page_prot) !=
> +           pgprot_val(dst_vma->vm_page_prot))
> +               goto out;

Does this check intentionally allow moving pages from a
PROT_READ|PROT_WRITE anonymous private VMA into a PROT_READ anonymous
private VMA (on architectures like x86 and arm64 where CoW memory has
the same protection flags as read-only memory), but forbid moving them
from a PROT_READ|PROT_EXEC VMA into a PROT_READ VMA? I think this
check needs at least a comment to explain what's going on here.

> +       /* only allow remapping if both are mlocked or both aren't */
> +       if ((src_vma->vm_flags & VM_LOCKED) ^ (dst_vma->vm_flags & VM_LOCKED))
> +               goto out;
> +
> +       /*
> +        * Be strict and only allow remap_pages if either the src or
> +        * dst range is registered in the userfaultfd to prevent
> +        * userland errors going unnoticed. As far as the VM
> +        * consistency is concerned, it would be perfectly safe to
> +        * remove this check, but there's no useful usage for
> +        * remap_pages ouside of userfaultfd registered ranges. This
> +        * is after all why it is an ioctl belonging to the
> +        * userfaultfd and not a syscall.
> +        *
> +        * Allow both vmas to be registered in the userfaultfd, just
> +        * in case somebody finds a way to make such a case useful.
> +        * Normally only one of the two vmas would be registered in
> +        * the userfaultfd.
> +        */
> +       if (!dst_vma->vm_userfaultfd_ctx.ctx &&
> +           !src_vma->vm_userfaultfd_ctx.ctx)
> +               goto out;
> +
> +       /*
> +        * FIXME: only allow remapping across anonymous vmas,
> +        * tmpfs should be added.
> +        */
> +       if (src_vma->vm_ops || dst_vma->vm_ops)
> +               goto out;

I don't think it's okay to check for anonymous VMAs by checking
->vm_ops. There are some weird drivers whose ->mmap helpers don't set
->vm_ops and instead just shove all the necessary PTEs into the VMA
right on ->mmap, so I think they end up with ->vm_ops==NULL. For
example, kcov_mmap() looks that way. I'm not sure how common this is.

Though, uuuuuh, I guess if that's true, the existing
vma_is_anonymous() is broken, since that also just checks ->vm_ops?
I'm not sure what the consequences of that would be... Either way,
vma_is_anonymous() might be the better way to check for anonymous VMAs
here, and someone should figure out whether vma_is_anonymous() needs
to be fixed.

> +       /*
> +        * Ensure the dst_vma has a anon_vma or this page
> +        * would get a NULL anon_vma when moved in the
> +        * dst_vma.
> +        */
> +       err = -ENOMEM;
> +       if (unlikely(anon_vma_prepare(dst_vma)))
> +               goto out;
> +
> +       for (src_addr = src_start, dst_addr = dst_start;
> +            src_addr < src_start + len;) {
> +               spinlock_t *ptl;
> +               pmd_t dst_pmdval;
> +
> +               BUG_ON(dst_addr >= dst_start + len);
> +               src_pmd = mm_find_pmd(src_mm, src_addr);

(this would blow up pretty badly if we could have transparent huge PUD
in the region but I think that's limited to file VMAs so it's fine as
it currently is)

> +               if (unlikely(!src_pmd)) {
> +                       if (!(mode & UFFDIO_REMAP_MODE_ALLOW_SRC_HOLES)) {
> +                               err = -ENOENT;
> +                               break;
> +                       }
> +                       src_pmd = mm_alloc_pmd(src_mm, src_addr);
> +                       if (unlikely(!src_pmd)) {
> +                               err = -ENOMEM;
> +                               break;
> +                       }
> +               }
> +               dst_pmd = mm_alloc_pmd(dst_mm, dst_addr);
> +               if (unlikely(!dst_pmd)) {
> +                       err = -ENOMEM;
> +                       break;
> +               }
> +
> +               dst_pmdval = pmdp_get_lockless(dst_pmd);
> +               /*
> +                * If the dst_pmd is mapped as THP don't
> +                * override it and just be strict.
> +                */
> +               if (unlikely(pmd_trans_huge(dst_pmdval))) {
> +                       err = -EEXIST;
> +                       break;
> +               }

This check is racy because the dst_pmd can still change at this point,
from previously pointing to a zeroed PMD to now pointing to a
hugepage, right? And we rely on remap_pages_pte() and
remap_pages_huge_pmd() to recheck for that?
If yes, maybe add a comment noting this and explaining why we want this check.

> +               ptl = pmd_trans_huge_lock(src_pmd, src_vma);
> +               if (ptl) {
> +                       /*
> +                        * Check if we can move the pmd without
> +                        * splitting it. First check the address
> +                        * alignment to be the same in src/dst.  These
> +                        * checks don't actually need the PT lock but
> +                        * it's good to do it here to optimize this
> +                        * block away at build time if
> +                        * CONFIG_TRANSPARENT_HUGEPAGE is not set.
> +                        */
> +                       if (thp_aligned == -1)
> +                               thp_aligned = ((src_addr & ~HPAGE_PMD_MASK) ==
> +                                              (dst_addr & ~HPAGE_PMD_MASK));
> +                       if (!thp_aligned || (src_addr & ~HPAGE_PMD_MASK) ||

This seems overly complicated, the only case when you can move a huge
PMD is if both addresses are hugepage-aligned and you have enough
length for one hugepage:

(src_addr & ~HPAGE_PMD_MASK) == 0 && (dst_addr & ~HPAGE_PMD_MASK) == 0
&& (src_start + len - src_addr >= HPAGE_PMD_SIZE).

> +                           !pmd_none(dst_pmdval) ||
> +                           src_start + len - src_addr < HPAGE_PMD_SIZE) {
> +                               spin_unlock(ptl);
> +                               /* Fall through */
> +                               split_huge_pmd(src_vma, src_pmd, src_addr);
> +                       } else {
> +                               err = remap_pages_huge_pmd(dst_mm,
> +                                                          src_mm,
> +                                                          dst_pmd,
> +                                                          src_pmd,
> +                                                          dst_pmdval,
> +                                                          dst_vma,
> +                                                          src_vma,
> +                                                          dst_addr,
> +                                                          src_addr);
> +                               cond_resched();
> +
> +                               if (!err) {
> +                                       dst_addr += HPAGE_PMD_SIZE;
> +                                       src_addr += HPAGE_PMD_SIZE;
> +                                       moved += HPAGE_PMD_SIZE;
> +                               }
> +
> +                               if ((!err || err == -EAGAIN) &&
> +                                   fatal_signal_pending(current))
> +                                       err = -EINTR;
> +
> +                               if (err && err != -EAGAIN)
> +                                       break;
> +
> +                               continue;
> +                       }
> +               }
> +
> +               if (pmd_none(*src_pmd)) {
> +                       if (!(mode & UFFDIO_REMAP_MODE_ALLOW_SRC_HOLES)) {
> +                               err = -ENOENT;
> +                               break;
> +                       }
> +                       if (unlikely(__pte_alloc(src_mm, src_pmd))) {
> +                               err = -ENOMEM;
> +                               break;
> +                       }
> +               }
> +
> +               if (unlikely(pmd_none(dst_pmdval)) &&
> +                   unlikely(__pte_alloc(dst_mm, dst_pmd))) {

Maybe just use pte_alloc() here?

> +                       err = -ENOMEM;
> +                       break;
> +               }
> +
> +               err = remap_pages_pte(dst_mm, src_mm,
> +                                     dst_pmd, src_pmd,
> +                                     dst_vma, src_vma,
> +                                     dst_addr, src_addr,
> +                                     mode);
> +
> +               cond_resched();
> +
> +               if (!err) {
> +                       dst_addr += PAGE_SIZE;
> +                       src_addr += PAGE_SIZE;
> +                       moved += PAGE_SIZE;
> +               }
> +
> +               if ((!err || err == -EAGAIN) &&
> +                   fatal_signal_pending(current))
> +                       err = -EINTR;
> +
> +               if (err && err != -EAGAIN)
> +                       break;
> +       }
> +
> +out:
> +       mmap_read_unlock(dst_mm);
> +       if (dst_mm != src_mm)
> +               mmap_read_unlock(src_mm);
> +       BUG_ON(moved < 0);
> +       BUG_ON(err > 0);
> +       BUG_ON(!moved && !err);
> +       return moved ? moved : err;
> +}

Maybe you could try whether this function would look simpler with a
shape roughly like:

for (src_addr = ...; src_addr < ...;) {
  unsigned long step_size;

  if (hugepage case) {
    if (have to split) {
      split it;
      continue;
    }
    step_size = HPAGE_PMD_SIZE;
    ...
  } else {
    ... 4k case ...
    step_size = PAGE_SIZE;
  }
  ...
  cond_resched();
  if (!err) {
    dst_addr += step_size;
    src_addr += step_size;
    moved += step_size;
  }
  ...
}