Re: [PATCH] mm/mremap: fix unaccount of memory on vma_merge() failure

[Brian Geffon]

On Tue, Aug 29, 2023 at 8:46 PM Anthony Yznaga
<anthony.yznaga@xxxxxxxxxx> wrote:
>
> Fix mremap so that only accounted memory is unaccounted if the
> mapping is expandable but vma_merge() fails.
>
> Fixes: fdbef6149135 ("mm/mremap: don't account pages in vma_to_resize()")
> Signed-off-by: Anthony Yznaga <anthony.yznaga@xxxxxxxxxx>

Looks correct to me.

Acked-by: Brian Geffon <bgeffon@xxxxxxxxxx>

> ---
>  mm/mremap.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/mm/mremap.c b/mm/mremap.c
> index 056478c106ee..07cdb04d4ab5 100644
> --- a/mm/mremap.c
> +++ b/mm/mremap.c
> @@ -1037,12 +1037,14 @@ SYSCALL_DEFINE5(mremap, unsigned long, addr, unsigned long, old_len,
>                         pgoff_t extension_pgoff = vma->vm_pgoff +
>                                 ((extension_start - vma->vm_start) >> PAGE_SHIFT);
>                         VMA_ITERATOR(vmi, mm, extension_start);
> +                       long charged = 0;
>
>                         if (vma->vm_flags & VM_ACCOUNT) {
>                                 if (security_vm_enough_memory_mm(mm, pages)) {
>                                         ret = -ENOMEM;
>                                         goto out;
>                                 }
> +                               charged = pages;
>                         }
>
>                         /*
> @@ -1058,7 +1060,7 @@ SYSCALL_DEFINE5(mremap, unsigned long, addr, unsigned long, old_len,
>                                 vma->vm_file, extension_pgoff, vma_policy(vma),
>                                 vma->vm_userfaultfd_ctx, anon_vma_name(vma));
>                         if (!vma) {
> -                               vm_unacct_memory(pages);
> +                               vm_unacct_memory(charged);
>                                 ret = -ENOMEM;
>                                 goto out;
>                         }
> --
> 2.39.3
>