Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220606214414.736109-1-shy828301@gmail.com> <20220606214414.736109-5-shy828301@gmail.com>
In-Reply-To: <20220606214414.736109-5-shy828301@gmail.com>
From:   "Zach O'Keefe" <zokeefe@google.com>
Date:   Thu, 9 Jun 2022 18:51:31 -0700
Message-ID: <CAAa6QmSCk+7HLtLOzbTYev7dAM0aff-f4USV3AXAQdSKu7_6HA@mail.gmail.com>
Subject: Re: [v3 PATCH 4/7] mm: khugepaged: use transhuge_vma_suitable replace open-code
To:     Yang Shi <shy828301@gmail.com>
Cc:     vbabka@suse.cz, kirill.shutemov@linux.intel.com,
        willy@infradead.org, akpm@linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Jun 6, 2022 at 2:44 PM Yang Shi <shy828301@gmail.com> wrote:
>
> The hugepage_vma_revalidate() needs to check if the address is still in
> the aligned HPAGE_PMD_SIZE area of the vma when reacquiring mmap_lock,
> but it was open-coded, use transhuge_vma_suitable() to do the job.  And
> add proper comments for transhuge_vma_suitable().
>
> Signed-off-by: Yang Shi <shy828301@gmail.com>
> ---
>  include/linux/huge_mm.h | 6 ++++++
>  mm/khugepaged.c         | 5 +----
>  2 files changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index a8f61db47f2a..79d5919beb83 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -128,6 +128,12 @@ static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
>         return false;
>  }
>
> +/*
> + * Do the below checks:
> + *   - For non-anon vma, check if the vm_pgoff is HPAGE_PMD_NR aligned.
> + *   - For all vmas, check if the haddr is in an aligned HPAGE_PMD_SIZE
> + *     area.
> + */

AFAIK we aren't checking if vm_pgoff is HPAGE_PMD_NR aligned, but
rather that linear_page_index(vma, round_up(vma->vm_start,
HPAGE_PMD_SIZE)) is HPAGE_PMD_NR aligned within vma->vm_file. I was
pretty confused about this (hopefully I have it right now - if not -
case and point :) ), so it might be a good opportunity to add some
extra commentary to help future travelers understand why this
constraint exists.

Also I wonder while we're at it if we can rename this to
transhuge_addr_aligned() or transhuge_addr_suitable() or something.

Otherwise I think the change is a nice cleanup.

>  static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
>                 unsigned long addr)
>  {
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 7a5d1c1a1833..ca1754d3a827 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -951,7 +951,6 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
>                 struct vm_area_struct **vmap)
>  {
>         struct vm_area_struct *vma;
> -       unsigned long hstart, hend;
>
>         if (unlikely(khugepaged_test_exit(mm)))
>                 return SCAN_ANY_PROCESS;
> @@ -960,9 +959,7 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
>         if (!vma)
>                 return SCAN_VMA_NULL;
>
> -       hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK;
> -       hend = vma->vm_end & HPAGE_PMD_MASK;
> -       if (address < hstart || address + HPAGE_PMD_SIZE > hend)
> +       if (!transhuge_vma_suitable(vma, address))
>                 return SCAN_ADDRESS_RANGE;
>         if (!hugepage_vma_check(vma, vma->vm_flags))
>                 return SCAN_VMA_CHECK;
> --
> 2.26.3
>
>