[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZrZOUaj2Kfe0CimQ@google.com>
Date: Fri, 9 Aug 2024 18:13:53 +0100
From: Vincent Donnefort <vdonnefort@...gle.com>
To: David Hildenbrand <david@...hat.com>, g@...gle.com
Cc: linux-kernel@...r.kernel.org, linux-mm@...ck.org,
Andrew Morton <akpm@...ux-foundation.org>,
"Matthew Wilcox (Oracle)" <willy@...radead.org>,
Hugh Dickins <hughd@...gle.com>,
Ryan Roberts <ryan.roberts@....com>,
Yin Fengwei <fengwei.yin@...el.com>,
Mike Kravetz <mike.kravetz@...cle.com>,
Muchun Song <muchun.song@...ux.dev>, Peter Xu <peterx@...hat.com>,
surenb@...gle.com
Subject: Re: [PATCH v2 08/40] mm/memory: page_add_file_rmap() ->
folio_add_file_rmap_[pte|pmd]()
Hi,
Sorry, reviving this thread as I have ran into something weird:
On Wed, Dec 20, 2023 at 11:44:32PM +0100, David Hildenbrand wrote:
> Let's convert insert_page_into_pte_locked() and do_set_pmd(). While at it,
> perform some folio conversion.
>
> Reviewed-by: Yin Fengwei <fengwei.yin@...el.com>
> Reviewed-by: Ryan Roberts <ryan.roberts@....com>
> Signed-off-by: David Hildenbrand <david@...hat.com>
> ---
> mm/memory.c | 14 ++++++++------
> 1 file changed, 8 insertions(+), 6 deletions(-)
>
> diff --git a/mm/memory.c b/mm/memory.c
> index 7f957e5a84311..c77d3952d261f 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
[...]
> vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page *page)
> {
> + struct folio *folio = page_folio(page);
> struct vm_area_struct *vma = vmf->vma;
> bool write = vmf->flags & FAULT_FLAG_WRITE;
> unsigned long haddr = vmf->address & HPAGE_PMD_MASK;
> @@ -4418,8 +4421,7 @@ vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page *page)
> if (!thp_vma_suitable_order(vma, haddr, PMD_ORDER))
> return ret;
>
> - page = compound_head(page);
> - if (compound_order(page) != HPAGE_PMD_ORDER)
> + if (page != &folio->page || folio_order(folio) != HPAGE_PMD_ORDER)
> return ret;
Is this `page != &folio->page` expected? I believe this check wasn't there
before as we had `page = compound_head()`.
It breaks the installation of a PMD level mapping for shmem when the fault
address is in the middle of this block. In its fault path, shmem sets
vmf->page = folio_file_page(folio, vmf->pgoff)
which fails this test above.
>
> /*
> @@ -4428,7 +4430,7 @@ vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page *page)
> * check. This kind of THP just can be PTE mapped. Access to
> * the corrupted subpage should trigger SIGBUS as expected.
> */
> - if (unlikely(PageHasHWPoisoned(page)))
> + if (unlikely(folio_test_has_hwpoisoned(folio)))
> return ret;
>
> /*
> @@ -4452,7 +4454,7 @@ vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page *page)
> entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
>
> add_mm_counter(vma->vm_mm, mm_counter_file(page), HPAGE_PMD_NR);
> - page_add_file_rmap(page, vma, true);
> + folio_add_file_rmap_pmd(folio, page, vma);
>
> /*
> * deposit and withdraw with pmd lock held
> --
> 2.43.0
>
Powered by blists - more mailing lists