linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Qi Zheng <zhengqi.arch@bytedance.com>
To: david@redhat.com, hughd@google.com, willy@infradead.org,
	muchun.song@linux.dev, vbabka@kernel.org,
	akpm@linux-foundation.org, rppt@kernel.org,
	vishal.moola@gmail.com, peterx@redhat.com, ryan.roberts@arm.com,
	christophe.leroy2@cs-soprasteria.com
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
	linux-arm-kernel@lists.infradead.org,
	linuxppc-dev@lists.ozlabs.org,
	Qi Zheng <zhengqi.arch@bytedance.com>
Subject: [PATCH v5 07/13] mm: khugepaged: collapse_pte_mapped_thp() use pte_offset_map_rw_nolock()
Date: Thu, 26 Sep 2024 14:46:20 +0800	[thread overview]
Message-ID: <055e42db68da00ac8ecab94bd2633c7cd965eb1c.1727332572.git.zhengqi.arch@bytedance.com> (raw)
In-Reply-To: <cover.1727332572.git.zhengqi.arch@bytedance.com>

In collapse_pte_mapped_thp(), we may modify the pte and pmd entry after
acquring the ptl, so convert it to using pte_offset_map_rw_nolock(). At
this time, the pte_same() check is not performed after the PTL held. So we
should get pgt_pmd and do pmd_same() check after the ptl held.

Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
---
 mm/khugepaged.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 6498721d4783a..854577f39957d 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -1605,7 +1605,7 @@ int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 	if (userfaultfd_armed(vma) && !(vma->vm_flags & VM_SHARED))
 		pml = pmd_lock(mm, pmd);
 
-	start_pte = pte_offset_map_nolock(mm, pmd, haddr, &ptl);
+	start_pte = pte_offset_map_rw_nolock(mm, pmd, haddr, &pgt_pmd, &ptl);
 	if (!start_pte)		/* mmap_lock + page lock should prevent this */
 		goto abort;
 	if (!pml)
@@ -1613,6 +1613,9 @@ int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 	else if (ptl != pml)
 		spin_lock_nested(ptl, SINGLE_DEPTH_NESTING);
 
+	if (unlikely(!pmd_same(pgt_pmd, pmdp_get_lockless(pmd))))
+		goto abort;
+
 	/* step 2: clear page table and adjust rmap */
 	for (i = 0, addr = haddr, pte = start_pte;
 	     i < HPAGE_PMD_NR; i++, addr += PAGE_SIZE, pte++) {
@@ -1645,7 +1648,6 @@ int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 		nr_ptes++;
 	}
 
-	pte_unmap(start_pte);
 	if (!pml)
 		spin_unlock(ptl);
 
@@ -1658,14 +1660,19 @@ int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 	/* step 4: remove empty page table */
 	if (!pml) {
 		pml = pmd_lock(mm, pmd);
-		if (ptl != pml)
+		if (ptl != pml) {
 			spin_lock_nested(ptl, SINGLE_DEPTH_NESTING);
+			if (unlikely(!pmd_same(pgt_pmd, pmdp_get_lockless(pmd)))) {
+				flush_tlb_mm(mm);
+				goto unlock;
+			}
+		}
 	}
 	pgt_pmd = pmdp_collapse_flush(vma, haddr, pmd);
 	pmdp_get_lockless_sync();
+	pte_unmap_unlock(start_pte, ptl);
 	if (ptl != pml)
-		spin_unlock(ptl);
-	spin_unlock(pml);
+		spin_unlock(pml);
 
 	mmu_notifier_invalidate_range_end(&range);
 
@@ -1685,6 +1692,7 @@ int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 		folio_ref_sub(folio, nr_ptes);
 		add_mm_counter(mm, mm_counter_file(folio), -nr_ptes);
 	}
+unlock:
 	if (start_pte)
 		pte_unmap_unlock(start_pte, ptl);
 	if (pml && pml != ptl)
-- 
2.20.1



  parent reply	other threads:[~2024-09-26  6:47 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-09-26  6:46 [PATCH v5 00/13] introduce pte_offset_map_{ro|rw}_nolock() Qi Zheng
2024-09-26  6:46 ` [PATCH v5 01/13] mm: pgtable: " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 02/13] powerpc: assert_pte_locked() use pte_offset_map_ro_nolock() Qi Zheng
2024-09-26  6:46 ` [PATCH v5 03/13] mm: filemap: filemap_fault_recheck_pte_none() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 04/13] mm: khugepaged: __collapse_huge_page_swapin() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 05/13] arm: adjust_pte() use pte_offset_map_rw_nolock() Qi Zheng
2024-09-26  6:46 ` [PATCH v5 06/13] mm: handle_pte_fault() " Qi Zheng
2024-09-26  6:46 ` Qi Zheng [this message]
2024-09-26  7:07   ` [PATCH v5 07/13] mm: khugepaged: collapse_pte_mapped_thp() " Muchun Song
2024-09-26  6:46 ` [PATCH v5 08/13] mm: copy_pte_range() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 09/13] mm: mremap: move_ptes() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 10/13] mm: page_vma_mapped_walk: map_pte() " Qi Zheng
2024-09-26  7:08   ` Muchun Song
2024-09-26  6:46 ` [PATCH v5 11/13] mm: userfaultfd: move_pages_pte() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 12/13] mm: multi-gen LRU: walk_pte_range() " Qi Zheng
2024-09-26  6:46 ` [PATCH v5 13/13] mm: pgtable: remove pte_offset_map_nolock() Qi Zheng

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=055e42db68da00ac8ecab94bd2633c7cd965eb1c.1727332572.git.zhengqi.arch@bytedance.com \
    --to=zhengqi.arch@bytedance.com \
    --cc=akpm@linux-foundation.org \
    --cc=christophe.leroy2@cs-soprasteria.com \
    --cc=david@redhat.com \
    --cc=hughd@google.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=muchun.song@linux.dev \
    --cc=peterx@redhat.com \
    --cc=rppt@kernel.org \
    --cc=ryan.roberts@arm.com \
    --cc=vbabka@kernel.org \
    --cc=vishal.moola@gmail.com \
    --cc=willy@infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox