The patch titled
Subject: madvise: cleanup swapin_walk_pmd_entry()
has been removed from the -mm tree. Its filename was
madvise-cleanup-swapin_walk_pmd_entry.patch
This patch was dropped because it was nacked
------------------------------------------------------
From: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx>
Subject: madvise: cleanup swapin_walk_pmd_entry()
With the recent update on page table walker, we can use common code for
the walking more. Unlike many other users, this swapin_walk expects to
handle swap entries. As a result we should be careful about ptl locking.
Swapin operation, read_swap_cache_async(), could cause page reclaim, so we
can't keep holding ptl throughout this pte loop. In order to properly
handle ptl in pte_entry(), this patch adds two new members on struct
mm_walk.
This cleanup is necessary to get to the final form of page table walker,
where we should do all caller's specific work on leaf entries (IOW, all
pmd_entry() should be used for trans_pmd.)
Signed-off-by: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx>
Cc: Hugh Dickins <hughd@xxxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxx>
Cc: "Kirill A. Shutemov" <kirill@xxxxxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxx>
Cc: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@xxxxxxxxxxxxxxxxxx>
Cc: Pavel Emelyanov <xemul@xxxxxxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxx>
Cc: Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/mm.h | 4 +++
mm/madvise.c | 56 ++++++++++++++++++-------------------------
mm/pagewalk.c | 5 ++-
3 files changed, 31 insertions(+), 34 deletions(-)
diff -puN include/linux/mm.h~madvise-cleanup-swapin_walk_pmd_entry include/linux/mm.h
--- a/include/linux/mm.h~madvise-cleanup-swapin_walk_pmd_entry
+++ a/include/linux/mm.h
@@ -1112,6 +1112,8 @@ void unmap_vmas(struct mmu_gather *tlb,
* walk_page_test() for how the skip control works.
* @mm: mm_struct representing the target process of page table walk
* @vma: vma currently walked
+ * @pmd: current pmd entry
+ * @ptl: page table lock associated with current entry
* @control: walk control flag
* @private: private data for callbacks' use
*
@@ -1130,6 +1132,8 @@ struct mm_walk {
struct mm_walk *walk);
struct mm_struct *mm;
struct vm_area_struct *vma;
+ pmd_t *pmd;
+ spinlock_t *ptl;
int control;
void *private;
};
diff -puN mm/madvise.c~madvise-cleanup-swapin_walk_pmd_entry mm/madvise.c
--- a/mm/madvise.c~madvise-cleanup-swapin_walk_pmd_entry
+++ a/mm/madvise.c
@@ -135,38 +135,31 @@ out:
}
#ifdef CONFIG_SWAP
-static int swapin_walk_pmd_entry(pmd_t *pmd, unsigned long start,
+/*
+ * Assuming that page table walker holds page table lock.
+ */
+static int swapin_walk_pte_entry(pte_t *pte, unsigned long start,
unsigned long end, struct mm_walk *walk)
{
- pte_t *orig_pte;
- struct vm_area_struct *vma = walk->private;
- unsigned long index;
-
- if (pmd_none_or_trans_huge_or_clear_bad(pmd))
- return 0;
-
- for (index = start; index != end; index += PAGE_SIZE) {
- pte_t pte;
- swp_entry_t entry;
- struct page *page;
- spinlock_t *ptl;
-
- orig_pte = pte_offset_map_lock(vma->vm_mm, pmd, start, &ptl);
- pte = *(orig_pte + ((index - start) / PAGE_SIZE));
- pte_unmap_unlock(orig_pte, ptl);
-
- if (pte_present(pte) || pte_none(pte) || pte_file(pte))
- continue;
- entry = pte_to_swp_entry(pte);
- if (unlikely(non_swap_entry(entry)))
- continue;
-
- page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
- vma, index);
- if (page)
- page_cache_release(page);
- }
-
+ pte_t ptent;
+ pte_t *orig_pte = pte - ((start & (PMD_SIZE - 1)) >> PAGE_SHIFT);
+ swp_entry_t entry;
+ struct page *page;
+
+ ptent = *pte;
+ pte_unmap_unlock(orig_pte, walk->ptl);
+ if (pte_present(ptent) || pte_none(ptent) || pte_file(ptent))
+ goto lock;
+ entry = pte_to_swp_entry(ptent);
+ if (unlikely(non_swap_entry(entry)))
+ goto lock;
+ page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
+ walk->vma, start);
+ if (page)
+ page_cache_release(page);
+lock:
+ pte_offset_map(walk->pmd, start & PMD_MASK);
+ spin_lock(walk->ptl);
return 0;
}
@@ -175,8 +168,7 @@ static void force_swapin_readahead(struc
{
struct mm_walk walk = {
.mm = vma->vm_mm,
- .pmd_entry = swapin_walk_pmd_entry,
- .private = vma,
+ .pte_entry = swapin_walk_pte_entry,
};
walk_page_range(start, end, &walk);
diff -puN mm/pagewalk.c~madvise-cleanup-swapin_walk_pmd_entry mm/pagewalk.c
--- a/mm/pagewalk.c~madvise-cleanup-swapin_walk_pmd_entry
+++ a/mm/pagewalk.c
@@ -20,7 +20,8 @@ static int walk_pte_range(pmd_t *pmd, un
spinlock_t *ptl;
int err = 0;
- orig_pte = pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
+ walk->pmd = pmd;
+ orig_pte = pte = pte_offset_map_lock(mm, pmd, addr, &walk->ptl);
do {
if (pte_none(*pte)) {
if (walk->pte_hole)
@@ -49,7 +50,7 @@ static int walk_pte_range(pmd_t *pmd, un
}
} while (pte++, addr += PAGE_SIZE, addr < end);
out_unlock:
- pte_unmap_unlock(orig_pte, ptl);
+ pte_unmap_unlock(orig_pte, walk->ptl);
cond_resched();
return addr == end ? 0 : err;
}
_
Patches currently in -mm which might be from n-horiguchi@xxxxxxxxxxxxx are
hwpoison-fix-the-handling-path-of-the-victimized-page-frame-that-belong-to-non-lur.patch
mm-hugetlbfs-fix-rmapping-for-anonymous-hugepages-with-page_pgoff.patch
mm-hugetlbfs-fix-rmapping-for-anonymous-hugepages-with-page_pgoff-v2.patch
mm-hugetlbfs-fix-rmapping-for-anonymous-hugepages-with-page_pgoff-v3.patch
mm-hugetlbfs-fix-rmapping-for-anonymous-hugepages-with-page_pgoff-v3-fix.patch
madvise-cleanup-swapin_walk_pmd_entry-fix.patch
memcg-separate-mem_cgroup_move_charge_pte_range.patch
memcg-separate-mem_cgroup_move_charge_pte_range-checkpatch-fixes.patch
arch-powerpc-mm-subpage-protc-cleanup-subpage_walk_pmd_entry.patch
mm-pagewalk-move-pmd_trans_huge_lock-from-callbacks-to-common-code.patch
mm-pagewalk-move-pmd_trans_huge_lock-from-callbacks-to-common-code-checkpatch-fixes.patch
mincore-apply-page-table-walker-on-do_mincore.patch
mincore-apply-page-table-walker-on-do_mincore-fix.patch
mm-introduce-do_shared_fault-and-drop-do_fault-fix-fix.patch
do_shared_fault-check-that-mmap_sem-is-held.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
