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