From mboxrd@z Thu Jan 1 00:00:00 1970 From: akpm@linux-foundation.org Subject: + mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages.patch added to -mm tree Date: Sat, 20 Jun 2020 17:18:48 -0700 Message-ID: <20200621001848.GzMgT%akpm@linux-foundation.org> Reply-To: linux-kernel@vger.kernel.org Return-path: Received: from mail.kernel.org ([198.145.29.99]:53066 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728480AbgFUASt (ORCPT ); Sat, 20 Jun 2020 20:18:49 -0400 Sender: mm-commits-owner@vger.kernel.org List-Id: mm-commits@vger.kernel.org To: mm-commits@vger.kernel.org, soheil@google.com, rientjes@google.com, hughd@google.com, edumazet@google.com, arjunroy@google.com The patch titled Subject: mm/memory.c: properly pte_offset_map_lock/unlock in vm_insert_pages() has been added to the -mm tree. Its filename is mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages.patch This patch should soon appear at http://ozlabs.org/~akpm/mmots/broken-out/mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages.patch and later at http://ozlabs.org/~akpm/mmotm/broken-out/mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages.patch Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next and is updated there every 3-4 working days ------------------------------------------------------ From: Arjun Roy Subject: mm/memory.c: properly pte_offset_map_lock/unlock in vm_insert_pages() Calls to pte_offset_map() in vm_insert_pages() are erroneously not matched with a call to pte_unmap(). This would cause problems on architectures where that is not a no-op. This patch does away with the non-traditional locking in the existing code, and instead uses pte_offset_map_lock/unlock() as usual, incrementing PTE as necessary. The PTE pointer is kept within bounds since we clamp it with PTRS_PER_PTE. Link: http://lkml.kernel.org/r/20200618220446.20284-1-arjunroy.kdev@gmail.com Fixes: 8cd3984d81d5 ("mm/memory.c: add vm_insert_pages()") Signed-off-by: Arjun Roy Acked-by: David Rientjes Cc: Eric Dumazet Cc: Hugh Dickins Cc: Soheil Hassas Yeganeh Signed-off-by: Andrew Morton --- mm/memory.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) --- a/mm/memory.c~mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages +++ a/mm/memory.c @@ -1498,7 +1498,7 @@ out: } #ifdef pte_index -static int insert_page_in_batch_locked(struct mm_struct *mm, pmd_t *pmd, +static int insert_page_in_batch_locked(struct mm_struct *mm, pte_t *pte, unsigned long addr, struct page *page, pgprot_t prot) { int err; @@ -1506,8 +1506,9 @@ static int insert_page_in_batch_locked(s if (!page_count(page)) return -EINVAL; err = validate_page_before_insert(page); - return err ? err : insert_page_into_pte_locked( - mm, pte_offset_map(pmd, addr), addr, page, prot); + if (err) + return err; + return insert_page_into_pte_locked(mm, pte, addr, page, prot); } /* insert_pages() amortizes the cost of spinlock operations @@ -1517,7 +1518,8 @@ static int insert_pages(struct vm_area_s struct page **pages, unsigned long *num, pgprot_t prot) { pmd_t *pmd = NULL; - spinlock_t *pte_lock = NULL; + pte_t *start_pte, *pte; + spinlock_t *pte_lock; struct mm_struct *const mm = vma->vm_mm; unsigned long curr_page_idx = 0; unsigned long remaining_pages_total = *num; @@ -1536,18 +1538,17 @@ more: ret = -ENOMEM; if (pte_alloc(mm, pmd)) goto out; - pte_lock = pte_lockptr(mm, pmd); while (pages_to_write_in_pmd) { int pte_idx = 0; const int batch_size = min_t(int, pages_to_write_in_pmd, 8); - spin_lock(pte_lock); - for (; pte_idx < batch_size; ++pte_idx) { - int err = insert_page_in_batch_locked(mm, pmd, + start_pte = pte_offset_map_lock(mm, pmd, addr, &pte_lock); + for (pte = start_pte; pte_idx < batch_size; ++pte, ++pte_idx) { + int err = insert_page_in_batch_locked(mm, pte, addr, pages[curr_page_idx], prot); if (unlikely(err)) { - spin_unlock(pte_lock); + pte_unmap_unlock(start_pte, pte_lock); ret = err; remaining_pages_total -= pte_idx; goto out; @@ -1555,7 +1556,7 @@ more: addr += PAGE_SIZE; ++curr_page_idx; } - spin_unlock(pte_lock); + pte_unmap_unlock(start_pte, pte_lock); pages_to_write_in_pmd -= batch_size; remaining_pages_total -= batch_size; } _ Patches currently in -mm which might be from arjunroy@google.com are mm-memoryc-properly-pte_offset_map_lock-unlock-in-vm_insert_pages.patch