Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1405539
| From | "Kirill A. Shutemov" <kirill@shutemov.name> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem |
| Date | 2016-05-23 19:50 +0200 |
| Message-ID | <rC29j-u0-15@gated-at.bofh.it> (permalink) |
| References | <rC1Gh-gw-3@gated-at.bofh.it> <rC1Gi-gw-11@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
On Mon, May 23, 2016 at 08:14:11PM +0300, Ebru Akagunduz wrote:
> Currently khugepaged makes swapin readahead under
> down_write. This patch supplies to make swapin
> readahead under down_read instead of down_write.
>
> The patch was tested with a test program that allocates
> 800MB of memory, writes to it, and then sleeps. The system
> was forced to swap out all. Afterwards, the test program
> touches the area by writing, it skips a page in each
> 20 pages of the area.
>
> Signed-off-by: Ebru Akagunduz <ebru.akagunduz@gmail.com>
> ---
> mm/huge_memory.c | 33 +++++++++++++++++++++++++++------
> 1 file changed, 27 insertions(+), 6 deletions(-)
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index feee44c..668bc07 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -2386,13 +2386,14 @@ static bool hugepage_vma_check(struct vm_area_struct *vma)
> * but with mmap_sem held to protect against vma changes.
> */
>
> -static void __collapse_huge_page_swapin(struct mm_struct *mm,
> +static bool __collapse_huge_page_swapin(struct mm_struct *mm,
> struct vm_area_struct *vma,
> unsigned long address, pmd_t *pmd)
> {
> unsigned long _address;
> pte_t *pte, pteval;
> int swapped_in = 0, ret = 0;
> + struct vm_area_struct *vma_orig = vma;
>
> pte = pte_offset_map(pmd, address);
> for (_address = address; _address < address + HPAGE_PMD_NR*PAGE_SIZE;
> @@ -2402,11 +2403,19 @@ static void __collapse_huge_page_swapin(struct mm_struct *mm,
> continue;
> swapped_in++;
> ret = do_swap_page(mm, vma, _address, pte, pmd,
> - FAULT_FLAG_ALLOW_RETRY|FAULT_FLAG_RETRY_NOWAIT,
> + FAULT_FLAG_ALLOW_RETRY,
> pteval);
> + /* do_swap_page returns VM_FAULT_RETRY with released mmap_sem */
> + if (ret & VM_FAULT_RETRY) {
> + down_read(&mm->mmap_sem);
> + vma = find_vma(mm, address);
> + /* vma is no longer available, don't continue to swapin */
> + if (vma != vma_orig)
> + return false;
> + }
> if (ret & VM_FAULT_ERROR) {
> trace_mm_collapse_huge_page_swapin(mm, swapped_in, 0);
> - return;
> + return false;
> }
> /* pte is unmapped now, we need to map it */
> pte = pte_offset_map(pmd, _address);
> @@ -2414,6 +2423,7 @@ static void __collapse_huge_page_swapin(struct mm_struct *mm,
> pte--;
> pte_unmap(pte);
> trace_mm_collapse_huge_page_swapin(mm, swapped_in, 1);
> + return true;
> }
>
> static void collapse_huge_page(struct mm_struct *mm,
> @@ -2459,7 +2469,7 @@ static void collapse_huge_page(struct mm_struct *mm,
> * gup_fast later hanlded by the ptep_clear_flush and the VM
> * handled by the anon_vma lock + PG_lock.
> */
> - down_write(&mm->mmap_sem);
> + down_read(&mm->mmap_sem);
> if (unlikely(khugepaged_test_exit(mm))) {
> result = SCAN_ANY_PROCESS;
> goto out;
> @@ -2490,9 +2500,20 @@ static void collapse_huge_page(struct mm_struct *mm,
> * Don't perform swapin readahead when the system is under pressure,
> * to avoid unnecessary resource consumption.
> */
> - if (allocstall == curr_allocstall && swap != 0)
> - __collapse_huge_page_swapin(mm, vma, address, pmd);
> + if (allocstall == curr_allocstall && swap != 0) {
> + /*
> + * __collapse_huge_page_swapin always returns with mmap_sem
> + * locked. If it fails, release mmap_sem and jump directly
> + * label out. Continuing to collapse causes inconsistency.
> + */
> + if (!__collapse_huge_page_swapin(mm, vma, address, pmd)) {
> + up_read(&mm->mmap_sem);
> + goto out;
> + }
> + }
>
> + up_read(&mm->mmap_sem);
> + down_write(&mm->mmap_sem);
That's the critical point.
How do you guarantee that the vma will not be destroyed (or changed)
between up_read() and down_write()?
You need at least find_vma() again.
> anon_vma_lock_write(vma->anon_vma);
>
> pte = pte_offset_map(pmd, address);
> --
> 1.9.1
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
--
Kirill A. Shutemov
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH 0/3] mm, thp: remove duplication and fix locking issues in swapin Ebru Akagunduz <ebru.akagunduz@gmail.com> - 2016-05-23 19:20 +0200
[PATCH 2/3] mm, thp: fix possible circular locking dependency caused by sum_vm_event() Ebru Akagunduz <ebru.akagunduz@gmail.com> - 2016-05-23 19:20 +0200
[PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Ebru Akagunduz <ebru.akagunduz@gmail.com> - 2016-05-23 19:20 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem "Kirill A. Shutemov" <kirill@shutemov.name> - 2016-05-23 19:50 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Michal Hocko <mhocko@kernel.org> - 2016-05-23 20:50 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Rik van Riel <riel@redhat.com> - 2016-05-23 20:50 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> - 2016-05-23 21:10 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Rik van Riel <riel@redhat.com> - 2016-05-23 21:30 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem "Kirill A. Shutemov" <kirill@shutemov.name> - 2016-05-23 22:10 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Rik van Riel <riel@redhat.com> - 2016-05-23 22:20 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> - 2016-05-23 23:50 +0200
Re: [PATCH 3/3] mm, thp: make swapin readahead under down_read of mmap_sem Andrea Arcangeli <aarcange@redhat.com> - 2016-05-24 01:10 +0200
[PATCH 1/3] mm, thp: remove duplication of included header Ebru Akagunduz <ebru.akagunduz@gmail.com> - 2016-05-23 19:20 +0200
Re: [PATCH 0/3] mm, thp: remove duplication and fix locking issues in swapin Ebru Akagunduz <ebru.akagunduz@gmail.com> - 2016-05-23 19:30 +0200
Re: [PATCH 0/3] mm, thp: remove duplication and fix locking issues in swapin Michal Hocko <mhocko@kernel.org> - 2016-05-27 15:20 +0200
csiph-web