From: Yang Shi <yang.shi@linux.alibaba.com>
To: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
akpm@linux-foundation.org, Andrea Arcangeli <aarcange@redhat.com>
Cc: Zi Yan <ziy@nvidia.com>, Ralph Campbell <rcampbell@nvidia.com>,
John Hubbard <jhubbard@nvidia.com>,
William Kucharski <william.kucharski@oracle.com>,
linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCHv3, RESEND 2/8] khugepaged: Do not stop collapse if less than half PTEs are referenced
Date: Wed, 15 Apr 2020 13:31:33 -0700 [thread overview]
Message-ID: <902cad73-c3ef-c274-7483-c948167639e9@linux.alibaba.com> (raw)
In-Reply-To: <20200413125220.663-3-kirill.shutemov@linux.intel.com>
On 4/13/20 5:52 AM, Kirill A. Shutemov wrote:
> __collapse_huge_page_swapin() checks the number of referenced PTE to
> decide if the memory range is hot enough to justify swapin.
>
> We have few problems with the approach:
>
> - It is way too late: we can do the check much earlier and safe time.
> khugepaged_scan_pmd() already knows if we have any pages to swap in
> and number of referenced page.
>
> - It stops collapse altogether if there's not enough referenced pages,
> not only swappingin.
>
> Fix it by making the right check early. We also can avoid additional
> page table scanning if khugepaged_scan_pmd() haven't found any swap
> entries.
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> Fixes: 0db501f7a34c ("mm, thp: convert from optimistic swapin collapsing to conservative")
> ---
> mm/khugepaged.c | 25 ++++++++++---------------
> 1 file changed, 10 insertions(+), 15 deletions(-)
Acked-by: Yang Shi <yang.shi@linux.alibaba.com>
Just a nit below.
>
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 99bab7e4d05b..5968ec5ddd6b 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -902,11 +902,6 @@ static bool __collapse_huge_page_swapin(struct mm_struct *mm,
> .pgoff = linear_page_index(vma, address),
> };
>
> - /* we only decide to swapin, if there is enough young ptes */
> - if (referenced < HPAGE_PMD_NR/2) {
> - trace_mm_collapse_huge_page_swapin(mm, swapped_in, referenced, 0);
> - return false;
> - }
> vmf.pte = pte_offset_map(pmd, address);
> for (; vmf.address < address + HPAGE_PMD_NR*PAGE_SIZE;
> vmf.pte++, vmf.address += PAGE_SIZE) {
> @@ -946,7 +941,7 @@ static bool __collapse_huge_page_swapin(struct mm_struct *mm,
> static void collapse_huge_page(struct mm_struct *mm,
> unsigned long address,
> struct page **hpage,
> - int node, int referenced)
> + int node, int referenced, int unmapped)
> {
> pmd_t *pmd, _pmd;
> pte_t *pte;
> @@ -1003,7 +998,8 @@ static void collapse_huge_page(struct mm_struct *mm,
> * If it fails, we release mmap_sem and jump out_nolock.
> * Continuing to collapse causes inconsistency.
> */
> - if (!__collapse_huge_page_swapin(mm, vma, address, pmd, referenced)) {
> + if (unmapped && !__collapse_huge_page_swapin(mm, vma, address,
> + pmd, referenced)) {
> mem_cgroup_cancel_charge(new_page, memcg, true);
> up_read(&mm->mmap_sem);
> goto out_nolock;
> @@ -1214,22 +1210,21 @@ static int khugepaged_scan_pmd(struct mm_struct *mm,
> mmu_notifier_test_young(vma->vm_mm, address))
> referenced++;
> }
> - if (writable) {
> - if (referenced) {
> + if (!writable) {
> + result = SCAN_PAGE_RO;
> + } else if (!referenced || (unmapped && referenced < HPAGE_PMD_NR/2)) {
> + result = SCAN_LACK_REFERENCED_PAGE;
> + } else {
> result = SCAN_SUCCEED;
> ret = 1;
Shall fix the indentation for the above two statements?
> - } else {
> - result = SCAN_LACK_REFERENCED_PAGE;
> - }
> - } else {
> - result = SCAN_PAGE_RO;
> }
> out_unmap:
> pte_unmap_unlock(pte, ptl);
> if (ret) {
> node = khugepaged_find_target_node();
> /* collapse_huge_page will return with the mmap_sem released */
> - collapse_huge_page(mm, address, hpage, node, referenced);
> + collapse_huge_page(mm, address, hpage, node,
> + referenced, unmapped);
> }
> out:
> trace_mm_khugepaged_scan_pmd(mm, page, writable, referenced,
next prev parent reply other threads:[~2020-04-15 20:31 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-04-13 12:52 [PATCHv3, RESEND 0/8] thp/khugepaged improvements and CoW semantics Kirill A. Shutemov
2020-04-13 12:52 ` [PATCHv3, RESEND 1/8] khugepaged: Add self test Kirill A. Shutemov
2020-04-15 15:39 ` Zi Yan
2020-04-15 20:11 ` Yang Shi
2020-04-13 12:52 ` [PATCHv3, RESEND 2/8] khugepaged: Do not stop collapse if less than half PTEs are referenced Kirill A. Shutemov
2020-04-15 20:31 ` Yang Shi [this message]
2020-04-13 12:52 ` [PATCHv3, RESEND 3/8] khugepaged: Drain all LRU caches before scanning pages Kirill A. Shutemov
2020-04-15 20:31 ` Yang Shi
2020-04-13 12:52 ` [PATCHv3, RESEND 4/8] khugepaged: Drain LRU add pagevec after swapin Kirill A. Shutemov
2020-04-15 20:32 ` Yang Shi
2020-04-13 12:52 ` [PATCHv3, RESEND 5/8] khugepaged: Allow to collapse a page shared across fork Kirill A. Shutemov
2020-04-13 20:48 ` John Hubbard
2020-04-14 21:35 ` Kirill A. Shutemov
2020-04-15 2:42 ` John Hubbard
2020-04-15 2:43 ` John Hubbard
2020-04-15 20:39 ` Yang Shi
2020-04-13 12:52 ` [PATCHv3, RESEND 6/8] khugepaged: Allow to collapse PTE-mapped compound pages Kirill A. Shutemov
2020-04-15 20:46 ` Yang Shi
2020-04-15 21:44 ` Andrew Morton
2020-04-15 21:52 ` Kirill A. Shutemov
2020-04-15 22:52 ` Andrew Morton
2020-04-15 22:59 ` Kirill A. Shutemov
2020-04-13 12:52 ` [PATCHv3, RESEND 7/8] thp: Change CoW semantics for anon-THP Kirill A. Shutemov
2020-04-15 21:19 ` Yang Shi
2020-04-13 12:52 ` [PATCHv3, RESEND 8/8] khugepaged: Introduce 'max_ptes_shared' tunable Kirill A. Shutemov
2020-04-15 21:25 ` Yang Shi
2020-04-16 15:52 ` Kirill A. Shutemov
2020-04-16 18:25 ` Yang Shi
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=902cad73-c3ef-c274-7483-c948167639e9@linux.alibaba.com \
--to=yang.shi@linux.alibaba.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=jhubbard@nvidia.com \
--cc=kirill.shutemov@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=rcampbell@nvidia.com \
--cc=william.kucharski@oracle.com \
--cc=ziy@nvidia.com \
/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;
as well as URLs for NNTP newsgroup(s).