From: Vlastimil Babka <vbabka@suse.cz>
To: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
Andrew Morton <akpm@linux-foundation.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Hugh Dickins <hughd@google.com>
Cc: Dave Hansen <dave.hansen@intel.com>, Mel Gorman <mgorman@suse.de>,
Rik van Riel <riel@redhat.com>, Christoph Lameter <cl@gentwo.org>,
Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>,
Steve Capper <steve.capper@linaro.org>,
"Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>,
Johannes Weiner <hannes@cmpxchg.org>,
Michal Hocko <mhocko@suse.cz>,
Jerome Marchand <jmarchan@redhat.com>,
Sasha Levin <sasha.levin@oracle.com>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCHv6 26/36] mm: rework mapcount accounting to enable 4k mapping of THPs
Date: Wed, 10 Jun 2015 15:47:06 +0200 [thread overview]
Message-ID: <55783FDA.3080700@suse.cz> (raw)
In-Reply-To: <1433351167-125878-27-git-send-email-kirill.shutemov@linux.intel.com>
On 06/03/2015 07:05 PM, Kirill A. Shutemov wrote:
> We're going to allow mapping of individual 4k pages of THP compound.
> It means we need to track mapcount on per small page basis.
>
> Straight-forward approach is to use ->_mapcount in all subpages to track
> how many time this subpage is mapped with PMDs or PTEs combined. But
> this is rather expensive: mapping or unmapping of a THP page with PMD
> would require HPAGE_PMD_NR atomic operations instead of single we have
> now.
>
> The idea is to store separately how many times the page was mapped as
> whole -- compound_mapcount. This frees up ->_mapcount in subpages to
> track PTE mapcount.
>
> We use the same approach as with compound page destructor and compound
> order to store compound_mapcount: use space in first tail page,
> ->mapping this time.
>
> Any time we map/unmap whole compound page (THP or hugetlb) -- we
> increment/decrement compound_mapcount. When we map part of compound page
> with PTE we operate on ->_mapcount of the subpage.
>
> page_mapcount() counts both: PTE and PMD mappings of the page.
>
> Basically, we have mapcount for a subpage spread over two counters.
> It makes tricky to detect when last mapcount for a page goes away.
>
> We introduced PageDoubleMap() for this. When we split THP PMD for the
> first time and there's other PMD mapping left we offset up ->_mapcount
> in all subpages by one and set PG_double_map on the compound page.
> These additional references go away with last compound_mapcount.
>
> This approach provides a way to detect when last mapcount goes away on
> per small page basis without introducing new overhead for most common
> cases.
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> ---
> include/linux/mm.h | 26 +++++++++++-
> include/linux/mm_types.h | 1 +
> include/linux/page-flags.h | 37 +++++++++++++++++
> include/linux/rmap.h | 4 +-
> mm/debug.c | 5 ++-
> mm/huge_memory.c | 2 +-
> mm/hugetlb.c | 4 +-
> mm/memory.c | 2 +-
> mm/migrate.c | 2 +-
> mm/page_alloc.c | 14 +++++--
> mm/rmap.c | 98 +++++++++++++++++++++++++++++++++++-----------
> 11 files changed, 160 insertions(+), 35 deletions(-)
>
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 31cd5be081cf..22cd540104ec 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -403,6 +403,19 @@ static inline int is_vmalloc_or_module_addr(const void *x)
>
> extern void kvfree(const void *addr);
>
> +static inline atomic_t *compound_mapcount_ptr(struct page *page)
> +{
> + return &page[1].compound_mapcount;
> +}
> +
> +static inline int compound_mapcount(struct page *page)
> +{
> + if (!PageCompound(page))
> + return 0;
> + page = compound_head(page);
> + return atomic_read(compound_mapcount_ptr(page)) + 1;
> +}
> +
> /*
> * The atomic page->_mapcount, starts from -1: so that transitions
> * both from it and to it can be tracked, using atomic_inc_and_test
> @@ -415,8 +428,17 @@ static inline void page_mapcount_reset(struct page *page)
>
> static inline int page_mapcount(struct page *page)
> {
> + int ret;
> VM_BUG_ON_PAGE(PageSlab(page), page);
> - return atomic_read(&page->_mapcount) + 1;
> +
> + ret = atomic_read(&page->_mapcount) + 1;
> + if (PageCompound(page)) {
> + page = compound_head(page);
> + ret += compound_mapcount(page);
compound_mapcount() means another PageCompound() and compound_head(),
which you just did. I've tried this to see the effect on a function that
"calls" (inlines) page_mapcount() once:
- ret += compound_mapcount(page);
+ ret += atomic_read(compound_mapcount_ptr(page)) + 1;
bloat-o-meter on compaction.o:
add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-59 (-59)
function old new delta
isolate_migratepages_block 1769 1710 -59
> + if (PageDoubleMap(page))
> + ret--;
> + }
> + return ret;
> }
>
> static inline int page_count(struct page *page)
> @@ -898,7 +920,7 @@ static inline pgoff_t page_file_index(struct page *page)
> */
> static inline int page_mapped(struct page *page)
> {
> - return atomic_read(&(page)->_mapcount) >= 0;
> + return atomic_read(&(page)->_mapcount) + compound_mapcount(page) >= 0;
> }
>
> /*
> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index 4b51a59160ab..4d182cd14c1f 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -56,6 +56,7 @@ struct page {
> * see PAGE_MAPPING_ANON below.
> */
> void *s_mem; /* slab first object */
> + atomic_t compound_mapcount; /* first tail page */
> };
>
> /* Second double word */
> diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
> index 74b7cece1dfa..a8d47c1edf6a 100644
> --- a/include/linux/page-flags.h
> +++ b/include/linux/page-flags.h
> @@ -127,6 +127,9 @@ enum pageflags {
>
> /* SLOB */
> PG_slob_free = PG_private,
> +
> + /* THP. Stored in first tail page's flags */
> + PG_double_map = PG_private_2,
Well, not just THP. Any user of compound pages must make sure not to use
PG_private_2 on the first tail page. At least where the page is going to
be user-mapped. And same thing about fields that are in union with
compound_mapcount. Should that be documented more prominently somewhere?
I guess there's no such user so far, right?
> };
>
> #ifndef __GENERATING_BOUNDS_H
[...]
> @@ -1167,6 +1194,41 @@ out:
> mem_cgroup_end_page_stat(memcg);
> }
>
> +static void page_remove_anon_compound_rmap(struct page *page)
> +{
> + int i, nr;
> +
> + if (!atomic_add_negative(-1, compound_mapcount_ptr(page)))
> + return;
> +
> + /* Hugepages are not counted in NR_ANON_PAGES for now. */
> + if (unlikely(PageHuge(page)))
> + return;
> +
> + if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
> + return;
> +
> + __dec_zone_page_state(page, NR_ANON_TRANSPARENT_HUGEPAGES);
> +
> + if (PageDoubleMap(page)) {
> + nr = 0;
> + ClearPageDoubleMap(page);
> + /*
> + * Subpages can be mapped with PTEs too. Check how many of
> + * themi are still mapped.
> + */
> + for (i = 0; i < HPAGE_PMD_NR; i++) {
> + if (atomic_add_negative(-1, &page[i]._mapcount))
> + nr++;
> + }
> + } else {
> + nr = HPAGE_PMD_NR;
> + }
> +
> + if (nr)
> + __mod_zone_page_state(page_zone(page), NR_ANON_PAGES, nr);
-nr as we discussed on IRC.
> +}
> +
> /**
> * page_remove_rmap - take down pte mapping from a page
> * @page: page to remove mapping from
> @@ -1176,33 +1238,25 @@ out:
> */
> void page_remove_rmap(struct page *page, bool compound)
> {
> - int nr = compound ? hpage_nr_pages(page) : 1;
> -
> if (!PageAnon(page)) {
> VM_BUG_ON_PAGE(compound && !PageHuge(page), page);
> page_remove_file_rmap(page);
> return;
> }
>
> + if (compound)
> + return page_remove_anon_compound_rmap(page);
> +
> /* page still mapped by someone else? */
> if (!atomic_add_negative(-1, &page->_mapcount))
> return;
>
> - /* Hugepages are not counted in NR_ANON_PAGES for now. */
> - if (unlikely(PageHuge(page)))
> - return;
> -
> /*
> * We use the irq-unsafe __{inc|mod}_zone_page_stat because
> * these counters are not modified in interrupt context, and
> * pte lock(a spinlock) is held, which implies preemption disabled.
> */
> - if (compound) {
> - VM_BUG_ON_PAGE(!PageTransHuge(page), page);
> - __dec_zone_page_state(page, NR_ANON_TRANSPARENT_HUGEPAGES);
> - }
> -
> - __mod_zone_page_state(page_zone(page), NR_ANON_PAGES, -nr);
> + __dec_zone_page_state(page, NR_ANON_PAGES);
>
> if (unlikely(PageMlocked(page)))
> clear_page_mlock(page);
> @@ -1643,7 +1697,7 @@ void hugepage_add_anon_rmap(struct page *page,
> BUG_ON(!PageLocked(page));
> BUG_ON(!anon_vma);
> /* address might be in next vma when migration races vma_adjust */
> - first = atomic_inc_and_test(&page->_mapcount);
> + first = atomic_inc_and_test(compound_mapcount_ptr(page));
> if (first)
> __hugepage_set_anon_rmap(page, vma, address, 0);
> }
> @@ -1652,7 +1706,7 @@ void hugepage_add_new_anon_rmap(struct page *page,
> struct vm_area_struct *vma, unsigned long address)
> {
> BUG_ON(address < vma->vm_start || address >= vma->vm_end);
> - atomic_set(&page->_mapcount, 0);
> + atomic_set(compound_mapcount_ptr(page), 0);
> __hugepage_set_anon_rmap(page, vma, address, 1);
> }
> #endif /* CONFIG_HUGETLB_PAGE */
>
next prev parent reply other threads:[~2015-06-10 13:47 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-03 17:05 [PATCHv6 00/36] THP refcounting redesign Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 01/36] mm, proc: adjust PSS calculation Kirill A. Shutemov
2015-06-09 12:29 ` Vlastimil Babka
2015-06-22 10:02 ` Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 02/36] rmap: add argument to charge compound page Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 03/36] memcg: adjust to support new THP refcounting Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 04/36] mm, thp: adjust conditions when we can reuse the page on WP fault Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 05/36] mm: adjust FOLL_SPLIT for new refcounting Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 06/36] mm: handle PTE-mapped tail pages in gerneric fast gup implementaiton Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 07/36] thp, mlock: do not allow huge pages in mlocked area Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 08/36] khugepaged: ignore pmd tables with THP mapped with ptes Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 09/36] thp: rename split_huge_page_pmd() to split_huge_pmd() Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 10/36] mm, vmstats: new THP splitting event Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 11/36] mm: temporally mark THP broken Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 12/36] thp: drop all split_huge_page()-related code Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 13/36] mm: drop tail page refcounting Kirill A. Shutemov
2015-06-09 13:59 ` Vlastimil Babka
2015-06-03 17:05 ` [PATCHv6 14/36] futex, thp: remove special case for THP in get_futex_key Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 15/36] ksm: prepare to new THP semantics Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 16/36] mm, thp: remove compound_lock Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 17/36] arm64, thp: remove infrastructure for handling splitting PMDs Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 18/36] arm, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 19/36] mips, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 20/36] powerpc, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 21/36] s390, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 22/36] sparc, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 23/36] tile, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 24/36] x86, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 25/36] mm, " Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 26/36] mm: rework mapcount accounting to enable 4k mapping of THPs Kirill A. Shutemov
2015-06-10 13:47 ` Vlastimil Babka [this message]
2015-06-22 10:22 ` Kirill A. Shutemov
2015-06-03 17:05 ` [PATCHv6 27/36] mm: differentiate page_mapped() from page_mapcount() for compound pages Kirill A. Shutemov
2015-06-09 10:58 ` Kirill A. Shutemov
2015-06-10 14:34 ` Vlastimil Babka
2015-06-03 17:05 ` [PATCHv6 28/36] mm, numa: skip PTE-mapped THP on numa fault Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 29/36] thp: implement split_huge_pmd() Kirill A. Shutemov
2015-06-11 9:49 ` Vlastimil Babka
2015-06-22 11:14 ` Kirill A. Shutemov
2015-06-22 16:01 ` Vlastimil Babka
2015-06-03 17:06 ` [PATCHv6 30/36] thp: add option to setup migration entiries during PMD split Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 31/36] thp, mm: split_huge_page(): caller need to lock page Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 32/36] thp: reintroduce split_huge_page() Kirill A. Shutemov
2015-06-10 15:44 ` Vlastimil Babka
2015-06-22 11:28 ` Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 33/36] migrate_pages: try to split pages on qeueuing Kirill A. Shutemov
2015-06-11 9:27 ` Vlastimil Babka
2015-06-22 11:35 ` Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 34/36] thp: introduce deferred_split_huge_page() Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 35/36] mm: re-enable THP Kirill A. Shutemov
2015-06-03 17:06 ` [PATCHv6 36/36] thp: update documentation Kirill A. Shutemov
2015-06-11 12:30 ` Vlastimil Babka
2015-06-22 13:18 ` Kirill A. Shutemov
2015-06-22 16:07 ` Vlastimil Babka
2015-06-16 13:17 ` [PATCHv6 00/36] THP refcounting redesign Jerome Marchand
2015-06-22 13:21 ` Kirill A. Shutemov
2015-06-22 13:32 ` Jerome Marchand
2015-06-22 13:39 ` Kirill A. Shutemov
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=55783FDA.3080700@suse.cz \
--to=vbabka@suse.cz \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=aneesh.kumar@linux.vnet.ibm.com \
--cc=cl@gentwo.org \
--cc=dave.hansen@intel.com \
--cc=hannes@cmpxchg.org \
--cc=hughd@google.com \
--cc=jmarchan@redhat.com \
--cc=kirill.shutemov@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mgorman@suse.de \
--cc=mhocko@suse.cz \
--cc=n-horiguchi@ah.jp.nec.com \
--cc=riel@redhat.com \
--cc=sasha.levin@oracle.com \
--cc=steve.capper@linaro.org \
/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).