linux/Documentation/mm/transhuge.rst
David Hildenbrand 05c5323b2a mm: track mapcount of large folios in single value
Let's track the mapcount of large folios in a single value.  The mapcount
of a large folio currently corresponds to the sum of the entire mapcount
and all page mapcounts.

This sum is what we actually want to know in folio_mapcount() and it is
also sufficient for implementing folio_mapped().

With PTE-mapped THP becoming more important and more widely used, we want
to avoid looping over all pages of a folio just to obtain the mapcount of
large folios.  The comment "In the common case, avoid the loop when no
pages mapped by PTE" in folio_total_mapcount() does no longer hold for
mTHP that are always mapped by PTE.

Further, we are planning on using folio_mapcount() more frequently, and
might even want to remove page mapcounts for large folios in some kernel
configs.  Therefore, allow for reading the mapcount of large folios
efficiently and atomically without looping over any pages.

Maintain the mapcount also for hugetlb pages for simplicity.  Use the new
mapcount to implement folio_mapcount() and folio_mapped().  Make
page_mapped() simply call folio_mapped().  We can now get rid of
folio_large_is_mapped().

_nr_pages_mapped is now only used in rmap code and for debugging purposes.
Keep folio_nr_pages_mapped() around, but document that its use should be
limited to rmap internals and debugging purposes.

This change implies one additional atomic add/sub whenever
mapping/unmapping (parts of) a large folio.

As we now batch RMAP operations for PTE-mapped THP during fork(), during
unmap/zap, and when PTE-remapping a PMD-mapped THP, and we adjust the
large mapcount for a PTE batch only once, the added overhead in the common
case is small.  Only when unmapping individual pages of a large folio
(e.g., during COW), the overhead might be bigger in comparison, but it's
essentially one additional atomic operation.

Note that before the new mapcount would overflow, already our refcount
would overflow: each mapping requires a folio reference.  Extend the
focumentation of folio_mapcount().

Link: https://lkml.kernel.org/r/20240409192301.907377-5-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Cc: Chris Zankel <chris@zankel.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Richard Chang <richardycc@google.com>
Cc: Rich Felker <dalias@libc.org>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-05-05 17:53:28 -07:00

170 lines
7.5 KiB
ReStructuredText

============================
Transparent Hugepage Support
============================
This document describes design principles for Transparent Hugepage (THP)
support and its interaction with other parts of the memory management
system.
Design principles
=================
- "graceful fallback": mm components which don't have transparent hugepage
knowledge fall back to breaking huge pmd mapping into table of ptes and,
if necessary, split a transparent hugepage. Therefore these components
can continue working on the regular pages or regular pte mappings.
- if a hugepage allocation fails because of memory fragmentation,
regular pages should be gracefully allocated instead and mixed in
the same vma without any failure or significant delay and without
userland noticing
- if some task quits and more hugepages become available (either
immediately in the buddy or through the VM), guest physical memory
backed by regular pages should be relocated on hugepages
automatically (with khugepaged)
- it doesn't require memory reservation and in turn it uses hugepages
whenever possible (the only possible reservation here is kernelcore=
to avoid unmovable pages to fragment all the memory but such a tweak
is not specific to transparent hugepage support and it's a generic
feature that applies to all dynamic high order allocations in the
kernel)
get_user_pages and follow_page
==============================
get_user_pages and follow_page if run on a hugepage, will return the
head or tail pages as usual (exactly as they would do on
hugetlbfs). Most GUP users will only care about the actual physical
address of the page and its temporary pinning to release after the I/O
is complete, so they won't ever notice the fact the page is huge. But
if any driver is going to mangle over the page structure of the tail
page (like for checking page->mapping or other bits that are relevant
for the head page and not the tail page), it should be updated to jump
to check head page instead. Taking a reference on any head/tail page would
prevent the page from being split by anyone.
.. note::
these aren't new constraints to the GUP API, and they match the
same constraints that apply to hugetlbfs too, so any driver capable
of handling GUP on hugetlbfs will also work fine on transparent
hugepage backed mappings.
Graceful fallback
=================
Code walking pagetables but unaware about huge pmds can simply call
split_huge_pmd(vma, pmd, addr) where the pmd is the one returned by
pmd_offset. It's trivial to make the code transparent hugepage aware
by just grepping for "pmd_offset" and adding split_huge_pmd where
missing after pmd_offset returns the pmd. Thanks to the graceful
fallback design, with a one liner change, you can avoid to write
hundreds if not thousands of lines of complex code to make your code
hugepage aware.
If you're not walking pagetables but you run into a physical hugepage
that you can't handle natively in your code, you can split it by
calling split_huge_page(page). This is what the Linux VM does before
it tries to swapout the hugepage for example. split_huge_page() can fail
if the page is pinned and you must handle this correctly.
Example to make mremap.c transparent hugepage aware with a one liner
change::
diff --git a/mm/mremap.c b/mm/mremap.c
--- a/mm/mremap.c
+++ b/mm/mremap.c
@@ -41,6 +41,7 @@ static pmd_t *get_old_pmd(struct mm_stru
return NULL;
pmd = pmd_offset(pud, addr);
+ split_huge_pmd(vma, pmd, addr);
if (pmd_none_or_clear_bad(pmd))
return NULL;
Locking in hugepage aware code
==============================
We want as much code as possible hugepage aware, as calling
split_huge_page() or split_huge_pmd() has a cost.
To make pagetable walks huge pmd aware, all you need to do is to call
pmd_trans_huge() on the pmd returned by pmd_offset. You must hold the
mmap_lock in read (or write) mode to be sure a huge pmd cannot be
created from under you by khugepaged (khugepaged collapse_huge_page
takes the mmap_lock in write mode in addition to the anon_vma lock). If
pmd_trans_huge returns false, you just fallback in the old code
paths. If instead pmd_trans_huge returns true, you have to take the
page table lock (pmd_lock()) and re-run pmd_trans_huge. Taking the
page table lock will prevent the huge pmd being converted into a
regular pmd from under you (split_huge_pmd can run in parallel to the
pagetable walk). If the second pmd_trans_huge returns false, you
should just drop the page table lock and fallback to the old code as
before. Otherwise, you can proceed to process the huge pmd and the
hugepage natively. Once finished, you can drop the page table lock.
Refcounts and transparent huge pages
====================================
Refcounting on THP is mostly consistent with refcounting on other compound
pages:
- get_page()/put_page() and GUP operate on the folio->_refcount.
- ->_refcount in tail pages is always zero: get_page_unless_zero() never
succeeds on tail pages.
- map/unmap of a PMD entry for the whole THP increment/decrement
folio->_entire_mapcount, increment/decrement folio->_large_mapcount
and also increment/decrement folio->_nr_pages_mapped by ENTIRELY_MAPPED
when _entire_mapcount goes from -1 to 0 or 0 to -1.
- map/unmap of individual pages with PTE entry increment/decrement
page->_mapcount, increment/decrement folio->_large_mapcount and also
increment/decrement folio->_nr_pages_mapped when page->_mapcount goes
from -1 to 0 or 0 to -1 as this counts the number of pages mapped by PTE.
split_huge_page internally has to distribute the refcounts in the head
page to the tail pages before clearing all PG_head/tail bits from the page
structures. It can be done easily for refcounts taken by page table
entries, but we don't have enough information on how to distribute any
additional pins (i.e. from get_user_pages). split_huge_page() fails any
requests to split pinned huge pages: it expects page count to be equal to
the sum of mapcount of all sub-pages plus one (split_huge_page caller must
have a reference to the head page).
split_huge_page uses migration entries to stabilize page->_refcount and
page->_mapcount of anonymous pages. File pages just get unmapped.
We are safe against physical memory scanners too: the only legitimate way
a scanner can get a reference to a page is get_page_unless_zero().
All tail pages have zero ->_refcount until atomic_add(). This prevents the
scanner from getting a reference to the tail page up to that point. After the
atomic_add() we don't care about the ->_refcount value. We already know how
many references should be uncharged from the head page.
For head page get_page_unless_zero() will succeed and we don't mind. It's
clear where references should go after split: it will stay on the head page.
Note that split_huge_pmd() doesn't have any limitations on refcounting:
pmd can be split at any point and never fails.
Partial unmap and deferred_split_folio()
========================================
Unmapping part of THP (with munmap() or other way) is not going to free
memory immediately. Instead, we detect that a subpage of THP is not in use
in folio_remove_rmap_*() and queue the THP for splitting if memory pressure
comes. Splitting will free up unused subpages.
Splitting the page right away is not an option due to locking context in
the place where we can detect partial unmap. It also might be
counterproductive since in many cases partial unmap happens during exit(2) if
a THP crosses a VMA boundary.
The function deferred_split_folio() is used to queue a folio for splitting.
The splitting itself will happen when we get memory pressure via shrinker
interface.