linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: David Hildenbrand <david@redhat.com>
To: Vlastimil Babka <vbabka@suse.cz>, linux-kernel@vger.kernel.org
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Hugh Dickins <hughd@google.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	David Rientjes <rientjes@google.com>,
	Shakeel Butt <shakeelb@google.com>,
	John Hubbard <jhubbard@nvidia.com>,
	Jason Gunthorpe <jgg@nvidia.com>,
	Mike Kravetz <mike.kravetz@oracle.com>,
	Mike Rapoport <rppt@linux.ibm.com>,
	Yang Shi <shy828301@gmail.com>,
	"Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>,
	Matthew Wilcox <willy@infradead.org>,
	Jann Horn <jannh@google.com>, Michal Hocko <mhocko@kernel.org>,
	Nadav Amit <namit@vmware.com>, Rik van Riel <riel@surriel.com>,
	Roman Gushchin <guro@fb.com>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Peter Xu <peterx@redhat.com>, Donald Dutile <ddutile@redhat.com>,
	Christoph Hellwig <hch@lst.de>, Oleg Nesterov <oleg@redhat.com>,
	Jan Kara <jack@suse.cz>, Liang Zhang <zhangliang5@huawei.com>,
	Pedro Gomes <pedrodemargomes@gmail.com>,
	Oded Gabbay <oded.gabbay@gmail.com>,
	linux-mm@kvack.org
Subject: Re: [PATCH v3 12/16] mm: remember exclusively mapped anonymous pages with PG_anon_exclusive
Date: Tue, 19 Apr 2022 18:46:43 +0200	[thread overview]
Message-ID: <219bd2d0-92ef-bcac-458a-0df6190fa387@redhat.com> (raw)
In-Reply-To: <5fc7d007-e59b-de8d-4d88-3f1b5adfa95b@suse.cz>

On 13.04.22 20:28, Vlastimil Babka wrote:
> On 4/13/22 18:39, David Hildenbrand wrote:
>>>> @@ -3035,10 +3083,19 @@ void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
>>>>  
>>>>  	flush_cache_range(vma, address, address + HPAGE_PMD_SIZE);
>>>>  	pmdval = pmdp_invalidate(vma, address, pvmw->pmd);
>>>> +
>>>> +	anon_exclusive = PageAnon(page) && PageAnonExclusive(page);
>>>> +	if (anon_exclusive && page_try_share_anon_rmap(page)) {
>>>> +		set_pmd_at(mm, address, pvmw->pmd, pmdval);
>>>> +		return;
>>>
>>> I am admittedly not too familiar with this code, but looks like this means
>>> we fail to migrate the THP, right? But we don't seem to be telling the
>>> caller, which is try_to_migrate_one(), so it will continue and not terminate
>>> the walk and return false?
>>
>> Right, we're not returning "false". Returning "false" would be an
>> optimization to make rmap_walk_anon() fail faster.
> 
> Ah right, that's what I missed, it's an optimization and we will realize
> elsewhere afterwards that the page has still mappings and we can't migrate...

I'll include that patch in v4 (to be tested):


From 08fb0e45404e3d0f85c2ad23a473e95053396376 Mon Sep 17 00:00:00 2001
From: David Hildenbrand <david@redhat.com>
Date: Tue, 19 Apr 2022 18:39:23 +0200
Subject: [PATCH] mm/rmap: fail try_to_migrate() early when setting a PMD
 migration entry fails

Let's fail right away in case we cannot clear PG_anon_exclusive because
the anon THP may be pinned. Right now, we continue trying to
install migration entries and the caller of try_to_migrate() will
realize that the page is still mapped and has to restore the migration
entries. Let's just fail fast just like for PTE migration entries.

Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/swapops.h | 4 ++--
 mm/huge_memory.c        | 8 +++++---
 mm/rmap.c               | 6 +++++-
 3 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/include/linux/swapops.h b/include/linux/swapops.h
index 06280fc1c99b..8b6e4cd1fab8 100644
--- a/include/linux/swapops.h
+++ b/include/linux/swapops.h
@@ -299,7 +299,7 @@ static inline bool is_pfn_swap_entry(swp_entry_t entry)
 struct page_vma_mapped_walk;
 
 #ifdef CONFIG_ARCH_ENABLE_THP_MIGRATION
-extern void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
+extern int set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 		struct page *page);
 
 extern void remove_migration_pmd(struct page_vma_mapped_walk *pvmw,
@@ -332,7 +332,7 @@ static inline int is_pmd_migration_entry(pmd_t pmd)
 	return !pmd_present(pmd) && is_migration_entry(pmd_to_swp_entry(pmd));
 }
 #else
-static inline void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
+static inline int set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 		struct page *page)
 {
 	BUILD_BUG();
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index c7ac1b462543..390f22334ee9 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -3080,7 +3080,7 @@ late_initcall(split_huge_pages_debugfs);
 #endif
 
 #ifdef CONFIG_ARCH_ENABLE_THP_MIGRATION
-void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
+int set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 		struct page *page)
 {
 	struct vm_area_struct *vma = pvmw->vma;
@@ -3092,7 +3092,7 @@ void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 	pmd_t pmdswp;
 
 	if (!(pvmw->pmd && !pvmw->pte))
-		return;
+		return 0;
 
 	flush_cache_range(vma, address, address + HPAGE_PMD_SIZE);
 	pmdval = pmdp_invalidate(vma, address, pvmw->pmd);
@@ -3100,7 +3100,7 @@ void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 	anon_exclusive = PageAnon(page) && PageAnonExclusive(page);
 	if (anon_exclusive && page_try_share_anon_rmap(page)) {
 		set_pmd_at(mm, address, pvmw->pmd, pmdval);
-		return;
+		return -EBUSY;
 	}
 
 	if (pmd_dirty(pmdval))
@@ -3118,6 +3118,8 @@ void set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
 	page_remove_rmap(page, vma, true);
 	put_page(page);
 	trace_set_migration_pmd(address, pmd_val(pmdswp));
+
+	return 0;
 }
 
 void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
diff --git a/mm/rmap.c b/mm/rmap.c
index 00418faaf4ce..68c2f61bf212 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -1814,7 +1814,11 @@ static bool try_to_migrate_one(struct folio *folio, struct vm_area_struct *vma,
 			VM_BUG_ON_FOLIO(folio_test_hugetlb(folio) ||
 					!folio_test_pmd_mappable(folio), folio);
 
-			set_pmd_migration_entry(&pvmw, subpage);
+			if (set_pmd_migration_entry(&pvmw, subpage)) {
+				ret = false;
+				page_vma_mapped_walk_done(&pvmw);
+				break;
+			}
 			continue;
 		}
 #endif
-- 
2.35.1



-- 
Thanks,

David / dhildenb



  reply	other threads:[~2022-04-19 16:46 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-29 16:04 [PATCH v3 00/16] mm: COW fixes part 2: reliable GUP pins of anonymous pages David Hildenbrand
2022-03-29 16:04 ` [PATCH v3 01/16] mm/rmap: fix missing swap_free() in try_to_unmap() after arch_unmap_one() failed David Hildenbrand
2022-04-11 16:04   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 02/16] mm/hugetlb: take src_mm->write_protect_seq in copy_hugetlb_page_range() David Hildenbrand
2022-04-11 16:15   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 03/16] mm/memory: slightly simplify copy_present_pte() David Hildenbrand
2022-04-11 16:38   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 04/16] mm/rmap: split page_dup_rmap() into page_dup_file_rmap() and page_try_dup_anon_rmap() David Hildenbrand
2022-04-11 18:18   ` Vlastimil Babka
2022-04-12  8:06     ` David Hildenbrand
2022-03-29 16:04 ` [PATCH v3 05/16] mm/rmap: convert RMAP flags to a proper distinct rmap_t type David Hildenbrand
2022-04-12  8:11   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 06/16] mm/rmap: remove do_page_add_anon_rmap() David Hildenbrand
2022-04-12  8:13   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 07/16] mm/rmap: pass rmap flags to hugepage_add_anon_rmap() David Hildenbrand
2022-04-12  8:37   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 08/16] mm/rmap: drop "compound" parameter from page_add_new_anon_rmap() David Hildenbrand
2022-04-12  8:47   ` Vlastimil Babka
2022-04-12  9:37     ` David Hildenbrand
2022-04-13 12:26       ` Matthew Wilcox
2022-04-13 12:28         ` David Hildenbrand
2022-04-13 12:48           ` Matthew Wilcox
2022-04-13 16:20             ` David Hildenbrand
2022-03-29 16:04 ` [PATCH v3 09/16] mm/rmap: use page_move_anon_rmap() when reusing a mapped PageAnon() page exclusively David Hildenbrand
2022-04-12  9:26   ` Vlastimil Babka
2022-04-12  9:28     ` David Hildenbrand
2022-03-29 16:04 ` [PATCH v3 10/16] mm/huge_memory: remove outdated VM_WARN_ON_ONCE_PAGE from unmap_page() David Hildenbrand
2022-04-12  9:37   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 11/16] mm/page-flags: reuse PG_mappedtodisk as PG_anon_exclusive for PageAnon() pages David Hildenbrand
2022-04-13  8:25   ` Vlastimil Babka
2022-04-13 10:28     ` David Hildenbrand
2022-04-13 14:55       ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 12/16] mm: remember exclusively mapped anonymous pages with PG_anon_exclusive David Hildenbrand
2022-04-13 16:28   ` Vlastimil Babka
2022-04-13 16:39     ` David Hildenbrand
2022-04-13 18:28       ` Vlastimil Babka
2022-04-19 16:46         ` David Hildenbrand [this message]
2022-04-13 18:29   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 13/16] mm/gup: disallow follow_page(FOLL_PIN) David Hildenbrand
2022-04-14 15:18   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 14/16] mm: support GUP-triggered unsharing of anonymous pages David Hildenbrand
2022-04-14 17:15   ` Vlastimil Babka
2022-04-19 16:29     ` David Hildenbrand
2022-04-19 16:31       ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 15/16] mm/gup: trigger FAULT_FLAG_UNSHARE when R/O-pinning a possibly shared anonymous page David Hildenbrand
2022-04-19 15:56   ` Vlastimil Babka
2022-03-29 16:04 ` [PATCH v3 16/16] mm/gup: sanity-check with CONFIG_DEBUG_VM that anonymous pages are exclusive when (un)pinning David Hildenbrand
2022-04-19 17:40   ` Vlastimil Babka
2022-04-21  9:15     ` David Hildenbrand
2022-04-22  6:54       ` Vlastimil Babka
2022-03-29 16:09 ` [PATCH v3 00/16] mm: COW fixes part 2: reliable GUP pins of anonymous pages David Hildenbrand

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=219bd2d0-92ef-bcac-458a-0df6190fa387@redhat.com \
    --to=david@redhat.com \
    --cc=aarcange@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=ddutile@redhat.com \
    --cc=guro@fb.com \
    --cc=hch@lst.de \
    --cc=hughd@google.com \
    --cc=jack@suse.cz \
    --cc=jannh@google.com \
    --cc=jgg@nvidia.com \
    --cc=jhubbard@nvidia.com \
    --cc=kirill.shutemov@linux.intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mhocko@kernel.org \
    --cc=mike.kravetz@oracle.com \
    --cc=namit@vmware.com \
    --cc=oded.gabbay@gmail.com \
    --cc=oleg@redhat.com \
    --cc=pedrodemargomes@gmail.com \
    --cc=peterx@redhat.com \
    --cc=riel@surriel.com \
    --cc=rientjes@google.com \
    --cc=rppt@linux.ibm.com \
    --cc=shakeelb@google.com \
    --cc=shy828301@gmail.com \
    --cc=torvalds@linux-foundation.org \
    --cc=vbabka@suse.cz \
    --cc=willy@infradead.org \
    --cc=zhangliang5@huawei.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).