[pve-devel] [PATCH kernel 4.13] fix #1633: potential deadlock with THPs

[Fabian Grünbichler]

see https://marc.info/?l=linux-mm&m=151683828707588

Signed-off-by: Fabian Grünbichler <f.gruenbichler at proxmox.com>
---
 ...ot-wait-for-lock_page-in-deferred_split_s.patch | 46 ++++++++++++++++++++++
 1 file changed, 46 insertions(+)
 create mode 100644 patches/kernel/0022-mm-thp-Do-not-wait-for-lock_page-in-deferred_split_s.patch

diff --git a/patches/kernel/0022-mm-thp-Do-not-wait-for-lock_page-in-deferred_split_s.patch b/patches/kernel/0022-mm-thp-Do-not-wait-for-lock_page-in-deferred_split_s.patch
new file mode 100644
index 0000000..239aa42
--- /dev/null
+++ b/patches/kernel/0022-mm-thp-Do-not-wait-for-lock_page-in-deferred_split_s.patch
@@ -0,0 +1,46 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: "Kirill A. Shutemov" <kirill.shutemov at linux.intel.com>
+Date: Thu, 15 Mar 2018 18:07:47 +0300
+Subject: [PATCH] mm/thp: Do not wait for lock_page() in deferred_split_scan()
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+deferred_split_scan() gets called from reclaim path. Waiting for page
+lock may lead to deadlock there.
+
+Replace lock_page() with trylock_page() and skip the page if we failed
+to lock it. We will get to the page on the next scan.
+
+Fixes: 9a982250f773 ("thp: introduce deferred_split_huge_page()")
+
+Signed-off-by: Kirill A. Shutemov <kirill.shutemov at linux.intel.com>
+Acked-by: Michal Hocko <mhocko at suse.com>
+(cherry-picked from https://patchwork.kernel.org/patch/10284703/)
+Signed-off-by: Fabian Grünbichler <f.gruenbichler at proxmox.com>
+---
+ mm/huge_memory.c | 4 +++-
+ 1 file changed, 3 insertions(+), 1 deletion(-)
+
+diff --git a/mm/huge_memory.c b/mm/huge_memory.c
+index 8b887db33383..5c4093e0be8d 100644
+--- a/mm/huge_memory.c
++++ b/mm/huge_memory.c
+@@ -2621,11 +2621,13 @@ static unsigned long deferred_split_scan(struct shrinker *shrink,
+ 
+ 	list_for_each_safe(pos, next, &list) {
+ 		page = list_entry((void *)pos, struct page, mapping);
+-		lock_page(page);
++		if (!trylock_page(page))
++			goto next;
+ 		/* split_huge_page() removes page from list on success */
+ 		if (!split_huge_page(page))
+ 			split++;
+ 		unlock_page(page);
++next:
+ 		put_page(page);
+ 	}
+ 
+-- 
+2.14.2
+
-- 
2.14.2