]> git.ipfire.org Git - thirdparty/kernel/stable-queue.git/commitdiff
delete queue-3.14/mm-fix-mlock-accouting.patch
authorGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Feb 2016 19:57:28 +0000 (11:57 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Feb 2016 19:57:28 +0000 (11:57 -0800)
queue-3.14/mm-fix-mlock-accouting.patch [deleted file]
queue-3.14/series

diff --git a/queue-3.14/mm-fix-mlock-accouting.patch b/queue-3.14/mm-fix-mlock-accouting.patch
deleted file mode 100644 (file)
index b9e0cee..0000000
+++ /dev/null
@@ -1,77 +0,0 @@
-From 7162a1e87b3e380133dadc7909081bb70d0a7041 Mon Sep 17 00:00:00 2001
-From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
-Date: Thu, 21 Jan 2016 16:40:27 -0800
-Subject: mm: fix mlock accouting
-
-From: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
-
-commit 7162a1e87b3e380133dadc7909081bb70d0a7041 upstream.
-
-Tetsuo Handa reported underflow of NR_MLOCK on munlock.
-
-Testcase:
-
-    #include <stdio.h>
-    #include <stdlib.h>
-    #include <sys/mman.h>
-
-    #define BASE ((void *)0x400000000000)
-    #define SIZE (1UL << 21)
-
-    int main(int argc, char *argv[])
-    {
-        void *addr;
-
-        system("grep Mlocked /proc/meminfo");
-        addr = mmap(BASE, SIZE, PROT_READ | PROT_WRITE,
-                MAP_ANONYMOUS | MAP_PRIVATE | MAP_LOCKED | MAP_FIXED,
-                -1, 0);
-        if (addr == MAP_FAILED)
-            printf("mmap() failed\n"), exit(1);
-        munmap(addr, SIZE);
-        system("grep Mlocked /proc/meminfo");
-        return 0;
-    }
-
-It happens on munlock_vma_page() due to unfortunate choice of nr_pages
-data type:
-
-    __mod_zone_page_state(zone, NR_MLOCK, -nr_pages);
-
-For unsigned int nr_pages, implicitly casted to long in
-__mod_zone_page_state(), it becomes something around UINT_MAX.
-
-munlock_vma_page() usually called for THP as small pages go though
-pagevec.
-
-Let's make nr_pages signed int.
-
-Similar fixes in 6cdb18ad98a4 ("mm/vmstat: fix overflow in
-mod_zone_page_state()") used `long' type, but `int' here is OK for a
-count of the number of sub-pages in a huge page.
-
-Fixes: ff6a6da60b89 ("mm: accelerate munlock() treatment of THP pages")
-Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
-Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
-Tested-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
-Cc: Michel Lespinasse <walken@google.com>
-Acked-by: Michal Hocko <mhocko@suse.com>
-Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
-Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
----
- mm/mlock.c |    2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
---- a/mm/mlock.c
-+++ b/mm/mlock.c
-@@ -172,7 +172,7 @@ static void __munlock_isolation_failed(s
-  */
- unsigned int munlock_vma_page(struct page *page)
- {
--      unsigned int nr_pages;
-+      int nr_pages;
-       struct zone *zone = page_zone(page);
-       /* For try_to_munlock() and to serialize with page migration */
index f6ab67a980e0f0f9176df8d5f628fbbe25896796..d98256c772e14ae4acd88b502a11334d03e12ec4 100644 (file)
@@ -49,7 +49,6 @@ arm-8517-1-icst-avoid-arithmetic-overflow-in-icst_hz.patch
 kvm-ppc-fix-emulation-of-h_set_dabr-x-on-power8.patch
 fuse-break-infinite-loop-in-fuse_fill_write_pages.patch
 mm-soft-offline-check-return-value-in-second-__get_any_page-call.patch
-mm-fix-mlock-accouting.patch
 input-elantech-add-fujitsu-lifebook-u745-to-force-crc_enabled.patch
 input-elantech-mark-protocols-v2-and-v3-as-semi-mt.patch
 input-i8042-add-fujitsu-lifebook-u745-to-the-nomux-list.patch