mm/memory-failure.c: shift page lock from head page to tail page after thp split

After thp split in hwpoison_user_mappings(), we hold page lock on the
raw error page only between try_to_unmap, hence we are in danger of race
condition.

I found in the RHEL7 MCE-relay testing that we have "bad page" error
when a memory error happens on a thp tail page used by qemu-kvm:

  Triggering MCE exception on CPU 10
  mce: [Hardware Error]: Machine check events logged
  MCE exception done on CPU 10
  MCE 0x38c535: Killing qemu-kvm:8418 due to hardware memory corruption
  MCE 0x38c535: dirty LRU page recovery: Recovered
  qemu-kvm[8418]: segfault at 20 ip 00007ffb0f0f229a sp 00007fffd6bc5240 error 4 in qemu-kvm[7ffb0ef14000+420000]
  BUG: Bad page state in process qemu-kvm  pfn:38c400
  page:ffffea000e310000 count:0 mapcount:0 mapping:          (null) index:0x7ffae3c00
  page flags: 0x2fffff0008001d(locked|referenced|uptodate|dirty|swapbacked)
  Modules linked in: hwpoison_inject mce_inject vhost_net macvtap macvlan ...
  CPU: 0 PID: 8418 Comm: qemu-kvm Tainted: G   M        --------------   3.10.0-54.0.1.el7.mce_test_fixed.x86_64 #1
  Hardware name: NEC NEC Express5800/R120b-1 [N8100-1719F]/MS-91E7-001, BIOS 4.6.3C19 02/10/2011
  Call Trace:
    dump_stack+0x19/0x1b
    bad_page.part.59+0xcf/0xe8
    free_pages_prepare+0x148/0x160
    free_hot_cold_page+0x31/0x140
    free_hot_cold_page_list+0x46/0xa0
    release_pages+0x1c1/0x200
    free_pages_and_swap_cache+0xad/0xd0
    tlb_flush_mmu.part.46+0x4c/0x90
    tlb_finish_mmu+0x55/0x60
    exit_mmap+0xcb/0x170
    mmput+0x67/0xf0
    vhost_dev_cleanup+0x231/0x260 [vhost_net]
    vhost_net_release+0x3f/0x90 [vhost_net]
    __fput+0xe9/0x270
    ____fput+0xe/0x10
    task_work_run+0xc4/0xe0
    do_exit+0x2bb/0xa40
    do_group_exit+0x3f/0xa0
    get_signal_to_deliver+0x1d0/0x6e0
    do_signal+0x48/0x5e0
    do_notify_resume+0x71/0xc0
    retint_signal+0x48/0x8c

The reason of this bug is that a page fault happens before unlocking the
head page at the end of memory_failure().  This strange page fault is
trying to access to address 0x20 and I'm not sure why qemu-kvm does
this, but anyway as a result the SIGSEGV makes qemu-kvm exit and on the
way we catch the bad page bug/warning because we try to free a locked
page (which was the former head page.)

To fix this, this patch suggests to shift page lock from head page to
tail page just after thp split.  SIGSEGV still happens, but it affects
only error affected VMs, not a whole system.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Wanpeng Li <liwanp@linux.vnet.ibm.com>
Cc: <stable@vger.kernel.org>        [3.9+] # a3e0f9e47d5ef "mm/memory-failure.c: transfer page count from head page to tail page after split thp"
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 11 insertions, 10 deletions

diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index b25ed321e66..4f08a2d6148 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -856,14 +856,14 @@ static int page_action(struct page_state *ps, struct page *p,
  * the pages and send SIGBUS to the processes if the data was dirty.
  */
 static int hwpoison_user_mappings(struct page *p, unsigned long pfn,
-				  int trapno, int flags)
+				  int trapno, int flags, struct page **hpagep)
 {
 	enum ttu_flags ttu = TTU_UNMAP | TTU_IGNORE_MLOCK | TTU_IGNORE_ACCESS;
 	struct address_space *mapping;
 	LIST_HEAD(tokill);
 	int ret;
 	int kill = 1, forcekill;
-	struct page *hpage = compound_head(p);
+	struct page *hpage = *hpagep;
 	struct page *ppage;
 
 	if (PageReserved(p) || PageSlab(p))
@@ -942,11 +942,14 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn,
 			 * We pinned the head page for hwpoison handling,
 			 * now we split the thp and we are interested in
 			 * the hwpoisoned raw page, so move the refcount
-			 * to it.
+			 * to it. Similarly, page lock is shifted.
 			 */
 			if (hpage != p) {
 				put_page(hpage);
 				get_page(p);
+				lock_page(p);
+				unlock_page(hpage);
+				*hpagep = p;
 			}
 			/* THP is split, so ppage should be the real poisoned page. */
 			ppage = p;
@@ -964,17 +967,11 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn,
 	if (kill)
 		collect_procs(ppage, &tokill);
 
-	if (hpage != ppage)
-		lock_page(ppage);
-
 	ret = try_to_unmap(ppage, ttu);
 	if (ret != SWAP_SUCCESS)
 		printk(KERN_ERR "MCE %#lx: failed to unmap page (mapcount=%d)\n",
 				pfn, page_mapcount(ppage));
 
-	if (hpage != ppage)
-		unlock_page(ppage);
-
 	/*
 	 * Now that the dirty bit has been propagated to the
 	 * struct page and all unmaps done we can decide if
@@ -1193,8 +1190,12 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
 	/*
 	 * Now take care of user space mappings.
 	 * Abort on fail: __delete_from_page_cache() assumes unmapped page.
+	 *
+	 * When the raw error page is thp tail page, hpage points to the raw
+	 * page after thp split.
 	 */
-	if (hwpoison_user_mappings(p, pfn, trapno, flags) != SWAP_SUCCESS) {
+	if (hwpoison_user_mappings(p, pfn, trapno, flags, &hpage)
+	    != SWAP_SUCCESS) {
 		printk(KERN_ERR "MCE %#lx: cannot unmap page, give up\n", pfn);
 		res = -EBUSY;
 		goto out;