[PATCH -next] mm/filemap: fix a data race in filemap_fault()

From: Qian Cai <cai@lca.pw>
To: akpm@linux-foundation.org
Cc: elver@google.com, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org, Qian Cai <cai@lca.pw>
Subject: [PATCH -next] mm/filemap: fix a data race in filemap_fault()
Date: Mon, 10 Feb 2020 12:00:29 -0500	[thread overview]
Message-ID: <1581354029-20154-1-git-send-email-cai@lca.pw> (raw)

struct file_ra_state ra.mmap_miss could be accessed concurrently during
page faults as noticed by KCSAN,

 BUG: KCSAN: data-race in filemap_fault / filemap_map_pages

 write to 0xffff9b1700a2c1b4 of 4 bytes by task 3292 on cpu 30:
  filemap_fault+0x920/0xfc0
  do_sync_mmap_readahead at mm/filemap.c:2384
  (inlined by) filemap_fault at mm/filemap.c:2486
  __xfs_filemap_fault+0x112/0x3e0 [xfs]
  xfs_filemap_fault+0x74/0x90 [xfs]
  __do_fault+0x9e/0x220
  do_fault+0x4a0/0x920
  __handle_mm_fault+0xc69/0xd00
  handle_mm_fault+0xfc/0x2f0
  do_page_fault+0x263/0x6f9
  page_fault+0x34/0x40

 read to 0xffff9b1700a2c1b4 of 4 bytes by task 3313 on cpu 32:
  filemap_map_pages+0xc2e/0xd80
  filemap_map_pages at mm/filemap.c:2625
  do_fault+0x3da/0x920
  __handle_mm_fault+0xc69/0xd00
  handle_mm_fault+0xfc/0x2f0
  do_page_fault+0x263/0x6f9
  page_fault+0x34/0x40

 Reported by Kernel Concurrency Sanitizer on:
 CPU: 32 PID: 3313 Comm: systemd-udevd Tainted: G        W    L 5.5.0-next-20200210+ #1
 Hardware name: HPE ProLiant DL385 Gen10/ProLiant DL385 Gen10, BIOS A40 07/10/2019

ra.mmap_miss is used to contribute the readahead decisions, a data race
could be undesirable. Since the stores are aligned and less than
word-size, assume they are safe. Thus, fixing it by adding READ_ONCE()
for the loads except those places comparing to zero where they are safe.

Signed-off-by: Qian Cai <cai@lca.pw>
---
 mm/filemap.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index 1784478270e1..b6c1d37f7ea3 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2380,14 +2380,14 @@ static struct file *do_sync_mmap_readahead(struct vm_fault *vmf)
 	}
 
 	/* Avoid banging the cache line if not needed */
-	if (ra->mmap_miss < MMAP_LOTSAMISS * 10)
+	if (READ_ONCE(ra->mmap_miss) < MMAP_LOTSAMISS * 10)
 		ra->mmap_miss++;
 
 	/*
 	 * Do we miss much more than hit in this file? If so,
 	 * stop bothering with read-ahead. It will only hurt.
 	 */
-	if (ra->mmap_miss > MMAP_LOTSAMISS)
+	if (READ_ONCE(ra->mmap_miss) > MMAP_LOTSAMISS)
 		return fpin;
 
 	/*
@@ -2418,7 +2418,7 @@ static struct file *do_async_mmap_readahead(struct vm_fault *vmf,
 	/* If we don't want any read-ahead, don't bother */
 	if (vmf->vma->vm_flags & VM_RAND_READ)
 		return fpin;
-	if (ra->mmap_miss > 0)
+	if (data_race(ra->mmap_miss > 0))
 		ra->mmap_miss--;
 	if (PageReadahead(page)) {
 		fpin = maybe_unlock_mmap_for_io(vmf, fpin);
@@ -2622,7 +2622,7 @@ void filemap_map_pages(struct vm_fault *vmf,
 		if (page->index >= max_idx)
 			goto unlock;
 
-		if (file->f_ra.mmap_miss > 0)
+		if (data_race(file->f_ra.mmap_miss > 0))
 			file->f_ra.mmap_miss--;
 
 		vmf->address += (xas.xa_index - last_pgoff) << PAGE_SHIFT;
-- 
1.8.3.1

