dm writecache: fix data corruption when reloading the target

The dm-writecache reads metadata in the target constructor. However, when
we reload the target, there could be another active instance running on
the same device. This is the sequence of operations when doing a reload:

1. construct new target
2. suspend old target
3. resume new target
4. destroy old target

Metadata that were written by the old target between steps 1 and 2 would
not be visible by the new target.

Fix the data corruption by loading the metadata in the resume handler.

Also, validate block_size is at least as large as both the devices'
logical block size and only read 1 block from the metadata during
target constructor -- no need to read entirety of metadata now that it
is done during resume.

Fixes: 48debafe4f2f ("dm: add writecache target")
Cc: stable@vger.kernel.org # v4.18+
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>

1 files changed, 37 insertions, 15 deletions

diff --git a/drivers/md/dm-writecache.c b/drivers/md/dm-writecache.c
index 114927da9cc9..613c171b1b6d 100644
--- a/drivers/md/dm-writecache.c
+++ b/drivers/md/dm-writecache.c
@@ -931,6 +931,24 @@ static int writecache_alloc_entries(struct dm_writecache *wc)
 	return 0;
 }
 
+static int writecache_read_metadata(struct dm_writecache *wc, sector_t n_sectors)
+{
+	struct dm_io_region region;
+	struct dm_io_request req;
+
+	region.bdev = wc->ssd_dev->bdev;
+	region.sector = wc->start_sector;
+	region.count = n_sectors;
+	req.bi_op = REQ_OP_READ;
+	req.bi_op_flags = REQ_SYNC;
+	req.mem.type = DM_IO_VMA;
+	req.mem.ptr.vma = (char *)wc->memory_map;
+	req.client = wc->dm_io;
+	req.notify.fn = NULL;
+
+	return dm_io(&req, 1, &region, NULL);
+}
+
 static void writecache_resume(struct dm_target *ti)
 {
 	struct dm_writecache *wc = ti->private;
@@ -941,8 +959,18 @@ static void writecache_resume(struct dm_target *ti)
 
 	wc_lock(wc);
 
-	if (WC_MODE_PMEM(wc))
+	if (WC_MODE_PMEM(wc)) {
 		persistent_memory_invalidate_cache(wc->memory_map, wc->memory_map_size);
+	} else {
+		r = writecache_read_metadata(wc, wc->metadata_sectors);
+		if (r) {
+			size_t sb_entries_offset;
+			writecache_error(wc, r, "unable to read metadata: %d", r);
+			sb_entries_offset = offsetof(struct wc_memory_superblock, entries);
+			memset((char *)wc->memory_map + sb_entries_offset, -1,
+			       (wc->metadata_sectors << SECTOR_SHIFT) - sb_entries_offset);
+		}
+	}
 
 	wc->tree = RB_ROOT;
 	INIT_LIST_HEAD(&wc->lru);
@@ -2102,6 +2130,12 @@ static int writecache_ctr(struct dm_target *ti, unsigned argc, char **argv)
 		ti->error = "Invalid block size";
 		goto bad;
 	}
+	if (wc->block_size < bdev_logical_block_size(wc->dev->bdev) ||
+	    wc->block_size < bdev_logical_block_size(wc->ssd_dev->bdev)) {
+		r = -EINVAL;
+		ti->error = "Block size is smaller than device logical block size";
+		goto bad;
+	}
 	wc->block_size_bits = __ffs(wc->block_size);
 
 	wc->max_writeback_jobs = MAX_WRITEBACK_JOBS;
@@ -2200,8 +2234,6 @@ invalid_optional:
 			goto bad;
 		}
 	} else {
-		struct dm_io_region region;
-		struct dm_io_request req;
 		size_t n_blocks, n_metadata_blocks;
 		uint64_t n_bitmap_bits;
 
@@ -2258,19 +2290,9 @@ invalid_optional:
 			goto bad;
 		}
 
-		region.bdev = wc->ssd_dev->bdev;
-		region.sector = wc->start_sector;
-		region.count = wc->metadata_sectors;
-		req.bi_op = REQ_OP_READ;
-		req.bi_op_flags = REQ_SYNC;
-		req.mem.type = DM_IO_VMA;
-		req.mem.ptr.vma = (char *)wc->memory_map;
-		req.client = wc->dm_io;
-		req.notify.fn = NULL;
-
-		r = dm_io(&req, 1, &region, NULL);
+		r = writecache_read_metadata(wc, wc->block_size >> SECTOR_SHIFT);
 		if (r) {
-			ti->error = "Unable to read metadata";
+			ti->error = "Unable to read first block of metadata";
 			goto bad;
 		}
 	}