[dm-devel] [PATCH] dm-writecache: support offset

[Mikulas Patocka]

Add a parameter "start_sector". It is offset in the unit of  512-byte
sectors from the start of the cache device. The sectors below the offset
are unused and they may be used for userspace metadata.

Signed-off-by: Mikulas Patocka <mpatocka at redhat.com>

---
 Documentation/device-mapper/writecache.txt |    2 +
 drivers/md/dm-writecache.c                 |   43 +++++++++++++++++++----------
 2 files changed, 31 insertions(+), 14 deletions(-)

Index: linux-2.6/drivers/md/dm-writecache.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-writecache.c	2018-06-27 20:54:00.000000000 +0200
+++ linux-2.6/drivers/md/dm-writecache.c	2018-06-29 02:57:46.260000000 +0200
@@ -136,6 +136,7 @@ struct dm_writecache {
 	struct dm_target *ti;
 	struct dm_dev *dev;
 	struct dm_dev *ssd_dev;
+	sector_t start_sector;
 	void *memory_map;
 	uint64_t memory_map_size;
 	size_t metadata_sectors;
@@ -293,6 +294,10 @@ static int persistent_memory_claim(struc
 	}
 
 	dax_read_unlock(id);
+
+	wc->memory_map += (size_t)wc->start_sector << SECTOR_SHIFT;
+	wc->memory_map_size -= (size_t)wc->start_sector << SECTOR_SHIFT;
+
 	return 0;
 err3:
 	kvfree(pages);
@@ -311,7 +316,7 @@ static int persistent_memory_claim(struc
 static void persistent_memory_release(struct dm_writecache *wc)
 {
 	if (wc->memory_vmapped)
-		vunmap(wc->memory_map);
+		vunmap(wc->memory_map - ((size_t)wc->start_sector << SECTOR_SHIFT));
 }
 
 static struct page *persistent_memory_page(void *addr)
@@ -359,7 +364,7 @@ static void *memory_data(struct dm_write
 
 static sector_t cache_sector(struct dm_writecache *wc, struct wc_entry *e)
 {
-	return wc->metadata_sectors +
+	return wc->start_sector + wc->metadata_sectors +
 		((sector_t)e->index << (wc->block_size_bits - SECTOR_SHIFT));
 }
 
@@ -471,6 +476,7 @@ static void ssd_commit_flushed(struct dm
 		if (unlikely(region.sector + region.count > wc->metadata_sectors))
 			region.count = wc->metadata_sectors - region.sector;
 
+		region.sector += wc->start_sector;
 		atomic_inc(&endio.count);
 		req.bi_op = REQ_OP_WRITE;
 		req.bi_op_flags = REQ_SYNC;
@@ -1946,14 +1952,6 @@ static int writecache_ctr(struct dm_targ
 	}
 	wc->memory_map_size = i_size_read(wc->ssd_dev->bdev->bd_inode);
 
-	if (WC_MODE_PMEM(wc)) {
-		r = persistent_memory_claim(wc);
-		if (r) {
-			ti->error = "Unable to map persistent memory for cache";
-			goto bad;
-		}
-	}
-
 	/*
 	 * Parse the cache block size
 	 */
@@ -1982,7 +1980,16 @@ static int writecache_ctr(struct dm_targ
 
 	while (opt_params) {
 		string = dm_shift_arg(&as), opt_params--;
-		if (!strcasecmp(string, "high_watermark") && opt_params >= 1) {
+		if (!strcasecmp(string, "start_sector") && opt_params >= 1) {
+			unsigned long long start_sector;
+			string = dm_shift_arg(&as), opt_params--;
+			if (sscanf(string, "%llu%c", &start_sector, &dummy) != 1)
+				goto invalid_optional;
+			wc->start_sector = start_sector;
+			if (wc->start_sector != start_sector ||
+			    wc->start_sector >= wc->memory_map_size >> SECTOR_SHIFT)
+				goto invalid_optional;
+		} else if (!strcasecmp(string, "high_watermark") && opt_params >= 1) {
 			string = dm_shift_arg(&as), opt_params--;
 			if (sscanf(string, "%d%c", &high_wm_percent, &dummy) != 1)
 				goto invalid_optional;
@@ -2039,12 +2046,20 @@ invalid_optional:
 		goto bad;
 	}
 
-	if (!WC_MODE_PMEM(wc)) {
+	if (WC_MODE_PMEM(wc)) {
+		r = persistent_memory_claim(wc);
+		if (r) {
+			ti->error = "Unable to map persistent memory for cache";
+			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;
 
+		wc->memory_map_size -= (uint64_t)wc->start_sector << SECTOR_SHIFT;
+
 		bio_list_init(&wc->flush_list);
 		wc->flush_thread = kthread_create(writecache_flush_thread, wc, "dm_writecache_flush");
 		if (IS_ERR(wc->flush_thread)) {
@@ -2097,7 +2112,7 @@ invalid_optional:
 		}
 
 		region.bdev = wc->ssd_dev->bdev;
-		region.sector = 0;
+		region.sector = wc->start_sector;
 		region.count = wc->metadata_sectors;
 		req.bi_op = REQ_OP_READ;
 		req.bi_op_flags = REQ_SYNC;
@@ -2265,7 +2280,7 @@ static void writecache_status(struct dm_
 
 static struct target_type writecache_target = {
 	.name			= "writecache",
-	.version		= {1, 0, 0},
+	.version		= {1, 1, 0},
 	.module			= THIS_MODULE,
 	.ctr			= writecache_ctr,
 	.dtr			= writecache_dtr,
Index: linux-2.6/Documentation/device-mapper/writecache.txt
===================================================================
--- linux-2.6.orig/Documentation/device-mapper/writecache.txt	2018-06-18 01:11:41.400000000 +0200
+++ linux-2.6/Documentation/device-mapper/writecache.txt	2018-06-29 02:57:04.360000000 +0200
@@ -15,6 +15,8 @@ Constructor parameters:
    size)
 5. the number of optional parameters (the parameters with an argument
    count as two)
+	start_sector n		(default: 0)
+		offset from the start of cache device in 512-byte sectors
 	high_watermark n	(default: 50)
 		start writeback when the number of used blocks reach this
 		watermark