[patch 0/8] dm-integrity patches

[patch 0/8] dm-integrity patches

[Mikulas Patocka]

Hi

Here I'm sending the dm-integrity patches for separate data and metadata
device.

Mikulas

[patch 1/8] dm-integrity: use kvfree

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-kvfree.patch --]
[-- Type: text/plain, Size: 815 bytes --]

Use kvfree instead of kfree because the structure is allocated with
kvmalloc.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org	# v4.12+

---
 drivers/md/dm-integrity.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-04-29 23:10:11.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-04-29 23:10:11.000000000 +0200
@@ -2440,7 +2440,7 @@ static void dm_integrity_free_journal_sc
 	unsigned i;
 	for (i = 0; i < ic->journal_sections; i++)
 		kvfree(sl[i]);
-	kfree(sl);
+	kvfree(sl);
 }
 
 static struct scatterlist **dm_integrity_alloc_journal_scatterlist(struct dm_integrity_c *ic, struct page_list *pl)

[patch 2/8] dm-integrity: change the variable suspending from bool to int

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-suspending.patch --]
[-- Type: text/plain, Size: 1182 bytes --]

The early alpha processors can't write a byte or short atomically, so
asynchronously-modified variables shouldn't really have byte type.
Change is to int.

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

---
 drivers/md/dm-integrity.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-05-01 00:23:32.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-05-01 00:38:57.000000000 +0200
@@ -178,7 +178,7 @@ struct dm_integrity_c {
 	__u8 sectors_per_block;
 
 	unsigned char mode;
-	bool suspending;
+	int suspending;
 
 	int failed;
 
@@ -2210,7 +2210,7 @@ static void dm_integrity_postsuspend(str
 
 	del_timer_sync(&ic->autocommit_timer);
 
-	ic->suspending = true;
+	WRITE_ONCE(ic->suspending, 1);
 
 	queue_work(ic->commit_wq, &ic->commit_work);
 	drain_workqueue(ic->commit_wq);
@@ -2220,7 +2220,7 @@ static void dm_integrity_postsuspend(str
 		dm_integrity_flush_buffers(ic);
 	}
 
-	ic->suspending = false;
+	WRITE_ONCE(ic->suspending, 0);
 
 	BUG_ON(!RB_EMPTY_ROOT(&ic->in_progress));
 

[patch 3/8] dm-integrity: report size in the status

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-report-size.patch --]
[-- Type: text/plain, Size: 827 bytes --]

Report provided data sectors in the status.

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

---
 drivers/md/dm-integrity.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-01-31 13:43:07.701126000 +0100
+++ linux-2.6/drivers/md/dm-integrity.c	2018-02-01 20:38:17.038213000 +0100
@@ -2243,7 +2243,9 @@ static void dm_integrity_status(struct d
 
 	switch (type) {
 	case STATUSTYPE_INFO:
-		DMEMIT("%llu", (unsigned long long)atomic64_read(&ic->number_of_mismatches));
+		DMEMIT("%llu %llu",
+			(unsigned long long)atomic64_read(&ic->number_of_mismatches),
+			(unsigned long long)ic->provided_data_sectors);
 		break;
 
 	case STATUSTYPE_TABLE: {

[patch 4/8] dm-integrity: add ic->start in get_data_sector

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-move-start-to-get_data_sector.patch --]
[-- Type: text/plain, Size: 1701 bytes --]

A small refactoring. Add the variable ic->start in get_data_sector and not
in the callers. This is a prerequisite for the patch that uses external
metadata device.

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

---
 drivers/md/dm-integrity.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-02-01 20:38:17.038213000 +0100
+++ linux-2.6/drivers/md/dm-integrity.c	2018-02-06 17:31:35.816018000 +0100
@@ -386,6 +386,8 @@ static sector_t get_data_sector(struct d
 		result += (area + 1) * ic->metadata_run;
 
 	result += (sector_t)ic->initial_sectors + offset;
+	result += ic->start;
+
 	return result;
 }
 
@@ -857,7 +859,7 @@ static void copy_from_journal(struct dm_
 	io_req.notify.context = data;
 	io_req.client = ic->io;
 	io_loc.bdev = ic->dev->bdev;
-	io_loc.sector = ic->start + target;
+	io_loc.sector = target;
 	io_loc.count = n_sectors;
 
 	r = dm_io(&io_req, 1, &io_loc, NULL);
@@ -1701,7 +1703,6 @@ sleep:
 	bio->bi_end_io = integrity_end_io;
 
 	bio->bi_iter.bi_size = dio->range.n_sectors << SECTOR_SHIFT;
-	bio->bi_iter.bi_sector += ic->start;
 	generic_make_request(bio);
 
 	if (need_sync_io) {
@@ -2339,7 +2340,7 @@ static int calculate_device_limits(struc
 	get_area_and_offset(ic, ic->provided_data_sectors - 1, &last_area, &last_offset);
 	last_sector = get_data_sector(ic, last_area, last_offset);
 
-	if (ic->start + last_sector < last_sector || ic->start + last_sector >= ic->device_sectors)
+	if (last_sector < ic->start || last_sector >= ic->device_sectors)
 		return -EINVAL;
 
 	return 0;

[patch 5/8] dm-integrity: allow external devices

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-separate-device.patch --]
[-- Type: text/plain, Size: 14615 bytes --]

This patch adds a possibility to use external device for metadata for
dm-integrity. It is activated with the option "meta_device:/dev/device".

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

---
 drivers/md/dm-integrity.c |  203 +++++++++++++++++++++++++++++++++-------------
 1 file changed, 149 insertions(+), 54 deletions(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-05-01 04:35:08.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-05-01 04:35:26.000000000 +0200
@@ -139,6 +139,7 @@ struct alg_spec {
 
 struct dm_integrity_c {
 	struct dm_dev *dev;
+	struct dm_dev *meta_dev;
 	unsigned tag_size;
 	__s8 log2_tag_size;
 	sector_t start;
@@ -170,7 +171,8 @@ struct dm_integrity_c {
 	unsigned short journal_section_sectors;
 	unsigned journal_sections;
 	unsigned journal_entries;
-	sector_t device_sectors;
+	sector_t data_device_sectors;
+	sector_t meta_device_sectors;
 	unsigned initial_sectors;
 	unsigned metadata_run;
 	__s8 log2_metadata_run;
@@ -337,10 +339,14 @@ static commit_id_t dm_integrity_commit_i
 static void get_area_and_offset(struct dm_integrity_c *ic, sector_t data_sector,
 				sector_t *area, sector_t *offset)
 {
-	__u8 log2_interleave_sectors = ic->sb->log2_interleave_sectors;
-
-	*area = data_sector >> log2_interleave_sectors;
-	*offset = (unsigned)data_sector & ((1U << log2_interleave_sectors) - 1);
+	if (!ic->meta_dev) {
+		__u8 log2_interleave_sectors = ic->sb->log2_interleave_sectors;
+		*area = data_sector >> log2_interleave_sectors;
+		*offset = (unsigned)data_sector & ((1U << log2_interleave_sectors) - 1);
+	} else {
+		*area = 0;
+		*offset = data_sector;
+	}
 }
 
 #define sector_to_block(ic, n)						\
@@ -379,6 +385,9 @@ static sector_t get_data_sector(struct d
 {
 	sector_t result;
 
+	if (ic->meta_dev)
+		return offset;
+
 	result = area << ic->sb->log2_interleave_sectors;
 	if (likely(ic->log2_metadata_run >= 0))
 		result += (area + 1) << ic->log2_metadata_run;
@@ -408,7 +417,7 @@ static int sync_rw_sb(struct dm_integrit
 	io_req.mem.ptr.addr = ic->sb;
 	io_req.notify.fn = NULL;
 	io_req.client = ic->io;
-	io_loc.bdev = ic->dev->bdev;
+	io_loc.bdev = ic->meta_dev ? ic->meta_dev->bdev : ic->dev->bdev;
 	io_loc.sector = ic->start;
 	io_loc.count = SB_SECTORS;
 
@@ -755,7 +764,7 @@ static void rw_journal(struct dm_integri
 		io_req.notify.fn = NULL;
 	}
 	io_req.client = ic->io;
-	io_loc.bdev = ic->dev->bdev;
+	io_loc.bdev = ic->meta_dev ? ic->meta_dev->bdev : ic->dev->bdev;
 	io_loc.sector = ic->start + SB_SECTORS + sector;
 	io_loc.count = n_sectors;
 
@@ -2254,12 +2263,15 @@ static void dm_integrity_status(struct d
 		watermark_percentage += ic->journal_entries / 2;
 		do_div(watermark_percentage, ic->journal_entries);
 		arg_count = 5;
+		arg_count += !!ic->meta_dev;
 		arg_count += ic->sectors_per_block != 1;
 		arg_count += !!ic->internal_hash_alg.alg_string;
 		arg_count += !!ic->journal_crypt_alg.alg_string;
 		arg_count += !!ic->journal_mac_alg.alg_string;
 		DMEMIT("%s %llu %u %c %u", ic->dev->name, (unsigned long long)ic->start,
 		       ic->tag_size, ic->mode, arg_count);
+		if (ic->meta_dev)
+			DMEMIT(" meta_device:%s", ic->meta_dev->name);
 		DMEMIT(" journal_sectors:%u", ic->initial_sectors - SB_SECTORS);
 		DMEMIT(" interleave_sectors:%u", 1U << ic->sb->log2_interleave_sectors);
 		DMEMIT(" buffer_sectors:%u", 1U << ic->log2_buffer_sectors);
@@ -2289,7 +2301,10 @@ static int dm_integrity_iterate_devices(
 {
 	struct dm_integrity_c *ic = ti->private;
 
-	return fn(ti, ic->dev, ic->start + ic->initial_sectors + ic->metadata_run, ti->len, data);
+	if (!ic->meta_dev)
+		return fn(ti, ic->dev, ic->start + ic->initial_sectors + ic->metadata_run, ti->len, data);
+	else
+		return fn(ti, ic->dev, 0, ti->len, data);
 }
 
 static void dm_integrity_io_hints(struct dm_target *ti, struct queue_limits *limits)
@@ -2322,26 +2337,38 @@ static void calculate_journal_section_si
 static int calculate_device_limits(struct dm_integrity_c *ic)
 {
 	__u64 initial_sectors;
-	sector_t last_sector, last_area, last_offset;
 
 	calculate_journal_section_size(ic);
 	initial_sectors = SB_SECTORS + (__u64)ic->journal_section_sectors * ic->journal_sections;
-	if (initial_sectors + METADATA_PADDING_SECTORS >= ic->device_sectors || initial_sectors > UINT_MAX)
+	if (initial_sectors + METADATA_PADDING_SECTORS >= ic->meta_device_sectors || initial_sectors > UINT_MAX)
 		return -EINVAL;
 	ic->initial_sectors = initial_sectors;
 
-	ic->metadata_run = roundup((__u64)ic->tag_size << (ic->sb->log2_interleave_sectors - ic->sb->log2_sectors_per_block),
-				   (__u64)(1 << SECTOR_SHIFT << METADATA_PADDING_SECTORS)) >> SECTOR_SHIFT;
-	if (!(ic->metadata_run & (ic->metadata_run - 1)))
-		ic->log2_metadata_run = __ffs(ic->metadata_run);
-	else
-		ic->log2_metadata_run = -1;
+	if (!ic->meta_dev) {
+		sector_t last_sector, last_area, last_offset;
 
-	get_area_and_offset(ic, ic->provided_data_sectors - 1, &last_area, &last_offset);
-	last_sector = get_data_sector(ic, last_area, last_offset);
+		ic->metadata_run = roundup((__u64)ic->tag_size << (ic->sb->log2_interleave_sectors - ic->sb->log2_sectors_per_block),
+					   (__u64)(1 << SECTOR_SHIFT << METADATA_PADDING_SECTORS)) >> SECTOR_SHIFT;
+		if (!(ic->metadata_run & (ic->metadata_run - 1)))
+			ic->log2_metadata_run = __ffs(ic->metadata_run);
+		else
+			ic->log2_metadata_run = -1;
 
-	if (last_sector < ic->start || last_sector >= ic->device_sectors)
-		return -EINVAL;
+		get_area_and_offset(ic, ic->provided_data_sectors - 1, &last_area, &last_offset);
+		last_sector = get_data_sector(ic, last_area, last_offset);
+		if (last_sector < ic->start || last_sector >= ic->meta_device_sectors)
+			return -EINVAL;
+	} else {
+		__u64 meta_size = ic->provided_data_sectors * ic->tag_size;
+		meta_size = (meta_size + ((1U << (ic->log2_buffer_sectors + SECTOR_SHIFT)) - 1))
+				>> (ic->log2_buffer_sectors + SECTOR_SHIFT);
+		meta_size <<= ic->log2_buffer_sectors;
+		if (ic->initial_sectors + meta_size < ic->initial_sectors ||
+		    ic->initial_sectors + meta_size > ic->meta_device_sectors)
+			return -EINVAL;
+		ic->metadata_run = 1;
+		ic->log2_metadata_run = 0;
+	}
 
 	return 0;
 }
@@ -2363,26 +2390,51 @@ static int initialize_superblock(struct
 	journal_sections = journal_sectors / ic->journal_section_sectors;
 	if (!journal_sections)
 		journal_sections = 1;
-	ic->sb->journal_sections = cpu_to_le32(journal_sections);
-
-	if (!interleave_sectors)
-		interleave_sectors = DEFAULT_INTERLEAVE_SECTORS;
-	ic->sb->log2_interleave_sectors = __fls(interleave_sectors);
-	ic->sb->log2_interleave_sectors = max((__u8)MIN_LOG2_INTERLEAVE_SECTORS, ic->sb->log2_interleave_sectors);
-	ic->sb->log2_interleave_sectors = min((__u8)MAX_LOG2_INTERLEAVE_SECTORS, ic->sb->log2_interleave_sectors);
 
-	ic->provided_data_sectors = 0;
-	for (test_bit = fls64(ic->device_sectors) - 1; test_bit >= 3; test_bit--) {
-		__u64 prev_data_sectors = ic->provided_data_sectors;
+	if (!ic->meta_dev) {
+		ic->sb->journal_sections = cpu_to_le32(journal_sections);
+		if (!interleave_sectors)
+			interleave_sectors = DEFAULT_INTERLEAVE_SECTORS;
+		ic->sb->log2_interleave_sectors = __fls(interleave_sectors);
+		ic->sb->log2_interleave_sectors = max((__u8)MIN_LOG2_INTERLEAVE_SECTORS, ic->sb->log2_interleave_sectors);
+		ic->sb->log2_interleave_sectors = min((__u8)MAX_LOG2_INTERLEAVE_SECTORS, ic->sb->log2_interleave_sectors);
+
+		ic->provided_data_sectors = 0;
+		for (test_bit = fls64(ic->meta_device_sectors) - 1; test_bit >= 3; test_bit--) {
+			__u64 prev_data_sectors = ic->provided_data_sectors;
+
+			ic->provided_data_sectors |= (sector_t)1 << test_bit;
+			if (calculate_device_limits(ic))
+				ic->provided_data_sectors = prev_data_sectors;
+		}
+		if (!ic->provided_data_sectors)
+			return -EINVAL;
+	} else {
+		ic->sb->log2_interleave_sectors = 0;
+		ic->provided_data_sectors = ic->data_device_sectors;
+		ic->provided_data_sectors &= ~(sector_t)(ic->sectors_per_block - 1);
+
+try_smaller_buffer:
+		ic->sb->journal_sections = cpu_to_le32(0);
+		for (test_bit = fls(journal_sections) - 1; test_bit >= 0; test_bit--) {
+			__u32 prev_journal_sections = le32_to_cpu(ic->sb->journal_sections);
+			__u32 test_journal_sections = prev_journal_sections | (1U << test_bit);
+			if (test_journal_sections > journal_sections)
+				continue;
+			ic->sb->journal_sections = cpu_to_le32(test_journal_sections);
+			if (calculate_device_limits(ic))
+				ic->sb->journal_sections = cpu_to_le32(prev_journal_sections);
 
-		ic->provided_data_sectors |= (sector_t)1 << test_bit;
-		if (calculate_device_limits(ic))
-			ic->provided_data_sectors = prev_data_sectors;
+		}
+		if (!le32_to_cpu(ic->sb->journal_sections)) {
+			if (ic->log2_buffer_sectors > 3) {
+				ic->log2_buffer_sectors--;
+				goto try_smaller_buffer;
+			}
+			return -EINVAL;
+		}
 	}
 
-	if (!ic->provided_data_sectors)
-		return -EINVAL;
-
 	ic->sb->provided_data_sectors = cpu_to_le64(ic->provided_data_sectors);
 
 	return 0;
@@ -2878,9 +2930,7 @@ static int dm_integrity_ctr(struct dm_ta
 		goto bad;
 	}
 
-	ic->device_sectors = i_size_read(ic->dev->bdev->bd_inode) >> SECTOR_SHIFT;
-	journal_sectors = min((sector_t)DEFAULT_MAX_JOURNAL_SECTORS,
-			ic->device_sectors >> DEFAULT_JOURNAL_SIZE_FACTOR);
+	journal_sectors = 0;
 	interleave_sectors = DEFAULT_INTERLEAVE_SECTORS;
 	buffer_sectors = DEFAULT_BUFFER_SECTORS;
 	journal_watermark = DEFAULT_JOURNAL_WATERMARK;
@@ -2903,7 +2953,7 @@ static int dm_integrity_ctr(struct dm_ta
 			goto bad;
 		}
 		if (sscanf(opt_string, "journal_sectors:%u%c", &val, &dummy) == 1)
-			journal_sectors = val;
+			journal_sectors = val ? val : 1;
 		else if (sscanf(opt_string, "interleave_sectors:%u%c", &val, &dummy) == 1)
 			interleave_sectors = val;
 		else if (sscanf(opt_string, "buffer_sectors:%u%c", &val, &dummy) == 1)
@@ -2912,7 +2962,17 @@ static int dm_integrity_ctr(struct dm_ta
 			journal_watermark = val;
 		else if (sscanf(opt_string, "commit_time:%u%c", &val, &dummy) == 1)
 			sync_msec = val;
-		else if (sscanf(opt_string, "block_size:%u%c", &val, &dummy) == 1) {
+		else if (!memcmp(opt_string, "meta_device:", strlen("meta_device:"))) {
+			if (ic->meta_dev) {
+				dm_put_device(ti, ic->meta_dev);
+				ic->meta_dev = NULL;
+			}
+			r = dm_get_device(ti, strchr(opt_string, ':') + 1, dm_table_get_mode(ti->table), &ic->meta_dev);
+			if (r) {
+				ti->error = "Device lookup failed";
+				goto bad;
+			}
+		} else if (sscanf(opt_string, "block_size:%u%c", &val, &dummy) == 1) {
 			if (val < 1 << SECTOR_SHIFT ||
 			    val > MAX_SECTORS_PER_BLOCK << SECTOR_SHIFT ||
 			    (val & (val -1))) {
@@ -2943,6 +3003,21 @@ static int dm_integrity_ctr(struct dm_ta
 		}
 	}
 
+	ic->data_device_sectors = i_size_read(ic->dev->bdev->bd_inode) >> SECTOR_SHIFT;
+	if (!ic->meta_dev)
+		ic->meta_device_sectors = ic->data_device_sectors;
+	else
+		ic->meta_device_sectors = i_size_read(ic->meta_dev->bdev->bd_inode) >> SECTOR_SHIFT;
+
+	if (!journal_sectors) {
+		journal_sectors = min((sector_t)DEFAULT_MAX_JOURNAL_SECTORS,
+			ic->data_device_sectors >> DEFAULT_JOURNAL_SIZE_FACTOR);
+	}
+
+	if (!buffer_sectors)
+		buffer_sectors = 1;
+	ic->log2_buffer_sectors = min((int)__fls(buffer_sectors), 31 - SECTOR_SHIFT);
+
 	r = get_mac(&ic->internal_hash, &ic->internal_hash_alg, &ti->error,
 		    "Invalid internal hash", "Error setting internal hash key");
 	if (r)
@@ -3079,11 +3154,19 @@ static int dm_integrity_ctr(struct dm_ta
 		goto bad;
 	}
 	/* make sure that ti->max_io_len doesn't overflow */
-	if (ic->sb->log2_interleave_sectors < MIN_LOG2_INTERLEAVE_SECTORS ||
-	    ic->sb->log2_interleave_sectors > MAX_LOG2_INTERLEAVE_SECTORS) {
-		r = -EINVAL;
-		ti->error = "Invalid interleave_sectors in the superblock";
-		goto bad;
+	if (!ic->meta_dev) {
+		if (ic->sb->log2_interleave_sectors < MIN_LOG2_INTERLEAVE_SECTORS ||
+		    ic->sb->log2_interleave_sectors > MAX_LOG2_INTERLEAVE_SECTORS) {
+			r = -EINVAL;
+			ti->error = "Invalid interleave_sectors in the superblock";
+			goto bad;
+		}
+	} else {
+		if (ic->sb->log2_interleave_sectors) {
+			r = -EINVAL;
+			ti->error = "Invalid interleave_sectors in the superblock";
+			goto bad;
+		}
 	}
 	ic->provided_data_sectors = le64_to_cpu(ic->sb->provided_data_sectors);
 	if (ic->provided_data_sectors != le64_to_cpu(ic->sb->provided_data_sectors)) {
@@ -3097,20 +3180,28 @@ static int dm_integrity_ctr(struct dm_ta
 		ti->error = "Journal mac mismatch";
 		goto bad;
 	}
+
+try_smaller_buffer:
 	r = calculate_device_limits(ic);
 	if (r) {
+		if (ic->meta_dev) {
+			if (ic->log2_buffer_sectors > 3) {
+				ic->log2_buffer_sectors--;
+				goto try_smaller_buffer;
+			}
+		}
 		ti->error = "The device is too small";
 		goto bad;
 	}
+	if (!ic->meta_dev)
+		ic->log2_buffer_sectors = min(ic->log2_buffer_sectors, (__u8)__ffs(ic->metadata_run));
+
 	if (ti->len > ic->provided_data_sectors) {
 		r = -EINVAL;
 		ti->error = "Not enough provided sectors for requested mapping size";
 		goto bad;
 	}
 
-	if (!buffer_sectors)
-		buffer_sectors = 1;
-	ic->log2_buffer_sectors = min3((int)__fls(buffer_sectors), (int)__ffs(ic->metadata_run), 31 - SECTOR_SHIFT);
 
 	threshold = (__u64)ic->journal_entries * (100 - journal_watermark);
 	threshold += 50;
@@ -3134,8 +3225,8 @@ static int dm_integrity_ctr(struct dm_ta
 		    (unsigned long long)ic->provided_data_sectors);
 	DEBUG_print("	log2_buffer_sectors %u\n", ic->log2_buffer_sectors);
 
-	ic->bufio = dm_bufio_client_create(ic->dev->bdev, 1U << (SECTOR_SHIFT + ic->log2_buffer_sectors),
-					   1, 0, NULL, NULL);
+	ic->bufio = dm_bufio_client_create(ic->meta_dev ? ic->meta_dev->bdev : ic->dev->bdev,
+			1U << (SECTOR_SHIFT + ic->log2_buffer_sectors), 1, 0, NULL, NULL);
 	if (IS_ERR(ic->bufio)) {
 		r = PTR_ERR(ic->bufio);
 		ti->error = "Cannot initialize dm-bufio";
@@ -3167,9 +3258,11 @@ static int dm_integrity_ctr(struct dm_ta
 		ic->just_formatted = true;
 	}
 
-	r = dm_set_target_max_io_len(ti, 1U << ic->sb->log2_interleave_sectors);
-	if (r)
-		goto bad;
+	if (!ic->meta_dev) {
+		r = dm_set_target_max_io_len(ti, 1U << ic->sb->log2_interleave_sectors);
+		if (r)
+			goto bad;
+	}
 
 	if (!ic->internal_hash)
 		dm_integrity_set(ti, ic);
@@ -3204,6 +3297,8 @@ static void dm_integrity_dtr(struct dm_t
 		dm_io_client_destroy(ic->io);
 	if (ic->dev)
 		dm_put_device(ti, ic->dev);
+	if (ic->meta_dev)
+		dm_put_device(ti, ic->meta_dev);
 	dm_integrity_free_page_list(ic, ic->journal);
 	dm_integrity_free_page_list(ic, ic->journal_io);
 	dm_integrity_free_page_list(ic, ic->journal_xor);

[patch 6/8] dm-integrity: use version 2 for separate metadata

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-sb-version.patch --]
[-- Type: text/plain, Size: 1901 bytes --]

Use version "2" in the superblock when data and metadata devices are
separate, so that the device is no accidentally read by older kernel
version.

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

---
 drivers/md/dm-integrity.c |   16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-05-01 04:34:46.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-05-01 04:34:46.000000000 +0200
@@ -44,7 +44,8 @@
  */
 
 #define SB_MAGIC			"integrt"
-#define SB_VERSION			1
+#define SB_VERSION_1			1
+#define SB_VERSION_2			2
 #define SB_SECTORS			8
 #define MAX_SECTORS_PER_BLOCK		8
 
@@ -406,6 +407,14 @@ static void wraparound_section(struct dm
 		*sec_ptr -= ic->journal_sections;
 }
 
+static void sb_set_version(struct dm_integrity_c *ic)
+{
+	if (ic->meta_dev)
+		ic->sb->version = SB_VERSION_2;
+	else
+		ic->sb->version = SB_VERSION_1;
+}
+
 static int sync_rw_sb(struct dm_integrity_c *ic, int op, int op_flags)
 {
 	struct dm_io_request io_req;
@@ -2380,7 +2389,6 @@ static int initialize_superblock(struct
 
 	memset(ic->sb, 0, SB_SECTORS << SECTOR_SHIFT);
 	memcpy(ic->sb->magic, SB_MAGIC, 8);
-	ic->sb->version = SB_VERSION;
 	ic->sb->integrity_tag_size = cpu_to_le16(ic->tag_size);
 	ic->sb->log2_sectors_per_block = __ffs(ic->sectors_per_block);
 	if (ic->journal_mac_alg.alg_string)
@@ -2437,6 +2445,8 @@ try_smaller_buffer:
 
 	ic->sb->provided_data_sectors = cpu_to_le64(ic->provided_data_sectors);
 
+	sb_set_version(ic);
+
 	return 0;
 }
 
@@ -3133,7 +3143,7 @@ static int dm_integrity_ctr(struct dm_ta
 			should_write_sb = true;
 	}
 
-	if (ic->sb->version != SB_VERSION) {
+	if (!ic->sb->version || ic->sb->version > SB_VERSION_2) {
 		r = -EINVAL;
 		ti->error = "Unknown version";
 		goto bad;

[patch 7/8] dm-integrity: flush journal on suspend when using separate device

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-flush-on-suspend.patch --]
[-- Type: text/plain, Size: 1324 bytes --]

Flush the journal on suspend when using separate data and metadata device,
so that the metadata device can be discarded and the table can be reloaded
with a linear target pointing to the data device.

Note: we deliberately don't flush the journal when using the same device
for metadata and data, so that the journal replay code is tested.

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

---
 drivers/md/dm-integrity.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-05-01 00:39:27.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-05-01 00:39:27.000000000 +0200
@@ -2000,7 +2000,7 @@ static void integrity_writer(struct work
 	unsigned prev_free_sectors;
 
 	/* the following test is not needed, but it tests the replay code */
-	if (READ_ONCE(ic->suspending))
+	if (READ_ONCE(ic->suspending) && !ic->meta_dev)
 		return;
 
 	spin_lock_irq(&ic->endio_wait.lock);
@@ -2235,6 +2235,8 @@ static void dm_integrity_postsuspend(str
 	drain_workqueue(ic->commit_wq);
 
 	if (ic->mode == 'J') {
+		if (ic->meta_dev)
+			queue_work(ic->writer_wq, &ic->writer_work);
 		drain_workqueue(ic->writer_wq);
 		dm_integrity_flush_buffers(ic);
 	}

[patch 8/8] dm-integrity: recalculate checksums

[Mikulas Patocka]

[-- Attachment #1: dm-integrity-recalc.patch --]
[-- Type: text/plain, Size: 8428 bytes --]

When using external metadata device and internal hash, recalculate the
checksums when the device is created - so that dm-integrity doesn't have
to overwrite the device. The superblock stores the last position when the
recalculation ended, so that it is properly restarted.

Integrity tags that haven't been recalculated yet are ignored.

This patch also increases the target version to 1.2.0.

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

---
 drivers/md/dm-integrity.c |  163 +++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 161 insertions(+), 2 deletions(-)

Index: linux-2.6/drivers/md/dm-integrity.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-integrity.c	2018-05-01 04:35:29.000000000 +0200
+++ linux-2.6/drivers/md/dm-integrity.c	2018-05-01 15:45:41.000000000 +0200
@@ -31,6 +31,8 @@
 #define MIN_LOG2_INTERLEAVE_SECTORS	3
 #define MAX_LOG2_INTERLEAVE_SECTORS	31
 #define METADATA_WORKQUEUE_MAX_ACTIVE	16
+#define RECALC_SECTORS			8192
+#define RECALC_WRITE_SUPER		16
 
 /*
  * Warning - DEBUG_PRINT prints security-sensitive data to the log,
@@ -58,9 +60,12 @@ struct superblock {
 	__u64 provided_data_sectors;	/* userspace uses this value */
 	__u32 flags;
 	__u8 log2_sectors_per_block;
+	__u8 pad[3];
+	__u64 recalc_sector;
 };
 
 #define SB_FLAG_HAVE_JOURNAL_MAC	0x1
+#define SB_FLAG_RECALCULATING		0x2
 
 #define	JOURNAL_ENTRY_ROUNDUP		8
 
@@ -213,6 +218,11 @@ struct dm_integrity_c {
 	struct workqueue_struct *writer_wq;
 	struct work_struct writer_work;
 
+	struct workqueue_struct *recalc_wq;
+	struct work_struct recalc_work;
+	u8 *recalc_buffer;
+	u8 *recalc_tags;
+
 	struct bio_list flush_bio_list;
 
 	unsigned long autocommit_jiffies;
@@ -409,7 +419,7 @@ static void wraparound_section(struct dm
 
 static void sb_set_version(struct dm_integrity_c *ic)
 {
-	if (ic->meta_dev)
+	if (ic->meta_dev || ic->sb->flags & cpu_to_le32(SB_FLAG_RECALCULATING))
 		ic->sb->version = SB_VERSION_2;
 	else
 		ic->sb->version = SB_VERSION_1;
@@ -1725,9 +1735,14 @@ sleep:
 
 	if (need_sync_io) {
 		wait_for_completion_io(&read_comp);
+		if (unlikely(ic->recalc_wq != NULL) &&
+		    ic->sb->flags & cpu_to_le32(SB_FLAG_RECALCULATING) &&
+		    dio->range.logical_sector + dio->range.n_sectors > le64_to_cpu(ic->sb->recalc_sector))
+			goto skip_check;
 		if (likely(!bio->bi_status))
 			integrity_metadata(&dio->work);
 		else
+skip_check:
 			dec_in_flight(dio);
 
 	} else {
@@ -2027,6 +2042,110 @@ static void integrity_writer(struct work
 	spin_unlock_irq(&ic->endio_wait.lock);
 }
 
+static void recalc_write_super(struct dm_integrity_c *ic)
+{
+	int r;
+
+	dm_integrity_flush_buffers(ic);
+	if (dm_integrity_failed(ic))
+		return;
+
+	sb_set_version(ic);
+	r = sync_rw_sb(ic, REQ_OP_WRITE, 0);
+	if (unlikely(r))
+		dm_integrity_io_error(ic, "writing superblock", r);
+}
+
+static void integrity_recalc(struct work_struct *w)
+{
+	struct dm_integrity_c *ic = container_of(w, struct dm_integrity_c, recalc_work);
+	struct dm_integrity_range range;
+	struct dm_io_request io_req;
+	struct dm_io_region io_loc;
+	sector_t area, offset;
+	sector_t metadata_block;
+	unsigned metadata_offset;
+	__u8 *t;
+	unsigned i;
+	int r;
+	unsigned super_counter = 0;
+
+	spin_lock_irq(&ic->endio_wait.lock);
+
+next_chunk:
+
+	if (unlikely(READ_ONCE(ic->suspending)))
+		goto unlock_ret;
+
+	range.logical_sector = le64_to_cpu(ic->sb->recalc_sector);
+	if (unlikely(range.logical_sector >= ic->provided_data_sectors)) {
+		ic->sb->flags &= ~cpu_to_le32(SB_FLAG_RECALCULATING);
+		ic->sb->recalc_sector = cpu_to_le64(0);
+		goto unlock_ret;
+	}
+	get_area_and_offset(ic, range.logical_sector, &area, &offset);
+	range.n_sectors = min((sector_t)RECALC_SECTORS, ic->provided_data_sectors - range.logical_sector);
+	if (!ic->meta_dev)
+		range.n_sectors = min(range.n_sectors, (1U << ic->sb->log2_interleave_sectors) - (unsigned)offset);
+
+	while (unlikely(!add_new_range(ic, &range)))
+		sleep_on_endio_wait(ic);
+
+	spin_unlock_irq(&ic->endio_wait.lock);
+
+	if (unlikely(++super_counter == RECALC_WRITE_SUPER)) {
+		recalc_write_super(ic);
+		super_counter = 0;
+	}
+
+	if (unlikely(dm_integrity_failed(ic)))
+		goto err;
+
+	io_req.bi_op = REQ_OP_READ;
+	io_req.bi_op_flags = 0;
+	io_req.mem.type = DM_IO_VMA;
+	io_req.mem.ptr.addr = ic->recalc_buffer;
+	io_req.notify.fn = NULL;
+	io_req.client = ic->io;
+	io_loc.bdev = ic->dev->bdev;
+	io_loc.sector = get_data_sector(ic, area, offset);
+	io_loc.count = range.n_sectors;
+
+	r = dm_io(&io_req, 1, &io_loc, NULL);
+	if (unlikely(r)) {
+		dm_integrity_io_error(ic, "reading data", r);
+		goto err;
+	}
+
+	t = ic->recalc_tags;
+	for (i = 0; i < range.n_sectors; i += ic->sectors_per_block) {
+		integrity_sector_checksum(ic, range.logical_sector + i, ic->recalc_buffer + (i << SECTOR_SHIFT), t);
+		t += ic->tag_size;
+	}
+
+	metadata_block = get_metadata_sector_and_offset(ic, area, offset, &metadata_offset);
+
+	r = dm_integrity_rw_tag(ic, ic->recalc_tags, &metadata_block, &metadata_offset, t - ic->recalc_tags, TAG_WRITE);
+	if (unlikely(r)) {
+		dm_integrity_io_error(ic, "writing tags", r);
+		goto err;
+	}
+
+	spin_lock_irq(&ic->endio_wait.lock);
+	remove_range_unlocked(ic, &range);
+	ic->sb->recalc_sector = cpu_to_le64(range.logical_sector + range.n_sectors);
+	goto next_chunk;
+
+err:
+	remove_range(ic, &range);
+	return;
+
+unlock_ret:
+	spin_unlock_irq(&ic->endio_wait.lock);
+
+	recalc_write_super(ic);
+}
+
 static void init_journal(struct dm_integrity_c *ic, unsigned start_section,
 			 unsigned n_sections, unsigned char commit_seq)
 {
@@ -2231,6 +2350,9 @@ static void dm_integrity_postsuspend(str
 
 	WRITE_ONCE(ic->suspending, 1);
 
+	if (ic->recalc_wq)
+		drain_workqueue(ic->recalc_wq);
+
 	queue_work(ic->commit_wq, &ic->commit_work);
 	drain_workqueue(ic->commit_wq);
 
@@ -2253,6 +2375,9 @@ static void dm_integrity_resume(struct d
 	struct dm_integrity_c *ic = (struct dm_integrity_c *)ti->private;
 
 	replay_journal(ic);
+
+	if (ic->recalc_wq && ic->sb->flags & cpu_to_le32(SB_FLAG_RECALCULATING))
+		queue_work(ic->recalc_wq, &ic->recalc_work);
 }
 
 static void dm_integrity_status(struct dm_target *ti, status_type_t type,
@@ -2420,6 +2545,11 @@ static int initialize_superblock(struct
 		if (!ic->provided_data_sectors)
 			return -EINVAL;
 	} else {
+		if (ic->internal_hash) {
+			ic->sb->flags |= cpu_to_le32(SB_FLAG_RECALCULATING);
+			ic->sb->recalc_sector = cpu_to_le64(0);
+		}
+
 		ic->sb->log2_interleave_sectors = 0;
 		ic->provided_data_sectors = ic->data_device_sectors;
 		ic->provided_data_sectors &= ~(sector_t)(ic->sectors_per_block - 1);
@@ -3237,6 +3367,29 @@ try_smaller_buffer:
 		    (unsigned long long)ic->provided_data_sectors);
 	DEBUG_print("	log2_buffer_sectors %u\n", ic->log2_buffer_sectors);
 
+
+	if (ic->sb->flags & cpu_to_le32(SB_FLAG_RECALCULATING)) {
+		ic->recalc_wq = alloc_workqueue("dm-intergrity-recalc", WQ_MEM_RECLAIM, 1);
+		if (!ic->recalc_wq ) {
+			ti->error = "Cannot allocate workqueue";
+			r = -ENOMEM;
+			goto bad;
+		}
+		INIT_WORK(&ic->recalc_work, integrity_recalc);
+		ic->recalc_buffer = vmalloc(RECALC_SECTORS << SECTOR_SHIFT);
+		if (!ic->recalc_buffer) {
+			ti->error = "Cannot allocate buffer for recalculating";
+			r = -ENOMEM;
+			goto bad;
+		}
+		ic->recalc_tags = kvmalloc((RECALC_SECTORS >> ic->sb->log2_sectors_per_block) * ic->tag_size, GFP_KERNEL);
+		if (!ic->recalc_tags) {
+			ti->error = "Cannot allocate tags for recalculating";
+			r = -ENOMEM;
+			goto bad;
+		}
+	}
+
 	ic->bufio = dm_bufio_client_create(ic->meta_dev ? ic->meta_dev->bdev : ic->dev->bdev,
 			1U << (SECTOR_SHIFT + ic->log2_buffer_sectors), 1, 0, NULL, NULL);
 	if (IS_ERR(ic->bufio)) {
@@ -3302,6 +3455,12 @@ static void dm_integrity_dtr(struct dm_t
 		destroy_workqueue(ic->commit_wq);
 	if (ic->writer_wq)
 		destroy_workqueue(ic->writer_wq);
+	if (ic->recalc_wq)
+		destroy_workqueue(ic->recalc_wq);
+	if (ic->recalc_buffer)
+		vfree(ic->recalc_buffer);
+	if (ic->recalc_tags)
+		kvfree(ic->recalc_tags);
 	if (ic->bufio)
 		dm_bufio_client_destroy(ic->bufio);
 	mempool_destroy(ic->journal_io_mempool);
@@ -3351,7 +3510,7 @@ static void dm_integrity_dtr(struct dm_t
 
 static struct target_type integrity_target = {
 	.name			= "integrity",
-	.version		= {1, 1, 0},
+	.version		= {1, 2, 0},
 	.module			= THIS_MODULE,
 	.features		= DM_TARGET_SINGLETON | DM_TARGET_INTEGRITY,
 	.ctr			= dm_integrity_ctr,