block: Allow devices to indicate whether discarded blocks are zeroed

The discard ioctl is used by mkfs utilities to clear a block device prior to putting metadata down. However, not all devices return zeroed blocks after a discard. Some drives return stale data, potentially containing old superblocks. It is therefore important to know whether discarded blocks are properly zeroed. Both ATA and SCSI drives have configuration bits that indicate whether zeroes are returned after a discard operation. Implement a block level interface that allows this information to be bubbled up the stack and queried via a new block device ioctl. Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

block: Allow devices to indicate whether discarded blocks are zeroed
The discard ioctl is used by mkfs utilities to clear a block device prior to putting metadata down. However, not all devices return zeroed blocks after a discard. Some drives return stale data, potentially containing old superblocks. It is therefore important to know whether discarded blocks are properly zeroed. Both ATA and SCSI drives have configuration bits that indicate whether zeroes are returned after a discard operation. Implement a block level interface that allows this information to be bubbled up the stack and queried via a new block device ioctl. Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Martin K. Petersen · Jens Axboe
1 parent 464191c65b
Showing 6 changed files with 32 additions and 0 deletions Side-by-side Diff
block/blk-settings.c
block/blk-sysfs.c
block/compat_ioctl.c
block/ioctl.c
include/linux/blkdev.h
include/linux/fs.h
@@ -101,6 +101,7 @@
 	lim->discard_granularity = 0;
 	lim->discard_alignment = 0;
 	lim->discard_misaligned = 0;
+	lim->discard_zeroes_data = -1;
 	lim->logical_block_size = lim->physical_block_size = lim->io_min = 512;
 	lim->bounce_pfn = (unsigned long)(BLK_BOUNCE_ANY >> PAGE_SHIFT);
 	lim->alignment_offset = 0;
@@ -544,6 +545,7 @@
  
 	t->io_min = max(t->io_min, b->io_min);
 	t->no_cluster |= b->no_cluster;
+	t->discard_zeroes_data &= b->discard_zeroes_data;
  
 	/* Bottom device offset aligned? */
 	if (offset &&
@@ -136,6 +136,11 @@
 	return queue_var_show(q->limits.max_discard_sectors << 9, page);
 }
  
+static ssize_t queue_discard_zeroes_data_show(struct request_queue *q, char *page)
+{
+	return queue_var_show(queue_discard_zeroes_data(q), page);
+}
+
 static ssize_t
 queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
 {
@@ -313,6 +318,11 @@
 	.show = queue_discard_max_show,
 };
  
+static struct queue_sysfs_entry queue_discard_zeroes_data_entry = {
+	.attr = {.name = "discard_zeroes_data", .mode = S_IRUGO },
+	.show = queue_discard_zeroes_data_show,
+};
+
 static struct queue_sysfs_entry queue_nonrot_entry = {
 	.attr = {.name = "rotational", .mode = S_IRUGO | S_IWUSR },
 	.show = queue_nonrot_show,
@@ -350,6 +360,7 @@
 	&queue_io_opt_entry.attr,
 	&queue_discard_granularity_entry.attr,
 	&queue_discard_max_entry.attr,
+	&queue_discard_zeroes_data_entry.attr,
 	&queue_nonrot_entry.attr,
 	&queue_nomerges_entry.attr,
 	&queue_rq_affinity_entry.attr,
@@ -747,6 +747,8 @@
 		return compat_put_uint(arg, bdev_io_opt(bdev));
 	case BLKALIGNOFF:
 		return compat_put_int(arg, bdev_alignment_offset(bdev));
+	case BLKDISCARDZEROES:
+		return compat_put_uint(arg, bdev_discard_zeroes_data(bdev));
 	case BLKFLSBUF:
 	case BLKROSET:
 	case BLKDISCARD:
@@ -280,6 +280,8 @@
 		return put_uint(arg, bdev_io_opt(bdev));
 	case BLKALIGNOFF:
 		return put_int(arg, bdev_alignment_offset(bdev));
+	case BLKDISCARDZEROES:
+		return put_uint(arg, bdev_discard_zeroes_data(bdev));
 	case BLKSECTGET:
 		return put_ushort(arg, queue_max_sectors(bdev_get_queue(bdev)));
 	case BLKRASET:
@@ -322,6 +322,7 @@
 	unsigned char		misaligned;
 	unsigned char		discard_misaligned;
 	unsigned char		no_cluster;
+	signed char		discard_zeroes_data;
 };
  
 struct request_queue
@@ -1148,6 +1149,19 @@
 {
 	return ((sector << 9) - q->limits.discard_alignment)
 		& (q->limits.discard_granularity - 1);
+}
+
+static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
+{
+	if (q->limits.discard_zeroes_data == 1)
+		return 1;
+
+	return 0;
+}
+
+static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
+{
+	return queue_discard_zeroes_data(bdev_get_queue(bdev));
 }
  
 static inline int queue_dma_alignment(struct request_queue *q)
@@ -304,6 +304,7 @@
 #define BLKIOOPT _IO(0x12,121)
 #define BLKALIGNOFF _IO(0x12,122)
 #define BLKPBSZGET _IO(0x12,123)
+#define BLKDISCARDZEROES _IO(0x12,124)
  
 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
...	...	@@ -101,6 +101,7 @@
101	101	lim->discard_granularity = 0;
102	102	lim->discard_alignment = 0;
103	103	lim->discard_misaligned = 0;
	104	+ lim->discard_zeroes_data = -1;
104	105	lim->logical_block_size = lim->physical_block_size = lim->io_min = 512;
105	106	lim->bounce_pfn = (unsigned long)(BLK_BOUNCE_ANY >> PAGE_SHIFT);
106	107	lim->alignment_offset = 0;
...	...	@@ -544,6 +545,7 @@
544	545
545	546	t->io_min = max(t->io_min, b->io_min);
546	547	t->no_cluster \|= b->no_cluster;
	548	+ t->discard_zeroes_data &= b->discard_zeroes_data;
547	549
548	550	/* Bottom device offset aligned? */
549	551	if (offset &&
...	...	@@ -136,6 +136,11 @@
136	136	return queue_var_show(q->limits.max_discard_sectors << 9, page);
137	137	}
138	138
	139	+static ssize_t queue_discard_zeroes_data_show(struct request_queue q, char page)
	140	+{
	141	+ return queue_var_show(queue_discard_zeroes_data(q), page);
	142	+}
	143	+
139	144	static ssize_t
140	145	queue_max_sectors_store(struct request_queue q, const char page, size_t count)
141	146	{
...	...	@@ -313,6 +318,11 @@
313	318	.show = queue_discard_max_show,
314	319	};
315	320
	321	+static struct queue_sysfs_entry queue_discard_zeroes_data_entry = {
	322	+ .attr = {.name = "discard_zeroes_data", .mode = S_IRUGO },
	323	+ .show = queue_discard_zeroes_data_show,
	324	+};
	325	+
316	326	static struct queue_sysfs_entry queue_nonrot_entry = {
317	327	.attr = {.name = "rotational", .mode = S_IRUGO \| S_IWUSR },
318	328	.show = queue_nonrot_show,
...	...	@@ -350,6 +360,7 @@
350	360	&queue_io_opt_entry.attr,
351	361	&queue_discard_granularity_entry.attr,
352	362	&queue_discard_max_entry.attr,
	363	+ &queue_discard_zeroes_data_entry.attr,
353	364	&queue_nonrot_entry.attr,
354	365	&queue_nomerges_entry.attr,
355	366	&queue_rq_affinity_entry.attr,
...	...	@@ -747,6 +747,8 @@
747	747	return compat_put_uint(arg, bdev_io_opt(bdev));
748	748	case BLKALIGNOFF:
749	749	return compat_put_int(arg, bdev_alignment_offset(bdev));
	750	+ case BLKDISCARDZEROES:
	751	+ return compat_put_uint(arg, bdev_discard_zeroes_data(bdev));
750	752	case BLKFLSBUF:
751	753	case BLKROSET:
752	754	case BLKDISCARD:
...	...	@@ -280,6 +280,8 @@
280	280	return put_uint(arg, bdev_io_opt(bdev));
281	281	case BLKALIGNOFF:
282	282	return put_int(arg, bdev_alignment_offset(bdev));
	283	+ case BLKDISCARDZEROES:
	284	+ return put_uint(arg, bdev_discard_zeroes_data(bdev));
283	285	case BLKSECTGET:
284	286	return put_ushort(arg, queue_max_sectors(bdev_get_queue(bdev)));
285	287	case BLKRASET:
...	...	@@ -322,6 +322,7 @@
322	322	unsigned char misaligned;
323	323	unsigned char discard_misaligned;
324	324	unsigned char no_cluster;
	325	+ signed char discard_zeroes_data;
325	326	};
326	327
327	328	struct request_queue
...	...	@@ -1148,6 +1149,19 @@
1148	1149	{
1149	1150	return ((sector << 9) - q->limits.discard_alignment)
1150	1151	& (q->limits.discard_granularity - 1);
	1152	+}
	1153	+
	1154	+static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
	1155	+{
	1156	+ if (q->limits.discard_zeroes_data == 1)
	1157	+ return 1;
	1158	+
	1159	+ return 0;
	1160	+}
	1161	+
	1162	+static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
	1163	+{
	1164	+ return queue_discard_zeroes_data(bdev_get_queue(bdev));
1151	1165	}
1152	1166
1153	1167	static inline int queue_dma_alignment(struct request_queue *q)
...	...	@@ -304,6 +304,7 @@
304	304	#define BLKIOOPT _IO(0x12,121)
305	305	#define BLKALIGNOFF _IO(0x12,122)
306	306	#define BLKPBSZGET _IO(0x12,123)
	307	+#define BLKDISCARDZEROES _IO(0x12,124)
307	308
308	309	#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
309	310	#define FIBMAP _IO(0x00,1) /* bmap access */