diff mbox

block: Check partition alignment on zoned block devices

Message ID 1480494139-1658-1-git-send-email-damien.lemoal@wdc.com (mailing list archive)
State New, archived
Headers show

Commit Message

Damien Le Moal Nov. 30, 2016, 8:22 a.m. UTC
Both blkdev_report_zones and blkdev_reset_zones can operate on a partition of
a zoned block device. However, the first and last zones reported for a
partition make sense only if the partition start sector and size are aligned
on the device zone size. The same applies for zone reset. Resetting the first
or the last zone of a partition straddling zones may impact neighboring
partitions. Finally, if a partition start sector is not at the beginning of a
sequential zone, it will be impossible to write to the first sectors of the
partition on a host-managed device.
Avoid all these problems and incoherencies by ignoring partitions that are not
zone aligned.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
---
 block/partition-generic.c | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

Comments

Jens Axboe Nov. 30, 2016, 3:20 p.m. UTC | #1
On 11/30/2016 01:22 AM, Damien Le Moal wrote:
> Both blkdev_report_zones and blkdev_reset_zones can operate on a partition of
> a zoned block device. However, the first and last zones reported for a
> partition make sense only if the partition start sector and size are aligned
> on the device zone size. The same applies for zone reset. Resetting the first
> or the last zone of a partition straddling zones may impact neighboring
> partitions. Finally, if a partition start sector is not at the beginning of a
> sequential zone, it will be impossible to write to the first sectors of the
> partition on a host-managed device.
> Avoid all these problems and incoherencies by ignoring partitions that are not
> zone aligned.

Let's try and keep the ZONED ifdefs to a minimum, I absolutely hate
having those sprinkled throughout the code. Less of an issue in the
headers.

bdev_is_zoned() is safe to call without the ifdef, and let's kill it
around the part_zone_aligned() function too. There's no need for either
one of the ifdefs below, as far as I can tell.
diff mbox

Patch

diff --git a/block/partition-generic.c b/block/partition-generic.c
index 71d9ed9..c5914b2 100644
--- a/block/partition-generic.c
+++ b/block/partition-generic.c
@@ -430,6 +430,29 @@  static int drop_partitions(struct gendisk *disk, struct block_device *bdev)
 	return 0;
 }
 
+#ifdef CONFIG_BLK_DEV_ZONED
+static bool part_zone_aligned(struct gendisk *disk,
+			      struct block_device *bdev,
+			      sector_t from, sector_t size)
+{
+	sector_t zone_size = bdev_zone_size(bdev);
+
+	/* Check partition start alignement */
+	if (from & (zone_size - 1))
+		return false;
+
+	/*
+	 * Check partition end alignement.
+	 * Ignore eventual last smaller zone.
+	 */
+	if ((from + size) < get_capacity(disk) &&
+	    (size & (zone_size - 1)))
+		return false;
+
+	return true;
+}
+#endif
+
 int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
 {
 	struct parsed_partitions *state = NULL;
@@ -529,6 +552,22 @@  int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
 			}
 		}
 
+#ifdef CONFIG_BLK_DEV_ZONED
+		/*
+		 * On a zoned block device, partitions should be aligned on the
+		 * device zone size (i.e. zone boundary crossing not allowed).
+		 * Otherwise, resetting the write pointer of the last zone of
+		 * one partition may impact the following partition.
+		 */
+		if (bdev_is_zoned(bdev) &&
+		    !part_zone_aligned(disk, bdev, from, size)) {
+			printk(KERN_WARNING
+			       "%s: p%d start %llu is not zone unaligned\n",
+			       disk->disk_name, p, (unsigned long long) from);
+			continue;
+		}
+#endif
+
 		part = add_partition(disk, p, from, size,
 				     state->parts[p].flags,
 				     &state->parts[p].info);