block/sd: Fix device-imposed transfer length limits

Commit 4f258a4634 ("sd: Fix maximum I/O size for BLOCK_PC requests")
had the unfortunate side-effect of removing an implicit clamp to
BLK_DEF_MAX_SECTORS for REQ_TYPE_FS requests in the block layer
code. This caused problems for some SMR drives.

Debugging this issue revealed a few problems with the existing
infrastructure since the block layer didn't know how to deal with
device-imposed limits, only limits set by the I/O controller.

 - Introduce a new queue limit, max_dev_sectors, which is used by the
   ULD to signal the maximum sectors for a REQ_TYPE_FS request.

 - Ensure that max_dev_sectors is correctly stacked and taken into
   account when overriding max_sectors through sysfs.

 - Rework sd_read_block_limits() so it saves the max_xfer and opt_xfer
   values for later processing.

 - In sd_revalidate() set the queue's max_dev_sectors based on the
   MAXIMUM TRANSFER LENGTH value in the Block Limits VPD. If this value
   is not reported, fall back to a cap based on the CDB TRANSFER LENGTH
   field size.

 - In sd_revalidate(), use OPTIMAL TRANSFER LENGTH from the Block Limits
   VPD--if reported and sane--to signal the preferred device transfer
   size for FS requests. Otherwise use BLK_DEF_MAX_SECTORS.

 - blk_limits_max_hw_sectors() is no longer used and can be removed.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=93581
Reviewed-by: Christoph Hellwig <hch@lst.de>
Tested-by: sweeneygj@gmx.com
Tested-by: Arzeets <anatol.pomozov@gmail.com>
Tested-by: David Eisner <david.eisner@oriel.oxon.org>
Tested-by: Mario Kicherer <dev@kicherer.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

block/blk-settings.c

@@ -91,7 +91,8 @@ void blk_set_default_limits(struct queue_limits *lim)
 	lim->seg_boundary_mask = BLK_SEG_BOUNDARY_MASK;
 	lim->virt_boundary_mask = 0;
 	lim->max_segment_size = BLK_MAX_SEGMENT_SIZE;
-	lim->max_sectors = lim->max_hw_sectors = BLK_SAFE_MAX_SECTORS;
+	lim->max_sectors = lim->max_dev_sectors = lim->max_hw_sectors =
+		BLK_SAFE_MAX_SECTORS;
 	lim->chunk_sectors = 0;
 	lim->max_write_same_sectors = 0;
 	lim->max_discard_sectors = 0;
@@ -127,6 +128,7 @@ void blk_set_stacking_limits(struct queue_limits *lim)
 	lim->max_hw_sectors = UINT_MAX;
 	lim->max_segment_size = UINT_MAX;
 	lim->max_sectors = UINT_MAX;
+	lim->max_dev_sectors = UINT_MAX;
 	lim->max_write_same_sectors = UINT_MAX;
 }
 EXPORT_SYMBOL(blk_set_stacking_limits);
@@ -214,8 +216,8 @@ void blk_queue_bounce_limit(struct request_queue *q, u64 max_addr)
 EXPORT_SYMBOL(blk_queue_bounce_limit);
 
 /**
- * blk_limits_max_hw_sectors - set hard and soft limit of max sectors for request
- * @limits: the queue limits
+ * blk_queue_max_hw_sectors - set max sectors for a request for this queue
+ * @q:  the request queue for the device
  * @max_hw_sectors:  max hardware sectors in the usual 512b unit
  *
  * Description:
@@ -224,13 +226,19 @@ EXPORT_SYMBOL(blk_queue_bounce_limit);
  *    the device driver based upon the capabilities of the I/O
  *    controller.
  *
+ *    max_dev_sectors is a hard limit imposed by the storage device for
+ *    READ/WRITE requests. It is set by the disk driver.
+ *
  *    max_sectors is a soft limit imposed by the block layer for
  *    filesystem type requests.  This value can be overridden on a
  *    per-device basis in /sys/block/<device>/queue/max_sectors_kb.
  *    The soft limit can not exceed max_hw_sectors.
  **/
-void blk_limits_max_hw_sectors(struct queue_limits *limits, unsigned int max_hw_sectors)
+void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_hw_sectors)
 {
+	struct queue_limits *limits = &q->limits;
+	unsigned int max_sectors;
+
 	if ((max_hw_sectors << 9) < PAGE_CACHE_SIZE) {
 		max_hw_sectors = 1 << (PAGE_CACHE_SHIFT - 9);
 		printk(KERN_INFO "%s: set to minimum %d\n",
@@ -238,22 +246,9 @@ void blk_limits_max_hw_sectors(struct queue_limits *limits, unsigned int max_hw_
 	}
 
 	limits->max_hw_sectors = max_hw_sectors;
-	limits->max_sectors = min_t(unsigned int, max_hw_sectors,
-				    BLK_DEF_MAX_SECTORS);
-}
-EXPORT_SYMBOL(blk_limits_max_hw_sectors);
-
-/**
- * blk_queue_max_hw_sectors - set max sectors for a request for this queue
- * @q:  the request queue for the device
- * @max_hw_sectors:  max hardware sectors in the usual 512b unit
- *
- * Description:
- *    See description for blk_limits_max_hw_sectors().
- **/
-void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_hw_sectors)
-{
-	blk_limits_max_hw_sectors(&q->limits, max_hw_sectors);
+	max_sectors = min_not_zero(max_hw_sectors, limits->max_dev_sectors);
+	max_sectors = min_t(unsigned int, max_sectors, BLK_DEF_MAX_SECTORS);
+	limits->max_sectors = max_sectors;
 }
 EXPORT_SYMBOL(blk_queue_max_hw_sectors);
 
@@ -527,6 +522,7 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
 
 	t->max_sectors = min_not_zero(t->max_sectors, b->max_sectors);
 	t->max_hw_sectors = min_not_zero(t->max_hw_sectors, b->max_hw_sectors);
+	t->max_dev_sectors = min_not_zero(t->max_dev_sectors, b->max_dev_sectors);
 	t->max_write_same_sectors = min(t->max_write_same_sectors,
 					b->max_write_same_sectors);
 	t->bounce_pfn = min_not_zero(t->bounce_pfn, b->bounce_pfn);

block/blk-sysfs.c

@@ -205,6 +205,9 @@ queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
 	if (ret < 0)
 		return ret;
 
+	max_hw_sectors_kb = min_not_zero(max_hw_sectors_kb, (unsigned long)
+					 q->limits.max_dev_sectors >> 1);
+
 	if (max_sectors_kb > max_hw_sectors_kb || max_sectors_kb < page_kb)
 		return -EINVAL;
 

drivers/scsi/sd.c

@@ -2238,11 +2238,8 @@ got_data:
 		}
 	}
 
-	if (sdkp->capacity > 0xffffffff) {
+	if (sdkp->capacity > 0xffffffff)
 		sdp->use_16_for_rw = 1;
-		sdkp->max_xfer_blocks = SD_MAX_XFER_BLOCKS;
-	} else
-		sdkp->max_xfer_blocks = SD_DEF_XFER_BLOCKS;
 
 	/* Rescale capacity to 512-byte units */
 	if (sector_size == 4096)
@@ -2559,7 +2556,6 @@ static void sd_read_block_limits(struct scsi_disk *sdkp)
 {
 	unsigned int sector_sz = sdkp->device->sector_size;
 	const int vpd_len = 64;
-	u32 max_xfer_length;
 	unsigned char *buffer = kmalloc(vpd_len, GFP_KERNEL);
 
 	if (!buffer ||
@@ -2567,14 +2563,11 @@ static void sd_read_block_limits(struct scsi_disk *sdkp)
 	    scsi_get_vpd_page(sdkp->device, 0xb0, buffer, vpd_len))
 		goto out;
 
-	max_xfer_length = get_unaligned_be32(&buffer[8]);
-	if (max_xfer_length)
-		sdkp->max_xfer_blocks = max_xfer_length;
-
 	blk_queue_io_min(sdkp->disk->queue,
 			 get_unaligned_be16(&buffer[6]) * sector_sz);
-	blk_queue_io_opt(sdkp->disk->queue,
-			 get_unaligned_be32(&buffer[12]) * sector_sz);
+
+	sdkp->max_xfer_blocks = get_unaligned_be32(&buffer[8]);
+	sdkp->opt_xfer_blocks = get_unaligned_be32(&buffer[12]);
 
 	if (buffer[3] == 0x3c) {
 		unsigned int lba_count, desc_count;
@@ -2723,6 +2716,11 @@ static int sd_try_extended_inquiry(struct scsi_device *sdp)
 	return 0;
 }
 
+static inline u32 logical_to_sectors(struct scsi_device *sdev, u32 blocks)
+{
+	return blocks << (ilog2(sdev->sector_size) - 9);
+}
+
 /**
  *	sd_revalidate_disk - called the first time a new disk is seen,
  *	performs disk spin up, read_capacity, etc.
@@ -2732,8 +2730,9 @@ static int sd_revalidate_disk(struct gendisk *disk)
 {
 	struct scsi_disk *sdkp = scsi_disk(disk);
 	struct scsi_device *sdp = sdkp->device;
+	struct request_queue *q = sdkp->disk->queue;
 	unsigned char *buffer;
-	unsigned int max_xfer;
+	unsigned int dev_max, rw_max;
 
 	SCSI_LOG_HLQUEUE(3, sd_printk(KERN_INFO, sdkp,
 				      "sd_revalidate_disk\n"));
@@ -2781,11 +2780,26 @@ static int sd_revalidate_disk(struct gendisk *disk)
 	 */
 	sd_set_flush_flag(sdkp);
 
-	max_xfer = sdkp->max_xfer_blocks;
-	max_xfer <<= ilog2(sdp->sector_size) - 9;
+	/* Initial block count limit based on CDB TRANSFER LENGTH field size. */
+	dev_max = sdp->use_16_for_rw ? SD_MAX_XFER_BLOCKS : SD_DEF_XFER_BLOCKS;
 
-	sdkp->disk->queue->limits.max_sectors =
-		min_not_zero(queue_max_hw_sectors(sdkp->disk->queue), max_xfer);
+	/* Some devices report a maximum block count for READ/WRITE requests. */
+	dev_max = min_not_zero(dev_max, sdkp->max_xfer_blocks);
+	q->limits.max_dev_sectors = logical_to_sectors(sdp, dev_max);
+
+	/*
+	 * Use the device's preferred I/O size for reads and writes
+	 * unless the reported value is unreasonably large (or garbage).
+	 */
+	if (sdkp->opt_xfer_blocks && sdkp->opt_xfer_blocks <= dev_max &&
+	    sdkp->opt_xfer_blocks <= SD_DEF_XFER_BLOCKS)
+		rw_max = q->limits.io_opt =
+			logical_to_sectors(sdp, sdkp->opt_xfer_blocks);
+	else
+		rw_max = BLK_DEF_MAX_SECTORS;
+
+	/* Combine with controller limits */
+	q->limits.max_sectors = min(rw_max, queue_max_hw_sectors(q));
 
 	set_capacity(disk, sdkp->capacity);
 	sd_config_write_same(sdkp);

drivers/scsi/sd.h

@@ -67,6 +67,7 @@ struct scsi_disk {
 	atomic_t	openers;
 	sector_t	capacity;	/* size in 512-byte sectors */
 	u32		max_xfer_blocks;
+	u32		opt_xfer_blocks;
 	u32		max_ws_blocks;
 	u32		max_unmap_blocks;
 	u32		unmap_granularity;

include/linux/blkdev.h

@@ -253,6 +253,7 @@ struct queue_limits {
 	unsigned long		virt_boundary_mask;
 
 	unsigned int		max_hw_sectors;
+	unsigned int		max_dev_sectors;
 	unsigned int		chunk_sectors;
 	unsigned int		max_sectors;
 	unsigned int		max_segment_size;
@@ -948,7 +949,6 @@ extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
 extern void blk_cleanup_queue(struct request_queue *);
 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
 extern void blk_queue_bounce_limit(struct request_queue *, u64);
-extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
 extern void blk_queue_max_segments(struct request_queue *, unsigned short);