[PATCH 1/1] target: fix max_unmap_lba_count calc overflow

[PATCH 1/1] target: fix max_unmap_lba_count calc overflow

[mchristi]

From: Mike Christie <mchristi@redhat.com>

max_discard_sectors only 32bits, and some non scsi backend
devices will set this to the max 0xffffffff, so we can end up
overflowing during the max_unmap_lba_count calculation.

This fixes a regression caused by my patch:

commit 8a9ebe717a133ba7bc90b06047f43cc6b8bcb8b3
Author: Mike Christie <mchristi@redhat.com>
Date:   Mon Jan 18 14:09:27 2016 -0600

    target: Fix WRITE_SAME/DISCARD conversion to linux 512b sectors

which can result in extra discards being sent to due the overflow
causing max_unmap_lba_count to be smaller than what the backing
device can actually support.

Signed-off-by: Mike Christie <mchristi@redhat.com>
Cc: stable@vger.kernel.org

---
 drivers/target/target_core_device.c  | 8 +++++---
 drivers/target/target_core_file.c    | 3 +--
 drivers/target/target_core_iblock.c  | 3 +--
 include/target/target_core_backend.h | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index a4046ca..6b42348 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -821,13 +821,15 @@ struct se_device *target_alloc_device(struct se_hba *hba, const char *name)
  * in ATA and we need to set TPE=1
  */
 bool target_configure_unmap_from_queue(struct se_dev_attrib *attrib,
-				       struct request_queue *q, int block_size)
+				       struct request_queue *q)
 {
+	int block_size = queue_logical_block_size(q);
+
 	if (!blk_queue_discard(q))
 		return false;
 
-	attrib->max_unmap_lba_count = (q->limits.max_discard_sectors << 9) /
-								block_size;
+	attrib->max_unmap_lba_count =
+		q->limits.max_discard_sectors >> (ilog2(block_size) - 9);
 	/*
 	 * Currently hardcoded to 1 in Linux/SCSI code..
 	 */
diff --git a/drivers/target/target_core_file.c b/drivers/target/target_core_file.c
index 75f0f08..7929186 100644
--- a/drivers/target/target_core_file.c
+++ b/drivers/target/target_core_file.c
@@ -161,8 +161,7 @@ static int fd_configure_device(struct se_device *dev)
 			dev_size, div_u64(dev_size, fd_dev->fd_block_size),
 			fd_dev->fd_block_size);
 
-		if (target_configure_unmap_from_queue(&dev->dev_attrib, q,
-						      fd_dev->fd_block_size))
+		if (target_configure_unmap_from_queue(&dev->dev_attrib, q))
 			pr_debug("IFILE: BLOCK Discard support available,"
 				 " disabled by default\n");
 		/*
diff --git a/drivers/target/target_core_iblock.c b/drivers/target/target_core_iblock.c
index 7c4efb4..2077bc2 100644
--- a/drivers/target/target_core_iblock.c
+++ b/drivers/target/target_core_iblock.c
@@ -121,8 +121,7 @@ static int iblock_configure_device(struct se_device *dev)
 	dev->dev_attrib.hw_max_sectors = queue_max_hw_sectors(q);
 	dev->dev_attrib.hw_queue_depth = q->nr_requests;
 
-	if (target_configure_unmap_from_queue(&dev->dev_attrib, q,
-					      dev->dev_attrib.hw_block_size))
+	if (target_configure_unmap_from_queue(&dev->dev_attrib, q))
 		pr_debug("IBLOCK: BLOCK Discard support available,"
 			 " disabled by default\n");
 
diff --git a/include/target/target_core_backend.h b/include/target/target_core_backend.h
index d8ab510..f6f3bc5 100644
--- a/include/target/target_core_backend.h
+++ b/include/target/target_core_backend.h
@@ -95,6 +95,6 @@ sense_reason_t passthrough_parse_cdb(struct se_cmd *cmd,
 bool target_sense_desc_format(struct se_device *dev);
 sector_t target_to_linux_sector(struct se_device *dev, sector_t lb);
 bool target_configure_unmap_from_queue(struct se_dev_attrib *attrib,
-				       struct request_queue *q, int block_size);
+				       struct request_queue *q);
 
 #endif /* TARGET_CORE_BACKEND_H */
-- 
2.7.2

Re: [PATCH 1/1] target: fix max_unmap_lba_count calc overflow

[Bart Van Assche]

On 06/02/2016 06:12 PM, mchristi@redhat.com wrote:
> max_discard_sectors only 32bits, and some non scsi backend
> devices will set this to the max 0xffffffff, so we can end up
> overflowing during the max_unmap_lba_count calculation.

Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>

Re: [PATCH 1/1] target: fix max_unmap_lba_count calc overflow

[Nicholas A. Bellinger]

On Thu, 2016-06-02 at 20:12 -0500, mchristi@redhat.com wrote:
> From: Mike Christie <mchristi@redhat.com>
> 
> max_discard_sectors only 32bits, and some non scsi backend
> devices will set this to the max 0xffffffff, so we can end up
> overflowing during the max_unmap_lba_count calculation.
> 
> This fixes a regression caused by my patch:
> 
> commit 8a9ebe717a133ba7bc90b06047f43cc6b8bcb8b3
> Author: Mike Christie <mchristi@redhat.com>
> Date:   Mon Jan 18 14:09:27 2016 -0600
> 
>     target: Fix WRITE_SAME/DISCARD conversion to linux 512b sectors
> 
> which can result in extra discards being sent to due the overflow
> causing max_unmap_lba_count to be smaller than what the backing
> device can actually support.
> 
> Signed-off-by: Mike Christie <mchristi@redhat.com>
> Cc: stable@vger.kernel.org
> 

Applied to target-pending/master.

Thanks MNC + Bart.