On 04/11/2018 03:05 PM, Qu Wenruo wrote:

On 2018年04月11日 15:03, Su Yue wrote:
This tests is most similar to xfstests generic/405.
It calls device mapper to create a thin provision device with small
backing size and big virtual size. mkfs.btrfs should fail on such

This test should pass after commit e805b143a4fe
("btrfs-progs: mkfs: return nozero value on thin provisioned device").

Signed-off-by: Su Yue <suy.f...@cn.fujitsu.com>
  .../test.sh                                   | 93 +++++++++++++++++++
  1 file changed, 93 insertions(+)
  create mode 100755 

diff --git 
new file mode 100755
index 000000000000..f2e044da5d17
--- /dev/null
+++ b/tests/mkfs-tests/017-small-backing-size-thin-provision-device/test.sh
@@ -0,0 +1,93 @@
+# mkfs.btrfs must failed on a thin provision device with very small
+# backing size and big virtual size.
+source "$TEST_TOP/common"
+check_prereq mkfs.btrfs
+# Backing data dev
+# Backing metadata dev
+# Backing pool dev (combination of above)
+# Thin volume
+       # wait for device to be fully settled
+       run_check $SUDO_HELPER udevadm settle
+       run_check $SUDO_HELPER dmsetup remove $DMTHIN_VOL_NAME
+       run_check $SUDO_HELPER dmsetup remove $DMTHIN_POOL_NAME
+       run_check $SUDO_HELPER dmsetup remove $DMTHIN_META_NAME
+       run_check $SUDO_HELPER dmsetup remove $DMTHIN_DATA_NAME
+sector_size=512                  # in bytes
+data_dev_size=$((1 * 1024 * 1024 / $sector_size))   # 1M
+virtual_size=$((1 * 1024 * 1024 * 1024 * 1024 / $sector_size))  # 1T
+cluster_size=1024        # 512k in sectors
+low_water=$((104857600 / $cluster_size/ $sector_size))  # 100M / 
$cluster_size, in sectors
+# Need to make linear metadata and data devs.  From kernel docs:
+# As a guide, we suggest you calculate the number of bytes to use in the
+# metadata device as 48 * $data_dev_size / $data_block_size but round it up
+# to 2MB (4096 sectors) if the answer is smaller.
+# So do that:
+meta_dev_size=$((48 * $data_dev_size / $cluster_size))
+if [ "$meta_dev_size" -lt "4096" ]; then
+        meta_dev_size=4096      # 2MB
+total_data_dev_size=$(($meta_dev_offset + $meta_dev_size + $data_dev_size))
+run_check truncate -s0 img
+chmod a+w img
+run_check truncate -s"$(($total_data_dev_size * $sector_size))" img
+dm_backing_dev=`run_check_stdout $SUDO_HELPER losetup --find --show img`
+# Metadata device
+DMTHIN_META_TABLE="0 $meta_dev_size linear $dm_backing_dev $meta_dev_offset"
+run_check $SUDO_HELPER dmsetup create $DMTHIN_META_NAME --table 

What about using lvm other than plain dmsetup to create thin provision
IIRC this should be much easier to implement, without all the linear
target hassles.

Yes, doing it by LVM is more convenient.
However, LVM seems too abstract and it's never used in btrfs-progs.

I have no idea whether LVM other tools should be used or not.



+# Data device
+data_dev_offset=$((meta_dev_offset + $meta_dev_size))
+DMTHIN_DATA_TABLE="0 $data_dev_size linear $dm_backing_dev $data_dev_offset"
+run_check $SUDO_HELPER dmsetup create $DMTHIN_DATA_NAME --table 
+# Zap the pool metadata dev
+run_check dd if=/dev/zero of=$DMTHIN_META_DEV bs=4096 count=1
+# Thin pool
+# "start length thin-pool metadata_dev data_dev data_block_size low_water_mark"
$cluster_size $low_water"
+run_check $SUDO_HELPER dmsetup create $DMTHIN_POOL_NAME --table 
+# Thin volume
+run_check $SUDO_HELPER dmsetup message $DMTHIN_POOL_DEV 0 "create_thin 
+# start length thin pool_dev dev_id [external_origin_dev]
+DMTHIN_VOL_TABLE="0 $virtual_size thin $DMTHIN_POOL_DEV $pool_id"
+run_check $SUDO_HELPER dmsetup create $DMTHIN_VOL_NAME --table 
+# mkfs.btrfs should fail due to the small backing device
+run_mustfail "should fail for samll backing size thin provision device" \
+            $SUDO_HELPER "$TOP/mkfs.btrfs" -f "$@" "$DMTHIN_VOL_DEV"
+run_mayfail $SUDO_HELPER losetup -d $dm_backing_dev
+run_check truncate -s0 img
+rm img

To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to