diff mbox series

fstests: btrfs: Verify falloc on multiple holes won't cause qgroup reserved data space leak

Message ID 20190913015151.15076-1-wqu@suse.com (mailing list archive)
State New, archived
Headers show
Series fstests: btrfs: Verify falloc on multiple holes won't cause qgroup reserved data space leak | expand

Commit Message

Qu Wenruo Sept. 13, 2019, 1:51 a.m. UTC 
Add a test case where falloc is called on multiple holes with qgroup
enabled.

This can cause qgroup reserved data space leak and false EDQUOT error
even we're not reaching the limit.

The fix is titled:
"btrfs: qgroup: Fix the wrong target io_tree when freeing
 reserved data space"

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 tests/btrfs/192     | 72 +++++++++++++++++++++++++++++++++++++++++++++
 tests/btrfs/192.out | 18 ++++++++++++
 tests/btrfs/group   |  1 +
 3 files changed, 91 insertions(+)
 create mode 100755 tests/btrfs/192
 create mode 100644 tests/btrfs/192.out

Comments

Josef Bacik Sept. 13, 2019, 4:12 p.m. UTC | #1 
On Fri, Sep 13, 2019 at 09:51:51AM +0800, Qu Wenruo wrote:
> Add a test case where falloc is called on multiple holes with qgroup
> enabled.
> 
> This can cause qgroup reserved data space leak and false EDQUOT error
> even we're not reaching the limit.
> 
> The fix is titled:
> "btrfs: qgroup: Fix the wrong target io_tree when freeing
>  reserved data space"
> 
> Signed-off-by: Qu Wenruo <wqu@suse.com>

Reviewed-by: Josef Bacik <josef@toxicpanda.com>

Thanks,

Josef
Eryu Guan Sept. 15, 2019, 4:36 a.m. UTC | #2 
On Fri, Sep 13, 2019 at 09:51:51AM +0800, Qu Wenruo wrote:
> Add a test case where falloc is called on multiple holes with qgroup
> enabled.
> 
> This can cause qgroup reserved data space leak and false EDQUOT error
> even we're not reaching the limit.
> 
> The fix is titled:
> "btrfs: qgroup: Fix the wrong target io_tree when freeing
>  reserved data space"
> 
> Signed-off-by: Qu Wenruo <wqu@suse.com>
> ---
>  tests/btrfs/192     | 72 +++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/192.out | 18 ++++++++++++
>  tests/btrfs/group   |  1 +
>  3 files changed, 91 insertions(+)
>  create mode 100755 tests/btrfs/192
>  create mode 100644 tests/btrfs/192.out
> 
> diff --git a/tests/btrfs/192 b/tests/btrfs/192
> new file mode 100755
> index 00000000..361b6d92
> --- /dev/null
> +++ b/tests/btrfs/192
> @@ -0,0 +1,72 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (C) 2019 SUSE Linux Products GmbH. All Rights Reserved.
> +#
> +# FS QA Test 192
> +#
> +# Test if btrfs is going to leak qgroup reserved data space when
> +# falloc on multiple holes fails.
> +# The fix is titled:
> +# "btrfs: qgroup: Fix the wrong target io_tree when freeing reserved data space"
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> +	cd /
> +	rm -f $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +# Modify as appropriate.
> +_supported_fs btrfs
> +_supported_os Linux
> +_require_scratch
> +_require_xfs_io_command falloc
> +
> +_scratch_mkfs > /dev/null
> +_scratch_mount
> +
> +$BTRFS_UTIL_PROG quota enable "$SCRATCH_MNT" > /dev/null
> +$BTRFS_UTIL_PROG quota rescan -w "$SCRATCH_MNT" > /dev/null
> +$BTRFS_UTIL_PROG qgroup limit -e 256M "$SCRATCH_MNT"
> +
> +for i in $(seq 3); do

Why do we need to loop 3 times? Some comments would be good, or we could
just remove the loop?

Other than that the test looks fine to me.

Thanks,
Eryu

> +	# Create a file with the following layout:
> +	# 0         128M      256M      384M
> +	# |  Hole   |4K| Hole |4K| Hole |
> +	# The total hole size will be 384M - 8k
> +	truncate -s 384m "$SCRATCH_MNT/file"
> +	$XFS_IO_PROG -c "pwrite 128m 4k" -c "pwrite 256m 4k" \
> +		"$SCRATCH_MNT/file" | _filter_xfs_io
> +
> +	# Falloc 0~384M range, it's going to fail due to the qgroup limit
> +	$XFS_IO_PROG -c "falloc 0 384m" "$SCRATCH_MNT/file" |\
> +		_filter_xfs_io_error
> +	rm "$SCRATCH_MNT/file"
> +
> +	# Ensure above delete reaches disk and free some space
> +	sync
> +done
> +
> +# We should be able to write at least 3/4 of the limit
> +$XFS_IO_PROG -f -c "pwrite 0 192m" "$SCRATCH_MNT/file" | _filter_xfs_io
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/btrfs/192.out b/tests/btrfs/192.out
> new file mode 100644
> index 00000000..13bc6036
> --- /dev/null
> +++ b/tests/btrfs/192.out
> @@ -0,0 +1,18 @@
> +QA output created by 192
> +wrote 4096/4096 bytes at offset 134217728
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +wrote 4096/4096 bytes at offset 268435456
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +fallocate: Disk quota exceeded
> +wrote 4096/4096 bytes at offset 134217728
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +wrote 4096/4096 bytes at offset 268435456
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +fallocate: Disk quota exceeded
> +wrote 4096/4096 bytes at offset 134217728
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +wrote 4096/4096 bytes at offset 268435456
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +fallocate: Disk quota exceeded
> +wrote 201326592/201326592 bytes at offset 0
> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index 2474d43e..160fe927 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -194,3 +194,4 @@
>  189 auto quick send clone
>  190 auto quick replay balance qgroup
>  191 auto quick send dedupe
> +192 auto qgroup fast enospc limit
> -- 
> 2.22.0
>
Qu Wenruo Sept. 15, 2019, 7:18 a.m. UTC | #3 
On 2019/9/15 下午12:36, Eryu Guan wrote:
> On Fri, Sep 13, 2019 at 09:51:51AM +0800, Qu Wenruo wrote:
>> Add a test case where falloc is called on multiple holes with qgroup
>> enabled.
>>
>> This can cause qgroup reserved data space leak and false EDQUOT error
>> even we're not reaching the limit.
>>
>> The fix is titled:
>> "btrfs: qgroup: Fix the wrong target io_tree when freeing
>>  reserved data space"
>>
>> Signed-off-by: Qu Wenruo <wqu@suse.com>
>> ---
>>  tests/btrfs/192     | 72 +++++++++++++++++++++++++++++++++++++++++++++
>>  tests/btrfs/192.out | 18 ++++++++++++
>>  tests/btrfs/group   |  1 +
>>  3 files changed, 91 insertions(+)
>>  create mode 100755 tests/btrfs/192
>>  create mode 100644 tests/btrfs/192.out
>>
>> diff --git a/tests/btrfs/192 b/tests/btrfs/192
>> new file mode 100755
>> index 00000000..361b6d92
>> --- /dev/null
>> +++ b/tests/btrfs/192
>> @@ -0,0 +1,72 @@
>> +#! /bin/bash
>> +# SPDX-License-Identifier: GPL-2.0
>> +# Copyright (C) 2019 SUSE Linux Products GmbH. All Rights Reserved.
>> +#
>> +# FS QA Test 192
>> +#
>> +# Test if btrfs is going to leak qgroup reserved data space when
>> +# falloc on multiple holes fails.
>> +# The fix is titled:
>> +# "btrfs: qgroup: Fix the wrong target io_tree when freeing reserved data space"
>> +#
>> +seq=`basename $0`
>> +seqres=$RESULT_DIR/$seq
>> +echo "QA output created by $seq"
>> +
>> +here=`pwd`
>> +tmp=/tmp/$$
>> +status=1	# failure is the default!
>> +trap "_cleanup; exit \$status" 0 1 2 3 15
>> +
>> +_cleanup()
>> +{
>> +	cd /
>> +	rm -f $tmp.*
>> +}
>> +
>> +# get standard environment, filters and checks
>> +. ./common/rc
>> +. ./common/filter
>> +
>> +# remove previous $seqres.full before test
>> +rm -f $seqres.full
>> +
>> +# real QA test starts here
>> +
>> +# Modify as appropriate.
>> +_supported_fs btrfs
>> +_supported_os Linux
>> +_require_scratch
>> +_require_xfs_io_command falloc
>> +
>> +_scratch_mkfs > /dev/null
>> +_scratch_mount
>> +
>> +$BTRFS_UTIL_PROG quota enable "$SCRATCH_MNT" > /dev/null
>> +$BTRFS_UTIL_PROG quota rescan -w "$SCRATCH_MNT" > /dev/null
>> +$BTRFS_UTIL_PROG qgroup limit -e 256M "$SCRATCH_MNT"
>> +
>> +for i in $(seq 3); do
> 
> Why do we need to loop 3 times? Some comments would be good, or we could
> just remove the loop?

One loop itself will leak around 128M (the first hole can be fulfilled
but the next whole will hit qgroup limit).

The 3 loop is just to make sure we can exhaust all usable space.

As you said, we can just remove the loop as even one single 128M leak
can already make the last pwrite verification fail.

So I'll update the test case to remove the loop.

Thanks,
Qu

> 
> Other than that the test looks fine to me.
> 
> Thanks,
> Eryu
> 
>> +	# Create a file with the following layout:
>> +	# 0         128M      256M      384M
>> +	# |  Hole   |4K| Hole |4K| Hole |
>> +	# The total hole size will be 384M - 8k
>> +	truncate -s 384m "$SCRATCH_MNT/file"
>> +	$XFS_IO_PROG -c "pwrite 128m 4k" -c "pwrite 256m 4k" \
>> +		"$SCRATCH_MNT/file" | _filter_xfs_io
>> +
>> +	# Falloc 0~384M range, it's going to fail due to the qgroup limit
>> +	$XFS_IO_PROG -c "falloc 0 384m" "$SCRATCH_MNT/file" |\
>> +		_filter_xfs_io_error
>> +	rm "$SCRATCH_MNT/file"
>> +
>> +	# Ensure above delete reaches disk and free some space
>> +	sync
>> +done
>> +
>> +# We should be able to write at least 3/4 of the limit
>> +$XFS_IO_PROG -f -c "pwrite 0 192m" "$SCRATCH_MNT/file" | _filter_xfs_io
>> +
>> +# success, all done
>> +status=0
>> +exit
>> diff --git a/tests/btrfs/192.out b/tests/btrfs/192.out
>> new file mode 100644
>> index 00000000..13bc6036
>> --- /dev/null
>> +++ b/tests/btrfs/192.out
>> @@ -0,0 +1,18 @@
>> +QA output created by 192
>> +wrote 4096/4096 bytes at offset 134217728
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +wrote 4096/4096 bytes at offset 268435456
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +fallocate: Disk quota exceeded
>> +wrote 4096/4096 bytes at offset 134217728
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +wrote 4096/4096 bytes at offset 268435456
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +fallocate: Disk quota exceeded
>> +wrote 4096/4096 bytes at offset 134217728
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +wrote 4096/4096 bytes at offset 268435456
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> +fallocate: Disk quota exceeded
>> +wrote 201326592/201326592 bytes at offset 0
>> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
>> diff --git a/tests/btrfs/group b/tests/btrfs/group
>> index 2474d43e..160fe927 100644
>> --- a/tests/btrfs/group
>> +++ b/tests/btrfs/group
>> @@ -194,3 +194,4 @@
>>  189 auto quick send clone
>>  190 auto quick replay balance qgroup
>>  191 auto quick send dedupe
>> +192 auto qgroup fast enospc limit
>> -- 
>> 2.22.0
>>
>
diff mbox series

Patch

diff --git a/tests/btrfs/192 b/tests/btrfs/192
new file mode 100755
index 00000000..361b6d92
--- /dev/null
+++ b/tests/btrfs/192
@@ -0,0 +1,72 @@ 
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (C) 2019 SUSE Linux Products GmbH. All Rights Reserved.
+#
+# FS QA Test 192
+#
+# Test if btrfs is going to leak qgroup reserved data space when
+# falloc on multiple holes fails.
+# The fix is titled:
+# "btrfs: qgroup: Fix the wrong target io_tree when freeing reserved data space"
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+
+# Modify as appropriate.
+_supported_fs btrfs
+_supported_os Linux
+_require_scratch
+_require_xfs_io_command falloc
+
+_scratch_mkfs > /dev/null
+_scratch_mount
+
+$BTRFS_UTIL_PROG quota enable "$SCRATCH_MNT" > /dev/null
+$BTRFS_UTIL_PROG quota rescan -w "$SCRATCH_MNT" > /dev/null
+$BTRFS_UTIL_PROG qgroup limit -e 256M "$SCRATCH_MNT"
+
+for i in $(seq 3); do
+	# Create a file with the following layout:
+	# 0         128M      256M      384M
+	# |  Hole   |4K| Hole |4K| Hole |
+	# The total hole size will be 384M - 8k
+	truncate -s 384m "$SCRATCH_MNT/file"
+	$XFS_IO_PROG -c "pwrite 128m 4k" -c "pwrite 256m 4k" \
+		"$SCRATCH_MNT/file" | _filter_xfs_io
+
+	# Falloc 0~384M range, it's going to fail due to the qgroup limit
+	$XFS_IO_PROG -c "falloc 0 384m" "$SCRATCH_MNT/file" |\
+		_filter_xfs_io_error
+	rm "$SCRATCH_MNT/file"
+
+	# Ensure above delete reaches disk and free some space
+	sync
+done
+
+# We should be able to write at least 3/4 of the limit
+$XFS_IO_PROG -f -c "pwrite 0 192m" "$SCRATCH_MNT/file" | _filter_xfs_io
+
+# success, all done
+status=0
+exit
diff --git a/tests/btrfs/192.out b/tests/btrfs/192.out
new file mode 100644
index 00000000..13bc6036
--- /dev/null
+++ b/tests/btrfs/192.out
@@ -0,0 +1,18 @@ 
+QA output created by 192
+wrote 4096/4096 bytes at offset 134217728
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+wrote 4096/4096 bytes at offset 268435456
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+fallocate: Disk quota exceeded
+wrote 4096/4096 bytes at offset 134217728
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+wrote 4096/4096 bytes at offset 268435456
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+fallocate: Disk quota exceeded
+wrote 4096/4096 bytes at offset 134217728
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+wrote 4096/4096 bytes at offset 268435456
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+fallocate: Disk quota exceeded
+wrote 201326592/201326592 bytes at offset 0
+XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
diff --git a/tests/btrfs/group b/tests/btrfs/group
index 2474d43e..160fe927 100644
--- a/tests/btrfs/group
+++ b/tests/btrfs/group
@@ -194,3 +194,4 @@ 
 189 auto quick send clone
 190 auto quick replay balance qgroup
 191 auto quick send dedupe
+192 auto qgroup fast enospc limit