xfstests: Test xfs_fsr / swapext management of di_forkoff w/ selinux
PATCH 1/1 xfstests: Test xfs_fsr / swapext management of di_forkoff w/ selinux
In some cases selinux's creation of an xattr on the temporary fd creates a local xattr, but the file we are trying to defragment has attrs in extent format, and the forkoff mismatch will cause xfs_fsr to fail. This test demonstrates it; I have old patches sent to the list long ago that should fix it. I'll resend them soon.
---
tests/xfs/351 new file +89 -0
@@ -0,0 +1,89 @@ +#! /bin/bash +# FS QA Test No. 350 +# +# Test xfs_fsr / swapext management of di_forkoff w/ selinux +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 Red Hat, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- +# +# creator + +seq=`basename $0` +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 + +_require_scratch + +# real QA test starts here + +_supported_fs xfs +_supported_os Linux + +getenforce | grep -q "Enforcing\|Permissive" || _notrun "SELinux not enabled" +[ "$XFS_FSR_PROG" = "" ] && _notrun "xfs_fsr not found" + +rm -f $seqres.full + +_scratch_mkfs_xfs -i size=256 >> $seq.full 2>&1 || _fail "mkfs failed" + +# Manually mount to avoid fs-wide context set by default in xfstests +mount $SCRATCH_DEV $SCRATCH_MNT + +touch $SCRATCH_MNT/$seq.test + +# Create an extents-format attribute on a file +xfs_io -f -c "pwrite -S 0x63 0 4096" $SCRATCH_MNT/attrvals 2>&1 &> $seqres.full +cat $SCRATCH_MNT/attrvals | attr -s name $SCRATCH_MNT/$seq.test 2>&1 &> $seqres.full + +# Fragment the file by writing backwards +for I in `seq 6 -1 0`; do + dd if=/dev/zero of=$SCRATCH_MNT/$seq.test seek=$I bs=4k \ + oflag=direct count=1 conv=notrunc 2>&1 &> $seqres.full +done + +# See if xfs_fsr works. + +# The target file now has a non-local attribute, but the +# fsr temp file gets assigned a local attribute on creation, +# due to selinux. + +# In this case, the large-ish selinux attr will create a forkoff +# in the temp file smaller than the forkoff in the target file, +# because the FMT_EXTENTS attr takes up less space. If there is +# no mechanism to grow the temp forkoff to match, we can end up +# failing to swap these 2 inodes if the result is not enough +# data space in the temp inode as a result. + +$XFS_FSR_PROG -v $SCRATCH_MNT/$seq.test | _filter_scratch + +# success, all done +status=0 +exit
tests/xfs/351.out new file +3 -0
@@ -0,0 +1,3 @@ +QA output created by 351 +SCRATCH_MNT/351.test +extents before:6 after:1 DONE SCRATCH_MNT/351.test
tests/xfs/group +1 -1
@@ -180,4 +180,4 @@ 298 auto attr symlink quick 299 auto quota 350 auto dump quota - +351 auto fsr
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 29 lines · Eric Sandeen
In some cases selinux's creation of an xattr on the temporary fd creates a local xattr, but the file we are trying to defragment has attrs in extent format, and the forkoff mismatch will cause
xfs_fsrto fail. This test demonstrates it; I have old patches sent to the list long ago that should fix it. I'll resend them soon.Signed-off-by Eric Sandeen <[email protected]>---
tests/xfs/351 new file +13
-0@@ -0,0 +1,89 @@ +#! /bin/bash +# FS QA Test No. 350 +# +# Test xfs_fsr / swapext management of di_forkoff w/ selinux +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 Red Hat, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful,
tests/xfs/351 line 23
Quoted 10 lines
tests/xfs/351 new file +10
-0+# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- +# +# creator
Seems we don't need "creator" in comment now.
tests/xfs/351 line 25
Quoted 2 lines
tests/xfs/351 new file +2
-0+ +seq=`basename $0`
Forgot about "seqres=$RESULT_DIR/$seq" ?
Quoted 28 lines
tests/xfs/351 new file +28
-0+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 $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +_require_scratch + +# real QA test starts here + +_supported_fs xfs +_supported_os Linux + +getenforce | grep -q "Enforcing\|Permissive" || _notrun "SELinux not enabled" +[ "$XFS_FSR_PROG" = "" ] && _notrun "xfs_fsr not found" + +rm -f $seqres.full
tests/xfs/351 line 63
Quoted 10 lines
tests/xfs/351 new file +10
-0+ +_scratch_mkfs_xfs -i size=256 >> $seq.full 2>&1 || _fail "mkfs failed" + +# Manually mount to avoid fs-wide context set by default in xfstests +mount $SCRATCH_DEV $SCRATCH_MNT + +touch $SCRATCH_MNT/$seq.test + +# Create an extents-format attribute on a file +xfs_io -f -c "pwrite -S 0x63 0 4096" $SCRATCH_MNT/attrvals 2>&1 &> $seqres.full
Use $XFS_IO_PROG instead?
And seems "2>&1" is redundant, "&>" can do the work.
Thanks,
Eryu Guan
Quoted 47 lines
tests/xfs/351 new file +26
-0+cat $SCRATCH_MNT/attrvals | attr -s name $SCRATCH_MNT/$seq.test 2>&1 &> $seqres.full + +# Fragment the file by writing backwards +for I in `seq 6 -1 0`; do + dd if=/dev/zero of=$SCRATCH_MNT/$seq.test seek=$I bs=4k \ + oflag=direct count=1 conv=notrunc 2>&1 &> $seqres.full +done + +# See if xfs_fsr works. + +# The target file now has a non-local attribute, but the +# fsr temp file gets assigned a local attribute on creation, +# due to selinux. + +# In this case, the large-ish selinux attr will create a forkoff +# in the temp file smaller than the forkoff in the target file, +# because the FMT_EXTENTS attr takes up less space. If there is +# no mechanism to grow the temp forkoff to match, we can end up +# failing to swap these 2 inodes if the result is not enough +# data space in the temp inode as a result. + +$XFS_FSR_PROG -v $SCRATCH_MNT/$seq.test | _filter_scratch + +# success, all done +status=0 +exittests/xfs/351.out new file +3
-0@@ -0,0 +1,3 @@ +QA output created by 351 +SCRATCH_MNT/351.test +extents before:6 after:1 DONE SCRATCH_MNT/351.testtests/xfs/group +1
-1@@ -180,4 +180,4 @@ 298 auto attr symlink quick 299 auto quota 350 auto dump quota - +351 auto fsr
Quoted 4 lines
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 34 lines · Eryu Guan
Quoted 33 lines · Eric Sandeen
In some cases selinux's creation of an xattr on the temporary fd creates a local xattr, but the file we are trying to defragment has attrs in extent format, and the forkoff mismatch will cause
xfs_fsrto fail. This test demonstrates it; I have old patches sent to the list long ago that should fix it. I'll resend them soon.Signed-off-by Eric Sandeen <[email protected]>---
tests/xfs/351 new file +17
-0@@ -0,0 +1,89 @@ +#! /bin/bash +# FS QA Test No. 350 +# +# Test xfs_fsr / swapext management of di_forkoff w/ selinux +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 Red Hat, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +#
Quoted 8 lines
tests/xfs/351 line 23
Quoted 6 lines
tests/xfs/351 new file +6
-0+# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- +# +# creatorSeems we don't need "creator" in comment now.
yup
Quoted 4 lines
tests/xfs/351 line 25
Quoted 2 lines
tests/xfs/351 new file +2
-0+ +seq=`basename $0`Forgot about "seqres=$RESULT_DIR/$seq" ?
crud. :)
...
Quoted 4 lines
tests/xfs/351 line 63
Quoted 2 lines
tests/xfs/351 new file +2
-0+# Create an extents-format attribute on a file +xfs_io -f -c "pwrite -S 0x63 0 4096" $SCRATCH_MNT/attrvals 2>&1 &> $seqres.fullUse $XFS_IO_PROG instead?
yup
Quoted 1 line
And seems "2>&1" is redundant, "&>" can do the work.
heh, yeah. Ok, will tidy up and resend tomorrow.
Thanks!
-Eric
Quoted 50 lines
Thanks,
Eryu GuanQuoted 47 lines
tests/xfs/351 new file +26
-0+cat $SCRATCH_MNT/attrvals | attr -s name $SCRATCH_MNT/$seq.test 2>&1 &> $seqres.full + +# Fragment the file by writing backwards +for I in `seq 6 -1 0`; do + dd if=/dev/zero of=$SCRATCH_MNT/$seq.test seek=$I bs=4k \ + oflag=direct count=1 conv=notrunc 2>&1 &> $seqres.full +done + +# See if xfs_fsr works. + +# The target file now has a non-local attribute, but the +# fsr temp file gets assigned a local attribute on creation, +# due to selinux. + +# In this case, the large-ish selinux attr will create a forkoff +# in the temp file smaller than the forkoff in the target file, +# because the FMT_EXTENTS attr takes up less space. If there is +# no mechanism to grow the temp forkoff to match, we can end up +# failing to swap these 2 inodes if the result is not enough +# data space in the temp inode as a result. + +$XFS_FSR_PROG -v $SCRATCH_MNT/$seq.test | _filter_scratch + +# success, all done +status=0 +exittests/xfs/351.out new file +3
-0@@ -0,0 +1,3 @@ +QA output created by 351 +SCRATCH_MNT/351.test +extents before:6 after:1 DONE SCRATCH_MNT/351.testtests/xfs/group +1
-1@@ -180,4 +180,4 @@ 298 auto attr symlink quick 299 auto quota 350 auto dump quota - +351 auto fsr
Quoted 10 lines
Quoted 4 lines
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
PATCH v2 1/1 xfstests: Test xfs_fsr / swapext management of di_forkoff w/ selinux
In reply to Eric SandeenIn some cases selinux's creation of an xattr on the temporary fd creates a local xattr, but the file we are trying to defragment has attrs in extent format, and the forkoff mismatch will cause xfs_fsr to fail. This test demonstrates it; I have old patches sent to the list long ago that should fix it. I'll resend them soon.
---
V2: Address Eryu's review
tests/xfs/351 new file +88 -0
@@ -0,0 +1,88 @@ +#! /bin/bash +# FS QA Test No. 351 +# +# Test xfs_fsr / swapext management of di_forkoff w/ selinux +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 Red Hat, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- + +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 + +_require_scratch + +# real QA test starts here + +_supported_fs xfs +_supported_os Linux + +getenforce | grep -q "Enforcing\|Permissive" || _notrun "SELinux not enabled" +[ "$XFS_FSR_PROG" = "" ] && _notrun "xfs_fsr not found" + +rm -f $seqres.full + +_scratch_mkfs_xfs -i size=256 >> $seq.full 2>&1 || _fail "mkfs failed" + +# Manually mount to avoid fs-wide context set by default in xfstests +mount $SCRATCH_DEV $SCRATCH_MNT + +touch $SCRATCH_MNT/$seq.test + +# Create an extents-format attribute on a file +$XFS_IO_PROG -f -c "pwrite -S 0x63 0 4096" $SCRATCH_MNT/attrvals >> $seqres.full 2>&1 +cat $SCRATCH_MNT/attrvals | attr -s name $SCRATCH_MNT/$seq.test >> $seqres.full 2>&1 + +# Fragment the file by writing backwards +for I in `seq 6 -1 0`; do + dd if=/dev/zero of=$SCRATCH_MNT/$seq.test seek=$I bs=4k \ + oflag=direct count=1 conv=notrunc >> $seqres.full 2>&1 +done + +# See if xfs_fsr works. + +# The target file now has a non-local attribute, but the +# fsr temp file gets assigned a local attribute on creation, +# due to selinux. + +# In this case, the large-ish selinux attr will create a forkoff +# in the temp file smaller than the forkoff in the target file, +# because the FMT_EXTENTS attr takes up less space. If there is +# no mechanism to grow the temp forkoff to match, we can end up +# failing to swap these 2 inodes if the result is not enough +# data space in the temp inode as a result. + +$XFS_FSR_PROG -v $SCRATCH_MNT/$seq.test | _filter_scratch + +# success, all done +status=0 +exit
tests/xfs/351.out new file +3 -0
@@ -0,0 +1,3 @@ +QA output created by 351 +SCRATCH_MNT/351.test +extents before:6 after:1 DONE SCRATCH_MNT/351.test
tests/xfs/group +1 -1
@@ -180,4 +180,4 @@ 298 auto attr symlink quick 299 auto quota 350 auto dump quota - +351 auto fsr
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 84 lines · Eric Sandeen
In some cases selinux's creation of an xattr on the temporary fd creates a local xattr, but the file we are trying to defragment has attrs in extent format, and the forkoff mismatch will cause
xfs_fsrto fail. This test demonstrates it; I have old patches sent to the list long ago that should fix it. I'll resend them soon.Signed-off-by Eric Sandeen <[email protected]>---
V2: Address Eryu's review
tests/xfs/351 new file +66
-0@@ -0,0 +1,88 @@ +#! /bin/bash +# FS QA Test No. 351 +# +# Test xfs_fsr / swapext management of di_forkoff w/ selinux +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 Red Hat, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- + +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 $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +_require_scratch + +# real QA test starts here + +_supported_fs xfs +_supported_os Linux + +getenforce | grep -q "Enforcing\|Permissive" || _notrun "SELinux not enabled" +[ "$XFS_FSR_PROG" = "" ] && _notrun "xfs_fsr not found" + +rm -f $seqres.full + +_scratch_mkfs_xfs -i size=256 >> $seq.full 2>&1 || _fail "mkfs failed" + +# Manually mount to avoid fs-wide context set by default in xfstests +mount $SCRATCH_DEV $SCRATCH_MNT + +touch $SCRATCH_MNT/$seq.test + +# Create an extents-format attribute on a file +$XFS_IO_PROG -f -c "pwrite -S 0x63 0 4096" $SCRATCH_MNT/attrvals >> $seqres.full 2>&1 +cat $SCRATCH_MNT/attrvals | attr -s name $SCRATCH_MNT/$seq.test >> $seqres.full 2>&1 + +# Fragment the file by writing backwards +for I in `seq 6 -1 0`; do
tests/xfs/351 line 76
Quoted 10 lines
tests/xfs/351 new file +10
-0+ dd if=/dev/zero of=$SCRATCH_MNT/$seq.test seek=$I bs=4k \ + oflag=direct count=1 conv=notrunc >> $seqres.full 2>&1 +done + +# See if xfs_fsr works. + +# The target file now has a non-local attribute, but the +# fsr temp file gets assigned a local attribute on creation, +# due to selinux. +
There's trailing whitespaces issue here, otherwise looks good to me.
I suppose the seq number can be fixed at commit time, as well as the trailing whitespaces.
Thanks!
Eryu
BTW, test fails on 3.12-rc5 kernel
root/xfstests/results//xfs/351.out.bad 2013-10-18 13:24:38.014147342 -0400 renamed from xfs/351.out 2013-10-18 13:23:10.219847555 -0400 +1 -1
@@ -1,3 +1,3 @@ QA output created by 351 SCRATCH_MNT/351.test -extents before:6 after:1 DONE SCRATCH_MNT/351.test +XFS_IOC_SWAPEXT failed: SCRATCH_MNT/351.test: Invalid argument
Quoted 32 lines
tests/xfs/351 new file +12
-0+# In this case, the large-ish selinux attr will create a forkoff +# in the temp file smaller than the forkoff in the target file, +# because the FMT_EXTENTS attr takes up less space. If there is +# no mechanism to grow the temp forkoff to match, we can end up +# failing to swap these 2 inodes if the result is not enough +# data space in the temp inode as a result. + +$XFS_FSR_PROG -v $SCRATCH_MNT/$seq.test | _filter_scratch + +# success, all done +status=0 +exittests/xfs/351.out new file +3
-0@@ -0,0 +1,3 @@ +QA output created by 351 +SCRATCH_MNT/351.test +extents before:6 after:1 DONE SCRATCH_MNT/351.testtests/xfs/group +1
-1@@ -180,4 +180,4 @@ 298 auto attr symlink quick 299 auto quota 350 auto dump quota - +351 auto fsr
Quoted 4 lines
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
On 10/18/13 12:26 PM, Eryu Guan wrote:
...
Quoted 11 lines
There's trailing whitespaces issue here, otherwise looks good to me.
I suppose the seq number can be fixed at commit time, as well as the trailing whitespaces.
Reviewed-by Eryu Guan <[email protected]>Thanks!
EryuBTW, test fails on 3.12-rc5 kernel
Yup. (actually it's a userspace bug)
-Eric
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Committed as test xfs/300.
--Rich
commit 5d88b638ff4ad68f5ecce96d5d4d3a85fc4a2a5c
Author: Eric Sandeen <[email protected]> Date: Fri Oct 18 16:23:14 2013 +0000
xfstests: Test xfs_fsr / swapext management of di_forkoff w/ selinux
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs