xfstests: xfs directory unbalance assert test
A message that is not in the archive (sent off-list or not yet received): [email protected]
PATCH 1/1 xfstests: xfs directory unbalance assert test
This tests triggers an assert in the XFS directory unbalance code. This test originally written by Brian Foster and suggestions from Micheal Semon.
--- tests/generic/319 | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/319.out | 2 + tests/generic/group | 1 3 files changed, 65 insertions(+)
tests/generic/319 new file +62 -0
@@ -0,0 +1,62 @@ +#! /bin/bash +# FS QA Test No. 319 +# +# Test directory code correctly handles fsstress filling the filesystem +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 SGI. 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 / +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter +_require_scratch + +# real QA test starts here + +_supported_fs generic +_supported_os IRIX Linux + +_scratch_unmount > /dev/null 2>&1 +_scratch_mkfs_sized 11g >> $seqres.full 2>&1 +_scratch_mount > /dev/null 2>&1 + +# Fill the filesystem. +FSSTRESS_ARGS="-z -s 1378390208 -fsymlink=1 -n7500000 -p4 -d $SCRATCH_MNT" +$FSSTRESS_PROG $FSSTRESS_ARGS >> $seqres.full 2>&1 + +cd $SCRATCH_MNT >> $seqres.full 2>&1 +sync +# A debug XFS may assert in the remove due to a directory bug. +rm -rf * +echo "--- silence is golden ---" +status=0 +exit
tests/generic/319.out new file +2 -0
@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---
tests/generic/group +1 -0
@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stress
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 54 lines · Mark Tinguely
This tests triggers an assert in the XFS directory unbalance code. This test originally written by Brian Foster and suggestions from Micheal Semon.
Signed-off-by Mark Tinguely <[email protected]>--- tests/generic/319 | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/319.out | 2 + tests/generic/group | 1 3 files changed, 65 insertions(+)Index: b/tests/generic/319
===================================================================tests/generic/319 new file +38
-0@@ -0,0 +1,62 @@ +#! /bin/bash +# FS QA Test No. 319 +# +# Test directory code correctly handles fsstress filling the filesystem +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 SGI. 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 / +} + +# get standard environment, filters and checks
tests/generic/319 line 48
Quoted 10 lines
tests/generic/319 new file +10
-0+. ./common/rc +. ./common/filter +_require_scratch + +# real QA test starts here + +_supported_fs generic +_supported_os IRIX Linux + +_scratch_unmount > /dev/null 2>&1
This is not necessary, _require_scratch has done the unmount work.
tests/generic/319 line 49
Quoted 1 line
tests/generic/319 new file +1
-0+_scratch_mkfs_sized 11g >> $seqres.full 2>&1
_scratch_mkfs_sized expects fssize in bytes, 11g is not a valid value The comments in common/rc about _scratch_mkfs_sized say
# _scratch_mkfs_sized <size in bytes> [optional blocksize]
Quoted 25 lines
tests/generic/319 new file +13
-0+_scratch_mount > /dev/null 2>&1 + +# Fill the filesystem. +FSSTRESS_ARGS="-z -s 1378390208 -fsymlink=1 -n7500000 -p4 -d $SCRATCH_MNT" +$FSSTRESS_PROG $FSSTRESS_ARGS >> $seqres.full 2>&1 + +cd $SCRATCH_MNT >> $seqres.full 2>&1 +sync +# A debug XFS may assert in the remove due to a directory bug. +rm -rf * +echo "--- silence is golden ---" +status=0 +exitIndex: b/tests/generic/319.out
===================================================================tests/generic/319.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---Index: b/tests/generic/group
===================================================================
tests/generic/group line 124
Quoted 4 lines
tests/generic/group +1
-0@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stress
Should be in auto group too I guess.
Thanks,
Eryu Guan
Quoted 6 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 1 line · Eryu Guan
On Tue, Sep 17, 2013 at 09:59:47AM -0500, Mark Tinguely wrote:
...
Quoted 3 lines
Quoted 1 line
+_scratch_unmount>
/dev/null2>&1This is not necessary,
_require_scratchhas done the unmount work.
okay, stole that from other tests.
Quoted 7 lines
Quoted 2 lines
+_scratch_mkfs_sized 11g>> $seqres.full 2>&1
_scratch_mkfs_sizedexpects fssize in bytes, 11g is not a valid value The comments in common/rc about_scratch_mkfs_sizedsay# _scratch_mkfs_sized<size in bytes> [optional blocksize]
That was a shortcut for xfs. Looking in common/rc. I see that it breaks the other filesystems that need the size in blocks.
...
Quoted 6 lines
Quoted 6 lines
tests/generic/319.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---Index: b/tests/generic/group
===================================================================
--- a/tests/generic/group
Quoted 8 lines
tests/generic/group line 124
Quoted 6 lines
+++ b/tests/generic/group
tests/generic/group +1
-0@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stressShould be in auto group too I guess.
It takes a very long time to run to completion, don't know if people want this in the auto run.
Quoted 2 lines
Thanks,
Eryu Guan
Thanks for the feedback
--Mark.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 11 lines · Mark Tinguely
Quoted 1 line · Eryu Guan
On Tue, Sep 17, 2013 at 09:59:47AM -0500, Mark Tinguely wrote:
...
Quoted 3 lines
Quoted 1 line
+_scratch_unmount>
/dev/null2>&1This is not necessary,
_require_scratchhas done the unmount work.okay, stole that from other tests.
Quoted 1 line
Quoted 8 lines
Quoted 6 lines
Quoted 1 line
+_scratch_mkfs_sized 11g>> $seqres.full 2>&1
_scratch_mkfs_sizedexpects fssize in bytes, 11g is not a valid value The comments in common/rc about_scratch_mkfs_sizedsay# _scratch_mkfs_sized<size in bytes> [optional blocksize]
That was a shortcut for xfs. Looking in common/rc. I see that it breaks the other filesystems that need the size in blocks.
at least mkfs.extN also understands "11g" but the helper does not, because it causes a failure in the device size check, (for any fs):
[ "$fssize" -gt "$devsize" ] && _notrun "Scratch device too small"
Quoted 17 lines
...
Quoted 6 lines
Quoted 6 lines
tests/generic/319.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---Index: b/tests/generic/group
===================================================================
--- a/tests/generic/groupQuoted 8 lines
tests/generic/group line 124
Quoted 6 lines
+++ b/tests/generic/group
tests/generic/group +1
-0@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stressShould be in auto group too I guess.
Quoted 1 line
It takes a very long time to run to completion, don't know if people want this in the auto run.
how long is long? We do have "quick" for people who want quick. I think auto is probably ok. Maybe we should add a "slow" group, and you can "-x slow" :)
-Eric
Quoted 12 lines
Quoted 2 lines
Thanks,
Eryu GuanThanks for the feedback
--Mark.
_______________________________________________
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 42 lines · Eric Sandeen
Quoted 11 lines · Mark Tinguely
Quoted 1 line · Eryu Guan
On Tue, Sep 17, 2013 at 09:59:47AM -0500, Mark Tinguely wrote:
...
Quoted 3 lines
Quoted 1 line
+_scratch_unmount>
/dev/null2>&1This is not necessary,
_require_scratchhas done the unmount work.okay, stole that from other tests.
Quoted 1 line
Quoted 8 lines
Quoted 6 lines
Quoted 1 line
+_scratch_mkfs_sized 11g>> $seqres.full 2>&1
_scratch_mkfs_sizedexpects fssize in bytes, 11g is not a valid value The comments in common/rc about_scratch_mkfs_sizedsay# _scratch_mkfs_sized<size in bytes> [optional blocksize]
That was a shortcut for xfs. Looking in common/rc. I see that it breaks the other filesystems that need the size in blocks.
at least mkfs.extN also understands "11g" but the helper does not, because it causes a failure in the device size check, (for any fs):
[ "$fssize" -gt "$devsize" ]&& _notrun "Scratch device too small"Quoted 17 lines
...
Quoted 14 lines
tests/generic/group line 124
Quoted 12 lines
tests/generic/319.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---Index: b/tests/generic/group
===================================================================tests/generic/group +1
-0@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stressShould be in auto group too I guess.
Quoted 13 lines
Quoted 1 line
It takes a very long time to run to completion, don't know if people want this in the auto run.
how long is long? We do have "quick" for people who want quick. I think auto is probably ok. Maybe we should add a "slow" group, and you can "-x slow" :)
-Eric
Quoted 6 lines
Quoted 2 lines
Thanks,
Eryu GuanThanks for the feedback
--Mark.
About 45 minutes.
--Mark.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 3 lines · Mark Tinguely
This tests triggers an assert in the XFS directory unbalance code. This test originally written by Brian Foster and suggestions from Micheal Semon.
cool, thanks. Comments below.
Quoted 38 lines
Signed-off-by Mark Tinguely <[email protected]>--- tests/generic/319 | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/319.out | 2 + tests/generic/group | 1 3 files changed, 65 insertions(+)Index: b/tests/generic/319
===================================================================tests/generic/319 new file +26
-0@@ -0,0 +1,62 @@ +#! /bin/bash +# FS QA Test No. 319 +# +# Test directory code correctly handles fsstress filling the filesystem +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 SGI. 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"
tests/generic/319 line 36
Quoted 10 lines
tests/generic/319 new file +10
-0+ +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / +}
That seems pointless; usually it's done w/ rm -f $tmp.* right after, but we have no tmpfile, so...
tests/generic/319 line 48
Quoted 12 lines
tests/generic/319 new file +12
-0+ +# get standard environment, filters and checks +. ./common/rc +. ./common/filter +_require_scratch + +# real QA test starts here + +_supported_fs generic +_supported_os IRIX Linux + +_scratch_unmount > /dev/null 2>&1
Aside:
I see this done both ways - is it required to unmount scratch at the beginning of a test? I don't think so (I know it's done in many tests, though, but again, C&P & cargo cult? Or not? I'm not sure :( )
I guess it doesn't hurt, but at some point I'd like to get it straight about who's required to umount scratch, and when (if at all).
tests/generic/319 line 49
Quoted 1 line
tests/generic/319 new file +1
-0+_scratch_mkfs_sized 11g >> $seqres.full 2>&1
_scratch_mkfs_sized doesn't take units like this ('g'), so the above fails to actually make an 11g fs:
# Create fs of certain size on scratch device
# _scratch_mkfs_sized <size in bytes> [optional blocksize] _scratch_mkfs_sized()
so we get this in 319.full:
expr: non-numeric argument
./common/rc: line 576: [: 11g: integer expression expected
but then it seems like mkfs carries on anyway w/ defaults. :(
Apparently the mkfs 11g part isn't actually critical? ;)
maybe _scratch_mkfs_sized needs something like this at the top:
re='^[0-9]+$'
if ! [[ $fssize =~ $re ]] ; then
_notrun "error: _scratch_mkfs_sized: $fssize not a number of bytes"
fi
tests/generic/319 line 50
Quoted 1 line
tests/generic/319 new file +1
-0+_scratch_mount > /dev/null 2>&1
is ignore-all-output really the right thing to do? When does _scratch_mount emit anything? (more cargo cult)? :)
tests/generic/319 line 56
Quoted 5 lines
tests/generic/319 new file +5
-0+# Fill the filesystem. +FSSTRESS_ARGS="-z -s 1378390208 -fsymlink=1 -n7500000 -p4 -d $SCRATCH_MNT" +$FSSTRESS_PROG $FSSTRESS_ARGS >> $seqres.full 2>&1 + +cd $SCRATCH_MNT >> $seqres.full 2>&1
cd doesn't emit anything except on error, right, and if there's an error we'd better stop the test right here!
tests/generic/319 line 59
Quoted 3 lines
tests/generic/319 new file +3
-0+sync +# A debug XFS may assert in the remove due to a directory bug. +rm -rf *
I'd feel a whole lot better if you did rm -rf $SCRATCH_MNT/* just in case we somehow ended up in the wrong place here.
Or even better if you pointed fsstress at $SCRATCH_MNT/$seq.dir, and then did rm -rf $SCRATCH_MNT/$seq.dir - just to avoid that nasty should-never-happen-still-scary "rm -rf /*"
-Eric
Quoted 21 lines
tests/generic/319 new file +3
-0+echo "--- silence is golden ---" +status=0 +exitIndex: b/tests/generic/319.out
===================================================================tests/generic/319.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 319 +--- silence is golden ---Index: b/tests/generic/group
===================================================================tests/generic/group +1
-0@@ -121,3 +121,4 @@ 316 auto quick 317 auto metadata quick 318 acl attr auto quick +319 stress
Quoted 5 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 50 lines · Eric Sandeen
Quoted 3 lines · Mark Tinguely
This tests triggers an assert in the XFS directory unbalance code. This test originally written by Brian Foster and suggestions from Micheal Semon.
cool, thanks. Comments below.
Quoted 43 lines
Signed-off-by Mark Tinguely<[email protected]>--- tests/generic/319 | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/319.out | 2 + tests/generic/group | 1 3 files changed, 65 insertions(+)Index: b/tests/generic/319
===================================================================tests/generic/319 new file +31
-0@@ -0,0 +1,62 @@ +#! /bin/bash +# FS QA Test No. 319 +# +# Test directory code correctly handles fsstress filling the filesystem +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 SGI. 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
Quoted 8 lines
tests/generic/319 line 36
Quoted 5 lines
tests/generic/319 new file +5
-0+ +_cleanup() +{ + cd / +}That seems pointless; usually it's done w/ rm -f $tmp.* right after, but we have no tmpfile, so...
Yeah, no cleanup is needed.
Quoted 15 lines
tests/generic/319
Quoted 12 lines
tests/generic/319 new file +12
-0+ +# get standard environment, filters and checks +. ./common/rc +. ./common/filter +_require_scratch + +# real QA test starts here + +_supported_fs generic +_supported_os IRIX Linux + +_scratch_unmount> /dev/null 2>&1Aside:
Quoted 6 lines
I see this done both ways - is it required to unmount scratch at the beginning of a test? I don't think so (I know it's done in many tests, though, but again, C&P& cargo cult? Or not? I'm not sure :( )
I guess it doesn't hurt, but at some point I'd like to get it straight about who's required to umount scratch, and when (if at all).
Have to unmount for the mkfs, as noted by Eryu, it is already done. I would rather manually unmount it than be surprised when someone changes the common files.
Quoted 10 lines
tests/generic/319 line 49
Quoted 2 lines
tests/generic/319 new file +1
-0+_scratch_mkfs_sized 11g>> $seqres.full 2>&1
_scratch_mkfs_sizeddoesn't take units like this ('g'), so the above fails to actually make an 11g fs:# Create fs of certain size on scratch device
# _scratch_mkfs_sized<size in bytes> [optional blocksize]_scratch_mkfs_sized()
Quoted 8 lines
so we get this in 319.full:
expr: non-numeric argument
./common/rc: line 576: [: 11g: integer expression expectedbut then it seems like mkfs carries on anyway w/ defaults. :(
Apparently the mkfs 11g part isn't actually critical? ;)
it works on xfs because mkfs.xfs size can take those values, but yes it breaks on other filesystem. my bad.
Quoted 29 lines
maybe
_scratch_mkfs_sizedneeds something like this at the top:re='^[0-9]+$'
if ! [[ $fssize =~ $re ]] ; then
_notrun"error:_scratch_mkfs_sized: $fssize not a number of bytes"
fitests/generic/319 line 50
Quoted 1 line
tests/generic/319 new file +1
-0+_scratch_mount> /dev/null 2>&1is ignore-all-output really the right thing to do? When does
_scratch_mountemit anything? (more cargo cult)? :)tests/generic/319
Quoted 5 lines
tests/generic/319 new file +5
-0+# Fill the filesystem. +FSSTRESS_ARGS="-z -s 1378390208 -fsymlink=1 -n7500000 -p4 -d $SCRATCH_MNT" +$FSSTRESS_PROG $FSSTRESS_ARGS>> $seqres.full 2>&1 + +cd $SCRATCH_MNT>> $seqres.full 2>&1cd doesn't emit anything except on error, right, and if there's an error we'd better stop the test right here!
tests/generic/319 line 59
Quoted 3 lines
tests/generic/319 new file +3
-0+sync +# A debug XFS may assert in the remove due to a directory bug. +rm -rf *I'd feel a whole lot better if you did rm -rf $SCRATCH_MNT/* just in case we somehow ended up in the wrong place here.
Quoted 4 lines
Or even better if you pointed fsstress at $SCRATCH_MNT/$seq.dir, and then did rm -rf $SCRATCH_MNT/$seq.dir - just to avoid that nasty should-never-happen-still-scary "rm -rf /*"
Makes a lot of sense.
Quoted 1 line
-Eric
Thanks Eric.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 81 lines · Mark Tinguely
Quoted 50 lines · Eric Sandeen
Quoted 3 lines · Mark Tinguely
This tests triggers an assert in the XFS directory unbalance code. This test originally written by Brian Foster and suggestions from Micheal Semon.
cool, thanks. Comments below.
Quoted 43 lines
Signed-off-by Mark Tinguely<[email protected]>--- tests/generic/319 | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/319.out | 2 + tests/generic/group | 1 3 files changed, 65 insertions(+)Index: b/tests/generic/319
===================================================================tests/generic/319 new file +31
-0@@ -0,0 +1,62 @@ +#! /bin/bash +# FS QA Test No. 319 +# +# Test directory code correctly handles fsstress filling the filesystem +# +#----------------------------------------------------------------------- +# Copyright (c) 2013 SGI. 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 15Quoted 8 lines
tests/generic/319 line 36
Quoted 5 lines
tests/generic/319 new file +5
-0+ +_cleanup() +{ + cd / +}That seems pointless; usually it's done w/ rm -f $tmp.* right after, but we have no tmpfile, so...
Yeah, no cleanup is needed.
Quoted 19 lines
tests/generic/319
Quoted 12 lines
tests/generic/319 new file +12
-0+ +# get standard environment, filters and checks +. ./common/rc +. ./common/filter +_require_scratch + +# real QA test starts here + +_supported_fs generic +_supported_os IRIX Linux + +_scratch_unmount> /dev/null 2>&1Aside:
I see this done both ways - is it required to unmount scratch at the beginning of a test? I don't think so (I know it's done in many tests, though, but again, C&P& cargo cult? Or not? I'm not sure :( )
Quoted 6 lines
Quoted 2 lines
I guess it doesn't hurt, but at some point I'd like to get it straight about who's required to umount scratch, and when (if at all).
Have to unmount for the mkfs, as noted by Eryu, it is already done. I would rather manually unmount it than be surprised when someone changes the common files.
If that happens, tons of tests will break. I'd really just remove it for clarity, but *shrug*
Quoted 12 lines
Quoted 12 lines
tests/generic/319 line 49
Quoted 2 lines
tests/generic/319 new file +1
-0+_scratch_mkfs_sized 11g>> $seqres.full 2>&1
_scratch_mkfs_sizeddoesn't take units like this ('g'), so the above fails to actually make an 11g fs:# Create fs of certain size on scratch device
# _scratch_mkfs_sized<size in bytes> [optional blocksize]_scratch_mkfs_sized()so we get this in 319.full:
Quoted 8 lines
Quoted 6 lines
expr: non-numeric argument
./common/rc: line 576: [: 11g: integer expression expectedbut then it seems like mkfs carries on anyway w/ defaults. :(
Apparently the mkfs 11g part isn't actually critical? ;)
it works on xfs because mkfs.xfs size can take those values, but yes it breaks on other filesystem. my bad.
One other nitpick in this area, please remove $seqres.full before you start so it doesn't grow each time the test is run.
(hm maybe we should add that to ./check or something; so many tests miss this)
But - no, it doesn't work for xfs either, at least not in all cases, because it doesn't do the device size check. Here's xfs output on a < 11G device:
expr: non-numeric argument
./common/rc: line 582: [: 11g: integer expression expected
** mkfs failed with extra mkfs options added to "-bsize=4096" by test 319 ** ** attempting to mkfs using only test 319 options: -d size=11g -b size=4096 ** size 11g specified for data subvolume is too large, maximum is 1048241 blocks
Usage: mkfs.xfs
/* blocksize */ [-b log=n|size=num]
...
and w/o error checking (2>&1 and no || _fail) the test just carries on w/o a fresh mkfs, on whatever size it happens to be.
-Eric
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
tests/generic/319 line 59
Quoted 8 lines · Mark Tinguely
tests/generic/319 new file +8
-0+# Fill the filesystem. +FSSTRESS_ARGS="-z -s 1378390208 -fsymlink=1 -n7500000 -p4 -d $SCRATCH_MNT" +$FSSTRESS_PROG $FSSTRESS_ARGS >> $seqres.full 2>&1 + +cd $SCRATCH_MNT >> $seqres.full 2>&1 +sync +# A debug XFS may assert in the remove due to a directory bug. +rm -rf *
Wouldn't this be better:
sync
rm -rf $SCRATCH_MNT/*
Remember, xfstests runs as root and so having a "rm -rf *" anywhere in a test is a disaster just waiting to happen....
Cheers,
Dave.
Signature · 7 lines
Dave Chinner [email protected] _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs