On Wed, May 17, 2017 at 01:20:36PM -0400, J. Bruce Fields wrote:
> On Wed, May 17, 2017 at 11:38:54AM +0800, Eryu Guan wrote:
> > On Tue, May 16, 2017 at 11:07:45AM -0400, J. Bruce Fields wrote:
> > > On Tue, May 16, 2017 at 08:09:09PM +0800, Eryu Guan wrote:
> > > > Test nfs4_getfacl gets ACL list correctly from server when the ACL
> > > > length is close enough to the end of a page. On buggy NFS client
> > > > getxattr could return ERANGE. Upstream commit ed92d8c137b7 ("NFSv4:
> > > > fix getacl ERANGE for some ACL buffer sizes") fixed this bug in 4.11
> > > > kernel.
> > >
> > > Thanks.
> > >
> > > > Note that this reproducer was originally written by J. Bruce Fields.
> > > >
> > > > Cc: J. Bruce Fields <bfields@xxxxxxxxxx>
> > > > Cc: Weston Andros Adamson <dros@xxxxxxxxxxxxxxx>
> > > > Cc: linux-nfs@xxxxxxxxxxxxxxx
> > > > Signed-off-by: Eryu Guan <eguan@xxxxxxxxxx>
> > > > ---
> > > >
> > > > Hi all,
> > > >
> > > > This is the first attempt to add NFS specific tests to fstests with the hope
> > > > that we'll have more NFS tests go to fstests and more people testing NFS by
> > > > running fstests, it would be great if fstests could be part of the Linux NFS
> > > > development workflow too.
> > >
> > > What's the difference between fstests and xfstests?
> >
> > Sorry, I should have mentioned that xfstests has been renamed to
> > fstests in 2014[1], so they're the same thing.
>
> That could be better advertised. A google search for "fstests" for
> example turns up mostly irrelevant results. And I can't recall anyone
> at LSF/MM meetings, for example, calling it anything other than
> "xfstests"?
(Sorry for getting back to this late.)
Yes, that's a problem, it's been 'xfstests' for at least 14 years and
'fstests' is only there for 3 years, sometimes (or most of the times) we
still call it 'xfstests' for a better understanding.
But at least this thread does 'advertise' fstests to NFS list, I guess :)
>
> Anyway, thanks for adding this. I do a partial fstests run as part of
Can I take it as a "Reviewed-by" tag? :)
> my regular testing but I'm not great about keeping it updated.
fstests is updated weekly, but usually only one or two new tests (or
none) are added in a update, so not keeping it up-to-date is not a big
issue. Running existing tests would be good enough too.
For a quick smoke test
'./check -nfs -g quick'
(run tests in 'quick' group) will do, it only takes around 10 minutes for me to complete.
And for something more thoroughly, you can do
'./check -nfs -g auto'
(run tests in 'auto' group), which will take longer time but runs many
more tests.
I test every RC release with local-mounted NFS, with NFSv3,
NFSv4.0/4.1/4.2. Here're the 'known issues' I have for 4.11 kernel (I
think some failures are false alert, the tests need updates, some
failures may long standing & low priority bugs on NFS? But I'm not sure
which is which)
Failures for all NFS versions:
generic/035 test fail in rename syscall test, test reported
"t_rename_overwrite: fstat(3): Stale file handle"
generic/294 expect EEXIST not EROFS
Failures for NFSv3
generic/258 NFSv3 does not support timestamp before epoch
generic/375 sgid not cleared on setfacl, Permission denied
Failures for NFSv4.0/4.1/4.2
generic/011 sometimes direstress test failed, rarely happened
generic/401 NFSv4 reports . and .. dtype as unknown
I use this 'local.config' file for fstests to test NFS, and update the
'vers=3' mount option as needed
----->8------
TEST_DEV=localhost:/mnt/nfsexport
TEST_DIR=/mnt/testarea/test
SCRATCH_DEV=localhost:/mnt/nfsscratch
SCRATCH_MNT=/mnt/testarea/scratch
TEST_FS_MOUNT_OPTS="-o vers=3"
MOUNT_OPTIONS="-o vers=3"
-----8<------
>
> --b.
>
> > > > fstests does find NFS bugs (some are regression bugs), I noticed at least three
> > > > regressions introduced in 4.12-rc1 release, one was reported to nfs list[1].
Some updates on these bugs.
> > > > The other two regressions are:
> > > > a) generic/095 crashes NFS (not sure server or client yet, not reproduced
> > > > manually yet)
This is a bit hard to reproduce, I bisected this to commit "1df631a mm:
make rmap_walk() return void", but I suspected the result is wrong.. I'm
doing further testings to confirm.
> > > > b) copy_file_range(2) hang on NFSv4.2 mount, caught by just added
> > > > copy_file_range tests.
There's a patch to fix this bug, thanks for the quick response!
Thanks,
Eryu
> > > >
> > > > I'm still working on bisecting them now, will report to NFS list once I get
> > > > something.
> > > >
> > > > Thanks,
> > > > Eryu
> > > >
> > > > [1] https://www.spinics.net/lists/linux-nfs/msg63767.html
> > > >
> > > > common/config | 2 ++
> > > > common/nfs | 35 +++++++++++++++++++++++++
> > > > common/rc | 1 +
> > > > tests/nfs/001 | 77 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > > tests/nfs/001.out | 2 ++
> > > > tests/nfs/Makefile | 20 ++++++++++++++
> > > > tests/nfs/group | 6 +++++
> > > > 7 files changed, 143 insertions(+)
> > > > create mode 100644 common/nfs
> > > > create mode 100755 tests/nfs/001
> > > > create mode 100644 tests/nfs/001.out
> > > > create mode 100644 tests/nfs/Makefile
> > > > create mode 100644 tests/nfs/group
> > > >
> > > > diff --git a/common/config b/common/config
> > > > index 8211356..c3e9997 100644
> > > > --- a/common/config
> > > > +++ b/common/config
> > > > @@ -196,6 +196,8 @@ export XZ_PROG="`set_prog_path xz`"
> > > > export FLOCK_PROG="`set_prog_path flock`"
> > > > export LDD_PROG="`set_prog_path ldd`"
> > > > export TIMEOUT_PROG="`set_prog_path timeout`"
> > > > +export NFS4_SETFACL_PROG="`set_prog_path nfs4_setfacl`"
> > > > +export NFS4_GETFACL_PROG="`set_prog_path nfs4_getfacl`"
> > > >
> > > > # use 'udevadm settle' or 'udevsettle' to wait for lv to be settled.
> > > > # newer systems have udevadm command but older systems like RHEL5 don't.
> > > > diff --git a/common/nfs b/common/nfs
> > > > new file mode 100644
> > > > index 0000000..a03eaf4
> > > > --- /dev/null
> > > > +++ b/common/nfs
> > > > @@ -0,0 +1,35 @@
> > > > +#
> > > > +# NFS specific common functions.
> > > > +#
> > > > +
> > > > +# given an NFS export and a NFS version number, return 0 if
> > > > +# the mounted NFS is the given version. e.g.
> > > > +# _is_nfs_version $SCRATCH_DEV 4.2
> > > > +_is_nfs_version()
> > > > +{
> > > > + local nfs_exp=$1
> > > > + local nfs_ver=$2
> > > > + _fs_options $1 | grep -q "vers=$2"
> > > > +}
> > > > +
> > > > +# require $TEST_DEV mounted as a given NFS version
> > > > +# e.g. _require_test_nfs_version 4.2
> > > > +_require_test_nfs_version()
> > > > +{
> > > > + _require_test
> > > > + if ! _is_nfs_version $TEST_DEV $1; then
> > > > + _notrun "Test requires \$TEST_DEV mounted as NFSv$1"
> > > > + fi
> > > > +}
> > > > +
> > > > +# require $SCRATCH_DEV mounted as a given NFS version
> > > > +# e.g. _require_scratch_nfs_version 4
> > > > +_require_scratch_nfs_version()
> > > > +{
> > > > + _require_scratch
> > > > + _scratch_mount
> > > > + if ! _is_nfs_version $SCRATCH_DEV $1; then
> > > > + _notrun "Test requires \$SCRATCH_DEV mounted as NFSv$1"
> > > > + fi
> > > > + _scratch_unmount
> > > > +}
> > > > diff --git a/common/rc b/common/rc
> > > > index aa35b5e..ca0a284 100644
> > > > --- a/common/rc
> > > > +++ b/common/rc
> > > > @@ -162,6 +162,7 @@ case "$FSTYP" in
> > > > [ "$MKFS_F2FS_PROG" = "" ] && _fatal "mkfs.f2fs not found"
> > > > ;;
> > > > nfs)
> > > > + . ./common/nfs
> > > > ;;
> > > > cifs)
> > > > ;;
> > > > diff --git a/tests/nfs/001 b/tests/nfs/001
> > > > new file mode 100755
> > > > index 0000000..764cde3
> > > > --- /dev/null
> > > > +++ b/tests/nfs/001
> > > > @@ -0,0 +1,77 @@
> > > > +#! /bin/bash
> > > > +# FS QA Test 001
> > > > +#
> > > > +# Test nfs4_getfacl gets ACL list correctly from server when the ACL length is
> > > > +# close enough to the end of a page. On buggy NFS client getxattr could return
> > > > +# ERANGE. Upstream commit ed92d8c137b7 ("NFSv4: fix getacl ERANGE for some ACL
> > > > +# buffer sizes") fixed this bug.
> > > > +#
> > > > +#-----------------------------------------------------------------------
> > > > +# Copyright (c) 2017 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.* $acltest.*
> > > > +}
> > > > +
> > > > +# 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
> > > > +_supported_fs nfs
> > > > +_supported_os Linux
> > > > +_require_test_nfs_version 4
> > > > +_require_command $NFS4_SETFACL_PROG "nfs4_setfacl"
> > > > +_require_command $NFS4_GETFACL_PROG "nfs4_getfacl"
> > > > +
> > > > +acltest=$TEST_DIR/acltest.$seq
> > > > +rm -f $acltest.*
> > > > +touch $acltest.file $acltest.list
> > > > +
> > > > +# Setup a carefully made ACE list to make attr buffer length is close enough to
> > > > +# PAGE_SIZE (size is 4088 in this case, and this only works for 4k page size)
> > > > +# that leaves too little room for adding extra attr bitmap in this page
> > > > +echo "A::OWNER@:RW" >$acltest.list
> > > > +for ((i=9802; i < 10002; i++)); do echo "A::$i:RW" >>$acltest.list; done
> > > > +echo "A::GROUP@:RW" >>$acltest.list
> > > > +echo "A::EVERYONE@:RW" >>$acltest.list
> > > > +
> > > > +# Save ACL to testfile
> > > > +$NFS4_SETFACL_PROG -S $acltest.list $acltest.file
> > > > +
> > > > +# Dump ACL, expect correct number of ACE entries
> > > > +$NFS4_GETFACL_PROG $acltest.file >>$seqres.full 2>&1
> > > > +$NFS4_GETFACL_PROG $acltest.file | wc -l
> > > > +
> > > > +# success, all done
> > > > +status=0
> > > > +exit
> > > > diff --git a/tests/nfs/001.out b/tests/nfs/001.out
> > > > new file mode 100644
> > > > index 0000000..741ade0
> > > > --- /dev/null
> > > > +++ b/tests/nfs/001.out
> > > > @@ -0,0 +1,2 @@
> > > > +QA output created by 001
> > > > +203
> > > > diff --git a/tests/nfs/Makefile b/tests/nfs/Makefile
> > > > new file mode 100644
> > > > index 0000000..754f2b2
> > > > --- /dev/null
> > > > +++ b/tests/nfs/Makefile
> > > > @@ -0,0 +1,20 @@
> > > > +#
> > > > +# Copyright (c) 2017 Red Hat, Inc. All Rights Reserved.
> > > > +#
> > > > +
> > > > +TOPDIR = ../..
> > > > +include $(TOPDIR)/include/builddefs
> > > > +
> > > > +NFS_DIR = nfs
> > > > +TARGET_DIR = $(PKG_LIB_DIR)/$(TESTS_DIR)/$(NFS_DIR)
> > > > +
> > > > +include $(BUILDRULES)
> > > > +
> > > > +install:
> > > > + $(INSTALL) -m 755 -d $(TARGET_DIR)
> > > > + $(INSTALL) -m 755 $(TESTS) $(TARGET_DIR)
> > > > + $(INSTALL) -m 644 group $(TARGET_DIR)
> > > > + $(INSTALL) -m 644 $(OUTFILES) $(TARGET_DIR)
> > > > +
> > > > +# Nothing.
> > > > +install-dev install-lib:
> > > > diff --git a/tests/nfs/group b/tests/nfs/group
> > > > new file mode 100644
> > > > index 0000000..2619eae
> > > > --- /dev/null
> > > > +++ b/tests/nfs/group
> > > > @@ -0,0 +1,6 @@
> > > > +# QA groups control file
> > > > +# Defines test groups and nominal group owners
> > > > +# - do not start group names with a digit
> > > > +# - comment line before each group is "new" description
> > > > +#
> > > > +001 auto quick nfs4_acl acl
> > > > --
> > > > 2.9.3
> > > >
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe fstests" in
> > > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > > More majordomo info at http://vger.kernel.org/majordomo-info.html
> --
> To unsubscribe from this list: send the line "unsubscribe fstests" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
