* [PATCH 1/3] Check pwrite parameters @ 2017-11-29 16:33 Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 2/3] generic/470: Test RWF_NOWAIT Goldwyn Rodrigues ` (2 more replies) 0 siblings, 3 replies; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-11-29 16:33 UTC (permalink / raw) To: darrick.wong, eguan; +Cc: fstests, Goldwyn Rodrigues From: Goldwyn Rodrigues <rgoldwyn@suse.com> There are some parameters added with xfs_io. Check if the pwrite parameters are available. For some cases, xfs_io now returns "command -%c not supported", so added "not supported" to count as error. Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> Change since v2: - More comments - opts to become local variable --- common/rc | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/common/rc b/common/rc index 4c053a53..90408b7e 100644 --- a/common/rc +++ b/common/rc @@ -2035,6 +2035,7 @@ _require_xfs_io_command() shift local param="$*" local param_checked=0 + local opts="" testfile=$TEST_DIR/$$.xfs_io case $command in @@ -2079,6 +2080,15 @@ _require_xfs_io_command() echo $testio | grep -q "invalid option" && \ _notrun "xfs_io $command support is missing" ;; + "pwrite") + # -N (RWF_NOWAIT) only works with direct I/O writes + if [ "$param" == "-N" ]; then + opts+=" -d" + fi + testio=`$XFS_IO_PROG -f $opts -c "pwrite $param 0 1M" \ + $testfile 2>&1` + param_checked=1 + ;; "scrub"|"repair") testio=`$XFS_IO_PROG -x -c "$command probe 0" $TEST_DIR 2>&1` echo $testio | grep -q "Inappropriate ioctl" && \ @@ -2109,7 +2119,9 @@ _require_xfs_io_command() $XFS_IO_PROG -c "help $command" | grep -q "^ $param --" || \ _notrun "xfs_io $command doesn't support $param" else - echo $testio | grep -q "invalid option" && \ + # xfs_io could result in "command %c not supported" if it was + # built on kernels not supporting pwritev2() calls + echo $testio | grep -q "\(invalid option\|not supported\)" && \ _notrun "xfs_io $command doesn't support $param" fi } -- 2.14.2 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 2/3] generic/470: Test RWF_NOWAIT 2017-11-29 16:33 [PATCH 1/3] Check pwrite parameters Goldwyn Rodrigues @ 2017-11-29 16:33 ` Goldwyn Rodrigues 2017-12-06 10:05 ` Eryu Guan 2017-11-29 16:33 ` [PATCH 3/3] generic/471: Partial direct write test Goldwyn Rodrigues 2017-12-06 9:55 ` [PATCH 1/3] Check pwrite parameters Eryu Guan 2 siblings, 1 reply; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-11-29 16:33 UTC (permalink / raw) To: darrick.wong, eguan; +Cc: fstests, Goldwyn Rodrigues From: Goldwyn Rodrigues <rgoldwyn@suse.com> Tests the RWF_NOWAIT flag so the I/O returns immediately with -EAGAIN on a new file since it requires block allocation. It creates a file, syncs it, and overwrites the file with RWF_NOWAIT. This should succeed. Finally, read the contents to make sure the overwrite is successful. Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> Changes since v1: - Fix testdir name - use of $XFS_IO_PROG instead of xfs_io - check pwrite accepts -N Changes since v2: - corrected test description and improved documentation - new leaner look ;) - rw group --- tests/generic/470 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/470.out | 13 +++++++++ tests/generic/group | 1 + 3 files changed, 88 insertions(+) create mode 100755 tests/generic/470 create mode 100644 tests/generic/470.out diff --git a/tests/generic/470 b/tests/generic/470 new file mode 100755 index 00000000..5580718b --- /dev/null +++ b/tests/generic/470 @@ -0,0 +1,74 @@ +#! /bin/bash +# FS QA Test No. 470 +# +# write a file with RWF_NOWAIT and it would fail because there are no +# blocks allocated. Create a file with direct I/O and re-write it +# using RWF_NOWAIT. I/O should finish within 50 microsecods since +# block allocations are already performed. +# +#----------------------------------------------------------------------- +# Copyright (c) 2017, SUSE Linux Products. 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! + +# get standard environment, filters and checks +. ./common/rc +. ./common/populate +. ./common/filter +. ./common/attr + +# real QA test starts here +_supported_os Linux +_require_odirect +_require_test +_require_xfs_io_command pwrite -N + +# Remove reminiscence of previously run tests +testdir=$TEST_DIR/$seq +if [ -e $testdir ]; then + rm -Rf $testdir +fi + +mkdir $testdir + +# Create a file with pwrite nowait (will fail with EAGAIN) +$XFS_IO_PROG -f -d -c "pwrite -N -V 1 -b 1M 0 1M" $testdir/f1 + +# Write the file without nowait +$XFS_IO_PROG -f -d -c "pwrite -S 0xaa -W -w -V 1 -b 1M 0 8M" $testdir/f1 | _filter_xfs_io + +time_taken=`$XFS_IO_PROG -d -c "pwrite -S 0xbb -N -V 1 -b 1M 2M 1M" $testdir/f1 | awk '/^1/ {print $5}'` + +# RWF_NOWAIT should finish within a short period of time. Anything longer +# means it is waiting for something in the kernel which would be a fail. +if (( $(echo "$time_taken < 0.05" | bc -l) )); then + echo "RWF_NOWAIT time is within limits." +else + echo "RWF_NOWAIT took $time_taken seconds" +fi + +$XFS_IO_PROG -c "pread -v 0 8M" $testdir/f1 | _filter_xfs_io_unique + +# success, all done +status=0 +exit diff --git a/tests/generic/470.out b/tests/generic/470.out new file mode 100644 index 00000000..e46622cb --- /dev/null +++ b/tests/generic/470.out @@ -0,0 +1,13 @@ +QA output created by 470 +pwrite: Resource temporarily unavailable +wrote 8388608/8388608 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +RWF_NOWAIT time is within limits. +00000000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ +* +00200000: bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb ................ +* +00300000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ +* +read 8388608/8388608 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) diff --git a/tests/generic/group b/tests/generic/group index 6c3bb03a..cf8529ea 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -472,3 +472,4 @@ 467 auto quick exportfs 468 shutdown auto quick metadata 469 auto quick +470 auto quick rw -- 2.14.2 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 2/3] generic/470: Test RWF_NOWAIT 2017-11-29 16:33 ` [PATCH 2/3] generic/470: Test RWF_NOWAIT Goldwyn Rodrigues @ 2017-12-06 10:05 ` Eryu Guan 2017-12-06 16:35 ` Goldwyn Rodrigues 0 siblings, 1 reply; 10+ messages in thread From: Eryu Guan @ 2017-12-06 10:05 UTC (permalink / raw) To: Goldwyn Rodrigues; +Cc: darrick.wong, fstests, Goldwyn Rodrigues On Wed, Nov 29, 2017 at 10:33:57AM -0600, Goldwyn Rodrigues wrote: > From: Goldwyn Rodrigues <rgoldwyn@suse.com> > > Tests the RWF_NOWAIT flag so the I/O returns immediately with -EAGAIN > on a new file since it requires block allocation. > > It creates a file, syncs it, and overwrites the file with RWF_NOWAIT. > This should succeed. > > Finally, read the contents to make sure the overwrite is successful. > > Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> > > Changes since v1: > - Fix testdir name > - use of $XFS_IO_PROG instead of xfs_io > - check pwrite accepts -N > > Changes since v2: > - corrected test description and improved documentation > - new leaner look ;) > - rw group > --- > tests/generic/470 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/generic/470.out | 13 +++++++++ > tests/generic/group | 1 + > 3 files changed, 88 insertions(+) > create mode 100755 tests/generic/470 > create mode 100644 tests/generic/470.out > > diff --git a/tests/generic/470 b/tests/generic/470 > new file mode 100755 > index 00000000..5580718b > --- /dev/null > +++ b/tests/generic/470 > @@ -0,0 +1,74 @@ > +#! /bin/bash > +# FS QA Test No. 470 > +# > +# write a file with RWF_NOWAIT and it would fail because there are no > +# blocks allocated. Create a file with direct I/O and re-write it > +# using RWF_NOWAIT. I/O should finish within 50 microsecods since > +# block allocations are already performed. > +# > +#----------------------------------------------------------------------- > +# Copyright (c) 2017, SUSE Linux Products. 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! > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/populate > +. ./common/filter > +. ./common/attr > + > +# real QA test starts here > +_supported_os Linux > +_require_odirect > +_require_test > +_require_xfs_io_command pwrite -N > + > +# Remove reminiscence of previously run tests > +testdir=$TEST_DIR/$seq > +if [ -e $testdir ]; then > + rm -Rf $testdir > +fi > + > +mkdir $testdir > + > +# Create a file with pwrite nowait (will fail with EAGAIN) > +$XFS_IO_PROG -f -d -c "pwrite -N -V 1 -b 1M 0 1M" $testdir/f1 It's weird that I got "Operation not supported" message here and test failed as @@ -1,5 +1,5 @@ QA output created by 470 -pwrite: Resource temporarily unavailable +pwrite: Operation not supported wrote 8388608/8388608 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) RWF_NOWAIT time is within limits. I was using latest for-next branch of upstream xfsprogs, and a pre-v4.15-rc1 kernel shipped by Fedora rawhide. A simple test showed: rm -f /mnt/xfs/testfile xfs_io -fdc "pwrite -N -V 1 -b 4k 0 4k" /mnt/xfs/testfile pwrite: Operation not supported But an strace run showed pwritev2 did -1 and set EAGAIN as errno, but printed "Operation not supported" as error message: ... pwritev2(3, [{iov_base="\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315"..., iov_len=4096}], 1, 0, RWF_NOWAIT) = -1 EAGAIN (Resource temporarily unavailable) dup(2) = 4 fcntl(4, F_GETFL) = 0x402 (flags O_RDWR|O_APPEND) openat(AT_FDCWD, "/usr/share/locale/en_US.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en_US.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en_US/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) fstat(4, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0 write(4, "pwrite: Operation not supported\n", 32pwrite: Operation not supported ... Not sure what happened yet, did I miss anything? > + > +# Write the file without nowait > +$XFS_IO_PROG -f -d -c "pwrite -S 0xaa -W -w -V 1 -b 1M 0 8M" $testdir/f1 | _filter_xfs_io I think "-w" (fdatasync) can be removed? We already use "-W" (fsync) here. > + > +time_taken=`$XFS_IO_PROG -d -c "pwrite -S 0xbb -N -V 1 -b 1M 2M 1M" $testdir/f1 | awk '/^1/ {print $5}'` > + > +# RWF_NOWAIT should finish within a short period of time. Anything longer > +# means it is waiting for something in the kernel which would be a fail. > +if (( $(echo "$time_taken < 0.05" | bc -l) )); then Better to describe where does this 0.05 come from. Thanks, Eryu > + echo "RWF_NOWAIT time is within limits." > +else > + echo "RWF_NOWAIT took $time_taken seconds" > +fi > + > +$XFS_IO_PROG -c "pread -v 0 8M" $testdir/f1 | _filter_xfs_io_unique > + > +# success, all done > +status=0 > +exit > diff --git a/tests/generic/470.out b/tests/generic/470.out > new file mode 100644 > index 00000000..e46622cb > --- /dev/null > +++ b/tests/generic/470.out > @@ -0,0 +1,13 @@ > +QA output created by 470 > +pwrite: Resource temporarily unavailable > +wrote 8388608/8388608 bytes at offset 0 > +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) > +RWF_NOWAIT time is within limits. > +00000000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ > +* > +00200000: bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb ................ > +* > +00300000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ > +* > +read 8388608/8388608 bytes at offset 0 > +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) > diff --git a/tests/generic/group b/tests/generic/group > index 6c3bb03a..cf8529ea 100644 > --- a/tests/generic/group > +++ b/tests/generic/group > @@ -472,3 +472,4 @@ > 467 auto quick exportfs > 468 shutdown auto quick metadata > 469 auto quick > +470 auto quick rw > -- > 2.14.2 > ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 2/3] generic/470: Test RWF_NOWAIT 2017-12-06 10:05 ` Eryu Guan @ 2017-12-06 16:35 ` Goldwyn Rodrigues 2017-12-07 4:17 ` Eryu Guan 0 siblings, 1 reply; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-12-06 16:35 UTC (permalink / raw) To: Eryu Guan; +Cc: darrick.wong, fstests, Goldwyn Rodrigues On 12/06/2017 04:05 AM, Eryu Guan wrote: > On Wed, Nov 29, 2017 at 10:33:57AM -0600, Goldwyn Rodrigues wrote: >> From: Goldwyn Rodrigues <rgoldwyn@suse.com> >> >> Tests the RWF_NOWAIT flag so the I/O returns immediately with -EAGAIN >> on a new file since it requires block allocation. >> >> It creates a file, syncs it, and overwrites the file with RWF_NOWAIT. >> This should succeed. >> >> Finally, read the contents to make sure the overwrite is successful. >> >> Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> >> >> Changes since v1: >> - Fix testdir name >> - use of $XFS_IO_PROG instead of xfs_io >> - check pwrite accepts -N >> >> Changes since v2: >> - corrected test description and improved documentation >> - new leaner look ;) >> - rw group >> --- >> tests/generic/470 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++ >> tests/generic/470.out | 13 +++++++++ >> tests/generic/group | 1 + >> 3 files changed, 88 insertions(+) >> create mode 100755 tests/generic/470 >> create mode 100644 tests/generic/470.out >> >> diff --git a/tests/generic/470 b/tests/generic/470 >> new file mode 100755 >> index 00000000..5580718b >> --- /dev/null >> +++ b/tests/generic/470 >> @@ -0,0 +1,74 @@ >> +#! /bin/bash >> +# FS QA Test No. 470 >> +# >> +# write a file with RWF_NOWAIT and it would fail because there are no >> +# blocks allocated. Create a file with direct I/O and re-write it >> +# using RWF_NOWAIT. I/O should finish within 50 microsecods since >> +# block allocations are already performed. >> +# >> +#----------------------------------------------------------------------- >> +# Copyright (c) 2017, SUSE Linux Products. 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! >> + >> +# get standard environment, filters and checks >> +. ./common/rc >> +. ./common/populate >> +. ./common/filter >> +. ./common/attr >> + >> +# real QA test starts here >> +_supported_os Linux >> +_require_odirect >> +_require_test >> +_require_xfs_io_command pwrite -N >> + >> +# Remove reminiscence of previously run tests >> +testdir=$TEST_DIR/$seq >> +if [ -e $testdir ]; then >> + rm -Rf $testdir >> +fi >> + >> +mkdir $testdir >> + >> +# Create a file with pwrite nowait (will fail with EAGAIN) >> +$XFS_IO_PROG -f -d -c "pwrite -N -V 1 -b 1M 0 1M" $testdir/f1 > > It's weird that I got "Operation not supported" message here and test > failed as > > @@ -1,5 +1,5 @@ > QA output created by 470 > -pwrite: Resource temporarily unavailable > +pwrite: Operation not supported > wrote 8388608/8388608 bytes at offset 0 > XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) > RWF_NOWAIT time is within limits. > > I was using latest for-next branch of upstream xfsprogs, and a > pre-v4.15-rc1 kernel shipped by Fedora rawhide. A simple test showed: What is the output of uname -a It should be compatible with 4.13 onward. OTOH, _require_xfs_io_command will catch that. > > rm -f /mnt/xfs/testfile > xfs_io -fdc "pwrite -N -V 1 -b 4k 0 4k" /mnt/xfs/testfile > pwrite: Operation not supported For kernels which do not support RWF_NOWAIT, it will return EOPNOTSUPP. > > But an strace run showed pwritev2 did -1 and set EAGAIN as errno, but > printed "Operation not supported" as error message: > ... > pwritev2(3, [{iov_base="\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315"..., iov_len=4096}], 1, 0, RWF_NOWAIT) = -1 EAGAIN (Resource temporarily unavailable) > dup(2) = 4 > fcntl(4, F_GETFL) = 0x402 (flags O_RDWR|O_APPEND) > openat(AT_FDCWD, "/usr/share/locale/en_US.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > openat(AT_FDCWD, "/usr/share/locale/en_US.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > openat(AT_FDCWD, "/usr/share/locale/en_US/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > openat(AT_FDCWD, "/usr/share/locale/en.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > openat(AT_FDCWD, "/usr/share/locale/en.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > openat(AT_FDCWD, "/usr/share/locale/en/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > fstat(4, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0 > write(4, "pwrite: Operation not supported\n", 32pwrite: Operation not supported > ... > > Not sure what happened yet, did I miss anything? Did you trim any lines in between? If you had another error in between before you collect -EAGAIN, it most likely overwrote the value. Not sure what is going on though.. > >> + >> +# Write the file without nowait >> +$XFS_IO_PROG -f -d -c "pwrite -S 0xaa -W -w -V 1 -b 1M 0 8M" $testdir/f1 | _filter_xfs_io > > I think "-w" (fdatasync) can be removed? We already use "-W" (fsync) here. > >> + >> +time_taken=`$XFS_IO_PROG -d -c "pwrite -S 0xbb -N -V 1 -b 1M 2M 1M" $testdir/f1 | awk '/^1/ {print $5}'` >> + >> +# RWF_NOWAIT should finish within a short period of time. Anything longer >> +# means it is waiting for something in the kernel which would be a fail. >> +if (( $(echo "$time_taken < 0.05" | bc -l) )); then > > Better to describe where does this 0.05 come from. > No significance as such, but a conservative value for the operation to be completed. If you want me to explicitly put 50 ms, I can. > Thanks, > Eryu > >> + echo "RWF_NOWAIT time is within limits." >> +else >> + echo "RWF_NOWAIT took $time_taken seconds" >> +fi >> + >> +$XFS_IO_PROG -c "pread -v 0 8M" $testdir/f1 | _filter_xfs_io_unique >> + >> +# success, all done >> +status=0 >> +exit >> diff --git a/tests/generic/470.out b/tests/generic/470.out >> new file mode 100644 >> index 00000000..e46622cb >> --- /dev/null >> +++ b/tests/generic/470.out >> @@ -0,0 +1,13 @@ >> +QA output created by 470 >> +pwrite: Resource temporarily unavailable >> +wrote 8388608/8388608 bytes at offset 0 >> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) >> +RWF_NOWAIT time is within limits. >> +00000000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ >> +* >> +00200000: bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb ................ >> +* >> +00300000: aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa ................ >> +* >> +read 8388608/8388608 bytes at offset 0 >> +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) >> diff --git a/tests/generic/group b/tests/generic/group >> index 6c3bb03a..cf8529ea 100644 >> --- a/tests/generic/group >> +++ b/tests/generic/group >> @@ -472,3 +472,4 @@ >> 467 auto quick exportfs >> 468 shutdown auto quick metadata >> 469 auto quick >> +470 auto quick rw >> -- >> 2.14.2 >> -- Goldwyn ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 2/3] generic/470: Test RWF_NOWAIT 2017-12-06 16:35 ` Goldwyn Rodrigues @ 2017-12-07 4:17 ` Eryu Guan 2017-12-07 11:57 ` Goldwyn Rodrigues 0 siblings, 1 reply; 10+ messages in thread From: Eryu Guan @ 2017-12-07 4:17 UTC (permalink / raw) To: Goldwyn Rodrigues; +Cc: darrick.wong, fstests, Goldwyn Rodrigues On Wed, Dec 06, 2017 at 10:35:06AM -0600, Goldwyn Rodrigues wrote: > > > On 12/06/2017 04:05 AM, Eryu Guan wrote: > > On Wed, Nov 29, 2017 at 10:33:57AM -0600, Goldwyn Rodrigues wrote: > >> From: Goldwyn Rodrigues <rgoldwyn@suse.com> > >> > >> Tests the RWF_NOWAIT flag so the I/O returns immediately with -EAGAIN > >> on a new file since it requires block allocation. > >> > >> It creates a file, syncs it, and overwrites the file with RWF_NOWAIT. > >> This should succeed. > >> > >> Finally, read the contents to make sure the overwrite is successful. > >> > >> Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> > >> > >> Changes since v1: > >> - Fix testdir name > >> - use of $XFS_IO_PROG instead of xfs_io > >> - check pwrite accepts -N > >> > >> Changes since v2: > >> - corrected test description and improved documentation > >> - new leaner look ;) > >> - rw group > >> --- > >> tests/generic/470 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++ > >> tests/generic/470.out | 13 +++++++++ > >> tests/generic/group | 1 + > >> 3 files changed, 88 insertions(+) > >> create mode 100755 tests/generic/470 > >> create mode 100644 tests/generic/470.out > >> > >> diff --git a/tests/generic/470 b/tests/generic/470 > >> new file mode 100755 > >> index 00000000..5580718b > >> --- /dev/null > >> +++ b/tests/generic/470 > >> @@ -0,0 +1,74 @@ > >> +#! /bin/bash > >> +# FS QA Test No. 470 > >> +# > >> +# write a file with RWF_NOWAIT and it would fail because there are no > >> +# blocks allocated. Create a file with direct I/O and re-write it > >> +# using RWF_NOWAIT. I/O should finish within 50 microsecods since > >> +# block allocations are already performed. > >> +# > >> +#----------------------------------------------------------------------- > >> +# Copyright (c) 2017, SUSE Linux Products. 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! > >> + > >> +# get standard environment, filters and checks > >> +. ./common/rc > >> +. ./common/populate > >> +. ./common/filter > >> +. ./common/attr > >> + > >> +# real QA test starts here > >> +_supported_os Linux > >> +_require_odirect > >> +_require_test > >> +_require_xfs_io_command pwrite -N > >> + > >> +# Remove reminiscence of previously run tests > >> +testdir=$TEST_DIR/$seq > >> +if [ -e $testdir ]; then > >> + rm -Rf $testdir > >> +fi > >> + > >> +mkdir $testdir > >> + > >> +# Create a file with pwrite nowait (will fail with EAGAIN) > >> +$XFS_IO_PROG -f -d -c "pwrite -N -V 1 -b 1M 0 1M" $testdir/f1 > > > > It's weird that I got "Operation not supported" message here and test > > failed as > > > > @@ -1,5 +1,5 @@ > > QA output created by 470 > > -pwrite: Resource temporarily unavailable > > +pwrite: Operation not supported > > wrote 8388608/8388608 bytes at offset 0 > > XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) > > RWF_NOWAIT time is within limits. > > > > I was using latest for-next branch of upstream xfsprogs, and a > > pre-v4.15-rc1 kernel shipped by Fedora rawhide. A simple test showed: > > What is the output of uname -a > It should be compatible with 4.13 onward. OTOH, _require_xfs_io_command > will catch that. Linux localhost 4.15.0-0.rc0.git6.1.fc28.x86_64 #1 SMP Mon Nov 20 18:08:48 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux _require_xfs_io_command doesn't run "pwrite -N" with "-V 1" option, so test there passes, it only fails with vectors specified # rm /mnt/xfs/testfile # xfs_io -fdc "pwrite -N 0 4k" /mnt/xfs/testfile wrote 4096/4096 bytes at offset 0 4 KiB, 1 ops; 0.0000 sec (787.402 KiB/sec and 196.8504 ops/sec) # rm /mnt/xfs/testfile # xfs_io -fdc "pwrite -N -V 1 0 4k" /mnt/xfs/testfile pwrite: Operation not supported That's why I was asking if we should include "-V 1" in _require_xfs_io_command in patch 1. > > > > > rm -f /mnt/xfs/testfile > > xfs_io -fdc "pwrite -N -V 1 -b 4k 0 4k" /mnt/xfs/testfile > > pwrite: Operation not supported > > For kernels which do not support RWF_NOWAIT, it will return EOPNOTSUPP. > > > > > But an strace run showed pwritev2 did -1 and set EAGAIN as errno, but > > printed "Operation not supported" as error message: > > ... > > pwritev2(3, [{iov_base="\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315"..., iov_len=4096}], 1, 0, RWF_NOWAIT) = -1 EAGAIN (Resource temporarily unavailable) > > dup(2) = 4 > > fcntl(4, F_GETFL) = 0x402 (flags O_RDWR|O_APPEND) > > openat(AT_FDCWD, "/usr/share/locale/en_US.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > openat(AT_FDCWD, "/usr/share/locale/en_US.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > openat(AT_FDCWD, "/usr/share/locale/en_US/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > openat(AT_FDCWD, "/usr/share/locale/en.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > openat(AT_FDCWD, "/usr/share/locale/en.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > openat(AT_FDCWD, "/usr/share/locale/en/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) > > fstat(4, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0 > > write(4, "pwrite: Operation not supported\n", 32pwrite: Operation not supported > > ... > > > > Not sure what happened yet, did I miss anything? > > Did you trim any lines in between? If you had another error in between > before you collect -EAGAIN, it most likely overwrote the value. Not sure > what is going on though.. No, I didn't trim any lines in between. > > > > >> + > >> +# Write the file without nowait > >> +$XFS_IO_PROG -f -d -c "pwrite -S 0xaa -W -w -V 1 -b 1M 0 8M" $testdir/f1 | _filter_xfs_io > > > > I think "-w" (fdatasync) can be removed? We already use "-W" (fsync) here. > > > >> + > >> +time_taken=`$XFS_IO_PROG -d -c "pwrite -S 0xbb -N -V 1 -b 1M 2M 1M" $testdir/f1 | awk '/^1/ {print $5}'` > >> + > >> +# RWF_NOWAIT should finish within a short period of time. Anything longer > >> +# means it is waiting for something in the kernel which would be a fail. > >> +if (( $(echo "$time_taken < 0.05" | bc -l) )); then > > > > Better to describe where does this 0.05 come from. > > > > No significance as such, but a conservative value for the operation to > be completed. If you want me to explicitly put 50 ms, I can. There should be a reason for this 50ms estimation. IMHO, as simple as "a conservative value for the operation to be completed" would be better than nothing :) Thanks, Eryu ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 2/3] generic/470: Test RWF_NOWAIT 2017-12-07 4:17 ` Eryu Guan @ 2017-12-07 11:57 ` Goldwyn Rodrigues 0 siblings, 0 replies; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-12-07 11:57 UTC (permalink / raw) To: Eryu Guan, Goldwyn Rodrigues; +Cc: darrick.wong, fstests On 12/06/2017 10:17 PM, Eryu Guan wrote: > On Wed, Dec 06, 2017 at 10:35:06AM -0600, Goldwyn Rodrigues wrote: >> >> >> On 12/06/2017 04:05 AM, Eryu Guan wrote: >>> On Wed, Nov 29, 2017 at 10:33:57AM -0600, Goldwyn Rodrigues wrote: >>>> From: Goldwyn Rodrigues <rgoldwyn@suse.com> >>>> >>>> Tests the RWF_NOWAIT flag so the I/O returns immediately with -EAGAIN >>>> on a new file since it requires block allocation. >>>> >>>> It creates a file, syncs it, and overwrites the file with RWF_NOWAIT. >>>> This should succeed. >>>> >>>> Finally, read the contents to make sure the overwrite is successful. >>>> >>>> Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> >>>> >>>> Changes since v1: >>>> - Fix testdir name >>>> - use of $XFS_IO_PROG instead of xfs_io >>>> - check pwrite accepts -N >>>> >>>> Changes since v2: >>>> - corrected test description and improved documentation >>>> - new leaner look ;) >>>> - rw group >>>> --- >>>> tests/generic/470 | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++ >>>> tests/generic/470.out | 13 +++++++++ >>>> tests/generic/group | 1 + >>>> 3 files changed, 88 insertions(+) >>>> create mode 100755 tests/generic/470 >>>> create mode 100644 tests/generic/470.out >>>> >>>> diff --git a/tests/generic/470 b/tests/generic/470 >>>> new file mode 100755 >>>> index 00000000..5580718b >>>> --- /dev/null >>>> +++ b/tests/generic/470 >>>> @@ -0,0 +1,74 @@ >>>> +#! /bin/bash >>>> +# FS QA Test No. 470 >>>> +# >>>> +# write a file with RWF_NOWAIT and it would fail because there are no >>>> +# blocks allocated. Create a file with direct I/O and re-write it >>>> +# using RWF_NOWAIT. I/O should finish within 50 microsecods since >>>> +# block allocations are already performed. >>>> +# >>>> +#----------------------------------------------------------------------- >>>> +# Copyright (c) 2017, SUSE Linux Products. 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! >>>> + >>>> +# get standard environment, filters and checks >>>> +. ./common/rc >>>> +. ./common/populate >>>> +. ./common/filter >>>> +. ./common/attr >>>> + >>>> +# real QA test starts here >>>> +_supported_os Linux >>>> +_require_odirect >>>> +_require_test >>>> +_require_xfs_io_command pwrite -N >>>> + >>>> +# Remove reminiscence of previously run tests >>>> +testdir=$TEST_DIR/$seq >>>> +if [ -e $testdir ]; then >>>> + rm -Rf $testdir >>>> +fi >>>> + >>>> +mkdir $testdir >>>> + >>>> +# Create a file with pwrite nowait (will fail with EAGAIN) >>>> +$XFS_IO_PROG -f -d -c "pwrite -N -V 1 -b 1M 0 1M" $testdir/f1 >>> >>> It's weird that I got "Operation not supported" message here and test >>> failed as >>> >>> @@ -1,5 +1,5 @@ >>> QA output created by 470 >>> -pwrite: Resource temporarily unavailable >>> +pwrite: Operation not supported >>> wrote 8388608/8388608 bytes at offset 0 >>> XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) >>> RWF_NOWAIT time is within limits. >>> >>> I was using latest for-next branch of upstream xfsprogs, and a >>> pre-v4.15-rc1 kernel shipped by Fedora rawhide. A simple test showed: >> >> What is the output of uname -a >> It should be compatible with 4.13 onward. OTOH, _require_xfs_io_command >> will catch that. > > Linux localhost 4.15.0-0.rc0.git6.1.fc28.x86_64 #1 SMP Mon Nov 20 18:08:48 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux > > _require_xfs_io_command doesn't run "pwrite -N" with "-V 1" option, so > test there passes, it only fails with vectors specified > > # rm /mnt/xfs/testfile > # xfs_io -fdc "pwrite -N 0 4k" /mnt/xfs/testfile > wrote 4096/4096 bytes at offset 0 > 4 KiB, 1 ops; 0.0000 sec (787.402 KiB/sec and 196.8504 ops/sec) > # rm /mnt/xfs/testfile > # xfs_io -fdc "pwrite -N -V 1 0 4k" /mnt/xfs/testfile > pwrite: Operation not supported > > That's why I was asking if we should include "-V 1" in > _require_xfs_io_command in patch 1. > I am all for including your suggestion in patch 1. However, it does not explain two things: 1. Why is the error code in strace changing from -EAGAIN to -EOPNOTSUPP? 2. Why is it happening on a kernel version greater than 4.14? Anyways, I shall add the changes and send it across again. >> >>> >>> rm -f /mnt/xfs/testfile >>> xfs_io -fdc "pwrite -N -V 1 -b 4k 0 4k" /mnt/xfs/testfile >>> pwrite: Operation not supported >> >> For kernels which do not support RWF_NOWAIT, it will return EOPNOTSUPP. >> >>> >>> But an strace run showed pwritev2 did -1 and set EAGAIN as errno, but >>> printed "Operation not supported" as error message: >>> ... >>> pwritev2(3, [{iov_base="\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315\315"..., iov_len=4096}], 1, 0, RWF_NOWAIT) = -1 EAGAIN (Resource temporarily unavailable) >>> dup(2) = 4 >>> fcntl(4, F_GETFL) = 0x402 (flags O_RDWR|O_APPEND) >>> openat(AT_FDCWD, "/usr/share/locale/en_US.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> openat(AT_FDCWD, "/usr/share/locale/en_US.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> openat(AT_FDCWD, "/usr/share/locale/en_US/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> openat(AT_FDCWD, "/usr/share/locale/en.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> openat(AT_FDCWD, "/usr/share/locale/en.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> openat(AT_FDCWD, "/usr/share/locale/en/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory) >>> fstat(4, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0 >>> write(4, "pwrite: Operation not supported\n", 32pwrite: Operation not supported >>> ... >>> >>> Not sure what happened yet, did I miss anything? >> >> Did you trim any lines in between? If you had another error in between >> before you collect -EAGAIN, it most likely overwrote the value. Not sure >> what is going on though.. > > No, I didn't trim any lines in between. > >> >>> >>>> + >>>> +# Write the file without nowait >>>> +$XFS_IO_PROG -f -d -c "pwrite -S 0xaa -W -w -V 1 -b 1M 0 8M" $testdir/f1 | _filter_xfs_io >>> >>> I think "-w" (fdatasync) can be removed? We already use "-W" (fsync) here. >>> >>>> + >>>> +time_taken=`$XFS_IO_PROG -d -c "pwrite -S 0xbb -N -V 1 -b 1M 2M 1M" $testdir/f1 | awk '/^1/ {print $5}'` >>>> + >>>> +# RWF_NOWAIT should finish within a short period of time. Anything longer >>>> +# means it is waiting for something in the kernel which would be a fail. >>>> +if (( $(echo "$time_taken < 0.05" | bc -l) )); then >>> >>> Better to describe where does this 0.05 come from. >>> >> >> No significance as such, but a conservative value for the operation to >> be completed. If you want me to explicitly put 50 ms, I can. > > There should be a reason for this 50ms estimation. IMHO, as simple as "a > conservative value for the operation to be completed" would be better > than nothing :) > > Thanks, > Eryu > -- Goldwyn ^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH 3/3] generic/471: Partial direct write test 2017-11-29 16:33 [PATCH 1/3] Check pwrite parameters Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 2/3] generic/470: Test RWF_NOWAIT Goldwyn Rodrigues @ 2017-11-29 16:33 ` Goldwyn Rodrigues 2017-12-06 9:55 ` [PATCH 1/3] Check pwrite parameters Eryu Guan 2 siblings, 0 replies; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-11-29 16:33 UTC (permalink / raw) To: darrick.wong, eguan; +Cc: fstests, Goldwyn Rodrigues From: Goldwyn Rodrigues <rgoldwyn@suse.com> Getting an error midway through a direct write would return an error and the error-code is returned in the write() call. However, partial data is over-written during the call. This tests simulates the ENOSPC error to check for partial direct write consistency. Changes since v1: - Check for option pwrite -O - Create a 200M file instead of multiple 40M files. Changes since v2: - new leaner look ;) - use _fill_fs after reserving some space to create an almost full FS - _require_scratch for reseting system Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> --- tests/generic/471 | 95 +++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/471.out | 9 +++++ tests/generic/group | 1 + 3 files changed, 105 insertions(+) create mode 100755 tests/generic/471 create mode 100644 tests/generic/471.out diff --git a/tests/generic/471 b/tests/generic/471 new file mode 100755 index 00000000..51723548 --- /dev/null +++ b/tests/generic/471 @@ -0,0 +1,95 @@ +#! /bin/bash +# FS QA Test No. 471 +# +# write() to a file opened with O_DIRECT with count > remaining +# bytes. In case of a bug, the write returns ENOSPC wheras the +# showing no data is written, but the file contents are updated. +# Result should be the write should return remaining bytes +# (to aligned bytes) instead of ENOSPC error. +# +#----------------------------------------------------------------------- +# Copyright (c) 2017, SUSE Linux Products. 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 -rf $tmp.* $testdir1 +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/populate +. ./common/filter +. ./common/attr + +# real QA test starts here +_supported_os Linux +_require_odirect +_require_xfs_io_command pwrite -O +_require_scratch + +rm -f $seqres.full + +echo "Reformat with 320M size" +sz_bytes=$((320 * 1024 * 1024)) +_scratch_mkfs_sized $sz_bytes >> $seqres.full 2>&1 +_scratch_mount >> $seqres.full 2>&1 +testdir=$SCRATCH_MNT/$seq +mkdir $testdir + +# Reserve some space to remove later +$XFS_IO_PROG -f -c "pwrite 0 2M" $testdir/reserve > /dev/null + +mkdir $testdir/fillspace +echo "Fill the filesystem" +_fill_fs $(( 1024 * 1024 * 1024 )) $testdir/fillspace 4096 0 > /dev/null 2>&1 + +# Remove the reserve file to work on almost full filesystem +rm $testdir/reserve +sync + +# Create a file using buffered I/O which succeeds only partially +$XFS_IO_PROG -f -c "pwrite -w -S 0xaa 0 4M" $testdir/partial + +# re-write using direct I/O with another pattern using one single buffer, once +write_size=`$XFS_IO_PROG -d -c "pwrite -O -S 0xbb -V 1 -b 4M 0 4M" $testdir/partial | awk '/^wrote/ {split($2, bytes, "/"); print bytes[1]}'` +if [ -z $write_size ]; then + write_size=0 +fi + +# If bug is not fixed, pwrite will return error and but still write data +if [ $write_size -gt 0 ]; then + echo "pwrite wrote more than zero bytes." +else + echo "pwrite wrote zero bytes" +fi + +# read the partial file to check if data written is of last write +$XFS_IO_PROG -c "pread 0 1M -v" $testdir/partial | _filter_xfs_io_unique + +# success, all done +status=0 +exit diff --git a/tests/generic/471.out b/tests/generic/471.out new file mode 100644 index 00000000..b846eb32 --- /dev/null +++ b/tests/generic/471.out @@ -0,0 +1,9 @@ +QA output created by 471 +Reformat with 320M size +Fill the filesystem +pwrite: No space left on device +pwrite wrote more than zero bytes. +00000000: bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb ................ +* +read 1048576/1048576 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) diff --git a/tests/generic/group b/tests/generic/group index cf8529ea..160e0e7b 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -473,3 +473,4 @@ 468 shutdown auto quick metadata 469 auto quick 470 auto quick rw +471 auto quick rw -- 2.14.2 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 1/3] Check pwrite parameters 2017-11-29 16:33 [PATCH 1/3] Check pwrite parameters Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 2/3] generic/470: Test RWF_NOWAIT Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 3/3] generic/471: Partial direct write test Goldwyn Rodrigues @ 2017-12-06 9:55 ` Eryu Guan 2 siblings, 0 replies; 10+ messages in thread From: Eryu Guan @ 2017-12-06 9:55 UTC (permalink / raw) To: Goldwyn Rodrigues; +Cc: darrick.wong, fstests, Goldwyn Rodrigues On Wed, Nov 29, 2017 at 10:33:56AM -0600, Goldwyn Rodrigues wrote: > From: Goldwyn Rodrigues <rgoldwyn@suse.com> > > There are some parameters added with xfs_io. Check if the pwrite > parameters are available. For some cases, xfs_io now returns "command > -%c not supported", so added "not supported" to count as error. > > Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> > > Change since v2: > - More comments > - opts to become local variable > > --- > common/rc | 14 +++++++++++++- > 1 file changed, 13 insertions(+), 1 deletion(-) > > diff --git a/common/rc b/common/rc > index 4c053a53..90408b7e 100644 > --- a/common/rc > +++ b/common/rc > @@ -2035,6 +2035,7 @@ _require_xfs_io_command() > shift > local param="$*" > local param_checked=0 > + local opts="" > > testfile=$TEST_DIR/$$.xfs_io > case $command in > @@ -2079,6 +2080,15 @@ _require_xfs_io_command() > echo $testio | grep -q "invalid option" && \ > _notrun "xfs_io $command support is missing" > ;; > + "pwrite") > + # -N (RWF_NOWAIT) only works with direct I/O writes > + if [ "$param" == "-N" ]; then > + opts+=" -d" > + fi > + testio=`$XFS_IO_PROG -f $opts -c "pwrite $param 0 1M" \ > + $testfile 2>&1` Do we need to specify the vector count when testing "-N" support? e.g. $XFS_IO_PROG -f $opts -c "pwrite $param -V 1 0 1M" ... >From xfsprogs code, it seems to me that we need to do so io/pwrite.c::do_pwrite do_pwrite( ... { if (!vectors) return pwrite(fd, ...); return do_pwritev(fd, ..., pritev2_flags); } and 'vectors' is a global variable with 0 as default value (no vector number specified via "-V N" option). So without "-V N", we call pwrite(2) not pwritev2(2) and RWF_NOWAIT doesn't get tested. Thanks, Eryu > + param_checked=1 > + ;; > "scrub"|"repair") > testio=`$XFS_IO_PROG -x -c "$command probe 0" $TEST_DIR 2>&1` > echo $testio | grep -q "Inappropriate ioctl" && \ > @@ -2109,7 +2119,9 @@ _require_xfs_io_command() > $XFS_IO_PROG -c "help $command" | grep -q "^ $param --" || \ > _notrun "xfs_io $command doesn't support $param" > else > - echo $testio | grep -q "invalid option" && \ > + # xfs_io could result in "command %c not supported" if it was > + # built on kernels not supporting pwritev2() calls > + echo $testio | grep -q "\(invalid option\|not supported\)" && \ > _notrun "xfs_io $command doesn't support $param" > fi > } > -- > 2.14.2 > ^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH 1/3] Check pwrite parameters @ 2017-11-27 18:49 Goldwyn Rodrigues 2017-11-29 9:49 ` Eryu Guan 0 siblings, 1 reply; 10+ messages in thread From: Goldwyn Rodrigues @ 2017-11-27 18:49 UTC (permalink / raw) To: fstests; +Cc: darrick.wong, Goldwyn Rodrigues From: Goldwyn Rodrigues <rgoldwyn@suse.com> There are some parameters added with xfs_io. Check if the pwrite parameters are available. For some cases, xfs_io now returns "command -%c not supported", so added "not supported" to count as error. Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> --- common/rc | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/common/rc b/common/rc index 4c053a53..b6778f07 100644 --- a/common/rc +++ b/common/rc @@ -2079,6 +2079,14 @@ _require_xfs_io_command() echo $testio | grep -q "invalid option" && \ _notrun "xfs_io $command support is missing" ;; + "pwrite") + if [ "$param" == "-N" ]; then + opts="-d" + fi + testio=`$XFS_IO_PROG -f $opts -c "pwrite $param 0 1M" \ + $testfile 2>&1` + param_checked=1 + ;; "scrub"|"repair") testio=`$XFS_IO_PROG -x -c "$command probe 0" $TEST_DIR 2>&1` echo $testio | grep -q "Inappropriate ioctl" && \ @@ -2109,7 +2117,7 @@ _require_xfs_io_command() $XFS_IO_PROG -c "help $command" | grep -q "^ $param --" || \ _notrun "xfs_io $command doesn't support $param" else - echo $testio | grep -q "invalid option" && \ + echo $testio | grep -q "\(invalid option\|not supported\)" && \ _notrun "xfs_io $command doesn't support $param" fi } -- 2.14.2 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 1/3] Check pwrite parameters 2017-11-27 18:49 Goldwyn Rodrigues @ 2017-11-29 9:49 ` Eryu Guan 0 siblings, 0 replies; 10+ messages in thread From: Eryu Guan @ 2017-11-29 9:49 UTC (permalink / raw) To: Goldwyn Rodrigues; +Cc: fstests, darrick.wong, Goldwyn Rodrigues On Mon, Nov 27, 2017 at 12:49:18PM -0600, Goldwyn Rodrigues wrote: > From: Goldwyn Rodrigues <rgoldwyn@suse.com> > > There are some parameters added with xfs_io. Check if the pwrite > parameters are available. For some cases, xfs_io now returns "command > -%c not supported", so added "not supported" to count as error. > > Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com> > --- > common/rc | 10 +++++++++- > 1 file changed, 9 insertions(+), 1 deletion(-) > > diff --git a/common/rc b/common/rc > index 4c053a53..b6778f07 100644 > --- a/common/rc > +++ b/common/rc > @@ -2079,6 +2079,14 @@ _require_xfs_io_command() > echo $testio | grep -q "invalid option" && \ > _notrun "xfs_io $command support is missing" > ;; > + "pwrite") > + if [ "$param" == "-N" ]; then > + opts="-d" I think we'd need some comments on why direct I/O is needed for "-N" option. And better to declare 'opts' as a local variable. > + fi > + testio=`$XFS_IO_PROG -f $opts -c "pwrite $param 0 1M" \ > + $testfile 2>&1` > + param_checked=1 > + ;; > "scrub"|"repair") > testio=`$XFS_IO_PROG -x -c "$command probe 0" $TEST_DIR 2>&1` > echo $testio | grep -q "Inappropriate ioctl" && \ > @@ -2109,7 +2117,7 @@ _require_xfs_io_command() > $XFS_IO_PROG -c "help $command" | grep -q "^ $param --" || \ > _notrun "xfs_io $command doesn't support $param" > else > - echo $testio | grep -q "invalid option" && \ > + echo $testio | grep -q "\(invalid option\|not supported\)" && \ Need comments on the "not supported" case, because we've checked "Operation not supported" case above, it's a bit confusing why the two "not supported" can't be merged into one. Thanks, Eryu > _notrun "xfs_io $command doesn't support $param" > fi > } > -- > 2.14.2 > > -- > To unsubscribe from this list: send the line "unsubscribe fstests" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2017-12-07 11:57 UTC | newest] Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2017-11-29 16:33 [PATCH 1/3] Check pwrite parameters Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 2/3] generic/470: Test RWF_NOWAIT Goldwyn Rodrigues 2017-12-06 10:05 ` Eryu Guan 2017-12-06 16:35 ` Goldwyn Rodrigues 2017-12-07 4:17 ` Eryu Guan 2017-12-07 11:57 ` Goldwyn Rodrigues 2017-11-29 16:33 ` [PATCH 3/3] generic/471: Partial direct write test Goldwyn Rodrigues 2017-12-06 9:55 ` [PATCH 1/3] Check pwrite parameters Eryu Guan -- strict thread matches above, loose matches on Subject: below -- 2017-11-27 18:49 Goldwyn Rodrigues 2017-11-29 9:49 ` Eryu Guan
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.