| Message ID | 20170731130956.30494-1-zlang@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Jul 31, 2017 at 09:09:56PM +0800, Zorro Lang wrote: > As posix standard, if the file offset is at or past the end of file, > no bytes are read, and read() returns zero. > > This case reads from range within EOF, past EOF and at EOF, to make > sure the return value as expected, especially read from past/at EOF > returns 0. > > Signed-off-by: Zorro Lang <zlang@redhat.com> > --- > > Hi, > > V3 did below changes: > 1) Remove truncate/fallocate/resvsp/alloc operations, only leave pwrite. > 2) Only test on sector size <= (block size / 4), generally that means > 512b sector size and 4k block size. > 3) remove all unaligned test, due to ext4 and xfs got different results > when read past EOF with unaligned offset/length. And I can get a pretty > sure expected result for this situation. > > For the 3rd situation (*unaligned* read from past EOF): > - RHEL-7.3 ext4 return 0, xfs return error (known bug). > - Upstream 4.13-rc1 ext4 return EINVAL, xfs return 0. > > I can't find any standard explain how a unaligned read from past EOF should > be. So I decide to skip this gray area for now. > This looks mostly good to me. Thanks for the rework. Just some nits... > Thanks, > Zorro > > tests/generic/450 | 149 ++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/generic/450.out | 11 ++++ > tests/generic/group | 1 + > 3 files changed, 161 insertions(+) > create mode 100755 tests/generic/450 > create mode 100644 tests/generic/450.out > > diff --git a/tests/generic/450 b/tests/generic/450 > new file mode 100755 > index 00000000..0aa2e81c > --- /dev/null > +++ b/tests/generic/450 > @@ -0,0 +1,149 @@ > +#! /bin/bash > +# FS QA Test 450 > +# > +# Test read around EOF. If the file offset is at or past the end of file, > +# no bytes are read, and read() returns zero. > +# > +#----------------------------------------------------------------------- > +# 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.* > + rm -f $tfile > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + Is ./common/filter used in this test? > +# remove previous $seqres.full before test > +rm -f $seqres.full > + > +# real QA test starts here > +_supported_fs generic > +_supported_os Linux > +_require_test > + > +tfile=$TEST_DIR/testfile_${seq} > +ssize=`_min_dio_alignment $TEST_DEV` > +bsize=`_get_block_size $TEST_DIR` > + > +if [ $ssize -gt $((bsize/4)) ]; then > + _notrun "Only test on sector size < half of block size" > +fi > + > +rm -f $tfile 2>/dev/null > + > +# check xfs_io pread result, especially for > +# Param1: expected pread return > +# Param2: expected pread count > +# Param3: expected pread offset > +# > +# If any of above values are not as expected, the output keeps > +# using the real value > +check_xfs_io_read() > +{ > + RETURN=$1 > + COUNT=$2 > + OFFSET=$3 > + > + $AWK_PROG -v ret="$RETURN" -v cnt="$COUNT" -v off="$OFFSET" ' > + /read/{ > + split($2, bytes, "/") > + > + retval=bytes[1] > + count=bytes[2] > + offset=$NF > + > + if(retval != ret || count != cnt || offset != off) > + printf("expect [%s, %s, %s], but get [%s, %s, %s]\n", ret, cnt, off, retval, count, offset) I'm not terribly familiar with the syntax here, but can we split this long line up without breaking it? We could also shorten it up a bit with an output format that looks like: expect [...], got [...] > + > + next > + } > + ' > +} > + > +# +-------------------------------------------------------+ > +# | block | block | > +# +-------------------------------------------------------+ > +# | sect | sect | sect | sect | sect | sect | sect | sect | > +# | > +# EOF > +# |<--------------- move EOF -------------->| xxxxxxxxxxx | > +# [pread1] > +# [ pread2 ] > +# [pread3] > +# [pread4] ... [pread5] > +# > +# Run below steps with differnt $operation and $openflag different > +# > +# 1) write 2 blocks (6 sectors) data to move EOF to the penultimate sector > +# 2) read (pread1) the first sector within EOF > +# 3) read (pread2) the second block contain EOF > +# 4) read (pread3) a sector at (after) EOF > +# 6) read (pread4) the last sector past EOF > +# 7) read (pread5) at far away from EOF > +# > +# NOTE: This case generally test on sector size < block size, but equal is fine Is this still valid..? > +# > +asize=$((bsize * 2)) > +tsize=$((asize - ssize * 2)) > + > +read_test() > +{ > + echo "## pread1 $oflag ##" > + $XFS_IO_PROG $oflag -c "pread 0 $ssize" $tfile | \ > + check_xfs_io_read "$ssize" "$ssize" "0" > + > + echo "## pread2 $oflag ##" > + $XFS_IO_PROG $oflag -c "pread $bsize $bsize" $tfile | \ > + check_xfs_io_read "$((bsize - ssize * 2))" "$bsize" "$bsize" An expected return value of $((tsize - bsize)) seems a bit more clear to me (i.e., file size - offset). > + > + echo "## pread3 $oflag ##" > + $XFS_IO_PROG $oflag -c "pread $tsize $ssize" $tfile | \ > + check_xfs_io_read "0" "$ssize" "$tsize" > + > + echo "## pread4 $oflag ##" > + $XFS_IO_PROG $oflag -c "pread $((asize - ssize)) $ssize" $tfile | \ $((tsize + ssize)) ? > + check_xfs_io_read "0" "$ssize" "$((asize - ssize))" > + > + echo "## pread5 $oflag ##" > + $XFS_IO_PROG $oflag -c "pread $((bsize * 100)) $ssize" $tfile | \ > + check_xfs_io_read "0" "$ssize" "$((bsize * 100))" > +} > + > +# Test buffer/direct I/O read > +for oflag in "" "-d"; do > + $XFS_IO_PROG -ft -c "pwrite 0 ${tsize}" $tfile >>$seqres.full > + read_test > +done I don't see why we need to write the file for each iteration of the loop. As it is, any writes beyond the initial iteration look like overwrites that otherwise won't modify the state of the file. Also, I think we should -c fsync the file to ensure consistent test behavior and it's slightly cleaner to pass oflag to read_test() so it's obvious how it's used. E.g.: $XFS_IO_PROG -ft -c "pwrite ..." -c fsync ... for oflag in "" "-d"; do read_test $oflag done Brian > + > +# success, all done > +status=0 > +exit > diff --git a/tests/generic/450.out b/tests/generic/450.out > new file mode 100644 > index 00000000..ce3571bc > --- /dev/null > +++ b/tests/generic/450.out > @@ -0,0 +1,11 @@ > +QA output created by 450 > +## pread1 ## > +## pread2 ## > +## pread3 ## > +## pread4 ## > +## pread5 ## > +## pread1 -d ## > +## pread2 -d ## > +## pread3 -d ## > +## pread4 -d ## > +## pread5 -d ## > diff --git a/tests/generic/group b/tests/generic/group > index a04cc900..0b29be0a 100644 > --- a/tests/generic/group > +++ b/tests/generic/group > @@ -452,3 +452,4 @@ > 447 auto quick clone > 448 auto quick rw > 449 auto quick acl enospc > +450 auto quick rw > -- > 2.13.3 > > -- > 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 -- 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
diff --git a/tests/generic/450 b/tests/generic/450 new file mode 100755 index 00000000..0aa2e81c --- /dev/null +++ b/tests/generic/450 @@ -0,0 +1,149 @@ +#! /bin/bash +# FS QA Test 450 +# +# Test read around EOF. If the file offset is at or past the end of file, +# no bytes are read, and read() returns zero. +# +#----------------------------------------------------------------------- +# 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.* + rm -f $tfile +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here +_supported_fs generic +_supported_os Linux +_require_test + +tfile=$TEST_DIR/testfile_${seq} +ssize=`_min_dio_alignment $TEST_DEV` +bsize=`_get_block_size $TEST_DIR` + +if [ $ssize -gt $((bsize/4)) ]; then + _notrun "Only test on sector size < half of block size" +fi + +rm -f $tfile 2>/dev/null + +# check xfs_io pread result, especially for +# Param1: expected pread return +# Param2: expected pread count +# Param3: expected pread offset +# +# If any of above values are not as expected, the output keeps +# using the real value +check_xfs_io_read() +{ + RETURN=$1 + COUNT=$2 + OFFSET=$3 + + $AWK_PROG -v ret="$RETURN" -v cnt="$COUNT" -v off="$OFFSET" ' + /read/{ + split($2, bytes, "/") + + retval=bytes[1] + count=bytes[2] + offset=$NF + + if(retval != ret || count != cnt || offset != off) + printf("expect [%s, %s, %s], but get [%s, %s, %s]\n", ret, cnt, off, retval, count, offset) + + next + } + ' +} + +# +-------------------------------------------------------+ +# | block | block | +# +-------------------------------------------------------+ +# | sect | sect | sect | sect | sect | sect | sect | sect | +# | +# EOF +# |<--------------- move EOF -------------->| xxxxxxxxxxx | +# [pread1] +# [ pread2 ] +# [pread3] +# [pread4] ... [pread5] +# +# Run below steps with differnt $operation and $openflag +# +# 1) write 2 blocks (6 sectors) data to move EOF to the penultimate sector +# 2) read (pread1) the first sector within EOF +# 3) read (pread2) the second block contain EOF +# 4) read (pread3) a sector at (after) EOF +# 6) read (pread4) the last sector past EOF +# 7) read (pread5) at far away from EOF +# +# NOTE: This case generally test on sector size < block size, but equal is fine +# +asize=$((bsize * 2)) +tsize=$((asize - ssize * 2)) + +read_test() +{ + echo "## pread1 $oflag ##" + $XFS_IO_PROG $oflag -c "pread 0 $ssize" $tfile | \ + check_xfs_io_read "$ssize" "$ssize" "0" + + echo "## pread2 $oflag ##" + $XFS_IO_PROG $oflag -c "pread $bsize $bsize" $tfile | \ + check_xfs_io_read "$((bsize - ssize * 2))" "$bsize" "$bsize" + + echo "## pread3 $oflag ##" + $XFS_IO_PROG $oflag -c "pread $tsize $ssize" $tfile | \ + check_xfs_io_read "0" "$ssize" "$tsize" + + echo "## pread4 $oflag ##" + $XFS_IO_PROG $oflag -c "pread $((asize - ssize)) $ssize" $tfile | \ + check_xfs_io_read "0" "$ssize" "$((asize - ssize))" + + echo "## pread5 $oflag ##" + $XFS_IO_PROG $oflag -c "pread $((bsize * 100)) $ssize" $tfile | \ + check_xfs_io_read "0" "$ssize" "$((bsize * 100))" +} + +# Test buffer/direct I/O read +for oflag in "" "-d"; do + $XFS_IO_PROG -ft -c "pwrite 0 ${tsize}" $tfile >>$seqres.full + read_test +done + +# success, all done +status=0 +exit diff --git a/tests/generic/450.out b/tests/generic/450.out new file mode 100644 index 00000000..ce3571bc --- /dev/null +++ b/tests/generic/450.out @@ -0,0 +1,11 @@ +QA output created by 450 +## pread1 ## +## pread2 ## +## pread3 ## +## pread4 ## +## pread5 ## +## pread1 -d ## +## pread2 -d ## +## pread3 -d ## +## pread4 -d ## +## pread5 -d ## diff --git a/tests/generic/group b/tests/generic/group index a04cc900..0b29be0a 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -452,3 +452,4 @@ 447 auto quick clone 448 auto quick rw 449 auto quick acl enospc +450 auto quick rw
As posix standard, if the file offset is at or past the end of file, no bytes are read, and read() returns zero. This case reads from range within EOF, past EOF and at EOF, to make sure the return value as expected, especially read from past/at EOF returns 0. Signed-off-by: Zorro Lang <zlang@redhat.com> --- Hi, V3 did below changes: 1) Remove truncate/fallocate/resvsp/alloc operations, only leave pwrite. 2) Only test on sector size <= (block size / 4), generally that means 512b sector size and 4k block size. 3) remove all unaligned test, due to ext4 and xfs got different results when read past EOF with unaligned offset/length. And I can get a pretty sure expected result for this situation. For the 3rd situation (*unaligned* read from past EOF): - RHEL-7.3 ext4 return 0, xfs return error (known bug). - Upstream 4.13-rc1 ext4 return EINVAL, xfs return 0. I can't find any standard explain how a unaligned read from past EOF should be. So I decide to skip this gray area for now. Thanks, Zorro tests/generic/450 | 149 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/450.out | 11 ++++ tests/generic/group | 1 + 3 files changed, 161 insertions(+) create mode 100755 tests/generic/450 create mode 100644 tests/generic/450.out