xfstests: introduce 279 for SEEK_DATA/SEEK_HOLE sanity check
PATCH v2 1/2 xfstests: introduce 279 for SEEK_DATA/SEEK_HOLE sanity check
Introduce 279 for SEEK_DATA/SEEK_HOLE sanity check.
--- 279 | 63 +++++ 279.out | 116 ++++++++ group | 1 + src/Makefile | 2 +- src/seek_sanity_tester.c | 683 ++++++++++++++++++++++++++++++++++++++++++++++ 5 files changed, 864 insertions(+), 1 deletions(-) create mode 100755 279 create mode 100644 279.out create mode 100644 src/seek_sanity_tester.c
279 new file +63 -0
@@ -0,0 +1,63 @@ +#! /bin/bash +# FS QA Test No. 279 +# +# SEEK_DATA/SEEK_HOLE sanity checker. +# +# Improved by [email protected] +# Creater: [email protected] +# +#----------------------------------------------------------------------- +# Copyright (c) 2011 Oracle Inc. All Rights Reserved. +# Copyright (c) 2011 Red Hat. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- +# +# creator [email protected] + +seq=`basename $0` +echo "QA output created by $seq" + +here=`pwd` +status=0 # success is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +# real QA test starts here + +# Modify as appropriate. +# FIXME: _supported_os should include Solaris too. +_supported_fs generic +_supported_os Linux Solaris + +base_test_path=$TEST_DIR/seek_sanity_testfile + +[ -x $here/src/seek_sanity_tester ] || _notrun "seek_sanitfy_tester not built" + +_cleanup() +{ + rm -f $base_test_path.* +} + +rm -rf $seq.out +echo "QA output created by $seq" > $seq.out +$here/src/seek_sanity_tester $base_test_path 2>&1 | tee -a $seq.out + +# success, all done +status=0 +exit
279.out new file +116 -0
116 lines not on this page · raw message
group +1 -0
@@ -389,3 +389,4 @@ deprecated 273 auto rw 274 auto rw 275 auto rw +279 auto rw
src/Makefile +1 -1
@@ -17,7 +17,7 @@ LINUX_TARGETS = xfsctl bstat t_mtab getdevicesize preallo_rw_pattern_reader \ preallo_rw_pattern_writer ftrunc trunc fs_perms testx looptest \ locktest unwritten_mmap bulkstat_unlink_test t_stripealign \ bulkstat_unlink_test_modified t_dir_offset t_futimens t_immutable \ - stale_handle pwrite_mmap_blocked fstrim t_dir_offset2 + stale_handle pwrite_mmap_blocked fstrim t_dir_offset2 seek_sanity_tester SUBDIRS =
src/seek_sanity_tester.c new file +683 -0
683 lines not on this page · raw message
Signature · 6 lines
1.7.9 _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Quoted 1 line · Jeff Liu
Introduce 279 for SEEK_DATA/SEEK_HOLE sanity check.
.....
279 line 34
Quoted 4 lines
279 new file +4
-0+echo "QA output created by $seq" + +here=`pwd` +status=0 # success is the default!
Why? failure should be the default, and it is in the new test template...
$ grep failure new status=1 # failure is the default!
Quoted 1 line
+trap "
_cleanup; exit \$status" 0 1 2 3 15
Indeed, we want the test to fail if it is interrupted.
279 line 46
Quoted 11 lines
279 new file +11
-0+ +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +# real QA test starts here + +# Modify as appropriate. +# FIXME: _supported_os should include Solaris too. +_supported_fs generic +_supported_os Linux Solaris
What is the FIXME for?
279 line 58
Quoted 11 lines
279 new file +11
-0+base_test_path=$TEST_DIR/seek_sanity_testfile + +[ -x $here/src/seek_sanity_tester ] || _notrun "seek_sanitfy_tester not built" + +_cleanup() +{ + rm -f $base_test_path.* +} + +rm -rf $seq.out +echo "QA output created by $seq" > $seq.out
Anything output on stdout automatically gets put into $seq.out by the test harness. The $seq.out file is truncated before the test, so this is not necessary. Indeed, doing this is probably why the tee command below only results in a single copy of the output in $seq.out.
Quoted 1 line
+$here/src/seek_sanity_tester $base_test_path 2>&1 | tee -a $seq.out
This only needs to be:
$here/src/seek_sanity_tester $base_test_path 2>&1
to redirect both stdout and stderr to $seq.out.
279.out line 3
Quoted 6 lines
279.out new file +3
-0@@ -0,0 +1,116 @@ +QA output created by 279 +File system supports the default behavior. +File system magic#: 0x58465342
You can't put the filesystem magic number in the output. It is different for XFS, ext4, ext3, etc. Either it needs to be removed or filtered.
279.out line 4
Quoted 1 line
279.out new file +1
-0+Allocation size: 4096
That's no good, either, as filesystems can easily return return something other than 4k there as well.
Quoted 13 lines
279.out new file +13
-0+01. Test empty file +01.01 SEEK_DATA expected -1 with errno -6, got -6. succ +01.02 SEEK_HOLE expected -1 with errno -6, got -6. succ +01.03 SEEK_HOLE expected -1 with errno -6, got -6. succ + +02. Test a tiny full file +02.01 SEEK_HOLE expected 8 or 8, got 8. succ +02.02 SEEK_DATA expected 0 or 0, got 0. succ +02.03 SEEK_DATA expected 1 or 1, got 1. succ +02.04 SEEK_HOLE expected 8 or 8, got 8. succ +02.05 SEEK_DATA expected 7 or 7, got 7. succ +02.06 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.07 SEEK_DATA expected -1 with errno -6, got -6. succ
279.out line 28
Quoted 10 lines
279.out new file +10
-0+02.08 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.09 SEEK_DATA expected -1 with errno -6, got -6. succ + +03. Test a larger full file +03.01 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.02 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.03 SEEK_DATA expected 0 or 0, got 0. succ +03.04 SEEK_DATA expected 1 or 1, got 1. succ +03.05 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.06 SEEK_DATA expected 8291 or 8291, got 8291. succ
Hmmm, these are all numbers that are based on an allocation size of 4k. So this test is guaranteed to fail on configurations that don't report a 4k block size from fstat().
I'd suggest that what you need to do here is have the test exit with a 1 or 0 to indicate success, and test for that in the 279 script, and pipe all this output to $seq.full so it can be used for debugging when a failure occurs.
Basically, if numbers can change between different test configs, then they either need to be filtered out of the golden output or directed to the $seq.full and the test does something like:
status=1 # failure is the default
rm -f $seq.full
....run_test >> $seq.full 2>&1 || _fail "run_test failed!" status=0
exit
Cheers,
Dave.
Signature · 7 lines
Dave Chinner [email protected] _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Hi Dave,
Sorry! I missed your response for this patch.
Quoted 10 lines · Dave Chinner
Quoted 1 line · Jeff Liu
Introduce 279 for SEEK_DATA/SEEK_HOLE sanity check.
.....
279 line 34
Quoted 4 lines
279 new file +4
-0+echo "QA output created by $seq" + +here=`pwd` +status=0 # success is the default!Why? failure should be the default, and it is in the new test template...
That's my mistake, I manually changed this default setting at that time somehow. :(
Quoted 10 lines
$ grep failure new status=1 # failure is the default!Quoted 1 line
+trap "
_cleanup; exit \$status" 0 1 2 3 15Indeed, we want the test to fail if it is interrupted.
Quoted 13 lines
279 line 46
Quoted 11 lines
279 new file +11
-0+ +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +# real QA test starts here + +# Modify as appropriate. +# FIXME: _supported_os should include Solaris too. +_supported_fs generic +_supported_os Linux SolarisWhat is the FIXME for?
At first, I was consider to let it also support Solaris lseek(2) tests, but xfstests only works on IRIX && Linux, so "FIXME" should be removed.
Quoted 19 lines
279 line 58
Quoted 12 lines
279 new file +11
-0+base_test_path=$TEST_DIR/seek_sanity_testfile + +[ -x $here/src/seek_sanity_tester ] || _notrun "seek_sanitfy_tester not built" + +_cleanup() +{ + rm -f $base_test_path.* +} + +rm -rf $seq.out +echo "QA output created by $seq" > $seq.outAnything output on stdout automatically gets put into $seq.out by the test harness. The $seq.out file is truncated before the test, so this is not necessary. Indeed, doing this is probably why the tee command below only results in a single copy of the output in $seq.out.
Quoted 7 lines
Quoted 1 line
+$here/src/seek_sanity_tester $base_test_path 2>&1 | tee -a $seq.out
This only needs to be:
$here/src/seek_sanity_tester $base_test_path 2>&1
to redirect both stdout and stderr to $seq.out.
Ok.
Quoted 11 lines
279.out line 3
Quoted 7 lines
279.out new file +3
-0@@ -0,0 +1,116 @@ +QA output created by 279 +File system supports the default behavior. +File system magic#: 0x58465342You can't put the filesystem magic number in the output. It is different for XFS, ext4, ext3, etc. Either it needs to be removed or filtered.
I'll remove it. :)
Quoted 25 lines
279.out line 4
Quoted 2 lines
279.out new file +1
-0+Allocation size: 4096That's no good, either, as filesystems can easily return return something other than 4k there as well.
Quoted 19 lines
279.out new file +19
-0+01. Test empty file +01.01 SEEK_DATA expected -1 with errno -6, got -6. succ +01.02 SEEK_HOLE expected -1 with errno -6, got -6. succ +01.03 SEEK_HOLE expected -1 with errno -6, got -6. succ + +02. Test a tiny full file +02.01 SEEK_HOLE expected 8 or 8, got 8. succ +02.02 SEEK_DATA expected 0 or 0, got 0. succ +02.03 SEEK_DATA expected 1 or 1, got 1. succ +02.04 SEEK_HOLE expected 8 or 8, got 8. succ +02.05 SEEK_DATA expected 7 or 7, got 7. succ +02.06 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.07 SEEK_DATA expected -1 with errno -6, got -6. succ +02.08 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.09 SEEK_DATA expected -1 with errno -6, got -6. succ + +03. Test a larger full file +03.01 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.02 SEEK_HOLE expected 8292 or 8292, got 8292. succ
Quoted 8 lines
279.out line 28
Quoted 4 lines
279.out new file +4
-0+03.03 SEEK_DATA expected 0 or 0, got 0. succ +03.04 SEEK_DATA expected 1 or 1, got 1. succ +03.05 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.06 SEEK_DATA expected 8291 or 8291, got 8291. succHmmm, these are all numbers that are based on an allocation size of 4k. So this test is guaranteed to fail on configurations that don't report a 4k block size from
fstat().
Hmm, although the numbers(like 8292, 8291) are based on an allocation size(4k in this case). But they will adjusted to other numbers depending on the detected allocation size on the fly.
For example, if a file system blksize is 8k, then the old result:
03.01 SEEK_HOLE expected 8292 or 8292, got 8292.
will show as:
03.01 SEEK_HOLE expected 16484 or 16484, got 16484.
Frankly speaking, I have not yet tried it on other allocation size, that is just the desired behavior per current implementation. I'll fix it up as you suggested if it run failed. :)
Thanks,
-Jeff
Quoted 18 lines
I'd suggest that what you need to do here is have the test exit with a 1 or 0 to indicate success, and test for that in the 279 script, and pipe all this output to $seq.full so it can be used for debugging when a failure occurs.
Basically, if numbers can change between different test configs, then they either need to be filtered out of the golden output or directed to the $seq.full and the test does something like:
status=1 # failure is the default
rm -f $seq.full
....run_test>> $seq.full 2>&1 ||_fail"run_testfailed!" status=0
exit
Quoted 3 lines
Cheers,
Dave.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 134 lines · Jeff Liu
Hi Dave,
Sorry! I missed your response for this patch.
Quoted 10 lines · Dave Chinner
Quoted 1 line · Jeff Liu
Introduce 279 for SEEK_DATA/SEEK_HOLE sanity check.
.....
279 line 34
Quoted 4 lines
279 new file +4
-0+echo "QA output created by $seq" + +here=`pwd` +status=0 # success is the default!Why? failure should be the default, and it is in the new test template...
That's my mistake, I manually changed this default setting at that time somehow. :(
Quoted 10 lines
$ grep failure new status=1 # failure is the default!Quoted 1 line
+trap "
_cleanup; exit \$status" 0 1 2 3 15Indeed, we want the test to fail if it is interrupted.
Quoted 13 lines
279 line 46
Quoted 11 lines
279 new file +11
-0+ +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +# real QA test starts here + +# Modify as appropriate. +# FIXME: _supported_os should include Solaris too. +_supported_fs generic +_supported_os Linux SolarisWhat is the FIXME for?
At first, I was consider to let it also support Solaris lseek(2) tests, but xfstests only works on IRIX && Linux, so "FIXME" should be removed.
Quoted 19 lines
279 line 58
Quoted 12 lines
279 new file +11
-0+base_test_path=$TEST_DIR/seek_sanity_testfile + +[ -x $here/src/seek_sanity_tester ] || _notrun "seek_sanitfy_tester not built" + +_cleanup() +{ + rm -f $base_test_path.* +} + +rm -rf $seq.out +echo "QA output created by $seq" > $seq.outAnything output on stdout automatically gets put into $seq.out by the test harness. The $seq.out file is truncated before the test, so this is not necessary. Indeed, doing this is probably why the tee command below only results in a single copy of the output in $seq.out.
Quoted 7 lines
Quoted 1 line
+$here/src/seek_sanity_tester $base_test_path 2>&1 | tee -a $seq.out
This only needs to be:
$here/src/seek_sanity_tester $base_test_path 2>&1
to redirect both stdout and stderr to $seq.out.
Ok.
Quoted 11 lines
279.out line 3
Quoted 7 lines
279.out new file +3
-0@@ -0,0 +1,116 @@ +QA output created by 279 +File system supports the default behavior. +File system magic#: 0x58465342You can't put the filesystem magic number in the output. It is different for XFS, ext4, ext3, etc. Either it needs to be removed or filtered.
I'll remove it. :)
Quoted 25 lines
279.out line 4
Quoted 2 lines
279.out new file +1
-0+Allocation size: 4096That's no good, either, as filesystems can easily return return something other than 4k there as well.
Quoted 19 lines
279.out new file +19
-0+01. Test empty file +01.01 SEEK_DATA expected -1 with errno -6, got -6. succ +01.02 SEEK_HOLE expected -1 with errno -6, got -6. succ +01.03 SEEK_HOLE expected -1 with errno -6, got -6. succ + +02. Test a tiny full file +02.01 SEEK_HOLE expected 8 or 8, got 8. succ +02.02 SEEK_DATA expected 0 or 0, got 0. succ +02.03 SEEK_DATA expected 1 or 1, got 1. succ +02.04 SEEK_HOLE expected 8 or 8, got 8. succ +02.05 SEEK_DATA expected 7 or 7, got 7. succ +02.06 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.07 SEEK_DATA expected -1 with errno -6, got -6. succ +02.08 SEEK_HOLE expected -1 with errno -6, got -6. succ +02.09 SEEK_DATA expected -1 with errno -6, got -6. succ + +03. Test a larger full file +03.01 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.02 SEEK_HOLE expected 8292 or 8292, got 8292. succQuoted 8 lines
279.out line 28
Quoted 4 lines
279.out new file +4
-0+03.03 SEEK_DATA expected 0 or 0, got 0. succ +03.04 SEEK_DATA expected 1 or 1, got 1. succ +03.05 SEEK_HOLE expected 8292 or 8292, got 8292. succ +03.06 SEEK_DATA expected 8291 or 8291, got 8291. succHmmm, these are all numbers that are based on an allocation size of 4k. So this test is guaranteed to fail on configurations that don't report a 4k block size from
fstat().Hmm, although the
numbers(like 8292, 8291)are based on an allocation size(4k in this case). But they will adjusted to other numbers depending on the detected allocation size on the fly.For example, if a file system blksize is 8k, then the old result:
03.01SEEK_HOLEexpected 8292 or 8292, got 8292.
will show as:
03.01SEEK_HOLEexpected 16484 or 16484, got 16484.
Quoted 3 lines
Frankly speaking, I have not yet tried it on other allocation size, that is just the desired behavior per current implementation. I'll fix it up as you suggested if it run failed. :)
Ok, I just tried on an ext4 file system with 2k blksize, looks fine, the number of expected size were shown as following:
$ ./seek_test File system magic#: 0xef53 Allocation unit: 2048 bytes File system supports the default behavior.
03. Test a larger full file
03.01 SEEK_HOLE expected 4196 or 4196, got 4196. succ 03.02 SEEK_HOLE expected 4196 or 4196, got 4196. succ 03.03 SEEK_DATA expected 0 or 0, got 0. succ 03.04 SEEK_DATA expected 1 or 1, got 1. succ 03.05 SEEK_HOLE expected 4196 or 4196, got 4196. succ 03.06 SEEK_DATA expected 4195 or 4195, got 4195. succ
Strange, I also tried to build XFS with 2k which shown as following:
$ sudo mkfs.xfs -b size=2k -n size=2k -f /dev/sda7
$ xfs_info /dev/sda7
meta-data=/dev/sda7 isize=256 agcount=4, agsize=1418736 blks
= sectsz=512 attr=2
data = bsize=2048 blocks=5674944, imaxpct=25
= sunit=0 swidth=0 blks
naming =version 2 bsize=2048 ascii-ci=0
log =internal bsize=2048 blocks=5120, version=2
= sectsz=512 sunit=0 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0But the blksize still shown as 4k, I think I must have missed something, will try it tomorrow.
Have a nice day!
-Jeff
Quoted 27 lines
Thanks,
-JeffQuoted 21 lines
I'd suggest that what you need to do here is have the test exit with a 1 or 0 to indicate success, and test for that in the 279 script, and pipe all this output to $seq.full so it can be used for debugging when a failure occurs.
Basically, if numbers can change between different test configs, then they either need to be filtered out of the golden output or directed to the $seq.full and the test does something like:
status=1 # failure is the default
rm -f $seq.full
....run_test>> $seq.full 2>&1 ||_fail"run_testfailed!" status=0
exitCheers,
Dave.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 8 lines · Jeff Liu
Strange, I also tried to build XFS with 2k which shown as following:
$ sudo mkfs.xfs -b size=2k -n size=2k -f /dev/sda7$ xfs_info /dev/sda7 meta-data=/dev/sda7 isize=256 agcount=4, agsize=1418736 blks = sectsz=512 attr=2 data = bsize=2048 blocks=5674944, imaxpct=25
^^^^^^^^^^
Quoted 2 lines
= sunit=0 swidth=0 blks naming =version 2 bsize=2048 ascii-ci=0
^^^^^^^^^^
Quoted 1 line
log =internal bsize=2048 blocks=5120, version=2
^^^^^^^^^^
The block size for data, metadata, directories and the log is 2k, just like you asked.
Quoted 2 lines
= sectsz=512 sunit=0 blks, lazy-count=1 realtime =none extsz=4096 blocks=0, rtextents=0
For the non-existent realtime device, the default is an extent size of 4k. However, extent size for the real time device is separate from the data/log device block size - it's more like the bigalloc ext4 functionality in that it defines the default extent allocation size and alignment for the RT device.
Cheers,
Dave.
Signature · 7 lines
Dave Chinner [email protected] _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Quoted 11 lines · Dave Chinner
Quoted 8 lines · Jeff Liu
Strange, I also tried to build XFS with 2k which shown as following:
$ sudo mkfs.xfs -b size=2k -n size=2k -f /dev/sda7$ xfs_info /dev/sda7 meta-data=/dev/sda7 isize=256 agcount=4, agsize=1418736 blks = sectsz=512 attr=2 data = bsize=2048 blocks=5674944, imaxpct=25^^^^^^^^^^
Quoted 8 lines
Quoted 2 lines
= sunit=0 swidth=0 blks naming =version 2 bsize=2048 ascii-ci=0^^^^^^^^^^Quoted 1 line
log =internal bsize=2048 blocks=5120, version=2
^^^^^^^^^^The block size for data, metadata, directories and the log is 2k, just like you asked.
Sorry, I mislead you.
Yes, the block size for data and metadata, etc are ok for me, but the allocate unit at "struct stat.st_blksize" is 4k, It should match data->bsize=2k IMHO.
File system magic#: 0x58465342
Allocation unit: 4096 bytes
File system supports the default behavior.
$ stat --print "%o\n" /xfs/seek_test 4096
Quoted 9 lines
Quoted 3 lines
= sectsz=512 sunit=0 blks, lazy-count=1 realtime =none extsz=4096 blocks=0, rtextents=0For the non-existent realtime device, the default is an extent size of 4k. However, extent size for the real time device is separate from the data/log device block size - it's more like the bigalloc ext4 functionality in that it defines the default extent allocation size and alignment for the RT device.
Thanks for those patiently explaining!
Cheers,
-Jeff
Quoted 4 lines
Cheers,
Dave.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 22 lines · Jeff Liu
Quoted 11 lines · Dave Chinner
Quoted 8 lines · Jeff Liu
Strange, I also tried to build XFS with 2k which shown as following:
$ sudo mkfs.xfs -b size=2k -n size=2k -f /dev/sda7$ xfs_info /dev/sda7 meta-data=/dev/sda7 isize=256 agcount=4, agsize=1418736 blks = sectsz=512 attr=2 data = bsize=2048 blocks=5674944, imaxpct=25^^^^^^^^^^Quoted 8 lines
Quoted 2 lines
= sunit=0 swidth=0 blks naming =version 2 bsize=2048 ascii-ci=0^^^^^^^^^^Quoted 1 line
log =internal bsize=2048 blocks=5120, version=2
^^^^^^^^^^The block size for data, metadata, directories and the log is 2k, just like you asked.
Quoted 5 lines
Sorry, I mislead you.
Yes, the block size for data and metadata, etc are ok for me, but the allocate unit at "struct stat.st_blksize" is 4k, It should match data->bsize=2k IMHO.
That field has nothing to do with the filesystem block size. According to the stat(2) man page:
'The st_blksize field gives the "preferred" blocksize for efficient file system I/O.'
Giving a value of less than PAGE_SIZE for this field leads to inefficient IO because it forces the page cache to do read-modify-write cycles for single filesystem block writes. Hence on a 4k page size machine, it needs to report 4k as a minimum to avoid this. On a 64k page size machine, you'll find that value is 64k.
Indeed, XFS gives you some control over what is actually reported here. If your file lies on a real-time device, then XFS will export the extent allocation size (either the mkfs default of the per inode hint if it is set) in this field. For files on the data device, if you mount with the "largeio" mount option, XFS will export the stripe width if it is set, the biosize if that mount option is used or the PAGE_SIZE if neither are set. These are all different but valid definitions of "preferred blocksize for efficient IO".
If you want to know the real block size of the filesystem, use statfs(2).
Cheers,
Dave.
Signature · 7 lines
Dave Chinner [email protected] _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Quoted 35 lines · Dave Chinner
Quoted 22 lines · Jeff Liu
Quoted 11 lines · Dave Chinner
Quoted 8 lines · Jeff Liu
Strange, I also tried to build XFS with 2k which shown as following:
$ sudo mkfs.xfs -b size=2k -n size=2k -f /dev/sda7$ xfs_info /dev/sda7 meta-data=/dev/sda7 isize=256 agcount=4, agsize=1418736 blks = sectsz=512 attr=2 data = bsize=2048 blocks=5674944, imaxpct=25^^^^^^^^^^Quoted 8 lines
Quoted 2 lines
= sunit=0 swidth=0 blks naming =version 2 bsize=2048 ascii-ci=0^^^^^^^^^^Quoted 1 line
log =internal bsize=2048 blocks=5120, version=2
^^^^^^^^^^The block size for data, metadata, directories and the log is 2k, just like you asked.
Quoted 5 lines
Sorry, I mislead you.
Yes, the block size for data and metadata, etc are ok for me, but the allocate unit at "struct stat.st_blksize" is 4k, It should match data->bsize=2k IMHO.
That field has nothing to do with the filesystem block size. According to the stat(2) man page:
'The
st_blksizefield gives the "preferred" blocksize for efficient file system I/O.'
Quoted 6 lines
Giving a value of less than
PAGE_SIZEfor this field leads to inefficient IO because it forces the page cache to do read-modify-write cycles for single filesystem block writes. Hence on a 4k page size machine, it needs to report 4k as a minimum to avoid this. On a 64k page size machine, you'll find that value is 64k.
Sigh, I was misled by EXT4's output for stat(2), since its st_blksize is 2k which is equal to the mkfs formating value even on a 4k page size machine. :(
Quoted 10 lines
Indeed, XFS gives you some control over what is actually reported here. If your file lies on a real-time device, then XFS will export the extent allocation size (either the mkfs default of the per inode hint if it is set) in this field. For files on the data device, if you mount with the "largeio" mount option, XFS will export the stripe width if it is set, the biosize if that mount option is used or the
PAGE_SIZEif neither are set. These are all different but valid definitions of "preferred blocksize for efficient IO".
Quoted 2 lines
If you want to know the real block size of the filesystem, use statfs(2).
Definitely, thanks a lot!!
-Jeff
Quoted 4 lines
Cheers,
Dave.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs