This patch adds punch hole tests that are based off the test scripts that were
used to test and develop punch hole for ext4. The tests use the new common
routines in common.punch to run through various corners cases for punching holes.
Signed-off-by: Allison Henderson <[email protected]>
---
:000000 100755 0000000... 32e8224... A 254
:000000 100644 0000000... 020e6da... A 254.out
:100644 100644 9f88e75... 29954dc... M group
254 | 293 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
254.out | 27 ++++++
group | 1 +
3 files changed, 321 insertions(+), 0 deletions(-)
diff --git a/254 b/254
new file mode 100755
index 0000000..32e8224
--- /dev/null
+++ b/254
@@ -0,0 +1,293 @@
+#! /bin/bash
+# FS QA Test No. 254
+#
+# Test hole punching through fallocate
+#
+# This test is a collection of corner case tests for
+# hole punching through fallocate
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2010 IBM Corporation. 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`
+tmp=/tmp/$$
+status=1 # failure is the default!
+
+_cleanup()
+{
+ rm -f $tmp.*
+}
+
+trap "_cleanup ; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common.rc
+. ./common.filter
+. ./common.punch
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+
+_require_xfs_io_falloc_punch
+
+testfile=$TEST_DIR/254.$$
+
+# Small Hole Test
+#
+# In this test a hole is punched in a 10K file from bytes
+# 8180 to 16216 resulting in some blocks being removed, and
+# the non block aligned data to be cleared.
+#
+# This test is a general purpose punch hole test to verify
+# that the code can successfully split the extents, and
+# remove the affected blocks, as well as clearing the
+# associated pages
+#----------------------------------------------------------------------
+echo Running Small Hole Test
+file_len=40960
+hole_offset=8180
+hole_len=8180
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo Small Hole Test OK
+
+# Big Hole and Hole Removal Test
+#
+# A hole large hole is punched in a large file,
+# Resulting in all but a few blocks being punched out
+# (exact number depends on the block size of the filesystem).
+# After the hole is punched out, the file is deleted
+#
+# This verifies that the code can properly
+# punch out a hole covering multiple extents, and
+# also ensure the all of the extents are properly
+# removed when the file is deleted.
+#
+# This test is successful when the following conditions are met:
+# - df shows that the number of available blocks is the same before the test file is copied
+# into the file system, and after the file is deleted
+#----------------------------------------------------------------------
+echo Running Big Hole and Hole Removal Test
+file_len=638162710
+hole_offset=6144
+hole_len=638150422
+
+block_size=`stat -f $TEST_DEV | grep "Block size" | cut -d " " -f3`
+
+rm $testfile*
+
+#record how many blocks free blocks the filesystem has
+fs_used_before_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $3 }'`
+fs_avail_before_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $4 }'`
+
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+rm $testfile*
+
+#record how many blocks free blocks the filesystem has after the test
+fs_used_after_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $3 }'`
+fs_avail_after_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $4 }'`
+
+if [ $fs_used_before_test != $fs_used_after_test ] || \
+ [ $fs_avail_before_test != $fs_avail_after_test ]; then
+
+ echo Error: File system did not reclaim the correct number of blocks
+ echo The number of used and available blocks should be the same before and after the test
+ echo Used blocks before test: $fs_used_before_test
+ echo Available blocks before test: $fs_avail_before_test
+ echo Used blocks after test: $fs_used_after_test
+ echo Available blocks after test: $fs_avail_after_test
+ status=1 ; exit
+else
+ echo Big Hole and Hole Removal Test OK
+fi
+
+# Last Block Hole Test
+#
+# This test cases punches a hole in 10K file from bytes 32768 to 40960
+# resulting in a hole that extends to the end of the file. This test
+# verifies that the code appropriately handles removing blocks at the
+# end of the file
+#----------------------------------------------------------------------
+echo Running Last Block Hole Test
+file_len=40960
+hole_offset=32768
+hole_len=8192
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo Last Block Hole Test OK
+
+# First Block Hole Test
+#
+# This test cases punches a hole in 10K file from bytes 0 to 8180
+# This test verifies that the code appropriately handles punching
+# holes appearing at the beginning of the file.
+#----------------------------------------------------------------------
+echo Running First Block Hole Test
+file_len=40960
+hole_offset=0
+hole_len=8180
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo First Block Hole Test OK
+
+# Cold Cache Hole Test
+#
+# This test is similar to the Small Hole Test, but the drive is
+# unmounted and remounted before the hole is punched out to make
+# sure that there are no pages in the cache. This test verifies
+# that the hole contains zeros even when the pages are cold.
+#----------------------------------------------------------------------
+echo Running Cold Cache Hole Test
+file_len=40960
+hole_offset=8180
+hole_len=8180
+_hole_test -m -s -c $file_len $testfile $hole_offset $hole_len
+echo Cold Cache Hole Test OK
+
+# Extended Last Block Hole Test
+#
+# This test cases punches a hole in 10K file from bytes 32768 to 45056
+# resulting in a hole that extends beyond the end of the file. This
+# test verifies that the code appropriately handles punching holes
+# that extend beyond the end of the file
+#----------------------------------------------------------------------
+echo Running Last Block Hole Test
+file_len=40960
+hole_offset=32768
+hole_len=12288
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo Last Block Hole Test OK
+
+# Off Edge Hole Test
+#
+# This test case punches a hole that is beyond the edge of the file.
+# This test is successful when ls, df, filefrag and the file contents
+# show no changes
+#----------------------------------------------------------------------
+echo Running Off Edge Hole Test
+file_len=40960
+hole_offset=49052
+hole_len=8180
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo Off Edge Hole Test OK
+
+# One Block Hole Test
+#
+# This test punches a hole in a file that is only one block in size
+#----------------------------------------------------------------------
+echo Running One Block Hole Test
+block_size=`stat -f $TEST_DEV | grep "Block size" | cut -d " " -f3`
+file_len=$block_size
+hole_offset=$(( $block_size / 4 ))
+hole_len=$(( $hole_offset * 2 ))
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+echo One Block Hole Test OK
+
+# Delayed Allocation Hole Test
+#
+# This test is similar to the Small Hole Test, but the file is not sync'd
+# to the disk before the hole is punched out. This will test the codes
+# ability to punch holes in delayed extents and also ensure that dirty
+# pages are zero'd properly
+#----------------------------------------------------------------------
+echo Running Delayed Allocation Hole Test
+file_len=40960
+hole_offset=8180
+hole_len=8180
+_hole_test -c $file_len $testfile $hole_offset $hole_len
+echo Delayed Allocation Hole Test OK
+
+# Pre Allocation Hole Test
+#
+# This test is similar to the Small Hole Test, but the file is
+# preallocated before the hole is punched out.
+#----------------------------------------------------------------------
+echo Running Pre Allocation Hole Test
+file_len=40960
+hole_offset=8180
+hole_len=8180
+_hole_test -p $file_len $testfile $hole_offset $hole_len
+echo Pre Allocation Hole Test OK
+
+# Multi Hole Test
+#
+# This test verifies the code ability to handle existing holes
+# The test punches out 5 consecutive overlapping holes in a 10K file
+#----------------------------------------------------------------------
+echo Running Multi Hole Test
+file_len=40960
+hole_offset=8192
+hole_len=8192
+
+_hole_test -c $file_len $testfile $hole_offset $hole_len
+
+hole_offset=8192
+hole_len=8192
+_hole_test $testfile $hole_offset $hole_len
+
+hole_offset=8180
+hole_len=8192
+_hole_test $testfile $hole_offset $hole_len
+
+hole_offset=8204
+hole_len=8192
+_hole_test $testfile $hole_offset $hole_len
+
+hole_offset=12288
+hole_len=12288
+_hole_test $testfile $hole_offset $hole_len
+
+echo Multi Hole Test OK
+
+
+# Multi Hole Test Front
+#
+# This test verifies the codes ability to handle existing holes
+# at the front of the file.
+#----------------------------------------------------------------------
+echo Running Multi Hole Test Front
+file_len=40960
+hole_offset=0
+hole_len=8192
+_hole_test -s -c $file_len $testfile $hole_offset $hole_len
+_hole_test $testfile $hole_offset $hole_len
+echo Multi Hole Test Front OK
+
+
+# Multi Hole Test Delayed
+#
+# This test verifies the code ability to handle existing holes
+# in delayed blocks. This test is similar to the Delayed Allocation
+# Hole Test, but the hole is punched twice
+#----------------------------------------------------------------------
+echo Running Multi Hole Test Delayed
+file_len=40960
+hole_offset=8180
+hole_len=8180
+_hole_test -s -p $file_len $testfile $hole_offset $hole_len
+_hole_test $testfile $hole_offset $hole_len
+echo Multi Hole Test Delayed OK
+
+status=0 ; exit
+
+
+
diff --git a/254.out b/254.out
new file mode 100644
index 0000000..020e6da
--- /dev/null
+++ b/254.out
@@ -0,0 +1,27 @@
+QA output created by 254
+Running Small Hole Test
+Small Hole Test OK
+Running Big Hole and Hole Removal Test
+Big Hole and Hole Removal Test OK
+Running Last Block Hole Test
+Last Block Hole Test OK
+Running First Block Hole Test
+First Block Hole Test OK
+Running Cold Cache Hole Test
+Cold Cache Hole Test OK
+Running Last Block Hole Test
+Last Block Hole Test OK
+Running Off Edge Hole Test
+Off Edge Hole Test OK
+Running One Block Hole Test
+One Block Hole Test OK
+Running Delayed Allocation Hole Test
+Delayed Allocation Hole Test OK
+Running Pre Allocation Hole Test
+Pre Allocation Hole Test OK
+Running Multi Hole Test
+Multi Hole Test OK
+Running Multi Hole Test Front
+Multi Hole Test Front OK
+Running Multi Hole Test Delayed
+Multi Hole Test Delayed OK
diff --git a/group b/group
index 9f88e75..29954dc 100644
--- a/group
+++ b/group
@@ -367,3 +367,4 @@ deprecated
251 ioctl trim
252 auto quick prealloc
253 auto quick
+254 other
--
1.7.1
On Wed, May 18, 2011 at 01:58:45PM -0700, Allison Henderson wrote:
> This patch adds punch hole tests that are based off the test scripts that were
> used to test and develop punch hole for ext4. The tests use the new common
> routines in common.punch to run through various corners cases for punching holes.
It's really just a slightly different set of corner cases to what
252 tests, and there's quite a bit of overlap. What your tests don't
do is test different combinations of extent type conversions and he
splits and merges which, historically, have been the major source of
bugs with hole punching...
Adding your various cases to 252 (generic_test_punch) would be a much
better idea, I think.
With that in mind:
> +# Small Hole Test
Covered by 252, case 2 "into allocated space".
> +# Big Hole and Hole Removal Test
Covered by 252, cases 11, 12 and 13. it doesn't need to write 600MB
files to do this, and in the case of XFS, a 600MB file is still only
a single extent, so the test is behaving according to your desire on
XFS.
> +# also ensure the all of the extents are properly
> +# removed when the file is deleted.
That's covered by the removing of the test file between test
cases in 252.
> +#record how many blocks free blocks the filesystem has after the test
> +fs_used_after_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $3 }'`
> +fs_avail_after_test=`df --block-size=$block_size $TEST_DEV | grep $TEST_DEV | awk '{ print $4 }'`
> +
> +if [ $fs_used_before_test != $fs_used_after_test ] || \
> + [ $fs_avail_before_test != $fs_avail_after_test ]; then
> +
> + echo Error: File system did not reclaim the correct number of blocks
> + echo The number of used and available blocks should be the same before and after the test
> + echo Used blocks before test: $fs_used_before_test
> + echo Available blocks before test: $fs_avail_before_test
> + echo Used blocks after test: $fs_used_after_test
> + echo Available blocks after test: $fs_avail_after_test
> + status=1 ; exit
> +else
> + echo Big Hole and Hole Removal Test OK
> +fi
Again, the filesystem free block count tests are simply not
reliable. I'm pretty sure that this simply won't work on btrfs....
> +# Last Block Hole Test
Not directly handled by 252. Easily added, though.
echo " 14. data -> hole @ EOF"
rm -f $testfile
$XFS_IO_PROG $xfs_io_opt -f -c "truncate 20k" \
-c "pwrite 0 20k" -c "fsync" \
-c "$zero_cmd 12k 8k" \
-c "$map_cmd -v" $testfile | $filter_cmd
[ $? -ne 0 ] && die_now
> +# First Block Hole Test
Same here
echo " 15. data -> hole @ 0"
rm -f $testfile
$XFS_IO_PROG $xfs_io_opt -f -c "truncate 20k" \
-c "pwrite 0 20k" -c "fsync" \
-c "$zero_cmd 0k 8k" \
-c "$map_cmd -v" $testfile | $filter_cmd
[ $? -ne 0 ] && die_now
> +# Cold Cache Hole Test
> +#
> +# This test is similar to the Small Hole Test, but the drive is
> +# unmounted and remounted before the hole is punched out to make
> +# sure that there are no pages in the cache. This test verifies
> +# that the hole contains zeros even when the pages are cold.
> +#----------------------------------------------------------------------
> +echo Running Cold Cache Hole Test
> +file_len=40960
> +hole_offset=8180
> +hole_len=8180
> +_hole_test -m -s -c $file_len $testfile $hole_offset $hole_len
> +echo Cold Cache Hole Test OK
Not directly tested in 252, but easily added via:
echo " 15. data -> cache cold ->hole"
rm -f $testfile
rm -f $testfile.2
$XFS_IO_PROG $xfs_io_opt -f -c "truncate 20k" \
-c "pwrite 8k 12k" -c "fsync" $testfile.2
$XFS_IO_PROG $xfs_io_opt -f -c "truncate 20k" \
-c "pwrite 0 20k" -c "fsync" \
-c "$zero_cmd 0k 8k" \
-c "fadvise -d" \
-c "$map_cmd -v" $testfile | $filter_cmd
diff $testfile $testfile.2
[ $? -ne 0 ] && die_now
rm -f $testfile.2
> +# Extended Last Block Hole Test
> +#
> +# This test cases punches a hole in 10K file from bytes 32768 to 45056
> +# resulting in a hole that extends beyond the end of the file. This
> +# test verifies that the code appropriately handles punching holes
> +# that extend beyond the end of the file
> +#----------------------------------------------------------------------
> +echo Running Last Block Hole Test
> +file_len=40960
> +hole_offset=32768
> +hole_len=12288
> +_hole_test -s -c $file_len $testfile $hole_offset $hole_len
> +echo Last Block Hole Test OK
That's effectively a repeat of the "data -> hole @ EOF" test.
Realistically, this test needs to be run with preallocation beyond
EOF to show that it can punch extents beyond EOF....
> +# Off Edge Hole Test
Just using a non-block aligned start hole offset in the above test
will cover this one as well. No need for multiple tests that
effectively do the same thing.
> +# One Block Hole Test
If anyone is doing this in an application, they deserve what they
get :/
echo " 16. data -> hole in single block file"
rm -f $testfile
$XFS_IO_PROG $xfs_io_opt -f -c "truncate 512" \
-c "pwrite 0 512" -c "fsync" \
-c "$zero_cmd 128 128" \
-c "$map_cmd -v" $testfile | $filter_cmd
[ $? -ne 0 ] && die_now
> +# Delayed Allocation Hole Test
> +#
> +# This test is similar to the Small Hole Test, but the file is not sync'd
> +# to the disk before the hole is punched out. This will test the codes
> +# ability to punch holes in delayed extents and also ensure that dirty
> +# pages are zero'd properly
> +#----------------------------------------------------------------------
> +echo Running Delayed Allocation Hole Test
> +file_len=40960
> +hole_offset=8180
> +hole_len=8180
> +_hole_test -c $file_len $testfile $hole_offset $hole_len
> +echo Delayed Allocation Hole Test OK
If you want to test this properly, you need to run all the 252 tests
without the fsync after the data write so the extents are still
delalloc before the other operations are done. i.e. make the fsync
command a variable like map_cmd and zero_cmd.
> +
> +# Pre Allocation Hole Test
Already covered by case 3 "into unwritten space".
> +# Multi Hole Test
Far better to redo the whole punch series without removing the file
in-between the individual tests. That will cover far more edge
conditions that "same hole, 12 bytes lower, 12 bytes higher and 4k
higher." As it is, the last test is already covered by 252.
> +# Multi Hole Test Front
And in doing so would cover this case.
> +# Multi Hole Test Delayed
and this case without needing any extra work.
And the good part about extending 252 (from my persepctive) is that
test 242 (for the XFS_IOC_ZERO_RANGE ioctl) will also get better
coverage. And if I add the fallocate zero-range equivalent, it will
also be easy to test for all these cases, too, just by running
test_generic_punch with a different zero command....
Cheers,
Dave.
--
Dave Chinner
[email protected]