| Message ID | 1535085644-23131-1-git-send-email-amir73il@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v3] overlay/016: split into two tests | expand |
On Fri, Aug 24, 2018 at 7:38 AM Amir Goldstein <amir73il@gmail.com> wrote: > > overlay/016 tests two cases of ro/rw fd data inconsistecies - > one using pread and one using mmap read (i.e. mread). > > The first case now passes with stacked overlay file operations > patch set merged. The second case will still fail. > > By splitting the two test cases we get one regression test for > the common case of ro/rw fd data inconsistecy with pread and > one test to track the remaining non-standard behavior of > overlayfs w.r.t mmap. > > Signed-off-by: Amir Goldstein <amir73il@gmail.com> > --- > > Eryu, > > The overlayfs ro/rw fd data inconsistecies patches were > finally merged, so we can now merge this test change to > reflect new behavior. > > After this change on current master: > - overlay/16 is expected to pass > - overlay/61 is expected to fail > > FYI, overlay/60 (new metacopy feature test) is also expected to > run and pass on master. > Eryu, More FYI from testing overlayfs over ext4/xfs on master: - With base fs xfs, swap group tests crash the kernel - sent out a fix - With base fs xfs, overlay/019 stress test hits lockdep "circular dependency" and "downgrading a read lock" warnings - it seems like a 4.18-rc1 regression I will need more time to investigate - With base fs ext4, the 6 tests that do "_test_generic_punch -d" fail - all expect for generic/009 are regressions from this cycle The above generic tests failures are the only failures I observed with "check -overlay -g quick" apart from test that also fail on base fs (e.g. generic/484). Thanks, Amir.
On Fri, Aug 24, 2018 at 01:51:04PM +0300, Amir Goldstein wrote: > On Fri, Aug 24, 2018 at 7:38 AM Amir Goldstein <amir73il@gmail.com> wrote: > > > > overlay/016 tests two cases of ro/rw fd data inconsistecies - > > one using pread and one using mmap read (i.e. mread). > > > > The first case now passes with stacked overlay file operations > > patch set merged. The second case will still fail. > > > > By splitting the two test cases we get one regression test for > > the common case of ro/rw fd data inconsistecy with pread and > > one test to track the remaining non-standard behavior of > > overlayfs w.r.t mmap. > > > > Signed-off-by: Amir Goldstein <amir73il@gmail.com> > > --- > > > > Eryu, > > > > The overlayfs ro/rw fd data inconsistecies patches were > > finally merged, so we can now merge this test change to > > reflect new behavior. > > > > After this change on current master: > > - overlay/16 is expected to pass > > - overlay/61 is expected to fail > > > > FYI, overlay/60 (new metacopy feature test) is also expected to > > run and pass on master. > > > > Eryu, > > More FYI from testing overlayfs over ext4/xfs on master: > - With base fs xfs, swap group tests crash the kernel - sent out a fix > - With base fs xfs, overlay/019 stress test hits lockdep "circular dependency" > and "downgrading a read lock" warnings - it seems like a 4.18-rc1 regression > I will need more time to investigate > - With base fs ext4, the 6 tests that do "_test_generic_punch -d" fail - > all expect for generic/009 are regressions from this cycle > > The above generic tests failures are the only failures I observed with > "check -overlay -g quick" apart from test that also fail on base fs > (e.g. generic/484). Thanks a lot for the detailed information! Eryu
On Sun, Aug 26, 2018 at 3:39 PM Eryu Guan <guaneryu@gmail.com> wrote: > > On Fri, Aug 24, 2018 at 01:51:04PM +0300, Amir Goldstein wrote: > > On Fri, Aug 24, 2018 at 7:38 AM Amir Goldstein <amir73il@gmail.com> wrote: > > > > > > overlay/016 tests two cases of ro/rw fd data inconsistecies - > > > one using pread and one using mmap read (i.e. mread). > > > > > > The first case now passes with stacked overlay file operations > > > patch set merged. The second case will still fail. > > > > > > By splitting the two test cases we get one regression test for > > > the common case of ro/rw fd data inconsistecy with pread and > > > one test to track the remaining non-standard behavior of > > > overlayfs w.r.t mmap. > > > > > > Signed-off-by: Amir Goldstein <amir73il@gmail.com> > > > --- > > > > > > Eryu, > > > > > > The overlayfs ro/rw fd data inconsistecies patches were > > > finally merged, so we can now merge this test change to > > > reflect new behavior. > > > > > > After this change on current master: > > > - overlay/16 is expected to pass > > > - overlay/61 is expected to fail > > > > > > FYI, overlay/60 (new metacopy feature test) is also expected to > > > run and pass on master. > > > > > > > Eryu, > > > > More FYI from testing overlayfs over ext4/xfs on master: Some updates: > > - With base fs xfs, swap group tests crash the kernel - sent out a fix v2 of ovl fixes will be posted shortly. v2 doesn't fix overlayfs swap file support - it disables it - so the swap group is due to not run if the fix is merged. > > - With base fs xfs, overlay/019 stress test hits lockdep "circular dependency" > > and "downgrading a read lock" warnings - it seems like a 4.18-rc1 regression > > I will need more time to investigate > > - With base fs ext4, the 6 tests that do "_test_generic_punch -d" fail - > > all expect for generic/009 are regressions from this cycle Will post an ovl fix shortly. generic/009 was failing on overlay over ext4 because the test is not "overlay friendly" (it has "$FSTYP" = "ext4") - will send a fix. > > > > The above generic tests failures are the only failures I observed with > > "check -overlay -g quick" apart from test that also fail on base fs > > (e.g. generic/484). > FYI2, there are ~150 generic tests that are notrun when running overlay over xfs-reflink compared to xfs-reflink for various reasons (mostly _require_block_device), but found one test generic/474 that is notrun on overlay over ext4 that is not expected and that test was written specifically to cover an overlayfs bug fix. Will follow up on that. Thanks, Amir.
diff --git a/tests/overlay/016 b/tests/overlay/016 index 5c5d9ab..4fcb31e 100755 --- a/tests/overlay/016 +++ b/tests/overlay/016 @@ -43,13 +43,11 @@ _scratch_mkfs >>$seqres.full 2>&1 lowerdir=$OVL_BASE_SCRATCH_MNT/$OVL_LOWER mkdir -p $lowerdir echo "This is old news" > $lowerdir/foo -echo "This is old news" > $lowerdir/bar _scratch_mount # -# case #1: # open file for read (rofd) # open file for write (rwfd) # write to rwfd @@ -62,18 +60,5 @@ $XFS_IO_PROG -r $SCRATCH_MNT/foo \ -C "pread -v 0 16" \ | _filter_xfs_io | _filter_scratch -# -# case #2: -# mmap MAP_SHARED|PROT_READ of rofd -# write to rwfd -# read from mapped memory -# -$XFS_IO_PROG -r $SCRATCH_MNT/bar \ - -C "mmap -r 0 16" \ - -C "open $SCRATCH_MNT/bar" \ - -C "pwrite -S 0x61 0 16" \ - -C "mread -v 0 16" \ -| _filter_xfs_io | _filter_scratch - status=0 exit diff --git a/tests/overlay/016.out b/tests/overlay/016.out index a28aaa6..02b2ae2 100644 --- a/tests/overlay/016.out +++ b/tests/overlay/016.out @@ -6,6 +6,3 @@ XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) 00000000: 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 aaaaaaaaaaaaaaaa read 16/16 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) -wrote 16/16 bytes at offset 0 -XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) -00000000: 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 aaaaaaaaaaaaaaaa diff --git a/tests/overlay/061 b/tests/overlay/061 new file mode 100755 index 0000000..4b653ab --- /dev/null +++ b/tests/overlay/061 @@ -0,0 +1,65 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (C) 2018 CTERA Networks. All Rights Reserved. +# +# FSQA Test No. 061 +# +# Test ro/rw fd data inconsistecies +# +# This is a variant of test overlay/016 with mread instead of pread. +# +# This simple test demonstrates a known issue with overlayfs: +# - process A opens file F for read + mmap +# - process B writes new data to file F +# - process A reads old data from mapped memory +# +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs overlay +_supported_os Linux +_require_scratch +_require_xfs_io_command "open" + +rm -f $seqres.full + +_scratch_mkfs >>$seqres.full 2>&1 + +# Create our test files. +lowerdir=$OVL_BASE_SCRATCH_MNT/$OVL_LOWER +mkdir -p $lowerdir +echo "This is old news" > $lowerdir/foo + +_scratch_mount + + +# +# mmap MAP_SHARED|PROT_READ of rofd +# write to rwfd +# read from mapped memory +# +$XFS_IO_PROG -r $SCRATCH_MNT/foo \ + -C "mmap -r 0 16" \ + -C "open $SCRATCH_MNT/foo" \ + -C "pwrite -S 0x61 0 16" \ + -C "mread -v 0 16" \ +| _filter_xfs_io | _filter_scratch + +status=0 +exit diff --git a/tests/overlay/061.out b/tests/overlay/061.out new file mode 100644 index 0000000..0662721 --- /dev/null +++ b/tests/overlay/061.out @@ -0,0 +1,4 @@ +QA output created by 061 +wrote 16/16 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +00000000: 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 61 aaaaaaaaaaaaaaaa diff --git a/tests/overlay/group b/tests/overlay/group index b73d7e9..ccc71f3 100644 --- a/tests/overlay/group +++ b/tests/overlay/group @@ -63,3 +63,4 @@ 058 auto quick exportfs 059 auto quick copyup 060 auto quick metacopy +061 auto quick copyup
overlay/016 tests two cases of ro/rw fd data inconsistecies - one using pread and one using mmap read (i.e. mread). The first case now passes with stacked overlay file operations patch set merged. The second case will still fail. By splitting the two test cases we get one regression test for the common case of ro/rw fd data inconsistecy with pread and one test to track the remaining non-standard behavior of overlayfs w.r.t mmap. Signed-off-by: Amir Goldstein <amir73il@gmail.com> --- Eryu, The overlayfs ro/rw fd data inconsistecies patches were finally merged, so we can now merge this test change to reflect new behavior. After this change on current master: - overlay/16 is expected to pass - overlay/61 is expected to fail FYI, overlay/60 (new metacopy feature test) is also expected to run and pass on master. Thanks, Amir. Changes from v2: - remove unimplemented feature copy_up_shared tests/overlay/016 | 15 ------------ tests/overlay/016.out | 3 --- tests/overlay/061 | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++ tests/overlay/061.out | 4 ++++ tests/overlay/group | 1 + 5 files changed, 70 insertions(+), 18 deletions(-) create mode 100755 tests/overlay/061 create mode 100644 tests/overlay/061.out