All of lore.kernel.org
 help / color / mirror / Atom feed
From: Chengguang Xu <cgxu519@icloud.com>
To: Eryu Guan <eguan@redhat.com>, Amir Goldstein <amir73il@gmail.com>
Cc: fstests <fstests@vger.kernel.org>,
	overlayfs <linux-unionfs@vger.kernel.org>
Subject: Re: [PATCH v3 1/3] common/rc: add scratch shutdown support for overlayfs
Date: Wed, 03 Jan 2018 20:44:40 +0800	[thread overview]
Message-ID: <DB72F77F-23E3-4F5E-93CB-F61B94F9EA86@icloud.com> (raw)
In-Reply-To: <20171212133911.GU2749@eguan.usersys.redhat.com>


> 在 2017年12月12日,下午9:39,Eryu Guan <eguan@redhat.com> 写道:
> 
> On Tue, Dec 12, 2017 at 06:20:51PM +0800, Chengguang Xu wrote:
>>> 
>>> 在 2017年12月12日,下午5:16,Eryu Guan <eguan@redhat.com> 写道:
>>> 
>>> On Tue, Dec 12, 2017 at 11:09:37AM +0800, Chengguang Xu wrote:
>>>> OverlayFS overrides SCRATCH_MNT variable to it's own,
>>>> so adjust target to underlying filesystem(OVL_BASE_SCRATCH_MNT)
>>>> when running shutdown on overlayfs.
>>>> 
>>>> Signed-off-by: Chengguang Xu <cgxu519@icloud.com>
>>>> ---
>>>> 
>>>> Changes since v2:
>>>> 1. Make option for _scratch_shutdown(), so caller can 
>>>> specify option when calling.
>>>> 2. Add comment for why overlay requires special handling.
>>>> 3. Add commit log.
>>>> 
>>>> Changes since v1:
>>>> _scratch_shutdown() does not call notrun.
>>>> 
>>>> common/rc | 37 +++++++++++++++++++++++++++++++++++--
>>>> 1 file changed, 35 insertions(+), 2 deletions(-)
>>>> 
>>>> diff --git a/common/rc b/common/rc
>>>> index 4c053a5..a5c4ace 100644
>>>> --- a/common/rc
>>>> +++ b/common/rc
>>>> @@ -382,6 +382,24 @@ _scratch_cycle_mount()
>>>>    _scratch_mount "$opts"
>>>> }
>>>> 
>>>> +_scratch_shutdown()
>>>> +{
>>>> +	if [ $FSTYP = "overlay" ]; then
>>>> +		# In lagacy overlay usage, it may specify directory as 
>>> 
>>> Trailing whitespace in above line.
>>> 
>>>> +		# SCRATCH_DEV, in this case OVL_BASE_SCRATCH_DEV
>>>> +		# will be null, so check OVL_BASE_SCRATCH_DEV before
>>>> +		# running shutdown to avoid shutting down base fs accidently.
>>>> +		if [ -z $OVL_BASE_SCRATCH_DEV ]; then
>>>> +			echo -n "Ignore shutdown because "
>>>> +			echo "$SCRATCH_DEV is not a block device"
>>> 
>>> Hmm, I think a '_fail "<message>"' would be more appropriate here, so
>>> that test fails and exits immediately. Test should call
>>> _require_scratch_shutdown first before calling _scratch_shutdown, if
>>> not, it indicates a useage error of the helper, we need to prompt the
>>> test writer to call _require_scratch_shutdown first in error message.
>>> 
>>>> +		else
>>>> +			src/godown $* $OVL_BASE_SCRATCH_MNT
>>>> +		fi
>>>> +	else
>>>> +		src/godown $* $SCRATCH_MNT
>>>> +	fi
>>>> +}
>>>> +
>>>> _test_mount()
>>>> {
>>>>    if [ "$FSTYP" == "overlay" ]; then
>>>> @@ -2908,8 +2926,23 @@ _require_scratch_shutdown()
>>>> 
>>>> 	_scratch_mkfs > /dev/null 2>&1
>>>> 	_scratch_mount
>>>> -	src/godown -f $SCRATCH_MNT 2>&1 \
>>>> -		|| _notrun "$FSTYP does not support shutdown"
>>>> +
>>>> +	if [ $FSTYP = "overlay" ]; then
>>>> +		if [ -z $OVL_BASE_SCRATCH_DEV ]; then
>>>> +			# In lagacy overlay usage, it may specify directory as 
>>> 
>>> Trailing whitespace in above line.
>>> 
>>>> +			# SCRATCH_DEV, in this case OVL_BASE_SCRATCH_DEV
>>>> +			# will be null, so check OVL_BASE_SCRATCH_DEV before
>>>> +			# running shutdown to avoid shutting down base fs accidently.
>>>> +			_notrun "$SCRATCH_DEV is not a block device"
>>>> +		else
>>>> +			src/godown -f $OVL_BASE_SCRATCH_MNT 2>&1 \
>>>> +			|| _notrun "Underlying filesystem does not support shutdown"
>>>> +		fi
>>>> +	else
>>>> +		src/godown -f $SCRATCH_MNT 2>&1 \
>>>> +			|| _notrun "$FSTYP does not support shutdown"
>>>> +	fi
>>>> +
>>>> 	_scratch_unmount
>>>> }
>>> 
>>> Sorry for not bringing this up earlier, but I think we need to convert
>>> all existing direct callers of 'godown' to use this new helper,
>>> otherwise tests will call godown on an overlay merged dir and result in
>>> test failure as:
>>> 
>>> +/root/xfstests/src/godown: error on xfsctl(GOINGDOWN) of "/mnt/scratch/ovl-mnt": Inappropriate ioctl for device
>>> 
>>> You can go through all tests in 'shutdown' group in generic dir. But I
>>> noticed that not all tests in 'shutdown' group can be converted to
>>> _scratch_shutdown straightforwardly, e.g. generic/042 and generic/050,
>>> they require $SCRATCH_DEV to be a local device file (either block device
>>> or char device (ubifs uses char device)) rather than a dir or an NFS
>>> export, we need an additional _require rule in both of the tests, so
>>> that they won't run on overlay:
>>> 
>>> _require_local_device $SCRATCH_DEV
>>> 
>>> Other tests in 'shutdown' group can be converted easily.
>> 
>> We should fix the improper godown calling for overlayfs in generic test cases,
>> but it’s not directly caused by this patch, so I suggest merging this patch
> 
> IMO, it's just this patch that causes these kind of false failures in
> other generic shutdown tests. It's this patch that makes generic
> shutdown tests running on overlay possible, thus causes false failures.
> 
> We should fix all these problems in one patch, otherwise anyone tests
> overlay using xfs/ext4 (or other filesystems that support shutdown) as
> backing filesystems could see these failures.  OTOH, the overlay support
> in _require_metadata_journaling issue is a relative corner case, I think
> it's safe to fix that in a separate patch.

In order to add overlay support in some requirement checks like  _require_metadata_journaling, 
I think it’s better save underlying filesystem type to $OVL_BASE_FSTYP and doing proper check based
on it.

Currently $OVL_BASE_FSTYP and $FSTYP all set to “overlay”, Is there any specific reason for it? 



 


Thanks,
Chengguang.

  reply	other threads:[~2018-01-03 12:44 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-12-12  3:09 [PATCH v3 1/3] common/rc: add scratch shutdown support for overlayfs Chengguang Xu
2017-12-12  3:09 ` [PATCH v3 2/3] common/rc: add a check case in _require_xfs_io_command() to support syncfs Chengguang Xu
2017-12-12  3:09 ` [PATCH v3 3/3] generic/470: add syncfs test Chengguang Xu
2017-12-12  6:00   ` Amir Goldstein
2017-12-12  6:07     ` Chengguang Xu
2017-12-12  9:24   ` [PATCH v3 3/3] generic/471: " Eryu Guan
2017-12-12  9:56     ` Amir Goldstein
2017-12-12 10:05       ` Chengguang Xu
2017-12-12 11:00         ` Amir Goldstein
2017-12-12 13:30         ` Eryu Guan
2017-12-12  5:58 ` [PATCH v3 1/3] common/rc: add scratch shutdown support for overlayfs Amir Goldstein
2017-12-12  9:16 ` Eryu Guan
2017-12-12 10:20   ` Chengguang Xu
2017-12-12 13:39     ` Eryu Guan
2018-01-03 12:44       ` Chengguang Xu [this message]
2018-01-03 12:58         ` Amir Goldstein
2018-01-03 14:25           ` Chengguang Xu
2018-01-03 14:54             ` Amir Goldstein
2018-01-10  2:35               ` Chengguang Xu
2017-12-14  3:35   ` Chengguang Xu
2017-12-14  4:32     ` Eryu Guan
2017-12-14  5:31       ` Chengguang Xu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=DB72F77F-23E3-4F5E-93CB-F61B94F9EA86@icloud.com \
    --to=cgxu519@icloud.com \
    --cc=amir73il@gmail.com \
    --cc=eguan@redhat.com \
    --cc=fstests@vger.kernel.org \
    --cc=linux-unionfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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.