Linux-Fsdevel Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Amir Goldstein <amir73il@gmail.com>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: Jeff Layton <jlayton@poochiereds.net>,
	"J . Bruce Fields" <bfields@fieldses.org>,
	overlayfs <linux-unionfs@vger.kernel.org>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>
Subject: Re: [PATCH v2 04/17] ovl: decode connected upper dir file handles
Date: Thu, 18 Jan 2018 10:47:54 +0200	[thread overview]
Message-ID: <CAOQ4uxi3V=s5t32DbNSqYum7EXjJTSqjE3jrJKLV9a0R5m4-QQ@mail.gmail.com> (raw)
In-Reply-To: <CAJfpegsJURzWdC0a3AiEMv_BnCO1rdXqCr1-Y0HtrG3vKxJd5g@mail.gmail.com>

On Thu, Jan 18, 2018 at 10:22 AM, Miklos Szeredi <miklos@szeredi.hu> wrote:
> On Wed, Jan 17, 2018 at 5:34 PM, Amir Goldstein <amir73il@gmail.com> wrote:
>> On Wed, Jan 17, 2018 at 5:42 PM, Miklos Szeredi <miklos@szeredi.hu> wrote:
>>> On Wed, Jan 17, 2018 at 12:18 PM, Amir Goldstein <amir73il@gmail.com> wrote:
>>>> On Mon, Jan 15, 2018 at 4:56 PM, Miklos Szeredi <miklos@szeredi.hu> wrote:
>>>>>
>>>> [...]
>>>>> >>
>>>>> >> So, a working algorithm would be going up to the first connected
>>>>> >> parent or root, lock parent, lookup name and restart.  Not guaranteed
>>>>> >> to finish, since not protected against always racing with renames.
>>>>> >> Can we take s_vfs_rename_sem on ovl to prevent that?
>>>>> >>
>>>>> >
>>>>> > Sounds like a simple and good enough solution.
>>>>> > Do we really need the locking of parent and restart connect if
>>>>> > we take s_vfs_rename_sem around ovl_lookup_real()?
>>>>>
>>>>> No, but s_vfs_rename_sem is a really heavyweight solution, we should
>>>>> do better than that for decoding a file handle.
>>>>>
>>>>> And we probably don't need anything else, since rename on ancestor
>>>>> means renamed dir is connected, and hopefully not evicted from the
>>>>> cache until we repeat the walk up.
>>>>>
>>>>> So need to lock parent, lookup ovl dentry, verify we got the same
>>>>> upper, if not retry icache lookup.
>>>>>
>>>>> Not sure we need to worry about that "hopefully".  Hopefully not.
>>>>>
>>>>
>>>> Something like this??
>>>>
>>>> This is just the raw fix to patch 4/17 without the icache lookup
>>>> that is added by later patches.
>>>>
>>>> I added rename_lock seqlock around backwalk to connected ancestor
>>>> and take_dentry_name_snapshot() for the stability of real name
>>>> during overlay lookup.
>>>>
>>>> I considered also storing OVL_I(d_inode(connected))->version
>>>> inside seqlock and comparing it to version in case lookup of child
>>>> failed. This could help us distinguish between overlay rename and
>>>> underlying rename (overlay dir version did not change) and return
>>>> ESTALE instead of restarting lookup in the latter case.
>>>> Wasn't sure if that was a good idea and what we loose if we leave it out.
>>>>
>>>> I tested this code, but only with upper file handles of course
>>>> (xfstest generic/467).
>>>>
>>>> Please let me know what you think.
>>>>
>>>> Thanks,
>>>> Amir.
>>>>
>>>> ================================================================
>>>>
>>>> From 337543c3fcdf9323d3720d17ab6fc13e287bbec1 Mon Sep 17 00:00:00 2001
>>>> From: Amir Goldstein <amir73il@gmail.com>
>>>> Date: Thu, 28 Dec 2017 18:36:16 +0200
>>>> Subject: [PATCH v3 4/17] ovl: decode connected upper dir file handles
>>>>
>>>> Until this change, we decoded upper file handles by instantiating an
>>>> overlay dentry from the real upper dentry. This is sufficient to handle
>>>> pure upper files, but insufficient to handle merge/impure dirs.
>>>>
>>>> To that end, if decoded real upper dir is connected and hashed, we
>>>> lookup an overlay dentry with the same path as the real upper dir.
>>>> If decoded real upper is non-dir, we instantiate a disconnected overlay
>>>> dentry as before this change.
>>>>
>>>> Because ovl_fh_to_dentry() returns a connected overlay dir dentry,
>>>> exportfs never needs to call get_parent() and get_name() to reconnect an
>>>> upper overlay dir. Because connectable non-dir file handles are not
>>>> supported, exportfs will not be able to use fh_to_parent() and get_name()
>>>> methods to reconnect a disconnected non-dir to its parent. Therefore, the
>>>> methods get_parent() and get_name() are implemented just to print out a
>>>> sanity warning and the method fh_to_parent() is implemented to warn the
>>>> user that using the 'subtree_check' exportfs option is not supported.
>>>>
>>>> An alternative approach could have been to implement instantiating of
>>>> an overlay directory inode from origin/index and implement get_parent()
>>>> and get_name() by calling into underlying fs operations and them
>>>> instantiating the overlay parent dir.
>>>>
>>>> The reasons for not choosing the get_parent() approach were:
>>>> - Obtaining a disconnected overlay dir dentry would requires a
>>>>   delicate re-factoring of ovl_lookup() to get a dentry with overlay
>>>>   parent info. It was preferred to avoid doing that re-factoring unless
>>>>   it was proven worthy.
>>>> - Going down the path of disconnected dir would mean that the (non
>>>>   trivial) code path of d_splice_alias() could be traveled and that
>>>>   meant writing more tests and introduces race cases that are very hard
>>>>   to hit on purpose. Taking the path of connecting overlay dentry by
>>>>   forward lookup is therefore the safe and boring way to avoid surprises.
>>>>
>>>> The culprit of the chosen "connected overlay dentry" approach:
>>>> - We need to take special care to rename of ancestors while connecting
>>>>   the overlay dentry by real dentry path. These subtleties are usually
>>>>   handled by generic exportfs and VFS code.
>>>>
>>>> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
>>>> ---
[...]
>>>> +/*
>>>> + * Lookup an overlay dentry whose real dentry is @real.
>>>> + * If @real is on upper layer, we lookup a child overlay dentry with the same
>>>> + * path the real dentry. Otherwise, we need to consult index for lookup.
>>>> + */
>>>> +static struct dentry *ovl_lookup_real(struct super_block *sb,
>>>> +                                     struct dentry *real,
>>>> +                                     struct ovl_layer *layer)
>>>> +{
>>>> +       struct dentry *connected;
>>>> +       unsigned int seq;
>>>> +       int err = 0;
>>>> +
>>>> +       /* TODO: use index when looking up by lower real dentry */
>>>> +       if (layer->idx)
>>>> +               return ERR_PTR(-EACCES);
>>>> +
>>>> +       connected = dget(sb->s_root);
>>>> +       while (!err) {
>>>> +               struct dentry *next, *this;
>>>> +               struct dentry *parent = NULL;
>>>> +               struct dentry *real_connected = layer->mnt->mnt_root;
>>>> +
>>>> +               if (real_connected == real)
>>>> +                       break;
>>>
>>> This loop will never finish, since real_connected is mnt_root now.
>>> Would be nice if there was a guaranteed way to finish this without
>>> icache lookup, but I don't see how.
>>>
>>
>> That's a bug.
>> The correct code is:
>>
>> struct dentry *real_connected = ovl_dentry_upper(connected);
>
> No, that's still broken, because if something gets renamed below
> "connected", we'll fall off root.  So need to stop at real_connected
> AND at root of overlay.
>

Right. Will fix that.

>>
>>
>>>> +
>>>> +               /*
>>>> +                * Find the topmost dentry not yet connected. Taking rename_lock
>>>> +                * so at least we don't race with rename when walking back to
>>>> +                * 'real_connected'.
>>>> +                */
>>>> +               seq = read_seqbegin(&rename_lock);
>>>
>>> I don't see what we gain with this.
>>>
>>
>> I can't say that I do see it, but perhaps there is something yet
>> to be gained by adding this later for lower layers lookup.
>> Perhaps when looking on lower real layer, we can store the
>> overlay dir cache version of 'connected' (connected in this case
>> may be an indexed merge dir).
>> After we take 'connected' dir mutex, we cannot check that
>> real parent hasn't changes as an indication to no overlay rename
>> because overlay rename happens on upper, but we can compare
>> the dir cache version of 'connected' dir to the version we stored
>> under rename_lock.
>> Then we can tell if lower lookup has failed because of some
>> permanent error (e.g. middle layer redirect) or because of an
>> indexed rename, so we need to restart.
>> Maybe that gains us something?
>
> Sorry, couldn't follow that.
>
> I don't see need for additional locking apart from the one in
> ovl_lookup_by_real_one().  Because the only race remaining is eviction
> of overlay inode(s) from the icache, and no locking is going to
> prevent that.
>

Right. No need for the rename seqlock.
The reason we need the cache version check for connecting by
lower real is to end the retry loop in case there is a permanent
decode error due to middle layer redirects and we did not mitigate
it with copy up on encode.

BTW, with read-only lower-only overlay mount we cannot mitigate
middle layer redirect with copy up, so we need to either fail encode
or fail decode. Failing decode is better, because it still allows nfsd
to work when dentry remains in cache.

Thanks,
Amir.

  reply	other threads:[~2018-01-18  8:47 UTC|newest]

Thread overview: 68+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-04 17:20 [PATCH v2 00/17] Overlayfs NFS export support Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 01/17] ovl: document NFS export Amir Goldstein
2018-01-11 16:06   ` Miklos Szeredi
2018-01-11 16:26     ` Amir Goldstein
2018-01-12 15:43       ` Miklos Szeredi
2018-01-12 15:49         ` Miklos Szeredi
2018-01-12 18:50           ` Amir Goldstein
2018-01-13  8:54           ` Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 02/17] ovl: encode pure upper file handles Amir Goldstein
2018-01-18 10:31   ` Miklos Szeredi
2018-01-04 17:20 ` [PATCH v2 03/17] ovl: decode " Amir Goldstein
2018-01-18 14:09   ` Miklos Szeredi
2018-01-18 14:34     ` Amir Goldstein
2018-01-18 14:39       ` Miklos Szeredi
2018-01-18 19:49         ` Amir Goldstein
2018-01-18 20:10           ` Miklos Szeredi
2018-01-18 20:35             ` Amir Goldstein
2018-01-18 22:57               ` Amir Goldstein
2018-01-19  0:23                 ` Amir Goldstein
2018-01-19 10:39                   ` Miklos Szeredi
2018-01-19 11:07                     ` Amir Goldstein
2018-01-19 20:10                       ` Amir Goldstein
2018-01-24 10:34                         ` Miklos Szeredi
2018-01-24 11:04                           ` Amir Goldstein
2018-01-24 11:18                             ` Amir Goldstein
2018-01-24 11:55                               ` Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 04/17] ovl: decode connected upper dir " Amir Goldstein
2018-01-05 12:33   ` Amir Goldstein
2018-01-05 15:18     ` J . Bruce Fields
2018-01-05 15:34       ` Amir Goldstein
2018-01-15 11:41     ` Miklos Szeredi
2018-01-15 11:33   ` Miklos Szeredi
2018-01-15 12:20     ` Amir Goldstein
2018-01-15 14:56       ` Miklos Szeredi
2018-01-17 11:18         ` Amir Goldstein
2018-01-17 12:20           ` Amir Goldstein
2018-01-17 13:29             ` Amir Goldstein
2018-01-17 15:42           ` Miklos Szeredi
2018-01-17 16:34             ` Amir Goldstein
2018-01-17 21:36               ` Amir Goldstein
2018-01-18  8:22               ` Miklos Szeredi
2018-01-18  8:47                 ` Amir Goldstein [this message]
2018-01-18  9:12                   ` Miklos Szeredi
2018-01-18 10:28                     ` Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 05/17] ovl: encode non-indexed upper " Amir Goldstein
2018-01-15 11:58   ` Miklos Szeredi
2018-01-15 12:07     ` Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 06/17] ovl: copy up before encoding dir file handle when ofs->numlower > 1 Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 07/17] ovl: encode lower file handles Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 08/17] ovl: decode lower non-dir " Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 09/17] ovl: decode indexed " Amir Goldstein
2018-01-18 13:11   ` Miklos Szeredi
2018-01-04 17:20 ` [PATCH v2 10/17] ovl: decode lower file handles of unlinked but open files Amir Goldstein
2018-01-16  9:16   ` Miklos Szeredi
2018-01-16  9:37     ` Amir Goldstein
2018-01-16 10:10       ` Miklos Szeredi
2018-01-16 10:40         ` Amir Goldstein
2018-01-16 11:07           ` Miklos Szeredi
2018-01-17 21:05         ` Amir Goldstein
2018-01-18 14:18   ` Amir Goldstein
2018-02-27 11:35     ` Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 11/17] ovl: decode indexed dir file handles Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 12/17] ovl: decode pure lower " Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 13/17] ovl: hash directory inodes for NFS export Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 14/17] ovl: lookup connected ancestor of dir in inode cache Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 15/17] ovl: lookup indexed ancestor of lower dir Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 16/17] ovl: wire up NFS export support Amir Goldstein
2018-01-04 17:20 ` [PATCH v2 17/17] nfsd: encode stat->mtime for getattr instead of inode->i_mtime Amir Goldstein

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='CAOQ4uxi3V=s5t32DbNSqYum7EXjJTSqjE3jrJKLV9a0R5m4-QQ@mail.gmail.com' \
    --to=amir73il@gmail.com \
    --cc=bfields@fieldses.org \
    --cc=jlayton@poochiereds.net \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-unionfs@vger.kernel.org \
    --cc=miklos@szeredi.hu \
    --subject='Re: [PATCH v2 04/17] ovl: decode connected upper dir file handles' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).