From: Jan Kara <jack@suse.cz>
To: Amir Goldstein <amir73il@gmail.com>
Cc: Jan Kara <jack@suse.cz>,
linux-fsdevel <linux-fsdevel@vger.kernel.org>,
Lino Sanfilippo <LinoSanfilippo@gmx.de>,
Miklos Szeredi <miklos@szeredi.hu>,
Paul Moore <paul@paul-moore.com>
Subject: Re: [PATCH 08/22] fsnotify: Attach marks to object via dedicated head structure
Date: Wed, 4 Jan 2017 14:38:17 +0100 [thread overview]
Message-ID: <20170104133817.GT3780@quack2.suse.cz> (raw)
In-Reply-To: <20161223133407.GE22679@quack2.suse.cz>
On Fri 23-12-16 14:34:07, Jan Kara wrote:
> On Fri 23-12-16 07:48:43, Amir Goldstein wrote:
> > On Thu, Dec 22, 2016 at 11:15 AM, Jan Kara <jack@suse.cz> wrote:
> > > Currently notification marks are attached to object (inode or vfsmnt) by
> > > a hlist_head in the object. The list is also protected by a spinlock in
> > > the object. So while there is any mark attached to the list of marks,
> > > the object must be pinned in memory (and thus e.g. last iput() deleting
> > > inode cannot happen). Also for list iteration in fsnotify() to work, we
> > > must hold fsnotify_mark_srcu lock so that mark itself and
> > > mark->obj_list.next cannot get freed. Thus we are required to wait for
> > > response to fanotify events from userspace process with
> > > fsnotify_mark_srcu lock held. That causes issues when userspace process
> > > is buggy and does not reply to some event - basically the whole
> > > notification subsystem gets eventually stuck.
> > >
> > > So to be able to drop fsnotify_mark_srcu lock while waiting for
> > > response, we have to pin the mark in memory and make sure it stays in
> > > the object list (as removing the mark waiting for response could lead to
> > > lost notification events for groups later in the list). However we don't
> > > want inode reclaim to block on such mark as that would lead to system
> > > just locking up elsewhere.
> > >
> > > This commit tries to pave a way towards solving these conflicting
> > > lifetime needs. Instead of anchoring the list of marks directly in the
> > > object, we anchor it in a dedicated structure (fsnotify_mark_list) and
> > > just point to that structure from the object. Also the list is protected
> > > by a spinlock contained in that structure. With this, we can detach
> > > notification marks from object without having to modify the list itself.
> > >
> >
> > The structural change looks very good to.
> > It makes the code much easier to manage IMO.
> >
> > I am only half way though this big change, but I wanted to make one meta
> > comment.
> >
> > I have a problem with the choice of naming for the new struct.
> > 'list' is really an overloaded term and the use of 'list' as a name of
> > a class that
> > contains a list head makes for some really confusing constructs like
> > list->list and mark->obj_list_head which is not a list_head struct.
>
> OK, I'll think about better naming. I agree it may be slightly confusing.
So how about naming the type fsnotify_mark_connector? We can use 'conn' as
a name for local variables. I think that is not as overloaded as 'list'
and it describes that it is a structure used for connecting marks with
inode / vfsmount. Would that make things more comprehensive for you?
Honza
--
Jan Kara <jack@suse.com>
SUSE Labs, CR
next prev parent reply other threads:[~2017-01-04 13:38 UTC|newest]
Thread overview: 79+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-12-22 9:15 [PATCH 0/22] fsnotify: Avoid SRCU stalls with fanotify permission events Jan Kara
2016-12-22 9:15 ` [PATCH 01/22] fsnotify: Remove unnecessary tests when showing fdinfo Jan Kara
2016-12-22 12:59 ` Amir Goldstein
2016-12-22 15:16 ` Jan Kara
2016-12-22 15:54 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 02/22] inotify: Remove inode pointers from debug messages Jan Kara
2016-12-22 15:31 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 03/22] fanotify: Move recalculation of inode / vfsmount mask under mark_mutex Jan Kara
2016-12-22 16:27 ` Amir Goldstein
2016-12-22 17:31 ` Jan Kara
2016-12-22 19:08 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 04/22] fsnotify: Remove fsnotify_duplicate_mark() Jan Kara
2016-12-22 23:13 ` Paul Moore
2016-12-23 13:22 ` Jan Kara
2016-12-23 14:01 ` Paul Moore
2016-12-22 9:15 ` [PATCH 05/22] audit: Fix sleep in atomic Jan Kara
2016-12-22 23:18 ` Paul Moore
2016-12-23 13:24 ` Jan Kara
2016-12-23 14:17 ` Paul Moore
2016-12-26 16:33 ` Paul Moore
2017-01-02 18:21 ` Jan Kara
2017-01-03 21:11 ` Paul Moore
2017-01-04 8:50 ` Jan Kara
2017-01-05 2:14 ` Paul Moore
2016-12-22 9:15 ` [PATCH 06/22] audit: Abstract hash key handling Jan Kara
2016-12-22 23:27 ` Paul Moore
2016-12-23 13:27 ` Jan Kara
2016-12-23 14:13 ` Paul Moore
2017-01-03 17:34 ` Jan Kara
2017-01-05 2:06 ` Paul Moore
2016-12-22 9:15 ` [PATCH 07/22] fsnotify: Update comments Jan Kara
2016-12-23 4:45 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 08/22] fsnotify: Attach marks to object via dedicated head structure Jan Kara
2016-12-23 5:48 ` Amir Goldstein
2016-12-23 13:34 ` Jan Kara
2017-01-04 13:38 ` Jan Kara [this message]
2017-01-04 15:29 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 09/22] inotify: Do not drop mark reference under idr_lock Jan Kara
2016-12-23 8:04 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 10/22] fsnotify: Detach mark from object list when last reference is dropped Jan Kara
2016-12-23 10:51 ` Amir Goldstein
2016-12-23 13:42 ` Jan Kara
2016-12-22 9:15 ` [PATCH 11/22] fsnotify: Remove special handling of mark destruction on group shutdown Jan Kara
2016-12-23 12:12 ` Amir Goldstein
2016-12-23 13:31 ` Jan Kara
2016-12-22 9:15 ` [PATCH 12/22] fsnotify: Move queueing of mark for destruction into fsnotify_put_mark() Jan Kara
2016-12-26 14:15 ` Amir Goldstein
2017-01-04 10:28 ` Jan Kara
2017-01-04 12:22 ` Jan Kara
2016-12-22 9:15 ` [PATCH 13/22] fsnotify: Provide framework for dropping SRCU lock in ->handle_event Jan Kara
2016-12-26 15:01 ` Amir Goldstein
2016-12-26 15:11 ` Amir Goldstein
2017-01-04 9:03 ` Jan Kara
2017-01-04 10:50 ` Amir Goldstein
2017-01-04 11:45 ` Jan Kara
2016-12-26 18:37 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 14/22] fsnotify: Pass SRCU index into handle_event handler Jan Kara
2016-12-26 15:13 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 15/22] fanotify: Release SRCU lock when waiting for userspace response Jan Kara
2016-12-26 15:22 ` Amir Goldstein
2017-01-04 9:05 ` Jan Kara
2016-12-22 9:15 ` [PATCH 16/22] fsnotify: Remove fsnotify_set_mark_{,ignored_}mask_locked() Jan Kara
2016-12-26 16:42 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 17/22] fsnotify: Remove fsnotify_recalc_{inode|vfsmount}_mask() Jan Kara
2016-12-26 16:44 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 18/22] fsnotify: Inline fsnotify_clear_{inode|vfsmount|_mark_group() Jan Kara
2016-12-26 16:57 ` Amir Goldstein
2017-01-04 9:28 ` Jan Kara
2016-12-22 9:15 ` [PATCH 19/22] fsnotify: Remove fsnotify_find_{inode|vfsmount}_mark() Jan Kara
2016-12-26 17:14 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 20/22] fsnotify: Drop inode_mark.c Jan Kara
2016-12-26 17:15 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 21/22] fsnotify: Add group pointer in fsnotify_init_mark() Jan Kara
2016-12-26 17:34 ` Amir Goldstein
2016-12-22 9:15 ` [PATCH 22/22] fsnotify: Move ->free_mark callback to fsnotify_ops Jan Kara
2016-12-26 17:39 ` Amir Goldstein
2016-12-22 20:58 ` [PATCH 0/22] fsnotify: Avoid SRCU stalls with fanotify permission events Paul Moore
2016-12-22 21:05 ` Amir Goldstein
2016-12-22 23:04 ` Paul Moore
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=20170104133817.GT3780@quack2.suse.cz \
--to=jack@suse.cz \
--cc=LinoSanfilippo@gmx.de \
--cc=amir73il@gmail.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=paul@paul-moore.com \
/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 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).