All of lore.kernel.org
 help / color / mirror / Atom feed
From: ebiederm@xmission.com (Eric W. Biederman)
To: "J. Bruce Fields" <bfields@fieldses.org>
Cc: Miklos Szeredi <miklos@szeredi.hu>,
	Al Viro <viro@zeniv.linux.org.uk>,
	"Serge E. Hallyn" <serge@hallyn.com>,
	Linux-Fsdevel <linux-fsdevel@vger.kernel.org>,
	Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Andy Lutomirski <luto@amacapital.net>,
	Rob Landley <rob@landley.net>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Christoph Hellwig <hch@infradead.org>,
	Karel Zak <kzak@redhat.com>
Subject: Re: [PATCH 08/11] vfs: Merge check_submounts_and_drop and d_invalidate
Date: Wed, 26 Feb 2014 18:05:17 -0800	[thread overview]
Message-ID: <87vbw1xqci.fsf@xmission.com> (raw)
In-Reply-To: <20140226193740.GA24456@fieldses.org> (J. Bruce Fields's message of "Wed, 26 Feb 2014 14:37:40 -0500")

"J. Bruce Fields" <bfields@fieldses.org> writes:

> On Tue, Feb 25, 2014 at 02:03:36PM -0800, Eric W. Biederman wrote:
>> "J. Bruce Fields" <bfields@fieldses.org> writes:
>> 
>> > On Mon, Feb 24, 2014 at 04:01:29PM -0800, Eric W. Biederman wrote:
>> >> Miklos Szeredi <miklos@szeredi.hu> writes:
>> >> 
>> >> >
>> >> > You can optimize this by including the negative check within the above d_locked
>> >> > region and calling __d_drop() instead.
>> >> 
>> >> For this patch just moving the code and not changing it is the corret
>> >> thing to do because it helps with review and understanding the code.
>> >> 
>> >> There are two ways I could see going with optimizing the preamble.
>> >> Simply dropping the d_lock from around the d_unhashed test as a pointer
>> >> dereference should be atomic, and the test is racy against
>> >> d_materialise_unique.
>> >
>> > Could you explain?  What's the race, and what are the consequences?
>
> Actually I was just confused as to whether the above was "is racy" was
> claiming the existance of some bug.
>
> I believe I should have read the above as more like "the test is already
> racy against d_materialise_unique, but it's a harmless race, and
> dropping the d_lock wouldn't make it any worse".
>
>> >> (We don't always hold the parent directories inode mutex when d_invalidate is called).
>> 
>> d_unhashed is not a permanent condition because of d_materialise_unique,
>> and d_splice_alias.
>> 
>> d_invalidate can be called on an unhashed dentry in one of two ways
>> (either d_revalidate dropped the dentry or another routine that drops
>> the dentry beat the current invocation of d_invalidate to the job).
>> 
>> 
>> There are 3 places d_revalidate is called.
>> 
>> Once on the rcu path with with the appropriate flag set.
>> 
>> Once without out the parent i_mutex held, just off of the rcu path,
>> on that path d_invalidate is when d_revalidate fails.
>> 
>> Once during lookup with the parent directory i_mutex held.
>> 
>> 
>> Because the parent direcories i_mutex is not always held accross
>> d_revalidate and the following d_invalidate it happens that d_invalidate
>> is not always an atomic operation.
>> 
>> 
>> At worst the race results in a dentry that is dropped when it could be
>> hashed,
>
> Because somebody not holding the i_mutex calls d_invalidate based on old
> information and unhashes something that
> d_materialise_unique/d_splice_alias just hashed?

More likely today somebody not holding i_mutex and not in rcu context
calls d_revalidate.  d_revalidate drops the dentry and before we
d_invalidate d_materialise_unique/d_splice_alias rehashes it.

After my changes it looks like it takes 3 processes two instances
of d_invalidate and a instance of d_materialise_unique/d_spliace_alias
to trigger this case.

In either case the window is very small and the outcome is effectively
harmless.  So I don't see this as a problem.

>> that we will resurrect next time someone attempts to look it
>> up and d_materialise_unique/d_splice_alias is called.
>
> OK.
>
>> None of that really matters for optimizing d_invalidate, but it is part
>> of the background in which d_invalidate lives. All that is significant
>> in d_invalidate is knowing that d_materialise_unique, and possibly
>> d_splice_alias may run concurrently with d_invalidate.  It is unlikely
>> and essentially harmless.
>> 
>> 
>> After my patchset (because I removed all of the d_drop's from
>> .d_revalidate) the only race that should remain is between two parallel
>> calls of d_invalidate.  Which probably means we can remove the test for
>> d_unhashed altogether.
>> 
>> Right now I just want to make this first big step and make certain the
>> code is solid.  After that optimization is easy.
>
> Thanks for the explanation!

Welcome.

Eric


  reply	other threads:[~2014-02-27  2:05 UTC|newest]

Thread overview: 181+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-08-14 17:42 DoS with unprivileged mounts Miklos Szeredi
2013-08-14 19:26 ` Andy Lutomirski
2013-08-14 19:53   ` Eric W. Biederman
2013-08-14 20:25     ` Andy Lutomirski
2013-08-14 21:54       ` Eric W. Biederman
2013-08-14 19:32 ` Eric W. Biederman
2013-08-15  4:59   ` Miklos Szeredi
2013-08-15  6:45     ` Eric W. Biederman
2013-08-15  6:52       ` Andy Lutomirski
2013-08-15  7:55         ` Eric W. Biederman
2013-08-15  9:29       ` Miklos Szeredi
2013-10-04 22:41         ` [RFC][PATCH 0/3] vfs: Detach mounts on unlink Eric W. Biederman
2013-10-04 22:42           ` [RFC][PATCH 1/3] vfs: Keep a list of mounts on a mount point Eric W. Biederman
2013-10-08 13:44             ` Miklos Szeredi
2013-10-08 20:31               ` Eric W. Biederman
2013-10-04 22:43           ` [RFC][PATCH 2/3] vfs: Add a function to lazily unmount all mounts from any dentry Eric W. Biederman
2013-10-07  4:22             ` Serge E. Hallyn
2013-10-07  6:15               ` Eric W. Biederman
2013-10-08 13:56             ` Miklos Szeredi
2013-10-08 20:32               ` Eric W. Biederman
2013-10-04 22:43           ` [RFC][PATCH 3/3] vfs: Lazily remove mounts on unlinked files and directories Eric W. Biederman
2013-10-08 15:50             ` Miklos Szeredi
2013-10-08 21:47               ` Eric W. Biederman
2013-10-09 19:12                 ` Grrrr fusermount Eric W. Biederman
2013-10-09 20:09                   ` Andy Lutomirski
2013-10-09 23:35                     ` Eric W. Biederman
2013-10-04 23:20           ` [RFC][PATCH 0/3] vfs: Detach mounts on unlink Linus Torvalds
2013-10-05  0:03             ` Eric W. Biederman
2013-10-05  1:57               ` Eric W. Biederman
2013-10-05 23:42               ` Rob Landley
2013-10-06  0:44                 ` Eric W. Biederman
2013-10-08  8:03                 ` Karel Zak
2013-10-10  6:46                   ` Rob Landley
2013-10-05  2:34           ` [RFC][PATCH 4/3] vfs: Allow rmdir to remove mounts in all but the current mount namespace Eric W. Biederman
2013-10-05 15:44             ` Serge E. Hallyn
2013-10-07  4:39             ` Serge E. Hallyn
2013-10-07  6:55               ` Eric W. Biederman
2013-10-07 16:53                 ` Andy Lutomirski
2013-10-07 22:25                   ` Eric W. Biederman
2013-10-08 16:06                     ` Miklos Szeredi
2013-10-08 16:06                       ` Andy Lutomirski
2013-10-08 16:11                         ` Miklos Szeredi
2013-10-08 20:50                           ` Eric W. Biederman
2013-10-10 10:02                             ` Miklos Szeredi
2013-10-10 11:43                               ` Eric W. Biederman
2013-10-10 11:57                                 ` Miklos Szeredi
2013-10-12  1:04                                   ` Eric W. Biederman
2013-10-12  1:39                                     ` Eric W. Biederman
     [not found]                                       ` <87d2nb8dxy.fsf-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-15 20:15                                         ` [REVIEW][PATCH 0/4] vfs: Detach mounts on unlink Eric W. Biederman
2013-10-15 20:15                                           ` Eric W. Biederman
2013-10-15 20:16                                           ` [REVIEW][PATCH 1/4] vfs: Don't allow overwriting mounts in the current mount namespace Eric W. Biederman
2013-11-03  3:54                                             ` Al Viro
     [not found]                                               ` <20131103035406.GA8537-3bDd1+5oDREiFSDQTTA3OLVCufUGDwFn@public.gmane.org>
2013-11-08 20:51                                                 ` Eric W. Biederman
2013-11-08 20:51                                                   ` Eric W. Biederman
2013-11-08 21:35                                                   ` Al Viro
     [not found]                                                     ` <20131108213551.GR13318-3bDd1+5oDREiFSDQTTA3OLVCufUGDwFn@public.gmane.org>
2013-11-08 22:17                                                       ` Eric W. Biederman
2013-11-08 22:17                                                         ` Eric W. Biederman
     [not found]                                                         ` <87fvr61qtg.fsf-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-11-09  8:49                                                           ` Christoph Hellwig
2013-11-09  8:49                                                             ` Christoph Hellwig
     [not found]                                                             ` <20131109084916.GA21413-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2013-11-21 20:58                                                               ` Eric W. Biederman
2013-11-21 20:58                                                                 ` Eric W. Biederman
     [not found]                                                   ` <87bo1u8vmf.fsf-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-11-08 21:35                                                     ` Al Viro
2013-11-21 20:49                                                 ` Eric W. Biederman
2013-11-21 20:49                                                   ` Eric W. Biederman
     [not found]                                             ` <87d2n6xpan.fsf_-_-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-22 19:04                                               ` Serge E. Hallyn
2013-10-22 19:04                                                 ` Serge E. Hallyn
2013-11-03  3:54                                               ` Al Viro
     [not found]                                           ` <87iowyxpci.fsf_-_-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-15 20:16                                             ` Eric W. Biederman
2013-10-15 20:17                                             ` [REVIEW][PATCH 2/4] vfs: Keep a list of mounts on a mount point Eric W. Biederman
2013-10-15 20:17                                               ` Eric W. Biederman
     [not found]                                               ` <877gdexp9s.fsf_-_-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-22 19:06                                                 ` Serge E. Hallyn
2013-10-22 19:06                                                   ` Serge E. Hallyn
2013-10-15 20:17                                             ` [REVIEW][PATCH 3/4] vfs: Add a function to lazily unmount all mounts from any dentry. v3 Eric W. Biederman
2013-10-15 20:17                                               ` Eric W. Biederman
     [not found]                                               ` <871u3mxp8s.fsf_-_-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-22 19:08                                                 ` Serge E. Hallyn
2013-10-22 19:08                                                   ` Serge E. Hallyn
2013-10-15 20:18                                             ` [REVIEW][PATCH 4/4] vfs: Lazily remove mounts on unlinked files and directories. v2 Eric W. Biederman
2013-10-15 20:18                                               ` Eric W. Biederman
     [not found]                                               ` <87vc0ywan7.fsf_-_-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org>
2013-10-22 19:13                                                 ` Serge E. Hallyn
2013-10-22 19:13                                                   ` Serge E. Hallyn
2013-10-08 10:42                 ` [RFC][PATCH 4/3] vfs: Allow rmdir to remove mounts in all but the current mount namespace Matthias Schniedermeyer
2013-10-08 13:13                   ` Eric W. Biederman
2013-10-05 23:07           ` [RFC][PATCH 0/3] vfs: Detach mounts on unlink Rob Landley
2013-10-05 23:17             ` Linus Torvalds
2013-10-05 23:22               ` Linus Torvalds
2013-10-06  0:18                 ` Rob Landley
2013-10-06  0:37                   ` Linus Torvalds
2013-10-05 23:24               ` Al Viro
2013-10-06  0:22                 ` Rob Landley
2013-10-06  0:12               ` Rob Landley
2013-10-05 23:19             ` Al Viro
2013-10-06  0:13               ` Rob Landley
2014-02-15 21:34           ` [PATCH 0/11] Detaching " Eric W. Biederman
2014-02-15 21:35             ` [PATCH 01/11] vfs: Document the effect of d_revalidate on d_find_alias Eric W. Biederman
2014-02-15 21:36             ` [PATCH 02/11] vfs: More precise tests in d_invalidate Eric W. Biederman
2014-02-15 22:51               ` Linus Torvalds
2014-02-15 22:59                 ` Linus Torvalds
2014-02-15 23:23                   ` Eric W. Biederman
2014-02-15 23:39                     ` Eric W. Biederman
2014-02-16  0:03                     ` Linus Torvalds
2014-02-16  1:22                       ` Eric W. Biederman
2014-02-15 21:36             ` [PATCH 03/11] vfs: Don't allow overwriting mounts in the current mount namespace Eric W. Biederman
2014-02-18 17:12               ` Miklos Szeredi
2014-02-18 21:28                 ` Eric W. Biederman
2014-02-18 22:20                   ` Linus Torvalds
2014-02-19  2:23                     ` Stephen Rothwell
2014-02-24 23:43                       ` Eric W. Biederman
2014-02-15 21:37             ` [PATCH 04/11] vfs: Keep a list of mounts on a mount point Eric W. Biederman
2014-02-15 21:37             ` [PATCH 05/11] vfs: Add a function to lazily unmount all mounts from any dentry Eric W. Biederman
2014-02-16  1:50               ` Eric W. Biederman
2014-02-18 17:17               ` Miklos Szeredi
2014-02-25  3:24                 ` Eric W. Biederman
2014-02-15 21:38             ` [PATCH 06/11] vfs: Lazily remove mounts on unlinked files and directories Eric W. Biederman
2014-02-15 21:38             ` [PATCH 07/11] vfs: Remove unnecessary calls of check_submounts_and_drop Eric W. Biederman
2014-02-15 21:39             ` [PATCH 08/11] vfs: Merge check_submounts_and_drop and d_invalidate Eric W. Biederman
2014-02-18 17:40               ` Miklos Szeredi
2014-02-25  0:01                 ` Eric W. Biederman
2014-02-25 15:13                   ` J. Bruce Fields
2014-02-25 22:03                     ` Eric W. Biederman
2014-02-26 19:37                       ` J. Bruce Fields
2014-02-27  2:05                         ` Eric W. Biederman [this message]
2014-02-15 21:39             ` [PATCH 09/11] vfs: Make d_invalidate return void Eric W. Biederman
2014-02-15 21:40             ` [PATCH 10/11] vfs: Remove d_drop calls from d_revalidate implementations Eric W. Biederman
2014-02-15 21:40             ` [PATCH 11/11] proc: Update proc_flush_task_mnt to use d_invalidate Eric W. Biederman
2014-02-18 17:43             ` [PATCH 0/11] Detaching mounts on unlink Miklos Szeredi
2014-02-25  9:33             ` [PATCH 0/12] Detaching mounts on unlink (for 3.15 v2) Eric W. Biederman
2014-02-25  9:34               ` [PATCH 01/12] vfs: Document the effect of d_revalidate on d_find_alias Eric W. Biederman
2014-02-25  9:35               ` [PATCH 02/12] vfs: More precise tests in d_invalidate Eric W. Biederman
2014-02-25  9:35               ` [PATCH 03/12] vfs: Don't allow overwriting mounts in the current mount namespace Eric W. Biederman
2014-02-25  9:36               ` [PATCH 04/12] vfs: Keep a list of mounts on a mount point Eric W. Biederman
2014-02-25  9:37               ` [PATCH 05/12] vfs: factor out lookup_mountpoint from new_mountpoint Eric W. Biederman
2014-02-25  9:38               ` [PATCH 06/12] vfs: Add a function to lazily unmount all mounts from any dentry Eric W. Biederman
2014-02-25  9:38               ` [PATCH 07/12] vfs: Lazily remove mounts on unlinked files and directories Eric W. Biederman
2014-02-25  9:39               ` [PATCH 08/12] vfs: Remove unnecessary calls of check_submounts_and_drop Eric W. Biederman
2014-02-25  9:40               ` [PATCH 09/12] vfs: Merge check_submounts_and_drop and d_invalidate Eric W. Biederman
2014-02-25  9:40               ` [PATCH 10/12] vfs: Make d_invalidate return void Eric W. Biederman
2014-02-25  9:41               ` [PATCH 11/12] vfs: Remove d_drop calls from d_revalidate implementations Eric W. Biederman
2014-02-25  9:41               ` [PATCH 12/12] proc: Update proc_flush_task_mnt to use d_invalidate Eric W. Biederman
2014-04-09  0:21               ` [GIT PULL] Detaching mounts on unlink for 3.15-rc1 Eric W. Biederman
2014-04-09  2:30                 ` Al Viro
2014-04-09  2:39                   ` Al Viro
2014-04-09  9:02                     ` Eric W. Biederman
2014-04-09 17:32                     ` Eric W. Biederman
2014-04-09 17:53                       ` Al Viro
2014-04-09 18:28                         ` Al Viro
2014-04-09 22:49                           ` Eric W. Biederman
2014-04-09 22:58                             ` [RFC][PATCH] vfs: In mntput run deactivate_super on a shallow stack Eric W. Biederman
2014-04-09 23:24                               ` Al Viro
2014-04-10  1:36                                 ` Eric W. Biederman
2014-04-12 22:15                                 ` Eric W. Biederman
2014-04-13  5:39                                   ` Al Viro
2014-04-13  5:49                                     ` Al Viro
2014-04-13  7:01                                     ` Eric W. Biederman
2014-04-13  7:51                                       ` Eric W. Biederman
2014-04-13 21:52                                         ` Al Viro
2014-04-14  7:38                                           ` [RFC][PATCH 0/4] No I/O from mntput Eric W. Biederman
2014-04-14  7:40                                             ` [PATCH 1/4] vfs: Remove useless loop in mntput_no_expire Eric W. Biederman
2014-04-14  7:40                                             ` [PATCH 2/4] vfs: Move autoclose of BSD accounting into a work queue Eric W. Biederman
2014-04-14  7:41                                             ` [PATCH 3/4] vfs: In mntput run deactivate_super on a shallow stack Eric W. Biederman
2014-04-14  7:42                                             ` [PATCH 4/4] vfs: Block intuitively in the case of BSD accounting files Eric W. Biederman
2014-04-16 22:03                                             ` [RFC][PATCH 0/4] No I/O from mntput Eric W. Biederman
2014-04-17 20:05                                               ` [GIT PULL] Detaching mounts on unlink for 3.15 Eric W. Biederman
2014-04-17 20:22                                                 ` Al Viro
2014-04-17 21:23                                                   ` Eric W. Biederman
2014-04-17 22:12                                                     ` Al Viro
2014-04-17 22:14                                                       ` Al Viro
2014-04-18  0:37                                                       ` Al Viro
2014-04-18  0:58                                                         ` Al Viro
2014-04-19  1:35                                                         ` Al Viro
2014-04-19  2:16                                                           ` Al Viro
2014-04-19  7:05                                                             ` Al Viro
2014-04-20  5:41                                                       ` Al Viro
2014-05-11 16:45                                                         ` Al Viro
2014-05-11 17:05                                                           ` Al Viro
2014-08-09  9:34                                                           ` Al Viro
2014-08-09  9:55                                                             ` Eric W. Biederman
2014-08-12 10:17                                                             ` Eric W. Biederman
2014-08-13 13:18                                                               ` Al Viro
2014-08-13 17:18                                                                 ` Eric W. Biederman
2014-04-09 22:01                       ` [GIT PULL] Detaching mounts on unlink for 3.15-rc1 Dave Chinner
2013-08-15 21:23 ` DoS with unprivileged mounts Rob Landley

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=87vbw1xqci.fsf@xmission.com \
    --to=ebiederm@xmission.com \
    --cc=bfields@fieldses.org \
    --cc=hch@infradead.org \
    --cc=kzak@redhat.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=luto@amacapital.net \
    --cc=miklos@szeredi.hu \
    --cc=rob@landley.net \
    --cc=serge@hallyn.com \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@zeniv.linux.org.uk \
    /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.