From: Ian Kent <raven@themaw.net>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Tejun Heo <tj@kernel.org>, Eric Sandeen <sandeen@sandeen.net>,
Fox Chen <foxhlchen@gmail.com>,
Brice Goglin <brice.goglin@gmail.com>,
Al Viro <viro@zeniv.linux.org.uk>,
Rick Lindsley <ricklind@linux.vnet.ibm.com>,
David Howells <dhowells@redhat.com>,
Marcelo Tosatti <mtosatti@redhat.com>,
linux-fsdevel <linux-fsdevel@vger.kernel.org>,
Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [REPOST PATCH v4 4/5] kernfs: use i_lock to protect concurrent inode updates
Date: Wed, 02 Jun 2021 13:41:38 +0800 [thread overview]
Message-ID: <b92354fb396cd9a93fce1b3d2bb2744f0535d22f.camel@themaw.net> (raw)
In-Reply-To: <CAJfpegshedor_ZiQ_8EdLGRG0AEWb5Sy5Pa4SwPg9+f196_mGg@mail.gmail.com>
On Tue, 2021-06-01 at 15:18 +0200, Miklos Szeredi wrote:
> On Fri, 28 May 2021 at 08:34, Ian Kent <raven@themaw.net> wrote:
> >
> > The inode operations .permission() and .getattr() use the kernfs
> > node
> > write lock but all that's needed is to keep the rb tree stable
> > while
> > updating the inode attributes as well as protecting the update
> > itself
> > against concurrent changes.
> >
> > And .permission() is called frequently during path walks and can
> > cause
> > quite a bit of contention between kernfs node operations and path
> > walks when the number of concurrent walks is high.
> >
> > To change kernfs_iop_getattr() and kernfs_iop_permission() to take
> > the rw sem read lock instead of the write lock an additional lock
> > is
> > needed to protect against multiple processes concurrently updating
> > the inode attributes and link count in kernfs_refresh_inode().
> >
> > The inode i_lock seems like the sensible thing to use to protect
> > these
> > inode attribute updates so use it in kernfs_refresh_inode().
> >
> > Signed-off-by: Ian Kent <raven@themaw.net>
> > ---
> > fs/kernfs/inode.c | 10 ++++++----
> > fs/kernfs/mount.c | 4 ++--
> > 2 files changed, 8 insertions(+), 6 deletions(-)
> >
> > diff --git a/fs/kernfs/inode.c b/fs/kernfs/inode.c
> > index 3b01e9e61f14e..6728ecd81eb37 100644
> > --- a/fs/kernfs/inode.c
> > +++ b/fs/kernfs/inode.c
> > @@ -172,6 +172,7 @@ static void kernfs_refresh_inode(struct
> > kernfs_node *kn, struct inode *inode)
> > {
> > struct kernfs_iattrs *attrs = kn->iattr;
> >
> > + spin_lock(&inode->i_lock);
> > inode->i_mode = kn->mode;
> > if (attrs)
> > /*
> > @@ -182,6 +183,7 @@ static void kernfs_refresh_inode(struct
> > kernfs_node *kn, struct inode *inode)
> >
> > if (kernfs_type(kn) == KERNFS_DIR)
> > set_nlink(inode, kn->dir.subdirs + 2);
> > + spin_unlock(&inode->i_lock);
> > }
> >
> > int kernfs_iop_getattr(struct user_namespace *mnt_userns,
> > @@ -191,9 +193,9 @@ int kernfs_iop_getattr(struct user_namespace
> > *mnt_userns,
> > struct inode *inode = d_inode(path->dentry);
> > struct kernfs_node *kn = inode->i_private;
> >
> > - down_write(&kernfs_rwsem);
> > + down_read(&kernfs_rwsem);
> > kernfs_refresh_inode(kn, inode);
> > - up_write(&kernfs_rwsem);
> > + up_read(&kernfs_rwsem);
> >
> > generic_fillattr(&init_user_ns, inode, stat);
> > return 0;
> > @@ -284,9 +286,9 @@ int kernfs_iop_permission(struct user_namespace
> > *mnt_userns,
> >
> > kn = inode->i_private;
> >
> > - down_write(&kernfs_rwsem);
> > + down_read(&kernfs_rwsem);
> > kernfs_refresh_inode(kn, inode);
> > - up_write(&kernfs_rwsem);
> > + up_read(&kernfs_rwsem);
> >
> > return generic_permission(&init_user_ns, inode, mask);
> > }
> > diff --git a/fs/kernfs/mount.c b/fs/kernfs/mount.c
> > index baa4155ba2edf..f2f909d09f522 100644
> > --- a/fs/kernfs/mount.c
> > +++ b/fs/kernfs/mount.c
> > @@ -255,9 +255,9 @@ static int kernfs_fill_super(struct super_block
> > *sb, struct kernfs_fs_context *k
> > sb->s_shrink.seeks = 0;
> >
> > /* get root inode, initialize and unlock it */
> > - down_write(&kernfs_rwsem);
> > + down_read(&kernfs_rwsem);
> > inode = kernfs_get_inode(sb, info->root->kn);
> > - up_write(&kernfs_rwsem);
> > + up_read(&kernfs_rwsem);
> > if (!inode) {
> > pr_debug("kernfs: could not get root inode\n");
> > return -ENOMEM;
> >
>
> This last hunk is not mentioned in the patch header. Why is this
> needed?
Yes, that's right.
The lock is needed to keep the node rb tree stable.
kernfs_get_inode() calls kernfs_refresh_inode() indirectly so
since the i_lock is probably not needed here this hunk could
just as well have gone into the rwsem change but because of
that kernfs_refresh_inode() call it also makes sense to put
it here.
I'd prefer to keep it here and clearly what's going on isn't
as obvious as I thought so I can add this reasoning to the
description if you still think it's worth while?
>
> Otherwise looks good.
>
> Thanks,
> Miklos
next prev parent reply other threads:[~2021-06-02 5:42 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-28 6:33 [REPOST PATCH v4 0/5] kernfs: proposed locking and concurrency improvement Ian Kent
2021-05-28 6:33 ` [REPOST PATCH v4 1/5] kernfs: move revalidate to be near lookup Ian Kent
2021-06-03 14:50 ` Eric W. Biederman
2021-06-04 2:29 ` Ian Kent
2021-05-28 6:34 ` [REPOST PATCH v4 2/5] kernfs: use VFS negative dentry caching Ian Kent
2021-06-01 12:41 ` Miklos Szeredi
2021-06-02 3:44 ` Ian Kent
2021-06-02 8:58 ` Miklos Szeredi
2021-06-02 10:57 ` Ian Kent
2021-06-03 2:15 ` Ian Kent
2021-06-03 23:57 ` Ian Kent
2021-06-04 1:07 ` Ian Kent
2021-06-03 17:26 ` Eric W. Biederman
2021-06-03 18:06 ` Miklos Szeredi
2021-06-03 22:02 ` Eric W. Biederman
2021-06-04 3:14 ` Ian Kent
2021-06-04 14:28 ` Eric W. Biederman
2021-06-05 3:19 ` Ian Kent
2021-06-05 20:52 ` Eric W. Biederman
2021-05-28 6:34 ` [REPOST PATCH v4 3/5] kernfs: switch kernfs to use an rwsem Ian Kent
2021-06-01 13:11 ` Miklos Szeredi
2021-06-03 16:59 ` Eric W. Biederman
2021-05-28 6:34 ` [REPOST PATCH v4 4/5] kernfs: use i_lock to protect concurrent inode updates Ian Kent
2021-05-31 14:53 ` [kernfs] 9a658329cd: stress-ng.get.ops_per_sec 191.4% improvement kernel test robot
2021-06-01 13:18 ` [REPOST PATCH v4 4/5] kernfs: use i_lock to protect concurrent inode updates Miklos Szeredi
2021-06-02 5:41 ` Ian Kent [this message]
2021-05-28 6:34 ` [REPOST PATCH v4 5/5] kernfs: add kernfs_need_inode_refresh() Ian Kent
2021-05-28 8:56 ` [REPOST PATCH v4 0/5] kernfs: proposed locking and concurrency improvement Greg Kroah-Hartman
2021-05-28 11:56 ` Fox Chen
2021-05-30 4:44 ` Fox Chen
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=b92354fb396cd9a93fce1b3d2bb2744f0535d22f.camel@themaw.net \
--to=raven@themaw.net \
--cc=brice.goglin@gmail.com \
--cc=dhowells@redhat.com \
--cc=foxhlchen@gmail.com \
--cc=gregkh@linuxfoundation.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=mtosatti@redhat.com \
--cc=ricklind@linux.vnet.ibm.com \
--cc=sandeen@sandeen.net \
--cc=tj@kernel.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 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).