From: Vivek Goyal <vgoyal@redhat.com>
To: Amir Goldstein <amir73il@gmail.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>,
Al Viro <viro@zeniv.linux.org.uk>,
linux-unionfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
"#v4 . 16" <stable@vger.kernel.org>
Subject: Re: [PATCH v3 1/4] ovl: use insert_inode_locked4() to hash a newly created inode
Date: Tue, 15 May 2018 09:23:28 -0400 [thread overview]
Message-ID: <20180515132328.GA11678@redhat.com> (raw)
In-Reply-To: <1526379972-20923-2-git-send-email-amir73il@gmail.com>
On Tue, May 15, 2018 at 01:26:09PM +0300, Amir Goldstein wrote:
> Currently, there is a small window where ovl_obtain_alias() can
> race with ovl_instantiate() and create two different overlay inodes
> with the same underlying real non-dir non-hardlink inode.
>
> The race requires an adversary to guess the file handle of the
> yet to be created upper inode and decode the guessed file handle
> after ovl_creat_real(), but before ovl_instantiate().
>
> This patch fixes the race, by using insert_inode_locked4() to add
> a newly created inode to icache.
>
> If the newly created inode apears to already exist in icache (hashed
> by the same real upper inode), we export this error to user instead
> of silently not hashing the new inode.
So we might return an error to user saying operation failed, but still
create file on upper. Does that sound little odd?
I am wondering why can't we call ovl_get_inode() in object creation
path. That should take care of race between creation path and file
handle decode and only one of the paths will get to instantiate and
initialize ovl_inode and other path will wait.
Thanks
Vivek
>
> This race does not affect overlay directory inodes, because those
> are decoded via ovl_lookup_real() and not with ovl_obtain_alias(),
> so avoid using the new helper d_instantiate_new() to reduce backport
> dependencies.
>
> Backporting only makes sense for v4.16 where NFS export was introduced.
>
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Cc: <stable@vger.kernel.org> #v4.16
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
> fs/overlayfs/dir.c | 24 ++++++++++++++++++------
> fs/overlayfs/inode.c | 18 ++++++++++++++++++
> fs/overlayfs/overlayfs.h | 1 +
> 3 files changed, 37 insertions(+), 6 deletions(-)
>
> diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c
> index 47dc980e8b33..62e6733b755c 100644
> --- a/fs/overlayfs/dir.c
> +++ b/fs/overlayfs/dir.c
> @@ -183,14 +183,24 @@ static int ovl_set_opaque(struct dentry *dentry, struct dentry *upperdentry)
> }
>
> /* Common operations required to be done after creation of file on upper */
> -static void ovl_instantiate(struct dentry *dentry, struct inode *inode,
> - struct dentry *newdentry, bool hardlink)
> +static int ovl_instantiate(struct dentry *dentry, struct inode *inode,
> + struct dentry *newdentry, bool hardlink)
> {
> ovl_dir_modified(dentry->d_parent, false);
> - ovl_copyattr(d_inode(newdentry), inode);
> ovl_dentry_set_upper_alias(dentry);
> if (!hardlink) {
> - ovl_inode_update(inode, newdentry);
> + int err;
> +
> + ovl_inode_init(inode, newdentry, NULL);
> + /*
> + * XXX: if we ever use ovl_obtain_alias() to decode directory
> + * file handles, need to use ovl_insert_inode_locked() and
> + * d_instantiate_new() here to prevent ovl_obtain_alias()
> + * from sneaking in before d_instantiate().
> + */
> + err = ovl_insert_inode(inode, d_inode(newdentry));
> + if (err)
> + return err;
> } else {
> WARN_ON(ovl_inode_real(inode) != d_inode(newdentry));
> dput(newdentry);
> @@ -200,6 +210,8 @@ static void ovl_instantiate(struct dentry *dentry, struct inode *inode,
> /* Force lookup of new upper hardlink to find its lower */
> if (hardlink)
> d_drop(dentry);
> +
> + return 0;
> }
>
> static bool ovl_type_merge(struct dentry *dentry)
> @@ -238,7 +250,7 @@ static int ovl_create_upper(struct dentry *dentry, struct inode *inode,
> ovl_set_opaque(dentry, newdentry);
> }
>
> - ovl_instantiate(dentry, inode, newdentry, !!hardlink);
> + err = ovl_instantiate(dentry, inode, newdentry, !!hardlink);
> newdentry = NULL;
> out_dput:
> dput(newdentry);
> @@ -439,7 +451,7 @@ static int ovl_create_over_whiteout(struct dentry *dentry, struct inode *inode,
> if (err)
> goto out_cleanup;
> }
> - ovl_instantiate(dentry, inode, newdentry, !!hardlink);
> + err = ovl_instantiate(dentry, inode, newdentry, !!hardlink);
> newdentry = NULL;
> out_dput2:
> dput(upper);
> diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c
> index 7abcf96e94fc..060c534998d1 100644
> --- a/fs/overlayfs/inode.c
> +++ b/fs/overlayfs/inode.c
> @@ -741,6 +741,24 @@ static bool ovl_verify_inode(struct inode *inode, struct dentry *lowerdentry,
> return true;
> }
>
> +static int ovl_insert_inode_locked(struct inode *inode, struct inode *realinode)
> +{
> + return insert_inode_locked4(inode, (unsigned long) realinode,
> + ovl_inode_test, realinode);
> +}
> +
> +int ovl_insert_inode(struct inode *inode, struct inode *realinode)
> +{
> + int err;
> +
> + err = ovl_insert_inode_locked(inode, realinode);
> + if (err)
> + return err;
> +
> + unlock_new_inode(inode);
> + return 0;
> +}
> +
> struct inode *ovl_lookup_inode(struct super_block *sb, struct dentry *real,
> bool is_upper)
> {
> diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
> index caaa47cea2aa..642b25702092 100644
> --- a/fs/overlayfs/overlayfs.h
> +++ b/fs/overlayfs/overlayfs.h
> @@ -343,6 +343,7 @@ int ovl_update_time(struct inode *inode, struct timespec *ts, int flags);
> bool ovl_is_private_xattr(const char *name);
>
> struct inode *ovl_new_inode(struct super_block *sb, umode_t mode, dev_t rdev);
> +int ovl_insert_inode(struct inode *inode, struct inode *realinode);
> struct inode *ovl_lookup_inode(struct super_block *sb, struct dentry *real,
> bool is_upper);
> struct inode *ovl_get_inode(struct super_block *sb, struct dentry *upperdentry,
> --
> 2.7.4
>
next prev parent reply other threads:[~2018-05-15 13:23 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-15 10:26 [PATCH v3 0/4] Overlayfs mkdir related fixes Amir Goldstein
2018-05-15 10:26 ` [PATCH v3 1/4] ovl: use insert_inode_locked4() to hash a newly created inode Amir Goldstein
2018-05-15 13:23 ` Vivek Goyal [this message]
2018-05-15 13:37 ` Amir Goldstein
2018-05-16 8:34 ` Miklos Szeredi
2018-05-16 9:51 ` Amir Goldstein
2018-05-16 10:14 ` Miklos Szeredi
2018-05-16 11:03 ` Amir Goldstein
2018-05-17 6:03 ` Amir Goldstein
2018-05-17 8:10 ` Miklos Szeredi
2018-05-17 8:45 ` Amir Goldstein
2018-05-17 8:53 ` Miklos Szeredi
2018-05-17 8:58 ` Amir Goldstein
2018-05-17 9:07 ` Miklos Szeredi
2018-05-17 16:14 ` Amir Goldstein
2018-05-15 10:26 ` [PATCH v3 2/4] ovl: relax WARN_ON() real inode attributes mismatch Amir Goldstein
2018-05-15 12:48 ` Vivek Goyal
2018-05-15 12:55 ` Amir Goldstein
2018-05-16 10:29 ` Miklos Szeredi
2018-05-16 11:06 ` Amir Goldstein
2018-05-16 11:18 ` Miklos Szeredi
2018-05-16 13:46 ` Amir Goldstein
2018-05-15 10:26 ` [PATCH v3 3/4] ovl: create helper ovl_create_temp() Amir Goldstein
2018-05-16 10:41 ` Miklos Szeredi
2018-05-16 11:15 ` Amir Goldstein
2018-05-16 11:37 ` Miklos Szeredi
2018-05-15 10:26 ` [PATCH v3 4/4] ovl: make ovl_create_real() cope with vfs_mkdir() safely 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=20180515132328.GA11678@redhat.com \
--to=vgoyal@redhat.com \
--cc=amir73il@gmail.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-unionfs@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=stable@vger.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).