[v7,2/6] ocfs2: Switch to security_inode_init_security()

Message ID 20221201104125.919483-3-roberto.sassu@huaweicloud.com
State New
Headers
Series evm: Do HMAC of multiple per LSM xattrs for new inodes |

Commit Message

Roberto Sassu Dec. 1, 2022, 10:41 a.m. UTC
  From: Roberto Sassu <roberto.sassu@huawei.com>

In preparation for removing security_old_inode_init_security(), switch to
security_inode_init_security().

Extend the existing ocfs2_initxattrs() to take the
ocfs2_security_xattr_info structure from fs_info, and populate the
name/value/len triple with the first xattr provided by LSMs.

As fs_info was not used before, ocfs2_initxattrs() can now handle the case
of replicating the behavior of security_old_inode_init_security(), i.e.
just obtaining the xattr, in addition to setting all xattrs provided by
LSMs.

Supporting multiple xattrs is not currently supported where
security_old_inode_init_security() was called (mknod, symlink), as it
requires non-trivial changes that can be done at a later time. Like for
reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
not the first to set it, its xattr will be discarded; if it is the first,
it does not have xattrs to calculate the HMAC on).

Finally, modify the handling of the return value from
ocfs2_init_security_get(). As security_inode_init_security() does not
return -EOPNOTSUPP, remove this case and directly handle the error if the
return value is not zero.

However, the previous case of receiving -EOPNOTSUPP should be still
taken into account, as security_inode_init_security() could return zero
without setting xattrs and ocfs2 would consider it as if the xattr was set.

Instead, if security_inode_init_security() returned zero, look at the xattr
if it was set, and behave accordingly, i.e. set si->enable to zero to
notify to the functions following ocfs2_init_security_get() that the xattr
is not available (same as if security_old_inode_init_security() returned
-EOPNOTSUPP).

Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
---
 fs/ocfs2/namei.c | 18 ++++++------------
 fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
 2 files changed, 32 insertions(+), 16 deletions(-)
  

Comments

Roberto Sassu Jan. 10, 2023, 8:55 a.m. UTC | #1
On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> From: Roberto Sassu <roberto.sassu@huawei.com>
> 
> In preparation for removing security_old_inode_init_security(), switch to
> security_inode_init_security().
> 
> Extend the existing ocfs2_initxattrs() to take the
> ocfs2_security_xattr_info structure from fs_info, and populate the
> name/value/len triple with the first xattr provided by LSMs.

Hi Mark, Joel, Joseph

some time ago I sent this patch set to switch to the newer
function security_inode_init_security(). Almost all the other parts of
this patch set have been reviewed, and the patch set itself should be
ready to be merged.

I kindly ask if you could have a look at this patch and give your
Reviewed-by, so that Paul could take the patch set.

Thanks a lot!

Roberto

> As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> of replicating the behavior of security_old_inode_init_security(), i.e.
> just obtaining the xattr, in addition to setting all xattrs provided by
> LSMs.
> 
> Supporting multiple xattrs is not currently supported where
> security_old_inode_init_security() was called (mknod, symlink), as it
> requires non-trivial changes that can be done at a later time. Like for
> reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> not the first to set it, its xattr will be discarded; if it is the first,
> it does not have xattrs to calculate the HMAC on).
> 
> Finally, modify the handling of the return value from
> ocfs2_init_security_get(). As security_inode_init_security() does not
> return -EOPNOTSUPP, remove this case and directly handle the error if the
> return value is not zero.
> 
> However, the previous case of receiving -EOPNOTSUPP should be still
> taken into account, as security_inode_init_security() could return zero
> without setting xattrs and ocfs2 would consider it as if the xattr was set.
> 
> Instead, if security_inode_init_security() returned zero, look at the xattr
> if it was set, and behave accordingly, i.e. set si->enable to zero to
> notify to the functions following ocfs2_init_security_get() that the xattr
> is not available (same as if security_old_inode_init_security() returned
> -EOPNOTSUPP).
> 
> Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
> ---
>  fs/ocfs2/namei.c | 18 ++++++------------
>  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
>  2 files changed, 32 insertions(+), 16 deletions(-)
> 
> diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> index 05f32989bad6..55fba81cd2d1 100644
> --- a/fs/ocfs2/namei.c
> +++ b/fs/ocfs2/namei.c
> @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>  	int want_meta = 0;
>  	int xattr_credits = 0;
>  	struct ocfs2_security_xattr_info si = {
> +		.name = NULL,
>  		.enable = 1,
>  	};
>  	int did_quota_inode = 0;
> @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>  	/* get security xattr */
>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>  	if (status) {
> -		if (status == -EOPNOTSUPP)
> -			si.enable = 0;
> -		else {
> -			mlog_errno(status);
> -			goto leave;
> -		}
> +		mlog_errno(status);
> +		goto leave;
>  	}
>  
>  	/* calculate meta data/clusters for setting security and acl xattr */
> @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>  	int want_clusters = 0;
>  	int xattr_credits = 0;
>  	struct ocfs2_security_xattr_info si = {
> +		.name = NULL,
>  		.enable = 1,
>  	};
>  	int did_quota = 0, did_quota_inode = 0;
> @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>  	/* get security xattr */
>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>  	if (status) {
> -		if (status == -EOPNOTSUPP)
> -			si.enable = 0;
> -		else {
> -			mlog_errno(status);
> -			goto bail;
> -		}
> +		mlog_errno(status);
> +		goto bail;
>  	}
>  
>  	/* calculate meta data/clusters for setting security xattr */
> diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> index 95d0611c5fc7..55699c573541 100644
> --- a/fs/ocfs2/xattr.c
> +++ b/fs/ocfs2/xattr.c
> @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
>  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
>  		     void *fs_info)
>  {
> +	struct ocfs2_security_xattr_info *si = fs_info;
>  	const struct xattr *xattr;
>  	int err = 0;
>  
> +	if (si) {
> +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> +				    GFP_KERNEL);
> +		if (!si->value)
> +			return -ENOMEM;
> +
> +		si->name = xattr_array->name;
> +		si->value_len = xattr_array->value_len;
> +		return 0;
> +	}
> +
>  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
>  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
>  				      xattr->name, xattr->value,
> @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
>  			    const struct qstr *qstr,
>  			    struct ocfs2_security_xattr_info *si)
>  {
> +	int ret;
> +
>  	/* check whether ocfs2 support feature xattr */
>  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
>  		return -EOPNOTSUPP;
> -	if (si)
> -		return security_old_inode_init_security(inode, dir, qstr,
> -							&si->name, &si->value,
> -							&si->value_len);
> +	if (si) {
> +		ret = security_inode_init_security(inode, dir, qstr,
> +						   &ocfs2_initxattrs, si);
> +		/*
> +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> +		 * we have to check the xattr ourselves.
> +		 */
> +		if (!ret && !si->name)
> +			si->enable = 0;
> +
> +		return ret;
> +	}
>  
>  	return security_inode_init_security(inode, dir, qstr,
>  					    &ocfs2_initxattrs, NULL);
  
Paul Moore Jan. 12, 2023, 5:21 p.m. UTC | #2
On Tue, Jan 10, 2023 at 3:56 AM Roberto Sassu
<roberto.sassu@huaweicloud.com> wrote:
> On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > From: Roberto Sassu <roberto.sassu@huawei.com>
> >
> > In preparation for removing security_old_inode_init_security(), switch to
> > security_inode_init_security().
> >
> > Extend the existing ocfs2_initxattrs() to take the
> > ocfs2_security_xattr_info structure from fs_info, and populate the
> > name/value/len triple with the first xattr provided by LSMs.
>
> Hi Mark, Joel, Joseph
>
> some time ago I sent this patch set to switch to the newer
> function security_inode_init_security(). Almost all the other parts of
> this patch set have been reviewed, and the patch set itself should be
> ready to be merged.
>
> I kindly ask if you could have a look at this patch and give your
> Reviewed-by, so that Paul could take the patch set.

I've been pushing to clean up some of the LSM interfaces to try and
simplify things and remove as many special cases as possible,
Roberto's work in this patchset is part of that.  I would really
appreciate it if the vfs/ocfs2 folks could give patch 2/6 a quick look
to make sure you are okay with the changes.

I realize that the various end-of-year holidays tend to slow things
down a bit, but this patchset has been on the lists for over a month
now; if I don't hear anything in the next week or two I'll assume you
folks are okay with these patches ...
  
Roberto Sassu Feb. 8, 2023, 2:33 p.m. UTC | #3
On Thu, 2023-01-12 at 12:21 -0500, Paul Moore wrote:
> On Tue, Jan 10, 2023 at 3:56 AM Roberto Sassu
> <roberto.sassu@huaweicloud.com> wrote:
> > On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > > From: Roberto Sassu <roberto.sassu@huawei.com>
> > > 
> > > In preparation for removing security_old_inode_init_security(), switch to
> > > security_inode_init_security().
> > > 
> > > Extend the existing ocfs2_initxattrs() to take the
> > > ocfs2_security_xattr_info structure from fs_info, and populate the
> > > name/value/len triple with the first xattr provided by LSMs.
> > 
> > Hi Mark, Joel, Joseph
> > 
> > some time ago I sent this patch set to switch to the newer
> > function security_inode_init_security(). Almost all the other parts of
> > this patch set have been reviewed, and the patch set itself should be
> > ready to be merged.
> > 
> > I kindly ask if you could have a look at this patch and give your
> > Reviewed-by, so that Paul could take the patch set.
> 
> I've been pushing to clean up some of the LSM interfaces to try and
> simplify things and remove as many special cases as possible,
> Roberto's work in this patchset is part of that.  I would really
> appreciate it if the vfs/ocfs2 folks could give patch 2/6 a quick look
> to make sure you are okay with the changes.
> 
> I realize that the various end-of-year holidays tend to slow things
> down a bit, but this patchset has been on the lists for over a month
> now; if I don't hear anything in the next week or two I'll assume you
> folks are okay with these patches ...

Hi Paul

is this patch set going to land in 6.3?

Thanks

Roberto
  
Paul Moore Feb. 9, 2023, 9:05 p.m. UTC | #4
On Wed, Feb 8, 2023 at 9:33 AM Roberto Sassu
<roberto.sassu@huaweicloud.com> wrote:
> On Thu, 2023-01-12 at 12:21 -0500, Paul Moore wrote:
> > On Tue, Jan 10, 2023 at 3:56 AM Roberto Sassu
> > <roberto.sassu@huaweicloud.com> wrote:
> > > On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > > > From: Roberto Sassu <roberto.sassu@huawei.com>
> > > >
> > > > In preparation for removing security_old_inode_init_security(), switch to
> > > > security_inode_init_security().
> > > >
> > > > Extend the existing ocfs2_initxattrs() to take the
> > > > ocfs2_security_xattr_info structure from fs_info, and populate the
> > > > name/value/len triple with the first xattr provided by LSMs.
> > >
> > > Hi Mark, Joel, Joseph
> > >
> > > some time ago I sent this patch set to switch to the newer
> > > function security_inode_init_security(). Almost all the other parts of
> > > this patch set have been reviewed, and the patch set itself should be
> > > ready to be merged.
> > >
> > > I kindly ask if you could have a look at this patch and give your
> > > Reviewed-by, so that Paul could take the patch set.
> >
> > I've been pushing to clean up some of the LSM interfaces to try and
> > simplify things and remove as many special cases as possible,
> > Roberto's work in this patchset is part of that.  I would really
> > appreciate it if the vfs/ocfs2 folks could give patch 2/6 a quick look
> > to make sure you are okay with the changes.
> >
> > I realize that the various end-of-year holidays tend to slow things
> > down a bit, but this patchset has been on the lists for over a month
> > now; if I don't hear anything in the next week or two I'll assume you
> > folks are okay with these patches ...
>
> Hi Paul
>
> is this patch set going to land in 6.3?

Hi Roberto,

I had really hoped the vfs/ocfs2 folks would have commented on this by
now, but it's been over two months now with no comments that I can see
so I think we have to do it ourselves via the LSM tree.  It's
obviously too late for the upcoming merge window, so no v6.3, but I
think we can merge it *after* the upcoming merge window closes,
assuming we get ACKs from Mimi on the EVM bits (I still need to review
it too, but I'm not expecting anything too bad).
  
Mimi Zohar Feb. 17, 2023, 7:51 p.m. UTC | #5
On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> From: Roberto Sassu <roberto.sassu@huawei.com>
> 
> In preparation for removing security_old_inode_init_security(), switch to
> security_inode_init_security().
> 
> Extend the existing ocfs2_initxattrs() to take the
> ocfs2_security_xattr_info structure from fs_info, and populate the
> name/value/len triple with the first xattr provided by LSMs.
> 
> As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> of replicating the behavior of security_old_inode_init_security(), i.e.
> just obtaining the xattr, in addition to setting all xattrs provided by
> LSMs.
> 
> Supporting multiple xattrs is not currently supported where
> security_old_inode_init_security() was called (mknod, symlink), as it
> requires non-trivial changes that can be done at a later time. Like for
> reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> not the first to set it, its xattr will be discarded; if it is the first,
> it does not have xattrs to calculate the HMAC on).
> 
> Finally, modify the handling of the return value from
> ocfs2_init_security_get(). As security_inode_init_security() does not
> return -EOPNOTSUPP, remove this case and directly handle the error if the
> return value is not zero.
> 
> However, the previous case of receiving -EOPNOTSUPP should be still
> taken into account, as security_inode_init_security() could return zero
> without setting xattrs and ocfs2 would consider it as if the xattr was set.
> 
> Instead, if security_inode_init_security() returned zero, look at the xattr
> if it was set, and behave accordingly, i.e. set si->enable to zero to
> notify to the functions following ocfs2_init_security_get() that the xattr
> is not available (same as if security_old_inode_init_security() returned
> -EOPNOTSUPP).
> 
> Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>

My previous review missed a couple of concerns.

> ---
>  fs/ocfs2/namei.c | 18 ++++++------------
>  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
>  2 files changed, 32 insertions(+), 16 deletions(-)
> 
> diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> index 05f32989bad6..55fba81cd2d1 100644
> --- a/fs/ocfs2/namei.c
> +++ b/fs/ocfs2/namei.c
> @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>  	int want_meta = 0;
>  	int xattr_credits = 0;
>  	struct ocfs2_security_xattr_info si = {
> +		.name = NULL,
>  		.enable = 1,
>  	};
>  	int did_quota_inode = 0;
> @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>  	/* get security xattr */
>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>  	if (status) {
> -		if (status == -EOPNOTSUPP)
> -			si.enable = 0;
> -		else {
> -			mlog_errno(status);
> -			goto leave;
> -		}

Although security_inode_init_security() does not return -EOPNOTSUPP, 
ocfs2_init_security_get() could.  Refer to commit 8154da3d2114 ("ocfs2:
Add incompatible flag for extended attribute").   It was added as a
temporary solution back in 2008, so it is highly unlikely that it is
still needed.

> +		mlog_errno(status);
> +		goto leave;

Without the -EOPNOTSUPP test, ocfs2_mknod() would not create the inode;
and similarly ocfs2_symlink(), below, would not create the symlink.  It
would be safer not to remove the -EOPNOTSUPP test.

>  	}
>  
>  	/* calculate meta data/clusters for setting security and acl xattr */
> @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>  	int want_clusters = 0;
>  	int xattr_credits = 0;
>  	struct ocfs2_security_xattr_info si = {
> +		.name = NULL,
>  		.enable = 1,
>  	};
>  	int did_quota = 0, did_quota_inode = 0;
> @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>  	/* get security xattr */
>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>  	if (status) {
> -		if (status == -EOPNOTSUPP)
> -			si.enable = 0;
> -		else {
> -			mlog_errno(status);
> -			goto bail;
> -		}
> +		mlog_errno(status);
> +		goto bail;
>  	}
>  
>  	/* calculate meta data/clusters for setting security xattr */
> diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> index 95d0611c5fc7..55699c573541 100644
> --- a/fs/ocfs2/xattr.c
> +++ b/fs/ocfs2/xattr.c
> @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
>  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
>  		     void *fs_info)
>  {
> +	struct ocfs2_security_xattr_info *si = fs_info;
>  	const struct xattr *xattr;
>  	int err = 0;
>  
> +	if (si) {
> +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> +				    GFP_KERNEL);
> +		if (!si->value)
> +			return -ENOMEM;
> +
> +		si->name = xattr_array->name;
> +		si->value_len = xattr_array->value_len;
> +		return 0;
> +	}
> +
>  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
>  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
>  				      xattr->name, xattr->value,
> @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
>  			    const struct qstr *qstr,
>  			    struct ocfs2_security_xattr_info *si)
>  {
> +	int ret;
> +
>  	/* check whether ocfs2 support feature xattr */
>  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
>  		return -EOPNOTSUPP;
> -	if (si)
> -		return security_old_inode_init_security(inode, dir, qstr,
> -							&si->name, &si->value,
> -							&si->value_len);
> +	if (si) {
> +		ret = security_inode_init_security(inode, dir, qstr,
> +						   &ocfs2_initxattrs, si);

The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
security_old_inode_init_security() and security_inode_init_security(),
but return different values.  In the former case, it returns
-EOPNOTSUPP.  In the latter case, it returns 0.  The question is
whether or not we need to be concerned about private inodes on ocfs2.  
If private inodes on ocfs2 are possible, then ocsf2_mknod() or
ocfs2_symlink() would fail to create the inode or symlink.

> +		/*
> +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> +		 * we have to check the xattr ourselves.
> +		 */
> +		if (!ret && !si->name)
> +			si->enable = 0;
> +
> +		return ret;
> +	}
>  
>  	return security_inode_init_security(inode, dir, qstr,
>  					    &ocfs2_initxattrs, NULL);
  
Mimi Zohar Feb. 17, 2023, 9:30 p.m. UTC | #6
On Fri, 2023-02-17 at 14:51 -0500, Mimi Zohar wrote:
> On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > From: Roberto Sassu <roberto.sassu@huawei.com>
> > 
> > In preparation for removing security_old_inode_init_security(), switch to
> > security_inode_init_security().
> > 
> > Extend the existing ocfs2_initxattrs() to take the
> > ocfs2_security_xattr_info structure from fs_info, and populate the
> > name/value/len triple with the first xattr provided by LSMs.
> > 
> > As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> > of replicating the behavior of security_old_inode_init_security(), i.e.
> > just obtaining the xattr, in addition to setting all xattrs provided by
> > LSMs.
> > 
> > Supporting multiple xattrs is not currently supported where
> > security_old_inode_init_security() was called (mknod, symlink), as it
> > requires non-trivial changes that can be done at a later time. Like for
> > reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> > not the first to set it, its xattr will be discarded; if it is the first,
> > it does not have xattrs to calculate the HMAC on).
> > 
> > Finally, modify the handling of the return value from
> > ocfs2_init_security_get(). As security_inode_init_security() does not
> > return -EOPNOTSUPP, remove this case and directly handle the error if the
> > return value is not zero.
> > 
> > However, the previous case of receiving -EOPNOTSUPP should be still
> > taken into account, as security_inode_init_security() could return zero
> > without setting xattrs and ocfs2 would consider it as if the xattr was set.
> > 
> > Instead, if security_inode_init_security() returned zero, look at the xattr
> > if it was set, and behave accordingly, i.e. set si->enable to zero to
> > notify to the functions following ocfs2_init_security_get() that the xattr
> > is not available (same as if security_old_inode_init_security() returned
> > -EOPNOTSUPP).
> > 
> > Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> > Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
> 
> My previous review missed a couple of concerns.
> 
> > ---
> >  fs/ocfs2/namei.c | 18 ++++++------------
> >  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
> >  2 files changed, 32 insertions(+), 16 deletions(-)
> > 
> > diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> > index 05f32989bad6..55fba81cd2d1 100644
> > --- a/fs/ocfs2/namei.c
> > +++ b/fs/ocfs2/namei.c
> > @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> >  	int want_meta = 0;
> >  	int xattr_credits = 0;
> >  	struct ocfs2_security_xattr_info si = {
> > +		.name = NULL,
> >  		.enable = 1,
> >  	};
> >  	int did_quota_inode = 0;
> > @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> >  	/* get security xattr */
> >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> >  	if (status) {
> > -		if (status == -EOPNOTSUPP)
> > -			si.enable = 0;
> > -		else {
> > -			mlog_errno(status);
> > -			goto leave;
> > -		}
> 
> Although security_inode_init_security() does not return -EOPNOTSUPP, 
> ocfs2_init_security_get() could.  Refer to commit 8154da3d2114 ("ocfs2:
> Add incompatible flag for extended attribute").   It was added as a
> temporary solution back in 2008, so it is highly unlikely that it is
> still needed.
> 
> > +		mlog_errno(status);
> > +		goto leave;
> 
> Without the -EOPNOTSUPP test, ocfs2_mknod() would not create the inode;
> and similarly ocfs2_symlink(), below, would not create the symlink.  It
> would be safer not to remove the -EOPNOTSUPP test.
> 
> >  	}
> >  
> >  	/* calculate meta data/clusters for setting security and acl xattr */
> > @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> >  	int want_clusters = 0;
> >  	int xattr_credits = 0;
> >  	struct ocfs2_security_xattr_info si = {
> > +		.name = NULL,
> >  		.enable = 1,
> >  	};
> >  	int did_quota = 0, did_quota_inode = 0;
> > @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> >  	/* get security xattr */
> >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> >  	if (status) {
> > -		if (status == -EOPNOTSUPP)
> > -			si.enable = 0;
> > -		else {
> > -			mlog_errno(status);
> > -			goto bail;
> > -		}
> > +		mlog_errno(status);
> > +		goto bail;
> >  	}
> >  
> >  	/* calculate meta data/clusters for setting security xattr */
> > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > index 95d0611c5fc7..55699c573541 100644
> > --- a/fs/ocfs2/xattr.c
> > +++ b/fs/ocfs2/xattr.c
> > @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
> >  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
> >  		     void *fs_info)
> >  {
> > +	struct ocfs2_security_xattr_info *si = fs_info;
> >  	const struct xattr *xattr;
> >  	int err = 0;
> >  
> > +	if (si) {
> > +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> > +				    GFP_KERNEL);
> > +		if (!si->value)
> > +			return -ENOMEM;
> > +
> > +		si->name = xattr_array->name;
> > +		si->value_len = xattr_array->value_len;
> > +		return 0;
> > +	}
> > +
> >  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
> >  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
> >  				      xattr->name, xattr->value,
> > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> >  			    const struct qstr *qstr,
> >  			    struct ocfs2_security_xattr_info *si)
> >  {
> > +	int ret;
> > +
> >  	/* check whether ocfs2 support feature xattr */
> >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> >  		return -EOPNOTSUPP;
> > -	if (si)
> > -		return security_old_inode_init_security(inode, dir, qstr,
> > -							&si->name, &si->value,
> > -							&si->value_len);
> > +	if (si) {
> > +		ret = security_inode_init_security(inode, dir, qstr,
> > +						   &ocfs2_initxattrs, si);
> 
> The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> security_old_inode_init_security() and security_inode_init_security(),
> but return different values.  In the former case, it returns
> -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> whether or not we need to be concerned about private inodes on ocfs2.  
> If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> ocfs2_symlink() would fail to create the inode or symlink.

Correction, previously when returning -EOPNOTSUPP for private inodes,
xattrs would not be wrriten.  By returning 0 without setting si->enable 
to 0, xattrs will be written.
 
> 
> > +		/*
> > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > +		 * we have to check the xattr ourselves.
> > +		 */
> > +		if (!ret && !si->name)
> > +			si->enable = 0;
> > +
> > +		return ret;
> > +	}
> >  
> >  	return security_inode_init_security(inode, dir, qstr,
> >  					    &ocfs2_initxattrs, NULL);
>
  
Roberto Sassu Feb. 20, 2023, 8:34 a.m. UTC | #7
On Fri, 2023-02-17 at 14:51 -0500, Mimi Zohar wrote:
> On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > From: Roberto Sassu <roberto.sassu@huawei.com>
> > 
> > In preparation for removing security_old_inode_init_security(), switch to
> > security_inode_init_security().
> > 
> > Extend the existing ocfs2_initxattrs() to take the
> > ocfs2_security_xattr_info structure from fs_info, and populate the
> > name/value/len triple with the first xattr provided by LSMs.
> > 
> > As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> > of replicating the behavior of security_old_inode_init_security(), i.e.
> > just obtaining the xattr, in addition to setting all xattrs provided by
> > LSMs.
> > 
> > Supporting multiple xattrs is not currently supported where
> > security_old_inode_init_security() was called (mknod, symlink), as it
> > requires non-trivial changes that can be done at a later time. Like for
> > reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> > not the first to set it, its xattr will be discarded; if it is the first,
> > it does not have xattrs to calculate the HMAC on).
> > 
> > Finally, modify the handling of the return value from
> > ocfs2_init_security_get(). As security_inode_init_security() does not
> > return -EOPNOTSUPP, remove this case and directly handle the error if the
> > return value is not zero.
> > 
> > However, the previous case of receiving -EOPNOTSUPP should be still
> > taken into account, as security_inode_init_security() could return zero
> > without setting xattrs and ocfs2 would consider it as if the xattr was set.
> > 
> > Instead, if security_inode_init_security() returned zero, look at the xattr
> > if it was set, and behave accordingly, i.e. set si->enable to zero to
> > notify to the functions following ocfs2_init_security_get() that the xattr
> > is not available (same as if security_old_inode_init_security() returned
> > -EOPNOTSUPP).
> > 
> > Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> > Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
> 
> My previous review missed a couple of concerns.
> 
> > ---
> >  fs/ocfs2/namei.c | 18 ++++++------------
> >  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
> >  2 files changed, 32 insertions(+), 16 deletions(-)
> > 
> > diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> > index 05f32989bad6..55fba81cd2d1 100644
> > --- a/fs/ocfs2/namei.c
> > +++ b/fs/ocfs2/namei.c
> > @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> >  	int want_meta = 0;
> >  	int xattr_credits = 0;
> >  	struct ocfs2_security_xattr_info si = {
> > +		.name = NULL,
> >  		.enable = 1,
> >  	};
> >  	int did_quota_inode = 0;
> > @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> >  	/* get security xattr */
> >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> >  	if (status) {
> > -		if (status == -EOPNOTSUPP)
> > -			si.enable = 0;
> > -		else {
> > -			mlog_errno(status);
> > -			goto leave;
> > -		}
> 
> Although security_inode_init_security() does not return -EOPNOTSUPP, 
> ocfs2_init_security_get() could.  Refer to commit 8154da3d2114 ("ocfs2:
> Add incompatible flag for extended attribute").   It was added as a
> temporary solution back in 2008, so it is highly unlikely that it is
> still needed.
> 
> > +		mlog_errno(status);
> > +		goto leave;
> 
> Without the -EOPNOTSUPP test, ocfs2_mknod() would not create the inode;
> and similarly ocfs2_symlink(), below, would not create the symlink.  It
> would be safer not to remove the -EOPNOTSUPP test.

You are absolutely right. Will add it back.

Thanks

Roberto

> >  	}
> >  
> >  	/* calculate meta data/clusters for setting security and acl xattr */
> > @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> >  	int want_clusters = 0;
> >  	int xattr_credits = 0;
> >  	struct ocfs2_security_xattr_info si = {
> > +		.name = NULL,
> >  		.enable = 1,
> >  	};
> >  	int did_quota = 0, did_quota_inode = 0;
> > @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> >  	/* get security xattr */
> >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> >  	if (status) {
> > -		if (status == -EOPNOTSUPP)
> > -			si.enable = 0;
> > -		else {
> > -			mlog_errno(status);
> > -			goto bail;
> > -		}
> > +		mlog_errno(status);
> > +		goto bail;
> >  	}
> >  
> >  	/* calculate meta data/clusters for setting security xattr */
> > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > index 95d0611c5fc7..55699c573541 100644
> > --- a/fs/ocfs2/xattr.c
> > +++ b/fs/ocfs2/xattr.c
> > @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
> >  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
> >  		     void *fs_info)
> >  {
> > +	struct ocfs2_security_xattr_info *si = fs_info;
> >  	const struct xattr *xattr;
> >  	int err = 0;
> >  
> > +	if (si) {
> > +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> > +				    GFP_KERNEL);
> > +		if (!si->value)
> > +			return -ENOMEM;
> > +
> > +		si->name = xattr_array->name;
> > +		si->value_len = xattr_array->value_len;
> > +		return 0;
> > +	}
> > +
> >  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
> >  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
> >  				      xattr->name, xattr->value,
> > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> >  			    const struct qstr *qstr,
> >  			    struct ocfs2_security_xattr_info *si)
> >  {
> > +	int ret;
> > +
> >  	/* check whether ocfs2 support feature xattr */
> >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> >  		return -EOPNOTSUPP;
> > -	if (si)
> > -		return security_old_inode_init_security(inode, dir, qstr,
> > -							&si->name, &si->value,
> > -							&si->value_len);
> > +	if (si) {
> > +		ret = security_inode_init_security(inode, dir, qstr,
> > +						   &ocfs2_initxattrs, si);
> 
> The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> security_old_inode_init_security() and security_inode_init_security(),
> but return different values.  In the former case, it returns
> -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> whether or not we need to be concerned about private inodes on ocfs2.  
> If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> ocfs2_symlink() would fail to create the inode or symlink.
> 
> > +		/*
> > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > +		 * we have to check the xattr ourselves.
> > +		 */
> > +		if (!ret && !si->name)
> > +			si->enable = 0;
> > +
> > +		return ret;
> > +	}
> >  
> >  	return security_inode_init_security(inode, dir, qstr,
> >  					    &ocfs2_initxattrs, NULL);
  
Roberto Sassu Feb. 20, 2023, 9:27 a.m. UTC | #8
On Fri, 2023-02-17 at 16:30 -0500, Mimi Zohar wrote:
> On Fri, 2023-02-17 at 14:51 -0500, Mimi Zohar wrote:
> > On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > > From: Roberto Sassu <roberto.sassu@huawei.com>
> > > 
> > > In preparation for removing security_old_inode_init_security(), switch to
> > > security_inode_init_security().
> > > 
> > > Extend the existing ocfs2_initxattrs() to take the
> > > ocfs2_security_xattr_info structure from fs_info, and populate the
> > > name/value/len triple with the first xattr provided by LSMs.
> > > 
> > > As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> > > of replicating the behavior of security_old_inode_init_security(), i.e.
> > > just obtaining the xattr, in addition to setting all xattrs provided by
> > > LSMs.
> > > 
> > > Supporting multiple xattrs is not currently supported where
> > > security_old_inode_init_security() was called (mknod, symlink), as it
> > > requires non-trivial changes that can be done at a later time. Like for
> > > reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> > > not the first to set it, its xattr will be discarded; if it is the first,
> > > it does not have xattrs to calculate the HMAC on).
> > > 
> > > Finally, modify the handling of the return value from
> > > ocfs2_init_security_get(). As security_inode_init_security() does not
> > > return -EOPNOTSUPP, remove this case and directly handle the error if the
> > > return value is not zero.
> > > 
> > > However, the previous case of receiving -EOPNOTSUPP should be still
> > > taken into account, as security_inode_init_security() could return zero
> > > without setting xattrs and ocfs2 would consider it as if the xattr was set.
> > > 
> > > Instead, if security_inode_init_security() returned zero, look at the xattr
> > > if it was set, and behave accordingly, i.e. set si->enable to zero to
> > > notify to the functions following ocfs2_init_security_get() that the xattr
> > > is not available (same as if security_old_inode_init_security() returned
> > > -EOPNOTSUPP).
> > > 
> > > Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> > > Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
> > 
> > My previous review missed a couple of concerns.
> > 
> > > ---
> > >  fs/ocfs2/namei.c | 18 ++++++------------
> > >  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
> > >  2 files changed, 32 insertions(+), 16 deletions(-)
> > > 
> > > diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> > > index 05f32989bad6..55fba81cd2d1 100644
> > > --- a/fs/ocfs2/namei.c
> > > +++ b/fs/ocfs2/namei.c
> > > @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> > >  	int want_meta = 0;
> > >  	int xattr_credits = 0;
> > >  	struct ocfs2_security_xattr_info si = {
> > > +		.name = NULL,
> > >  		.enable = 1,
> > >  	};
> > >  	int did_quota_inode = 0;
> > > @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> > >  	/* get security xattr */
> > >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> > >  	if (status) {
> > > -		if (status == -EOPNOTSUPP)
> > > -			si.enable = 0;
> > > -		else {
> > > -			mlog_errno(status);
> > > -			goto leave;
> > > -		}
> > 
> > Although security_inode_init_security() does not return -EOPNOTSUPP, 
> > ocfs2_init_security_get() could.  Refer to commit 8154da3d2114 ("ocfs2:
> > Add incompatible flag for extended attribute").   It was added as a
> > temporary solution back in 2008, so it is highly unlikely that it is
> > still needed.
> > 
> > > +		mlog_errno(status);
> > > +		goto leave;
> > 
> > Without the -EOPNOTSUPP test, ocfs2_mknod() would not create the inode;
> > and similarly ocfs2_symlink(), below, would not create the symlink.  It
> > would be safer not to remove the -EOPNOTSUPP test.
> > 
> > >  	}
> > >  
> > >  	/* calculate meta data/clusters for setting security and acl xattr */
> > > @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> > >  	int want_clusters = 0;
> > >  	int xattr_credits = 0;
> > >  	struct ocfs2_security_xattr_info si = {
> > > +		.name = NULL,
> > >  		.enable = 1,
> > >  	};
> > >  	int did_quota = 0, did_quota_inode = 0;
> > > @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> > >  	/* get security xattr */
> > >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> > >  	if (status) {
> > > -		if (status == -EOPNOTSUPP)
> > > -			si.enable = 0;
> > > -		else {
> > > -			mlog_errno(status);
> > > -			goto bail;
> > > -		}
> > > +		mlog_errno(status);
> > > +		goto bail;
> > >  	}
> > >  
> > >  	/* calculate meta data/clusters for setting security xattr */
> > > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > > index 95d0611c5fc7..55699c573541 100644
> > > --- a/fs/ocfs2/xattr.c
> > > +++ b/fs/ocfs2/xattr.c
> > > @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
> > >  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
> > >  		     void *fs_info)
> > >  {
> > > +	struct ocfs2_security_xattr_info *si = fs_info;
> > >  	const struct xattr *xattr;
> > >  	int err = 0;
> > >  
> > > +	if (si) {
> > > +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> > > +				    GFP_KERNEL);
> > > +		if (!si->value)
> > > +			return -ENOMEM;
> > > +
> > > +		si->name = xattr_array->name;
> > > +		si->value_len = xattr_array->value_len;
> > > +		return 0;
> > > +	}
> > > +
> > >  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
> > >  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
> > >  				      xattr->name, xattr->value,
> > > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> > >  			    const struct qstr *qstr,
> > >  			    struct ocfs2_security_xattr_info *si)
> > >  {
> > > +	int ret;
> > > +
> > >  	/* check whether ocfs2 support feature xattr */
> > >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> > >  		return -EOPNOTSUPP;
> > > -	if (si)
> > > -		return security_old_inode_init_security(inode, dir, qstr,
> > > -							&si->name, &si->value,
> > > -							&si->value_len);
> > > +	if (si) {
> > > +		ret = security_inode_init_security(inode, dir, qstr,
> > > +						   &ocfs2_initxattrs, si);
> > 
> > The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> > security_old_inode_init_security() and security_inode_init_security(),
> > but return different values.  In the former case, it returns
> > -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> > whether or not we need to be concerned about private inodes on ocfs2.  
> > If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> > ocfs2_symlink() would fail to create the inode or symlink.
> 
> Correction, previously when returning -EOPNOTSUPP for private inodes,
> xattrs would not be wrriten.  By returning 0 without setting si->enable 
> to 0, xattrs will be written.

Ok, but if there is a private inode, we would be setting si->enable to
zero. Should be ok, I guess.

Thanks

Roberto

> > > +		/*
> > > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > > +		 * we have to check the xattr ourselves.
> > > +		 */
> > > +		if (!ret && !si->name)
> > > +			si->enable = 0;
> > > +
> > > +		return ret;
> > > +	}
> > >  
> > >  	return security_inode_init_security(inode, dir, qstr,
> > >  					    &ocfs2_initxattrs, NULL);
  
Mimi Zohar Feb. 20, 2023, 11:08 a.m. UTC | #9
> > > > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > > > index 95d0611c5fc7..55699c573541 100644

> > > > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> > > >  			    const struct qstr *qstr,
> > > >  			    struct ocfs2_security_xattr_info *si)
> > > >  {
> > > > +	int ret;
> > > > +
> > > >  	/* check whether ocfs2 support feature xattr */
> > > >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> > > >  		return -EOPNOTSUPP;
> > > > -	if (si)
> > > > -		return security_old_inode_init_security(inode, dir, qstr,
> > > > -							&si->name, &si->value,
> > > > -							&si->value_len);
> > > > +	if (si) {
> > > > +		ret = security_inode_init_security(inode, dir, qstr,
> > > > +						   &ocfs2_initxattrs, si);
> > > 
> > > The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> > > security_old_inode_init_security() and security_inode_init_security(),
> > > but return different values.  In the former case, it returns
> > > -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> > > whether or not we need to be concerned about private inodes on ocfs2.  
> > > If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> > > ocfs2_symlink() would fail to create the inode or symlink.
> > 
> > Correction, previously when returning -EOPNOTSUPP for private inodes,
> > xattrs would not be wrriten.  By returning 0 without setting si->enable 
> > to 0, xattrs will be written.
> 
> Ok, but if there is a private inode, we would be setting si->enable to
> zero. Should be ok, I guess.

si->enable is being set to zero, below, but is conditional on !si-
>name.

This is the last concern, otherwise the patch set looks good.

> > > > +		/*
> > > > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > > > +		 * we have to check the xattr ourselves.
> > > > +		 */
> > > > +		if (!ret && !si->name)
> > > > +			si->enable = 0;
> > > > +
> > > > +		return ret;
> > > > +	}
> > > >  
> > > >  	return security_inode_init_security(inode, dir, qstr,
> > > >  					    &ocfs2_initxattrs, NULL);
>
  
Roberto Sassu Feb. 20, 2023, 12:20 p.m. UTC | #10
On Mon, 2023-02-20 at 06:08 -0500, Mimi Zohar wrote:
> > > > > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > > > > index 95d0611c5fc7..55699c573541 100644
> > > > > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> > > > >  			    const struct qstr *qstr,
> > > > >  			    struct ocfs2_security_xattr_info *si)
> > > > >  {
> > > > > +	int ret;
> > > > > +
> > > > >  	/* check whether ocfs2 support feature xattr */
> > > > >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> > > > >  		return -EOPNOTSUPP;
> > > > > -	if (si)
> > > > > -		return security_old_inode_init_security(inode, dir, qstr,
> > > > > -							&si->name, &si->value,
> > > > > -							&si->value_len);
> > > > > +	if (si) {
> > > > > +		ret = security_inode_init_security(inode, dir, qstr,
> > > > > +						   &ocfs2_initxattrs, si);
> > > > 
> > > > The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> > > > security_old_inode_init_security() and security_inode_init_security(),
> > > > but return different values.  In the former case, it returns
> > > > -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> > > > whether or not we need to be concerned about private inodes on ocfs2.  
> > > > If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> > > > ocfs2_symlink() would fail to create the inode or symlink.
> > > 
> > > Correction, previously when returning -EOPNOTSUPP for private inodes,
> > > xattrs would not be wrriten.  By returning 0 without setting si->enable 
> > > to 0, xattrs will be written.
> > 
> > Ok, but if there is a private inode, we would be setting si->enable to
> > zero. Should be ok, I guess.
> 
> si->enable is being set to zero, below, but is conditional on !si-
> > name.
> 
> This is the last concern, otherwise the patch set looks good.

Uhm, if the inode is private, security_inode_init_security() will
immediately return. So, the condition !si->name should be always true.

Thanks

Roberto

> > > > > +		/*
> > > > > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > > > > +		 * we have to check the xattr ourselves.
> > > > > +		 */
> > > > > +		if (!ret && !si->name)
> > > > > +			si->enable = 0;
> > > > > +
> > > > > +		return ret;
> > > > > +	}
> > > > >  
> > > > >  	return security_inode_init_security(inode, dir, qstr,
> > > > >  					    &ocfs2_initxattrs, NULL);
  
Mimi Zohar Feb. 20, 2023, 12:40 p.m. UTC | #11
On Mon, 2023-02-20 at 13:20 +0100, Roberto Sassu wrote:
> On Mon, 2023-02-20 at 06:08 -0500, Mimi Zohar wrote:
> > > > > > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > > > > > index 95d0611c5fc7..55699c573541 100644
> > > > > > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> > > > > >  			    const struct qstr *qstr,
> > > > > >  			    struct ocfs2_security_xattr_info *si)
> > > > > >  {
> > > > > > +	int ret;
> > > > > > +
> > > > > >  	/* check whether ocfs2 support feature xattr */
> > > > > >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> > > > > >  		return -EOPNOTSUPP;
> > > > > > -	if (si)
> > > > > > -		return security_old_inode_init_security(inode, dir, qstr,
> > > > > > -							&si->name, &si->value,
> > > > > > -							&si->value_len);
> > > > > > +	if (si) {
> > > > > > +		ret = security_inode_init_security(inode, dir, qstr,
> > > > > > +						   &ocfs2_initxattrs, si);
> > > > > 
> > > > > The "if (unlikely(IS_PRIVATE(inode))"  test exists in both
> > > > > security_old_inode_init_security() and security_inode_init_security(),
> > > > > but return different values.  In the former case, it returns
> > > > > -EOPNOTSUPP.  In the latter case, it returns 0.  The question is
> > > > > whether or not we need to be concerned about private inodes on ocfs2.  
> > > > > If private inodes on ocfs2 are possible, then ocsf2_mknod() or
> > > > > ocfs2_symlink() would fail to create the inode or symlink.
> > > > 
> > > > Correction, previously when returning -EOPNOTSUPP for private inodes,
> > > > xattrs would not be wrriten.  By returning 0 without setting si->enable 
> > > > to 0, xattrs will be written.
> > > 
> > > Ok, but if there is a private inode, we would be setting si->enable to
> > > zero. Should be ok, I guess.
> > 
> > si->enable is being set to zero, below, but is conditional on !si-
> > > name.
> > 
> > This is the last concern, otherwise the patch set looks good.
> 
> Uhm, if the inode is private, security_inode_init_security() will
> immediately return. So, the condition !si->name should be always true.

Ok, thanks for checking.

> 
> > > > > > +		/*
> > > > > > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > > > > > +		 * we have to check the xattr ourselves.
> > > > > > +		 */
> > > > > > +		if (!ret && !si->name)
> > > > > > +			si->enable = 0;
> > > > > > +
> > > > > > +		return ret;
> > > > > > +	}
> > > > > >  
> > > > > >  	return security_inode_init_security(inode, dir, qstr,
> > > > > >  					    &ocfs2_initxattrs, NULL);
>
  
Joseph Qi Feb. 21, 2023, 6:45 a.m. UTC | #12
Hi,

Sorry for the late reply.

I don't have much background on this thread. It seems that we have to
check EOPNOTSUPP since ocfs2_init_security_get() may return EOPNOTSUPP
if it doesn't support extended attribute feature for backward
compatibility.

Other looks good. So with above comments addressed, you can add:
Acked-by: Joseph Qi <joseph.qi@linux.alibaba.com>

On 1/10/23 4:55 PM, Roberto Sassu wrote:
> On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
>> From: Roberto Sassu <roberto.sassu@huawei.com>
>>
>> In preparation for removing security_old_inode_init_security(), switch to
>> security_inode_init_security().
>>
>> Extend the existing ocfs2_initxattrs() to take the
>> ocfs2_security_xattr_info structure from fs_info, and populate the
>> name/value/len triple with the first xattr provided by LSMs.
> 
> Hi Mark, Joel, Joseph
> 
> some time ago I sent this patch set to switch to the newer
> function security_inode_init_security(). Almost all the other parts of
> this patch set have been reviewed, and the patch set itself should be
> ready to be merged.
> 
> I kindly ask if you could have a look at this patch and give your
> Reviewed-by, so that Paul could take the patch set.
> 
> Thanks a lot!
> 
> Roberto
> 
>> As fs_info was not used before, ocfs2_initxattrs() can now handle the case
>> of replicating the behavior of security_old_inode_init_security(), i.e.
>> just obtaining the xattr, in addition to setting all xattrs provided by
>> LSMs.
>>
>> Supporting multiple xattrs is not currently supported where
>> security_old_inode_init_security() was called (mknod, symlink), as it
>> requires non-trivial changes that can be done at a later time. Like for
>> reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
>> not the first to set it, its xattr will be discarded; if it is the first,
>> it does not have xattrs to calculate the HMAC on).
>>
>> Finally, modify the handling of the return value from
>> ocfs2_init_security_get(). As security_inode_init_security() does not
>> return -EOPNOTSUPP, remove this case and directly handle the error if the
>> return value is not zero.
>>
>> However, the previous case of receiving -EOPNOTSUPP should be still
>> taken into account, as security_inode_init_security() could return zero
>> without setting xattrs and ocfs2 would consider it as if the xattr was set.
>>
>> Instead, if security_inode_init_security() returned zero, look at the xattr
>> if it was set, and behave accordingly, i.e. set si->enable to zero to
>> notify to the functions following ocfs2_init_security_get() that the xattr
>> is not available (same as if security_old_inode_init_security() returned
>> -EOPNOTSUPP).
>>
>> Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
>> Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
>> ---
>>  fs/ocfs2/namei.c | 18 ++++++------------
>>  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
>>  2 files changed, 32 insertions(+), 16 deletions(-)
>>
>> diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
>> index 05f32989bad6..55fba81cd2d1 100644
>> --- a/fs/ocfs2/namei.c
>> +++ b/fs/ocfs2/namei.c
>> @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>>  	int want_meta = 0;
>>  	int xattr_credits = 0;
>>  	struct ocfs2_security_xattr_info si = {
>> +		.name = NULL,
>>  		.enable = 1,
>>  	};
>>  	int did_quota_inode = 0;
>> @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
>>  	/* get security xattr */
>>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>>  	if (status) {
>> -		if (status == -EOPNOTSUPP)
>> -			si.enable = 0;
>> -		else {
>> -			mlog_errno(status);
>> -			goto leave;
>> -		}
>> +		mlog_errno(status);
>> +		goto leave;
>>  	}
>>  
>>  	/* calculate meta data/clusters for setting security and acl xattr */
>> @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>>  	int want_clusters = 0;
>>  	int xattr_credits = 0;
>>  	struct ocfs2_security_xattr_info si = {
>> +		.name = NULL,
>>  		.enable = 1,
>>  	};
>>  	int did_quota = 0, did_quota_inode = 0;
>> @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
>>  	/* get security xattr */
>>  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
>>  	if (status) {
>> -		if (status == -EOPNOTSUPP)
>> -			si.enable = 0;
>> -		else {
>> -			mlog_errno(status);
>> -			goto bail;
>> -		}
>> +		mlog_errno(status);
>> +		goto bail;
>>  	}
>>  
>>  	/* calculate meta data/clusters for setting security xattr */
>> diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
>> index 95d0611c5fc7..55699c573541 100644
>> --- a/fs/ocfs2/xattr.c
>> +++ b/fs/ocfs2/xattr.c
>> @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
>>  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
>>  		     void *fs_info)
>>  {
>> +	struct ocfs2_security_xattr_info *si = fs_info;
>>  	const struct xattr *xattr;
>>  	int err = 0;
>>  
>> +	if (si) {
>> +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
>> +				    GFP_KERNEL);
>> +		if (!si->value)
>> +			return -ENOMEM;
>> +
>> +		si->name = xattr_array->name;
>> +		si->value_len = xattr_array->value_len;
>> +		return 0;
>> +	}
>> +
>>  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
>>  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
>>  				      xattr->name, xattr->value,
>> @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
>>  			    const struct qstr *qstr,
>>  			    struct ocfs2_security_xattr_info *si)
>>  {
>> +	int ret;
>> +
>>  	/* check whether ocfs2 support feature xattr */
>>  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
>>  		return -EOPNOTSUPP;
>> -	if (si)
>> -		return security_old_inode_init_security(inode, dir, qstr,
>> -							&si->name, &si->value,
>> -							&si->value_len);
>> +	if (si) {
>> +		ret = security_inode_init_security(inode, dir, qstr,
>> +						   &ocfs2_initxattrs, si);
>> +		/*
>> +		 * security_inode_init_security() does not return -EOPNOTSUPP,
>> +		 * we have to check the xattr ourselves.
>> +		 */
>> +		if (!ret && !si->name)
>> +			si->enable = 0;
>> +
>> +		return ret;
>> +	}
>>  
>>  	return security_inode_init_security(inode, dir, qstr,
>>  					    &ocfs2_initxattrs, NULL);
  
Roberto Sassu Feb. 21, 2023, 7:51 a.m. UTC | #13
On Tue, 2023-02-21 at 14:45 +0800, Joseph Qi wrote:
> Hi,
> 
> Sorry for the late reply.
> 
> I don't have much background on this thread. It seems that we have to
> check EOPNOTSUPP since ocfs2_init_security_get() may return EOPNOTSUPP
> if it doesn't support extended attribute feature for backward
> compatibility.

Hi Joseph

yes, I already reintroduced the check. Thanks for having a look.

Roberto

> Other looks good. So with above comments addressed, you can add:
> Acked-by: Joseph Qi <joseph.qi@linux.alibaba.com>
> 
> On 1/10/23 4:55 PM, Roberto Sassu wrote:
> > On Thu, 2022-12-01 at 11:41 +0100, Roberto Sassu wrote:
> > > From: Roberto Sassu <roberto.sassu@huawei.com>
> > > 
> > > In preparation for removing security_old_inode_init_security(), switch to
> > > security_inode_init_security().
> > > 
> > > Extend the existing ocfs2_initxattrs() to take the
> > > ocfs2_security_xattr_info structure from fs_info, and populate the
> > > name/value/len triple with the first xattr provided by LSMs.
> > 
> > Hi Mark, Joel, Joseph
> > 
> > some time ago I sent this patch set to switch to the newer
> > function security_inode_init_security(). Almost all the other parts of
> > this patch set have been reviewed, and the patch set itself should be
> > ready to be merged.
> > 
> > I kindly ask if you could have a look at this patch and give your
> > Reviewed-by, so that Paul could take the patch set.
> > 
> > Thanks a lot!
> > 
> > Roberto
> > 
> > > As fs_info was not used before, ocfs2_initxattrs() can now handle the case
> > > of replicating the behavior of security_old_inode_init_security(), i.e.
> > > just obtaining the xattr, in addition to setting all xattrs provided by
> > > LSMs.
> > > 
> > > Supporting multiple xattrs is not currently supported where
> > > security_old_inode_init_security() was called (mknod, symlink), as it
> > > requires non-trivial changes that can be done at a later time. Like for
> > > reiserfs, even if EVM is invoked, it will not provide an xattr (if it is
> > > not the first to set it, its xattr will be discarded; if it is the first,
> > > it does not have xattrs to calculate the HMAC on).
> > > 
> > > Finally, modify the handling of the return value from
> > > ocfs2_init_security_get(). As security_inode_init_security() does not
> > > return -EOPNOTSUPP, remove this case and directly handle the error if the
> > > return value is not zero.
> > > 
> > > However, the previous case of receiving -EOPNOTSUPP should be still
> > > taken into account, as security_inode_init_security() could return zero
> > > without setting xattrs and ocfs2 would consider it as if the xattr was set.
> > > 
> > > Instead, if security_inode_init_security() returned zero, look at the xattr
> > > if it was set, and behave accordingly, i.e. set si->enable to zero to
> > > notify to the functions following ocfs2_init_security_get() that the xattr
> > > is not available (same as if security_old_inode_init_security() returned
> > > -EOPNOTSUPP).
> > > 
> > > Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> > > Reviewed-by: Casey Schaufler <casey@schaufler-ca.com>
> > > ---
> > >  fs/ocfs2/namei.c | 18 ++++++------------
> > >  fs/ocfs2/xattr.c | 30 ++++++++++++++++++++++++++----
> > >  2 files changed, 32 insertions(+), 16 deletions(-)
> > > 
> > > diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
> > > index 05f32989bad6..55fba81cd2d1 100644
> > > --- a/fs/ocfs2/namei.c
> > > +++ b/fs/ocfs2/namei.c
> > > @@ -242,6 +242,7 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> > >  	int want_meta = 0;
> > >  	int xattr_credits = 0;
> > >  	struct ocfs2_security_xattr_info si = {
> > > +		.name = NULL,
> > >  		.enable = 1,
> > >  	};
> > >  	int did_quota_inode = 0;
> > > @@ -315,12 +316,8 @@ static int ocfs2_mknod(struct user_namespace *mnt_userns,
> > >  	/* get security xattr */
> > >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> > >  	if (status) {
> > > -		if (status == -EOPNOTSUPP)
> > > -			si.enable = 0;
> > > -		else {
> > > -			mlog_errno(status);
> > > -			goto leave;
> > > -		}
> > > +		mlog_errno(status);
> > > +		goto leave;
> > >  	}
> > >  
> > >  	/* calculate meta data/clusters for setting security and acl xattr */
> > > @@ -1805,6 +1802,7 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> > >  	int want_clusters = 0;
> > >  	int xattr_credits = 0;
> > >  	struct ocfs2_security_xattr_info si = {
> > > +		.name = NULL,
> > >  		.enable = 1,
> > >  	};
> > >  	int did_quota = 0, did_quota_inode = 0;
> > > @@ -1875,12 +1873,8 @@ static int ocfs2_symlink(struct user_namespace *mnt_userns,
> > >  	/* get security xattr */
> > >  	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
> > >  	if (status) {
> > > -		if (status == -EOPNOTSUPP)
> > > -			si.enable = 0;
> > > -		else {
> > > -			mlog_errno(status);
> > > -			goto bail;
> > > -		}
> > > +		mlog_errno(status);
> > > +		goto bail;
> > >  	}
> > >  
> > >  	/* calculate meta data/clusters for setting security xattr */
> > > diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> > > index 95d0611c5fc7..55699c573541 100644
> > > --- a/fs/ocfs2/xattr.c
> > > +++ b/fs/ocfs2/xattr.c
> > > @@ -7259,9 +7259,21 @@ static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
> > >  static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
> > >  		     void *fs_info)
> > >  {
> > > +	struct ocfs2_security_xattr_info *si = fs_info;
> > >  	const struct xattr *xattr;
> > >  	int err = 0;
> > >  
> > > +	if (si) {
> > > +		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
> > > +				    GFP_KERNEL);
> > > +		if (!si->value)
> > > +			return -ENOMEM;
> > > +
> > > +		si->name = xattr_array->name;
> > > +		si->value_len = xattr_array->value_len;
> > > +		return 0;
> > > +	}
> > > +
> > >  	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
> > >  		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
> > >  				      xattr->name, xattr->value,
> > > @@ -7277,13 +7289,23 @@ int ocfs2_init_security_get(struct inode *inode,
> > >  			    const struct qstr *qstr,
> > >  			    struct ocfs2_security_xattr_info *si)
> > >  {
> > > +	int ret;
> > > +
> > >  	/* check whether ocfs2 support feature xattr */
> > >  	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
> > >  		return -EOPNOTSUPP;
> > > -	if (si)
> > > -		return security_old_inode_init_security(inode, dir, qstr,
> > > -							&si->name, &si->value,
> > > -							&si->value_len);
> > > +	if (si) {
> > > +		ret = security_inode_init_security(inode, dir, qstr,
> > > +						   &ocfs2_initxattrs, si);
> > > +		/*
> > > +		 * security_inode_init_security() does not return -EOPNOTSUPP,
> > > +		 * we have to check the xattr ourselves.
> > > +		 */
> > > +		if (!ret && !si->name)
> > > +			si->enable = 0;
> > > +
> > > +		return ret;
> > > +	}
> > >  
> > >  	return security_inode_init_security(inode, dir, qstr,
> > >  					    &ocfs2_initxattrs, NULL);
  

Patch

diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index 05f32989bad6..55fba81cd2d1 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -242,6 +242,7 @@  static int ocfs2_mknod(struct user_namespace *mnt_userns,
 	int want_meta = 0;
 	int xattr_credits = 0;
 	struct ocfs2_security_xattr_info si = {
+		.name = NULL,
 		.enable = 1,
 	};
 	int did_quota_inode = 0;
@@ -315,12 +316,8 @@  static int ocfs2_mknod(struct user_namespace *mnt_userns,
 	/* get security xattr */
 	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
 	if (status) {
-		if (status == -EOPNOTSUPP)
-			si.enable = 0;
-		else {
-			mlog_errno(status);
-			goto leave;
-		}
+		mlog_errno(status);
+		goto leave;
 	}
 
 	/* calculate meta data/clusters for setting security and acl xattr */
@@ -1805,6 +1802,7 @@  static int ocfs2_symlink(struct user_namespace *mnt_userns,
 	int want_clusters = 0;
 	int xattr_credits = 0;
 	struct ocfs2_security_xattr_info si = {
+		.name = NULL,
 		.enable = 1,
 	};
 	int did_quota = 0, did_quota_inode = 0;
@@ -1875,12 +1873,8 @@  static int ocfs2_symlink(struct user_namespace *mnt_userns,
 	/* get security xattr */
 	status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si);
 	if (status) {
-		if (status == -EOPNOTSUPP)
-			si.enable = 0;
-		else {
-			mlog_errno(status);
-			goto bail;
-		}
+		mlog_errno(status);
+		goto bail;
 	}
 
 	/* calculate meta data/clusters for setting security xattr */
diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index 95d0611c5fc7..55699c573541 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -7259,9 +7259,21 @@  static int ocfs2_xattr_security_set(const struct xattr_handler *handler,
 static int ocfs2_initxattrs(struct inode *inode, const struct xattr *xattr_array,
 		     void *fs_info)
 {
+	struct ocfs2_security_xattr_info *si = fs_info;
 	const struct xattr *xattr;
 	int err = 0;
 
+	if (si) {
+		si->value = kmemdup(xattr_array->value, xattr_array->value_len,
+				    GFP_KERNEL);
+		if (!si->value)
+			return -ENOMEM;
+
+		si->name = xattr_array->name;
+		si->value_len = xattr_array->value_len;
+		return 0;
+	}
+
 	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
 		err = ocfs2_xattr_set(inode, OCFS2_XATTR_INDEX_SECURITY,
 				      xattr->name, xattr->value,
@@ -7277,13 +7289,23 @@  int ocfs2_init_security_get(struct inode *inode,
 			    const struct qstr *qstr,
 			    struct ocfs2_security_xattr_info *si)
 {
+	int ret;
+
 	/* check whether ocfs2 support feature xattr */
 	if (!ocfs2_supports_xattr(OCFS2_SB(dir->i_sb)))
 		return -EOPNOTSUPP;
-	if (si)
-		return security_old_inode_init_security(inode, dir, qstr,
-							&si->name, &si->value,
-							&si->value_len);
+	if (si) {
+		ret = security_inode_init_security(inode, dir, qstr,
+						   &ocfs2_initxattrs, si);
+		/*
+		 * security_inode_init_security() does not return -EOPNOTSUPP,
+		 * we have to check the xattr ourselves.
+		 */
+		if (!ret && !si->name)
+			si->enable = 0;
+
+		return ret;
+	}
 
 	return security_inode_init_security(inode, dir, qstr,
 					    &ocfs2_initxattrs, NULL);