[Ocfs2-devel] [PATCH] ocfs2/security: Check xattr support in security_init.

[Ocfs2-devel] [PATCH] ocfs2/security: Check xattr support in security_init.

[Tao Ma]

If the system supports selinux, we will return sucessfully from
ocfs2_init_security_get if it is called for the mount point.
And in that case if the volume doesn't have xattr support, we
will not be able to create a new inode in the mount dir because
ocfs2_mknod will try to set security attributes for a new created
inode. This patch check xattr support in ocfs2_init_security_get,
so it will let ocfs2_mknod knows that we don't support xattr and
it don't need to init security for the new inode in that case.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
---
 fs/ocfs2/xattr.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index 5aec833..8a7db21 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -5289,6 +5289,9 @@ int ocfs2_init_security_get(struct inode *inode,
 			    struct inode *dir,
 			    struct ocfs2_security_xattr_info *si)
 {
+	if (!ocfs2_supports_xattr(OCFS2_SB(inode->i_sb)))
+		return -EOPNOTSUPP;
+
 	return security_inode_init_security(inode, dir, &si->name, &si->value,
 					    &si->value_len);
 }
-- 
1.5.4.4

[Ocfs2-devel] [PATCH] ocfs2/security: Check xattr support in security_init.

[Tiger Yang]

Hi, Tao

I already fixed this issue :)
http://oss.oracle.com/pipermail/ocfs2-devel/2008-December/003559.html

Thanks,
tiger

Tao Ma wrote:
> If the system supports selinux, we will return sucessfully from
> ocfs2_init_security_get if it is called for the mount point.
> And in that case if the volume doesn't have xattr support, we
> will not be able to create a new inode in the mount dir because
> ocfs2_mknod will try to set security attributes for a new created
> inode. This patch check xattr support in ocfs2_init_security_get,
> so it will let ocfs2_mknod knows that we don't support xattr and
> it don't need to init security for the new inode in that case.
> 
> Signed-off-by: Tao Ma <tao.ma@oracle.com>
> ---
>  fs/ocfs2/xattr.c |    3 +++
>  1 files changed, 3 insertions(+), 0 deletions(-)
> 
> diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
> index 5aec833..8a7db21 100644
> --- a/fs/ocfs2/xattr.c
> +++ b/fs/ocfs2/xattr.c
> @@ -5289,6 +5289,9 @@ int ocfs2_init_security_get(struct inode *inode,
>  			    struct inode *dir,
>  			    struct ocfs2_security_xattr_info *si)
>  {
> +	if (!ocfs2_supports_xattr(OCFS2_SB(inode->i_sb)))
> +		return -EOPNOTSUPP;
> +
>  	return security_inode_init_security(inode, dir, &si->name, &si->value,
>  					    &si->value_len);
>  }

[Ocfs2-devel] [PATCH] ocfs2/security: Check xattr support in security_init.

[Tao Ma]

Tiger Yang wrote:
> Hi, Tao
> 
> I already fixed this issue :)
> http://oss.oracle.com/pipermail/ocfs2-devel/2008-December/003559.html
aha, yes, It looks that my repository is a little older. Let me rebase it.

Sorry for the noise.

Regards,
Tao
> 
> Thanks,
> tiger
> 
> Tao Ma wrote:
>> If the system supports selinux, we will return sucessfully from
>> ocfs2_init_security_get if it is called for the mount point.
>> And in that case if the volume doesn't have xattr support, we
>> will not be able to create a new inode in the mount dir because
>> ocfs2_mknod will try to set security attributes for a new created
>> inode. This patch check xattr support in ocfs2_init_security_get,
>> so it will let ocfs2_mknod knows that we don't support xattr and
>> it don't need to init security for the new inode in that case.
>>
>> Signed-off-by: Tao Ma <tao.ma@oracle.com>
>> ---
>>  fs/ocfs2/xattr.c |    3 +++
>>  1 files changed, 3 insertions(+), 0 deletions(-)
>>
>> diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
>> index 5aec833..8a7db21 100644
>> --- a/fs/ocfs2/xattr.c
>> +++ b/fs/ocfs2/xattr.c
>> @@ -5289,6 +5289,9 @@ int ocfs2_init_security_get(struct inode *inode,
>>                  struct inode *dir,
>>                  struct ocfs2_security_xattr_info *si)
>>  {
>> +    if (!ocfs2_supports_xattr(OCFS2_SB(inode->i_sb)))
>> +        return -EOPNOTSUPP;
>> +
>>      return security_inode_init_security(inode, dir, &si->name, 
>> &si->value,
>>                          &si->value_len);
>>  }