linux-kernel - Re: [RFC PATCH] ceph: add support for encrypted snapshot names

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sat, 26 Feb 2022 15:06:18 +0000
From:   Luís Henriques <lhenriques@...e.de>
To:     Jeff Layton <jlayton@...nel.org>
Cc:     Xiubo Li <xiubli@...hat.com>, Ilya Dryomov <idryomov@...il.com>,
        ceph-devel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] ceph: add support for encrypted snapshot names

Jeff Layton <jlayton@...nel.org> writes:

> On Thu, 2022-02-24 at 11:21 +0000, Luís Henriques wrote:
>> Since filenames in encrypted directories are already encrypted and shown
>> as a base64-encoded string when the directory is locked, snapshot names
>> should show a similar behaviour.
>> 
>> Signed-off-by: Luís Henriques <lhenriques@...e.de>
>> ---
>>  fs/ceph/dir.c   | 15 +++++++++++++++
>>  fs/ceph/inode.c | 10 +++++++++-
>>  2 files changed, 24 insertions(+), 1 deletion(-)
>> 
>> Support on the MDS for names that'll be > MAX_NAME when base64 encoded is
>> still TBD.  I thought it would be something easy to do, but snapshots
>> don't seem to make use of the CDir/CDentry (which is where alternate_name
>> is stored on the MDS).  I'm still looking into this, but I may need some
>> help there :-(
>> 
>> Cheers,
>> --
>> Luís
>> 
>> diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
>> index a449f4a07c07..20ae600ee7cd 100644
>> --- a/fs/ceph/dir.c
>> +++ b/fs/ceph/dir.c
>> @@ -1065,6 +1065,13 @@ static int ceph_mkdir(struct user_namespace *mnt_userns, struct inode *dir,
>>  		op = CEPH_MDS_OP_MKSNAP;
>>  		dout("mksnap dir %p snap '%pd' dn %p\n", dir,
>>  		     dentry, dentry);
>> +		/* XXX missing support for alternate_name in snapshots */
>> +		if (IS_ENCRYPTED(dir) && (dentry->d_name.len >= 189)) {
>> +			dout("encrypted snapshot name too long: %pd len: %d\n",
>> +			     dentry, dentry->d_name.len);
>> +			err = -ENAMETOOLONG;
>> +			goto out;
>> +		}
>
> Where does 189 come from? You probably want to use CEPH_NOHASH_NAME_MAX.
>

Yeah, this is just a temporary workaround while the support for altnames
isn't implemented in snapshots.  (189 is the max size that will result in
a base64-encoded that is < MAX_NAME; 190 will be result in a filename that
is too long).

>>  	} else if (ceph_snap(dir) == CEPH_NOSNAP) {
>>  		dout("mkdir dir %p dn %p mode 0%ho\n", dir, dentry, mode);
>>  		op = CEPH_MDS_OP_MKDIR;
>> @@ -1109,6 +1116,14 @@ static int ceph_mkdir(struct user_namespace *mnt_userns, struct inode *dir,
>>  	    !req->r_reply_info.head->is_target &&
>>  	    !req->r_reply_info.head->is_dentry)
>>  		err = ceph_handle_notrace_create(dir, dentry);
>> +
>> +	/*
>> +	 * If we have created a snapshot we need to clear the cache, otherwise
>> +	 * snapshot will show encrypted filenames in readdir.
>> +	 */
>> +	if (ceph_snap(dir) == CEPH_SNAPDIR)
>> +		d_drop(dentry);
>> +
>
> This looks hacky, but I just caught up on the discussion between you and
> Xiubo, so I assume you're addressing that.

Right, I still need to investigate this further.  It may actually be a bug
somewhere else.  Right now I was trying to get the MDS code written and
decided to look at this later.  I just thought I could send out this RFC
anyway in case someone had an idea -- and Xiubo already gave some
suggestions (which I still have to look at...).

>
>>  out_req:
>>  	ceph_mdsc_put_request(req);
>>  out:
>> diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
>> index 8b0832271fdf..080824610b73 100644
>> --- a/fs/ceph/inode.c
>> +++ b/fs/ceph/inode.c
>> @@ -182,6 +182,13 @@ struct inode *ceph_get_snapdir(struct inode *parent)
>>  	ci->i_rbytes = 0;
>>  	ci->i_btime = ceph_inode(parent)->i_btime;
>>  
>> +	/* if encrypted, just borough fscrypt_auth from parent */
>> +	if (IS_ENCRYPTED(parent)) {
>> +		struct ceph_inode_info *pci = ceph_inode(parent);
>> +		inode->i_flags |= S_ENCRYPTED;
>> +		ci->fscrypt_auth_len = pci->fscrypt_auth_len;
>> +		ci->fscrypt_auth = pci->fscrypt_auth;
>> +	}
>>  	if (inode->i_state & I_NEW) {
>>  		inode->i_op = &ceph_snapdir_iops;
>>  		inode->i_fop = &ceph_snapdir_fops;
>> @@ -632,7 +639,8 @@ void ceph_free_inode(struct inode *inode)
>>  
>>  	kfree(ci->i_symlink);
>>  #ifdef CONFIG_FS_ENCRYPTION
>> -	kfree(ci->fscrypt_auth);
>> +	if (ceph_snap(inode) != CEPH_SNAPDIR)
>> +		kfree(ci->fscrypt_auth);
>
> Can a snapdir inode outlive its parent?

Good question.  That actually occurred to me and I assumed it can not.
But maybe a better/safer option is to create a new copy of fscrypt_auth
into the snapdir and kfree it here.

Cheers,
-- 
Luís

>
>>  #endif
>>  	fscrypt_free_inode(inode);
>>  	kmem_cache_free(ceph_inode_cachep, ci);
>
> -- 
> Jeff Layton <jlayton@...nel.org>