ceph: add support to readdir for encrypted names

To make it simpler to decrypt names in a readdir reply (i.e. before
we have a dentry), add a new ceph_encode_encrypted_fname()-like helper
that takes a qstr pointer instead of a dentry pointer.

Once we've decrypted the names in a readdir reply, we no longer need the
crypttext, so overwrite them in ceph_mds_reply_dir_entry with the
unencrypted names. Then in both ceph_readdir_prepopulate() and
ceph_readdir() we will use the dencrypted name directly.

[ jlayton: convert some BUG_ONs into error returns ]

Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Reviewed-and-tested-by: Luís Henriques <lhenriques@suse.de>
Reviewed-by: Milind Changire <mchangir@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

fs/ceph/crypto.c

@@ -192,15 +192,18 @@ void ceph_fscrypt_as_ctx_to_req(struct ceph_mds_request *req,
 	swap(req->r_fscrypt_auth, as->fscrypt_auth);
 }
 
-int ceph_encode_encrypted_fname(const struct inode *parent,
+int ceph_encode_encrypted_dname(const struct inode *parent,
-				struct dentry *dentry, char *buf)
+				struct qstr *d_name, char *buf)
 {
 	u32 len;
 	int elen;
 	int ret;
 	u8 *cryptbuf;
 
-	WARN_ON_ONCE(!fscrypt_has_encryption_key(parent));
+	if (!fscrypt_has_encryption_key(parent)) {
+		memcpy(buf, d_name->name, d_name->len);
+		return d_name->len;
+	}
 
 	/*
 	 * Convert cleartext d_name to ciphertext. If result is longer than
@@ -208,8 +211,7 @@ int ceph_encode_encrypted_fname(const struct inode *parent,
 	 *
 	 * See: fscrypt_setup_filename
 	 */
-	if (!fscrypt_fname_encrypted_size(parent, dentry->d_name.len, NAME_MAX,
+	if (!fscrypt_fname_encrypted_size(parent, d_name->len, NAME_MAX, &len))
-					  &len))
 		return -ENAMETOOLONG;
 
 	/* Allocate a buffer appropriate to hold the result */
@@ -218,7 +220,7 @@ int ceph_encode_encrypted_fname(const struct inode *parent,
 	if (!cryptbuf)
 		return -ENOMEM;
 
-	ret = fscrypt_fname_encrypt(parent, &dentry->d_name, cryptbuf, len);
+	ret = fscrypt_fname_encrypt(parent, d_name, cryptbuf, len);
 	if (ret) {
 		kfree(cryptbuf);
 		return ret;
@@ -245,6 +247,14 @@ int ceph_encode_encrypted_fname(const struct inode *parent,
 	return elen;
 }
 
+int ceph_encode_encrypted_fname(const struct inode *parent,
+				struct dentry *dentry, char *buf)
+{
+	WARN_ON_ONCE(!fscrypt_has_encryption_key(parent));
+
+	return ceph_encode_encrypted_dname(parent, &dentry->d_name, buf);
+}
+
 /**
  * ceph_fname_to_usr - convert a filename for userland presentation
  * @fname: ceph_fname to be converted
@@ -286,7 +296,10 @@ int ceph_fname_to_usr(const struct ceph_fname *fname, struct fscrypt_str *tname,
 	 * generating a nokey name via fscrypt.
 	 */
 	if (!fscrypt_has_encryption_key(fname->dir)) {
-		memcpy(oname->name, fname->name, fname->name_len);
+		if (fname->no_copy)
+			oname->name = fname->name;
+		else
+			memcpy(oname->name, fname->name, fname->name_len);
 		oname->len = fname->name_len;
 		if (is_nokey)
 			*is_nokey = true;

fs/ceph/crypto.h

@@ -19,6 +19,7 @@ struct ceph_fname {
 	unsigned char	*ctext;		// binary crypttext (if any)
 	u32		name_len;	// length of name buffer
 	u32		ctext_len;	// length of crypttext
+	bool		no_copy;
 };
 
 struct ceph_fscrypt_auth {
@@ -76,6 +77,8 @@ int ceph_fscrypt_prepare_context(struct inode *dir, struct inode *inode,
 				 struct ceph_acl_sec_ctx *as);
 void ceph_fscrypt_as_ctx_to_req(struct ceph_mds_request *req,
 				struct ceph_acl_sec_ctx *as);
+int ceph_encode_encrypted_dname(const struct inode *parent,
+				struct qstr *d_name, char *buf);
 int ceph_encode_encrypted_fname(const struct inode *parent,
 				struct dentry *dentry, char *buf);
 
@@ -121,6 +124,13 @@ static inline void ceph_fscrypt_as_ctx_to_req(struct ceph_mds_request *req,
 {
 }
 
+static inline int ceph_encode_encrypted_dname(const struct inode *parent,
+					      struct qstr *d_name, char *buf)
+{
+	memcpy(buf, d_name->name, d_name->len);
+	return d_name->len;
+}
+
 static inline int ceph_encode_encrypted_fname(const struct inode *parent,
 					      struct dentry *dentry, char *buf)
 {

fs/ceph/dir.c

@@ -9,6 +9,7 @@
 
 #include "super.h"
 #include "mds_client.h"
+#include "crypto.h"
 
 /*
  * Directory operations: readdir, lookup, create, link, unlink,
@@ -241,7 +242,9 @@ static int __dcache_readdir(struct file *file,  struct dir_context *ctx,
 		di = ceph_dentry(dentry);
 		if (d_unhashed(dentry) ||
 		    d_really_is_negative(dentry) ||
-		    di->lease_shared_gen != shared_gen) {
+		    di->lease_shared_gen != shared_gen ||
+		    ((dentry->d_flags & DCACHE_NOKEY_NAME) &&
+		     fscrypt_has_encryption_key(dir))) {
 			spin_unlock(&dentry->d_lock);
 			dput(dentry);
 			err = -EAGAIN;
@@ -340,6 +343,10 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 		ctx->pos = 2;
 	}
 
+	err = fscrypt_prepare_readdir(inode);
+	if (err)
+		return err;
+
 	spin_lock(&ci->i_ceph_lock);
 	/* request Fx cap. if have Fx, we don't need to release Fs cap
 	 * for later create/unlink. */
@@ -389,6 +396,7 @@ more:
 		req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
 		if (IS_ERR(req))
 			return PTR_ERR(req);
+
 		err = ceph_alloc_readdir_reply_buffer(req, inode);
 		if (err) {
 			ceph_mdsc_put_request(req);
@@ -402,11 +410,21 @@ more:
 			req->r_inode_drop = CEPH_CAP_FILE_EXCL;
 		}
 		if (dfi->last_name) {
-			req->r_path2 = kstrdup(dfi->last_name, GFP_KERNEL);
+			struct qstr d_name = { .name = dfi->last_name,
+					       .len = strlen(dfi->last_name) };
+
+			req->r_path2 = kzalloc(NAME_MAX + 1, GFP_KERNEL);
 			if (!req->r_path2) {
 				ceph_mdsc_put_request(req);
 				return -ENOMEM;
 			}
+
+			err = ceph_encode_encrypted_dname(inode, &d_name,
+							  req->r_path2);
+			if (err < 0) {
+				ceph_mdsc_put_request(req);
+				return err;
+			}
 		} else if (is_hash_order(ctx->pos)) {
 			req->r_args.readdir.offset_hash =
 				cpu_to_le32(fpos_hash(ctx->pos));
@@ -511,15 +529,20 @@ more:
 	for (; i < rinfo->dir_nr; i++) {
 		struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
 
-		BUG_ON(rde->offset < ctx->pos);
+		if (rde->offset < ctx->pos) {
+			pr_warn("%s: rde->offset 0x%llx ctx->pos 0x%llx\n",
+				__func__, rde->offset, ctx->pos);
+			return -EIO;
+		}
+
+		if (WARN_ON_ONCE(!rde->inode.in))
+			return -EIO;
 
 		ctx->pos = rde->offset;
 		dout("readdir (%d/%d) -> %llx '%.*s' %p\n",
 		     i, rinfo->dir_nr, ctx->pos,
 		     rde->name_len, rde->name, &rde->inode.in);
 
-		BUG_ON(!rde->inode.in);
-
 		if (!dir_emit(ctx, rde->name, rde->name_len,
 			      ceph_present_ino(inode->i_sb, le64_to_cpu(rde->inode.in->ino)),
 			      le32_to_cpu(rde->inode.in->mode) >> 12)) {
@@ -532,6 +555,8 @@ more:
 			dout("filldir stopping us...\n");
 			return 0;
 		}
+
+		/* Reset the lengths to their original allocated vals */
 		ctx->pos++;
 	}
 
@@ -586,7 +611,6 @@ more:
 					dfi->dir_ordered_count);
 		spin_unlock(&ci->i_ceph_lock);
 	}
-
 	dout("readdir %p file %p done.\n", inode, file);
 	return 0;
 }

fs/ceph/inode.c

@@ -1752,7 +1752,8 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 			     struct ceph_mds_session *session)
 {
 	struct dentry *parent = req->r_dentry;
-	struct ceph_inode_info *ci = ceph_inode(d_inode(parent));
+	struct inode *inode = d_inode(parent);
+	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
 	struct qstr dname;
 	struct dentry *dn;
@@ -1826,9 +1827,7 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 		tvino.snap = le64_to_cpu(rde->inode.in->snapid);
 
 		if (rinfo->hash_order) {
-			u32 hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
+			u32 hash = ceph_frag_value(rde->raw_hash);
-						 rde->name, rde->name_len);
-			hash = ceph_frag_value(hash);
 			if (hash != last_hash)
 				fpos_offset = 2;
 			last_hash = hash;
@@ -1851,6 +1850,11 @@ retry_lookup:
 				err = -ENOMEM;
 				goto out;
 			}
+			if (rde->is_nokey) {
+				spin_lock(&dn->d_lock);
+				dn->d_flags |= DCACHE_NOKEY_NAME;
+				spin_unlock(&dn->d_lock);
+			}
 		} else if (d_really_is_positive(dn) &&
 			   (ceph_ino(d_inode(dn)) != tvino.ino ||
 			    ceph_snap(d_inode(dn)) != tvino.snap)) {

fs/ceph/mds_client.c

@@ -440,20 +440,87 @@ static int parse_reply_info_readdir(void **p, void *end,
 
 	info->dir_nr = num;
 	while (num) {
+		struct inode *inode = d_inode(req->r_dentry);
+		struct ceph_inode_info *ci = ceph_inode(inode);
 		struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
+		struct fscrypt_str tname = FSTR_INIT(NULL, 0);
+		struct fscrypt_str oname = FSTR_INIT(NULL, 0);
+		struct ceph_fname fname;
+		u32 altname_len, _name_len;
+		u8 *altname, *_name;
+
 		/* dentry */
-		ceph_decode_32_safe(p, end, rde->name_len, bad);
+		ceph_decode_32_safe(p, end, _name_len, bad);
-		ceph_decode_need(p, end, rde->name_len, bad);
+		ceph_decode_need(p, end, _name_len, bad);
-		rde->name = *p;
+		_name = *p;
-		*p += rde->name_len;
+		*p += _name_len;
-		dout("parsed dir dname '%.*s'\n", rde->name_len, rde->name);
+		dout("parsed dir dname '%.*s'\n", _name_len, _name);
+
+		if (info->hash_order)
+			rde->raw_hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
+						      _name, _name_len);
 
 		/* dentry lease */
 		err = parse_reply_info_lease(p, end, &rde->lease, features,
-					     &rde->altname_len, &rde->altname);
+					     &altname_len, &altname);
 		if (err)
 			goto out_bad;
 
+		/*
+		 * Try to dencrypt the dentry names and update them
+		 * in the ceph_mds_reply_dir_entry struct.
+		 */
+		fname.dir = inode;
+		fname.name = _name;
+		fname.name_len = _name_len;
+		fname.ctext = altname;
+		fname.ctext_len = altname_len;
+		/*
+		 * The _name_len maybe larger than altname_len, such as
+		 * when the human readable name length is in range of
+		 * (CEPH_NOHASH_NAME_MAX, CEPH_NOHASH_NAME_MAX + SHA256_DIGEST_SIZE),
+		 * then the copy in ceph_fname_to_usr will corrupt the
+		 * data if there has no encryption key.
+		 *
+		 * Just set the no_copy flag and then if there has no
+		 * encryption key the oname.name will be assigned to
+		 * _name always.
+		 */
+		fname.no_copy = true;
+		if (altname_len == 0) {
+			/*
+			 * Set tname to _name, and this will be used
+			 * to do the base64_decode in-place. It's
+			 * safe because the decoded string should
+			 * always be shorter, which is 3/4 of origin
+			 * string.
+			 */
+			tname.name = _name;
+
+			/*
+			 * Set oname to _name too, and this will be
+			 * used to do the dencryption in-place.
+			 */
+			oname.name = _name;
+			oname.len = _name_len;
+		} else {
+			/*
+			 * This will do the decryption only in-place
+			 * from altname cryptext directly.
+			 */
+			oname.name = altname;
+			oname.len = altname_len;
+		}
+		rde->is_nokey = false;
+		err = ceph_fname_to_usr(&fname, &tname, &oname, &rde->is_nokey);
+		if (err) {
+			pr_err("%s unable to decode %.*s, got %d\n", __func__,
+			       _name_len, _name, err);
+			goto out_bad;
+		}
+		rde->name = oname.name;
+		rde->name_len = oname.len;
+
 		/* inode */
 		err = parse_reply_info_in(p, end, &rde->inode, features);
 		if (err < 0)
@@ -3671,7 +3738,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	if (err == 0) {
 		if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
 				    req->r_op == CEPH_MDS_OP_LSSNAP))
-			ceph_readdir_prepopulate(req, req->r_session);
+			err = ceph_readdir_prepopulate(req, req->r_session);
 	}
 	current->journal_info = NULL;
 	mutex_unlock(&req->r_fill_mutex);

fs/ceph/mds_client.h

@@ -96,10 +96,10 @@ struct ceph_mds_reply_info_in {
 };
 
 struct ceph_mds_reply_dir_entry {
+	bool			      is_nokey;
 	char                          *name;
-	u8			      *altname;
 	u32                           name_len;
-	u32			      altname_len;
+	u32			      raw_hash;
 	struct ceph_mds_reply_lease   *lease;
 	struct ceph_mds_reply_info_in inode;
 	loff_t			      offset;