[PATCH 24/32] vfs: syscall: Add fsopen() to prepare for superblock creation [ver #9]

[David Howells]

Provide an fsopen() system call that starts the process of preparing to
create a superblock that will then be mountable, using an fd as a context
handle.  fsopen() is given the name of the filesystem that will be used:

	int mfd = fsopen(const char *fsname, unsigned int flags);

where flags can be 0 or FSOPEN_CLOEXEC.

For example:

	sfd = fsopen("ext4", FSOPEN_CLOEXEC);
	write(sfd, "s /dev/sdb1"); // note I'm ignoring write's length arg
	write(sfd, "o noatime");
	write(sfd, "o acl");
	write(sfd, "o user_attr");
	write(sfd, "o iversion");
	write(sfd, "o ");
	write(sfd, "r /my/container"); // root inside the fs
	write(sfd, "x create"); // create the superblock
	fsinfo(sfd, NULL, ...); // query new superblock attributes
	mfd = fsmount(sfd, FSMOUNT_CLOEXEC, MS_RELATIME);
	move_mount(mfd, "", sfd, AT_FDCWD, "/mnt", MOVE_MOUNT_F_EMPTY_PATH);

	sfd = fsopen("afs", -1);
	write(sfd, "s %grand.central.org:root.cell");
	write(sfd, "o cell=grand.central.org");
	write(sfd, "r /");
	write(sfd, "x create");
	mfd = fsmount(sfd, 0, MS_NODEV);
	move_mount(mfd, "", sfd, AT_FDCWD, "/mnt", MOVE_MOUNT_F_EMPTY_PATH);

If an error is reported at any step, an error message may be available to be
read() back (ENODATA will be reported if there isn't an error available) in
the form:

	"e <subsys>:<problem>"
	"e SELinux:Mount on mountpoint not permitted"

Once fsmount() has been called, further write() calls will incur EBUSY,
even if the fsmount() fails.  read() is still possible to retrieve error
information.

The fsopen() syscall creates a mount context and hangs it of the fd that it
returns.

Netlink is not used because it is optional and would make the core VFS
dependent on the networking layer and also potentially add network
namespace issues.

Note that, for the moment, the caller must have SYS_CAP_ADMIN to use
fsopen().

Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
cc: linux-api@xxxxxxxxxxxxxxx
---

 arch/x86/entry/syscalls/syscall_32.tbl |    1 
 arch/x86/entry/syscalls/syscall_64.tbl |    1 
 fs/Makefile                            |    2 
 fs/fs_context.c                        |    4 +
 fs/fsopen.c                            |  209 ++++++++++++++++++++++++++++++++
 include/linux/fs_context.h             |    2 
 include/linux/syscalls.h               |    1 
 include/uapi/linux/fs.h                |    5 +
 8 files changed, 224 insertions(+), 1 deletion(-)
 create mode 100644 fs/fsopen.c

diff --git a/arch/x86/entry/syscalls/syscall_32.tbl b/arch/x86/entry/syscalls/syscall_32.tbl
index 76d092b7d1b0..1647fefd2969 100644
--- a/arch/x86/entry/syscalls/syscall_32.tbl
+++ b/arch/x86/entry/syscalls/syscall_32.tbl
@@ -400,3 +400,4 @@
 386	i386	rseq			sys_rseq			__ia32_sys_rseq
 387	i386	open_tree		sys_open_tree			__ia32_sys_open_tree
 388	i386	move_mount		sys_move_mount			__ia32_sys_move_mount
+389	i386	fsopen			sys_fsopen			__ia32_sys_fsopen
diff --git a/arch/x86/entry/syscalls/syscall_64.tbl b/arch/x86/entry/syscalls/syscall_64.tbl
index 37ba4e65eee6..235d33dbccb2 100644
--- a/arch/x86/entry/syscalls/syscall_64.tbl
+++ b/arch/x86/entry/syscalls/syscall_64.tbl
@@ -345,6 +345,7 @@
 334	common	rseq			__x64_sys_rseq
 335	common	open_tree		__x64_sys_open_tree
 336	common	move_mount		__x64_sys_move_mount
+337	common	fsopen			__x64_sys_fsopen
 
 #
 # x32-specific system call numbers start at 512 to avoid cache impact
diff --git a/fs/Makefile b/fs/Makefile
index 7e9ca59ac3a7..d3b33798998e 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -13,7 +13,7 @@ obj-y :=	open.o read_write.o file_table.o super.o \
 		seq_file.o xattr.o libfs.o fs-writeback.o \
 		pnode.o splice.o sync.o utimes.o d_path.o \
 		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o \
-		fs_context.o
+		fs_context.o fsopen.o
 
 ifeq ($(CONFIG_BLOCK),y)
 obj-y +=	buffer.o block_dev.o direct-io.o mpage.o
diff --git a/fs/fs_context.c b/fs/fs_context.c
index b7c84e0aa2f9..a2d745e6d356 100644
--- a/fs/fs_context.c
+++ b/fs/fs_context.c
@@ -251,6 +251,8 @@ struct fs_context *vfs_new_fs_context(struct file_system_type *fs_type,
 	fc->fs_type	= get_filesystem(fs_type);
 	fc->cred	= get_current_cred();
 
+	mutex_init(&fc->uapi_mutex);
+
 	switch (purpose) {
 	case FS_CONTEXT_FOR_KERNEL_MOUNT:
 		fc->sb_flags |= SB_KERNMOUNT;
@@ -335,6 +337,8 @@ struct fs_context *vfs_dup_fs_context(struct fs_context *src_fc)
 	if (!fc)
 		return ERR_PTR(-ENOMEM);
 
+	mutex_init(&fc->uapi_mutex);
+
 	fc->fs_private	= NULL;
 	fc->s_fs_info	= NULL;
 	fc->source	= NULL;
diff --git a/fs/fsopen.c b/fs/fsopen.c
new file mode 100644
index 000000000000..28bb72bda163
--- /dev/null
+++ b/fs/fsopen.c
@@ -0,0 +1,209 @@
+/* Filesystem access-by-fd.
+ *
+ * Copyright (C) 2017 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@xxxxxxxxxx)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public Licence
+ * as published by the Free Software Foundation; either version
+ * 2 of the Licence, or (at your option) any later version.
+ */
+
+#include <linux/fs_context.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <linux/syscalls.h>
+#include <linux/security.h>
+#include <linux/anon_inodes.h>
+#include "mount.h"
+
+/*
+ * Userspace writes configuration data and commands to the fd and we parse it
+ * here.  For the moment, we assume a single option or command per write.  Each
+ * line written is of the form
+ *
+ *	<command_type><space><stuff...>
+ *
+ *	s /dev/sda1				-- Source device
+ *	o noatime				-- Option without value
+ *	o cell=grand.central.org		-- Option with value
+ *	x create				-- Create a superblock
+ *	x reconfigure				-- Reconfigure a superblock
+ */
+static ssize_t fscontext_write(struct file *file,
+			       const char __user *_buf, size_t len, loff_t *pos)
+{
+	struct fs_context *fc = file->private_data;
+	char opt[2], *data;
+	ssize_t ret;
+
+	if (len < 3 || len > 4095)
+		return -EINVAL;
+
+	if (copy_from_user(opt, _buf, 2) != 0)
+		return -EFAULT;
+	switch (opt[0]) {
+	case 's':
+	case 'o':
+	case 'x':
+		break;
+	default:
+		return -EINVAL;
+	}
+	if (opt[1] != ' ')
+		return -EINVAL;
+
+	data = memdup_user_nul(_buf + 2, len - 2);
+	if (IS_ERR(data))
+		return PTR_ERR(data);
+
+	/* From this point onwards we need to lock the fd against someone
+	 * trying to mount it.
+	 */
+	ret = mutex_lock_interruptible(&fc->uapi_mutex);
+	if (ret < 0)
+		goto err_free;
+
+	if (fc->phase == FS_CONTEXT_AWAITING_RECONF) {
+		if (fc->fs_type->init_fs_context) {
+			ret = fc->fs_type->init_fs_context(fc, fc->root);
+			if (ret < 0) {
+				fc->phase = FS_CONTEXT_FAILED;
+				goto err_unlock;
+			}
+		} else {
+			/* Leave legacy context ops in place */
+		}
+
+		/* Do the security check last because ->init_fs_context may
+		 * change the namespace subscriptions.
+		 */
+		ret = security_fs_context_alloc(fc, fc->root);
+		if (ret < 0) {
+			fc->phase = FS_CONTEXT_FAILED;
+			goto err_unlock;
+		}
+
+		fc->phase = FS_CONTEXT_RECONF_PARAMS;
+	}
+
+	ret = -EINVAL;
+	switch (opt[0]) {
+	case 's':
+		if (fc->phase != FS_CONTEXT_CREATE_PARAMS &&
+		    fc->phase != FS_CONTEXT_RECONF_PARAMS)
+			goto wrong_phase;
+		ret = vfs_set_fs_source(fc, data, len - 2);
+		if (ret < 0)
+			goto err_unlock;
+		break;
+
+	case 'o':
+		if (fc->phase != FS_CONTEXT_CREATE_PARAMS &&
+		    fc->phase != FS_CONTEXT_RECONF_PARAMS)
+			goto wrong_phase;
+		ret = vfs_parse_fs_option(fc, data, len - 2);
+		if (ret < 0)
+			goto err_unlock;
+		break;
+
+	case 'x':
+		if (strcmp(data, "create") == 0) {
+			if (fc->phase != FS_CONTEXT_CREATE_PARAMS)
+				goto wrong_phase;
+			fc->phase = FS_CONTEXT_CREATING;
+			ret = vfs_get_tree(fc);
+			if (ret == 0)
+				fc->phase = FS_CONTEXT_AWAITING_MOUNT;
+			else
+				fc->phase = FS_CONTEXT_FAILED;
+		} else {
+			ret = -EOPNOTSUPP;
+		}
+		if (ret < 0)
+			goto err_unlock;
+		break;
+
+	default:
+		goto err_unlock;
+	}
+
+	ret = len;
+err_unlock:
+	mutex_unlock(&fc->uapi_mutex);
+err_free:
+	kfree(data);
+	return ret;
+
+wrong_phase:
+	ret = -EBUSY;
+	goto err_unlock;
+}
+
+static int fscontext_release(struct inode *inode, struct file *file)
+{
+	struct fs_context *fc = file->private_data;
+
+	if (fc) {
+		file->private_data = NULL;
+		put_fs_context(fc);
+	}
+	return 0;
+}
+
+const struct file_operations fscontext_fs_fops = {
+	.write		= fscontext_write,
+	.release	= fscontext_release,
+	.llseek		= no_llseek,
+};
+
+/*
+ * Attach a filesystem context to a file and an fd.
+ */
+static int fscontext_create_fd(struct fs_context *fc, unsigned int o_flags)
+{
+	int fd;
+
+	fd = anon_inode_getfd("fscontext", &fscontext_fs_fops, fc,
+			      O_RDWR | o_flags);
+	if (fd < 0)
+		put_fs_context(fc);
+	return fd;
+}
+
+/*
+ * Open a filesystem by name so that it can be configured for mounting.
+ *
+ * We are allowed to specify a container in which the filesystem will be
+ * opened, thereby indicating which namespaces will be used (notably, which
+ * network namespace will be used for network filesystems).
+ */
+SYSCALL_DEFINE2(fsopen, const char __user *, _fs_name, unsigned int, flags)
+{
+	struct file_system_type *fs_type;
+	struct fs_context *fc;
+	const char *fs_name;
+
+	if (!ns_capable(current->nsproxy->mnt_ns->user_ns, CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (flags & ~FSOPEN_CLOEXEC)
+		return -EINVAL;
+
+	fs_name = strndup_user(_fs_name, PAGE_SIZE);
+	if (IS_ERR(fs_name))
+		return PTR_ERR(fs_name);
+
+	fs_type = get_fs_type(fs_name);
+	kfree(fs_name);
+	if (!fs_type)
+		return -ENODEV;
+
+	fc = vfs_new_fs_context(fs_type, NULL, 0, FS_CONTEXT_FOR_USER_MOUNT);
+	put_filesystem(fs_type);
+	if (IS_ERR(fc))
+		return PTR_ERR(fc);
+
+	fc->phase = FS_CONTEXT_CREATE_PARAMS;
+	return fscontext_create_fd(fc, flags & FSOPEN_CLOEXEC ? O_CLOEXEC : 0);
+}
diff --git a/include/linux/fs_context.h b/include/linux/fs_context.h
index f157ff935a1e..387f25d7acc4 100644
--- a/include/linux/fs_context.h
+++ b/include/linux/fs_context.h
@@ -14,6 +14,7 @@
 
 #include <linux/kernel.h>
 #include <linux/errno.h>
+#include <linux/mutex.h>
 
 struct cred;
 struct dentry;
@@ -58,6 +59,7 @@ enum fs_context_phase {
  */
 struct fs_context {
 	const struct fs_context_operations *ops;
+	struct mutex		uapi_mutex;	/* Userspace access mutex */
 	struct file_system_type	*fs_type;
 	void			*fs_private;	/* The filesystem's context */
 	struct dentry		*root;		/* The root and superblock */
diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
index 3c0855d9b105..ad6c7ff33c01 100644
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -904,6 +904,7 @@ asmlinkage long sys_open_tree(int dfd, const char __user *path, unsigned flags);
 asmlinkage long sys_move_mount(int from_dfd, const char __user *from_path,
 			       int to_dfd, const char __user *to_path,
 			       unsigned int ms_flags);
+asmlinkage long sys_fsopen(const char __user *fs_name, unsigned int flags);
 
 /*
  * Architecture-specific system calls
diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
index 1c982eb44ff4..f8818e6cddd6 100644
--- a/include/uapi/linux/fs.h
+++ b/include/uapi/linux/fs.h
@@ -344,4 +344,9 @@ typedef int __bitwise __kernel_rwf_t;
 #define RWF_SUPPORTED	(RWF_HIPRI | RWF_DSYNC | RWF_SYNC | RWF_NOWAIT |\
 			 RWF_APPEND)
 
+/*
+ * Flags for fsopen() and co.
+ */
+#define FSOPEN_CLOEXEC		0x00000001
+
 #endif /* _UAPI_LINUX_FS_H */