xref: /src/sys/fs/tmpfs/tmpfs.h

/*	$NetBSD: tmpfs.h,v 1.36 2008/07/28 18:00:20 pooka Exp $	*/

/*
 * Copyright (c) 2005, 2006, 2007 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Julio M. Merino Vidal, developed as part of Google's Summer of Code
 * 2005 program.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

#ifndef _FS_TMPFS_TMPFS_H_
#define _FS_TMPFS_TMPFS_H_

#include <sys/dirent.h>
#include <sys/mount.h>
#include <sys/queue.h>
#include <sys/vnode.h>

#include <fs/tmpfs/tmpfs_pool.h>
#include <fs/tmpfs/tmpfs_args.h>

/* --------------------------------------------------------------------- */

/* Each entry in a directory has a cookie that identifies it.  Cookies
 * supersede offsets within directories because, given how tmpfs stores
 * directories in memory, there is no such thing as an offset.  (Emulating
 * a real offset could be very difficult.)
 *
 * The '.', '..' and the end of directory markers have fixed cookies which
 * cannot collide with the cookies generated by other entries.  The cookies
 * fot the other entries are generated based on the memory address on which
 * stores their information is stored.
 *
 * Ideally, using the entry's memory pointer as the cookie would be enough
 * to represent it and it wouldn't cause collisions in any system.
 * Unfortunately, this results in "offsets" with very large values which
 * later raise problems in the Linux compatibility layer (and maybe in other
 * places) as described in PR kern/32034.  Hence we need to workaround this
 * with a rather ugly hack.
 *
 * Linux 32-bit binaries, unless built with _FILE_OFFSET_BITS=64, have off_t
 * set to 'long', which is a 32-bit *signed* long integer.  Regardless of
 * the macro value, GLIBC (2.3 at least) always uses the getdents64
 * system call (when calling readdir) which internally returns off64_t
 * offsets.  In order to make 32-bit binaries work, *GLIBC* converts the
 * 64-bit values returned by the kernel to 32-bit ones and aborts with
 * EOVERFLOW if the conversion results in values that won't fit in 32-bit
 * integers (which it assumes is because the directory is extremely large).
 * This wouldn't cause problems if we were dealing with unsigned integers,
 * but as we have signed integers, this check fails due to sign expansion.
 *
 * For example, consider that the kernel returns the 0xc1234567 cookie to
 * userspace in a off64_t integer.  Later on, GLIBC casts this value to
 * off_t (remember, signed) with code similar to:
 *     system call returns the offset in kernel_value;
 *     off_t casted_value = kernel_value;
 *     if (sizeof(off_t) != sizeof(off64_t) &&
 *         kernel_value != casted_value)
 *             error!
 * In this case, casted_value still has 0xc1234567, but when it is compared
 * for equality against kernel_value, it is promoted to a 64-bit integer and
 * becomes 0xffffffffc1234567, which is different than 0x00000000c1234567.
 * Then, GLIBC assumes this is because the directory is very large.
 *
 * Given that all the above happens in user-space, we have no control over
 * it; therefore we must workaround the issue here.  We do this by
 * truncating the pointer value to a 32-bit integer and hope that there
 * won't be collisions.  In fact, this will not cause any problems in
 * 32-bit platforms but some might arise in 64-bit machines (I'm not sure
 * if they can happen at all in practice).
 *
 * XXX A nicer solution shall be attempted. */
#define	TMPFS_DIRCOOKIE_DOT	0
#define	TMPFS_DIRCOOKIE_DOTDOT	1
#define	TMPFS_DIRCOOKIE_EOF	2
static __inline
off_t
tmpfs_dircookie(struct tmpfs_dirent *de)
{
	off_t cookie;

	cookie = ((off_t)(uintptr_t)de >> 1) & 0x7FFFFFFF;
	KASSERT(cookie != TMPFS_DIRCOOKIE_DOT);
	KASSERT(cookie != TMPFS_DIRCOOKIE_DOTDOT);
	KASSERT(cookie != TMPFS_DIRCOOKIE_EOF);

	return cookie;
}

/* --------------------------------------------------------------------- */

LIST_HEAD(tmpfs_node_list, tmpfs_node);

/* --------------------------------------------------------------------- */

/*
 * Internal representation of a tmpfs mount point.
 */
struct tmpfs_mount {
	/* Maximum number of memory pages available for use by the file
	 * system, set during mount time.  This variable must never be
	 * used directly as it may be bigger than the current amount of
	 * free memory; in the extreme case, it will hold the SIZE_MAX
	 * value.  Instead, use the TMPFS_PAGES_MAX macro. */
	unsigned int		tm_pages_max;

	/* Number of pages in use by the file system.  Cannot be bigger
	 * than the value returned by TMPFS_PAGES_MAX in any case. */
	unsigned int		tm_pages_used;

	/* Pointer to the node representing the root directory of this
	 * file system. */
	struct tmpfs_node *	tm_root;

	/* Maximum number of possible nodes for this file system; set
	 * during mount time.  We need a hard limit on the maximum number
	 * of nodes to avoid allocating too much of them; their objects
	 * cannot be released until the file system is unmounted.
	 * Otherwise, we could easily run out of memory by creating lots
	 * of empty files and then simply removing them. */
	unsigned int		tm_nodes_max;

	/* Number of nodes currently allocated.  This number only grows.
	 * When it reaches tm_nodes_max, no more new nodes can be allocated.
	 * Of course, the old, unused ones can be reused. */
	unsigned int		tm_nodes_cnt;

	/* Node list. */
	kmutex_t		tm_lock;
	struct tmpfs_node_list	tm_nodes;

	/* Pools used to store file system meta data.  These are not shared
	 * across several instances of tmpfs for the reasons described in
	 * tmpfs_pool.c. */
	struct tmpfs_pool	tm_dirent_pool;
	struct tmpfs_pool	tm_node_pool;
	struct tmpfs_str_pool	tm_str_pool;
};

/* --------------------------------------------------------------------- */

/*
 * This structure maps a file identifier to a tmpfs node.  Used by the
 * NFS code.
 */
struct tmpfs_fid {
	uint16_t		tf_len;
	uint16_t		tf_pad;
	uint32_t		tf_gen;
	ino_t			tf_id;
};

/* --------------------------------------------------------------------- */

/*
 * Prototypes for tmpfs_subr.c.
 */

int	tmpfs_alloc_node(struct tmpfs_mount *, enum vtype,
	    uid_t uid, gid_t gid, mode_t mode, struct tmpfs_node *,
	    char *, dev_t, struct tmpfs_node **);
void	tmpfs_free_node(struct tmpfs_mount *, struct tmpfs_node *);
int	tmpfs_alloc_dirent(struct tmpfs_mount *, struct tmpfs_node *,
	    const char *, uint16_t, struct tmpfs_dirent **);
void	tmpfs_free_dirent(struct tmpfs_mount *, struct tmpfs_dirent *,
	    bool);
int	tmpfs_alloc_vp(struct mount *, struct tmpfs_node *, struct vnode **);
void	tmpfs_free_vp(struct vnode *);
int	tmpfs_alloc_file(struct vnode *, struct vnode **, struct vattr *,
	    struct componentname *, char *);
void	tmpfs_dir_attach(struct vnode *, struct tmpfs_dirent *);
void	tmpfs_dir_detach(struct vnode *, struct tmpfs_dirent *);
struct tmpfs_dirent *	tmpfs_dir_lookup(struct tmpfs_node *node,
			    struct componentname *cnp);
int	tmpfs_dir_getdotdent(struct tmpfs_node *, struct uio *);
int	tmpfs_dir_getdotdotdent(struct tmpfs_node *, struct uio *);
struct tmpfs_dirent *	tmpfs_dir_lookupbycookie(struct tmpfs_node *, off_t);
int	tmpfs_dir_getdents(struct tmpfs_node *, struct uio *, off_t *);
int	tmpfs_reg_resize(struct vnode *, off_t);
size_t	tmpfs_mem_info(bool);
int	tmpfs_chflags(struct vnode *, int, kauth_cred_t, struct lwp *);
int	tmpfs_chmod(struct vnode *, mode_t, kauth_cred_t, struct lwp *);
int	tmpfs_chown(struct vnode *, uid_t, gid_t, kauth_cred_t, struct lwp *);
int	tmpfs_chsize(struct vnode *, u_quad_t, kauth_cred_t, struct lwp *);
int	tmpfs_chtimes(struct vnode *, const struct timespec *,
    const struct timespec *, const struct timespec *, int, kauth_cred_t,
    struct lwp *);
void	tmpfs_itimes(struct vnode *, const struct timespec *,
	    const struct timespec *, const struct timespec *);

void	tmpfs_update(struct vnode *, const struct timespec *,
	    const struct timespec *, const struct timespec *, int);
int	tmpfs_truncate(struct vnode *, off_t);

/* --------------------------------------------------------------------- */

/*
 * Convenience macros to simplify some logical expressions.
 */
#define IMPLIES(a, b) (!(a) || (b))
#define IFF(a, b) (IMPLIES(a, b) && IMPLIES(b, a))

/* --------------------------------------------------------------------- */

/*
 * Checks that the directory entry pointed by 'de' matches the name 'name'
 * with a length of 'len'.
 */
#define TMPFS_DIRENT_MATCHES(de, name, len) \
    (de->td_namelen == (uint16_t)len && \
    memcmp((de)->td_name, (name), (de)->td_namelen) == 0)

/* --------------------------------------------------------------------- */

/*
 * Ensures that the node pointed by 'node' is a directory and that its
 * contents are consistent with respect to directories.
 */
#define TMPFS_VALIDATE_DIR(node) \
    KASSERT((node)->tn_type == VDIR); \
    KASSERT((node)->tn_size % sizeof(struct tmpfs_dirent) == 0); \
    KASSERT((node)->tn_spec.tn_dir.tn_readdir_lastp == NULL || \
        tmpfs_dircookie((node)->tn_spec.tn_dir.tn_readdir_lastp) == \
        (node)->tn_spec.tn_dir.tn_readdir_lastn);

/* --------------------------------------------------------------------- */

/*
 * Memory management stuff.
 */

/* Amount of memory pages to reserve for the system (e.g., to not use by
 * tmpfs).
 * XXX: Should this be tunable through sysctl, for instance? */
#define TMPFS_PAGES_RESERVED (4 * 1024 * 1024 / PAGE_SIZE)

/* Returns the maximum size allowed for a tmpfs file system.  This macro
 * must be used instead of directly retrieving the value from tm_pages_max.
 * The reason is that the size of a tmpfs file system is dynamic: it lets
 * the user store files as long as there is enough free memory (including
 * physical memory and swap space).  Therefore, the amount of memory to be
 * used is either the limit imposed by the user during mount time or the
 * amount of available memory, whichever is lower.  To avoid consuming all
 * the memory for a given mount point, the system will always reserve a
 * minimum of TMPFS_PAGES_RESERVED pages, which is also taken into account
 * by this macro (see above). */
static __inline size_t
TMPFS_PAGES_MAX(struct tmpfs_mount *tmp)
{
	size_t freepages;

	freepages = tmpfs_mem_info(false);
	if (freepages < TMPFS_PAGES_RESERVED)
		freepages = 0;
	else
		freepages -= TMPFS_PAGES_RESERVED;

	return MIN(tmp->tm_pages_max, freepages + tmp->tm_pages_used);
}

/* Returns the available space for the given file system. */
#define TMPFS_PAGES_AVAIL(tmp)		\
    ((ssize_t)(TMPFS_PAGES_MAX(tmp) - (tmp)->tm_pages_used))

/* --------------------------------------------------------------------- */

/*
 * Macros/functions to convert from generic data structures to tmpfs
 * specific ones.
 */

static __inline
struct tmpfs_mount *
VFS_TO_TMPFS(struct mount *mp)
{
	struct tmpfs_mount *tmp;

#ifdef KASSERT
	KASSERT((mp) != NULL && (mp)->mnt_data != NULL);
#endif
	tmp = (struct tmpfs_mount *)(mp)->mnt_data;
	return tmp;
}

static __inline
struct tmpfs_node *
VP_TO_TMPFS_DIR(struct vnode *vp)
{
	struct tmpfs_node *node;

	node = VP_TO_TMPFS_NODE(vp);
#ifdef KASSERT
	TMPFS_VALIDATE_DIR(node);
#endif
	return node;
}
#endif /* _FS_TMPFS_TMPFS_H_ */