Home | History | Annotate | Line # | Download | only in tmpfs
tmpfs.h revision 1.32.10.3
      1 /*	$NetBSD: tmpfs.h,v 1.32.10.3 2010/08/11 22:54:36 yamt Exp $	*/
      2 
      3 /*
      4  * Copyright (c) 2005, 2006, 2007 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Julio M. Merino Vidal, developed as part of Google's Summer of Code
      9  * 2005 program.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     21  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     22  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     23  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     24  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     25  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     26  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     27  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     28  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     29  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     30  * POSSIBILITY OF SUCH DAMAGE.
     31  */
     32 
     33 #ifndef _FS_TMPFS_TMPFS_H_
     34 #define _FS_TMPFS_TMPFS_H_
     35 
     36 #include <sys/dirent.h>
     37 #include <sys/mount.h>
     38 #include <sys/pool.h>
     39 #include <sys/queue.h>
     40 #include <sys/vnode.h>
     41 
     42 /*
     43  * Internal representation of a tmpfs directory entry.
     44  */
     45 struct tmpfs_dirent {
     46 	TAILQ_ENTRY(tmpfs_dirent)	td_entries;
     47 
     48 	/* Length of the name stored in this directory entry.  This avoids
     49 	 * the need to recalculate it every time the name is used. */
     50 	uint16_t			td_namelen;
     51 
     52 	/* The name of the entry, allocated from a string pool.  This
     53 	* string is not required to be zero-terminated; therefore, the
     54 	* td_namelen field must always be used when accessing its value. */
     55 	char *				td_name;
     56 
     57 	/* Pointer to the node this entry refers to. */
     58 	struct tmpfs_node *		td_node;
     59 };
     60 
     61 /* A directory in tmpfs holds a sorted list of directory entries, which in
     62  * turn point to other files (which can be directories themselves).
     63  *
     64  * In tmpfs, this list is managed by a tail queue, whose head is defined by
     65  * the struct tmpfs_dir type.
     66  *
     67  * It is imporant to notice that directories do not have entries for . and
     68  * .. as other file systems do.  These can be generated when requested
     69  * based on information available by other means, such as the pointer to
     70  * the node itself in the former case or the pointer to the parent directory
     71  * in the latter case.  This is done to simplify tmpfs's code and, more
     72  * importantly, to remove redundancy. */
     73 TAILQ_HEAD(tmpfs_dir, tmpfs_dirent);
     74 
     75 /* Each entry in a directory has a cookie that identifies it.  Cookies
     76  * supersede offsets within directories because, given how tmpfs stores
     77  * directories in memory, there is no such thing as an offset.  (Emulating
     78  * a real offset could be very difficult.)
     79  *
     80  * The '.', '..' and the end of directory markers have fixed cookies which
     81  * cannot collide with the cookies generated by other entries.  The cookies
     82  * fot the other entries are generated based on the memory address on which
     83  * stores their information is stored.
     84  *
     85  * Ideally, using the entry's memory pointer as the cookie would be enough
     86  * to represent it and it wouldn't cause collisions in any system.
     87  * Unfortunately, this results in "offsets" with very large values which
     88  * later raise problems in the Linux compatibility layer (and maybe in other
     89  * places) as described in PR kern/32034.  Hence we need to workaround this
     90  * with a rather ugly hack.
     91  *
     92  * Linux 32-bit binaries, unless built with _FILE_OFFSET_BITS=64, have off_t
     93  * set to 'long', which is a 32-bit *signed* long integer.  Regardless of
     94  * the macro value, GLIBC (2.3 at least) always uses the getdents64
     95  * system call (when calling readdir) which internally returns off64_t
     96  * offsets.  In order to make 32-bit binaries work, *GLIBC* converts the
     97  * 64-bit values returned by the kernel to 32-bit ones and aborts with
     98  * EOVERFLOW if the conversion results in values that won't fit in 32-bit
     99  * integers (which it assumes is because the directory is extremely large).
    100  * This wouldn't cause problems if we were dealing with unsigned integers,
    101  * but as we have signed integers, this check fails due to sign expansion.
    102  *
    103  * For example, consider that the kernel returns the 0xc1234567 cookie to
    104  * userspace in a off64_t integer.  Later on, GLIBC casts this value to
    105  * off_t (remember, signed) with code similar to:
    106  *     system call returns the offset in kernel_value;
    107  *     off_t casted_value = kernel_value;
    108  *     if (sizeof(off_t) != sizeof(off64_t) &&
    109  *         kernel_value != casted_value)
    110  *             error!
    111  * In this case, casted_value still has 0xc1234567, but when it is compared
    112  * for equality against kernel_value, it is promoted to a 64-bit integer and
    113  * becomes 0xffffffffc1234567, which is different than 0x00000000c1234567.
    114  * Then, GLIBC assumes this is because the directory is very large.
    115  *
    116  * Given that all the above happens in user-space, we have no control over
    117  * it; therefore we must workaround the issue here.  We do this by
    118  * truncating the pointer value to a 32-bit integer and hope that there
    119  * won't be collisions.  In fact, this will not cause any problems in
    120  * 32-bit platforms but some might arise in 64-bit machines (I'm not sure
    121  * if they can happen at all in practice).
    122  *
    123  * XXX A nicer solution shall be attempted. */
    124 #if defined(_KERNEL)
    125 #define	TMPFS_DIRCOOKIE_DOT	0
    126 #define	TMPFS_DIRCOOKIE_DOTDOT	1
    127 #define	TMPFS_DIRCOOKIE_EOF	2
    128 static __inline
    129 off_t
    130 tmpfs_dircookie(struct tmpfs_dirent *de)
    131 {
    132 	off_t cookie;
    133 
    134 	cookie = ((off_t)(uintptr_t)de >> 1) & 0x7FFFFFFF;
    135 	KASSERT(cookie != TMPFS_DIRCOOKIE_DOT);
    136 	KASSERT(cookie != TMPFS_DIRCOOKIE_DOTDOT);
    137 	KASSERT(cookie != TMPFS_DIRCOOKIE_EOF);
    138 
    139 	return cookie;
    140 }
    141 #endif /* defined(_KERNEL) */
    142 
    143 /* --------------------------------------------------------------------- */
    144 
    145 /*
    146  * Internal representation of a tmpfs file system node.
    147  *
    148  * This structure is splitted in two parts: one holds attributes common
    149  * to all file types and the other holds data that is only applicable to
    150  * a particular type.  The code must be careful to only access those
    151  * attributes that are actually allowed by the node's type.
    152  */
    153 struct tmpfs_node {
    154 	/* Doubly-linked list entry which links all existing nodes for a
    155 	 * single file system.  This is provided to ease the removal of
    156 	 * all nodes during the unmount operation. */
    157 	LIST_ENTRY(tmpfs_node)	tn_entries;
    158 
    159 	/* The node's type.  Any of 'VBLK', 'VCHR', 'VDIR', 'VFIFO',
    160 	 * 'VLNK', 'VREG' and 'VSOCK' is allowed.  The usage of vnode
    161 	 * types instead of a custom enumeration is to make things simpler
    162 	 * and faster, as we do not need to convert between two types. */
    163 	enum vtype		tn_type;
    164 
    165 	/* Node identifier. */
    166 	ino_t			tn_id;
    167 
    168 	/* Node's internal status.  This is used by several file system
    169 	 * operations to do modifications to the node in a delayed
    170 	 * fashion. */
    171 	int			tn_status;
    172 #define	TMPFS_NODE_ACCESSED	(1 << 1)
    173 #define	TMPFS_NODE_MODIFIED	(1 << 2)
    174 #define	TMPFS_NODE_CHANGED	(1 << 3)
    175 
    176 	/* The node size.  It does not necessarily match the real amount
    177 	 * of memory consumed by it. */
    178 	off_t			tn_size;
    179 
    180 	/* Generic node attributes. */
    181 	uid_t			tn_uid;
    182 	gid_t			tn_gid;
    183 	mode_t			tn_mode;
    184 	int			tn_flags;
    185 	nlink_t			tn_links;
    186 	struct timespec		tn_atime;
    187 	struct timespec		tn_mtime;
    188 	struct timespec		tn_ctime;
    189 	struct timespec		tn_birthtime;
    190 	unsigned long		tn_gen;
    191 
    192 	/* Head of byte-level lock list (used by tmpfs_advlock). */
    193 	struct lockf *		tn_lockf;
    194 
    195 	/* As there is a single vnode for each active file within the
    196 	 * system, care has to be taken to avoid allocating more than one
    197 	 * vnode per file.  In order to do this, a bidirectional association
    198 	 * is kept between vnodes and nodes.
    199 	 *
    200 	 * Whenever a vnode is allocated, its v_data field is updated to
    201 	 * point to the node it references.  At the same time, the node's
    202 	 * tn_vnode field is modified to point to the new vnode representing
    203 	 * it.  Further attempts to allocate a vnode for this same node will
    204 	 * result in returning a new reference to the value stored in
    205 	 * tn_vnode.
    206 	 *
    207 	 * May be NULL when the node is unused (that is, no vnode has been
    208 	 * allocated for it or it has been reclaimed). */
    209 	kmutex_t		tn_vlock;
    210 	struct vnode *		tn_vnode;
    211 
    212 	union {
    213 		/* Valid when tn_type == VBLK || tn_type == VCHR. */
    214 		struct {
    215 			dev_t			tn_rdev;
    216 		} tn_dev;
    217 
    218 		/* Valid when tn_type == VDIR. */
    219 		struct {
    220 			/* Pointer to the parent directory.  The root
    221 			 * directory has a pointer to itself in this field;
    222 			 * this property identifies the root node. */
    223 			struct tmpfs_node *	tn_parent;
    224 
    225 			/* Head of a tail-queue that links the contents of
    226 			 * the directory together.  See above for a
    227 			 * description of its contents. */
    228 			struct tmpfs_dir	tn_dir;
    229 
    230 			/* Number and pointer of the first directory entry
    231 			 * returned by the readdir operation if it were
    232 			 * called again to continue reading data from the
    233 			 * same directory as before.  This is used to speed
    234 			 * up reads of long directories, assuming that no
    235 			 * more than one read is in progress at a given time.
    236 			 * Otherwise, these values are discarded and a linear
    237 			 * scan is performed from the beginning up to the
    238 			 * point where readdir starts returning values. */
    239 			off_t			tn_readdir_lastn;
    240 			struct tmpfs_dirent *	tn_readdir_lastp;
    241 		} tn_dir;
    242 
    243 		/* Valid when tn_type == VLNK. */
    244 		struct tn_lnk {
    245 			/* The link's target, allocated from a string pool. */
    246 			char *			tn_link;
    247 		} tn_lnk;
    248 
    249 		/* Valid when tn_type == VREG. */
    250 		struct tn_reg {
    251 			/* The contents of regular files stored in a tmpfs
    252 			 * file system are represented by a single anonymous
    253 			 * memory object (aobj, for short).  The aobj provides
    254 			 * direct access to any position within the file,
    255 			 * because its contents are always mapped in a
    256 			 * contiguous region of virtual memory.  It is a task
    257 			 * of the memory management subsystem (see uvm(9)) to
    258 			 * issue the required page ins or page outs whenever
    259 			 * a position within the file is accessed. */
    260 			struct uvm_object *	tn_aobj;
    261 			size_t			tn_aobj_pages;
    262 		} tn_reg;
    263 	} tn_spec;
    264 };
    265 
    266 #if defined(_KERNEL)
    267 LIST_HEAD(tmpfs_node_list, tmpfs_node);
    268 
    269 /* --------------------------------------------------------------------- */
    270 
    271 /*
    272  * Internal representation of a tmpfs mount point.
    273  */
    274 struct tmpfs_mount {
    275 	/* Limit and number of bytes in use by the file system. */
    276 	uint64_t		tm_mem_limit;
    277 	uint64_t		tm_bytes_used;
    278 	kmutex_t		tm_acc_lock;
    279 
    280 	/* Pointer to the node representing the root directory of this
    281 	 * file system. */
    282 	struct tmpfs_node *	tm_root;
    283 
    284 	/* Maximum number of possible nodes for this file system; set
    285 	 * during mount time.  We need a hard limit on the maximum number
    286 	 * of nodes to avoid allocating too much of them; their objects
    287 	 * cannot be released until the file system is unmounted.
    288 	 * Otherwise, we could easily run out of memory by creating lots
    289 	 * of empty files and then simply removing them. */
    290 	unsigned int		tm_nodes_max;
    291 
    292 	/* Number of nodes currently allocated.  This number only grows.
    293 	 * When it reaches tm_nodes_max, no more new nodes can be allocated.
    294 	 * Of course, the old, unused ones can be reused. */
    295 	unsigned int		tm_nodes_cnt;
    296 
    297 	/* Node list. */
    298 	kmutex_t		tm_lock;
    299 	struct tmpfs_node_list	tm_nodes;
    300 
    301 	char			tm_dwchan[32];
    302 	struct pool		tm_dirent_pool;
    303 	char			tm_nwchan[32];
    304 	struct pool		tm_node_pool;
    305 };
    306 
    307 /* --------------------------------------------------------------------- */
    308 
    309 /*
    310  * This structure maps a file identifier to a tmpfs node.  Used by the
    311  * NFS code.
    312  */
    313 struct tmpfs_fid {
    314 	uint16_t		tf_len;
    315 	uint16_t		tf_pad;
    316 	uint32_t		tf_gen;
    317 	ino_t			tf_id;
    318 };
    319 
    320 /* --------------------------------------------------------------------- */
    321 
    322 /*
    323  * Prototypes for tmpfs_subr.c.
    324  */
    325 
    326 int	tmpfs_alloc_node(struct tmpfs_mount *, enum vtype,
    327 	    uid_t uid, gid_t gid, mode_t mode, struct tmpfs_node *,
    328 	    char *, dev_t, struct tmpfs_node **);
    329 void	tmpfs_free_node(struct tmpfs_mount *, struct tmpfs_node *);
    330 int	tmpfs_alloc_dirent(struct tmpfs_mount *, struct tmpfs_node *,
    331 	    const char *, uint16_t, struct tmpfs_dirent **);
    332 void	tmpfs_free_dirent(struct tmpfs_mount *, struct tmpfs_dirent *,
    333 	    bool);
    334 int	tmpfs_alloc_vp(struct mount *, struct tmpfs_node *, struct vnode **);
    335 void	tmpfs_free_vp(struct vnode *);
    336 int	tmpfs_alloc_file(struct vnode *, struct vnode **, struct vattr *,
    337 	    struct componentname *, char *);
    338 void	tmpfs_dir_attach(struct vnode *, struct tmpfs_dirent *);
    339 void	tmpfs_dir_detach(struct vnode *, struct tmpfs_dirent *);
    340 struct tmpfs_dirent *	tmpfs_dir_lookup(struct tmpfs_node *node,
    341 			    struct componentname *cnp);
    342 int	tmpfs_dir_getdotdent(struct tmpfs_node *, struct uio *);
    343 int	tmpfs_dir_getdotdotdent(struct tmpfs_node *, struct uio *);
    344 struct tmpfs_dirent *	tmpfs_dir_lookupbycookie(struct tmpfs_node *, off_t);
    345 int	tmpfs_dir_getdents(struct tmpfs_node *, struct uio *, off_t *);
    346 int	tmpfs_reg_resize(struct vnode *, off_t);
    347 int	tmpfs_chflags(struct vnode *, int, kauth_cred_t, struct lwp *);
    348 int	tmpfs_chmod(struct vnode *, mode_t, kauth_cred_t, struct lwp *);
    349 int	tmpfs_chown(struct vnode *, uid_t, gid_t, kauth_cred_t, struct lwp *);
    350 int	tmpfs_chsize(struct vnode *, u_quad_t, kauth_cred_t, struct lwp *);
    351 int	tmpfs_chtimes(struct vnode *, const struct timespec *,
    352     const struct timespec *, const struct timespec *, int, kauth_cred_t,
    353     struct lwp *);
    354 void	tmpfs_itimes(struct vnode *, const struct timespec *,
    355 	    const struct timespec *, const struct timespec *);
    356 
    357 void	tmpfs_update(struct vnode *, const struct timespec *,
    358 	    const struct timespec *, const struct timespec *, int);
    359 int	tmpfs_truncate(struct vnode *, off_t);
    360 
    361 /*
    362  * Prototypes for tmpfs_mem.c.
    363  */
    364 
    365 void		tmpfs_mntmem_init(struct tmpfs_mount *, uint64_t);
    366 void		tmpfs_mntmem_destroy(struct tmpfs_mount *);
    367 
    368 size_t		tmpfs_mem_info(bool);
    369 uint64_t	tmpfs_bytes_max(struct tmpfs_mount *);
    370 size_t		tmpfs_pages_avail(struct tmpfs_mount *);
    371 bool		tmpfs_mem_incr(struct tmpfs_mount *, size_t);
    372 void		tmpfs_mem_decr(struct tmpfs_mount *, size_t);
    373 
    374 struct tmpfs_dirent *tmpfs_dirent_get(struct tmpfs_mount *);
    375 void		tmpfs_dirent_put(struct tmpfs_mount *, struct tmpfs_dirent *);
    376 
    377 struct tmpfs_node *tmpfs_node_get(struct tmpfs_mount *);
    378 void		tmpfs_node_put(struct tmpfs_mount *, struct tmpfs_node *);
    379 
    380 char *		tmpfs_strname_alloc(struct tmpfs_mount *, size_t);
    381 void		tmpfs_strname_free(struct tmpfs_mount *, char *, size_t);
    382 bool		tmpfs_strname_neqlen(struct componentname *, struct componentname *);
    383 
    384 /* --------------------------------------------------------------------- */
    385 
    386 /*
    387  * Convenience macros to simplify some logical expressions.
    388  */
    389 #define IMPLIES(a, b) (!(a) || (b))
    390 #define IFF(a, b) (IMPLIES(a, b) && IMPLIES(b, a))
    391 
    392 /* --------------------------------------------------------------------- */
    393 
    394 /*
    395  * Checks that the directory entry pointed by 'de' matches the name 'name'
    396  * with a length of 'len'.
    397  */
    398 #define TMPFS_DIRENT_MATCHES(de, name, len) \
    399     (de->td_namelen == (uint16_t)len && \
    400     memcmp((de)->td_name, (name), (de)->td_namelen) == 0)
    401 
    402 /* --------------------------------------------------------------------- */
    403 
    404 /*
    405  * Ensures that the node pointed by 'node' is a directory and that its
    406  * contents are consistent with respect to directories.
    407  */
    408 #define TMPFS_VALIDATE_DIR(node) \
    409     KASSERT((node)->tn_type == VDIR); \
    410     KASSERT((node)->tn_size % sizeof(struct tmpfs_dirent) == 0); \
    411     KASSERT((node)->tn_spec.tn_dir.tn_readdir_lastp == NULL || \
    412         tmpfs_dircookie((node)->tn_spec.tn_dir.tn_readdir_lastp) == \
    413         (node)->tn_spec.tn_dir.tn_readdir_lastn);
    414 
    415 /* --------------------------------------------------------------------- */
    416 
    417 /*
    418  * Memory management stuff.
    419  */
    420 
    421 /* Amount of memory pages to reserve for the system (e.g., to not use by
    422  * tmpfs).
    423  * XXX: Should this be tunable through sysctl, for instance? */
    424 #define TMPFS_PAGES_RESERVED (4 * 1024 * 1024 / PAGE_SIZE)
    425 
    426 /*
    427  * Macros/functions to convert from generic data structures to tmpfs
    428  * specific ones.
    429  */
    430 
    431 static __inline
    432 struct tmpfs_mount *
    433 VFS_TO_TMPFS(struct mount *mp)
    434 {
    435 	struct tmpfs_mount *tmp;
    436 
    437 #ifdef KASSERT
    438 	KASSERT((mp) != NULL && (mp)->mnt_data != NULL);
    439 #endif
    440 	tmp = (struct tmpfs_mount *)(mp)->mnt_data;
    441 	return tmp;
    442 }
    443 
    444 #endif /* defined(_KERNEL) */
    445 
    446 static __inline
    447 struct tmpfs_node *
    448 VP_TO_TMPFS_NODE(struct vnode *vp)
    449 {
    450 	struct tmpfs_node *node;
    451 
    452 #ifdef KASSERT
    453 	KASSERT((vp) != NULL && (vp)->v_data != NULL);
    454 #endif
    455 	node = (struct tmpfs_node *)vp->v_data;
    456 	return node;
    457 }
    458 
    459 #if defined(_KERNEL)
    460 
    461 static __inline
    462 struct tmpfs_node *
    463 VP_TO_TMPFS_DIR(struct vnode *vp)
    464 {
    465 	struct tmpfs_node *node;
    466 
    467 	node = VP_TO_TMPFS_NODE(vp);
    468 #ifdef KASSERT
    469 	TMPFS_VALIDATE_DIR(node);
    470 #endif
    471 	return node;
    472 }
    473 
    474 #endif /* defined(_KERNEL) */
    475 #endif /* _FS_TMPFS_TMPFS_H_ */
    476