xref: /dragonfly/sys/vfs/tmpfs/tmpfs_vfsops.c (revision 87f62b1c)
1 /*	$NetBSD: tmpfs_vfsops.c,v 1.10 2005/12/11 12:24:29 christos Exp $	*/
2 
3 /*-
4  * Copyright (c) 2005 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 /*
34  * Efficient memory file system.
35  *
36  * tmpfs is a file system that uses virtual memory to store file data and
37  * metadata efficiently. It does not follow the structure of an on-disk
38  * file system because it simply does not need to. Instead, it uses
39  * memory-specific data structures and algorithms to automatically
40  * allocate and release resources.
41  */
42 
43 #include <sys/conf.h>
44 #include <sys/param.h>
45 #include <sys/limits.h>
46 #include <sys/lock.h>
47 #include <sys/mutex.h>
48 #include <sys/kernel.h>
49 #include <sys/stat.h>
50 #include <sys/systm.h>
51 #include <sys/sysctl.h>
52 #include <sys/objcache.h>
53 
54 #include <vm/vm.h>
55 #include <vm/vm_object.h>
56 #include <vm/vm_param.h>
57 
58 #if 0
59 #include <vfs/tmpfs/tmpfs.h>
60 #endif
61 #include "tmpfs.h"
62 #include <vfs/tmpfs/tmpfs_vnops.h>
63 #include <vfs/tmpfs/tmpfs_args.h>
64 
65 /*
66  * Default permission for root node
67  */
68 #define TMPFS_DEFAULT_ROOT_MODE	(S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH)
69 
70 MALLOC_DEFINE(M_TMPFSMNT, "tmpfs mount", "tmpfs mount structures");
71 
72 /* --------------------------------------------------------------------- */
73 
74 static int	tmpfs_mount(struct mount *, char *, caddr_t, struct ucred *);
75 static int	tmpfs_unmount(struct mount *, int);
76 static int	tmpfs_root(struct mount *, struct vnode **);
77 static int	tmpfs_fhtovp(struct mount *, struct vnode *, struct fid *, struct vnode **);
78 static int	tmpfs_statfs(struct mount *, struct statfs *, struct ucred *cred);
79 
80 /* --------------------------------------------------------------------- */
81 boolean_t
82 tmpfs_node_ctor(void *obj, void *privdata, int flags)
83 {
84 	struct tmpfs_node *node = obj;
85 
86 	node->tn_gen++;
87 	node->tn_size = 0;
88 	node->tn_status = 0;
89 	node->tn_flags = 0;
90 	node->tn_links = 0;
91 	node->tn_vnode = NULL;
92 	node->tn_vpstate = TMPFS_VNODE_WANT;
93 	bzero(&node->tn_spec, sizeof(node->tn_spec));
94 
95 	return (TRUE);
96 }
97 
98 static void
99 tmpfs_node_dtor(void *obj, void *privdata)
100 {
101 	struct tmpfs_node *node = (struct tmpfs_node *)obj;
102 	node->tn_type = VNON;
103 	node->tn_vpstate = TMPFS_VNODE_DOOMED;
104 }
105 
106 static void *
107 tmpfs_node_init(void *args, int flags)
108 {
109 	struct tmpfs_node *node = objcache_malloc_alloc(args, flags);
110 	if (node == NULL)
111 		return (NULL);
112 	node->tn_id = 0;
113 
114 	lockinit(&node->tn_interlock, "tmpfs node interlock", 0, LK_CANRECURSE);
115 	node->tn_gen = karc4random();
116 
117 	return node;
118 }
119 
120 static void
121 tmpfs_node_fini(void *obj, void *args)
122 {
123 	struct tmpfs_node *node = (struct tmpfs_node *)obj;
124 	lockuninit(&node->tn_interlock);
125 	objcache_malloc_free(obj, args);
126 }
127 
128 static int
129 tmpfs_mount(struct mount *mp, char *path, caddr_t data, struct ucred *cred)
130 {
131 	struct tmpfs_mount *tmp;
132 	struct tmpfs_node *root;
133 	struct tmpfs_args args;
134 	vm_pindex_t pages;
135 	vm_pindex_t pages_limit;
136 	ino_t nodes;
137 	u_int64_t	maxfsize;
138 	int error;
139 	/* Size counters. */
140 	ino_t	nodes_max;
141 	off_t	size_max;
142 	size_t	maxfsize_max;
143 	size_t	size;
144 
145 	/* Root node attributes. */
146 	uid_t	root_uid = cred->cr_uid;
147 	gid_t	root_gid = cred->cr_gid;
148 	mode_t	root_mode = (VREAD | VWRITE);
149 
150 	if (mp->mnt_flag & MNT_UPDATE) {
151 		/* XXX: There is no support yet to update file system
152 		 * settings.  Should be added. */
153 
154 		return EOPNOTSUPP;
155 	}
156 
157 	/*
158 	 * mount info
159 	 */
160 	bzero(&args, sizeof(args));
161 	size_max  = 0;
162 	nodes_max = 0;
163 	maxfsize_max = 0;
164 
165 	if (path) {
166 		if (data) {
167 			error = copyin(data, &args, sizeof(args));
168 			if (error)
169 				return (error);
170 		}
171 		size_max = args.ta_size_max;
172 		nodes_max = args.ta_nodes_max;
173 		maxfsize_max = args.ta_maxfsize_max;
174 		root_uid = args.ta_root_uid;
175 		root_gid = args.ta_root_gid;
176 		root_mode = args.ta_root_mode;
177 	}
178 
179 	/*
180 	 * If mount by non-root, then verify that user has necessary
181 	 * permissions on the device.
182 	 */
183 	if (cred->cr_uid != 0) {
184 		root_mode = VREAD;
185 		if ((mp->mnt_flag & MNT_RDONLY) == 0)
186 			root_mode |= VWRITE;
187 	}
188 
189 	pages_limit = vm_swap_max + vmstats.v_page_count / 2;
190 
191 	if (size_max == 0) {
192 		pages = pages_limit / 2;
193 	} else if (size_max < PAGE_SIZE) {
194 		pages = 1;
195 	} else if (OFF_TO_IDX(size_max) > pages_limit) {
196 		/*
197 		 * do not force pages = pages_limit for this case, otherwise
198 		 * we might not honor tmpfs size requests from /etc/fstab
199 		 * during boot because they are mounted prior to swap being
200 		 * turned on.
201 		 */
202 		pages = OFF_TO_IDX(size_max);
203 	} else {
204 		pages = OFF_TO_IDX(size_max);
205 	}
206 
207 	if (nodes_max == 0)
208 		nodes = 3 + pages * PAGE_SIZE / 1024;
209 	else if (nodes_max < 3)
210 		nodes = 3;
211 	else if (nodes_max > pages)
212 		nodes = pages;
213 	else
214 		nodes = nodes_max;
215 
216 	maxfsize = IDX_TO_OFF(pages_limit);
217 	if (maxfsize_max != 0 && maxfsize > maxfsize_max)
218 		maxfsize = maxfsize_max;
219 
220 	/* Allocate the tmpfs mount structure and fill it. */
221 	tmp = kmalloc(sizeof(*tmp), M_TMPFSMNT, M_WAITOK | M_ZERO);
222 
223 	tmp->tm_mount = mp;
224 	tmp->tm_nodes_max = nodes;
225 	tmp->tm_nodes_inuse = 0;
226 	tmp->tm_maxfilesize = maxfsize;
227 	LIST_INIT(&tmp->tm_nodes_used);
228 
229 	tmp->tm_pages_max = pages;
230 	tmp->tm_pages_used = 0;
231 
232 	kmalloc_create(&tmp->tm_node_zone, "tmpfs node");
233 	kmalloc_create(&tmp->tm_dirent_zone, "tmpfs dirent");
234 	kmalloc_create(&tmp->tm_name_zone, "tmpfs name zone");
235 
236 	kmalloc_raise_limit(tmp->tm_node_zone, sizeof(struct tmpfs_node) *
237 			    tmp->tm_nodes_max);
238 
239 	tmp->tm_node_zone_malloc_args.objsize = sizeof(struct tmpfs_node);
240 	tmp->tm_node_zone_malloc_args.mtype = tmp->tm_node_zone;
241 
242 	tmp->tm_dirent_zone_malloc_args.objsize = sizeof(struct tmpfs_dirent);
243 	tmp->tm_dirent_zone_malloc_args.mtype = tmp->tm_dirent_zone;
244 
245 	tmp->tm_dirent_pool =  objcache_create( "tmpfs dirent cache",
246 	    0, 0,
247 	    NULL, NULL, NULL,
248 	    objcache_malloc_alloc, objcache_malloc_free,
249 	    &tmp->tm_dirent_zone_malloc_args);
250 	tmp->tm_node_pool = objcache_create( "tmpfs node cache",
251 	    0, 0,
252 	    tmpfs_node_ctor, tmpfs_node_dtor, NULL,
253 	    tmpfs_node_init, tmpfs_node_fini,
254 	    &tmp->tm_node_zone_malloc_args);
255 
256 	tmp->tm_ino = TMPFS_ROOTINO;
257 
258 	/* Allocate the root node. */
259 	error = tmpfs_alloc_node(tmp, VDIR, root_uid, root_gid,
260 				 root_mode & ALLPERMS, NULL,
261 				 VNOVAL, VNOVAL, &root);
262 
263 	/*
264 	 * We are backed by swap, set snocache chflags flag so we
265 	 * don't trip over swapcache.
266 	 */
267 	root->tn_flags = SF_NOCACHE;
268 
269 	if (error != 0 || root == NULL) {
270 	    objcache_destroy(tmp->tm_node_pool);
271 	    objcache_destroy(tmp->tm_dirent_pool);
272 	    kfree(tmp, M_TMPFSMNT);
273 	    return error;
274 	}
275 	KASSERT(root->tn_id == TMPFS_ROOTINO,
276 		("tmpfs root with invalid ino: %ju", (uintmax_t)root->tn_id));
277 
278 	++root->tn_links;	/* prevent destruction */
279 	tmp->tm_root = root;
280 
281 	mp->mnt_flag |= MNT_LOCAL;
282 	mp->mnt_kern_flag |= MNTK_ALL_MPSAFE;
283 	mp->mnt_kern_flag |= MNTK_NOMSYNC;
284 	mp->mnt_kern_flag |= MNTK_THR_SYNC;	/* new vsyncscan semantics */
285 	mp->mnt_data = (qaddr_t)tmp;
286 	vfs_getnewfsid(mp);
287 
288 	vfs_add_vnodeops(mp, &tmpfs_vnode_vops, &mp->mnt_vn_norm_ops);
289 	vfs_add_vnodeops(mp, &tmpfs_fifo_vops, &mp->mnt_vn_fifo_ops);
290 
291 	copystr("tmpfs", mp->mnt_stat.f_mntfromname, MNAMELEN - 1, &size);
292 	bzero(mp->mnt_stat.f_mntfromname +size, MNAMELEN - size);
293 	bzero(mp->mnt_stat.f_mntonname, sizeof(mp->mnt_stat.f_mntonname));
294 	copyinstr(path, mp->mnt_stat.f_mntonname,
295 		  sizeof(mp->mnt_stat.f_mntonname) -1,
296 		  &size);
297 
298 	tmpfs_statfs(mp, &mp->mnt_stat, cred);
299 
300 	return 0;
301 }
302 
303 /* --------------------------------------------------------------------- */
304 
305 /* ARGSUSED2 */
306 static int
307 tmpfs_unmount(struct mount *mp, int mntflags)
308 {
309 	int error;
310 	int flags = 0;
311 	struct tmpfs_mount *tmp;
312 	struct tmpfs_node *node;
313 	struct vnode *vp;
314 	int isok;
315 
316 	tmp = VFS_TO_TMPFS(mp);
317 	TMPFS_LOCK(tmp);
318 
319 	/* Handle forced unmounts. */
320 	if (mntflags & MNT_FORCE)
321 		flags |= FORCECLOSE;
322 
323 	/*
324 	 * Finalize all pending I/O.  In the case of tmpfs we want
325 	 * to throw all the data away so clean out the buffer cache
326 	 * and vm objects before calling vflush().
327 	 */
328 	LIST_FOREACH(node, &tmp->tm_nodes_used, tn_entries) {
329 		/*
330 		 * tn_links is mnt_token protected
331 		 */
332 		TMPFS_NODE_LOCK(node);
333 		++node->tn_links;
334 		TMPFS_NODE_UNLOCK(node);
335 
336 		while (node->tn_type == VREG && node->tn_vnode) {
337 			vp = node->tn_vnode;
338 			vhold(vp);
339 			lwkt_yield();
340 
341 			/*
342 			 * vx_get/vx_put and tmpfs_truncate may block,
343 			 * releasing the tmpfs mountpoint token.
344 			 *
345 			 * Make sure the lock order is correct.
346 			 */
347 			vx_get(vp);		/* held vnode */
348 			TMPFS_NODE_LOCK(node);
349 			if (node->tn_vnode == vp) {
350 				tmpfs_truncate(vp, 0);
351 				isok = 1;
352 			} else {
353 				isok = 0;
354 			}
355 			TMPFS_NODE_UNLOCK(node);
356 			vx_put(vp);
357 			vdrop(vp);
358 			if (isok)
359 				break;
360 			/* retry */
361 		}
362 
363 		TMPFS_NODE_LOCK(node);
364 		--node->tn_links;
365 		TMPFS_NODE_UNLOCK(node);
366 	}
367 
368 	/*
369 	 * Flush all vnodes on the unmount.
370 	 *
371 	 * If we fail to flush, we cannot unmount, but all the nodes have
372 	 * already been truncated. Erroring out is the best we can do.
373 	 */
374 	error = vflush(mp, 0, flags);
375 	if (error != 0) {
376 		TMPFS_UNLOCK(tmp);
377 		return (error);
378 	}
379 
380 	/*
381 	 * First pass get rid of all the directory entries and
382 	 * vnode associations.  This will also destroy the
383 	 * directory topology and should drop all link counts
384 	 * to 0 except for the root.
385 	 *
386 	 * No vnodes should remain after the vflush above.
387 	 */
388 	LIST_FOREACH(node, &tmp->tm_nodes_used, tn_entries) {
389 		lwkt_yield();
390 
391 		TMPFS_NODE_LOCK(node);
392 		++node->tn_links;
393 		if (node->tn_type == VDIR) {
394 			struct tmpfs_dirent *de;
395 
396 			while ((de = RB_ROOT(&node->tn_dir.tn_dirtree)) != NULL)			{
397 				tmpfs_dir_detach(node, de);
398 				tmpfs_free_dirent(tmp, de);
399 			}
400 		}
401 		KKASSERT(node->tn_vnode == NULL);
402 
403 		--node->tn_links;
404 		TMPFS_NODE_UNLOCK(node);
405 	}
406 
407 	/*
408 	 * Allow the root node to be destroyed by dropping the link count
409 	 * we bumped in the mount code.
410 	 */
411 	KKASSERT(tmp->tm_root);
412 	TMPFS_NODE_LOCK(tmp->tm_root);
413 	--tmp->tm_root->tn_links;
414 	TMPFS_NODE_UNLOCK(tmp->tm_root);
415 
416 	/*
417 	 * At this point all nodes, including the root node, should have a
418 	 * link count of 0.  The root is not necessarily going to be last.
419 	 */
420 	while ((node = LIST_FIRST(&tmp->tm_nodes_used)) != NULL) {
421 		if (node->tn_links)
422 			panic("tmpfs: Dangling nodes during umount (%p)!\n",
423 			      node);
424 
425 		TMPFS_NODE_LOCK(node);
426 		tmpfs_free_node(tmp, node);
427 		/* eats lock */
428 		lwkt_yield();
429 	}
430 	KKASSERT(tmp->tm_root == NULL);
431 
432 	objcache_destroy(tmp->tm_dirent_pool);
433 	objcache_destroy(tmp->tm_node_pool);
434 
435 	kmalloc_destroy(&tmp->tm_name_zone);
436 	kmalloc_destroy(&tmp->tm_dirent_zone);
437 	kmalloc_destroy(&tmp->tm_node_zone);
438 
439 	tmp->tm_node_zone = tmp->tm_dirent_zone = NULL;
440 
441 	KKASSERT(tmp->tm_pages_used == 0);
442 	KKASSERT(tmp->tm_nodes_inuse == 0);
443 
444 	TMPFS_UNLOCK(tmp);
445 
446 	/* Throw away the tmpfs_mount structure. */
447 	kfree(tmp, M_TMPFSMNT);
448 	mp->mnt_data = NULL;
449 
450 	mp->mnt_flag &= ~MNT_LOCAL;
451 	return 0;
452 }
453 
454 /* --------------------------------------------------------------------- */
455 
456 static int
457 tmpfs_root(struct mount *mp, struct vnode **vpp)
458 {
459 	struct tmpfs_mount *tmp;
460 	int error;
461 
462 	tmp = VFS_TO_TMPFS(mp);
463 	if (tmp->tm_root == NULL) {
464 		kprintf("tmpfs_root: called without root node %p\n", mp);
465 		print_backtrace(-1);
466 		*vpp = NULL;
467 		error = EINVAL;
468 	} else {
469 		error = tmpfs_alloc_vp(mp, tmp->tm_root, LK_EXCLUSIVE, vpp);
470 		(*vpp)->v_flag |= VROOT;
471 		(*vpp)->v_type = VDIR;
472 	}
473 	return error;
474 }
475 
476 /* --------------------------------------------------------------------- */
477 
478 static int
479 tmpfs_fhtovp(struct mount *mp, struct vnode *rootvp, struct fid *fhp,
480 	     struct vnode **vpp)
481 {
482 	boolean_t found;
483 	struct tmpfs_fid *tfhp;
484 	struct tmpfs_mount *tmp;
485 	struct tmpfs_node *node;
486 	int rc;
487 
488 	tmp = VFS_TO_TMPFS(mp);
489 
490 	tfhp = (struct tmpfs_fid *) fhp;
491 	if (tfhp->tf_len != sizeof(struct tmpfs_fid))
492 		return EINVAL;
493 
494 	if (tfhp->tf_id >= tmp->tm_nodes_max)
495 		return EINVAL;
496 
497 	rc = EINVAL;
498 	found = FALSE;
499 
500 	TMPFS_LOCK(tmp);
501 	LIST_FOREACH(node, &tmp->tm_nodes_used, tn_entries) {
502 		if (node->tn_id == tfhp->tf_id &&
503 		    node->tn_gen == tfhp->tf_gen) {
504 			found = TRUE;
505 			break;
506 		}
507 	}
508 
509 	if (found)
510 		rc = tmpfs_alloc_vp(mp, node, LK_EXCLUSIVE, vpp);
511 
512 	TMPFS_UNLOCK(tmp);
513 
514 	return (rc);
515 }
516 
517 /* --------------------------------------------------------------------- */
518 
519 /* ARGSUSED2 */
520 static int
521 tmpfs_statfs(struct mount *mp, struct statfs *sbp, struct ucred *cred)
522 {
523 	fsfilcnt_t freenodes;
524 	struct tmpfs_mount *tmp;
525 
526 	tmp = VFS_TO_TMPFS(mp);
527 
528 	TMPFS_LOCK(tmp);
529 	sbp->f_iosize = PAGE_SIZE;
530 	sbp->f_bsize = PAGE_SIZE;
531 
532 	sbp->f_blocks = tmp->tm_pages_max;
533 	sbp->f_bavail = tmp->tm_pages_max - tmp->tm_pages_used;
534 	sbp->f_bfree = sbp->f_bavail;
535 
536 	freenodes = tmp->tm_nodes_max - tmp->tm_nodes_inuse;
537 
538 	sbp->f_files = freenodes + tmp->tm_nodes_inuse;
539 	sbp->f_ffree = freenodes;
540 	sbp->f_owner = tmp->tm_root->tn_uid;
541 
542 	TMPFS_UNLOCK(tmp);
543 
544 	return 0;
545 }
546 
547 /* --------------------------------------------------------------------- */
548 
549 static int
550 tmpfs_vptofh(struct vnode *vp, struct fid *fhp)
551 {
552 	struct tmpfs_node *node;
553 	struct tmpfs_fid tfh;
554 	node = VP_TO_TMPFS_NODE(vp);
555 	memset(&tfh, 0, sizeof(tfh));
556 	tfh.tf_len = sizeof(struct tmpfs_fid);
557 	tfh.tf_gen = node->tn_gen;
558 	tfh.tf_id = node->tn_id;
559 	memcpy(fhp, &tfh, sizeof(tfh));
560 	return (0);
561 }
562 
563 /* --------------------------------------------------------------------- */
564 
565 static int
566 tmpfs_checkexp(struct mount *mp, struct sockaddr *nam, int *exflagsp,
567 	       struct ucred **credanonp)
568 {
569 	struct tmpfs_mount *tmp;
570 	struct netcred *nc;
571 
572 	tmp = (struct tmpfs_mount *) mp->mnt_data;
573 	nc = vfs_export_lookup(mp, &tmp->tm_export, nam);
574 	if (nc == NULL)
575 		return (EACCES);
576 
577 	*exflagsp = nc->netc_exflags;
578 	*credanonp = &nc->netc_anon;
579 
580 	return (0);
581 }
582 
583 /* --------------------------------------------------------------------- */
584 
585 /*
586  * tmpfs vfs operations.
587  */
588 
589 static struct vfsops tmpfs_vfsops = {
590 	.vfs_mount =			tmpfs_mount,
591 	.vfs_unmount =			tmpfs_unmount,
592 	.vfs_root =			tmpfs_root,
593 	.vfs_statfs =			tmpfs_statfs,
594 	.vfs_fhtovp =			tmpfs_fhtovp,
595 	.vfs_vptofh =			tmpfs_vptofh,
596 	.vfs_checkexp =			tmpfs_checkexp,
597 };
598 
599 VFS_SET(tmpfs_vfsops, tmpfs, VFCF_MPSAFE);
600 MODULE_VERSION(tmpfs, 1);
601