Blame view
fs/notify/mark.c
24.9 KB
c82ee6d3b
|
1 |
// SPDX-License-Identifier: GPL-2.0-or-later |
5444e2981
|
2 3 |
/* * Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@redhat.com> |
5444e2981
|
4 5 6 7 8 9 |
*/ /* * fsnotify inode mark locking/lifetime/and refcnting * * REFCNT: |
9756b9187
|
10 11 12 13 14 15 |
* The group->recnt and mark->refcnt tell how many "things" in the kernel * currently are referencing the objects. Both kind of objects typically will * live inside the kernel with a refcnt of 2, one for its creation and one for * the reference a group and a mark hold to each other. * If you are holding the appropriate locks, you can take a reference and the * object itself is guaranteed to survive until the reference is dropped. |
5444e2981
|
16 17 |
* * LOCKING: |
9756b9187
|
18 19 |
* There are 3 locks involved with fsnotify inode marks and they MUST be taken * in order as follows: |
5444e2981
|
20 |
* |
9756b9187
|
21 |
* group->mark_mutex |
5444e2981
|
22 |
* mark->lock |
04662cab5
|
23 |
* mark->connector->lock |
5444e2981
|
24 |
* |
9756b9187
|
25 26 27 28 29 30 31 32 |
* group->mark_mutex protects the marks_list anchored inside a given group and * each mark is hooked via the g_list. It also protects the groups private * data (i.e group limits). * mark->lock protects the marks attributes like its masks and flags. * Furthermore it protects the access to a reference of the group that the mark * is assigned to as well as the access to a reference of the inode/vfsmount * that is being watched by the mark. |
5444e2981
|
33 |
* |
04662cab5
|
34 35 |
* mark->connector->lock protects the list of marks anchored inside an * inode / vfsmount and each mark is hooked via the i_list. |
5444e2981
|
36 |
* |
04662cab5
|
37 38 |
* A list of notification marks relating to inode / mnt is contained in * fsnotify_mark_connector. That structure is alive as long as there are any |
6b3f05d24
|
39 40 41 42 43 44 45 |
* marks in the list and is also protected by fsnotify_mark_srcu. A mark gets * detached from fsnotify_mark_connector when last reference to the mark is * dropped. Thus having mark reference is enough to protect mark->connector * pointer and to make sure fsnotify_mark_connector cannot disappear. Also * because we remove mark from g_list before dropping mark reference associated * with that, any mark found through g_list is guaranteed to have * mark->connector set until we drop group->mark_mutex. |
5444e2981
|
46 47 48 |
* * LIFETIME: * Inode marks survive between when they are added to an inode and when their |
c1f33073a
|
49 |
* refcnt==0. Marks are also protected by fsnotify_mark_srcu. |
5444e2981
|
50 51 52 53 54 55 56 |
* * The inode mark can be cleared for a number of different reasons including: * - The inode is unlinked for the last time. (fsnotify_inode_remove) * - The inode is being evicted from cache. (fsnotify_inode_delete) * - The fs the inode is on is unmounted. (fsnotify_inode_delete/fsnotify_unmount_inodes) * - Something explicitly requests that it be removed. (fsnotify_destroy_mark) * - The fsnotify_group associated with the mark is going away and all such marks |
2e37c6ca8
|
57 |
* need to be cleaned up. (fsnotify_clear_marks_by_group) |
5444e2981
|
58 |
* |
5444e2981
|
59 60 61 62 63 64 65 |
* This has the very interesting property of being able to run concurrently with * any (or all) other directions. */ #include <linux/fs.h> #include <linux/init.h> #include <linux/kernel.h> |
75c1be487
|
66 |
#include <linux/kthread.h> |
5444e2981
|
67 68 69 70 |
#include <linux/module.h> #include <linux/mutex.h> #include <linux/slab.h> #include <linux/spinlock.h> |
75c1be487
|
71 |
#include <linux/srcu.h> |
77115225a
|
72 |
#include <linux/ratelimit.h> |
5444e2981
|
73 |
|
60063497a
|
74 |
#include <linux/atomic.h> |
5444e2981
|
75 76 77 |
#include <linux/fsnotify_backend.h> #include "fsnotify.h" |
0918f1c30
|
78 |
#define FSNOTIFY_REAPER_DELAY (1) /* 1 jiffy */ |
75c1be487
|
79 |
struct srcu_struct fsnotify_mark_srcu; |
9dd813c15
|
80 |
struct kmem_cache *fsnotify_mark_connector_cachep; |
13d34ac6e
|
81 82 |
static DEFINE_SPINLOCK(destroy_lock); static LIST_HEAD(destroy_list); |
08991e83b
|
83 |
static struct fsnotify_mark_connector *connector_destroy_list; |
0918f1c30
|
84 |
|
35e481761
|
85 86 |
static void fsnotify_mark_destroy_workfn(struct work_struct *work); static DECLARE_DELAYED_WORK(reaper_work, fsnotify_mark_destroy_workfn); |
75c1be487
|
87 |
|
08991e83b
|
88 89 |
static void fsnotify_connector_destroy_workfn(struct work_struct *work); static DECLARE_WORK(connector_reaper_work, fsnotify_connector_destroy_workfn); |
5444e2981
|
90 91 |
void fsnotify_get_mark(struct fsnotify_mark *mark) { |
ab97f8732
|
92 93 |
WARN_ON_ONCE(!refcount_read(&mark->refcnt)); refcount_inc(&mark->refcnt); |
5444e2981
|
94 |
} |
3ac70bfcd
|
95 96 97 98 99 100 |
static __u32 *fsnotify_conn_mask_p(struct fsnotify_mark_connector *conn) { if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) return &fsnotify_conn_inode(conn)->i_fsnotify_mask; else if (conn->type == FSNOTIFY_OBJ_TYPE_VFSMOUNT) return &fsnotify_conn_mount(conn)->mnt_fsnotify_mask; |
1e6cb7239
|
101 102 |
else if (conn->type == FSNOTIFY_OBJ_TYPE_SB) return &fsnotify_conn_sb(conn)->s_fsnotify_mask; |
3ac70bfcd
|
103 104 105 106 107 108 109 110 111 112 |
return NULL; } __u32 fsnotify_conn_mask(struct fsnotify_mark_connector *conn) { if (WARN_ON(!fsnotify_valid_obj_type(conn->type))) return 0; return *fsnotify_conn_mask_p(conn); } |
a242677bb
|
113 |
static void __fsnotify_recalc_mask(struct fsnotify_mark_connector *conn) |
0809ab69a
|
114 115 116 |
{ u32 new_mask = 0; struct fsnotify_mark *mark; |
04662cab5
|
117 |
assert_spin_locked(&conn->lock); |
d3bc0fa84
|
118 119 120 |
/* We can get detached connector here when inode is getting unlinked. */ if (!fsnotify_valid_obj_type(conn->type)) return; |
6b3f05d24
|
121 122 123 124 |
hlist_for_each_entry(mark, &conn->list, obj_list) { if (mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED) new_mask |= mark->mask; } |
3ac70bfcd
|
125 |
*fsnotify_conn_mask_p(conn) = new_mask; |
a242677bb
|
126 127 128 129 |
} /* * Calculate mask of events for a list of marks. The caller must make sure |
36f10f55f
|
130 |
* connector and connector->obj cannot disappear under us. Callers achieve |
6b3f05d24
|
131 132 |
* this by holding a mark->lock or mark->group->mark_mutex for a mark on this * list. |
a242677bb
|
133 134 135 136 137 |
*/ void fsnotify_recalc_mask(struct fsnotify_mark_connector *conn) { if (!conn) return; |
04662cab5
|
138 |
spin_lock(&conn->lock); |
a242677bb
|
139 |
__fsnotify_recalc_mask(conn); |
04662cab5
|
140 |
spin_unlock(&conn->lock); |
d6f7b98bc
|
141 |
if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) |
36f10f55f
|
142 143 |
__fsnotify_update_child_dentry_flags( fsnotify_conn_inode(conn)); |
0809ab69a
|
144 |
} |
08991e83b
|
145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 |
/* Free all connectors queued for freeing once SRCU period ends */ static void fsnotify_connector_destroy_workfn(struct work_struct *work) { struct fsnotify_mark_connector *conn, *free; spin_lock(&destroy_lock); conn = connector_destroy_list; connector_destroy_list = NULL; spin_unlock(&destroy_lock); synchronize_srcu(&fsnotify_mark_srcu); while (conn) { free = conn; conn = conn->destroy_next; kmem_cache_free(fsnotify_mark_connector_cachep, free); } } |
721fb6fbf
|
162 163 164 |
static void *fsnotify_detach_connector_from_object( struct fsnotify_mark_connector *conn, unsigned int *type) |
08991e83b
|
165 166 |
{ struct inode *inode = NULL; |
721fb6fbf
|
167 |
*type = conn->type; |
36f10f55f
|
168 169 |
if (conn->type == FSNOTIFY_OBJ_TYPE_DETACHED) return NULL; |
d6f7b98bc
|
170 |
if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) { |
36f10f55f
|
171 |
inode = fsnotify_conn_inode(conn); |
08991e83b
|
172 |
inode->i_fsnotify_mask = 0; |
721fb6fbf
|
173 |
atomic_long_inc(&inode->i_sb->s_fsnotify_inode_refs); |
d6f7b98bc
|
174 |
} else if (conn->type == FSNOTIFY_OBJ_TYPE_VFSMOUNT) { |
36f10f55f
|
175 |
fsnotify_conn_mount(conn)->mnt_fsnotify_mask = 0; |
1e6cb7239
|
176 177 |
} else if (conn->type == FSNOTIFY_OBJ_TYPE_SB) { fsnotify_conn_sb(conn)->s_fsnotify_mask = 0; |
08991e83b
|
178 |
} |
36f10f55f
|
179 180 181 |
rcu_assign_pointer(*(conn->obj), NULL); conn->obj = NULL; conn->type = FSNOTIFY_OBJ_TYPE_DETACHED; |
08991e83b
|
182 183 |
return inode; } |
6b3f05d24
|
184 185 |
static void fsnotify_final_mark_destroy(struct fsnotify_mark *mark) { |
054c636e5
|
186 187 188 189 190 191 |
struct fsnotify_group *group = mark->group; if (WARN_ON_ONCE(!group)) return; group->ops->free_mark(mark); fsnotify_put_group(group); |
6b3f05d24
|
192 |
} |
721fb6fbf
|
193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 |
/* Drop object reference originally held by a connector */ static void fsnotify_drop_object(unsigned int type, void *objp) { struct inode *inode; struct super_block *sb; if (!objp) return; /* Currently only inode references are passed to be dropped */ if (WARN_ON_ONCE(type != FSNOTIFY_OBJ_TYPE_INODE)) return; inode = objp; sb = inode->i_sb; iput(inode); if (atomic_long_dec_and_test(&sb->s_fsnotify_inode_refs)) wake_up_var(&sb->s_fsnotify_inode_refs); } |
6b3f05d24
|
210 |
void fsnotify_put_mark(struct fsnotify_mark *mark) |
8212a6097
|
211 |
{ |
b1da6a518
|
212 |
struct fsnotify_mark_connector *conn = READ_ONCE(mark->connector); |
721fb6fbf
|
213 214 |
void *objp = NULL; unsigned int type = FSNOTIFY_OBJ_TYPE_DETACHED; |
08991e83b
|
215 |
bool free_conn = false; |
8212a6097
|
216 |
|
6b3f05d24
|
217 |
/* Catch marks that were actually never attached to object */ |
b1da6a518
|
218 |
if (!conn) { |
ab97f8732
|
219 |
if (refcount_dec_and_test(&mark->refcnt)) |
6b3f05d24
|
220 221 222 223 224 225 226 227 |
fsnotify_final_mark_destroy(mark); return; } /* * We have to be careful so that traversals of obj_list under lock can * safely grab mark reference. */ |
b1da6a518
|
228 |
if (!refcount_dec_and_lock(&mark->refcnt, &conn->lock)) |
6b3f05d24
|
229 |
return; |
8212a6097
|
230 231 |
hlist_del_init_rcu(&mark->obj_list); if (hlist_empty(&conn->list)) { |
721fb6fbf
|
232 |
objp = fsnotify_detach_connector_from_object(conn, &type); |
08991e83b
|
233 234 235 |
free_conn = true; } else { __fsnotify_recalc_mask(conn); |
8212a6097
|
236 |
} |
b1da6a518
|
237 |
WRITE_ONCE(mark->connector, NULL); |
04662cab5
|
238 |
spin_unlock(&conn->lock); |
8212a6097
|
239 |
|
721fb6fbf
|
240 |
fsnotify_drop_object(type, objp); |
6b3f05d24
|
241 |
|
08991e83b
|
242 243 244 245 246 247 248 |
if (free_conn) { spin_lock(&destroy_lock); conn->destroy_next = connector_destroy_list; connector_destroy_list = conn; spin_unlock(&destroy_lock); queue_work(system_unbound_wq, &connector_reaper_work); } |
6b3f05d24
|
249 250 251 252 253 254 255 256 257 258 259 |
/* * Note that we didn't update flags telling whether inode cares about * what's happening with children. We update these flags from * __fsnotify_parent() lazily when next event happens on one of our * children. */ spin_lock(&destroy_lock); list_add(&mark->g_list, &destroy_list); spin_unlock(&destroy_lock); queue_delayed_work(system_unbound_wq, &reaper_work, FSNOTIFY_REAPER_DELAY); |
8212a6097
|
260 |
} |
b72679ee8
|
261 |
EXPORT_SYMBOL_GPL(fsnotify_put_mark); |
8212a6097
|
262 |
|
24c20305c
|
263 264 265 266 267 268 269 270 |
/* * Get mark reference when we found the mark via lockless traversal of object * list. Mark can be already removed from the list by now and on its way to be * destroyed once SRCU period ends. * * Also pin the group so it doesn't disappear under us. */ static bool fsnotify_get_mark_safe(struct fsnotify_mark *mark) |
abc77577a
|
271 |
{ |
24c20305c
|
272 273 |
if (!mark) return true; |
abc77577a
|
274 |
|
ab97f8732
|
275 |
if (refcount_inc_not_zero(&mark->refcnt)) { |
9a31d7ad9
|
276 277 278 279 280 281 282 283 284 285 |
spin_lock(&mark->lock); if (mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED) { /* mark is attached, group is still alive then */ atomic_inc(&mark->group->user_waits); spin_unlock(&mark->lock); return true; } spin_unlock(&mark->lock); fsnotify_put_mark(mark); } |
24c20305c
|
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 |
return false; } /* * Puts marks and wakes up group destruction if necessary. * * Pairs with fsnotify_get_mark_safe() */ static void fsnotify_put_mark_wake(struct fsnotify_mark *mark) { if (mark) { struct fsnotify_group *group = mark->group; fsnotify_put_mark(mark); /* * We abuse notification_waitq on group shutdown for waiting for * all marks pinned when waiting for userspace. */ if (atomic_dec_and_test(&group->user_waits) && group->shutdown) wake_up(&group->notification_waitq); |
abc77577a
|
306 |
} |
24c20305c
|
307 308 309 |
} bool fsnotify_prepare_user_wait(struct fsnotify_iter_info *iter_info) |
00e0afb65
|
310 |
__releases(&fsnotify_mark_srcu) |
24c20305c
|
311 |
{ |
47d9c7cc4
|
312 313 314 315 |
int type; fsnotify_foreach_obj_type(type) { /* This can fail if mark is being removed */ |
00e0afb65
|
316 317 |
if (!fsnotify_get_mark_safe(iter_info->marks[type])) { __release(&fsnotify_mark_srcu); |
47d9c7cc4
|
318 |
goto fail; |
00e0afb65
|
319 |
} |
abc77577a
|
320 321 322 323 324 325 326 327 328 329 |
} /* * Now that both marks are pinned by refcount in the inode / vfsmount * lists, we can drop SRCU lock, and safely resume the list iteration * once userspace returns. */ srcu_read_unlock(&fsnotify_mark_srcu, iter_info->srcu_idx); return true; |
47d9c7cc4
|
330 331 332 333 334 |
fail: for (type--; type >= 0; type--) fsnotify_put_mark_wake(iter_info->marks[type]); return false; |
abc77577a
|
335 336 337 |
} void fsnotify_finish_user_wait(struct fsnotify_iter_info *iter_info) |
00e0afb65
|
338 |
__acquires(&fsnotify_mark_srcu) |
abc77577a
|
339 |
{ |
47d9c7cc4
|
340 |
int type; |
abc77577a
|
341 |
iter_info->srcu_idx = srcu_read_lock(&fsnotify_mark_srcu); |
47d9c7cc4
|
342 343 |
fsnotify_foreach_obj_type(type) fsnotify_put_mark_wake(iter_info->marks[type]); |
abc77577a
|
344 |
} |
5444e2981
|
345 |
/* |
6b3f05d24
|
346 347 348 349 350 |
* Mark mark as detached, remove it from group list. Mark still stays in object * list until its last reference is dropped. Note that we rely on mark being * removed from group list before corresponding reference to it is dropped. In * particular we rely on mark->connector being valid while we hold * group->mark_mutex if we found the mark through g_list. |
4712e722f
|
351 |
* |
11375145a
|
352 353 |
* Must be called with group->mark_mutex held. The caller must either hold * reference to the mark or be protected by fsnotify_mark_srcu. |
5444e2981
|
354 |
*/ |
4712e722f
|
355 |
void fsnotify_detach_mark(struct fsnotify_mark *mark) |
5444e2981
|
356 |
{ |
4712e722f
|
357 |
struct fsnotify_group *group = mark->group; |
5444e2981
|
358 |
|
11375145a
|
359 360 |
WARN_ON_ONCE(!mutex_is_locked(&group->mark_mutex)); WARN_ON_ONCE(!srcu_read_lock_held(&fsnotify_mark_srcu) && |
ab97f8732
|
361 |
refcount_read(&mark->refcnt) < 1 + |
11375145a
|
362 |
!!(mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)); |
d5a335b84
|
363 |
|
104d06f08
|
364 |
spin_lock(&mark->lock); |
700307a29
|
365 |
/* something else already called this function on this mark */ |
4712e722f
|
366 |
if (!(mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)) { |
5444e2981
|
367 |
spin_unlock(&mark->lock); |
e2a29943e
|
368 |
return; |
5444e2981
|
369 |
} |
4712e722f
|
370 |
mark->flags &= ~FSNOTIFY_MARK_FLAG_ATTACHED; |
5444e2981
|
371 |
list_del_init(&mark->g_list); |
5444e2981
|
372 |
spin_unlock(&mark->lock); |
d5a335b84
|
373 |
|
4712e722f
|
374 |
atomic_dec(&group->num_marks); |
11375145a
|
375 376 377 |
/* Drop mark reference acquired in fsnotify_add_mark_locked() */ fsnotify_put_mark(mark); |
4712e722f
|
378 379 380 |
} /* |
11375145a
|
381 382 383 |
* Free fsnotify mark. The mark is actually only marked as being freed. The * freeing is actually happening only once last reference to the mark is * dropped from a workqueue which first waits for srcu period end. |
35e481761
|
384 |
* |
11375145a
|
385 386 |
* Caller must have a reference to the mark or be protected by * fsnotify_mark_srcu. |
4712e722f
|
387 |
*/ |
11375145a
|
388 |
void fsnotify_free_mark(struct fsnotify_mark *mark) |
4712e722f
|
389 390 391 392 393 394 395 |
{ struct fsnotify_group *group = mark->group; spin_lock(&mark->lock); /* something else already called this function on this mark */ if (!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE)) { spin_unlock(&mark->lock); |
11375145a
|
396 |
return; |
4712e722f
|
397 398 399 |
} mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE; spin_unlock(&mark->lock); |
5444e2981
|
400 |
|
d725e66c0
|
401 402 403 404 405 406 407 |
/* * Some groups like to know that marks are being freed. This is a * callback to the group function to let it know that this mark * is being freed. */ if (group->ops->freeing_mark) group->ops->freeing_mark(mark, group); |
d5a335b84
|
408 409 410 411 412 |
} void fsnotify_destroy_mark(struct fsnotify_mark *mark, struct fsnotify_group *group) { |
6960b0d90
|
413 |
mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); |
4712e722f
|
414 |
fsnotify_detach_mark(mark); |
d5a335b84
|
415 |
mutex_unlock(&group->mark_mutex); |
4712e722f
|
416 |
fsnotify_free_mark(mark); |
5444e2981
|
417 |
} |
b72679ee8
|
418 |
EXPORT_SYMBOL_GPL(fsnotify_destroy_mark); |
5444e2981
|
419 420 |
/* |
8edc6e168
|
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 |
* Sorting function for lists of fsnotify marks. * * Fanotify supports different notification classes (reflected as priority of * notification group). Events shall be passed to notification groups in * decreasing priority order. To achieve this marks in notification lists for * inodes and vfsmounts are sorted so that priorities of corresponding groups * are descending. * * Furthermore correct handling of the ignore mask requires processing inode * and vfsmount marks of each group together. Using the group address as * further sort criterion provides a unique sorting order and thus we can * merge inode and vfsmount lists of marks in linear time and find groups * present in both lists. * * A return value of 1 signifies that b has priority over a. * A return value of 0 signifies that the two marks have to be handled together. * A return value of -1 signifies that a has priority over b. */ int fsnotify_compare_groups(struct fsnotify_group *a, struct fsnotify_group *b) { if (a == b) return 0; if (!a) return 1; if (!b) return -1; if (a->priority < b->priority) return 1; if (a->priority > b->priority) return -1; if (a < b) return 1; return -1; } |
9b6e54345
|
455 |
static int fsnotify_attach_connector_to_object(fsnotify_connp_t *connp, |
77115225a
|
456 457 |
unsigned int type, __kernel_fsid_t *fsid) |
9dd813c15
|
458 |
{ |
b812a9f58
|
459 |
struct inode *inode = NULL; |
9dd813c15
|
460 |
struct fsnotify_mark_connector *conn; |
755b5bc68
|
461 |
conn = kmem_cache_alloc(fsnotify_mark_connector_cachep, GFP_KERNEL); |
9dd813c15
|
462 463 |
if (!conn) return -ENOMEM; |
04662cab5
|
464 |
spin_lock_init(&conn->lock); |
9dd813c15
|
465 |
INIT_HLIST_HEAD(&conn->list); |
b812a9f58
|
466 |
conn->type = type; |
36f10f55f
|
467 |
conn->obj = connp; |
77115225a
|
468 |
/* Cache fsid of filesystem containing the object */ |
c285a2f01
|
469 |
if (fsid) { |
77115225a
|
470 |
conn->fsid = *fsid; |
c285a2f01
|
471 472 |
conn->flags = FSNOTIFY_CONN_FLAG_HAS_FSID; } else { |
77115225a
|
473 |
conn->fsid.val[0] = conn->fsid.val[1] = 0; |
c285a2f01
|
474 475 |
conn->flags = 0; } |
b812a9f58
|
476 |
if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) |
36f10f55f
|
477 |
inode = igrab(fsnotify_conn_inode(conn)); |
9dd813c15
|
478 |
/* |
04662cab5
|
479 480 |
* cmpxchg() provides the barrier so that readers of *connp can see * only initialized structure |
9dd813c15
|
481 |
*/ |
04662cab5
|
482 483 |
if (cmpxchg(connp, NULL, conn)) { /* Someone else created list structure for us */ |
08991e83b
|
484 485 |
if (inode) iput(inode); |
755b5bc68
|
486 |
kmem_cache_free(fsnotify_mark_connector_cachep, conn); |
04662cab5
|
487 |
} |
9dd813c15
|
488 489 490 491 492 |
return 0; } /* |
08991e83b
|
493 494 495 496 497 498 |
* Get mark connector, make sure it is alive and return with its lock held. * This is for users that get connector pointer from inode or mount. Users that * hold reference to a mark on the list may directly lock connector->lock as * they are sure list cannot go away under them. */ static struct fsnotify_mark_connector *fsnotify_grab_connector( |
9b6e54345
|
499 |
fsnotify_connp_t *connp) |
08991e83b
|
500 501 502 503 504 505 506 507 508 |
{ struct fsnotify_mark_connector *conn; int idx; idx = srcu_read_lock(&fsnotify_mark_srcu); conn = srcu_dereference(*connp, &fsnotify_mark_srcu); if (!conn) goto out; spin_lock(&conn->lock); |
d6f7b98bc
|
509 |
if (conn->type == FSNOTIFY_OBJ_TYPE_DETACHED) { |
08991e83b
|
510 511 512 513 514 515 516 517 518 519 |
spin_unlock(&conn->lock); srcu_read_unlock(&fsnotify_mark_srcu, idx); return NULL; } out: srcu_read_unlock(&fsnotify_mark_srcu, idx); return conn; } /* |
9dd813c15
|
520 521 522 523 524 |
* Add mark into proper place in given list of marks. These marks may be used * for the fsnotify backend to determine which event types should be delivered * to which group and for which inodes. These marks are ordered according to * priority, highest number first, and then by the group's location in memory. */ |
755b5bc68
|
525 |
static int fsnotify_add_mark_list(struct fsnotify_mark *mark, |
b812a9f58
|
526 |
fsnotify_connp_t *connp, unsigned int type, |
77115225a
|
527 |
int allow_dups, __kernel_fsid_t *fsid) |
0809ab69a
|
528 529 |
{ struct fsnotify_mark *lmark, *last = NULL; |
9dd813c15
|
530 |
struct fsnotify_mark_connector *conn; |
0809ab69a
|
531 |
int cmp; |
755b5bc68
|
532 |
int err = 0; |
b812a9f58
|
533 |
if (WARN_ON(!fsnotify_valid_obj_type(type))) |
755b5bc68
|
534 |
return -EINVAL; |
77115225a
|
535 536 537 538 |
/* Backend is expected to check for zero fsid (e.g. tmpfs) */ if (fsid && WARN_ON_ONCE(!fsid->val[0] && !fsid->val[1])) return -ENODEV; |
08991e83b
|
539 540 541 542 543 |
restart: spin_lock(&mark->lock); conn = fsnotify_grab_connector(connp); if (!conn) { spin_unlock(&mark->lock); |
77115225a
|
544 |
err = fsnotify_attach_connector_to_object(connp, type, fsid); |
9dd813c15
|
545 546 |
if (err) return err; |
08991e83b
|
547 |
goto restart; |
c285a2f01
|
548 549 550 551 552 553 |
} else if (fsid && !(conn->flags & FSNOTIFY_CONN_FLAG_HAS_FSID)) { conn->fsid = *fsid; /* Pairs with smp_rmb() in fanotify_get_fsid() */ smp_wmb(); conn->flags |= FSNOTIFY_CONN_FLAG_HAS_FSID; } else if (fsid && (conn->flags & FSNOTIFY_CONN_FLAG_HAS_FSID) && |
77115225a
|
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 |
(fsid->val[0] != conn->fsid.val[0] || fsid->val[1] != conn->fsid.val[1])) { /* * Backend is expected to check for non uniform fsid * (e.g. btrfs), but maybe we missed something? * Only allow setting conn->fsid once to non zero fsid. * inotify and non-fid fanotify groups do not set nor test * conn->fsid. */ pr_warn_ratelimited("%s: fsid mismatch on object of type %u: " "%x.%x != %x.%x ", __func__, conn->type, fsid->val[0], fsid->val[1], conn->fsid.val[0], conn->fsid.val[1]); err = -EXDEV; goto out_err; |
9dd813c15
|
570 |
} |
0809ab69a
|
571 572 |
/* is mark the first mark? */ |
9dd813c15
|
573 574 |
if (hlist_empty(&conn->list)) { hlist_add_head_rcu(&mark->obj_list, &conn->list); |
86ffe245c
|
575 |
goto added; |
0809ab69a
|
576 577 578 |
} /* should mark be in the middle of the current list? */ |
9dd813c15
|
579 |
hlist_for_each_entry(lmark, &conn->list, obj_list) { |
0809ab69a
|
580 |
last = lmark; |
6b3f05d24
|
581 582 583 |
if ((lmark->group == mark->group) && (lmark->flags & FSNOTIFY_MARK_FLAG_ATTACHED) && !allow_dups) { |
755b5bc68
|
584 585 586 |
err = -EEXIST; goto out_err; } |
0809ab69a
|
587 588 589 590 |
cmp = fsnotify_compare_groups(lmark->group, mark->group); if (cmp >= 0) { hlist_add_before_rcu(&mark->obj_list, &lmark->obj_list); |
86ffe245c
|
591 |
goto added; |
0809ab69a
|
592 593 594 595 596 597 |
} } BUG_ON(last == NULL); /* mark should be the last entry. last is the current last entry */ hlist_add_behind_rcu(&mark->obj_list, &last->obj_list); |
86ffe245c
|
598 |
added: |
11a6f8e2d
|
599 600 601 602 603 |
/* * Since connector is attached to object using cmpxchg() we are * guaranteed that connector initialization is fully visible by anyone * seeing mark->connector set. */ |
b1da6a518
|
604 |
WRITE_ONCE(mark->connector, conn); |
755b5bc68
|
605 |
out_err: |
04662cab5
|
606 |
spin_unlock(&conn->lock); |
755b5bc68
|
607 608 |
spin_unlock(&mark->lock); return err; |
0809ab69a
|
609 |
} |
8edc6e168
|
610 |
/* |
5444e2981
|
611 612 613 614 |
* Attach an initialized mark to a given group and fs object. * These marks may be used for the fsnotify backend to determine which * event types should be delivered to which group. */ |
b812a9f58
|
615 616 |
int fsnotify_add_mark_locked(struct fsnotify_mark *mark, fsnotify_connp_t *connp, unsigned int type, |
77115225a
|
617 |
int allow_dups, __kernel_fsid_t *fsid) |
5444e2981
|
618 |
{ |
7b1293234
|
619 |
struct fsnotify_group *group = mark->group; |
5444e2981
|
620 |
int ret = 0; |
d5a335b84
|
621 |
BUG_ON(!mutex_is_locked(&group->mark_mutex)); |
5444e2981
|
622 623 |
/* |
5444e2981
|
624 |
* LOCKING ORDER!!!! |
986ab0980
|
625 |
* group->mark_mutex |
104d06f08
|
626 |
* mark->lock |
04662cab5
|
627 |
* mark->connector->lock |
5444e2981
|
628 |
*/ |
104d06f08
|
629 |
spin_lock(&mark->lock); |
4712e722f
|
630 |
mark->flags |= FSNOTIFY_MARK_FLAG_ALIVE | FSNOTIFY_MARK_FLAG_ATTACHED; |
700307a29
|
631 |
|
5444e2981
|
632 633 |
list_add(&mark->g_list, &group->marks_list); atomic_inc(&group->num_marks); |
6b3f05d24
|
634 |
fsnotify_get_mark(mark); /* for g_list */ |
5444e2981
|
635 |
spin_unlock(&mark->lock); |
77115225a
|
636 |
ret = fsnotify_add_mark_list(mark, connp, type, allow_dups, fsid); |
755b5bc68
|
637 638 |
if (ret) goto err; |
a242677bb
|
639 640 |
if (mark->mask) fsnotify_recalc_mask(mark->connector); |
5444e2981
|
641 642 643 |
return ret; err: |
9cf90cef3
|
644 |
spin_lock(&mark->lock); |
11375145a
|
645 646 |
mark->flags &= ~(FSNOTIFY_MARK_FLAG_ALIVE | FSNOTIFY_MARK_FLAG_ATTACHED); |
5444e2981
|
647 |
list_del_init(&mark->g_list); |
9cf90cef3
|
648 |
spin_unlock(&mark->lock); |
5444e2981
|
649 |
atomic_dec(&group->num_marks); |
5444e2981
|
650 |
|
11375145a
|
651 |
fsnotify_put_mark(mark); |
5444e2981
|
652 653 |
return ret; } |
b812a9f58
|
654 |
int fsnotify_add_mark(struct fsnotify_mark *mark, fsnotify_connp_t *connp, |
77115225a
|
655 |
unsigned int type, int allow_dups, __kernel_fsid_t *fsid) |
d5a335b84
|
656 657 |
{ int ret; |
7b1293234
|
658 |
struct fsnotify_group *group = mark->group; |
d5a335b84
|
659 |
mutex_lock(&group->mark_mutex); |
77115225a
|
660 |
ret = fsnotify_add_mark_locked(mark, connp, type, allow_dups, fsid); |
d5a335b84
|
661 662 663 |
mutex_unlock(&group->mark_mutex); return ret; } |
b72679ee8
|
664 |
EXPORT_SYMBOL_GPL(fsnotify_add_mark); |
d5a335b84
|
665 |
|
5444e2981
|
666 |
/* |
0809ab69a
|
667 668 669 |
* Given a list of marks, find the mark associated with given group. If found * take a reference to that mark and return it, else return NULL. */ |
9b6e54345
|
670 671 |
struct fsnotify_mark *fsnotify_find_mark(fsnotify_connp_t *connp, struct fsnotify_group *group) |
0809ab69a
|
672 |
{ |
08991e83b
|
673 |
struct fsnotify_mark_connector *conn; |
0809ab69a
|
674 |
struct fsnotify_mark *mark; |
08991e83b
|
675 |
conn = fsnotify_grab_connector(connp); |
9dd813c15
|
676 677 678 679 |
if (!conn) return NULL; hlist_for_each_entry(mark, &conn->list, obj_list) { |
6b3f05d24
|
680 681 |
if (mark->group == group && (mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)) { |
0809ab69a
|
682 |
fsnotify_get_mark(mark); |
04662cab5
|
683 |
spin_unlock(&conn->lock); |
0809ab69a
|
684 685 686 |
return mark; } } |
04662cab5
|
687 |
spin_unlock(&conn->lock); |
0809ab69a
|
688 689 |
return NULL; } |
b72679ee8
|
690 |
EXPORT_SYMBOL_GPL(fsnotify_find_mark); |
0809ab69a
|
691 |
|
d6f7b98bc
|
692 |
/* Clear any marks in a group with given type mask */ |
18f2e0d3a
|
693 |
void fsnotify_clear_marks_by_group(struct fsnotify_group *group, |
d6f7b98bc
|
694 |
unsigned int type_mask) |
5444e2981
|
695 696 |
{ struct fsnotify_mark *lmark, *mark; |
8f2f3eb59
|
697 |
LIST_HEAD(to_free); |
2e37c6ca8
|
698 |
struct list_head *head = &to_free; |
5444e2981
|
699 |
|
2e37c6ca8
|
700 |
/* Skip selection step if we want to clear all marks. */ |
d6f7b98bc
|
701 |
if (type_mask == FSNOTIFY_OBJ_ALL_TYPES_MASK) { |
2e37c6ca8
|
702 703 704 |
head = &group->marks_list; goto clear; } |
8f2f3eb59
|
705 706 707 708 709 710 711 712 713 |
/* * We have to be really careful here. Anytime we drop mark_mutex, e.g. * fsnotify_clear_marks_by_inode() can come and free marks. Even in our * to_free list so we have to use mark_mutex even when accessing that * list. And freeing mark requires us to drop mark_mutex. So we can * reliably free only the first mark in the list. That's why we first * move marks to free to to_free list in one go and then free marks in * to_free list one by one. */ |
6960b0d90
|
714 |
mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); |
5444e2981
|
715 |
list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) { |
d6f7b98bc
|
716 |
if ((1U << mark->connector->type) & type_mask) |
8f2f3eb59
|
717 |
list_move(&mark->g_list, &to_free); |
5444e2981
|
718 |
} |
986ab0980
|
719 |
mutex_unlock(&group->mark_mutex); |
8f2f3eb59
|
720 |
|
2e37c6ca8
|
721 |
clear: |
8f2f3eb59
|
722 723 |
while (1) { mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); |
2e37c6ca8
|
724 |
if (list_empty(head)) { |
8f2f3eb59
|
725 726 727 |
mutex_unlock(&group->mark_mutex); break; } |
2e37c6ca8
|
728 |
mark = list_first_entry(head, struct fsnotify_mark, g_list); |
8f2f3eb59
|
729 |
fsnotify_get_mark(mark); |
4712e722f
|
730 |
fsnotify_detach_mark(mark); |
8f2f3eb59
|
731 |
mutex_unlock(&group->mark_mutex); |
4712e722f
|
732 |
fsnotify_free_mark(mark); |
8f2f3eb59
|
733 734 |
fsnotify_put_mark(mark); } |
5444e2981
|
735 |
} |
9b6e54345
|
736 737 |
/* Destroy all marks attached to an object via connector */ void fsnotify_destroy_marks(fsnotify_connp_t *connp) |
0810b4f9f
|
738 |
{ |
08991e83b
|
739 |
struct fsnotify_mark_connector *conn; |
6b3f05d24
|
740 |
struct fsnotify_mark *mark, *old_mark = NULL; |
721fb6fbf
|
741 742 |
void *objp; unsigned int type; |
0810b4f9f
|
743 |
|
6b3f05d24
|
744 745 746 747 748 749 750 751 752 753 754 |
conn = fsnotify_grab_connector(connp); if (!conn) return; /* * We have to be careful since we can race with e.g. * fsnotify_clear_marks_by_group() and once we drop the conn->lock, the * list can get modified. However we are holding mark reference and * thus our mark cannot be removed from obj_list so we can continue * iteration after regaining conn->lock. */ hlist_for_each_entry(mark, &conn->list, obj_list) { |
0810b4f9f
|
755 |
fsnotify_get_mark(mark); |
04662cab5
|
756 |
spin_unlock(&conn->lock); |
6b3f05d24
|
757 758 759 |
if (old_mark) fsnotify_put_mark(old_mark); old_mark = mark; |
0810b4f9f
|
760 |
fsnotify_destroy_mark(mark, mark->group); |
6b3f05d24
|
761 |
spin_lock(&conn->lock); |
0810b4f9f
|
762 |
} |
6b3f05d24
|
763 764 765 766 767 |
/* * Detach list from object now so that we don't pin inode until all * mark references get dropped. It would lead to strange results such * as delaying inode deletion or blocking unmount. */ |
721fb6fbf
|
768 |
objp = fsnotify_detach_connector_from_object(conn, &type); |
6b3f05d24
|
769 770 771 |
spin_unlock(&conn->lock); if (old_mark) fsnotify_put_mark(old_mark); |
721fb6fbf
|
772 |
fsnotify_drop_object(type, objp); |
0810b4f9f
|
773 |
} |
5444e2981
|
774 775 776 777 |
/* * Nothing fancy, just initialize lists and locks and counters. */ void fsnotify_init_mark(struct fsnotify_mark *mark, |
054c636e5
|
778 |
struct fsnotify_group *group) |
5444e2981
|
779 |
{ |
ba643f04c
|
780 |
memset(mark, 0, sizeof(*mark)); |
5444e2981
|
781 |
spin_lock_init(&mark->lock); |
ab97f8732
|
782 |
refcount_set(&mark->refcnt, 1); |
7b1293234
|
783 784 |
fsnotify_get_group(group); mark->group = group; |
b1da6a518
|
785 |
WRITE_ONCE(mark->connector, NULL); |
5444e2981
|
786 |
} |
b72679ee8
|
787 |
EXPORT_SYMBOL_GPL(fsnotify_init_mark); |
13d34ac6e
|
788 |
|
35e481761
|
789 790 791 792 |
/* * Destroy all marks in destroy_list, waits for SRCU period to finish before * actually freeing marks. */ |
f09b04a03
|
793 |
static void fsnotify_mark_destroy_workfn(struct work_struct *work) |
13d34ac6e
|
794 795 796 |
{ struct fsnotify_mark *mark, *next; struct list_head private_destroy_list; |
0918f1c30
|
797 798 799 800 |
spin_lock(&destroy_lock); /* exchange the list head */ list_replace_init(&destroy_list, &private_destroy_list); spin_unlock(&destroy_lock); |
13d34ac6e
|
801 |
|
0918f1c30
|
802 |
synchronize_srcu(&fsnotify_mark_srcu); |
13d34ac6e
|
803 |
|
0918f1c30
|
804 805 |
list_for_each_entry_safe(mark, next, &private_destroy_list, g_list) { list_del_init(&mark->g_list); |
6b3f05d24
|
806 |
fsnotify_final_mark_destroy(mark); |
13d34ac6e
|
807 |
} |
13d34ac6e
|
808 |
} |
35e481761
|
809 |
|
f09b04a03
|
810 811 |
/* Wait for all marks queued for destruction to be actually destroyed */ void fsnotify_wait_marks_destroyed(void) |
35e481761
|
812 |
{ |
f09b04a03
|
813 |
flush_delayed_work(&reaper_work); |
35e481761
|
814 |
} |
b72679ee8
|
815 |
EXPORT_SYMBOL_GPL(fsnotify_wait_marks_destroyed); |