1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
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
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
|
// SPDX-License-Identifier: GPL-2.0-or-later
/*
* NET3: Garbage Collector For AF_UNIX sockets
*
* Garbage Collector:
* Copyright (C) Barak A. Pearlmutter.
*
* Chopped about by Alan Cox 22/3/96 to make it fit the AF_UNIX socket problem.
* If it doesn't work blame me, it worked when Barak sent it.
*
* Assumptions:
*
* - object w/ a bit
* - free list
*
* Current optimizations:
*
* - explicit stack instead of recursion
* - tail recurse on first born instead of immediate push/pop
* - we gather the stuff that should not be killed into tree
* and stack is just a path from root to the current pointer.
*
* Future optimizations:
*
* - don't just push entire root set; process in place
*
* Fixes:
* Alan Cox 07 Sept 1997 Vmalloc internal stack as needed.
* Cope with changing max_files.
* Al Viro 11 Oct 1998
* Graph may have cycles. That is, we can send the descriptor
* of foo to bar and vice versa. Current code chokes on that.
* Fix: move SCM_RIGHTS ones into the separate list and then
* skb_free() them all instead of doing explicit fput's.
* Another problem: since fput() may block somebody may
* create a new unix_socket when we are in the middle of sweep
* phase. Fix: revert the logic wrt MARKED. Mark everything
* upon the beginning and unmark non-junk ones.
*
* [12 Oct 1998] AAARGH! New code purges all SCM_RIGHTS
* sent to connect()'ed but still not accept()'ed sockets.
* Fixed. Old code had slightly different problem here:
* extra fput() in situation when we passed the descriptor via
* such socket and closed it (descriptor). That would happen on
* each unix_gc() until the accept(). Since the struct file in
* question would go to the free list and might be reused...
* That might be the reason of random oopses on filp_close()
* in unrelated processes.
*
* AV 28 Feb 1999
* Kill the explicit allocation of stack. Now we keep the tree
* with root in dummy + pointer (gc_current) to one of the nodes.
* Stack is represented as path from gc_current to dummy. Unmark
* now means "add to tree". Push == "make it a son of gc_current".
* Pop == "move gc_current to parent". We keep only pointers to
* parents (->gc_tree).
* AV 1 Mar 1999
* Damn. Added missing check for ->dead in listen queues scanning.
*
* Miklos Szeredi 25 Jun 2007
* Reimplement with a cycle collecting algorithm. This should
* solve several problems with the previous code, like being racy
* wrt receive and holding up unrelated socket operations.
*/
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/un.h>
#include <linux/net.h>
#include <linux/fs.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <linux/file.h>
#include <linux/proc_fs.h>
#include <linux/mutex.h>
#include <linux/wait.h>
#include <net/sock.h>
#include <net/af_unix.h>
#include <net/scm.h>
#include <net/tcp_states.h>
struct unix_sock *unix_get_socket(struct file *filp)
{
struct inode *inode = file_inode(filp);
/* Socket ? */
if (S_ISSOCK(inode->i_mode) && !(filp->f_mode & FMODE_PATH)) {
struct socket *sock = SOCKET_I(inode);
const struct proto_ops *ops;
struct sock *sk = sock->sk;
ops = READ_ONCE(sock->ops);
/* PF_UNIX ? */
if (sk && ops && ops->family == PF_UNIX)
return unix_sk(sk);
}
return NULL;
}
static LIST_HEAD(unix_unvisited_vertices);
enum unix_vertex_index {
UNIX_VERTEX_INDEX_UNVISITED,
UNIX_VERTEX_INDEX_START,
};
static void unix_add_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
{
struct unix_vertex *vertex = edge->predecessor->vertex;
if (!vertex) {
vertex = list_first_entry(&fpl->vertices, typeof(*vertex), entry);
vertex->out_degree = 0;
INIT_LIST_HEAD(&vertex->edges);
list_move_tail(&vertex->entry, &unix_unvisited_vertices);
edge->predecessor->vertex = vertex;
}
vertex->out_degree++;
list_add_tail(&edge->vertex_entry, &vertex->edges);
}
static void unix_del_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
{
struct unix_vertex *vertex = edge->predecessor->vertex;
list_del(&edge->vertex_entry);
vertex->out_degree--;
if (!vertex->out_degree) {
edge->predecessor->vertex = NULL;
list_move_tail(&vertex->entry, &fpl->vertices);
}
}
static void unix_free_vertices(struct scm_fp_list *fpl)
{
struct unix_vertex *vertex, *next_vertex;
list_for_each_entry_safe(vertex, next_vertex, &fpl->vertices, entry) {
list_del(&vertex->entry);
kfree(vertex);
}
}
DEFINE_SPINLOCK(unix_gc_lock);
unsigned int unix_tot_inflight;
void unix_add_edges(struct scm_fp_list *fpl, struct unix_sock *receiver)
{
int i = 0, j = 0;
spin_lock(&unix_gc_lock);
if (!fpl->count_unix)
goto out;
do {
struct unix_sock *inflight = unix_get_socket(fpl->fp[j++]);
struct unix_edge *edge;
if (!inflight)
continue;
edge = fpl->edges + i++;
edge->predecessor = inflight;
edge->successor = receiver;
unix_add_edge(fpl, edge);
} while (i < fpl->count_unix);
WRITE_ONCE(unix_tot_inflight, unix_tot_inflight + fpl->count_unix);
out:
WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight + fpl->count);
spin_unlock(&unix_gc_lock);
fpl->inflight = true;
unix_free_vertices(fpl);
}
void unix_del_edges(struct scm_fp_list *fpl)
{
int i = 0;
spin_lock(&unix_gc_lock);
if (!fpl->count_unix)
goto out;
do {
struct unix_edge *edge = fpl->edges + i++;
unix_del_edge(fpl, edge);
} while (i < fpl->count_unix);
WRITE_ONCE(unix_tot_inflight, unix_tot_inflight - fpl->count_unix);
out:
WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight - fpl->count);
spin_unlock(&unix_gc_lock);
fpl->inflight = false;
}
int unix_prepare_fpl(struct scm_fp_list *fpl)
{
struct unix_vertex *vertex;
int i;
if (!fpl->count_unix)
return 0;
for (i = 0; i < fpl->count_unix; i++) {
vertex = kmalloc(sizeof(*vertex), GFP_KERNEL);
if (!vertex)
goto err;
list_add(&vertex->entry, &fpl->vertices);
}
fpl->edges = kvmalloc_array(fpl->count_unix, sizeof(*fpl->edges),
GFP_KERNEL_ACCOUNT);
if (!fpl->edges)
goto err;
return 0;
err:
unix_free_vertices(fpl);
return -ENOMEM;
}
void unix_destroy_fpl(struct scm_fp_list *fpl)
{
if (fpl->inflight)
unix_del_edges(fpl);
kvfree(fpl->edges);
unix_free_vertices(fpl);
}
static LIST_HEAD(unix_visited_vertices);
static void __unix_walk_scc(struct unix_vertex *vertex)
{
unsigned long index = UNIX_VERTEX_INDEX_START;
struct unix_edge *edge;
LIST_HEAD(edge_stack);
next_vertex:
vertex->index = index;
index++;
/* Explore neighbour vertices (receivers of the current vertex's fd). */
list_for_each_entry(edge, &vertex->edges, vertex_entry) {
struct unix_vertex *next_vertex = edge->successor->vertex;
if (!next_vertex)
continue;
if (next_vertex->index == UNIX_VERTEX_INDEX_UNVISITED) {
/* Iterative deepening depth first search
*
* 1. Push a forward edge to edge_stack and set
* the successor to vertex for the next iteration.
*/
list_add(&edge->stack_entry, &edge_stack);
vertex = next_vertex;
goto next_vertex;
/* 2. Pop the edge directed to the current vertex
* and restore the ancestor for backtracking.
*/
prev_vertex:
edge = list_first_entry(&edge_stack, typeof(*edge), stack_entry);
list_del_init(&edge->stack_entry);
vertex = edge->predecessor->vertex;
}
}
/* Don't restart DFS from this vertex in unix_walk_scc(). */
list_move_tail(&vertex->entry, &unix_visited_vertices);
/* Need backtracking ? */
if (!list_empty(&edge_stack))
goto prev_vertex;
}
static void unix_walk_scc(void)
{
struct unix_vertex *vertex;
list_for_each_entry(vertex, &unix_unvisited_vertices, entry)
vertex->index = UNIX_VERTEX_INDEX_UNVISITED;
/* Visit every vertex exactly once.
* __unix_walk_scc() moves visited vertices to unix_visited_vertices.
*/
while (!list_empty(&unix_unvisited_vertices)) {
vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry);
__unix_walk_scc(vertex);
}
list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices);
}
static LIST_HEAD(gc_candidates);
static LIST_HEAD(gc_inflight_list);
/* Keep the number of times in flight count for the file
* descriptor if it is for an AF_UNIX socket.
*/
void unix_inflight(struct user_struct *user, struct file *filp)
{
struct unix_sock *u = unix_get_socket(filp);
spin_lock(&unix_gc_lock);
if (u) {
if (!u->inflight) {
WARN_ON_ONCE(!list_empty(&u->link));
list_add_tail(&u->link, &gc_inflight_list);
} else {
WARN_ON_ONCE(list_empty(&u->link));
}
u->inflight++;
}
spin_unlock(&unix_gc_lock);
}
void unix_notinflight(struct user_struct *user, struct file *filp)
{
struct unix_sock *u = unix_get_socket(filp);
spin_lock(&unix_gc_lock);
if (u) {
WARN_ON_ONCE(!u->inflight);
WARN_ON_ONCE(list_empty(&u->link));
u->inflight--;
if (!u->inflight)
list_del_init(&u->link);
}
spin_unlock(&unix_gc_lock);
}
static void scan_inflight(struct sock *x, void (*func)(struct unix_sock *),
struct sk_buff_head *hitlist)
{
struct sk_buff *skb;
struct sk_buff *next;
spin_lock(&x->sk_receive_queue.lock);
skb_queue_walk_safe(&x->sk_receive_queue, skb, next) {
/* Do we have file descriptors ? */
if (UNIXCB(skb).fp) {
bool hit = false;
/* Process the descriptors of this socket */
int nfd = UNIXCB(skb).fp->count;
struct file **fp = UNIXCB(skb).fp->fp;
while (nfd--) {
/* Get the socket the fd matches if it indeed does so */
struct unix_sock *u = unix_get_socket(*fp++);
/* Ignore non-candidates, they could have been added
* to the queues after starting the garbage collection
*/
if (u && test_bit(UNIX_GC_CANDIDATE, &u->gc_flags)) {
hit = true;
func(u);
}
}
if (hit && hitlist != NULL) {
__skb_unlink(skb, &x->sk_receive_queue);
__skb_queue_tail(hitlist, skb);
}
}
}
spin_unlock(&x->sk_receive_queue.lock);
}
static void scan_children(struct sock *x, void (*func)(struct unix_sock *),
struct sk_buff_head *hitlist)
{
if (x->sk_state != TCP_LISTEN) {
scan_inflight(x, func, hitlist);
} else {
struct sk_buff *skb;
struct sk_buff *next;
struct unix_sock *u;
LIST_HEAD(embryos);
/* For a listening socket collect the queued embryos
* and perform a scan on them as well.
*/
spin_lock(&x->sk_receive_queue.lock);
skb_queue_walk_safe(&x->sk_receive_queue, skb, next) {
u = unix_sk(skb->sk);
/* An embryo cannot be in-flight, so it's safe
* to use the list link.
*/
WARN_ON_ONCE(!list_empty(&u->link));
list_add_tail(&u->link, &embryos);
}
spin_unlock(&x->sk_receive_queue.lock);
while (!list_empty(&embryos)) {
u = list_entry(embryos.next, struct unix_sock, link);
scan_inflight(&u->sk, func, hitlist);
list_del_init(&u->link);
}
}
}
static void dec_inflight(struct unix_sock *usk)
{
usk->inflight--;
}
static void inc_inflight(struct unix_sock *usk)
{
usk->inflight++;
}
static void inc_inflight_move_tail(struct unix_sock *u)
{
u->inflight++;
/* If this still might be part of a cycle, move it to the end
* of the list, so that it's checked even if it was already
* passed over
*/
if (test_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags))
list_move_tail(&u->link, &gc_candidates);
}
static bool gc_in_progress;
static void __unix_gc(struct work_struct *work)
{
struct sk_buff_head hitlist;
struct unix_sock *u, *next;
LIST_HEAD(not_cycle_list);
struct list_head cursor;
spin_lock(&unix_gc_lock);
unix_walk_scc();
/* First, select candidates for garbage collection. Only
* in-flight sockets are considered, and from those only ones
* which don't have any external reference.
*
* Holding unix_gc_lock will protect these candidates from
* being detached, and hence from gaining an external
* reference. Since there are no possible receivers, all
* buffers currently on the candidates' queues stay there
* during the garbage collection.
*
* We also know that no new candidate can be added onto the
* receive queues. Other, non candidate sockets _can_ be
* added to queue, so we must make sure only to touch
* candidates.
*/
list_for_each_entry_safe(u, next, &gc_inflight_list, link) {
long total_refs;
total_refs = file_count(u->sk.sk_socket->file);
WARN_ON_ONCE(!u->inflight);
WARN_ON_ONCE(total_refs < u->inflight);
if (total_refs == u->inflight) {
list_move_tail(&u->link, &gc_candidates);
__set_bit(UNIX_GC_CANDIDATE, &u->gc_flags);
__set_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags);
}
}
/* Now remove all internal in-flight reference to children of
* the candidates.
*/
list_for_each_entry(u, &gc_candidates, link)
scan_children(&u->sk, dec_inflight, NULL);
/* Restore the references for children of all candidates,
* which have remaining references. Do this recursively, so
* only those remain, which form cyclic references.
*
* Use a "cursor" link, to make the list traversal safe, even
* though elements might be moved about.
*/
list_add(&cursor, &gc_candidates);
while (cursor.next != &gc_candidates) {
u = list_entry(cursor.next, struct unix_sock, link);
/* Move cursor to after the current position. */
list_move(&cursor, &u->link);
if (u->inflight) {
list_move_tail(&u->link, ¬_cycle_list);
__clear_bit(UNIX_GC_MAYBE_CYCLE, &u->gc_flags);
scan_children(&u->sk, inc_inflight_move_tail, NULL);
}
}
list_del(&cursor);
/* Now gc_candidates contains only garbage. Restore original
* inflight counters for these as well, and remove the skbuffs
* which are creating the cycle(s).
*/
skb_queue_head_init(&hitlist);
list_for_each_entry(u, &gc_candidates, link) {
scan_children(&u->sk, inc_inflight, &hitlist);
#if IS_ENABLED(CONFIG_AF_UNIX_OOB)
if (u->oob_skb) {
kfree_skb(u->oob_skb);
u->oob_skb = NULL;
}
#endif
}
/* not_cycle_list contains those sockets which do not make up a
* cycle. Restore these to the inflight list.
*/
while (!list_empty(¬_cycle_list)) {
u = list_entry(not_cycle_list.next, struct unix_sock, link);
__clear_bit(UNIX_GC_CANDIDATE, &u->gc_flags);
list_move_tail(&u->link, &gc_inflight_list);
}
spin_unlock(&unix_gc_lock);
/* Here we are. Hitlist is filled. Die. */
__skb_queue_purge(&hitlist);
spin_lock(&unix_gc_lock);
/* All candidates should have been detached by now. */
WARN_ON_ONCE(!list_empty(&gc_candidates));
/* Paired with READ_ONCE() in wait_for_unix_gc(). */
WRITE_ONCE(gc_in_progress, false);
spin_unlock(&unix_gc_lock);
}
static DECLARE_WORK(unix_gc_work, __unix_gc);
void unix_gc(void)
{
WRITE_ONCE(gc_in_progress, true);
queue_work(system_unbound_wq, &unix_gc_work);
}
#define UNIX_INFLIGHT_TRIGGER_GC 16000
#define UNIX_INFLIGHT_SANE_USER (SCM_MAX_FD * 8)
void wait_for_unix_gc(struct scm_fp_list *fpl)
{
/* If number of inflight sockets is insane,
* force a garbage collect right now.
*
* Paired with the WRITE_ONCE() in unix_inflight(),
* unix_notinflight(), and __unix_gc().
*/
if (READ_ONCE(unix_tot_inflight) > UNIX_INFLIGHT_TRIGGER_GC &&
!READ_ONCE(gc_in_progress))
unix_gc();
/* Penalise users who want to send AF_UNIX sockets
* but whose sockets have not been received yet.
*/
if (!fpl || !fpl->count_unix ||
READ_ONCE(fpl->user->unix_inflight) < UNIX_INFLIGHT_SANE_USER)
return;
if (READ_ONCE(gc_in_progress))
flush_work(&unix_gc_work);
}
|