xref: /linux/net/unix/garbage.c (revision 3f1c07fc21c68bd3bd2df9d2c9441f6485e934d9)
1 // SPDX-License-Identifier: GPL-2.0-or-later
2 /*
3  * NET3:	Garbage Collector For AF_UNIX sockets
4  *
5  * Garbage Collector:
6  *	Copyright (C) Barak A. Pearlmutter.
7  *
8  * Chopped about by Alan Cox 22/3/96 to make it fit the AF_UNIX socket problem.
9  * If it doesn't work blame me, it worked when Barak sent it.
10  *
11  * Assumptions:
12  *
13  *  - object w/ a bit
14  *  - free list
15  *
16  * Current optimizations:
17  *
18  *  - explicit stack instead of recursion
19  *  - tail recurse on first born instead of immediate push/pop
20  *  - we gather the stuff that should not be killed into tree
21  *    and stack is just a path from root to the current pointer.
22  *
23  *  Future optimizations:
24  *
25  *  - don't just push entire root set; process in place
26  *
27  *  Fixes:
28  *	Alan Cox	07 Sept	1997	Vmalloc internal stack as needed.
29  *					Cope with changing max_files.
30  *	Al Viro		11 Oct 1998
31  *		Graph may have cycles. That is, we can send the descriptor
32  *		of foo to bar and vice versa. Current code chokes on that.
33  *		Fix: move SCM_RIGHTS ones into the separate list and then
34  *		skb_free() them all instead of doing explicit fput's.
35  *		Another problem: since fput() may block somebody may
36  *		create a new unix_socket when we are in the middle of sweep
37  *		phase. Fix: revert the logic wrt MARKED. Mark everything
38  *		upon the beginning and unmark non-junk ones.
39  *
40  *		[12 Oct 1998] AAARGH! New code purges all SCM_RIGHTS
41  *		sent to connect()'ed but still not accept()'ed sockets.
42  *		Fixed. Old code had slightly different problem here:
43  *		extra fput() in situation when we passed the descriptor via
44  *		such socket and closed it (descriptor). That would happen on
45  *		each unix_gc() until the accept(). Since the struct file in
46  *		question would go to the free list and might be reused...
47  *		That might be the reason of random oopses on filp_close()
48  *		in unrelated processes.
49  *
50  *	AV		28 Feb 1999
51  *		Kill the explicit allocation of stack. Now we keep the tree
52  *		with root in dummy + pointer (gc_current) to one of the nodes.
53  *		Stack is represented as path from gc_current to dummy. Unmark
54  *		now means "add to tree". Push == "make it a son of gc_current".
55  *		Pop == "move gc_current to parent". We keep only pointers to
56  *		parents (->gc_tree).
57  *	AV		1 Mar 1999
58  *		Damn. Added missing check for ->dead in listen queues scanning.
59  *
60  *	Miklos Szeredi 25 Jun 2007
61  *		Reimplement with a cycle collecting algorithm. This should
62  *		solve several problems with the previous code, like being racy
63  *		wrt receive and holding up unrelated socket operations.
64  */
65 
66 #include <linux/fs.h>
67 #include <linux/list.h>
68 #include <linux/skbuff.h>
69 #include <linux/socket.h>
70 #include <linux/workqueue.h>
71 #include <net/af_unix.h>
72 #include <net/scm.h>
73 #include <net/tcp_states.h>
74 
75 #include "af_unix.h"
76 
77 struct unix_vertex {
78 	struct list_head edges;
79 	struct list_head entry;
80 	struct list_head scc_entry;
81 	unsigned long out_degree;
82 	unsigned long index;
83 	unsigned long scc_index;
84 };
85 
86 struct unix_edge {
87 	struct unix_sock *predecessor;
88 	struct unix_sock *successor;
89 	struct list_head vertex_entry;
90 	struct list_head stack_entry;
91 };
92 
unix_get_socket(struct file * filp)93 struct unix_sock *unix_get_socket(struct file *filp)
94 {
95 	struct inode *inode = file_inode(filp);
96 
97 	/* Socket ? */
98 	if (S_ISSOCK(inode->i_mode) && !(filp->f_mode & FMODE_PATH)) {
99 		struct socket *sock = SOCKET_I(inode);
100 		const struct proto_ops *ops;
101 		struct sock *sk = sock->sk;
102 
103 		ops = READ_ONCE(sock->ops);
104 
105 		/* PF_UNIX ? */
106 		if (sk && ops && ops->family == PF_UNIX)
107 			return unix_sk(sk);
108 	}
109 
110 	return NULL;
111 }
112 
unix_edge_successor(struct unix_edge * edge)113 static struct unix_vertex *unix_edge_successor(struct unix_edge *edge)
114 {
115 	/* If an embryo socket has a fd,
116 	 * the listener indirectly holds the fd's refcnt.
117 	 */
118 	if (edge->successor->listener)
119 		return unix_sk(edge->successor->listener)->vertex;
120 
121 	return edge->successor->vertex;
122 }
123 
124 enum {
125 	UNIX_GRAPH_NOT_CYCLIC,
126 	UNIX_GRAPH_MAYBE_CYCLIC,
127 	UNIX_GRAPH_CYCLIC,
128 };
129 
130 static unsigned char unix_graph_state;
131 
unix_update_graph(struct unix_vertex * vertex)132 static void unix_update_graph(struct unix_vertex *vertex)
133 {
134 	/* If the receiver socket is not inflight, no cyclic
135 	 * reference could be formed.
136 	 */
137 	if (!vertex)
138 		return;
139 
140 	WRITE_ONCE(unix_graph_state, UNIX_GRAPH_MAYBE_CYCLIC);
141 }
142 
143 static LIST_HEAD(unix_unvisited_vertices);
144 
145 enum unix_vertex_index {
146 	UNIX_VERTEX_INDEX_MARK1,
147 	UNIX_VERTEX_INDEX_MARK2,
148 	UNIX_VERTEX_INDEX_START,
149 };
150 
151 static unsigned long unix_vertex_unvisited_index = UNIX_VERTEX_INDEX_MARK1;
152 static unsigned long unix_vertex_max_scc_index = UNIX_VERTEX_INDEX_START;
153 
unix_add_edge(struct scm_fp_list * fpl,struct unix_edge * edge)154 static void unix_add_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
155 {
156 	struct unix_vertex *vertex = edge->predecessor->vertex;
157 
158 	if (!vertex) {
159 		vertex = list_first_entry(&fpl->vertices, typeof(*vertex), entry);
160 		vertex->index = unix_vertex_unvisited_index;
161 		vertex->scc_index = ++unix_vertex_max_scc_index;
162 		vertex->out_degree = 0;
163 		INIT_LIST_HEAD(&vertex->edges);
164 		INIT_LIST_HEAD(&vertex->scc_entry);
165 
166 		list_move_tail(&vertex->entry, &unix_unvisited_vertices);
167 		edge->predecessor->vertex = vertex;
168 	}
169 
170 	vertex->out_degree++;
171 	list_add_tail(&edge->vertex_entry, &vertex->edges);
172 
173 	unix_update_graph(unix_edge_successor(edge));
174 }
175 
unix_del_edge(struct scm_fp_list * fpl,struct unix_edge * edge)176 static void unix_del_edge(struct scm_fp_list *fpl, struct unix_edge *edge)
177 {
178 	struct unix_vertex *vertex = edge->predecessor->vertex;
179 
180 	if (!fpl->dead)
181 		unix_update_graph(unix_edge_successor(edge));
182 
183 	list_del(&edge->vertex_entry);
184 	vertex->out_degree--;
185 
186 	if (!vertex->out_degree) {
187 		edge->predecessor->vertex = NULL;
188 		list_move_tail(&vertex->entry, &fpl->vertices);
189 	}
190 }
191 
unix_free_vertices(struct scm_fp_list * fpl)192 static void unix_free_vertices(struct scm_fp_list *fpl)
193 {
194 	struct unix_vertex *vertex, *next_vertex;
195 
196 	list_for_each_entry_safe(vertex, next_vertex, &fpl->vertices, entry) {
197 		list_del(&vertex->entry);
198 		kfree(vertex);
199 	}
200 }
201 
202 static DEFINE_SPINLOCK(unix_gc_lock);
203 
unix_add_edges(struct scm_fp_list * fpl,struct unix_sock * receiver)204 void unix_add_edges(struct scm_fp_list *fpl, struct unix_sock *receiver)
205 {
206 	int i = 0, j = 0;
207 
208 	spin_lock(&unix_gc_lock);
209 
210 	if (!fpl->count_unix)
211 		goto out;
212 
213 	do {
214 		struct unix_sock *inflight = unix_get_socket(fpl->fp[j++]);
215 		struct unix_edge *edge;
216 
217 		if (!inflight)
218 			continue;
219 
220 		edge = fpl->edges + i++;
221 		edge->predecessor = inflight;
222 		edge->successor = receiver;
223 
224 		unix_add_edge(fpl, edge);
225 	} while (i < fpl->count_unix);
226 
227 	receiver->scm_stat.nr_unix_fds += fpl->count_unix;
228 out:
229 	WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight + fpl->count);
230 
231 	spin_unlock(&unix_gc_lock);
232 
233 	fpl->inflight = true;
234 
235 	unix_free_vertices(fpl);
236 }
237 
unix_del_edges(struct scm_fp_list * fpl)238 void unix_del_edges(struct scm_fp_list *fpl)
239 {
240 	struct unix_sock *receiver;
241 	int i = 0;
242 
243 	spin_lock(&unix_gc_lock);
244 
245 	if (!fpl->count_unix)
246 		goto out;
247 
248 	do {
249 		struct unix_edge *edge = fpl->edges + i++;
250 
251 		unix_del_edge(fpl, edge);
252 	} while (i < fpl->count_unix);
253 
254 	if (!fpl->dead) {
255 		receiver = fpl->edges[0].successor;
256 		receiver->scm_stat.nr_unix_fds -= fpl->count_unix;
257 	}
258 out:
259 	WRITE_ONCE(fpl->user->unix_inflight, fpl->user->unix_inflight - fpl->count);
260 
261 	spin_unlock(&unix_gc_lock);
262 
263 	fpl->inflight = false;
264 }
265 
unix_update_edges(struct unix_sock * receiver)266 void unix_update_edges(struct unix_sock *receiver)
267 {
268 	/* nr_unix_fds is only updated under unix_state_lock().
269 	 * If it's 0 here, the embryo socket is not part of the
270 	 * inflight graph, and GC will not see it, so no lock needed.
271 	 */
272 	if (!receiver->scm_stat.nr_unix_fds) {
273 		receiver->listener = NULL;
274 	} else {
275 		spin_lock(&unix_gc_lock);
276 		unix_update_graph(unix_sk(receiver->listener)->vertex);
277 		receiver->listener = NULL;
278 		spin_unlock(&unix_gc_lock);
279 	}
280 }
281 
unix_prepare_fpl(struct scm_fp_list * fpl)282 int unix_prepare_fpl(struct scm_fp_list *fpl)
283 {
284 	struct unix_vertex *vertex;
285 	int i;
286 
287 	if (!fpl->count_unix)
288 		return 0;
289 
290 	for (i = 0; i < fpl->count_unix; i++) {
291 		vertex = kmalloc(sizeof(*vertex), GFP_KERNEL);
292 		if (!vertex)
293 			goto err;
294 
295 		list_add(&vertex->entry, &fpl->vertices);
296 	}
297 
298 	fpl->edges = kvmalloc_array(fpl->count_unix, sizeof(*fpl->edges),
299 				    GFP_KERNEL_ACCOUNT);
300 	if (!fpl->edges)
301 		goto err;
302 
303 	unix_schedule_gc(fpl->user);
304 
305 	return 0;
306 
307 err:
308 	unix_free_vertices(fpl);
309 	return -ENOMEM;
310 }
311 
unix_destroy_fpl(struct scm_fp_list * fpl)312 void unix_destroy_fpl(struct scm_fp_list *fpl)
313 {
314 	if (fpl->inflight)
315 		unix_del_edges(fpl);
316 
317 	kvfree(fpl->edges);
318 	unix_free_vertices(fpl);
319 }
320 
unix_vertex_dead(struct unix_vertex * vertex)321 static bool unix_vertex_dead(struct unix_vertex *vertex)
322 {
323 	struct unix_edge *edge;
324 	struct unix_sock *u;
325 	long total_ref;
326 
327 	list_for_each_entry(edge, &vertex->edges, vertex_entry) {
328 		struct unix_vertex *next_vertex = unix_edge_successor(edge);
329 
330 		/* The vertex's fd can be received by a non-inflight socket. */
331 		if (!next_vertex)
332 			return false;
333 
334 		/* The vertex's fd can be received by an inflight socket in
335 		 * another SCC.
336 		 */
337 		if (next_vertex->scc_index != vertex->scc_index)
338 			return false;
339 	}
340 
341 	/* No receiver exists out of the same SCC. */
342 
343 	edge = list_first_entry(&vertex->edges, typeof(*edge), vertex_entry);
344 	u = edge->predecessor;
345 	total_ref = file_count(u->sk.sk_socket->file);
346 
347 	/* If not close()d, total_ref > out_degree. */
348 	if (total_ref != vertex->out_degree)
349 		return false;
350 
351 	return true;
352 }
353 
unix_collect_skb(struct list_head * scc,struct sk_buff_head * hitlist)354 static void unix_collect_skb(struct list_head *scc, struct sk_buff_head *hitlist)
355 {
356 	struct unix_vertex *vertex;
357 
358 	list_for_each_entry_reverse(vertex, scc, scc_entry) {
359 		struct sk_buff_head *queue;
360 		struct unix_edge *edge;
361 		struct unix_sock *u;
362 
363 		edge = list_first_entry(&vertex->edges, typeof(*edge), vertex_entry);
364 		u = edge->predecessor;
365 		queue = &u->sk.sk_receive_queue;
366 
367 		spin_lock(&queue->lock);
368 
369 		if (u->sk.sk_state == TCP_LISTEN) {
370 			struct sk_buff *skb;
371 
372 			skb_queue_walk(queue, skb) {
373 				struct sk_buff_head *embryo_queue = &skb->sk->sk_receive_queue;
374 
375 				spin_lock(&embryo_queue->lock);
376 				skb_queue_splice_init(embryo_queue, hitlist);
377 				spin_unlock(&embryo_queue->lock);
378 			}
379 		} else {
380 			skb_queue_splice_init(queue, hitlist);
381 		}
382 
383 		spin_unlock(&queue->lock);
384 	}
385 }
386 
unix_scc_cyclic(struct list_head * scc)387 static bool unix_scc_cyclic(struct list_head *scc)
388 {
389 	struct unix_vertex *vertex;
390 	struct unix_edge *edge;
391 
392 	/* SCC containing multiple vertices ? */
393 	if (!list_is_singular(scc))
394 		return true;
395 
396 	vertex = list_first_entry(scc, typeof(*vertex), scc_entry);
397 
398 	/* Self-reference or a embryo-listener circle ? */
399 	list_for_each_entry(edge, &vertex->edges, vertex_entry) {
400 		if (unix_edge_successor(edge) == vertex)
401 			return true;
402 	}
403 
404 	return false;
405 }
406 
407 static LIST_HEAD(unix_visited_vertices);
408 static unsigned long unix_vertex_grouped_index = UNIX_VERTEX_INDEX_MARK2;
409 
__unix_walk_scc(struct unix_vertex * vertex,unsigned long * last_index,struct sk_buff_head * hitlist)410 static unsigned long __unix_walk_scc(struct unix_vertex *vertex,
411 				     unsigned long *last_index,
412 				     struct sk_buff_head *hitlist)
413 {
414 	unsigned long cyclic_sccs = 0;
415 	LIST_HEAD(vertex_stack);
416 	struct unix_edge *edge;
417 	LIST_HEAD(edge_stack);
418 
419 next_vertex:
420 	/* Push vertex to vertex_stack and mark it as on-stack
421 	 * (index >= UNIX_VERTEX_INDEX_START).
422 	 * The vertex will be popped when finalising SCC later.
423 	 */
424 	list_add(&vertex->scc_entry, &vertex_stack);
425 
426 	vertex->index = *last_index;
427 	vertex->scc_index = *last_index;
428 	(*last_index)++;
429 
430 	/* Explore neighbour vertices (receivers of the current vertex's fd). */
431 	list_for_each_entry(edge, &vertex->edges, vertex_entry) {
432 		struct unix_vertex *next_vertex = unix_edge_successor(edge);
433 
434 		if (!next_vertex)
435 			continue;
436 
437 		if (next_vertex->index == unix_vertex_unvisited_index) {
438 			/* Iterative deepening depth first search
439 			 *
440 			 *   1. Push a forward edge to edge_stack and set
441 			 *      the successor to vertex for the next iteration.
442 			 */
443 			list_add(&edge->stack_entry, &edge_stack);
444 
445 			vertex = next_vertex;
446 			goto next_vertex;
447 
448 			/*   2. Pop the edge directed to the current vertex
449 			 *      and restore the ancestor for backtracking.
450 			 */
451 prev_vertex:
452 			edge = list_first_entry(&edge_stack, typeof(*edge), stack_entry);
453 			list_del_init(&edge->stack_entry);
454 
455 			next_vertex = vertex;
456 			vertex = edge->predecessor->vertex;
457 
458 			/* If the successor has a smaller scc_index, two vertices
459 			 * are in the same SCC, so propagate the smaller scc_index
460 			 * to skip SCC finalisation.
461 			 */
462 			vertex->scc_index = min(vertex->scc_index, next_vertex->scc_index);
463 		} else if (next_vertex->index != unix_vertex_grouped_index) {
464 			/* Loop detected by a back/cross edge.
465 			 *
466 			 * The successor is on vertex_stack, so two vertices are in
467 			 * the same SCC.  If the successor has a smaller *scc_index*,
468 			 * propagate it to skip SCC finalisation.
469 			 */
470 			vertex->scc_index = min(vertex->scc_index, next_vertex->scc_index);
471 		} else {
472 			/* The successor was already grouped as another SCC */
473 		}
474 	}
475 
476 	if (vertex->index == vertex->scc_index) {
477 		struct unix_vertex *v;
478 		struct list_head scc;
479 		bool scc_dead = true;
480 
481 		/* SCC finalised.
482 		 *
483 		 * If the scc_index was not updated, all the vertices above on
484 		 * vertex_stack are in the same SCC.  Group them using scc_entry.
485 		 */
486 		__list_cut_position(&scc, &vertex_stack, &vertex->scc_entry);
487 
488 		list_for_each_entry_reverse(v, &scc, scc_entry) {
489 			/* Don't restart DFS from this vertex in unix_walk_scc(). */
490 			list_move_tail(&v->entry, &unix_visited_vertices);
491 
492 			/* Mark vertex as off-stack. */
493 			v->index = unix_vertex_grouped_index;
494 
495 			if (scc_dead)
496 				scc_dead = unix_vertex_dead(v);
497 		}
498 
499 		if (scc_dead) {
500 			unix_collect_skb(&scc, hitlist);
501 		} else {
502 			if (unix_vertex_max_scc_index < vertex->scc_index)
503 				unix_vertex_max_scc_index = vertex->scc_index;
504 
505 			if (unix_scc_cyclic(&scc))
506 				cyclic_sccs++;
507 		}
508 
509 		list_del(&scc);
510 	}
511 
512 	/* Need backtracking ? */
513 	if (!list_empty(&edge_stack))
514 		goto prev_vertex;
515 
516 	return cyclic_sccs;
517 }
518 
519 static unsigned long unix_graph_cyclic_sccs;
520 
unix_walk_scc(struct sk_buff_head * hitlist)521 static void unix_walk_scc(struct sk_buff_head *hitlist)
522 {
523 	unsigned long last_index = UNIX_VERTEX_INDEX_START;
524 	unsigned long cyclic_sccs = 0;
525 
526 	unix_vertex_max_scc_index = UNIX_VERTEX_INDEX_START;
527 
528 	/* Visit every vertex exactly once.
529 	 * __unix_walk_scc() moves visited vertices to unix_visited_vertices.
530 	 */
531 	while (!list_empty(&unix_unvisited_vertices)) {
532 		struct unix_vertex *vertex;
533 
534 		vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry);
535 		cyclic_sccs += __unix_walk_scc(vertex, &last_index, hitlist);
536 	}
537 
538 	list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices);
539 	swap(unix_vertex_unvisited_index, unix_vertex_grouped_index);
540 
541 	WRITE_ONCE(unix_graph_cyclic_sccs, cyclic_sccs);
542 	WRITE_ONCE(unix_graph_state,
543 		   cyclic_sccs ? UNIX_GRAPH_CYCLIC : UNIX_GRAPH_NOT_CYCLIC);
544 }
545 
unix_walk_scc_fast(struct sk_buff_head * hitlist)546 static void unix_walk_scc_fast(struct sk_buff_head *hitlist)
547 {
548 	unsigned long cyclic_sccs = unix_graph_cyclic_sccs;
549 
550 	while (!list_empty(&unix_unvisited_vertices)) {
551 		struct unix_vertex *vertex;
552 		struct list_head scc;
553 		bool scc_dead = true;
554 
555 		vertex = list_first_entry(&unix_unvisited_vertices, typeof(*vertex), entry);
556 		list_add(&scc, &vertex->scc_entry);
557 
558 		list_for_each_entry_reverse(vertex, &scc, scc_entry) {
559 			list_move_tail(&vertex->entry, &unix_visited_vertices);
560 
561 			if (scc_dead)
562 				scc_dead = unix_vertex_dead(vertex);
563 		}
564 
565 		if (scc_dead) {
566 			cyclic_sccs--;
567 			unix_collect_skb(&scc, hitlist);
568 		}
569 
570 		list_del(&scc);
571 	}
572 
573 	list_replace_init(&unix_visited_vertices, &unix_unvisited_vertices);
574 
575 	WRITE_ONCE(unix_graph_cyclic_sccs, cyclic_sccs);
576 	WRITE_ONCE(unix_graph_state,
577 		   cyclic_sccs ? UNIX_GRAPH_CYCLIC : UNIX_GRAPH_NOT_CYCLIC);
578 }
579 
580 static bool gc_in_progress;
581 
unix_gc(struct work_struct * work)582 static void unix_gc(struct work_struct *work)
583 {
584 	struct sk_buff_head hitlist;
585 	struct sk_buff *skb;
586 
587 	spin_lock(&unix_gc_lock);
588 
589 	if (unix_graph_state == UNIX_GRAPH_NOT_CYCLIC) {
590 		spin_unlock(&unix_gc_lock);
591 		goto skip_gc;
592 	}
593 
594 	__skb_queue_head_init(&hitlist);
595 
596 	if (unix_graph_state == UNIX_GRAPH_CYCLIC)
597 		unix_walk_scc_fast(&hitlist);
598 	else
599 		unix_walk_scc(&hitlist);
600 
601 	spin_unlock(&unix_gc_lock);
602 
603 	skb_queue_walk(&hitlist, skb) {
604 		if (UNIXCB(skb).fp)
605 			UNIXCB(skb).fp->dead = true;
606 	}
607 
608 	__skb_queue_purge_reason(&hitlist, SKB_DROP_REASON_SOCKET_CLOSE);
609 skip_gc:
610 	WRITE_ONCE(gc_in_progress, false);
611 }
612 
613 static DECLARE_WORK(unix_gc_work, unix_gc);
614 
615 #define UNIX_INFLIGHT_SANE_USER		(SCM_MAX_FD * 8)
616 
unix_schedule_gc(struct user_struct * user)617 void unix_schedule_gc(struct user_struct *user)
618 {
619 	if (READ_ONCE(unix_graph_state) == UNIX_GRAPH_NOT_CYCLIC)
620 		return;
621 
622 	/* Penalise users who want to send AF_UNIX sockets
623 	 * but whose sockets have not been received yet.
624 	 */
625 	if (user &&
626 	    READ_ONCE(user->unix_inflight) < UNIX_INFLIGHT_SANE_USER)
627 		return;
628 
629 	if (!READ_ONCE(gc_in_progress)) {
630 		WRITE_ONCE(gc_in_progress, true);
631 		queue_work(system_dfl_wq, &unix_gc_work);
632 	}
633 
634 	if (user && READ_ONCE(unix_graph_cyclic_sccs))
635 		flush_work(&unix_gc_work);
636 }
637