aboutsummaryrefslogtreecommitdiffstats
path: root/net/unix/garbage.c
blob: 00734e22ec151cfb6d711f73f7b70a478391eed1 (plain)
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
/*
 * NET3:	Garbage Collector For AF_UNIX sockets
 *
 * Garbage Collector:
 *	Copyright (C) Barak A. Pearlmutter.
 *	Released under the GPL version 2 or later.
 *
 * 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
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 *
 *  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/slab.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <linux/file.h>
#include <linux/proc_fs.h>
#include <linux/mutex.h>

#include <net/sock.h>
#include <net/af_unix.h>
#include <net/scm.h>
#include <net/tcp_states.h>

/* Internal data structures and random procedures: */

static LIST_HEAD(gc_inflight_list);
static LIST_HEAD(gc_candidates);
static DEFINE_SPINLOCK(unix_gc_lock);

unsigned int unix_tot_inflight;


static struct sock *unix_get_socket(struct file *filp)
{
	struct sock *u_sock = NULL;
	struct inode *inode = filp->f_path.dentry->d_inode;

	/*
	 *	Socket ?
	 */
	if (S_ISSOCK(inode->i_mode)) {
		struct socket *sock = SOCKET_I(inode);
		struct sock *s = sock->sk;

		/*
		 *	PF_UNIX ?
		 */
		if (s && sock->ops && sock->ops->family == PF_UNIX)
			u_sock = s;
	}
	return u_sock;
}

/*
 *	Keep the number of times in flight count for the file
 *	descriptor if it is for an AF_UNIX socket.
 */

void unix_inflight(struct file *fp)
{
	struct sock *s = unix_get_socket(fp);
	if (s) {
		struct unix_sock *u = unix_sk(s);
		spin_lock(&unix_gc_lock);
		if (atomic_long_inc_return(&u->inflight) == 1) {
			BUG_ON(!list_empty(&u->link));
			list_add_tail(&u->link, &gc_inflight_list);
		} else {
			BUG_ON(list_empty(&u->link));
		}
		unix_tot_inflight++;
		spin_unlock(&unix_gc_lock);
	}
}

void unix_notinflight(struct file *fp)
{
	struct sock *s = unix_get_socket(fp);
	if (s) {
		struct unix_sock *u = unix_sk(s);
		spin_lock(&unix_gc_lock);
		BUG_ON(list_empty(&u->link));
		if (atomic_long_dec_and_test(&u->inflight))
			list_del_init(&u->link);
		unix_tot_inflight--;
		spin_unlock(&unix_gc_lock);
	}
}

static inline struct sk_buff *sock_queue_head(struct sock *sk)
{
	return (struct sk_buff *)&sk->sk_receive_queue;
}

#define receive_queue_for_each_skb(sk, next, skb) \
	for (skb = sock_queue_head(sk)->next, next = skb->next; \
	     skb != sock_queue_head(sk); skb = next, next = skb->next)

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);
	receive_queue_for_each_skb(x, next, skb) {
		/*
		 *	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 sock *sk = unix_get_socket(*fp++);
				if (sk) {
					hit = true;
					func(unix_sk(sk));
				}
			}
			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);
		receive_queue_for_each_skb(x, next, skb) {
			u = unix_sk(skb->sk);

			/*
			 * An embryo cannot be in-flight, so it's safe
			 * to use the list link.
			 */
			BUG_ON(!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)
{
	atomic_long_dec(&usk->inflight);
}

static void inc_inflight(struct unix_sock *usk)
{
	atomic_long_inc(&usk->inflight);
}

static void inc_inflight_move_tail(struct unix_sock *u)
{
	atomic_long_inc(&u->inflight);
	/*
	 * If this is still a candidate, move it to the end of the
	 * list, so that it's checked even if it was already passed
	 * over
	 */
	if (u->gc_candidate)
		list_move_tail(&u->link, &gc_candidates);
}

/* The external entry point: unix_gc() */

void unix_gc(void)
{
	static bool gc_in_progress = false;

	struct unix_sock *u;
	struct unix_sock *next;
	struct sk_buff_head hitlist;
	struct list_head cursor;

	spin_lock(&unix_gc_lock);

	/* Avoid a recursive GC. */
	if (gc_in_progress)
		goto out;

	gc_in_progress = true;
	/*
	 * 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.  This also means, that since there are no
	 * possible receivers, the receive queues of these sockets are
	 * static during the GC, even though the dequeue is done
	 * before the detach without atomicity guarantees.
	 */
	list_for_each_entry_safe(u, next, &gc_inflight_list, link) {
		long total_refs;
		long inflight_refs;

		total_refs = file_count(u->sk.sk_socket->file);
		inflight_refs = atomic_long_read(&u->inflight);

		BUG_ON(inflight_refs < 1);
		BUG_ON(total_refs < inflight_refs);
		if (total_refs == inflight_refs) {
			list_move_tail(&u->link, &gc_candidates);
			u->gc_candidate = 1;
		}
	}

	/*
	 * 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 (atomic_long_read(&u->inflight) > 0) {
			list_move_tail(&u->link, &gc_inflight_list);
			u->gc_candidate = 0;
			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);

	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. */
	BUG_ON(!list_empty(&gc_candidates));
	gc_in_progress = false;

 out:
	spin_unlock(&unix_gc_lock);
}