summaryrefslogtreecommitdiffhomepage
path: root/pkg/sentry/socket/unix/transport/queue.go
blob: e27b1c714d861382b5bafd5c4601f4b2049961a3 (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
// Copyright 2018 The gVisor Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

package transport

import (
	"sync"

	"gvisor.dev/gvisor/pkg/refs"
	"gvisor.dev/gvisor/pkg/syserr"
	"gvisor.dev/gvisor/pkg/waiter"
)

// queue is a buffer queue.
//
// +stateify savable
type queue struct {
	refs.AtomicRefCount

	ReaderQueue *waiter.Queue
	WriterQueue *waiter.Queue

	mu       sync.Mutex `state:"nosave"`
	closed   bool
	unread   bool
	used     int64
	limit    int64
	dataList messageList
}

// Close closes q for reading and writing. It is immediately not writable and
// will become unreadable when no more data is pending.
//
// Both the read and write queues must be notified after closing:
// q.ReaderQueue.Notify(waiter.EventIn)
// q.WriterQueue.Notify(waiter.EventOut)
func (q *queue) Close() {
	q.mu.Lock()
	q.closed = true
	q.mu.Unlock()
}

// Reset empties the queue and Releases all of the Entries.
//
// Both the read and write queues must be notified after resetting:
// q.ReaderQueue.Notify(waiter.EventIn)
// q.WriterQueue.Notify(waiter.EventOut)
func (q *queue) Reset() {
	q.mu.Lock()
	for cur := q.dataList.Front(); cur != nil; cur = cur.Next() {
		cur.Release()
	}
	q.dataList.Reset()
	q.used = 0
	q.mu.Unlock()
}

// DecRef implements RefCounter.DecRef with destructor q.Reset.
func (q *queue) DecRef() {
	q.DecRefWithDestructor(q.Reset)
	// We don't need to notify after resetting because no one cares about
	// this queue after all references have been dropped.
}

// IsReadable determines if q is currently readable.
func (q *queue) IsReadable() bool {
	q.mu.Lock()
	defer q.mu.Unlock()

	return q.closed || q.dataList.Front() != nil
}

// bufWritable returns true if there is space for writing.
//
// N.B. Linux only considers a unix socket "writable" if >75% of the buffer is
// free.
//
// See net/unix/af_unix.c:unix_writeable.
func (q *queue) bufWritable() bool {
	return 4*q.used < q.limit
}

// IsWritable determines if q is currently writable.
func (q *queue) IsWritable() bool {
	q.mu.Lock()
	defer q.mu.Unlock()

	return q.closed || q.bufWritable()
}

// Enqueue adds an entry to the data queue if room is available.
//
// If truncate is true, Enqueue may truncate the message before enqueuing it.
// Otherwise, the entire message must fit. If n < e.Length(), err indicates why.
//
// If notify is true, ReaderQueue.Notify must be called:
// q.ReaderQueue.Notify(waiter.EventIn)
func (q *queue) Enqueue(e *message, truncate bool) (l int64, notify bool, err *syserr.Error) {
	q.mu.Lock()

	if q.closed {
		q.mu.Unlock()
		return 0, false, syserr.ErrClosedForSend
	}

	free := q.limit - q.used

	l = e.Length()

	if l > free && truncate {
		if free == 0 {
			// Message can't fit right now.
			q.mu.Unlock()
			return 0, false, syserr.ErrWouldBlock
		}

		e.Truncate(free)
		l = e.Length()
		err = syserr.ErrWouldBlock
	}

	if l > q.limit {
		// Message is too big to ever fit.
		q.mu.Unlock()
		return 0, false, syserr.ErrMessageTooLong
	}

	if l > free {
		// Message can't fit right now.
		q.mu.Unlock()
		return 0, false, syserr.ErrWouldBlock
	}

	notify = q.dataList.Front() == nil
	q.used += l
	q.dataList.PushBack(e)

	q.mu.Unlock()

	return l, notify, err
}

// Dequeue removes the first entry in the data queue, if one exists.
//
// If notify is true, WriterQueue.Notify must be called:
// q.WriterQueue.Notify(waiter.EventOut)
func (q *queue) Dequeue() (e *message, notify bool, err *syserr.Error) {
	q.mu.Lock()

	if q.dataList.Front() == nil {
		err := syserr.ErrWouldBlock
		if q.closed {
			err = syserr.ErrClosedForReceive
			if q.unread {
				err = syserr.ErrConnectionReset
			}
		}
		q.mu.Unlock()

		return nil, false, err
	}

	notify = !q.bufWritable()

	e = q.dataList.Front()
	q.dataList.Remove(e)
	q.used -= e.Length()

	notify = notify && q.bufWritable()

	q.mu.Unlock()

	return e, notify, nil
}

// Peek returns the first entry in the data queue, if one exists.
func (q *queue) Peek() (*message, *syserr.Error) {
	q.mu.Lock()
	defer q.mu.Unlock()

	if q.dataList.Front() == nil {
		err := syserr.ErrWouldBlock
		if q.closed {
			if err = syserr.ErrClosedForReceive; q.unread {
				err = syserr.ErrConnectionReset
			}
		}
		return nil, err
	}

	return q.dataList.Front().Peek(), nil
}

// QueuedSize returns the number of bytes currently in the queue, that is, the
// number of readable bytes.
func (q *queue) QueuedSize() int64 {
	q.mu.Lock()
	defer q.mu.Unlock()
	return q.used
}

// MaxQueueSize returns the maximum number of bytes storable in the queue.
func (q *queue) MaxQueueSize() int64 {
	return q.limit
}

// CloseUnread sets flag to indicate that the peer is closed (not shutdown)
// with unread data. So if read on this queue shall return ECONNRESET error.
func (q *queue) CloseUnread() {
	q.mu.Lock()
	defer q.mu.Unlock()
	q.unread = true
}