summaryrefslogtreecommitdiffhomepage
path: root/pkg/tcpip/link/sharedmem/sharedmem_test.go
blob: 0f72d4e95960fc4a1e7af8cc89d084ba11f045fd (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
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
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
// 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.

// +build linux

package sharedmem

import (
	"bytes"
	"io/ioutil"
	"math/rand"
	"os"
	"strings"
	"testing"
	"time"

	"golang.org/x/sys/unix"
	"gvisor.dev/gvisor/pkg/sync"
	"gvisor.dev/gvisor/pkg/tcpip"
	"gvisor.dev/gvisor/pkg/tcpip/buffer"
	"gvisor.dev/gvisor/pkg/tcpip/header"
	"gvisor.dev/gvisor/pkg/tcpip/link/sharedmem/pipe"
	"gvisor.dev/gvisor/pkg/tcpip/link/sharedmem/queue"
	"gvisor.dev/gvisor/pkg/tcpip/stack"
)

const (
	localLinkAddr  = "\xde\xad\xbe\xef\x56\x78"
	remoteLinkAddr = "\xde\xad\xbe\xef\x12\x34"

	queueDataSize = 1024 * 1024
	queuePipeSize = 4096
)

type queueBuffers struct {
	data []byte
	rx   pipe.Tx
	tx   pipe.Rx
}

func initQueue(t *testing.T, q *queueBuffers, c *QueueConfig) {
	// Prepare tx pipe.
	b, err := getBuffer(c.TxPipeFD)
	if err != nil {
		t.Fatalf("getBuffer failed: %v", err)
	}
	q.tx.Init(b)

	// Prepare rx pipe.
	b, err = getBuffer(c.RxPipeFD)
	if err != nil {
		t.Fatalf("getBuffer failed: %v", err)
	}
	q.rx.Init(b)

	// Get data slice.
	q.data, err = getBuffer(c.DataFD)
	if err != nil {
		t.Fatalf("getBuffer failed: %v", err)
	}
}

func (q *queueBuffers) cleanup() {
	unix.Munmap(q.tx.Bytes())
	unix.Munmap(q.rx.Bytes())
	unix.Munmap(q.data)
}

type packetInfo struct {
	addr       tcpip.LinkAddress
	proto      tcpip.NetworkProtocolNumber
	data       buffer.View
	linkHeader buffer.View
}

type testContext struct {
	t     *testing.T
	ep    *endpoint
	txCfg QueueConfig
	rxCfg QueueConfig
	txq   queueBuffers
	rxq   queueBuffers

	packetCh chan struct{}
	mu       sync.Mutex
	packets  []packetInfo
}

func newTestContext(t *testing.T, mtu, bufferSize uint32, addr tcpip.LinkAddress) *testContext {
	var err error
	c := &testContext{
		t:        t,
		packetCh: make(chan struct{}, 1000000),
	}
	c.txCfg = createQueueFDs(t, queueSizes{
		dataSize:       queueDataSize,
		txPipeSize:     queuePipeSize,
		rxPipeSize:     queuePipeSize,
		sharedDataSize: 4096,
	})

	c.rxCfg = createQueueFDs(t, queueSizes{
		dataSize:       queueDataSize,
		txPipeSize:     queuePipeSize,
		rxPipeSize:     queuePipeSize,
		sharedDataSize: 4096,
	})

	initQueue(t, &c.txq, &c.txCfg)
	initQueue(t, &c.rxq, &c.rxCfg)

	ep, err := New(mtu, bufferSize, addr, c.txCfg, c.rxCfg)
	if err != nil {
		t.Fatalf("New failed: %v", err)
	}

	c.ep = ep.(*endpoint)
	c.ep.Attach(c)

	return c
}

func (c *testContext) DeliverNetworkPacket(remoteLinkAddr, localLinkAddr tcpip.LinkAddress, proto tcpip.NetworkProtocolNumber, pkt *stack.PacketBuffer) {
	c.mu.Lock()
	c.packets = append(c.packets, packetInfo{
		addr:  remoteLinkAddr,
		proto: proto,
		data:  pkt.Data().AsRange().ToOwnedView(),
	})
	c.mu.Unlock()

	c.packetCh <- struct{}{}
}

func (c *testContext) DeliverOutboundPacket(remoteLinkAddr, localLinkAddr tcpip.LinkAddress, proto tcpip.NetworkProtocolNumber, pkt *stack.PacketBuffer) {
	panic("unimplemented")
}

func (c *testContext) cleanup() {
	c.ep.Close()
	closeFDs(&c.txCfg)
	closeFDs(&c.rxCfg)
	c.txq.cleanup()
	c.rxq.cleanup()
}

func (c *testContext) waitForPackets(n int, to <-chan time.Time, errorStr string) {
	for i := 0; i < n; i++ {
		select {
		case <-c.packetCh:
		case <-to:
			c.t.Fatalf(errorStr)
		}
	}
}

func (c *testContext) pushRxCompletion(size uint32, bs []queue.RxBuffer) {
	b := c.rxq.rx.Push(queue.RxCompletionSize(len(bs)))
	queue.EncodeRxCompletion(b, size, 0)
	for i := range bs {
		queue.EncodeRxCompletionBuffer(b, i, queue.RxBuffer{
			Offset: bs[i].Offset,
			Size:   bs[i].Size,
			ID:     bs[i].ID,
		})
	}
}

func randomFill(b []byte) {
	for i := range b {
		b[i] = byte(rand.Intn(256))
	}
}

func shuffle(b []int) {
	for i := len(b) - 1; i >= 0; i-- {
		j := rand.Intn(i + 1)
		b[i], b[j] = b[j], b[i]
	}
}

func createFile(t *testing.T, size int64, initQueue bool) int {
	tmpDir, ok := os.LookupEnv("TEST_TMPDIR")
	if !ok {
		tmpDir = os.Getenv("TMPDIR")
	}
	f, err := ioutil.TempFile(tmpDir, "sharedmem_test")
	if err != nil {
		t.Fatalf("TempFile failed: %v", err)
	}
	defer f.Close()
	unix.Unlink(f.Name())

	if initQueue {
		// Write the "slot-free" flag in the initial queue.
		_, err := f.WriteAt([]byte{0, 0, 0, 0, 0, 0, 0, 0x80}, 0)
		if err != nil {
			t.Fatalf("WriteAt failed: %v", err)
		}
	}

	fd, err := unix.Dup(int(f.Fd()))
	if err != nil {
		t.Fatalf("Dup failed: %v", err)
	}

	if err := unix.Ftruncate(fd, size); err != nil {
		unix.Close(fd)
		t.Fatalf("Ftruncate failed: %v", err)
	}

	return fd
}

func closeFDs(c *QueueConfig) {
	unix.Close(c.DataFD)
	unix.Close(c.EventFD)
	unix.Close(c.TxPipeFD)
	unix.Close(c.RxPipeFD)
	unix.Close(c.SharedDataFD)
}

type queueSizes struct {
	dataSize       int64
	txPipeSize     int64
	rxPipeSize     int64
	sharedDataSize int64
}

func createQueueFDs(t *testing.T, s queueSizes) QueueConfig {
	fd, _, err := unix.RawSyscall(unix.SYS_EVENTFD2, 0, 0, 0)
	if err != 0 {
		t.Fatalf("eventfd failed: %v", error(err))
	}

	return QueueConfig{
		EventFD:      int(fd),
		DataFD:       createFile(t, s.dataSize, false),
		TxPipeFD:     createFile(t, s.txPipeSize, true),
		RxPipeFD:     createFile(t, s.rxPipeSize, true),
		SharedDataFD: createFile(t, s.sharedDataSize, false),
	}
}

// TestSimpleSend sends 1000 packets with random header and payload sizes,
// then checks that the right payload is received on the shared memory queues.
func TestSimpleSend(t *testing.T) {
	c := newTestContext(t, 20000, 1500, localLinkAddr)
	defer c.cleanup()

	// Prepare route.
	var r stack.RouteInfo
	r.RemoteLinkAddress = remoteLinkAddr

	for iters := 1000; iters > 0; iters-- {
		func() {
			hdrLen, dataLen := rand.Intn(10000), rand.Intn(10000)

			// Prepare and send packet.
			hdrBuf := buffer.NewView(hdrLen)
			randomFill(hdrBuf)

			data := buffer.NewView(dataLen)
			randomFill(data)

			pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
				ReserveHeaderBytes: hdrLen + int(c.ep.MaxHeaderLength()),
				Data:               data.ToVectorisedView(),
			})
			copy(pkt.NetworkHeader().Push(hdrLen), hdrBuf)

			proto := tcpip.NetworkProtocolNumber(rand.Intn(0x10000))
			if err := c.ep.WritePacket(r, proto, pkt); err != nil {
				t.Fatalf("WritePacket failed: %v", err)
			}

			// Receive packet.
			desc := c.txq.tx.Pull()
			pi := queue.DecodeTxPacketHeader(desc)
			if pi.Reserved != 0 {
				t.Fatalf("Reserved value is non-zero: 0x%x", pi.Reserved)
			}
			contents := make([]byte, 0, pi.Size)
			for i := 0; i < pi.BufferCount; i++ {
				bi := queue.DecodeTxBufferHeader(desc, i)
				contents = append(contents, c.txq.data[bi.Offset:][:bi.Size]...)
			}
			c.txq.tx.Flush()

			defer func() {
				// Tell the endpoint about the completion of the write.
				b := c.txq.rx.Push(8)
				queue.EncodeTxCompletion(b, pi.ID)
				c.txq.rx.Flush()
			}()

			// Check the ethernet header.
			ethTemplate := make(header.Ethernet, header.EthernetMinimumSize)
			ethTemplate.Encode(&header.EthernetFields{
				SrcAddr: localLinkAddr,
				DstAddr: remoteLinkAddr,
				Type:    proto,
			})
			if got := contents[:header.EthernetMinimumSize]; !bytes.Equal(got, []byte(ethTemplate)) {
				t.Fatalf("Bad ethernet header in packet: got %x, want %x", got, ethTemplate)
			}

			// Compare contents skipping the ethernet header added by the
			// endpoint.
			merged := append(hdrBuf, data...)
			if uint32(len(contents)) < pi.Size {
				t.Fatalf("Sum of buffers is less than packet size: %v < %v", len(contents), pi.Size)
			}
			contents = contents[:pi.Size][header.EthernetMinimumSize:]

			if !bytes.Equal(contents, merged) {
				t.Fatalf("Buffers are different: got %x (%v bytes), want %x (%v bytes)", contents, len(contents), merged, len(merged))
			}
		}()
	}
}

// TestPreserveSrcAddressInSend calls WritePacket once with LocalLinkAddress
// set in Route (using much of the same code as TestSimpleSend), then checks
// that the encoded ethernet header received includes the correct SrcAddr.
func TestPreserveSrcAddressInSend(t *testing.T) {
	c := newTestContext(t, 20000, 1500, localLinkAddr)
	defer c.cleanup()

	newLocalLinkAddress := tcpip.LinkAddress(strings.Repeat("0xFE", 6))
	// Set both remote and local link address in route.
	var r stack.RouteInfo
	r.LocalLinkAddress = newLocalLinkAddress
	r.RemoteLinkAddress = remoteLinkAddr

	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
		// WritePacket panics given a prependable with anything less than
		// the minimum size of the ethernet header.
		ReserveHeaderBytes: header.EthernetMinimumSize,
	})

	proto := tcpip.NetworkProtocolNumber(rand.Intn(0x10000))
	if err := c.ep.WritePacket(r, proto, pkt); err != nil {
		t.Fatalf("WritePacket failed: %v", err)
	}

	// Receive packet.
	desc := c.txq.tx.Pull()
	pi := queue.DecodeTxPacketHeader(desc)
	if pi.Reserved != 0 {
		t.Fatalf("Reserved value is non-zero: 0x%x", pi.Reserved)
	}
	contents := make([]byte, 0, pi.Size)
	for i := 0; i < pi.BufferCount; i++ {
		bi := queue.DecodeTxBufferHeader(desc, i)
		contents = append(contents, c.txq.data[bi.Offset:][:bi.Size]...)
	}
	c.txq.tx.Flush()

	defer func() {
		// Tell the endpoint about the completion of the write.
		b := c.txq.rx.Push(8)
		queue.EncodeTxCompletion(b, pi.ID)
		c.txq.rx.Flush()
	}()

	// Check that the ethernet header contains the expected SrcAddr.
	ethTemplate := make(header.Ethernet, header.EthernetMinimumSize)
	ethTemplate.Encode(&header.EthernetFields{
		SrcAddr: newLocalLinkAddress,
		DstAddr: remoteLinkAddr,
		Type:    proto,
	})
	if got := contents[:header.EthernetMinimumSize]; !bytes.Equal(got, []byte(ethTemplate)) {
		t.Fatalf("Bad ethernet header in packet: got %x, want %x", got, ethTemplate)
	}
}

// TestFillTxQueue sends packets until the queue is full.
func TestFillTxQueue(t *testing.T) {
	c := newTestContext(t, 20000, 1500, localLinkAddr)
	defer c.cleanup()

	// Prepare to send a packet.
	var r stack.RouteInfo
	r.RemoteLinkAddress = remoteLinkAddr

	buf := buffer.NewView(100)

	// Each packet is uses no more than 40 bytes, so write that many packets
	// until the tx queue if full.
	ids := make(map[uint64]struct{})
	for i := queuePipeSize / 40; i > 0; i-- {
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})

		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}

		// Check that they have different IDs.
		desc := c.txq.tx.Pull()
		pi := queue.DecodeTxPacketHeader(desc)
		if _, ok := ids[pi.ID]; ok {
			t.Fatalf("ID (%v) reused", pi.ID)
		}
		ids[pi.ID] = struct{}{}
	}

	// Next attempt to write must fail.
	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
		ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
		Data:               buf.ToVectorisedView(),
	})
	err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt)
	if _, ok := err.(*tcpip.ErrWouldBlock); !ok {
		t.Fatalf("got WritePacket(...) = %v, want %s", err, &tcpip.ErrWouldBlock{})
	}
}

// TestFillTxQueueAfterBadCompletion sends a bad completion, then sends packets
// until the queue is full.
func TestFillTxQueueAfterBadCompletion(t *testing.T) {
	c := newTestContext(t, 20000, 1500, localLinkAddr)
	defer c.cleanup()

	// Send a bad completion.
	queue.EncodeTxCompletion(c.txq.rx.Push(8), 1)
	c.txq.rx.Flush()

	// Prepare to send a packet.
	var r stack.RouteInfo
	r.RemoteLinkAddress = remoteLinkAddr

	buf := buffer.NewView(100)

	// Send two packets so that the id slice has at least two slots.
	for i := 2; i > 0; i-- {
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})
		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}
	}

	// Complete the two writes twice.
	for i := 2; i > 0; i-- {
		pi := queue.DecodeTxPacketHeader(c.txq.tx.Pull())

		queue.EncodeTxCompletion(c.txq.rx.Push(8), pi.ID)
		queue.EncodeTxCompletion(c.txq.rx.Push(8), pi.ID)
		c.txq.rx.Flush()
	}
	c.txq.tx.Flush()

	// Each packet is uses no more than 40 bytes, so write that many packets
	// until the tx queue if full.
	ids := make(map[uint64]struct{})
	for i := queuePipeSize / 40; i > 0; i-- {
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})
		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}

		// Check that they have different IDs.
		desc := c.txq.tx.Pull()
		pi := queue.DecodeTxPacketHeader(desc)
		if _, ok := ids[pi.ID]; ok {
			t.Fatalf("ID (%v) reused", pi.ID)
		}
		ids[pi.ID] = struct{}{}
	}

	// Next attempt to write must fail.
	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
		ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
		Data:               buf.ToVectorisedView(),
	})
	err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt)
	if _, ok := err.(*tcpip.ErrWouldBlock); !ok {
		t.Fatalf("got WritePacket(...) = %v, want %s", err, &tcpip.ErrWouldBlock{})
	}
}

// TestFillTxMemory sends packets until the we run out of shared memory.
func TestFillTxMemory(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	defer c.cleanup()

	// Prepare to send a packet.
	var r stack.RouteInfo
	r.RemoteLinkAddress = remoteLinkAddr

	buf := buffer.NewView(100)

	// Each packet is uses up one buffer, so write as many as possible until
	// we fill the memory.
	ids := make(map[uint64]struct{})
	for i := queueDataSize / bufferSize; i > 0; i-- {
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})
		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}

		// Check that they have different IDs.
		desc := c.txq.tx.Pull()
		pi := queue.DecodeTxPacketHeader(desc)
		if _, ok := ids[pi.ID]; ok {
			t.Fatalf("ID (%v) reused", pi.ID)
		}
		ids[pi.ID] = struct{}{}
		c.txq.tx.Flush()
	}

	// Next attempt to write must fail.
	pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
		ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
		Data:               buf.ToVectorisedView(),
	})
	err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt)
	if _, ok := err.(*tcpip.ErrWouldBlock); !ok {
		t.Fatalf("got WritePacket(...) = %v, want %s", err, &tcpip.ErrWouldBlock{})
	}
}

// TestFillTxMemoryWithMultiBuffer sends packets until the we run out of
// shared memory for a 2-buffer packet, but still with room for a 1-buffer
// packet.
func TestFillTxMemoryWithMultiBuffer(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	defer c.cleanup()

	// Prepare to send a packet.
	var r stack.RouteInfo
	r.RemoteLinkAddress = remoteLinkAddr

	buf := buffer.NewView(100)

	// Each packet is uses up one buffer, so write as many as possible
	// until there is only one buffer left.
	for i := queueDataSize/bufferSize - 1; i > 0; i-- {
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})
		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}

		// Pull the posted buffer.
		c.txq.tx.Pull()
		c.txq.tx.Flush()
	}

	// Attempt to write a two-buffer packet. It must fail.
	{
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buffer.NewView(bufferSize).ToVectorisedView(),
		})
		err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt)
		if _, ok := err.(*tcpip.ErrWouldBlock); !ok {
			t.Fatalf("got WritePacket(...) = %v, want %s", err, &tcpip.ErrWouldBlock{})
		}
	}

	// Attempt to write the one-buffer packet again. It must succeed.
	{
		pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
			ReserveHeaderBytes: int(c.ep.MaxHeaderLength()),
			Data:               buf.ToVectorisedView(),
		})
		if err := c.ep.WritePacket(r, header.IPv4ProtocolNumber, pkt); err != nil {
			t.Fatalf("WritePacket failed unexpectedly: %v", err)
		}
	}
}

func pollPull(t *testing.T, p *pipe.Rx, to <-chan time.Time, errStr string) []byte {
	t.Helper()

	for {
		b := p.Pull()
		if b != nil {
			return b
		}

		select {
		case <-time.After(10 * time.Millisecond):
		case <-to:
			t.Fatal(errStr)
		}
	}
}

// TestSimpleReceive completes 1000 different receives with random payload and
// random number of buffers. It checks that the contents match the expected
// values.
func TestSimpleReceive(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	defer c.cleanup()

	// Check that buffers have been posted.
	limit := c.ep.rx.q.PostedBuffersLimit()
	for i := uint64(0); i < limit; i++ {
		timeout := time.After(2 * time.Second)
		bi := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, timeout, "Timeout waiting for all buffers to be posted"))

		if want := i * bufferSize; want != bi.Offset {
			t.Fatalf("Bad posted offset: got %v, want %v", bi.Offset, want)
		}

		if want := i; want != bi.ID {
			t.Fatalf("Bad posted ID: got %v, want %v", bi.ID, want)
		}

		if bufferSize != bi.Size {
			t.Fatalf("Bad posted bufferSize: got %v, want %v", bi.Size, bufferSize)
		}
	}
	c.rxq.tx.Flush()

	// Create a slice with the indices 0..limit-1.
	idx := make([]int, limit)
	for i := range idx {
		idx[i] = i
	}

	// Complete random packets 1000 times.
	for iters := 1000; iters > 0; iters-- {
		timeout := time.After(2 * time.Second)
		// Prepare a random packet.
		shuffle(idx)
		n := 1 + rand.Intn(10)
		bufs := make([]queue.RxBuffer, n)
		contents := make([]byte, bufferSize*n-rand.Intn(500))
		randomFill(contents)
		for i := range bufs {
			j := idx[i]
			bufs[i].Size = bufferSize
			bufs[i].Offset = uint64(bufferSize * j)
			bufs[i].ID = uint64(j)

			copy(c.rxq.data[bufs[i].Offset:][:bufferSize], contents[i*bufferSize:])
		}

		// Push completion.
		c.pushRxCompletion(uint32(len(contents)), bufs)
		c.rxq.rx.Flush()
		unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

		// Wait for packet to be received, then check it.
		c.waitForPackets(1, time.After(5*time.Second), "Timeout waiting for packet")
		c.mu.Lock()
		rcvd := []byte(c.packets[0].data)
		c.packets = c.packets[:0]
		c.mu.Unlock()

		if contents := contents[header.EthernetMinimumSize:]; !bytes.Equal(contents, rcvd) {
			t.Fatalf("Unexpected buffer contents: got %x, want %x", rcvd, contents)
		}

		// Check that buffers have been reposted.
		for i := range bufs {
			bi := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, timeout, "Timeout waiting for buffers to be reposted"))
			if bi != bufs[i] {
				t.Fatalf("Unexpected buffer reposted: got %x, want %x", bi, bufs[i])
			}
		}
		c.rxq.tx.Flush()
	}
}

// TestRxBuffersReposted tests that rx buffers get reposted after they have been
// completed.
func TestRxBuffersReposted(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	defer c.cleanup()

	// Receive all posted buffers.
	limit := c.ep.rx.q.PostedBuffersLimit()
	buffers := make([]queue.RxBuffer, 0, limit)
	for i := limit; i > 0; i-- {
		timeout := time.After(2 * time.Second)
		buffers = append(buffers, queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, timeout, "Timeout waiting for all buffers")))
	}
	c.rxq.tx.Flush()

	// Check that all buffers are reposted when individually completed.
	for i := range buffers {
		timeout := time.After(2 * time.Second)
		// Complete the buffer.
		c.pushRxCompletion(buffers[i].Size, buffers[i:][:1])
		c.rxq.rx.Flush()
		unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

		// Wait for it to be reposted.
		bi := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, timeout, "Timeout waiting for buffer to be reposted"))
		if bi != buffers[i] {
			t.Fatalf("Different buffer posted: got %v, want %v", bi, buffers[i])
		}
	}
	c.rxq.tx.Flush()

	// Check that all buffers are reposted when completed in pairs.
	for i := 0; i < len(buffers)/2; i++ {
		timeout := time.After(2 * time.Second)
		// Complete with two buffers.
		c.pushRxCompletion(2*bufferSize, buffers[2*i:][:2])
		c.rxq.rx.Flush()
		unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

		// Wait for them to be reposted.
		for j := 0; j < 2; j++ {
			bi := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, timeout, "Timeout waiting for buffer to be reposted"))
			if bi != buffers[2*i+j] {
				t.Fatalf("Different buffer posted: got %v, want %v", bi, buffers[2*i+j])
			}
		}
	}
	c.rxq.tx.Flush()
}

// TestReceivePostingIsFull checks that the endpoint will properly handle the
// case when a received buffer cannot be immediately reposted because it hasn't
// been pulled from the tx pipe yet.
func TestReceivePostingIsFull(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	defer c.cleanup()

	// Complete first posted buffer before flushing it from the tx pipe.
	first := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, time.After(time.Second), "Timeout waiting for first buffer to be posted"))
	c.pushRxCompletion(first.Size, []queue.RxBuffer{first})
	c.rxq.rx.Flush()
	unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

	// Check that packet is received.
	c.waitForPackets(1, time.After(time.Second), "Timeout waiting for completed packet")

	// Complete another buffer.
	second := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, time.After(time.Second), "Timeout waiting for second buffer to be posted"))
	c.pushRxCompletion(second.Size, []queue.RxBuffer{second})
	c.rxq.rx.Flush()
	unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

	// Check that no packet is received yet, as the worker is blocked trying
	// to repost.
	select {
	case <-time.After(500 * time.Millisecond):
	case <-c.packetCh:
		t.Fatalf("Unexpected packet received")
	}

	// Flush tx queue, which will allow the first buffer to be reposted,
	// and the second completion to be pulled.
	c.rxq.tx.Flush()
	unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

	// Check that second packet completes.
	c.waitForPackets(1, time.After(time.Second), "Timeout waiting for second completed packet")
}

// TestCloseWhileWaitingToPost closes the endpoint while it is waiting to
// repost a buffer. Make sure it backs out.
func TestCloseWhileWaitingToPost(t *testing.T) {
	const bufferSize = 1500
	c := newTestContext(t, 20000, bufferSize, localLinkAddr)
	cleaned := false
	defer func() {
		if !cleaned {
			c.cleanup()
		}
	}()

	// Complete first posted buffer before flushing it from the tx pipe.
	bi := queue.DecodeRxBufferHeader(pollPull(t, &c.rxq.tx, time.After(time.Second), "Timeout waiting for initial buffer to be posted"))
	c.pushRxCompletion(bi.Size, []queue.RxBuffer{bi})
	c.rxq.rx.Flush()
	unix.Write(c.rxCfg.EventFD, []byte{1, 0, 0, 0, 0, 0, 0, 0})

	// Wait for packet to be indicated.
	c.waitForPackets(1, time.After(time.Second), "Timeout waiting for completed packet")

	// Cleanup and wait for worker to complete.
	c.cleanup()
	cleaned = true
	c.ep.Wait()
}