/
unix.go
887 lines (776 loc) · 24 KB
/
unix.go
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
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
// 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 unix provides an implementation of the socket.Socket interface for
// the AF_UNIX protocol family.
package unix
import (
"bytes"
"fmt"
"golang.org/x/sys/unix"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/context"
"gvisor.dev/gvisor/pkg/errors/linuxerr"
"gvisor.dev/gvisor/pkg/fspath"
"gvisor.dev/gvisor/pkg/hostarch"
"gvisor.dev/gvisor/pkg/log"
"gvisor.dev/gvisor/pkg/marshal"
"gvisor.dev/gvisor/pkg/sentry/arch"
"gvisor.dev/gvisor/pkg/sentry/fsimpl/sockfs"
"gvisor.dev/gvisor/pkg/sentry/inet"
"gvisor.dev/gvisor/pkg/sentry/kernel"
ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
"gvisor.dev/gvisor/pkg/sentry/socket"
"gvisor.dev/gvisor/pkg/sentry/socket/control"
"gvisor.dev/gvisor/pkg/sentry/socket/netstack"
"gvisor.dev/gvisor/pkg/sentry/socket/unix/transport"
"gvisor.dev/gvisor/pkg/sentry/vfs"
"gvisor.dev/gvisor/pkg/syserr"
"gvisor.dev/gvisor/pkg/usermem"
"gvisor.dev/gvisor/pkg/waiter"
)
// Socket implements socket.Socket (and by extension,
// vfs.FileDescriptionImpl) for Unix sockets.
//
// +stateify savable
type Socket struct {
vfsfd vfs.FileDescription
vfs.FileDescriptionDefaultImpl
vfs.DentryMetadataFileDescriptionImpl
vfs.LockFD
socket.SendReceiveTimeout
socketRefs
namespace *inet.Namespace
ep transport.Endpoint
stype linux.SockType
// abstractName and abstractNamespace indicate the name and namespace of the
// socket if it is bound to an abstract socket namespace. Once the socket is
// bound, they cannot be modified.
abstractName string
abstractBound bool
}
var _ = socket.Socket(&Socket{})
// NewSockfsFile creates a new socket file in the global sockfs mount and
// returns a corresponding file description.
func NewSockfsFile(t *kernel.Task, ep transport.Endpoint, stype linux.SockType) (*vfs.FileDescription, *syserr.Error) {
mnt := t.Kernel().SocketMount()
d := sockfs.NewDentry(t, mnt)
defer d.DecRef(t)
ns := t.GetNetworkNamespace()
fd, err := NewFileDescription(ep, stype, linux.O_RDWR, ns, mnt, d, &vfs.FileLocks{})
if err != nil {
ns.DecRef(t)
return nil, syserr.FromError(err)
}
return fd, nil
}
// NewFileDescription creates and returns a socket file description
// corresponding to the given mount and dentry.
func NewFileDescription(ep transport.Endpoint, stype linux.SockType, flags uint32, ns *inet.Namespace, mnt *vfs.Mount, d *vfs.Dentry, locks *vfs.FileLocks) (*vfs.FileDescription, error) {
// You can create AF_UNIX, SOCK_RAW sockets. They're the same as
// SOCK_DGRAM and don't require CAP_NET_RAW.
if stype == linux.SOCK_RAW {
stype = linux.SOCK_DGRAM
}
sock := &Socket{
ep: ep,
stype: stype,
namespace: ns,
}
sock.InitRefs()
sock.LockFD.Init(locks)
vfsfd := &sock.vfsfd
if err := vfsfd.Init(sock, flags, mnt, d, &vfs.FileDescriptionOptions{
DenyPRead: true,
DenyPWrite: true,
UseDentryMetadata: true,
}); err != nil {
return nil, err
}
return vfsfd, nil
}
// DecRef implements RefCounter.DecRef.
func (s *Socket) DecRef(ctx context.Context) {
s.socketRefs.DecRef(func() {
kernel.KernelFromContext(ctx).DeleteSocket(&s.vfsfd)
s.ep.Close(ctx)
if s.abstractBound {
s.namespace.AbstractSockets().Remove(s.abstractName, s)
}
if s.namespace != nil {
s.namespace.DecRef(ctx)
}
})
}
// Release implements vfs.FileDescriptionImpl.Release.
func (s *Socket) Release(ctx context.Context) {
// Release only decrements a reference on s because s may be referenced in
// the abstract socket namespace.
s.DecRef(ctx)
}
// GetSockOpt implements the linux syscall getsockopt(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) GetSockOpt(t *kernel.Task, level, name int, outPtr hostarch.Addr, outLen int) (marshal.Marshallable, *syserr.Error) {
return netstack.GetSockOpt(t, s, s.ep, linux.AF_UNIX, s.ep.Type(), level, name, outPtr, outLen)
}
// blockingAccept implements a blocking version of accept(2), that is, if no
// connections are ready to be accept, it will block until one becomes ready.
func (s *Socket) blockingAccept(t *kernel.Task, peerAddr *transport.Address) (transport.Endpoint, *syserr.Error) {
// Register for notifications.
e, ch := waiter.NewChannelEntry(waiter.ReadableEvents)
s.EventRegister(&e)
defer s.EventUnregister(&e)
// Try to accept the connection; if it fails, then wait until we get a
// notification.
for {
if ep, err := s.ep.Accept(t, peerAddr); err != syserr.ErrWouldBlock {
return ep, err
}
if err := t.Block(ch); err != nil {
return nil, syserr.FromError(err)
}
}
}
// Accept implements the linux syscall accept(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) Accept(t *kernel.Task, peerRequested bool, flags int, blocking bool) (int32, linux.SockAddr, uint32, *syserr.Error) {
var peerAddr *transport.Address
if peerRequested {
peerAddr = &transport.Address{}
}
ep, err := s.ep.Accept(t, peerAddr)
if err != nil {
if err != syserr.ErrWouldBlock || !blocking {
return 0, nil, 0, err
}
var err *syserr.Error
ep, err = s.blockingAccept(t, peerAddr)
if err != nil {
return 0, nil, 0, err
}
}
ns, err := NewSockfsFile(t, ep, s.stype)
if err != nil {
return 0, nil, 0, err
}
defer ns.DecRef(t)
if flags&linux.SOCK_NONBLOCK != 0 {
ns.SetStatusFlags(t, t.Credentials(), linux.SOCK_NONBLOCK)
}
var addr linux.SockAddr
var addrLen uint32
if peerAddr != nil {
addr, addrLen = convertAddress(*peerAddr)
}
fd, e := t.NewFDFrom(0, ns, kernel.FDFlags{
CloseOnExec: flags&linux.SOCK_CLOEXEC != 0,
})
if e != nil {
return 0, nil, 0, syserr.FromError(e)
}
t.Kernel().RecordSocket(ns)
return fd, addr, addrLen, nil
}
// Bind implements the linux syscall bind(2) for unix sockets.
func (s *Socket) Bind(t *kernel.Task, sockaddr []byte) *syserr.Error {
p, e := extractPath(sockaddr)
if e != nil {
return e
}
bep, ok := s.ep.(transport.BoundEndpoint)
if !ok {
// This socket can't be bound.
return syserr.ErrInvalidArgument
}
// If path is empty, the socket is autobound to an abstract address.
if len(p) == 0 || p[0] == 0 {
// Abstract socket. See net/unix/af_unix.c:unix_bind_abstract().
asn := s.namespace.AbstractSockets()
p, err := asn.Bind(t, p, bep, s)
if err != nil {
return err
}
name := p[1:]
if err := s.ep.Bind(transport.Address{Addr: p}); err != nil {
asn.Remove(name, s)
return err
}
// The socket has been successfully bound. We can update the following.
s.abstractName = name
s.abstractBound = true
return nil
}
// See net/unix/af_unix.c:unix_bind_bsd().
path := fspath.Parse(p)
root := t.FSContext().RootDirectory()
defer root.DecRef(t)
start := root
relPath := !path.Absolute
if relPath {
start = t.FSContext().WorkingDirectory()
defer start.DecRef(t)
}
pop := vfs.PathOperation{
Root: root,
Start: start,
Path: path,
}
stat, err := s.vfsfd.Stat(t, vfs.StatOptions{Mask: linux.STATX_MODE})
if err != nil {
return syserr.FromError(err)
}
err = t.Kernel().VFS().MknodAt(t, t.Credentials(), &pop, &vfs.MknodOptions{
Mode: linux.FileMode(linux.S_IFSOCK | uint(stat.Mode)&^t.FSContext().Umask()),
Endpoint: bep,
})
if linuxerr.Equals(linuxerr.EEXIST, err) {
return syserr.ErrAddressInUse
}
if err != nil {
return syserr.FromError(err)
}
if err := s.ep.Bind(transport.Address{Addr: p}); err != nil {
if unlinkErr := t.Kernel().VFS().UnlinkAt(t, t.Credentials(), &pop); unlinkErr != nil {
log.Warningf("failed to unlink socket file created for bind(%q): %v", p, unlinkErr)
}
return err
}
return nil
}
// Ioctl implements vfs.FileDescriptionImpl.
func (s *Socket) Ioctl(ctx context.Context, uio usermem.IO, sysno uintptr, args arch.SyscallArguments) (uintptr, error) {
return netstack.Ioctl(ctx, s.ep, uio, sysno, args)
}
// PRead implements vfs.FileDescriptionImpl.
func (s *Socket) PRead(ctx context.Context, dst usermem.IOSequence, offset int64, opts vfs.ReadOptions) (int64, error) {
return 0, linuxerr.ESPIPE
}
// Read implements vfs.FileDescriptionImpl.
func (s *Socket) Read(ctx context.Context, dst usermem.IOSequence, opts vfs.ReadOptions) (int64, error) {
// All flags other than RWF_NOWAIT should be ignored.
// TODO(gvisor.dev/issue/2601): Support RWF_NOWAIT.
if opts.Flags != 0 {
return 0, linuxerr.EOPNOTSUPP
}
if dst.NumBytes() == 0 {
return 0, nil
}
r := &EndpointReader{
Ctx: ctx,
Endpoint: s.ep,
NumRights: 0,
Peek: false,
}
n, err := dst.CopyOutFrom(ctx, r)
if r.Notify != nil {
r.Notify()
}
// Drop any unused rights messages.
for _, rm := range r.UnusedRights {
rm.Release(ctx)
}
// Drop control messages.
r.Control.Release(ctx)
return n, err
}
// PWrite implements vfs.FileDescriptionImpl.
func (s *Socket) PWrite(ctx context.Context, src usermem.IOSequence, offset int64, opts vfs.WriteOptions) (int64, error) {
return 0, linuxerr.ESPIPE
}
// Write implements vfs.FileDescriptionImpl.
func (s *Socket) Write(ctx context.Context, src usermem.IOSequence, opts vfs.WriteOptions) (int64, error) {
// All flags other than RWF_NOWAIT should be ignored.
// TODO(gvisor.dev/issue/2601): Support RWF_NOWAIT.
if opts.Flags != 0 {
return 0, linuxerr.EOPNOTSUPP
}
t := kernel.TaskFromContext(ctx)
ctrl := control.New(t, s.ep)
if src.NumBytes() == 0 {
nInt, notify, err := s.ep.SendMsg(ctx, [][]byte{}, ctrl, nil)
if notify != nil {
notify()
}
return int64(nInt), err.ToError()
}
w := &EndpointWriter{
Ctx: ctx,
Endpoint: s.ep,
Control: ctrl,
To: nil,
}
n, err := src.CopyInTo(ctx, w)
if w.Notify != nil {
w.Notify()
}
return n, err
}
// Epollable implements FileDescriptionImpl.Epollable.
func (s *Socket) Epollable() bool {
return true
}
// SetSockOpt implements the linux syscall setsockopt(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) SetSockOpt(t *kernel.Task, level int, name int, optVal []byte) *syserr.Error {
return netstack.SetSockOpt(t, s, s.ep, level, name, optVal)
}
// provider is a unix domain socket provider.
type provider struct{}
func (*provider) Socket(t *kernel.Task, stype linux.SockType, protocol int) (*vfs.FileDescription, *syserr.Error) {
// Check arguments.
if protocol != 0 && protocol != linux.AF_UNIX /* PF_UNIX */ {
return nil, syserr.ErrProtocolNotSupported
}
// Create the endpoint and socket.
var ep transport.Endpoint
switch stype {
case linux.SOCK_DGRAM, linux.SOCK_RAW:
ep = transport.NewConnectionless(t)
case linux.SOCK_SEQPACKET, linux.SOCK_STREAM:
ep = transport.NewConnectioned(t, stype, t.Kernel())
default:
return nil, syserr.ErrInvalidArgument
}
f, err := NewSockfsFile(t, ep, stype)
if err != nil {
ep.Close(t)
return nil, err
}
return f, nil
}
// Pair creates a new pair of AF_UNIX connected sockets.
func (*provider) Pair(t *kernel.Task, stype linux.SockType, protocol int) (*vfs.FileDescription, *vfs.FileDescription, *syserr.Error) {
// Check arguments.
if protocol != 0 && protocol != linux.AF_UNIX /* PF_UNIX */ {
return nil, nil, syserr.ErrProtocolNotSupported
}
switch stype {
case linux.SOCK_STREAM, linux.SOCK_DGRAM, linux.SOCK_SEQPACKET, linux.SOCK_RAW:
// Ok
default:
return nil, nil, syserr.ErrInvalidArgument
}
// Create the endpoints and sockets.
ep1, ep2 := transport.NewPair(t, stype, t.Kernel())
s1, err := NewSockfsFile(t, ep1, stype)
if err != nil {
ep1.Close(t)
ep2.Close(t)
return nil, nil, err
}
s2, err := NewSockfsFile(t, ep2, stype)
if err != nil {
s1.DecRef(t)
ep2.Close(t)
return nil, nil, err
}
return s1, s2, nil
}
func (s *Socket) isPacket() bool {
switch s.stype {
case linux.SOCK_DGRAM, linux.SOCK_SEQPACKET:
return true
case linux.SOCK_STREAM:
return false
default:
// We shouldn't have allowed any other socket types during creation.
panic(fmt.Sprintf("Invalid socket type %d", s.stype))
}
}
// Endpoint extracts the transport.Endpoint.
func (s *Socket) Endpoint() transport.Endpoint {
return s.ep
}
// extractPath extracts and validates the address.
func extractPath(sockaddr []byte) (string, *syserr.Error) {
addr, family, err := addressAndFamily(sockaddr)
if err != nil {
if err == syserr.ErrAddressFamilyNotSupported {
err = syserr.ErrInvalidArgument
}
return "", err
}
if family != linux.AF_UNIX {
return "", syserr.ErrInvalidArgument
}
// The address is trimmed by GetAddress.
p := addr.Addr
if len(p) > 0 && p[len(p)-1] == '/' {
// Weird, they tried to bind '/a/b/c/'?
return "", syserr.ErrIsDir
}
return p, nil
}
func addressAndFamily(addr []byte) (transport.Address, uint16, *syserr.Error) {
// Make sure we have at least 2 bytes for the address family.
if len(addr) < 2 {
return transport.Address{}, 0, syserr.ErrInvalidArgument
}
// Get the rest of the fields based on the address family.
switch family := hostarch.ByteOrder.Uint16(addr); family {
case linux.AF_UNIX:
path := addr[2:]
if len(path) > linux.UnixPathMax {
return transport.Address{}, family, syserr.ErrInvalidArgument
}
// Drop the terminating NUL (if one exists) and everything after
// it for filesystem (non-abstract) addresses.
if len(path) > 0 && path[0] != 0 {
if n := bytes.IndexByte(path[1:], 0); n >= 0 {
path = path[:n+1]
}
}
return transport.Address{
Addr: string(path),
}, family, nil
}
return transport.Address{}, 0, syserr.ErrAddressFamilyNotSupported
}
// GetPeerName implements the linux syscall getpeername(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) GetPeerName(t *kernel.Task) (linux.SockAddr, uint32, *syserr.Error) {
addr, err := s.ep.GetRemoteAddress()
if err != nil {
return nil, 0, syserr.TranslateNetstackError(err)
}
a, l := convertAddress(addr)
return a, l, nil
}
// GetSockName implements the linux syscall getsockname(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) GetSockName(t *kernel.Task) (linux.SockAddr, uint32, *syserr.Error) {
addr, err := s.ep.GetLocalAddress()
if err != nil {
return nil, 0, syserr.TranslateNetstackError(err)
}
a, l := convertAddress(addr)
return a, l, nil
}
// Listen implements the linux syscall listen(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) Listen(t *kernel.Task, backlog int) *syserr.Error {
return s.ep.Listen(t, backlog)
}
// extractEndpoint retrieves the transport.BoundEndpoint associated with a Unix
// socket path. The Release must be called on the transport.BoundEndpoint when
// the caller is done with it.
func (s *Socket) extractEndpoint(t *kernel.Task, sockaddr []byte) (transport.BoundEndpoint, *syserr.Error) {
path, err := extractPath(sockaddr)
if err != nil {
return nil, err
}
if path == "" {
// Not allowed.
return nil, syserr.ErrInvalidArgument
}
// Is it abstract?
if path[0] == 0 {
ep := s.namespace.AbstractSockets().BoundEndpoint(path[1:])
if ep == nil {
// No socket found.
return nil, syserr.ErrConnectionRefused
}
return ep, nil
}
p := fspath.Parse(path)
root := t.FSContext().RootDirectory()
start := root
relPath := !p.Absolute
if relPath {
start = t.FSContext().WorkingDirectory()
}
pop := vfs.PathOperation{
Root: root,
Start: start,
Path: p,
FollowFinalSymlink: true,
}
ep, e := t.Kernel().VFS().BoundEndpointAt(t, t.Credentials(), &pop, &vfs.BoundEndpointOptions{path})
root.DecRef(t)
if relPath {
start.DecRef(t)
}
if e != nil {
return nil, syserr.FromError(e)
}
return ep, nil
}
// Connect implements the linux syscall connect(2) for unix sockets.
func (s *Socket) Connect(t *kernel.Task, sockaddr []byte, blocking bool) *syserr.Error {
ep, err := s.extractEndpoint(t, sockaddr)
if err != nil {
return err
}
defer ep.Release(t)
// Connect the server endpoint.
err = s.ep.Connect(t, ep)
if err == syserr.ErrWrongProtocolForSocket {
// Linux for abstract sockets returns ErrConnectionRefused
// instead of ErrWrongProtocolForSocket.
path, _ := extractPath(sockaddr)
if len(path) > 0 && path[0] == 0 {
err = syserr.ErrConnectionRefused
}
}
return err
}
// SendMsg implements the linux syscall sendmsg(2) for unix sockets backed by
// a transport.Endpoint.
func (s *Socket) SendMsg(t *kernel.Task, src usermem.IOSequence, to []byte, flags int, haveDeadline bool, deadline ktime.Time, controlMessages socket.ControlMessages) (int, *syserr.Error) {
w := EndpointWriter{
Ctx: t,
Endpoint: s.ep,
Control: controlMessages.Unix,
To: nil,
}
if len(to) > 0 {
switch s.stype {
case linux.SOCK_SEQPACKET:
// to is ignored.
case linux.SOCK_STREAM:
if s.State() == linux.SS_CONNECTED {
return 0, syserr.ErrAlreadyConnected
}
return 0, syserr.ErrNotSupported
default:
ep, err := s.extractEndpoint(t, to)
if err != nil {
return 0, err
}
defer ep.Release(t)
w.To = ep
if ep.Passcred() && w.Control.Credentials == nil {
w.Control.Credentials = control.MakeCreds(t)
}
}
}
n, err := src.CopyInTo(t, &w)
if w.Notify != nil {
w.Notify()
}
if err != linuxerr.ErrWouldBlock || flags&linux.MSG_DONTWAIT != 0 {
return int(n), syserr.FromError(err)
}
// Only send SCM Rights once (see net/unix/af_unix.c:unix_stream_sendmsg).
w.Control.Rights = nil
// We'll have to block. Register for notification and keep trying to
// send all the data.
e, ch := waiter.NewChannelEntry(waiter.WritableEvents)
s.EventRegister(&e)
defer s.EventUnregister(&e)
total := n
for {
// Shorten src to reflect bytes previously written.
src = src.DropFirst64(n)
n, err = src.CopyInTo(t, &w)
if w.Notify != nil {
w.Notify()
}
total += n
if err != linuxerr.ErrWouldBlock {
break
}
if err = t.BlockWithDeadline(ch, haveDeadline, deadline); err != nil {
if linuxerr.Equals(linuxerr.ETIMEDOUT, err) {
err = linuxerr.ErrWouldBlock
}
break
}
}
return int(total), syserr.FromError(err)
}
// Passcred implements transport.Credentialer.Passcred.
func (s *Socket) Passcred() bool {
return s.ep.Passcred()
}
// ConnectedPasscred implements transport.Credentialer.ConnectedPasscred.
func (s *Socket) ConnectedPasscred() bool {
return s.ep.ConnectedPasscred()
}
// Readiness implements waiter.Waitable.Readiness.
func (s *Socket) Readiness(mask waiter.EventMask) waiter.EventMask {
return s.ep.Readiness(mask)
}
// EventRegister implements waiter.Waitable.EventRegister.
func (s *Socket) EventRegister(e *waiter.Entry) error {
return s.ep.EventRegister(e)
}
// EventUnregister implements waiter.Waitable.EventUnregister.
func (s *Socket) EventUnregister(e *waiter.Entry) {
s.ep.EventUnregister(e)
}
// Shutdown implements the linux syscall shutdown(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) Shutdown(t *kernel.Task, how int) *syserr.Error {
f, err := netstack.ConvertShutdown(how)
if err != nil {
return err
}
// Issue shutdown request.
return s.ep.Shutdown(f)
}
// RecvMsg implements the linux syscall recvmsg(2) for sockets backed by
// a transport.Endpoint.
func (s *Socket) RecvMsg(t *kernel.Task, dst usermem.IOSequence, flags int, haveDeadline bool, deadline ktime.Time, senderRequested bool, controlDataLen uint64) (n int, msgFlags int, senderAddr linux.SockAddr, senderAddrLen uint32, controlMessages socket.ControlMessages, err *syserr.Error) {
trunc := flags&linux.MSG_TRUNC != 0
peek := flags&linux.MSG_PEEK != 0
dontWait := flags&linux.MSG_DONTWAIT != 0
waitAll := flags&linux.MSG_WAITALL != 0
isPacket := s.isPacket()
// Calculate the number of FDs for which we have space and if we are
// requesting credentials.
var wantCreds bool
rightsLen := int(controlDataLen) - unix.SizeofCmsghdr
if s.Passcred() {
// Credentials take priority if they are enabled and there is space.
wantCreds = rightsLen > 0
if !wantCreds {
msgFlags |= linux.MSG_CTRUNC
}
credLen := unix.CmsgSpace(unix.SizeofUcred)
rightsLen -= credLen
}
// FDs are 32 bit (4 byte) ints.
numRights := rightsLen / 4
if numRights < 0 {
numRights = 0
}
r := EndpointReader{
Ctx: t,
Endpoint: s.ep,
Creds: wantCreds,
NumRights: numRights,
Peek: peek,
}
doRead := func() (int64, error) {
n, err := dst.CopyOutFrom(t, &r)
if r.Notify != nil {
r.Notify()
}
return n, err
}
// Drop any unused rights messages after reading.
defer func() {
for _, rm := range r.UnusedRights {
rm.Release(t)
}
}()
// If MSG_TRUNC is set with a zero byte destination then we still need
// to read the message and discard it, or in the case where MSG_PEEK is
// set, leave it be. In both cases the full message length must be
// returned.
if trunc && dst.Addrs.NumBytes() == 0 {
doRead = func() (int64, error) {
err := r.Truncate()
// Always return zero for bytes read since the destination size is
// zero.
return 0, err
}
}
var total int64
if n, err := doRead(); err != linuxerr.ErrWouldBlock || dontWait {
var from linux.SockAddr
var fromLen uint32
if senderRequested && len([]byte(r.From.Addr)) != 0 {
from, fromLen = convertAddress(r.From)
}
if r.ControlTrunc {
msgFlags |= linux.MSG_CTRUNC
}
if err != nil || dontWait || !waitAll || isPacket || n >= dst.NumBytes() {
if isPacket && n < int64(r.MsgSize) {
msgFlags |= linux.MSG_TRUNC
}
if trunc {
n = int64(r.MsgSize)
}
return int(n), msgFlags, from, fromLen, socket.ControlMessages{Unix: r.Control}, syserr.FromError(err)
}
// Don't overwrite any data we received.
dst = dst.DropFirst64(n)
total += n
}
// We'll have to block. Register for notification and keep trying to
// send all the data.
e, ch := waiter.NewChannelEntry(waiter.ReadableEvents)
s.EventRegister(&e)
defer s.EventUnregister(&e)
for {
if n, err := doRead(); err != linuxerr.ErrWouldBlock {
var from linux.SockAddr
var fromLen uint32
if senderRequested {
from, fromLen = convertAddress(r.From)
}
if r.ControlTrunc {
msgFlags |= linux.MSG_CTRUNC
}
if trunc {
// n and r.MsgSize are the same for streams.
total += int64(r.MsgSize)
} else {
total += n
}
streamPeerClosed := s.stype == linux.SOCK_STREAM && n == 0 && err == nil
if err != nil || !waitAll || isPacket || n >= dst.NumBytes() || streamPeerClosed {
if total > 0 {
err = nil
}
if isPacket && n < int64(r.MsgSize) {
msgFlags |= linux.MSG_TRUNC
}
return int(total), msgFlags, from, fromLen, socket.ControlMessages{Unix: r.Control}, syserr.FromError(err)
}
// Don't overwrite any data we received.
dst = dst.DropFirst64(n)
}
if err := t.BlockWithDeadline(ch, haveDeadline, deadline); err != nil {
if total > 0 {
err = nil
}
if linuxerr.Equals(linuxerr.ETIMEDOUT, err) {
return int(total), msgFlags, nil, 0, socket.ControlMessages{}, syserr.ErrTryAgain
}
return int(total), msgFlags, nil, 0, socket.ControlMessages{}, syserr.FromError(err)
}
}
}
// State implements socket.Socket.State.
func (s *Socket) State() uint32 {
return s.ep.State()
}
// Type implements socket.Socket.Type.
func (s *Socket) Type() (family int, skType linux.SockType, protocol int) {
// Unix domain sockets always have a protocol of 0.
return linux.AF_UNIX, s.stype, 0
}
func convertAddress(addr transport.Address) (linux.SockAddr, uint32) {
var out linux.SockAddrUnix
out.Family = linux.AF_UNIX
l := len([]byte(addr.Addr))
for i := 0; i < l; i++ {
out.Path[i] = int8(addr.Addr[i])
}
// Linux returns the used length of the address struct (including the
// null terminator) for filesystem paths. The Family field is 2 bytes.
// It is sometimes allowed to exclude the null terminator if the
// address length is the max. Abstract and empty paths always return
// the full exact length.
if l == 0 || out.Path[0] == 0 || l == len(out.Path) {
return &out, uint32(2 + l)
}
return &out, uint32(3 + l)
}
func init() {
socket.RegisterProvider(linux.AF_UNIX, &provider{})
}