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
//! Declarations for traits that we need our runtimes to implement.
use async_trait;
use Framed;
use stream;
use Spawn;
use ;
use Debug;
use ;
use net;
use ;
use unix;
/// A runtime for use by Tor client library code.
///
/// This trait comprises several other traits that we require all of our
/// runtimes to provide:
///
/// * [`futures::task::Spawn`] to launch new background tasks.
/// * [`SleepProvider`] to pause a task for a given amount of time.
/// * [`CoarseTimeProvider`] for a cheaper but less accurate notion of time.
/// * [`NetStreamProvider`] to launch and accept network connections.
/// * [`TlsProvider`] to launch TLS connections.
/// * [`Blocking`] to be able to run synchronous (cpubound or IO) code,
/// and *re*-enter the async context from synchronous thread
/// (This may become optional in the future, if/when we add WASM
/// support).
///
/// A value which is only `Runtime` cannot be used as an *entry point* to the runtime.
/// For that, it must also implement [`ToplevelBlockOn`],
/// making it a [`ToplevelRuntime`].
/// Since you can only [enter a runtime](ToplevelBlockOn::block_on) once,
/// typically you use a `ToplevelRuntime` to enter the runtime,
/// and use it as only a `Runtime` afterwards.
/// This means that library code should typically
/// deal with `Runtime` rather than `ToplevelRuntime`.
///
/// We require that every `Runtime` has an efficient [`Clone`] implementation
/// that gives a new opaque reference to the same underlying runtime.
///
/// Additionally, every `Runtime` is [`Send`] and [`Sync`], though these
/// requirements may be somewhat relaxed in the future.
///
/// At some future point,
/// Arti may require that the runtime `impl<S> TlsProvider<S>` (for suitable`S`),
/// rather than just for their own `TcpStream`s.
/// I.e., Arti may start to require that the runtime's TLS provider can wrap any streams,
/// not only the runtime's own TCP streams.
/// This might be expressed as an additional supertrait bound on `Runtime`,
/// eg when Rust supports GATs,
/// or as an additional bound on the Arti APIs that currently use `Runtime`.
/// For API future compatibility, if you `impl Runtime for MyRuntime`,
/// you should also ensure that you
/// ```ignore
/// impl<S> TlsProvider<S> for MyRuntime
/// where S: futures::AsyncRead + futures::AsyncWrite + Unpin + Send + 'static
/// ```
//
/// Perhaps we will need this if we make our own TLS connections *through* Tor,
/// rather than just channels to guards.
/// A runtime that we can use to run Tor as a client.
/// * [`ToplevelBlockOn`] to block on a top-level future and run it to completion
/// (This may become optional in the future, if/when we add WASM
/// support).
///
/// Trait for a runtime that can wait until a timer has expired.
///
/// Every `SleepProvider` also implements
/// [`SleepProviderExt`](crate::SleepProviderExt); see that trait
/// for other useful functions.
/// A provider of reduced-precision timestamps
///
/// This doesn't provide any facility for sleeping.
/// If you want to sleep based on reduced-precision timestamps,
/// convert the desired sleep duration to `std::time::Duration`
/// and use [`SleepProvider`].
/// Trait for a runtime that can be entered to block on a toplevel future.
///
/// This trait is *not* implied by `Runtime`, only by `ToplevelRuntime`.
/// `ToplevelRuntime` is available at the toplevel of each program,
/// typically, where a concrete async executor is selected.
/// Support for interacting with blocking (non-async) code
///
/// This supports two use cases: blocking IO and CPU-intensive activities.
/// (In both of these cases, simply calling the functions within an `async` task
/// is a bad idea, because that can block the whole async runtime.)
///
/// ### Blocking IO
///
/// `Blocking` can be used to interact with libraries or OS primitives
/// that only offer a synchronous, blocking, interface.
///
/// Use [`spawn_blocking`](Blocking::spawn_blocking)
/// when it is convenient to have a long-running thread,
/// for these operations.
///
/// Use [`blocking_io`](Blocking::blocking_io)
/// when the blocking code is usually expected to complete quickly,
/// and/or you will be switching back and forth a lot
/// between sync and async contexts.
/// Note that you cannot call back to async code from within `blocking_io`.
///
/// ### CPU-intensive activities
///
/// Perform CPU-intensive work, that ought not to block the program's main loop,
/// via [`Blocking::spawn_blocking`].
///
/// `spawn_blocking` does not apply any limiting or prioritisation;
/// its threads simply compete for CPU with other threads in the program.
/// That must be done by the caller; therefore:
///
/// **Limit the number of cpu threads** spawned
/// in order to limit the total amount of CPU time consumed by any part of the program.
/// For example, consider using one CPU thread per Tor Hidden Service.
///
/// It is most performant to spawn a long-running thread,
/// rather than to repeatedly spawn short-lived threads for individual work items.
/// This also makes it easier to limit the number of concurrente cpu threads.
/// For the same reason, [`Blocking::blocking_io`] should be avoided
/// for the CPU-intensive use case.
///
/// ### Mapping to concrete functions from underlying libraries
///
/// The semantics of `Blocking` are heavily influenced by Tokio
/// and by the desire to be able to use tor-rtmock's `MockExecutor` to test Arti code.
///
/// | `tor-rtcompat` | Tokio | `MockExecutor` |
/// |------------------------------|-----------------------|--------------------------------|
/// | `ToplevelBlockOn::block_on` | `Runtime::block_on` | `ToplevelBlockOn::block_on` |
/// | `Blocking::spawn_blocking` | `task::spawn_blocking` | `subthread_spawn` |
/// | `Blocking::reenter_block_on` | `Handle::block_on` | `subthread_block_on_future` |
/// | `Blocking::blocking_io` | `block_in_place` | `subthread_spawn` |
/// | (not available) | (not implemented) | `progress_until_stalled` etc. |
///
/// Re `block_on`, see also the docs for the underlying implementations in
/// [tokio][tokio-threadpool] and
/// [async-std][async-std-threadpool].
///
/// [tokio-threadpool]: https://docs.rs/tokio/latest/tokio/task/fn.spawn_blocking.html
/// [async-std-threadpool]: https://docs.rs/async-std/latest/async_std/task/fn.spawn_blocking.html
/// Trait providing additional operations on network sockets.
/// A [`StreamOps`] handle that always returns an error.
///
/// Returned from [`StreamOps::new_handle`] for types and platforms
/// that do not support `StreamOps`.
;
/// Error: Tried to perform a [`StreamOps`] operation on an unsupported stream type
/// or on an unsupported platform.
///
/// (For example, you can't call [`StreamOps::set_tcp_notsent_lowat`] on Windows
/// or on a stream type that is not backed by a TCP socket.)
/// Trait for a runtime that can create and accept connections
/// over network sockets.
///
/// (In Arti we use the [`AsyncRead`] and [`AsyncWrite`] traits from
/// [`futures::io`] as more standard, even though the ones from Tokio
/// can be a bit more efficient. Let's hope that they converge in the
/// future.)
// TODO: Use of async_trait is not ideal, since we have to box with every
// call. Still, async_io basically makes that necessary :/
/// Trait for a local socket that accepts incoming streams.
///
/// These objects are returned by instances of [`NetStreamProvider`]. To use
/// one,
/// use `incoming` to convert this object into a [`stream::Stream`].
/// Trait for a runtime that can send and receive UDP datagrams.
/// Trait for a locally bound Udp socket that can send and receive datagrams.
///
/// These objects are returned by instances of [`UdpProvider`].
//
// NOTE that UdpSocket objects are _necessarily_ un-connected. If you need to
// implement a connected Udp socket in the future, please make a new trait (and
// a new type.)
/// An object with a peer certificate: typically a TLS connection.
/// An object that knows how to wrap a TCP connection (where the type of said TCP
/// connection is `S`) with TLS.
///
/// # Usage notes
///
/// Note that because of Tor's peculiarities, this is not a
/// general-purpose TLS type. Unlike typical users, Tor does not want
/// its TLS library to check whether the certificates used in TLS are signed
/// within the web PKI hierarchy, or what their hostnames are, or even whether
/// they are valid. It *does*, however, check that the subject public key in the
/// certificate is indeed correctly used to authenticate the TLS handshake.
///
/// If you are implementing something other than Tor, this is **not** the
/// functionality you want.
///
/// How can this behavior be remotely safe, even in Tor? It only works for Tor
/// because the certificate that a Tor relay uses in TLS is not actually being
/// used to certify that relay's public key. Instead, the certificate only used
/// as a container for the relay's public key. The real certification happens
/// later, inside the TLS session, when the relay presents a CERTS cell.
///
/// Such sneakiness was especially necessary before TLS 1.3, which encrypts more
/// of the handshake, and before pluggable transports, which make
/// "innocuous-looking TLS handshakes" less important than they once were. Once
/// TLS 1.3 is completely ubiquitous, we might be able to specify a simpler link
/// handshake than Tor uses now.
/// Trait for a runtime that knows how to create TLS connections over
/// TCP streams of type `S`.
///
/// This is separate from [`TlsConnector`] because eventually we may
/// eventually want to support multiple `TlsConnector` implementations
/// that use a single [`Runtime`].
///
/// See the [`TlsConnector`] documentation for a discussion of the Tor-specific
/// limitations of this trait: If you are implementing something other than Tor,
/// this is **not** the functionality you want.