Subversion Repositories HelenOS-doc

Rev

Rev 43 | Rev 45 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
9 bondari 1
<?xml version="1.0" encoding="UTF-8"?>
41 jermar 2
<chapter id="sync">
3
  <?dbhtml filename="sync.html"?>
9 bondari 4
 
41 jermar 5
  <title>Mutual exclusion and synchronization</title>
9 bondari 6
 
41 jermar 7
  <section>
8
    <title>Introduction</title>
9 bondari 9
 
41 jermar 10
    <para>The HelenOS operating system is designed to make use of parallelism
11
    offered by hardware and to exploit concurrency of both the kernel and
12
    userspace tasks. This is achieved through multiprocessor support and
13
    several levels of multiprogramming (i.e. multitasking, multithreading and
14
    through userspace pseudo threads). However, such a highly concurrent
15
    environment needs safe and efficient ways to handle mutual exclusion and
16
    synchronization of many execution flows.</para>
17
  </section>
18
 
19
  <section>
20
    <title>Active kernel primitives</title>
21
 
9 bondari 22
    <section>
41 jermar 23
      <title>Spinlocks</title>
9 bondari 24
 
41 jermar 25
      <para>The basic mutual exclusion primitive is the spinlock. Spinlock
26
      implements busy waiting for an availability of a memory lock (i.e.
27
      simple variable) in a multiprocessor-safe manner. This safety is
28
      achieved through the use of a specialized, architecture-dependent,
29
      atomic test-and-set operation which either locks the spinlock (i.e. sets
30
      the variable) or, provided that it is already locked, leaves it
31
      unaltered. In any case, the test-and-set operation returns a value, thus
32
      signalling either success (i.e. zero return value) or failure (i.e.
33
      non-zero value) in acquiring the lock. Note that this makes the
34
      fundamental difference between the naive algorithm that doesn't use the
35
      atomic operation and the spinlock algortihm. While the naive algorithm
36
      is prone to race conditions on SMP configuratinos and thus is completely
37
      SMP-unsafe, the spinlock algorithm eliminates the possibility of race
38
      conditions and is suitable for mutual exclusion use.</para>
9 bondari 39
 
41 jermar 40
      <para>The semantics of the test-and-set operation is that the spinlock
41
      remains unavailable until this operation called on the respective
42
      spinlock returns zero. HelenOS builds two functions on top of
43
      test-and-set operation. The first is the unconditional attempt to
44
      acquire the spinlock and is called <emphasis>spinlock_lock</emphasis>.
45
      It simply loops until test-and-set returns zero. The other operation,
46
      <emphasis>spinlock_trylock</emphasis>, is the conditional lock operation
47
      and calls the test-and-set only once to find out wheter it managed to
48
      acquire the spinlock or not. The conditional operation is useful in
49
      situations in which an algorithm cannot acquire more spinlocks in the
50
      proper order and a deadlock cannot be avoided. In such a case, the
51
      algorithm would detect the danger and instead of possibly deadlocking
52
      the system it would simply release some spinlocks it already holds and
53
      retry the whole operation with the hope that it will succeed next time.
54
      The unlock operation, <emphasis>spinlock_unlock</emphasis>, is quite
55
      easy - it merely clears the spinlock variable.</para>
9 bondari 56
 
41 jermar 57
      <para>Nevertheless, there is a special issue related to hardware
58
      optimizations that modern processors implement. Particularily
59
      problematic is the out-of-order execution of instructions within the
60
      critical section protected by a spinlock. The processors are always
61
      self-consistent so that they can carry out speculatively executed
62
      instructions in the right order with regard to dependencies among those
63
      instructions. However, the dependency between instructions inside the
64
      critical section and those that implement locking and unlocking of the
65
      respective spinlock is not implicit on some processor architectures and
66
      the processor needs to be explicitly told about each occurrence of such
67
      a dependency. Therefore, HelenOS adds architecture-specific hooks to all
68
      <emphasis>spinlock_lock</emphasis>,
69
      <emphasis>spinlock_trylock</emphasis> and
70
      <emphasis>spinlock_unlock</emphasis> to prevent the instructions inside
71
      the critical section from bleeding out. On some architectures, these
72
      hooks can be a no-op because the dependencies are implicitly there
73
      because of the special properties of locking and unlocking instructions.
74
      However, other architectures need to instrument these hooks with
75
      different memory barriers, depending on what operations can bleed
76
      out.</para>
9 bondari 77
 
41 jermar 78
      <para>Spinlocks have one significant drawback: when held for longer time
79
      periods, they harm both parallelism and concurrency. Processor executing
80
      <emphasis>spinlock_lock</emphasis> does not do any fruitful work and is
81
      effectively halted until it can grab the lock and proceed. Similarily,
82
      other threads cannot execute on the processor that holds the spinlock
83
      because the kernel disables preemption on that processor when a spinlock
84
      is held. The reason behind disabling preemption is priority inversion
85
      problem avoidance. For the same reason, threads are strongly discouraged
86
      from sleeping when they hold a spinlock.</para>
9 bondari 87
 
41 jermar 88
      <para>To summarize, spinlocks represent very simple and essential mutual
89
      exclusion primitive for SMP systems. On the other hand, spinlocks scale
90
      poorly because of the active loop they are based on. Therefore,
91
      spinlocks are used in HelenOS only for a short-time mutual exclusion and
92
      in cases where the mutual exclusion is required out of thread context.
93
      Lastly, spinlocks are used in the construction of passive
94
      synchronization primitives.</para>
95
    </section>
96
  </section>
9 bondari 97
 
41 jermar 98
  <section>
99
    <title>Passive kernel synchronization</title>
9 bondari 100
 
41 jermar 101
    <section>
43 jermar 102
      <title>Wait queues</title>
9 bondari 103
 
43 jermar 104
      <para>A wait queue is the basic passive synchronization primitive on
105
      which all other passive synchronization primitives build. Simply put, it
106
      allows a thread to sleep until an event associated with the particular
107
      wait queue occurs. Multiple threads are notified about incoming events
108
      in first come, first served fashion. Moreover, should the event come
109
      before any thread waits for it, it is recorded in the wait queue as a
110
      missed wakeup and later forwarded to the first thread that decides to
111
      wait in the queue. The inner structures of the wait queue are protected
112
      by a spinlock.</para>
113
 
114
      <para>The thread that wants to wait for a wait queue event uses the
115
      <emphasis>waitq_sleep_timeout</emphasis> function. The algorithm then
116
      checks the wait queue's counter of missed wakeups and if there are any
117
      missed wakeups, the call returns immediately. The call also returns
118
      immediately if only a conditional wait was requested. Otherwise the
119
      thread is enqueued in the wait queue's list of sleeping threads and its
120
      state is changed to <emphasis>Sleeping</emphasis>. It then sleeps until
121
      one of the following events happens:</para>
122
 
123
      <orderedlist>
124
        <listitem>
125
          <para>another thread calls <emphasis>waitq_wakeup</emphasis> and the
126
          thread is the first thread in the wait queue's list of sleeping
127
          threads</para>
128
        </listitem>
129
 
130
        <listitem>
131
          <para>another thread calls
132
          <emphasis>waitq_interrupt_sleep</emphasis> on the sleeping
133
          thread</para>
134
        </listitem>
135
 
136
        <listitem>
137
          <para>the sleep timeouts provided that none of the previous occurred
138
          within a specified time limit; the limit can be infinity</para>
139
        </listitem>
140
      </orderedlist>
141
 
142
      <para>All five possibilities (immediate return on success, immediate
143
      return on failure, wakeup after sleep, interruption and timeout) are
144
      distinguishable by the return value of
145
      <emphasis>waitq_sleep_timeout</emphasis>. The ability to interrupt a
146
      sleeping thread is essential for externally initiated thread termination
147
      and the ability to wait only for a certain amount of time is used, for
148
      instance, to passively delay thread execution by several microseconds or
149
      even seconds in <emphasis>thread_sleep</emphasis> function. Because all
150
      other passive kernel synchronization primitives are based on wait
151
      queues, they also have the option of being interrutped and, more
152
      importantly, can timeout. All of them also implement the conditional
153
      operation. Furthemore, this very fundamental interface reaches up to the
154
      implementation of futexes - userspace synchronization primitive, which
155
      makes it possible for a userspace thread to request synchronization
156
      operation with a timeout or a conditional operation.</para>
157
 
158
      <para>From the description above, it should be apparent, that when a
159
      sleeping thread is woken by <emphasis>waitq_wakeup</emphasis> or when
160
      <emphasis>waitq_sleep_timeout</emphasis> succeeds immediatelly, the
161
      thread can be sure the event has come and the thread need not and should
162
      not verify this fact. This approach is called direct hand-off and is
163
      characteristic for all passive HelenOS synchronization primitives with
164
      one exception described below.</para>
41 jermar 165
    </section>
9 bondari 166
 
41 jermar 167
    <section>
168
      <title>Semaphores</title>
9 bondari 169
 
43 jermar 170
      <para>The interesting point about wait queues is that the number of
171
      missed wakeups is equal to the number of threads that will not block in
172
      <emphasis>watiq_sleep_timeout</emphasis> and would immediately succeed
173
      instead. On the other hand, semaphores are synchronization primitives
174
      that will let predefined amount of threads in its critical section and
175
      block any other threads above this count. However, these two cases are
176
      exactly the same. Semaphores in HelenOS are therefore implemented as
177
      wait queues with a single semantic change: their wait queue is
178
      initialized to have so many missed wakeups as is the number of threads
179
      that the semphore intends to let into its critical section
180
      simultaneously.</para>
181
 
182
      <para>In the semaphore language, the wait queue operation
183
      <emphasis>waitq_sleep_timeout</emphasis> corresponds to
184
      <emphasis><emphasis>semaphore</emphasis> down</emphasis> operation,
185
      represented by the function <emphasis>semaphore_down_timeout</emphasis>
186
      and by way of similitude the wait queue operation waitq_wakeup
187
      corresponds to semaphore <emphasis>up</emphasis> operation, represented
188
      by the function <emphasis>sempafore_up</emphasis>. The conditional down
189
      operation is called <emphasis>semaphore_trydown</emphasis>.</para>
41 jermar 190
    </section>
9 bondari 191
 
41 jermar 192
    <section>
43 jermar 193
      <title>Mutexes</title>
9 bondari 194
 
43 jermar 195
      <para>Mutexes are are sometimes referred to as binary sempahores. That
196
      means that mutexes are like semaphores that allow only one thread in its
197
      critical section. Indeed, mutexes in HelenOS are implemented exactly in
198
      this way: they are built atop semaphores. From another point of view,
199
      they can be viewed as spinlocks without busy waiting. Their semaphore
200
      heritage provides good basics for both conditional operation and
201
      operation with timeout. The locking operation is called
202
      <emphasis>mutex_lock</emphasis>, the conditional locking operation is
203
      called <emphasis>mutex_trylock</emphasis> and the unlocking operation is
204
      called <emphasis>mutex_unlock</emphasis>.</para>
41 jermar 205
    </section>
9 bondari 206
 
41 jermar 207
    <section>
43 jermar 208
      <title>Reader/writer locks</title>
9 bondari 209
 
43 jermar 210
      <para>Reader/writer locks, or rwlocks, are by far the most complicated
211
      synchronization primitive within the kernel. The goal of these locks is
212
      to improve concurrency of applications in which threads need to
213
      synchronize access to a shared resource and that access can be
214
      partitioned into a read-only mode and a write mode. Reader/writer locks
215
      should make it possible for several, possibly many, readers to enter the
216
      critical section, provided that no writer is currently in the critical
217
      section, or to be in the critical section contemporarily. Writers are
218
      allowed to enter the critical section only individually, provided that
219
      no reader is in the critical section already. Applications in which the
220
      majority of operations can be done in the read-only mode can benefit
221
      from increased concurrency created by reader/writer locks.</para>
222
 
223
      <para>During reader/writer locks construction, a decision should be made
224
      whether readers will be prefered over writers or whether writers will be
225
      prefered over readers in cases when the lock is not currently held and
226
      both a reader and a writer want to gain the lock. Some operating systems
227
      prefer one group over the other, creating thus a possibility for
228
      starving the unprefered group. In the HelenOS operating system, none of
229
      the two groups is prefered. The lock is granted on the first come, first
230
      served basis with the additional note that readers are granted the lock
231
      in biggest possible batches.</para>
232
 
233
      <para>With this policy and the timeout modes of operation, the direct
234
      hand-off becomes much more complicated. For instance, a writer leaving
235
      the critical section must wake up all leading readers in the rwlock's
236
      wait queue or one leading writer or no-one if no thread is waiting.
237
      Similarily, the last reader leaving the critical section must wakeup the
238
      sleeping writer, if there are any sleeping threads at all. As another
239
      example, if a writer at the beginning of the rwlock's wait queue
240
      timeouts and the lock is held by at least one reader, the timeouting
241
      writer must first wake up all readers that follow him in the queue prior
242
      to signalling the timeout itself and giving up.</para>
243
 
244
      <para>Because of the issues mentioned in the previous paragraph, the
245
      reader/writer locks imlpementation needs to walk the rwlock wait queue's
246
      list of sleeping threads directly in order to find out the type of
247
      access that the queueing threads demand. This makes the code difficult
248
      to understand and dependent on the internal implementation of the wait
249
      queue. Nevertheless, it remains unclear to the authors of HelenOS
250
      whether a simpler but equivalently fair solution exists.</para>
251
 
252
      <para>The implementation of rwlocks as it has been already put, makes
253
      use of one single wait queue for both readers and writers, thus avoiding
254
      any possibility of starvation. In fact, rwlocks use a mutex rather than
255
      a bare wait queue. This mutex is called <emphasis>exclusive</emphasis>
256
      and is used to synchronize writers. The writer's lock operation,
257
      <emphasis>rwlock_write_lock_timeout</emphasis>, simply tries to acquire
258
      the exclusive mutex. If it succeeds, the writer is granted the rwlock.
44 jermar 259
      However, if the operation fails (e.g. times out), the writer must check
260
      for potential readers at the head of the list of sleeping threads
261
      associated with the mutex's wait queue and proceed according to the
262
      procedure outlined above.</para>
43 jermar 263
 
264
      <para>The exclusive mutex plays an important role in reader
265
      synchronization as well. However, a reader doing the reader's lock
266
      operation, <emphasis>rwlock_read_lock_timeout</emphasis>, may bypass
267
      this mutex when it detects that:</para>
268
 
269
      <orderedlist>
270
        <listitem>
271
          <para>there are other readers in the critical section</para>
272
        </listitem>
273
 
274
        <listitem>
275
          <para>there are no sleeping threads waiting for the exclusive
276
          mutex</para>
277
        </listitem>
278
      </orderedlist>
279
 
280
      <para>If both conditions are true, the reader will bypass the mutex,
281
      increment the number of readers in the critical section and enter the
282
      critical section. Note that if there are any sleeping threads at the
283
      beginning of the wait queue, the first of them must be a writer. If the
284
      conditions are not fulfilled, the reader normally waits until the
285
      exclusive mutex is granted to it.</para>
41 jermar 286
    </section>
9 bondari 287
 
288
    <section>
41 jermar 289
      <title>Condition variables</title>
9 bondari 290
 
41 jermar 291
      <para>Condvars explanation</para>
9 bondari 292
    </section>
41 jermar 293
  </section>
9 bondari 294
 
41 jermar 295
  <section>
296
    <title>Userspace synchronization</title>
9 bondari 297
 
41 jermar 298
    <section>
299
      <title>Futexes</title>
300
 
301
      <para></para>
302
    </section>
303
  </section>
304
</chapter>