summaryrefslogtreecommitdiff
path: root/cpus-common.c
blob: 3e114529c917fb66b94bf07eb365b080a6085e7a (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
/*
 * CPU thread main loop - common bits for user and system mode emulation
 *
 *  Copyright (c) 2003-2005 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, see <http://www.gnu.org/licenses/>.
 */

#include "qemu/osdep.h"
#include "qemu/main-loop.h"
#include "exec/cpu-common.h"
#include "qom/cpu.h"
#include "sysemu/cpus.h"

static QemuMutex qemu_cpu_list_lock;
static QemuCond exclusive_cond;
static QemuCond exclusive_resume;
static QemuCond qemu_work_cond;

/* >= 1 if a thread is inside start_exclusive/end_exclusive.  Written
 * under qemu_cpu_list_lock, read with atomic operations.
 */
static int pending_cpus;

void qemu_init_cpu_list(void)
{
    /* This is needed because qemu_init_cpu_list is also called by the
     * child process in a fork.  */
    pending_cpus = 0;

    qemu_mutex_init(&qemu_cpu_list_lock);
    qemu_cond_init(&exclusive_cond);
    qemu_cond_init(&exclusive_resume);
    qemu_cond_init(&qemu_work_cond);
}

void cpu_list_lock(void)
{
    qemu_mutex_lock(&qemu_cpu_list_lock);
}

void cpu_list_unlock(void)
{
    qemu_mutex_unlock(&qemu_cpu_list_lock);
}

static bool cpu_index_auto_assigned;

static int cpu_get_free_index(void)
{
    CPUState *some_cpu;
    int cpu_index = 0;

    cpu_index_auto_assigned = true;
    CPU_FOREACH(some_cpu) {
        cpu_index++;
    }
    return cpu_index;
}

static void finish_safe_work(CPUState *cpu)
{
    cpu_exec_start(cpu);
    cpu_exec_end(cpu);
}

void cpu_list_add(CPUState *cpu)
{
    qemu_mutex_lock(&qemu_cpu_list_lock);
    if (cpu->cpu_index == UNASSIGNED_CPU_INDEX) {
        cpu->cpu_index = cpu_get_free_index();
        assert(cpu->cpu_index != UNASSIGNED_CPU_INDEX);
    } else {
        assert(!cpu_index_auto_assigned);
    }
    QTAILQ_INSERT_TAIL(&cpus, cpu, node);
    qemu_mutex_unlock(&qemu_cpu_list_lock);

    finish_safe_work(cpu);
}

void cpu_list_remove(CPUState *cpu)
{
    qemu_mutex_lock(&qemu_cpu_list_lock);
    if (!QTAILQ_IN_USE(cpu, node)) {
        /* there is nothing to undo since cpu_exec_init() hasn't been called */
        qemu_mutex_unlock(&qemu_cpu_list_lock);
        return;
    }

    assert(!(cpu_index_auto_assigned && cpu != QTAILQ_LAST(&cpus, CPUTailQ)));

    QTAILQ_REMOVE(&cpus, cpu, node);
    cpu->cpu_index = UNASSIGNED_CPU_INDEX;
    qemu_mutex_unlock(&qemu_cpu_list_lock);
}

struct qemu_work_item {
    struct qemu_work_item *next;
    run_on_cpu_func func;
    void *data;
    bool free, exclusive, done;
};

static void queue_work_on_cpu(CPUState *cpu, struct qemu_work_item *wi)
{
    qemu_mutex_lock(&cpu->work_mutex);
    if (cpu->queued_work_first == NULL) {
        cpu->queued_work_first = wi;
    } else {
        cpu->queued_work_last->next = wi;
    }
    cpu->queued_work_last = wi;
    wi->next = NULL;
    wi->done = false;
    qemu_mutex_unlock(&cpu->work_mutex);

    qemu_cpu_kick(cpu);
}

void do_run_on_cpu(CPUState *cpu, run_on_cpu_func func, void *data,
                   QemuMutex *mutex)
{
    struct qemu_work_item wi;

    if (qemu_cpu_is_self(cpu)) {
        func(cpu, data);
        return;
    }

    wi.func = func;
    wi.data = data;
    wi.done = false;
    wi.free = false;
    wi.exclusive = false;

    queue_work_on_cpu(cpu, &wi);
    while (!atomic_mb_read(&wi.done)) {
        CPUState *self_cpu = current_cpu;

        qemu_cond_wait(&qemu_work_cond, mutex);
        current_cpu = self_cpu;
    }
}

void async_run_on_cpu(CPUState *cpu, run_on_cpu_func func, void *data)
{
    struct qemu_work_item *wi;

    wi = g_malloc0(sizeof(struct qemu_work_item));
    wi->func = func;
    wi->data = data;
    wi->free = true;

    queue_work_on_cpu(cpu, wi);
}

/* Wait for pending exclusive operations to complete.  The CPU list lock
   must be held.  */
static inline void exclusive_idle(void)
{
    while (pending_cpus) {
        qemu_cond_wait(&exclusive_resume, &qemu_cpu_list_lock);
    }
}

/* Start an exclusive operation.
   Must only be called from outside cpu_exec.  */
void start_exclusive(void)
{
    CPUState *other_cpu;
    int running_cpus;

    qemu_mutex_lock(&qemu_cpu_list_lock);
    exclusive_idle();

    /* Make all other cpus stop executing.  */
    atomic_set(&pending_cpus, 1);

    /* Write pending_cpus before reading other_cpu->running.  */
    smp_mb();
    running_cpus = 0;
    CPU_FOREACH(other_cpu) {
        if (atomic_read(&other_cpu->running)) {
            other_cpu->has_waiter = true;
            running_cpus++;
            qemu_cpu_kick(other_cpu);
        }
    }

    atomic_set(&pending_cpus, running_cpus + 1);
    while (pending_cpus > 1) {
        qemu_cond_wait(&exclusive_cond, &qemu_cpu_list_lock);
    }

    /* Can release mutex, no one will enter another exclusive
     * section until end_exclusive resets pending_cpus to 0.
     */
    qemu_mutex_unlock(&qemu_cpu_list_lock);
}

/* Finish an exclusive operation.  */
void end_exclusive(void)
{
    qemu_mutex_lock(&qemu_cpu_list_lock);
    atomic_set(&pending_cpus, 0);
    qemu_cond_broadcast(&exclusive_resume);
    qemu_mutex_unlock(&qemu_cpu_list_lock);
}

/* Wait for exclusive ops to finish, and begin cpu execution.  */
void cpu_exec_start(CPUState *cpu)
{
    atomic_set(&cpu->running, true);

    /* Write cpu->running before reading pending_cpus.  */
    smp_mb();

    /* 1. start_exclusive saw cpu->running == true and pending_cpus >= 1.
     * After taking the lock we'll see cpu->has_waiter == true and run---not
     * for long because start_exclusive kicked us.  cpu_exec_end will
     * decrement pending_cpus and signal the waiter.
     *
     * 2. start_exclusive saw cpu->running == false but pending_cpus >= 1.
     * This includes the case when an exclusive item is running now.
     * Then we'll see cpu->has_waiter == false and wait for the item to
     * complete.
     *
     * 3. pending_cpus == 0.  Then start_exclusive is definitely going to
     * see cpu->running == true, and it will kick the CPU.
     */
    if (unlikely(atomic_read(&pending_cpus))) {
        qemu_mutex_lock(&qemu_cpu_list_lock);
        if (!cpu->has_waiter) {
            /* Not counted in pending_cpus, let the exclusive item
             * run.  Since we have the lock, just set cpu->running to true
             * while holding it; no need to check pending_cpus again.
             */
            atomic_set(&cpu->running, false);
            exclusive_idle();
            /* Now pending_cpus is zero.  */
            atomic_set(&cpu->running, true);
        } else {
            /* Counted in pending_cpus, go ahead and release the
             * waiter at cpu_exec_end.
             */
        }
        qemu_mutex_unlock(&qemu_cpu_list_lock);
    }
}

/* Mark cpu as not executing, and release pending exclusive ops.  */
void cpu_exec_end(CPUState *cpu)
{
    atomic_set(&cpu->running, false);

    /* Write cpu->running before reading pending_cpus.  */
    smp_mb();

    /* 1. start_exclusive saw cpu->running == true.  Then it will increment
     * pending_cpus and wait for exclusive_cond.  After taking the lock
     * we'll see cpu->has_waiter == true.
     *
     * 2. start_exclusive saw cpu->running == false but here pending_cpus >= 1.
     * This includes the case when an exclusive item started after setting
     * cpu->running to false and before we read pending_cpus.  Then we'll see
     * cpu->has_waiter == false and not touch pending_cpus.  The next call to
     * cpu_exec_start will run exclusive_idle if still necessary, thus waiting
     * for the item to complete.
     *
     * 3. pending_cpus == 0.  Then start_exclusive is definitely going to
     * see cpu->running == false, and it can ignore this CPU until the
     * next cpu_exec_start.
     */
    if (unlikely(atomic_read(&pending_cpus))) {
        qemu_mutex_lock(&qemu_cpu_list_lock);
        if (cpu->has_waiter) {
            cpu->has_waiter = false;
            atomic_set(&pending_cpus, pending_cpus - 1);
            if (pending_cpus == 1) {
                qemu_cond_signal(&exclusive_cond);
            }
        }
        qemu_mutex_unlock(&qemu_cpu_list_lock);
    }
}

void async_safe_run_on_cpu(CPUState *cpu, run_on_cpu_func func, void *data)
{
    struct qemu_work_item *wi;

    wi = g_malloc0(sizeof(struct qemu_work_item));
    wi->func = func;
    wi->data = data;
    wi->free = true;
    wi->exclusive = true;

    queue_work_on_cpu(cpu, wi);
}

void process_queued_cpu_work(CPUState *cpu)
{
    struct qemu_work_item *wi;

    if (cpu->queued_work_first == NULL) {
        return;
    }

    qemu_mutex_lock(&cpu->work_mutex);
    while (cpu->queued_work_first != NULL) {
        wi = cpu->queued_work_first;
        cpu->queued_work_first = wi->next;
        if (!cpu->queued_work_first) {
            cpu->queued_work_last = NULL;
        }
        qemu_mutex_unlock(&cpu->work_mutex);
        if (wi->exclusive) {
            /* Running work items outside the BQL avoids the following deadlock:
             * 1) start_exclusive() is called with the BQL taken while another
             * CPU is running; 2) cpu_exec in the other CPU tries to takes the
             * BQL, so it goes to sleep; start_exclusive() is sleeping too, so
             * neither CPU can proceed.
             */
            qemu_mutex_unlock_iothread();
            start_exclusive();
            wi->func(cpu, wi->data);
            end_exclusive();
            qemu_mutex_lock_iothread();
        } else {
            wi->func(cpu, wi->data);
        }
        qemu_mutex_lock(&cpu->work_mutex);
        if (wi->free) {
            g_free(wi);
        } else {
            atomic_mb_set(&wi->done, true);
        }
    }
    qemu_mutex_unlock(&cpu->work_mutex);
    qemu_cond_broadcast(&qemu_work_cond);
}