/* Now check return values. */
static void
do_ret_di (void)
{
if (__sync_val_compare_and_swap (AL+0, 0x100000002ll, 0x1234567890ll) !=
0x100000002ll) abort ();
if (__sync_bool_compare_and_swap (AL+1, 0x200000003ll, 0x1234567890ll) !=
1) abort ();
if (__sync_lock_test_and_set (AL+2, 1) != 0) abort ();
__sync_lock_release (AL+3); /* no return value, but keep to match results. */
/* The following tests should not change the value since the
original does NOT match. */
if (__sync_val_compare_and_swap (AL+4, 0x000000002ll, 0x1234567890ll) !=
0x100000002ll) abort ();
if (__sync_val_compare_and_swap (AL+5, 0x100000000ll, 0x1234567890ll) !=
0x100000002ll) abort ();
if (__sync_bool_compare_and_swap (AL+6, 0x000000002ll, 0x1234567890ll) !=
0) abort ();
if (__sync_bool_compare_and_swap (AL+7, 0x100000000ll, 0x1234567890ll) !=
0) abort ();
if (__sync_fetch_and_add (AL+8, 1) != 0) abort ();
if (__sync_fetch_and_add (AL+9, 0xb000e0000000ll) != 0x1000e0de0000ll) abort ();
if (__sync_fetch_and_sub (AL+10, 22) != 42) abort ();
if (__sync_fetch_and_sub (AL+11, 0xb000e0000000ll) != 0xc001c0de0000ll)
abort ();
if (__sync_fetch_and_and (AL+12, 0x300000007ll) != -1ll) abort ();
if (__sync_fetch_and_or (AL+13, 0x500000009ll) != 0) abort ();
if (__sync_fetch_and_xor (AL+14, 0xe00000001ll) != 0xff00ff0000ll) abort ();
if (__sync_fetch_and_nand (AL+15, 0xa00000007ll) != -1ll) abort ();
/* These should be the same as the fetch_and_* cases except for
return value. */
if (__sync_add_and_fetch (AL+16, 1) != 1) abort ();
if (__sync_add_and_fetch (AL+17, 0xb000e0000000ll) != 0xc001c0de0000ll)
abort ();
if (__sync_sub_and_fetch (AL+18, 22) != 20) abort ();
if (__sync_sub_and_fetch (AL+19, 0xb000e0000000ll) != 0x1000e0de0000ll)
abort ();
if (__sync_and_and_fetch (AL+20, 0x300000007ll) != 0x300000007ll) abort ();
if (__sync_or_and_fetch (AL+21, 0x500000009ll) != 0x500000009ll) abort ();
if (__sync_xor_and_fetch (AL+22, 0xe00000001ll) != 0xf100ff0001ll) abort ();
if (__sync_nand_and_fetch (AL+23, 0xa00000007ll) != ~0xa00000007ll) abort ();
}
static void
do_di (void)
{
if (__sync_val_compare_and_swap(AL+0, 0, 1) != 0)
abort ();
if (__sync_val_compare_and_swap(AL+0, 0, 1) != 1)
abort ();
if (__sync_bool_compare_and_swap(AL+1, 0, 1) != 1)
abort ();
if (__sync_bool_compare_and_swap(AL+1, 0, 1) != 0)
abort ();
if (__sync_lock_test_and_set(AL+2, 1) != 0)
abort ();
if (__sync_fetch_and_add(AL+4, 1) != 0)
abort ();
if (__sync_fetch_and_add(AL+5, 4) != 0)
abort ();
if (__sync_fetch_and_add(AL+6, 22) != 0)
abort ();
if (__sync_fetch_and_sub(AL+7, 12) != 0)
abort ();
if (__sync_fetch_and_and(AL+8, 7) != -1)
abort ();
if (__sync_fetch_and_or(AL+9, 8) != 0)
abort ();
if (__sync_fetch_and_xor(AL+10, 9) != 0)
abort ();
if (__sync_fetch_and_nand(AL+11, 7) != 0)
abort ();
if (__sync_add_and_fetch(AL+12, 1) != 1)
abort ();
if (__sync_sub_and_fetch(AL+13, 12) != -12)
abort ();
if (__sync_and_and_fetch(AL+14, 7) != 7)
abort ();
if (__sync_or_and_fetch(AL+15, 8) != 8)
abort ();
if (__sync_xor_and_fetch(AL+16, 9) != 9)
abort ();
if (__sync_nand_and_fetch(AL+17, 7) != 7)
abort ();
}
static int
reserve_id(struct socket_server *ss) {
int i;
for (i=0;i<MAX_SOCKET;i++) {
int id = __sync_add_and_fetch(&(ss->alloc_id), 1);
if (id < 0) {
id = __sync_and_and_fetch(&(ss->alloc_id), 0x7fffffff);
}
struct socket *s = &ss->slot[id % MAX_SOCKET];
if (s->type == SOCKET_TYPE_INVALID) {
if (__sync_bool_compare_and_swap(&s->type, SOCKET_TYPE_INVALID, SOCKET_TYPE_RESERVE)) {
return id;
} else {
// retry
--i;
}
}
}
return -1;
}
int socket_server::reserve_id()
{
for (int i = 0; i < MAX_SOCKET; i++) {
int id = __sync_add_and_fetch(&alloc_id, 1);
if (id < 0) {
id = __sync_and_and_fetch(&alloc_id, 0x7fffffff);
}
struct socket *s = &slot[HASH_ID(id)];
if (s->type == SOCKET_TYPE_INVALID) {
if (__sync_bool_compare_and_swap(&s->type, SOCKET_TYPE_INVALID, SOCKET_TYPE_RESERVE)) {
s->id = id;
s->fd = -1;
return id;
} else {
--i;
}
}
}
return -1;
}
/* First check they work in terms of what they do to memory. */
static void
do_noret_di (void)
{
__sync_val_compare_and_swap (AL+0, 0x100000002ll, 0x1234567890ll);
__sync_bool_compare_and_swap (AL+1, 0x200000003ll, 0x1234567890ll);
__sync_lock_test_and_set (AL+2, 1);
__sync_lock_release (AL+3);
/* The following tests should not change the value since the
original does NOT match. */
__sync_val_compare_and_swap (AL+4, 0x000000002ll, 0x1234567890ll);
__sync_val_compare_and_swap (AL+5, 0x100000000ll, 0x1234567890ll);
__sync_bool_compare_and_swap (AL+6, 0x000000002ll, 0x1234567890ll);
__sync_bool_compare_and_swap (AL+7, 0x100000000ll, 0x1234567890ll);
__sync_fetch_and_add (AL+8, 1);
__sync_fetch_and_add (AL+9, 0xb000e0000000ll); /* + to both halves & carry. */
__sync_fetch_and_sub (AL+10, 22);
__sync_fetch_and_sub (AL+11, 0xb000e0000000ll);
__sync_fetch_and_and (AL+12, 0x300000007ll);
__sync_fetch_and_or (AL+13, 0x500000009ll);
__sync_fetch_and_xor (AL+14, 0xe00000001ll);
__sync_fetch_and_nand (AL+15, 0xa00000007ll);
/* These should be the same as the fetch_and_* cases except for
return value. */
__sync_add_and_fetch (AL+16, 1);
/* add to both halves & carry. */
__sync_add_and_fetch (AL+17, 0xb000e0000000ll);
__sync_sub_and_fetch (AL+18, 22);
__sync_sub_and_fetch (AL+19, 0xb000e0000000ll);
__sync_and_and_fetch (AL+20, 0x300000007ll);
__sync_or_and_fetch (AL+21, 0x500000009ll);
__sync_xor_and_fetch (AL+22, 0xe00000001ll);
__sync_nand_and_fetch (AL+23, 0xa00000007ll);
}
// 从socket池中获取一个空的socket 并为其分配一个id 2^31-1
// 在socket池中的位置 池的大小是64K socket_id的范围远大与64K
static int
reserve_id(struct socket_server *ss) {
int i;
for (i=0;i<MAX_SOCKET;i++) {
int id = __sync_add_and_fetch(&(ss->alloc_id), 1); // 原子的++
// 小于0 已经最大了 说明 从0开始
if (id < 0) {
id = __sync_and_and_fetch(&(ss->alloc_id), 0x7fffffff);
}
struct socket *s = &ss->slot[id % MAX_SOCKET];// 从socket池中取出socket
if (s->type == SOCKET_TYPE_INVALID) {
// 如果相等就交换成 SOCKET_TYPE_RESERVE 设置为已用
// 这里由于没有加锁 可能多个线程操作 所以还需要判断一次
if (__sync_bool_compare_and_swap(&s->type, SOCKET_TYPE_INVALID, SOCKET_TYPE_RESERVE)) {
return id;
}
else {
--i;// retry
}
}
}
return -1;
}
void test_op_and_fetch (void)
{
sc = __sync_add_and_fetch (&sc, uc);
uc = __sync_add_and_fetch (&uc, uc);
ss = __sync_add_and_fetch (&ss, uc);
us = __sync_add_and_fetch (&us, uc);
si = __sync_add_and_fetch (&si, uc);
ui = __sync_add_and_fetch (&ui, uc);
sc = __sync_sub_and_fetch (&sc, uc);
uc = __sync_sub_and_fetch (&uc, uc);
ss = __sync_sub_and_fetch (&ss, uc);
us = __sync_sub_and_fetch (&us, uc);
si = __sync_sub_and_fetch (&si, uc);
ui = __sync_sub_and_fetch (&ui, uc);
sc = __sync_or_and_fetch (&sc, uc);
uc = __sync_or_and_fetch (&uc, uc);
ss = __sync_or_and_fetch (&ss, uc);
us = __sync_or_and_fetch (&us, uc);
si = __sync_or_and_fetch (&si, uc);
ui = __sync_or_and_fetch (&ui, uc);
sc = __sync_xor_and_fetch (&sc, uc);
uc = __sync_xor_and_fetch (&uc, uc);
ss = __sync_xor_and_fetch (&ss, uc);
us = __sync_xor_and_fetch (&us, uc);
si = __sync_xor_and_fetch (&si, uc);
ui = __sync_xor_and_fetch (&ui, uc);
sc = __sync_and_and_fetch (&sc, uc);
uc = __sync_and_and_fetch (&uc, uc);
ss = __sync_and_and_fetch (&ss, uc);
us = __sync_and_and_fetch (&us, uc);
si = __sync_and_and_fetch (&si, uc);
ui = __sync_and_and_fetch (&ui, uc);
sc = __sync_nand_and_fetch (&sc, uc);
uc = __sync_nand_and_fetch (&uc, uc);
ss = __sync_nand_and_fetch (&ss, uc);
us = __sync_nand_and_fetch (&us, uc);
si = __sync_nand_and_fetch (&si, uc);
ui = __sync_nand_and_fetch (&ui, uc);
}
void *thread_and_and_fetch(void *arg)
{
for(int i = 0; i < 10000; ++i)
__sync_and_and_fetch((T*)&and_and_fetch_data, *(T*)arg);
pthread_exit(0);
}
static void watchdog_reset(void)
{
__sync_and_and_fetch(&watchdog_ctx.ticks, 0);
}
void jobmanager::threadDone(int tid) {
uint64_t newmask = __sync_and_and_fetch(&_runningMask, ~(1 << tid));
if (newmask == 0)
sem_post(&_sem);
}
T andAndFetch(T value) { return __sync_and_and_fetch(&_value, value); }
int timeout_set(const double seconds)
{
struct timespec now, then;
struct itimerspec when;
double next;
int timeout, i;
/* Timeout must be in the future. */
if (seconds <= 0.0)
return -1;
/* Get current time, */
if (clock_gettime(CLOCK_REALTIME, &now))
return -1;
/* and calculate when the timeout should fire. */
then = now;
timespec_add(&then, seconds);
/* Find an unused timeout. */
for (timeout = 0; timeout < TIMEOUTS; timeout++)
if (!(__sync_fetch_and_or(&timeout_state[timeout], TIMEOUT_USED) & TIMEOUT_USED))
break;
/* No unused timeouts? */
if (timeout >= TIMEOUTS)
return -1;
/* Clear all but TIMEOUT_USED from the state, */
__sync_and_and_fetch(&timeout_state[timeout], TIMEOUT_USED);
/* update the timeout details, */
timeout_time[timeout] = then;
/* and mark the timeout armable. */
__sync_or_and_fetch(&timeout_state[timeout], TIMEOUT_ARMED);
/* How long till the next timeout? */
next = seconds;
for (i = 0; i < TIMEOUTS; i++)
if ((__sync_fetch_and_or(&timeout_state[i], 0) & (TIMEOUT_USED | TIMEOUT_ARMED | TIMEOUT_PASSED)) == (TIMEOUT_USED | TIMEOUT_ARMED)) {
const double secs = timespec_diff(timeout_time[i], now);
if (secs >= 0.0 && secs < next)
next = secs;
}
/* Calculate duration when to fire the timeout next, */
timespec_set(&when.it_value, next);
when.it_interval.tv_sec = 0;
when.it_interval.tv_nsec = 0L;
/* and arm the timer. */
if (timer_settime(timeout_timer, 0, &when, NULL)) {
/* Failed. */
__sync_and_and_fetch(&timeout_state[timeout], 0);
return -1;
}
/* Return the timeout number. */
return timeout;
}
UInt16 BitAndAtomic16(UInt32 val, UInt16* ptr)
{
return __sync_and_and_fetch(ptr, val);
}
UInt8 BitAndAtomic8(UInt32 val, UInt8* ptr)
{
return __sync_and_and_fetch(ptr, val);
}
void reset()
{
push(__sync_and_and_fetch(&data->v, 0));
}
int main()
{
{
T x = HILO(5, 3);
T y = __sync_add_and_fetch(&x, DUP(1));
assert(y == HILO(6, 4));
assert(x == HILO(6, 4));
volatile T n = HILO(2, 1);
if (emscripten_has_threading_support())
{
for(int i = 0; i < NUM_THREADS; ++i) pthread_create(&thread[i], NULL, thread_add_and_fetch, (void*)&n);
for(int i = 0; i < NUM_THREADS; ++i) pthread_join(thread[i], NULL);
printf("n: %llx\n", n);
assert(n == HILO(NUM_THREADS*10000ULL+2ULL, NUM_THREADS*10000ULL+1ULL));
}
}
{
T x = HILO(15, 13);
T y = __sync_sub_and_fetch(&x, HILO(10, 10));
assert(y == HILO(5, 3));
assert(x == HILO(5, 3));
volatile T n = HILO(NUM_THREADS*10000ULL+5ULL, NUM_THREADS*10000ULL+3ULL);
if (emscripten_has_threading_support())
{
for(int i = 0; i < NUM_THREADS; ++i) pthread_create(&thread[i], NULL, thread_sub_and_fetch, (void*)&n);
for(int i = 0; i < NUM_THREADS; ++i) pthread_join(thread[i], NULL);
printf("n: %llx\n", n);
assert(n == HILO(5,3));
}
}
{
T x = HILO(32768 + 5, 5);
T y = __sync_or_and_fetch(&x, HILO(65536 + 9, 9));
assert(y == HILO(32768 + 65536 + 13, 13));
assert(x == HILO(32768 + 65536 + 13, 13));
for(int x = 0; x < 100; ++x) // Test a few times for robustness, since this test is so short-lived.
{
or_and_fetch_data = HILO(65536 + (1<<NUM_THREADS), 1<<NUM_THREADS);
if (emscripten_has_threading_support())
{
for(int i = 0; i < NUM_THREADS; ++i)
{
threadArg[i] = DUP(1 << i);
pthread_create(&thread[i], NULL, thread_or_and_fetch, (void*)&threadArg[i]);
}
for(int i = 0; i < NUM_THREADS; ++i) pthread_join(thread[i], NULL);
assert(or_and_fetch_data == HILO(65536 + (1<<(NUM_THREADS+1))-1, (1<<(NUM_THREADS+1))-1));
}
}
}
{
T x = HILO(32768 + 5, 5);
T y = __sync_and_and_fetch(&x, HILO(32768 + 9, 9));
assert(y == HILO(32768 + 1, 1));
assert(x == HILO(32768 + 1, 1));
if (emscripten_has_threading_support())
{
for(int x = 0; x < 100; ++x) // Test a few times for robustness, since this test is so short-lived.
{
and_and_fetch_data = HILO(65536 + (1<<(NUM_THREADS+1))-1, (1<<(NUM_THREADS+1))-1);
for(int i = 0; i < NUM_THREADS; ++i)
{
threadArg[i] = DUP(~(1UL<<i));
pthread_create(&thread[i], NULL, thread_and_and_fetch, (void*)&threadArg[i]);
}
for(int i = 0; i < NUM_THREADS; ++i) pthread_join(thread[i], NULL);
assert(and_and_fetch_data == HILO(65536 + (1<<NUM_THREADS), 1<<NUM_THREADS));
}
}
}
{
T x = HILO(32768 + 5, 5);
T y = __sync_xor_and_fetch(&x, HILO(16384 + 9, 9));
assert(y == HILO(32768 + 16384 + 12, 12));
assert(x == HILO(32768 + 16384 + 12, 12));
if (emscripten_has_threading_support())
{
for(int x = 0; x < 100; ++x) // Test a few times for robustness, since this test is so short-lived.
{
xor_and_fetch_data = HILO(32768 + (1<<NUM_THREADS), 1<<NUM_THREADS);
for(int i = 0; i < NUM_THREADS; ++i)
{
threadArg[i] = DUP(~(1UL<<i));
pthread_create(&thread[i], NULL, thread_xor_and_fetch, (void*)&threadArg[i]);
}
for(int i = 0; i < NUM_THREADS; ++i) pthread_join(thread[i], NULL);
assert(xor_and_fetch_data == HILO(32768 + ((1<<(NUM_THREADS+1))-1), (1<<(NUM_THREADS+1))-1));
}
}
}
// XXX NAND support does not exist in Atomics API.
#if 0
{
T x = 5;
T y = __sync_nand_and_fetch(&x, 9);
assert(y == 5);
assert(x == -2);
const int oddNThreads = NUM_THREADS-1;
for(int x = 0; x < 100; ++x) // Test a few times for robustness, since this test is so short-lived.
{
nand_and_fetch_data = 0;
for(int i = 0; i < oddNThreads; ++i) pthread_create(&thread[i], NULL, thread_nand_and_fetch, (void*)-1);
for(int i = 0; i < oddNThreads; ++i) pthread_join(thread[i], NULL);
assert(nand_and_fetch_data == -1);
}
}
#endif
#ifdef REPORT_RESULT
REPORT_RESULT(0);
#endif
}
int audio_thread(void *param)
{
SND_EVENT evnt;
int buffsize;
int samples;
int err;
char *errstr;
int active;
if((err = CreateBuffer(snd_format|PCM_RING,0, &hBuff)) != 0)
{
errstr = "Cannot create sound buffer\n\r";
goto exit_whith_error;
};
SetVolume(hBuff,-900,-900);
if((err = GetBufferSize(hBuff, &buffsize)) != 0)
{
errstr = "Cannot get buffer size\n\r";
goto exit_whith_error;
};
__sync_or_and_fetch(&threads_running,AUDIO_THREAD);
resampler_size = buffsize = buffsize/2;
samples = buffsize/4;
while( player_state != CLOSED)
{
uint32_t offset;
double event_stamp, wait_stamp;
int too_late = 0;
switch(sound_state)
{
case PREPARE:
mutex_lock(&astream.lock);
if(astream.count < buffsize*2)
{
memset(astream.buffer+astream.count,
0, buffsize*2-astream.count);
astream.count = buffsize*2;
};
SetBuffer(hBuff, astream.buffer, 0, buffsize*2);
astream.count -= buffsize*2;
if(astream.count)
memcpy(astream.buffer, astream.buffer+buffsize*2, astream.count);
mutex_unlock(&astream.lock);
SetTimeBase(hBuff, audio_base);
case PAUSE_2_PLAY:
GetTimeStamp(hBuff, &last_time_stamp);
// printf("last audio time stamp %f\n", last_time_stamp);
if((err = PlayBuffer(hBuff, 0)) !=0 )
{
errstr = "Cannot play buffer\n\r";
goto exit_whith_error;
};
active = 1;
sync_audio(hBuff, buffsize);
sound_state = PLAY;
// printf("render: set audio latency to %f\n", audio_delta);
/* breaktrough */
case PLAY:
GetNotify(&evnt);
if(evnt.code != 0xFF000001)
{
printf("invalid event code %d\n\r", evnt.code);
continue;
}
if(evnt.stream != hBuff)
{
printf("invalid stream %x hBuff= %x\n\r",
evnt.stream, hBuff);
continue;
};
offset = evnt.offset;
mutex_lock(&astream.lock);
if(astream.count < buffsize)
{
memset(astream.buffer+astream.count,
0, buffsize-astream.count);
astream.count = buffsize;
};
SetBuffer(hBuff, astream.buffer, offset, buffsize);
{
double val = 0;
int16_t *src = (int16_t*)astream.buffer;
int samples = buffsize/2;
int i;
for(i = 0, val = 0; i < samples/2; i++, src++)
if(val < abs(*src))
val= abs(*src); // * *src;
sound_level_0 = val; //sqrt(val / (samples/2));
for(i = 0, val = 0; i < samples/2; i++, src++)
if(val < abs(*src))
val= abs(*src); // * *src;
sound_level_1 = val; //sqrt(val / (samples/2));
// printf("%d\n", sound_level);
};
samples_written+= buffsize/4;
astream.count -= buffsize;
if(astream.count)
memcpy(astream.buffer, astream.buffer+buffsize, astream.count);
mutex_unlock(&astream.lock);
break;
case PLAY_2_STOP:
if( active )
{
ResetBuffer(hBuff, SND_RESET_ALL);
audio_base = -1.0;
active = 0;
}
sound_state = STOP;
break;
case PLAY_2_PAUSE:
if( active )
{
StopBuffer(hBuff);
};
sound_state = PAUSE;
case PAUSE:
case STOP:
delay(1);
};
}
__sync_and_and_fetch(&threads_running,~AUDIO_THREAD);
StopBuffer(hBuff);
DestroyBuffer(hBuff);
return 0;
exit_whith_error:
printf(errstr);
return -1;
};
void TASunlock(){
//TASstate = 0;
__sync_and_and_fetch(&TASlockt, 0); // "set(false)"
}
uint8_t x1__atomic_clear_uint8_t_bits(uint8_t *p, uint8_t v)
{
return __sync_and_and_fetch(p, ~v);
}
void NANOS_atomic ( int op, int type, void * variable, void * operand )
{
if ( ( type == 0 ) && ( op == 1 || op == 2 || op == 5 || op == 6 || op == 7) )
{ // variable has integer type and the operation is some kind of the following compound assignments:
// plus, minus, and, ior, xor
printf("info: 'atomic' construct implemented using atomic builtins.\n");
int tmp = *((int *) operand);
switch (op)
{
case 1: __sync_add_and_fetch((int *) variable, tmp);
break;
case 2: __sync_sub_and_fetch((int *) variable, tmp);
break;
case 5: __sync_and_and_fetch((int *) variable, tmp);
break;
case 6: __sync_or_and_fetch((int *) variable, tmp);
break;
case 7: __sync_xor_and_fetch((int *) variable, tmp);
break;
};
}
else if ( ( type == 0 ) && ( op == 10 || op == 11) )
{ // variable has integer type and the operation is a pre-/post- incr-/decr- ement
printf("info: 'atomic' construct implemented using atomic builtins.\n");
if (op == 10)
__sync_add_and_fetch((int *) variable, 1);
else
__sync_sub_and_fetch((int *) variable, 1);
}
else
{ // any other case
printf("info: 'atomic' construct implemented using compare and exchange.\n");
if (type == 1)
{ // Float type
// float tmp = *((float *) operand);
printf("Nanos support for Atomic access to floats is not yet implemented\n");
abort();
}
else if (type == 2)
{ // Double type
double tmp = *((double *) operand);
double oldval, newval;
unsigned int sizeof_var = sizeof(variable);
do {
oldval = *((double *) variable);
switch (op)
{
case 1: newval = oldval + tmp;
break;
case 2: newval = oldval - tmp;
break;
case 3: newval = oldval * tmp;
break;
case 4: newval = oldval / tmp;
break;
case 10: newval = oldval + 1;
break;
case 11: newval = oldval - 1;
break;
default: printf("Unhandled operation type while generating Nanos code for OpenMP atomic contruct.");
abort();
}
__sync_synchronize();
} while ( (sizeof_var == 4) ? !__sync_bool_compare_and_swap_4((double *) variable,
*(unsigned int *) &oldval,
*(unsigned int *) &newval) :
(sizeof_var == 8) ? !__sync_bool_compare_and_swap_8((double *) variable,
*(unsigned long *) &oldval,
*(unsigned long *) &newval) :
0 );
}
else
{
printf("Unhandled variable type while generating Nanos code for OpenMP atomic contruct.");
abort( );
}
}
}
