1 |
/* |
2 |
* libetp implementation |
3 |
* |
4 |
* Copyright (c) 2007,2008,2009,2010,2011,2012,2013 Marc Alexander Lehmann <libetp@schmorp.de> |
5 |
* All rights reserved. |
6 |
* |
7 |
* Redistribution and use in source and binary forms, with or without modifica- |
8 |
* tion, are permitted provided that the following conditions are met: |
9 |
* |
10 |
* 1. Redistributions of source code must retain the above copyright notice, |
11 |
* this list of conditions and the following disclaimer. |
12 |
* |
13 |
* 2. Redistributions in binary form must reproduce the above copyright |
14 |
* notice, this list of conditions and the following disclaimer in the |
15 |
* documentation and/or other materials provided with the distribution. |
16 |
* |
17 |
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED |
18 |
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER- |
19 |
* CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO |
20 |
* EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE- |
21 |
* CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
22 |
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; |
23 |
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, |
24 |
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH- |
25 |
* ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED |
26 |
* OF THE POSSIBILITY OF SUCH DAMAGE. |
27 |
* |
28 |
* Alternatively, the contents of this file may be used under the terms of |
29 |
* the GNU General Public License ("GPL") version 2 or any later version, |
30 |
* in which case the provisions of the GPL are applicable instead of |
31 |
* the above. If you wish to allow the use of your version of this file |
32 |
* only under the terms of the GPL and not to allow others to use your |
33 |
* version of this file under the BSD license, indicate your decision |
34 |
* by deleting the provisions above and replace them with the notice |
35 |
* and other provisions required by the GPL. If you do not delete the |
36 |
* provisions above, a recipient may use your version of this file under |
37 |
* either the BSD or the GPL. |
38 |
*/ |
39 |
|
40 |
#ifndef ETP_API_DECL |
41 |
# define ETP_API_DECL static |
42 |
#endif |
43 |
|
44 |
#ifndef ETP_PRI_MIN |
45 |
# define ETP_PRI_MIN 0 |
46 |
# define ETP_PRI_MAX 0 |
47 |
#endif |
48 |
|
49 |
#ifndef ETP_TYPE_QUIT |
50 |
# define ETP_TYPE_QUIT 0 |
51 |
#endif |
52 |
|
53 |
#ifndef ETP_TYPE_GROUP |
54 |
# define ETP_TYPE_GROUP 1 |
55 |
#endif |
56 |
|
57 |
#define ETP_NUM_PRI (ETP_PRI_MAX - ETP_PRI_MIN + 1) |
58 |
|
59 |
#define ETP_TICKS ((1000000 + 1023) >> 10) |
60 |
|
61 |
/* calculate time difference in ~1/ETP_TICKS of a second */ |
62 |
ecb_inline int |
63 |
etp_tvdiff (struct timeval *tv1, struct timeval *tv2) |
64 |
{ |
65 |
return (tv2->tv_sec - tv1->tv_sec ) * ETP_TICKS |
66 |
+ ((tv2->tv_usec - tv1->tv_usec) >> 10); |
67 |
} |
68 |
|
69 |
static unsigned int started, idle, wanted = 4; |
70 |
|
71 |
static void (*want_poll_cb) (void); |
72 |
static void (*done_poll_cb) (void); |
73 |
|
74 |
static unsigned int max_poll_time; /* reslock */ |
75 |
static unsigned int max_poll_reqs; /* reslock */ |
76 |
|
77 |
static unsigned int nreqs; /* reqlock */ |
78 |
static unsigned int nready; /* reqlock */ |
79 |
static unsigned int npending; /* reqlock */ |
80 |
static unsigned int max_idle = 4; /* maximum number of threads that can idle indefinitely */ |
81 |
static unsigned int idle_timeout = 10; /* number of seconds after which an idle threads exit */ |
82 |
|
83 |
static xmutex_t wrklock; |
84 |
static xmutex_t reslock; |
85 |
static xmutex_t reqlock; |
86 |
static xcond_t reqwait; |
87 |
|
88 |
typedef struct etp_worker |
89 |
{ |
90 |
struct tmpbuf tmpbuf; |
91 |
|
92 |
/* locked by wrklock */ |
93 |
struct etp_worker *prev, *next; |
94 |
|
95 |
xthread_t tid; |
96 |
|
97 |
#ifdef ETP_WORKER_COMMON |
98 |
ETP_WORKER_COMMON |
99 |
#endif |
100 |
} etp_worker; |
101 |
|
102 |
static etp_worker wrk_first; /* NOT etp */ |
103 |
|
104 |
#define ETP_WORKER_LOCK(wrk) X_LOCK (wrklock) |
105 |
#define ETP_WORKER_UNLOCK(wrk) X_UNLOCK (wrklock) |
106 |
|
107 |
/* worker threads management */ |
108 |
|
109 |
static void |
110 |
etp_worker_clear (etp_worker *wrk) |
111 |
{ |
112 |
} |
113 |
|
114 |
static void ecb_cold |
115 |
etp_worker_free (etp_worker *wrk) |
116 |
{ |
117 |
free (wrk->tmpbuf.ptr); |
118 |
|
119 |
wrk->next->prev = wrk->prev; |
120 |
wrk->prev->next = wrk->next; |
121 |
|
122 |
free (wrk); |
123 |
} |
124 |
|
125 |
ETP_API_DECL unsigned int |
126 |
etp_nreqs (void) |
127 |
{ |
128 |
int retval; |
129 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
130 |
retval = nreqs; |
131 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
132 |
return retval; |
133 |
} |
134 |
|
135 |
ETP_API_DECL unsigned int |
136 |
etp_nready (void) |
137 |
{ |
138 |
unsigned int retval; |
139 |
|
140 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
141 |
retval = nready; |
142 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
143 |
|
144 |
return retval; |
145 |
} |
146 |
|
147 |
ETP_API_DECL unsigned int |
148 |
etp_npending (void) |
149 |
{ |
150 |
unsigned int retval; |
151 |
|
152 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
153 |
retval = npending; |
154 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
155 |
|
156 |
return retval; |
157 |
} |
158 |
|
159 |
ETP_API_DECL unsigned int |
160 |
etp_nthreads (void) |
161 |
{ |
162 |
unsigned int retval; |
163 |
|
164 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
165 |
retval = started; |
166 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
167 |
|
168 |
return retval; |
169 |
} |
170 |
|
171 |
/* |
172 |
* a somewhat faster data structure might be nice, but |
173 |
* with 8 priorities this actually needs <20 insns |
174 |
* per shift, the most expensive operation. |
175 |
*/ |
176 |
typedef struct { |
177 |
ETP_REQ *qs[ETP_NUM_PRI], *qe[ETP_NUM_PRI]; /* qstart, qend */ |
178 |
int size; |
179 |
} etp_reqq; |
180 |
|
181 |
static etp_reqq req_queue; |
182 |
static etp_reqq res_queue; |
183 |
|
184 |
static void ecb_noinline ecb_cold |
185 |
reqq_init (etp_reqq *q) |
186 |
{ |
187 |
int pri; |
188 |
|
189 |
for (pri = 0; pri < ETP_NUM_PRI; ++pri) |
190 |
q->qs[pri] = q->qe[pri] = 0; |
191 |
|
192 |
q->size = 0; |
193 |
} |
194 |
|
195 |
static int ecb_noinline |
196 |
reqq_push (etp_reqq *q, ETP_REQ *req) |
197 |
{ |
198 |
int pri = req->pri; |
199 |
req->next = 0; |
200 |
|
201 |
if (q->qe[pri]) |
202 |
{ |
203 |
q->qe[pri]->next = req; |
204 |
q->qe[pri] = req; |
205 |
} |
206 |
else |
207 |
q->qe[pri] = q->qs[pri] = req; |
208 |
|
209 |
return q->size++; |
210 |
} |
211 |
|
212 |
static ETP_REQ * ecb_noinline |
213 |
reqq_shift (etp_reqq *q) |
214 |
{ |
215 |
int pri; |
216 |
|
217 |
if (!q->size) |
218 |
return 0; |
219 |
|
220 |
--q->size; |
221 |
|
222 |
for (pri = ETP_NUM_PRI; pri--; ) |
223 |
{ |
224 |
ETP_REQ *req = q->qs[pri]; |
225 |
|
226 |
if (req) |
227 |
{ |
228 |
if (!(q->qs[pri] = (ETP_REQ *)req->next)) |
229 |
q->qe[pri] = 0; |
230 |
|
231 |
return req; |
232 |
} |
233 |
} |
234 |
|
235 |
abort (); |
236 |
} |
237 |
|
238 |
ETP_API_DECL int ecb_cold |
239 |
etp_init (void (*want_poll)(void), void (*done_poll)(void)) |
240 |
{ |
241 |
X_MUTEX_CREATE (wrklock); |
242 |
X_MUTEX_CREATE (reslock); |
243 |
X_MUTEX_CREATE (reqlock); |
244 |
X_COND_CREATE (reqwait); |
245 |
|
246 |
reqq_init (&req_queue); |
247 |
reqq_init (&res_queue); |
248 |
|
249 |
wrk_first.next = |
250 |
wrk_first.prev = &wrk_first; |
251 |
|
252 |
started = 0; |
253 |
idle = 0; |
254 |
nreqs = 0; |
255 |
nready = 0; |
256 |
npending = 0; |
257 |
|
258 |
want_poll_cb = want_poll; |
259 |
done_poll_cb = done_poll; |
260 |
|
261 |
return 0; |
262 |
} |
263 |
|
264 |
/* not yet in etp.c */ |
265 |
X_THREAD_PROC (etp_proc); |
266 |
|
267 |
static void ecb_cold |
268 |
etp_start_thread (void) |
269 |
{ |
270 |
etp_worker *wrk = calloc (1, sizeof (etp_worker)); |
271 |
|
272 |
/*TODO*/ |
273 |
assert (("unable to allocate worker thread data", wrk)); |
274 |
|
275 |
X_LOCK (wrklock); |
276 |
|
277 |
if (xthread_create (&wrk->tid, etp_proc, (void *)wrk)) |
278 |
{ |
279 |
wrk->prev = &wrk_first; |
280 |
wrk->next = wrk_first.next; |
281 |
wrk_first.next->prev = wrk; |
282 |
wrk_first.next = wrk; |
283 |
++started; |
284 |
} |
285 |
else |
286 |
free (wrk); |
287 |
|
288 |
X_UNLOCK (wrklock); |
289 |
} |
290 |
|
291 |
static void |
292 |
etp_maybe_start_thread (void) |
293 |
{ |
294 |
if (ecb_expect_true (etp_nthreads () >= wanted)) |
295 |
return; |
296 |
|
297 |
/* todo: maybe use idle here, but might be less exact */ |
298 |
if (ecb_expect_true (0 <= (int)etp_nthreads () + (int)etp_npending () - (int)etp_nreqs ())) |
299 |
return; |
300 |
|
301 |
etp_start_thread (); |
302 |
} |
303 |
|
304 |
static void ecb_cold |
305 |
etp_end_thread (void) |
306 |
{ |
307 |
ETP_REQ *req = calloc (1, sizeof (ETP_REQ)); /* will be freed by worker */ |
308 |
|
309 |
req->type = ETP_TYPE_QUIT; |
310 |
req->pri = ETP_PRI_MAX - ETP_PRI_MIN; |
311 |
|
312 |
X_LOCK (reqlock); |
313 |
reqq_push (&req_queue, req); |
314 |
X_COND_SIGNAL (reqwait); |
315 |
X_UNLOCK (reqlock); |
316 |
|
317 |
X_LOCK (wrklock); |
318 |
--started; |
319 |
X_UNLOCK (wrklock); |
320 |
} |
321 |
|
322 |
ETP_API_DECL int |
323 |
etp_poll (void) |
324 |
{ |
325 |
unsigned int maxreqs; |
326 |
unsigned int maxtime; |
327 |
struct timeval tv_start, tv_now; |
328 |
|
329 |
X_LOCK (reslock); |
330 |
maxreqs = max_poll_reqs; |
331 |
maxtime = max_poll_time; |
332 |
X_UNLOCK (reslock); |
333 |
|
334 |
if (maxtime) |
335 |
gettimeofday (&tv_start, 0); |
336 |
|
337 |
for (;;) |
338 |
{ |
339 |
ETP_REQ *req; |
340 |
|
341 |
etp_maybe_start_thread (); |
342 |
|
343 |
X_LOCK (reslock); |
344 |
req = reqq_shift (&res_queue); |
345 |
|
346 |
if (req) |
347 |
{ |
348 |
--npending; |
349 |
|
350 |
if (!res_queue.size && done_poll_cb) |
351 |
done_poll_cb (); |
352 |
} |
353 |
|
354 |
X_UNLOCK (reslock); |
355 |
|
356 |
if (!req) |
357 |
return 0; |
358 |
|
359 |
X_LOCK (reqlock); |
360 |
--nreqs; |
361 |
X_UNLOCK (reqlock); |
362 |
|
363 |
if (ecb_expect_false (req->type == ETP_TYPE_GROUP && req->size)) |
364 |
{ |
365 |
req->int1 = 1; /* mark request as delayed */ |
366 |
continue; |
367 |
} |
368 |
else |
369 |
{ |
370 |
int res = ETP_FINISH (req); |
371 |
if (ecb_expect_false (res)) |
372 |
return res; |
373 |
} |
374 |
|
375 |
if (ecb_expect_false (maxreqs && !--maxreqs)) |
376 |
break; |
377 |
|
378 |
if (maxtime) |
379 |
{ |
380 |
gettimeofday (&tv_now, 0); |
381 |
|
382 |
if (etp_tvdiff (&tv_start, &tv_now) >= maxtime) |
383 |
break; |
384 |
} |
385 |
} |
386 |
|
387 |
errno = EAGAIN; |
388 |
return -1; |
389 |
} |
390 |
|
391 |
ETP_API_DECL void |
392 |
etp_grp_cancel (ETP_REQ *grp); |
393 |
|
394 |
ETP_API_DECL void |
395 |
etp_cancel (ETP_REQ *req) |
396 |
{ |
397 |
req->cancelled = 1; |
398 |
|
399 |
etp_grp_cancel (req); |
400 |
} |
401 |
|
402 |
ETP_API_DECL void |
403 |
etp_grp_cancel (ETP_REQ *grp) |
404 |
{ |
405 |
for (grp = grp->grp_first; grp; grp = grp->grp_next) |
406 |
etp_cancel (grp); |
407 |
} |
408 |
|
409 |
ETP_API_DECL void |
410 |
etp_submit (ETP_REQ *req) |
411 |
{ |
412 |
req->pri -= ETP_PRI_MIN; |
413 |
|
414 |
if (ecb_expect_false (req->pri < ETP_PRI_MIN - ETP_PRI_MIN)) req->pri = ETP_PRI_MIN - ETP_PRI_MIN; |
415 |
if (ecb_expect_false (req->pri > ETP_PRI_MAX - ETP_PRI_MIN)) req->pri = ETP_PRI_MAX - ETP_PRI_MIN; |
416 |
|
417 |
if (ecb_expect_false (req->type == ETP_TYPE_GROUP)) |
418 |
{ |
419 |
/* I hope this is worth it :/ */ |
420 |
X_LOCK (reqlock); |
421 |
++nreqs; |
422 |
X_UNLOCK (reqlock); |
423 |
|
424 |
X_LOCK (reslock); |
425 |
|
426 |
++npending; |
427 |
|
428 |
if (!reqq_push (&res_queue, req) && want_poll_cb) |
429 |
want_poll_cb (); |
430 |
|
431 |
X_UNLOCK (reslock); |
432 |
} |
433 |
else |
434 |
{ |
435 |
X_LOCK (reqlock); |
436 |
++nreqs; |
437 |
++nready; |
438 |
reqq_push (&req_queue, req); |
439 |
X_COND_SIGNAL (reqwait); |
440 |
X_UNLOCK (reqlock); |
441 |
|
442 |
etp_maybe_start_thread (); |
443 |
} |
444 |
} |
445 |
|
446 |
ETP_API_DECL void ecb_cold |
447 |
etp_set_max_poll_time (double nseconds) |
448 |
{ |
449 |
if (WORDACCESS_UNSAFE) X_LOCK (reslock); |
450 |
max_poll_time = nseconds * ETP_TICKS; |
451 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reslock); |
452 |
} |
453 |
|
454 |
ETP_API_DECL void ecb_cold |
455 |
etp_set_max_poll_reqs (unsigned int maxreqs) |
456 |
{ |
457 |
if (WORDACCESS_UNSAFE) X_LOCK (reslock); |
458 |
max_poll_reqs = maxreqs; |
459 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reslock); |
460 |
} |
461 |
|
462 |
ETP_API_DECL void ecb_cold |
463 |
etp_set_max_idle (unsigned int nthreads) |
464 |
{ |
465 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
466 |
max_idle = nthreads; |
467 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
468 |
} |
469 |
|
470 |
ETP_API_DECL void ecb_cold |
471 |
etp_set_idle_timeout (unsigned int seconds) |
472 |
{ |
473 |
if (WORDACCESS_UNSAFE) X_LOCK (reqlock); |
474 |
idle_timeout = seconds; |
475 |
if (WORDACCESS_UNSAFE) X_UNLOCK (reqlock); |
476 |
} |
477 |
|
478 |
ETP_API_DECL void ecb_cold |
479 |
etp_set_min_parallel (unsigned int nthreads) |
480 |
{ |
481 |
if (wanted < nthreads) |
482 |
wanted = nthreads; |
483 |
} |
484 |
|
485 |
ETP_API_DECL void ecb_cold |
486 |
etp_set_max_parallel (unsigned int nthreads) |
487 |
{ |
488 |
if (wanted > nthreads) |
489 |
wanted = nthreads; |
490 |
|
491 |
while (started > wanted) |
492 |
etp_end_thread (); |
493 |
} |
494 |
|