Commit | Line | Data |
---|---|---|
734efb46 | 1 | /* |
2 | * linux/kernel/time/clocksource.c | |
3 | * | |
4 | * This file contains the functions which manage clocksource drivers. | |
5 | * | |
6 | * Copyright (C) 2004, 2005 IBM, John Stultz (johnstul@us.ibm.com) | |
7 | * | |
8 | * This program is free software; you can redistribute it and/or modify | |
9 | * it under the terms of the GNU General Public License as published by | |
10 | * the Free Software Foundation; either version 2 of the License, or | |
11 | * (at your option) any later version. | |
12 | * | |
13 | * This program is distributed in the hope that it will be useful, | |
14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | * GNU General Public License for more details. | |
17 | * | |
18 | * You should have received a copy of the GNU General Public License | |
19 | * along with this program; if not, write to the Free Software | |
20 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
21 | * | |
22 | * TODO WishList: | |
23 | * o Allow clocksource drivers to be unregistered | |
734efb46 | 24 | */ |
25 | ||
26 | #include <linux/clocksource.h> | |
27 | #include <linux/sysdev.h> | |
28 | #include <linux/init.h> | |
29 | #include <linux/module.h> | |
dc29a365 | 30 | #include <linux/sched.h> /* for spin_unlock_irq() using preempt_count() m68k */ |
79bf2bb3 | 31 | #include <linux/tick.h> |
01548f4d | 32 | #include <linux/kthread.h> |
734efb46 | 33 | |
a038a353 PO |
34 | void timecounter_init(struct timecounter *tc, |
35 | const struct cyclecounter *cc, | |
36 | u64 start_tstamp) | |
37 | { | |
38 | tc->cc = cc; | |
39 | tc->cycle_last = cc->read(cc); | |
40 | tc->nsec = start_tstamp; | |
41 | } | |
3586e0a9 | 42 | EXPORT_SYMBOL_GPL(timecounter_init); |
a038a353 PO |
43 | |
44 | /** | |
45 | * timecounter_read_delta - get nanoseconds since last call of this function | |
46 | * @tc: Pointer to time counter | |
47 | * | |
48 | * When the underlying cycle counter runs over, this will be handled | |
49 | * correctly as long as it does not run over more than once between | |
50 | * calls. | |
51 | * | |
52 | * The first call to this function for a new time counter initializes | |
53 | * the time tracking and returns an undefined result. | |
54 | */ | |
55 | static u64 timecounter_read_delta(struct timecounter *tc) | |
56 | { | |
57 | cycle_t cycle_now, cycle_delta; | |
58 | u64 ns_offset; | |
59 | ||
60 | /* read cycle counter: */ | |
61 | cycle_now = tc->cc->read(tc->cc); | |
62 | ||
63 | /* calculate the delta since the last timecounter_read_delta(): */ | |
64 | cycle_delta = (cycle_now - tc->cycle_last) & tc->cc->mask; | |
65 | ||
66 | /* convert to nanoseconds: */ | |
67 | ns_offset = cyclecounter_cyc2ns(tc->cc, cycle_delta); | |
68 | ||
69 | /* update time stamp of timecounter_read_delta() call: */ | |
70 | tc->cycle_last = cycle_now; | |
71 | ||
72 | return ns_offset; | |
73 | } | |
74 | ||
75 | u64 timecounter_read(struct timecounter *tc) | |
76 | { | |
77 | u64 nsec; | |
78 | ||
79 | /* increment time by nanoseconds since last call */ | |
80 | nsec = timecounter_read_delta(tc); | |
81 | nsec += tc->nsec; | |
82 | tc->nsec = nsec; | |
83 | ||
84 | return nsec; | |
85 | } | |
3586e0a9 | 86 | EXPORT_SYMBOL_GPL(timecounter_read); |
a038a353 PO |
87 | |
88 | u64 timecounter_cyc2time(struct timecounter *tc, | |
89 | cycle_t cycle_tstamp) | |
90 | { | |
91 | u64 cycle_delta = (cycle_tstamp - tc->cycle_last) & tc->cc->mask; | |
92 | u64 nsec; | |
93 | ||
94 | /* | |
95 | * Instead of always treating cycle_tstamp as more recent | |
96 | * than tc->cycle_last, detect when it is too far in the | |
97 | * future and treat it as old time stamp instead. | |
98 | */ | |
99 | if (cycle_delta > tc->cc->mask / 2) { | |
100 | cycle_delta = (tc->cycle_last - cycle_tstamp) & tc->cc->mask; | |
101 | nsec = tc->nsec - cyclecounter_cyc2ns(tc->cc, cycle_delta); | |
102 | } else { | |
103 | nsec = cyclecounter_cyc2ns(tc->cc, cycle_delta) + tc->nsec; | |
104 | } | |
105 | ||
106 | return nsec; | |
107 | } | |
3586e0a9 | 108 | EXPORT_SYMBOL_GPL(timecounter_cyc2time); |
a038a353 | 109 | |
7d2f944a TG |
110 | /** |
111 | * clocks_calc_mult_shift - calculate mult/shift factors for scaled math of clocks | |
112 | * @mult: pointer to mult variable | |
113 | * @shift: pointer to shift variable | |
114 | * @from: frequency to convert from | |
115 | * @to: frequency to convert to | |
5fdade95 | 116 | * @maxsec: guaranteed runtime conversion range in seconds |
7d2f944a TG |
117 | * |
118 | * The function evaluates the shift/mult pair for the scaled math | |
119 | * operations of clocksources and clockevents. | |
120 | * | |
121 | * @to and @from are frequency values in HZ. For clock sources @to is | |
122 | * NSEC_PER_SEC == 1GHz and @from is the counter frequency. For clock | |
123 | * event @to is the counter frequency and @from is NSEC_PER_SEC. | |
124 | * | |
5fdade95 | 125 | * The @maxsec conversion range argument controls the time frame in |
7d2f944a TG |
126 | * seconds which must be covered by the runtime conversion with the |
127 | * calculated mult and shift factors. This guarantees that no 64bit | |
128 | * overflow happens when the input value of the conversion is | |
129 | * multiplied with the calculated mult factor. Larger ranges may | |
130 | * reduce the conversion accuracy by chosing smaller mult and shift | |
131 | * factors. | |
132 | */ | |
133 | void | |
5fdade95 | 134 | clocks_calc_mult_shift(u32 *mult, u32 *shift, u32 from, u32 to, u32 maxsec) |
7d2f944a TG |
135 | { |
136 | u64 tmp; | |
137 | u32 sft, sftacc= 32; | |
138 | ||
139 | /* | |
140 | * Calculate the shift factor which is limiting the conversion | |
141 | * range: | |
142 | */ | |
5fdade95 | 143 | tmp = ((u64)maxsec * from) >> 32; |
7d2f944a TG |
144 | while (tmp) { |
145 | tmp >>=1; | |
146 | sftacc--; | |
147 | } | |
148 | ||
149 | /* | |
150 | * Find the conversion shift/mult pair which has the best | |
151 | * accuracy and fits the maxsec conversion range: | |
152 | */ | |
153 | for (sft = 32; sft > 0; sft--) { | |
154 | tmp = (u64) to << sft; | |
b5776c4a | 155 | tmp += from / 2; |
7d2f944a TG |
156 | do_div(tmp, from); |
157 | if ((tmp >> sftacc) == 0) | |
158 | break; | |
159 | } | |
160 | *mult = tmp; | |
161 | *shift = sft; | |
162 | } | |
163 | ||
734efb46 | 164 | /*[Clocksource internal variables]--------- |
165 | * curr_clocksource: | |
f1b82746 | 166 | * currently selected clocksource. |
734efb46 | 167 | * clocksource_list: |
168 | * linked list with the registered clocksources | |
75c5158f MS |
169 | * clocksource_mutex: |
170 | * protects manipulations to curr_clocksource and the clocksource_list | |
734efb46 | 171 | * override_name: |
172 | * Name of the user-specified clocksource. | |
173 | */ | |
f1b82746 | 174 | static struct clocksource *curr_clocksource; |
734efb46 | 175 | static LIST_HEAD(clocksource_list); |
75c5158f | 176 | static DEFINE_MUTEX(clocksource_mutex); |
734efb46 | 177 | static char override_name[32]; |
54a6bc0b | 178 | static int finished_booting; |
734efb46 | 179 | |
5d8b34fd | 180 | #ifdef CONFIG_CLOCKSOURCE_WATCHDOG |
f79e0258 MS |
181 | static void clocksource_watchdog_work(struct work_struct *work); |
182 | ||
5d8b34fd TG |
183 | static LIST_HEAD(watchdog_list); |
184 | static struct clocksource *watchdog; | |
185 | static struct timer_list watchdog_timer; | |
f79e0258 | 186 | static DECLARE_WORK(watchdog_work, clocksource_watchdog_work); |
5d8b34fd | 187 | static DEFINE_SPINLOCK(watchdog_lock); |
fb63a0eb | 188 | static int watchdog_running; |
b52f52a0 | 189 | |
01548f4d | 190 | static int clocksource_watchdog_kthread(void *data); |
d0981a1b | 191 | static void __clocksource_change_rating(struct clocksource *cs, int rating); |
c55c87c8 | 192 | |
5d8b34fd | 193 | /* |
35c35d1a | 194 | * Interval: 0.5sec Threshold: 0.0625s |
5d8b34fd TG |
195 | */ |
196 | #define WATCHDOG_INTERVAL (HZ >> 1) | |
35c35d1a | 197 | #define WATCHDOG_THRESHOLD (NSEC_PER_SEC >> 4) |
5d8b34fd | 198 | |
01548f4d MS |
199 | static void clocksource_watchdog_work(struct work_struct *work) |
200 | { | |
201 | /* | |
202 | * If kthread_run fails the next watchdog scan over the | |
203 | * watchdog_list will find the unstable clock again. | |
204 | */ | |
205 | kthread_run(clocksource_watchdog_kthread, NULL, "kwatchdog"); | |
206 | } | |
207 | ||
7285dd7f | 208 | static void __clocksource_unstable(struct clocksource *cs) |
5d8b34fd | 209 | { |
5d8b34fd | 210 | cs->flags &= ~(CLOCK_SOURCE_VALID_FOR_HRES | CLOCK_SOURCE_WATCHDOG); |
c55c87c8 | 211 | cs->flags |= CLOCK_SOURCE_UNSTABLE; |
54a6bc0b TG |
212 | if (finished_booting) |
213 | schedule_work(&watchdog_work); | |
5d8b34fd TG |
214 | } |
215 | ||
7285dd7f TG |
216 | static void clocksource_unstable(struct clocksource *cs, int64_t delta) |
217 | { | |
218 | printk(KERN_WARNING "Clocksource %s unstable (delta = %Ld ns)\n", | |
219 | cs->name, delta); | |
220 | __clocksource_unstable(cs); | |
221 | } | |
222 | ||
223 | /** | |
224 | * clocksource_mark_unstable - mark clocksource unstable via watchdog | |
225 | * @cs: clocksource to be marked unstable | |
226 | * | |
227 | * This function is called instead of clocksource_change_rating from | |
228 | * cpu hotplug code to avoid a deadlock between the clocksource mutex | |
229 | * and the cpu hotplug mutex. It defers the update of the clocksource | |
230 | * to the watchdog thread. | |
231 | */ | |
232 | void clocksource_mark_unstable(struct clocksource *cs) | |
233 | { | |
234 | unsigned long flags; | |
235 | ||
236 | spin_lock_irqsave(&watchdog_lock, flags); | |
237 | if (!(cs->flags & CLOCK_SOURCE_UNSTABLE)) { | |
238 | if (list_empty(&cs->wd_list)) | |
239 | list_add(&cs->wd_list, &watchdog_list); | |
240 | __clocksource_unstable(cs); | |
241 | } | |
242 | spin_unlock_irqrestore(&watchdog_lock, flags); | |
243 | } | |
244 | ||
5d8b34fd TG |
245 | static void clocksource_watchdog(unsigned long data) |
246 | { | |
c55c87c8 | 247 | struct clocksource *cs; |
5d8b34fd TG |
248 | cycle_t csnow, wdnow; |
249 | int64_t wd_nsec, cs_nsec; | |
c55c87c8 | 250 | int next_cpu; |
5d8b34fd TG |
251 | |
252 | spin_lock(&watchdog_lock); | |
fb63a0eb MS |
253 | if (!watchdog_running) |
254 | goto out; | |
5d8b34fd | 255 | |
c55c87c8 MS |
256 | list_for_each_entry(cs, &watchdog_list, wd_list) { |
257 | ||
258 | /* Clocksource already marked unstable? */ | |
01548f4d | 259 | if (cs->flags & CLOCK_SOURCE_UNSTABLE) { |
54a6bc0b TG |
260 | if (finished_booting) |
261 | schedule_work(&watchdog_work); | |
c55c87c8 | 262 | continue; |
01548f4d | 263 | } |
c55c87c8 | 264 | |
b5199515 | 265 | local_irq_disable(); |
8e19608e | 266 | csnow = cs->read(cs); |
b5199515 TG |
267 | wdnow = watchdog->read(watchdog); |
268 | local_irq_enable(); | |
b52f52a0 | 269 | |
8cf4e750 MS |
270 | /* Clocksource initialized ? */ |
271 | if (!(cs->flags & CLOCK_SOURCE_WATCHDOG)) { | |
272 | cs->flags |= CLOCK_SOURCE_WATCHDOG; | |
b5199515 TG |
273 | cs->wd_last = wdnow; |
274 | cs->cs_last = csnow; | |
b52f52a0 TG |
275 | continue; |
276 | } | |
277 | ||
b5199515 TG |
278 | wd_nsec = clocksource_cyc2ns((wdnow - cs->wd_last) & watchdog->mask, |
279 | watchdog->mult, watchdog->shift); | |
280 | ||
281 | cs_nsec = clocksource_cyc2ns((csnow - cs->cs_last) & | |
155ec602 | 282 | cs->mask, cs->mult, cs->shift); |
b5199515 TG |
283 | cs->cs_last = csnow; |
284 | cs->wd_last = wdnow; | |
285 | ||
286 | /* Check the deviation from the watchdog clocksource. */ | |
8cf4e750 MS |
287 | if (abs(cs_nsec - wd_nsec) > WATCHDOG_THRESHOLD) { |
288 | clocksource_unstable(cs, cs_nsec - wd_nsec); | |
289 | continue; | |
290 | } | |
291 | ||
292 | if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && | |
293 | (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS) && | |
294 | (watchdog->flags & CLOCK_SOURCE_IS_CONTINUOUS)) { | |
295 | cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES; | |
296 | /* | |
297 | * We just marked the clocksource as highres-capable, | |
298 | * notify the rest of the system as well so that we | |
299 | * transition into high-res mode: | |
300 | */ | |
301 | tick_clock_notify(); | |
5d8b34fd TG |
302 | } |
303 | } | |
304 | ||
c55c87c8 MS |
305 | /* |
306 | * Cycle through CPUs to check if the CPUs stay synchronized | |
307 | * to each other. | |
308 | */ | |
309 | next_cpu = cpumask_next(raw_smp_processor_id(), cpu_online_mask); | |
310 | if (next_cpu >= nr_cpu_ids) | |
311 | next_cpu = cpumask_first(cpu_online_mask); | |
312 | watchdog_timer.expires += WATCHDOG_INTERVAL; | |
313 | add_timer_on(&watchdog_timer, next_cpu); | |
fb63a0eb | 314 | out: |
5d8b34fd TG |
315 | spin_unlock(&watchdog_lock); |
316 | } | |
0f8e8ef7 | 317 | |
fb63a0eb MS |
318 | static inline void clocksource_start_watchdog(void) |
319 | { | |
320 | if (watchdog_running || !watchdog || list_empty(&watchdog_list)) | |
321 | return; | |
322 | init_timer(&watchdog_timer); | |
323 | watchdog_timer.function = clocksource_watchdog; | |
fb63a0eb MS |
324 | watchdog_timer.expires = jiffies + WATCHDOG_INTERVAL; |
325 | add_timer_on(&watchdog_timer, cpumask_first(cpu_online_mask)); | |
326 | watchdog_running = 1; | |
327 | } | |
328 | ||
329 | static inline void clocksource_stop_watchdog(void) | |
330 | { | |
331 | if (!watchdog_running || (watchdog && !list_empty(&watchdog_list))) | |
332 | return; | |
333 | del_timer(&watchdog_timer); | |
334 | watchdog_running = 0; | |
335 | } | |
336 | ||
0f8e8ef7 MS |
337 | static inline void clocksource_reset_watchdog(void) |
338 | { | |
339 | struct clocksource *cs; | |
340 | ||
341 | list_for_each_entry(cs, &watchdog_list, wd_list) | |
342 | cs->flags &= ~CLOCK_SOURCE_WATCHDOG; | |
343 | } | |
344 | ||
b52f52a0 TG |
345 | static void clocksource_resume_watchdog(void) |
346 | { | |
0f8e8ef7 MS |
347 | unsigned long flags; |
348 | ||
7b7422a5 TG |
349 | /* |
350 | * We use trylock here to avoid a potential dead lock when | |
351 | * kgdb calls this code after the kernel has been stopped with | |
352 | * watchdog_lock held. When watchdog_lock is held we just | |
353 | * return and accept, that the watchdog might trigger and mark | |
354 | * the monitored clock source (usually TSC) unstable. | |
355 | * | |
356 | * This does not affect the other caller clocksource_resume() | |
357 | * because at this point the kernel is UP, interrupts are | |
358 | * disabled and nothing can hold watchdog_lock. | |
359 | */ | |
360 | if (!spin_trylock_irqsave(&watchdog_lock, flags)) | |
361 | return; | |
0f8e8ef7 MS |
362 | clocksource_reset_watchdog(); |
363 | spin_unlock_irqrestore(&watchdog_lock, flags); | |
b52f52a0 TG |
364 | } |
365 | ||
fb63a0eb | 366 | static void clocksource_enqueue_watchdog(struct clocksource *cs) |
5d8b34fd | 367 | { |
5d8b34fd TG |
368 | unsigned long flags; |
369 | ||
370 | spin_lock_irqsave(&watchdog_lock, flags); | |
371 | if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) { | |
fb63a0eb | 372 | /* cs is a clocksource to be watched. */ |
5d8b34fd | 373 | list_add(&cs->wd_list, &watchdog_list); |
fb63a0eb | 374 | cs->flags &= ~CLOCK_SOURCE_WATCHDOG; |
948ac6d7 | 375 | } else { |
fb63a0eb | 376 | /* cs is a watchdog. */ |
948ac6d7 | 377 | if (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS) |
5d8b34fd | 378 | cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES; |
fb63a0eb | 379 | /* Pick the best watchdog. */ |
5d8b34fd | 380 | if (!watchdog || cs->rating > watchdog->rating) { |
5d8b34fd | 381 | watchdog = cs; |
5d8b34fd | 382 | /* Reset watchdog cycles */ |
0f8e8ef7 | 383 | clocksource_reset_watchdog(); |
5d8b34fd TG |
384 | } |
385 | } | |
fb63a0eb MS |
386 | /* Check if the watchdog timer needs to be started. */ |
387 | clocksource_start_watchdog(); | |
5d8b34fd TG |
388 | spin_unlock_irqrestore(&watchdog_lock, flags); |
389 | } | |
fb63a0eb MS |
390 | |
391 | static void clocksource_dequeue_watchdog(struct clocksource *cs) | |
392 | { | |
393 | struct clocksource *tmp; | |
394 | unsigned long flags; | |
395 | ||
396 | spin_lock_irqsave(&watchdog_lock, flags); | |
397 | if (cs->flags & CLOCK_SOURCE_MUST_VERIFY) { | |
398 | /* cs is a watched clocksource. */ | |
399 | list_del_init(&cs->wd_list); | |
400 | } else if (cs == watchdog) { | |
401 | /* Reset watchdog cycles */ | |
402 | clocksource_reset_watchdog(); | |
403 | /* Current watchdog is removed. Find an alternative. */ | |
404 | watchdog = NULL; | |
405 | list_for_each_entry(tmp, &clocksource_list, list) { | |
406 | if (tmp == cs || tmp->flags & CLOCK_SOURCE_MUST_VERIFY) | |
407 | continue; | |
408 | if (!watchdog || tmp->rating > watchdog->rating) | |
409 | watchdog = tmp; | |
410 | } | |
411 | } | |
412 | cs->flags &= ~CLOCK_SOURCE_WATCHDOG; | |
413 | /* Check if the watchdog timer needs to be stopped. */ | |
414 | clocksource_stop_watchdog(); | |
415 | spin_unlock_irqrestore(&watchdog_lock, flags); | |
416 | } | |
417 | ||
01548f4d | 418 | static int clocksource_watchdog_kthread(void *data) |
c55c87c8 MS |
419 | { |
420 | struct clocksource *cs, *tmp; | |
421 | unsigned long flags; | |
6ea41d25 | 422 | LIST_HEAD(unstable); |
c55c87c8 | 423 | |
d0981a1b | 424 | mutex_lock(&clocksource_mutex); |
c55c87c8 MS |
425 | spin_lock_irqsave(&watchdog_lock, flags); |
426 | list_for_each_entry_safe(cs, tmp, &watchdog_list, wd_list) | |
427 | if (cs->flags & CLOCK_SOURCE_UNSTABLE) { | |
428 | list_del_init(&cs->wd_list); | |
6ea41d25 | 429 | list_add(&cs->wd_list, &unstable); |
c55c87c8 MS |
430 | } |
431 | /* Check if the watchdog timer needs to be stopped. */ | |
432 | clocksource_stop_watchdog(); | |
6ea41d25 TG |
433 | spin_unlock_irqrestore(&watchdog_lock, flags); |
434 | ||
435 | /* Needs to be done outside of watchdog lock */ | |
436 | list_for_each_entry_safe(cs, tmp, &unstable, wd_list) { | |
437 | list_del_init(&cs->wd_list); | |
d0981a1b | 438 | __clocksource_change_rating(cs, 0); |
6ea41d25 | 439 | } |
d0981a1b | 440 | mutex_unlock(&clocksource_mutex); |
01548f4d | 441 | return 0; |
c55c87c8 MS |
442 | } |
443 | ||
fb63a0eb MS |
444 | #else /* CONFIG_CLOCKSOURCE_WATCHDOG */ |
445 | ||
446 | static void clocksource_enqueue_watchdog(struct clocksource *cs) | |
5d8b34fd TG |
447 | { |
448 | if (cs->flags & CLOCK_SOURCE_IS_CONTINUOUS) | |
449 | cs->flags |= CLOCK_SOURCE_VALID_FOR_HRES; | |
450 | } | |
b52f52a0 | 451 | |
fb63a0eb | 452 | static inline void clocksource_dequeue_watchdog(struct clocksource *cs) { } |
b52f52a0 | 453 | static inline void clocksource_resume_watchdog(void) { } |
54a6bc0b | 454 | static inline int clocksource_watchdog_kthread(void *data) { return 0; } |
fb63a0eb MS |
455 | |
456 | #endif /* CONFIG_CLOCKSOURCE_WATCHDOG */ | |
5d8b34fd | 457 | |
c54a42b1 MD |
458 | /** |
459 | * clocksource_suspend - suspend the clocksource(s) | |
460 | */ | |
461 | void clocksource_suspend(void) | |
462 | { | |
463 | struct clocksource *cs; | |
464 | ||
465 | list_for_each_entry_reverse(cs, &clocksource_list, list) | |
466 | if (cs->suspend) | |
467 | cs->suspend(cs); | |
468 | } | |
469 | ||
b52f52a0 TG |
470 | /** |
471 | * clocksource_resume - resume the clocksource(s) | |
472 | */ | |
473 | void clocksource_resume(void) | |
474 | { | |
2e197586 | 475 | struct clocksource *cs; |
b52f52a0 | 476 | |
75c5158f | 477 | list_for_each_entry(cs, &clocksource_list, list) |
b52f52a0 | 478 | if (cs->resume) |
17622339 | 479 | cs->resume(cs); |
b52f52a0 TG |
480 | |
481 | clocksource_resume_watchdog(); | |
b52f52a0 TG |
482 | } |
483 | ||
7c3078b6 JW |
484 | /** |
485 | * clocksource_touch_watchdog - Update watchdog | |
486 | * | |
487 | * Update the watchdog after exception contexts such as kgdb so as not | |
7b7422a5 TG |
488 | * to incorrectly trip the watchdog. This might fail when the kernel |
489 | * was stopped in code which holds watchdog_lock. | |
7c3078b6 JW |
490 | */ |
491 | void clocksource_touch_watchdog(void) | |
492 | { | |
493 | clocksource_resume_watchdog(); | |
494 | } | |
495 | ||
98962465 JH |
496 | /** |
497 | * clocksource_max_deferment - Returns max time the clocksource can be deferred | |
498 | * @cs: Pointer to clocksource | |
499 | * | |
500 | */ | |
501 | static u64 clocksource_max_deferment(struct clocksource *cs) | |
502 | { | |
503 | u64 max_nsecs, max_cycles; | |
504 | ||
505 | /* | |
506 | * Calculate the maximum number of cycles that we can pass to the | |
507 | * cyc2ns function without overflowing a 64-bit signed result. The | |
508 | * maximum number of cycles is equal to ULLONG_MAX/cs->mult which | |
509 | * is equivalent to the below. | |
510 | * max_cycles < (2^63)/cs->mult | |
511 | * max_cycles < 2^(log2((2^63)/cs->mult)) | |
512 | * max_cycles < 2^(log2(2^63) - log2(cs->mult)) | |
513 | * max_cycles < 2^(63 - log2(cs->mult)) | |
514 | * max_cycles < 1 << (63 - log2(cs->mult)) | |
515 | * Please note that we add 1 to the result of the log2 to account for | |
516 | * any rounding errors, ensure the above inequality is satisfied and | |
517 | * no overflow will occur. | |
518 | */ | |
519 | max_cycles = 1ULL << (63 - (ilog2(cs->mult) + 1)); | |
520 | ||
521 | /* | |
522 | * The actual maximum number of cycles we can defer the clocksource is | |
523 | * determined by the minimum of max_cycles and cs->mask. | |
524 | */ | |
525 | max_cycles = min_t(u64, max_cycles, (u64) cs->mask); | |
526 | max_nsecs = clocksource_cyc2ns(max_cycles, cs->mult, cs->shift); | |
527 | ||
528 | /* | |
529 | * To ensure that the clocksource does not wrap whilst we are idle, | |
530 | * limit the time the clocksource can be deferred by 12.5%. Please | |
531 | * note a margin of 12.5% is used because this can be computed with | |
532 | * a shift, versus say 10% which would require division. | |
533 | */ | |
534 | return max_nsecs - (max_nsecs >> 5); | |
535 | } | |
536 | ||
592913ec | 537 | #ifndef CONFIG_ARCH_USES_GETTIMEOFFSET |
734efb46 | 538 | |
734efb46 | 539 | /** |
f1b82746 | 540 | * clocksource_select - Select the best clocksource available |
734efb46 | 541 | * |
75c5158f | 542 | * Private function. Must hold clocksource_mutex when called. |
734efb46 | 543 | * |
92c7e002 TG |
544 | * Select the clocksource with the best rating, or the clocksource, |
545 | * which is selected by userspace override. | |
734efb46 | 546 | */ |
f1b82746 | 547 | static void clocksource_select(void) |
734efb46 | 548 | { |
f1b82746 | 549 | struct clocksource *best, *cs; |
5d8b34fd | 550 | |
75c5158f | 551 | if (!finished_booting || list_empty(&clocksource_list)) |
f1b82746 MS |
552 | return; |
553 | /* First clocksource on the list has the best rating. */ | |
554 | best = list_first_entry(&clocksource_list, struct clocksource, list); | |
555 | /* Check for the override clocksource. */ | |
556 | list_for_each_entry(cs, &clocksource_list, list) { | |
557 | if (strcmp(cs->name, override_name) != 0) | |
558 | continue; | |
559 | /* | |
560 | * Check to make sure we don't switch to a non-highres | |
561 | * capable clocksource if the tick code is in oneshot | |
562 | * mode (highres or nohz) | |
563 | */ | |
564 | if (!(cs->flags & CLOCK_SOURCE_VALID_FOR_HRES) && | |
565 | tick_oneshot_mode_active()) { | |
566 | /* Override clocksource cannot be used. */ | |
567 | printk(KERN_WARNING "Override clocksource %s is not " | |
568 | "HRT compatible. Cannot switch while in " | |
569 | "HRT/NOHZ mode\n", cs->name); | |
570 | override_name[0] = 0; | |
571 | } else | |
572 | /* Override clocksource can be used. */ | |
573 | best = cs; | |
574 | break; | |
575 | } | |
75c5158f MS |
576 | if (curr_clocksource != best) { |
577 | printk(KERN_INFO "Switching to clocksource %s\n", best->name); | |
578 | curr_clocksource = best; | |
579 | timekeeping_notify(curr_clocksource); | |
580 | } | |
f1b82746 | 581 | } |
734efb46 | 582 | |
592913ec | 583 | #else /* !CONFIG_ARCH_USES_GETTIMEOFFSET */ |
54a6bc0b TG |
584 | |
585 | static inline void clocksource_select(void) { } | |
586 | ||
587 | #endif | |
588 | ||
75c5158f MS |
589 | /* |
590 | * clocksource_done_booting - Called near the end of core bootup | |
591 | * | |
592 | * Hack to avoid lots of clocksource churn at boot time. | |
593 | * We use fs_initcall because we want this to start before | |
594 | * device_initcall but after subsys_initcall. | |
595 | */ | |
596 | static int __init clocksource_done_booting(void) | |
597 | { | |
ad6759fb | 598 | mutex_lock(&clocksource_mutex); |
599 | curr_clocksource = clocksource_default_clock(); | |
600 | mutex_unlock(&clocksource_mutex); | |
601 | ||
75c5158f | 602 | finished_booting = 1; |
54a6bc0b TG |
603 | |
604 | /* | |
605 | * Run the watchdog first to eliminate unstable clock sources | |
606 | */ | |
607 | clocksource_watchdog_kthread(NULL); | |
608 | ||
e6c73305 | 609 | mutex_lock(&clocksource_mutex); |
75c5158f | 610 | clocksource_select(); |
e6c73305 | 611 | mutex_unlock(&clocksource_mutex); |
75c5158f MS |
612 | return 0; |
613 | } | |
614 | fs_initcall(clocksource_done_booting); | |
615 | ||
92c7e002 TG |
616 | /* |
617 | * Enqueue the clocksource sorted by rating | |
734efb46 | 618 | */ |
f1b82746 | 619 | static void clocksource_enqueue(struct clocksource *cs) |
734efb46 | 620 | { |
f1b82746 MS |
621 | struct list_head *entry = &clocksource_list; |
622 | struct clocksource *tmp; | |
92c7e002 | 623 | |
f1b82746 | 624 | list_for_each_entry(tmp, &clocksource_list, list) |
92c7e002 | 625 | /* Keep track of the place, where to insert */ |
f1b82746 MS |
626 | if (tmp->rating >= cs->rating) |
627 | entry = &tmp->list; | |
628 | list_add(&cs->list, entry); | |
734efb46 | 629 | } |
630 | ||
d7e81c26 | 631 | /** |
852db46d | 632 | * __clocksource_updatefreq_scale - Used update clocksource with new freq |
d7e81c26 JS |
633 | * @t: clocksource to be registered |
634 | * @scale: Scale factor multiplied against freq to get clocksource hz | |
635 | * @freq: clocksource frequency (cycles per second) divided by scale | |
636 | * | |
852db46d | 637 | * This should only be called from the clocksource->enable() method. |
d7e81c26 JS |
638 | * |
639 | * This *SHOULD NOT* be called directly! Please use the | |
852db46d | 640 | * clocksource_updatefreq_hz() or clocksource_updatefreq_khz helper functions. |
d7e81c26 | 641 | */ |
852db46d | 642 | void __clocksource_updatefreq_scale(struct clocksource *cs, u32 scale, u32 freq) |
d7e81c26 | 643 | { |
c0e299b1 | 644 | u64 sec; |
724ed53e | 645 | |
d7e81c26 | 646 | /* |
724ed53e TG |
647 | * Calc the maximum number of seconds which we can run before |
648 | * wrapping around. For clocksources which have a mask > 32bit | |
649 | * we need to limit the max sleep time to have a good | |
650 | * conversion precision. 10 minutes is still a reasonable | |
651 | * amount. That results in a shift value of 24 for a | |
652 | * clocksource with mask >= 40bit and f >= 4GHz. That maps to | |
653 | * ~ 0.06ppm granularity for NTP. We apply the same 12.5% | |
654 | * margin as we do in clocksource_max_deferment() | |
d7e81c26 | 655 | */ |
724ed53e TG |
656 | sec = (cs->mask - (cs->mask >> 5)); |
657 | do_div(sec, freq); | |
658 | do_div(sec, scale); | |
659 | if (!sec) | |
660 | sec = 1; | |
661 | else if (sec > 600 && cs->mask > UINT_MAX) | |
662 | sec = 600; | |
663 | ||
d7e81c26 | 664 | clocks_calc_mult_shift(&cs->mult, &cs->shift, freq, |
724ed53e | 665 | NSEC_PER_SEC / scale, sec * scale); |
d7e81c26 | 666 | cs->max_idle_ns = clocksource_max_deferment(cs); |
852db46d JS |
667 | } |
668 | EXPORT_SYMBOL_GPL(__clocksource_updatefreq_scale); | |
669 | ||
670 | /** | |
671 | * __clocksource_register_scale - Used to install new clocksources | |
672 | * @t: clocksource to be registered | |
673 | * @scale: Scale factor multiplied against freq to get clocksource hz | |
674 | * @freq: clocksource frequency (cycles per second) divided by scale | |
675 | * | |
676 | * Returns -EBUSY if registration fails, zero otherwise. | |
677 | * | |
678 | * This *SHOULD NOT* be called directly! Please use the | |
679 | * clocksource_register_hz() or clocksource_register_khz helper functions. | |
680 | */ | |
681 | int __clocksource_register_scale(struct clocksource *cs, u32 scale, u32 freq) | |
682 | { | |
683 | ||
b595076a | 684 | /* Initialize mult/shift and max_idle_ns */ |
852db46d | 685 | __clocksource_updatefreq_scale(cs, scale, freq); |
d7e81c26 | 686 | |
852db46d | 687 | /* Add clocksource to the clcoksource list */ |
d7e81c26 JS |
688 | mutex_lock(&clocksource_mutex); |
689 | clocksource_enqueue(cs); | |
d7e81c26 | 690 | clocksource_enqueue_watchdog(cs); |
e05b2efb | 691 | clocksource_select(); |
d7e81c26 JS |
692 | mutex_unlock(&clocksource_mutex); |
693 | return 0; | |
694 | } | |
695 | EXPORT_SYMBOL_GPL(__clocksource_register_scale); | |
696 | ||
697 | ||
734efb46 | 698 | /** |
a2752549 | 699 | * clocksource_register - Used to install new clocksources |
734efb46 | 700 | * @t: clocksource to be registered |
701 | * | |
702 | * Returns -EBUSY if registration fails, zero otherwise. | |
703 | */ | |
f1b82746 | 704 | int clocksource_register(struct clocksource *cs) |
734efb46 | 705 | { |
98962465 JH |
706 | /* calculate max idle time permitted for this clocksource */ |
707 | cs->max_idle_ns = clocksource_max_deferment(cs); | |
708 | ||
75c5158f | 709 | mutex_lock(&clocksource_mutex); |
f1b82746 | 710 | clocksource_enqueue(cs); |
fb63a0eb | 711 | clocksource_enqueue_watchdog(cs); |
e05b2efb | 712 | clocksource_select(); |
75c5158f | 713 | mutex_unlock(&clocksource_mutex); |
f1b82746 | 714 | return 0; |
734efb46 | 715 | } |
a2752549 | 716 | EXPORT_SYMBOL(clocksource_register); |
734efb46 | 717 | |
d0981a1b TG |
718 | static void __clocksource_change_rating(struct clocksource *cs, int rating) |
719 | { | |
720 | list_del(&cs->list); | |
721 | cs->rating = rating; | |
722 | clocksource_enqueue(cs); | |
723 | clocksource_select(); | |
724 | } | |
725 | ||
734efb46 | 726 | /** |
92c7e002 | 727 | * clocksource_change_rating - Change the rating of a registered clocksource |
734efb46 | 728 | */ |
92c7e002 | 729 | void clocksource_change_rating(struct clocksource *cs, int rating) |
734efb46 | 730 | { |
75c5158f | 731 | mutex_lock(&clocksource_mutex); |
d0981a1b | 732 | __clocksource_change_rating(cs, rating); |
75c5158f | 733 | mutex_unlock(&clocksource_mutex); |
734efb46 | 734 | } |
fb63a0eb | 735 | EXPORT_SYMBOL(clocksource_change_rating); |
734efb46 | 736 | |
4713e22c TG |
737 | /** |
738 | * clocksource_unregister - remove a registered clocksource | |
739 | */ | |
740 | void clocksource_unregister(struct clocksource *cs) | |
741 | { | |
75c5158f | 742 | mutex_lock(&clocksource_mutex); |
fb63a0eb | 743 | clocksource_dequeue_watchdog(cs); |
4713e22c | 744 | list_del(&cs->list); |
f1b82746 | 745 | clocksource_select(); |
75c5158f | 746 | mutex_unlock(&clocksource_mutex); |
4713e22c | 747 | } |
fb63a0eb | 748 | EXPORT_SYMBOL(clocksource_unregister); |
4713e22c | 749 | |
2b013700 | 750 | #ifdef CONFIG_SYSFS |
734efb46 | 751 | /** |
752 | * sysfs_show_current_clocksources - sysfs interface for current clocksource | |
753 | * @dev: unused | |
754 | * @buf: char buffer to be filled with clocksource list | |
755 | * | |
756 | * Provides sysfs interface for listing current clocksource. | |
757 | */ | |
758 | static ssize_t | |
4a0b2b4d AK |
759 | sysfs_show_current_clocksources(struct sys_device *dev, |
760 | struct sysdev_attribute *attr, char *buf) | |
734efb46 | 761 | { |
5e2cb101 | 762 | ssize_t count = 0; |
734efb46 | 763 | |
75c5158f | 764 | mutex_lock(&clocksource_mutex); |
5e2cb101 | 765 | count = snprintf(buf, PAGE_SIZE, "%s\n", curr_clocksource->name); |
75c5158f | 766 | mutex_unlock(&clocksource_mutex); |
734efb46 | 767 | |
5e2cb101 | 768 | return count; |
734efb46 | 769 | } |
770 | ||
771 | /** | |
772 | * sysfs_override_clocksource - interface for manually overriding clocksource | |
773 | * @dev: unused | |
774 | * @buf: name of override clocksource | |
775 | * @count: length of buffer | |
776 | * | |
777 | * Takes input from sysfs interface for manually overriding the default | |
b71a8eb0 | 778 | * clocksource selection. |
734efb46 | 779 | */ |
780 | static ssize_t sysfs_override_clocksource(struct sys_device *dev, | |
4a0b2b4d | 781 | struct sysdev_attribute *attr, |
734efb46 | 782 | const char *buf, size_t count) |
783 | { | |
784 | size_t ret = count; | |
92c7e002 | 785 | |
734efb46 | 786 | /* strings from sysfs write are not 0 terminated! */ |
787 | if (count >= sizeof(override_name)) | |
788 | return -EINVAL; | |
789 | ||
790 | /* strip of \n: */ | |
791 | if (buf[count-1] == '\n') | |
792 | count--; | |
734efb46 | 793 | |
75c5158f | 794 | mutex_lock(&clocksource_mutex); |
734efb46 | 795 | |
92c7e002 TG |
796 | if (count > 0) |
797 | memcpy(override_name, buf, count); | |
734efb46 | 798 | override_name[count] = 0; |
f1b82746 | 799 | clocksource_select(); |
734efb46 | 800 | |
75c5158f | 801 | mutex_unlock(&clocksource_mutex); |
734efb46 | 802 | |
803 | return ret; | |
804 | } | |
805 | ||
806 | /** | |
807 | * sysfs_show_available_clocksources - sysfs interface for listing clocksource | |
808 | * @dev: unused | |
809 | * @buf: char buffer to be filled with clocksource list | |
810 | * | |
811 | * Provides sysfs interface for listing registered clocksources | |
812 | */ | |
813 | static ssize_t | |
4a0b2b4d AK |
814 | sysfs_show_available_clocksources(struct sys_device *dev, |
815 | struct sysdev_attribute *attr, | |
816 | char *buf) | |
734efb46 | 817 | { |
2e197586 | 818 | struct clocksource *src; |
5e2cb101 | 819 | ssize_t count = 0; |
734efb46 | 820 | |
75c5158f | 821 | mutex_lock(&clocksource_mutex); |
2e197586 | 822 | list_for_each_entry(src, &clocksource_list, list) { |
cd6d95d8 TG |
823 | /* |
824 | * Don't show non-HRES clocksource if the tick code is | |
825 | * in one shot mode (highres=on or nohz=on) | |
826 | */ | |
827 | if (!tick_oneshot_mode_active() || | |
828 | (src->flags & CLOCK_SOURCE_VALID_FOR_HRES)) | |
3f68535a | 829 | count += snprintf(buf + count, |
5e2cb101 MX |
830 | max((ssize_t)PAGE_SIZE - count, (ssize_t)0), |
831 | "%s ", src->name); | |
734efb46 | 832 | } |
75c5158f | 833 | mutex_unlock(&clocksource_mutex); |
734efb46 | 834 | |
5e2cb101 MX |
835 | count += snprintf(buf + count, |
836 | max((ssize_t)PAGE_SIZE - count, (ssize_t)0), "\n"); | |
734efb46 | 837 | |
5e2cb101 | 838 | return count; |
734efb46 | 839 | } |
840 | ||
841 | /* | |
842 | * Sysfs setup bits: | |
843 | */ | |
4f95f81a | 844 | static SYSDEV_ATTR(current_clocksource, 0644, sysfs_show_current_clocksources, |
f5f1a24a | 845 | sysfs_override_clocksource); |
734efb46 | 846 | |
4f95f81a | 847 | static SYSDEV_ATTR(available_clocksource, 0444, |
f5f1a24a | 848 | sysfs_show_available_clocksources, NULL); |
734efb46 | 849 | |
850 | static struct sysdev_class clocksource_sysclass = { | |
af5ca3f4 | 851 | .name = "clocksource", |
734efb46 | 852 | }; |
853 | ||
854 | static struct sys_device device_clocksource = { | |
855 | .id = 0, | |
856 | .cls = &clocksource_sysclass, | |
857 | }; | |
858 | ||
ad596171 | 859 | static int __init init_clocksource_sysfs(void) |
734efb46 | 860 | { |
861 | int error = sysdev_class_register(&clocksource_sysclass); | |
862 | ||
863 | if (!error) | |
864 | error = sysdev_register(&device_clocksource); | |
865 | if (!error) | |
866 | error = sysdev_create_file( | |
867 | &device_clocksource, | |
868 | &attr_current_clocksource); | |
869 | if (!error) | |
870 | error = sysdev_create_file( | |
871 | &device_clocksource, | |
872 | &attr_available_clocksource); | |
873 | return error; | |
874 | } | |
875 | ||
876 | device_initcall(init_clocksource_sysfs); | |
2b013700 | 877 | #endif /* CONFIG_SYSFS */ |
734efb46 | 878 | |
879 | /** | |
880 | * boot_override_clocksource - boot clock override | |
881 | * @str: override name | |
882 | * | |
883 | * Takes a clocksource= boot argument and uses it | |
884 | * as the clocksource override name. | |
885 | */ | |
886 | static int __init boot_override_clocksource(char* str) | |
887 | { | |
75c5158f | 888 | mutex_lock(&clocksource_mutex); |
734efb46 | 889 | if (str) |
890 | strlcpy(override_name, str, sizeof(override_name)); | |
75c5158f | 891 | mutex_unlock(&clocksource_mutex); |
734efb46 | 892 | return 1; |
893 | } | |
894 | ||
895 | __setup("clocksource=", boot_override_clocksource); | |
896 | ||
897 | /** | |
898 | * boot_override_clock - Compatibility layer for deprecated boot option | |
899 | * @str: override name | |
900 | * | |
901 | * DEPRECATED! Takes a clock= boot argument and uses it | |
902 | * as the clocksource override name | |
903 | */ | |
904 | static int __init boot_override_clock(char* str) | |
905 | { | |
5d0cf410 | 906 | if (!strcmp(str, "pmtmr")) { |
907 | printk("Warning: clock=pmtmr is deprecated. " | |
908 | "Use clocksource=acpi_pm.\n"); | |
909 | return boot_override_clocksource("acpi_pm"); | |
910 | } | |
911 | printk("Warning! clock= boot option is deprecated. " | |
912 | "Use clocksource=xyz\n"); | |
734efb46 | 913 | return boot_override_clocksource(str); |
914 | } | |
915 | ||
916 | __setup("clock=", boot_override_clock); |