| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 29 Mar 2022 12:49:25 +0800
From: kernel test robot <lkp@...el.com>
To: Rick Yiu <rickyiu@...gle.com>
Cc: kbuild-all@...ts.01.org,
GNU/Weeb Mailing List <gwml@...r.gnuweeb.org>,
linux-kernel@...r.kernel.org
Subject: [ammarfaizi2-block:google/android/kernel/common/android13-5.10
9999/9999] kernel/sched/pelt.c:220:5: warning: no previous prototype for
'___update_load_sum'
tree: https://github.com/ammarfaizi2/linux-block google/android/kernel/common/android13-5.10
head: e3356ca0a6c24e1ccdde6e9f80875701218ab090
commit: e3356ca0a6c24e1ccdde6e9f80875701218ab090 [9999/9999] ANDROID: sched: Add vendor hook for util-update related functions
config: x86_64-allyesconfig (https://download.01.org/0day-ci/archive/20220329/202203291253.hHVqFjjU-lkp@intel.com/config)
compiler: gcc-9 (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
reproduce (this is a W=1 build):
# https://github.com/ammarfaizi2/linux-block/commit/e3356ca0a6c24e1ccdde6e9f80875701218ab090
git remote add ammarfaizi2-block https://github.com/ammarfaizi2/linux-block
git fetch --no-tags ammarfaizi2-block google/android/kernel/common/android13-5.10
git checkout e3356ca0a6c24e1ccdde6e9f80875701218ab090
# save the config file to linux build tree
mkdir build_dir
make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash kernel/
If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@...el.com>
All warnings (new ones prefixed by >>):
>> kernel/sched/pelt.c:220:5: warning: no previous prototype for '___update_load_sum' [-Wmissing-prototypes]
220 | int ___update_load_sum(u64 now, struct sched_avg *sa,
| ^~~~~~~~~~~~~~~~~~
>> kernel/sched/pelt.c:297:6: warning: no previous prototype for '___update_load_avg' [-Wmissing-prototypes]
297 | void ___update_load_avg(struct sched_avg *sa, unsigned long load)
| ^~~~~~~~~~~~~~~~~~
vim +/___update_load_sum +220 kernel/sched/pelt.c
191
192 /*
193 * We can represent the historical contribution to runnable average as the
194 * coefficients of a geometric series. To do this we sub-divide our runnable
195 * history into segments of approximately 1ms (1024us); label the segment that
196 * occurred N-ms ago p_N, with p_0 corresponding to the current period, e.g.
197 *
198 * [<- 1024us ->|<- 1024us ->|<- 1024us ->| ...
199 * p0 p1 p2
200 * (now) (~1ms ago) (~2ms ago)
201 *
202 * Let u_i denote the fraction of p_i that the entity was runnable.
203 *
204 * We then designate the fractions u_i as our co-efficients, yielding the
205 * following representation of historical load:
206 * u_0 + u_1*y + u_2*y^2 + u_3*y^3 + ...
207 *
208 * We choose y based on the with of a reasonably scheduling period, fixing:
209 * y^32 = 0.5
210 *
211 * This means that the contribution to load ~32ms ago (u_32) will be weighted
212 * approximately half as much as the contribution to load within the last ms
213 * (u_0).
214 *
215 * When a period "rolls over" and we have new u_0`, multiplying the previous
216 * sum again by y is sufficient to update:
217 * load_avg = u_0` + y*(u_0 + u_1*y + u_2*y^2 + ... )
218 * = u_0 + u_1*y + u_2*y^2 + ... [re-labeling u_i --> u_{i+1}]
219 */
> 220 int ___update_load_sum(u64 now, struct sched_avg *sa,
221 unsigned long load, unsigned long runnable, int running)
222 {
223 u64 delta;
224
225 delta = now - sa->last_update_time;
226 /*
227 * This should only happen when time goes backwards, which it
228 * unfortunately does during sched clock init when we swap over to TSC.
229 */
230 if ((s64)delta < 0) {
231 sa->last_update_time = now;
232 return 0;
233 }
234
235 /*
236 * Use 1024ns as the unit of measurement since it's a reasonable
237 * approximation of 1us and fast to compute.
238 */
239 delta >>= 10;
240 if (!delta)
241 return 0;
242
243 sa->last_update_time += delta << 10;
244
245 /*
246 * running is a subset of runnable (weight) so running can't be set if
247 * runnable is clear. But there are some corner cases where the current
248 * se has been already dequeued but cfs_rq->curr still points to it.
249 * This means that weight will be 0 but not running for a sched_entity
250 * but also for a cfs_rq if the latter becomes idle. As an example,
251 * this happens during idle_balance() which calls
252 * update_blocked_averages().
253 *
254 * Also see the comment in accumulate_sum().
255 */
256 if (!load)
257 runnable = running = 0;
258
259 /*
260 * Now we know we crossed measurement unit boundaries. The *_avg
261 * accrues by two steps:
262 *
263 * Step 1: accumulate *_sum since last_update_time. If we haven't
264 * crossed period boundaries, finish.
265 */
266 if (!accumulate_sum(delta, sa, load, runnable, running))
267 return 0;
268
269 return 1;
270 }
271 EXPORT_SYMBOL_GPL(___update_load_sum);
272
273 /*
274 * When syncing *_avg with *_sum, we must take into account the current
275 * position in the PELT segment otherwise the remaining part of the segment
276 * will be considered as idle time whereas it's not yet elapsed and this will
277 * generate unwanted oscillation in the range [1002..1024[.
278 *
279 * The max value of *_sum varies with the position in the time segment and is
280 * equals to :
281 *
282 * LOAD_AVG_MAX*y + sa->period_contrib
283 *
284 * which can be simplified into:
285 *
286 * LOAD_AVG_MAX - 1024 + sa->period_contrib
287 *
288 * because LOAD_AVG_MAX*y == LOAD_AVG_MAX-1024
289 *
290 * The same care must be taken when a sched entity is added, updated or
291 * removed from a cfs_rq and we need to update sched_avg. Scheduler entities
292 * and the cfs rq, to which they are attached, have the same position in the
293 * time segment because they use the same clock. This means that we can use
294 * the period_contrib of cfs_rq when updating the sched_avg of a sched_entity
295 * if it's more convenient.
296 */
> 297 void ___update_load_avg(struct sched_avg *sa, unsigned long load)
298 {
299 u32 divider = get_pelt_divider(sa);
300
301 /*
302 * Step 2: update *_avg.
303 */
304 sa->load_avg = div_u64(load * sa->load_sum, divider);
305 sa->runnable_avg = div_u64(sa->runnable_sum, divider);
306 WRITE_ONCE(sa->util_avg, sa->util_sum / divider);
307 }
308 EXPORT_SYMBOL_GPL(___update_load_avg);
309
--
0-DAY CI Kernel Test Service
https://01.org/lkp
Powered by blists - more mailing lists