| 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
| ||
|
Message-ID: <202203171642.uVKJqHxx-lkp@intel.com>
Date: Thu, 17 Mar 2022 16:02:40 +0800
From: kernel test robot <lkp@...el.com>
To: Patrick Bellasi <patrick.bellasi@....com>
Cc: kbuild-all@...ts.01.org,
GNU/Weeb Mailing List <gwml@...r.gnuweeb.org>,
linux-kernel@...r.kernel.org,
Quentin Perret <quentin.perret@....com>,
Chris Redpath <chris.redpath@....com>
Subject: [ammarfaizi2-block:google/android/kernel/common/android-4.19-q-release
43/7636] kernel/sched/tune.c:290:33: warning: 'struct cgroup_taskset'
declared inside parameter list will not be visible outside of this
definition or declaration
tree: https://github.com/ammarfaizi2/linux-block google/android/kernel/common/android-4.19-q-release
head: 1a3500dc59324a85c20f025cfa6485a1a0debbdf
commit: 68dbff9ce92ae2f923321d150fb72a6d8b05e640 [43/7636] ANDROID: sched: fair/tune: Add schedtune with cgroups interface
config: s390-randconfig-r035-20220317 (https://download.01.org/0day-ci/archive/20220317/202203171642.uVKJqHxx-lkp@intel.com/config)
compiler: s390-linux-gcc (GCC) 11.2.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/ammarfaizi2/linux-block/commit/68dbff9ce92ae2f923321d150fb72a6d8b05e640
git remote add ammarfaizi2-block https://github.com/ammarfaizi2/linux-block
git fetch --no-tags ammarfaizi2-block google/android/kernel/common/android-4.19-q-release
git checkout 68dbff9ce92ae2f923321d150fb72a6d8b05e640
# save the config file to linux build tree
mkdir build_dir
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross O=build_dir ARCH=s390 SHELL=/bin/bash kernel/sched/
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/tune.c:81:36: error: field 'css' has incomplete type
81 | struct cgroup_subsys_state css;
| ^~~
kernel/sched/tune.c: In function 'task_schedtune':
kernel/sched/tune.c:101:23: error: implicit declaration of function 'task_css'; did you mean 'task_cpu'? [-Werror=implicit-function-declaration]
101 | return css_st(task_css(tsk, schedtune_cgrp_id));
| ^~~~~~~~
| task_cpu
kernel/sched/tune.c:101:37: error: 'schedtune_cgrp_id' undeclared (first use in this function); did you mean 'schedtune_spc_rdiv'?
101 | return css_st(task_css(tsk, schedtune_cgrp_id));
| ^~~~~~~~~~~~~~~~~
| schedtune_spc_rdiv
kernel/sched/tune.c:101:37: note: each undeclared identifier is reported only once for each function it appears in
kernel/sched/tune.c: At top level:
>> kernel/sched/tune.c:290:33: warning: 'struct cgroup_taskset' declared inside parameter list will not be visible outside of this definition or declaration
290 | int schedtune_can_attach(struct cgroup_taskset *tset)
| ^~~~~~~~~~~~~~
kernel/sched/tune.c:290:5: warning: no previous prototype for 'schedtune_can_attach' [-Wmissing-prototypes]
290 | int schedtune_can_attach(struct cgroup_taskset *tset)
| ^~~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c: In function 'schedtune_can_attach':
kernel/sched/tune.c:306:9: error: implicit declaration of function 'cgroup_taskset_for_each' [-Werror=implicit-function-declaration]
306 | cgroup_taskset_for_each(task, css, tset) {
| ^~~~~~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c:306:49: error: expected ';' before '{' token
306 | cgroup_taskset_for_each(task, css, tset) {
| ^~
| ;
kernel/sched/tune.c:300:13: warning: unused variable 'tasks' [-Wunused-variable]
300 | int tasks;
| ^~~~~
kernel/sched/tune.c:299:13: warning: unused variable 'dst_bg' [-Wunused-variable]
299 | int dst_bg; /* Destination boost group index */
| ^~~~~~
kernel/sched/tune.c:298:13: warning: unused variable 'src_bg' [-Wunused-variable]
298 | int src_bg; /* Source boost group index */
| ^~~~~~
kernel/sched/tune.c:297:20: warning: unused variable 'rq' [-Wunused-variable]
297 | struct rq *rq;
| ^~
kernel/sched/tune.c:296:22: warning: unused variable 'cpu' [-Wunused-variable]
296 | unsigned int cpu;
| ^~~
kernel/sched/tune.c:295:25: warning: unused variable 'rq_flags' [-Wunused-variable]
295 | struct rq_flags rq_flags;
| ^~~~~~~~
kernel/sched/tune.c:294:30: warning: unused variable 'bg' [-Wunused-variable]
294 | struct boost_groups *bg;
| ^~
kernel/sched/tune.c: At top level:
kernel/sched/tune.c:363:37: warning: 'struct cgroup_taskset' declared inside parameter list will not be visible outside of this definition or declaration
363 | void schedtune_cancel_attach(struct cgroup_taskset *tset)
| ^~~~~~~~~~~~~~
kernel/sched/tune.c:363:6: warning: no previous prototype for 'schedtune_cancel_attach' [-Wmissing-prototypes]
363 | void schedtune_cancel_attach(struct cgroup_taskset *tset)
| ^~~~~~~~~~~~~~~~~~~~~~~
>> kernel/sched/tune.c:444:58: warning: 'struct cftype' declared inside parameter list will not be visible outside of this definition or declaration
444 | prefer_idle_read(struct cgroup_subsys_state *css, struct cftype *cft)
| ^~~~~~
kernel/sched/tune.c:452:59: warning: 'struct cftype' declared inside parameter list will not be visible outside of this definition or declaration
452 | prefer_idle_write(struct cgroup_subsys_state *css, struct cftype *cft,
| ^~~~~~
kernel/sched/tune.c:462:52: warning: 'struct cftype' declared inside parameter list will not be visible outside of this definition or declaration
462 | boost_read(struct cgroup_subsys_state *css, struct cftype *cft)
| ^~~~~~
kernel/sched/tune.c:470:53: warning: 'struct cftype' declared inside parameter list will not be visible outside of this definition or declaration
470 | boost_write(struct cgroup_subsys_state *css, struct cftype *cft,
| ^~~~~~
kernel/sched/tune.c:486:22: error: array type has incomplete element type 'struct cftype'
486 | static struct cftype files[] = {
| ^~~~~
kernel/sched/tune.c:583:8: error: variable 'schedtune_cgrp_subsys' has initializer but incomplete type
583 | struct cgroup_subsys schedtune_cgrp_subsys = {
| ^~~~~~~~~~~~~
kernel/sched/tune.c:584:10: error: 'struct cgroup_subsys' has no member named 'css_alloc'
584 | .css_alloc = schedtune_css_alloc,
| ^~~~~~~~~
>> kernel/sched/tune.c:584:27: warning: excess elements in struct initializer
584 | .css_alloc = schedtune_css_alloc,
| ^~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c:584:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:585:10: error: 'struct cgroup_subsys' has no member named 'css_free'
585 | .css_free = schedtune_css_free,
| ^~~~~~~~
kernel/sched/tune.c:585:27: warning: excess elements in struct initializer
585 | .css_free = schedtune_css_free,
| ^~~~~~~~~~~~~~~~~~
kernel/sched/tune.c:585:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:586:10: error: 'struct cgroup_subsys' has no member named 'can_attach'
586 | .can_attach = schedtune_can_attach,
| ^~~~~~~~~~
kernel/sched/tune.c:586:27: warning: excess elements in struct initializer
586 | .can_attach = schedtune_can_attach,
| ^~~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c:586:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:587:10: error: 'struct cgroup_subsys' has no member named 'cancel_attach'
587 | .cancel_attach = schedtune_cancel_attach,
| ^~~~~~~~~~~~~
kernel/sched/tune.c:587:27: warning: excess elements in struct initializer
587 | .cancel_attach = schedtune_cancel_attach,
| ^~~~~~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c:587:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:588:10: error: 'struct cgroup_subsys' has no member named 'legacy_cftypes'
588 | .legacy_cftypes = files,
| ^~~~~~~~~~~~~~
kernel/sched/tune.c:588:27: warning: excess elements in struct initializer
588 | .legacy_cftypes = files,
| ^~~~~
kernel/sched/tune.c:588:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:589:10: error: 'struct cgroup_subsys' has no member named 'early_init'
589 | .early_init = 1,
| ^~~~~~~~~~
kernel/sched/tune.c:589:27: warning: excess elements in struct initializer
589 | .early_init = 1,
| ^
kernel/sched/tune.c:589:27: note: (near initialization for 'schedtune_cgrp_subsys')
kernel/sched/tune.c:583:22: error: storage size of 'schedtune_cgrp_subsys' isn't known
583 | struct cgroup_subsys schedtune_cgrp_subsys = {
| ^~~~~~~~~~~~~~~~~~~~~
kernel/sched/tune.c: In function 'task_schedtune':
kernel/sched/tune.c:102:1: warning: control reaches end of non-void function [-Wreturn-type]
102 | }
| ^
kernel/sched/tune.c: In function 'schedtune_can_attach':
kernel/sched/tune.c:361:1: warning: control reaches end of non-void function [-Wreturn-type]
361 | }
| ^
At top level:
kernel/sched/tune.c:486:22: warning: 'files' defined but not used [-Wunused-variable]
486 | static struct cftype files[] = {
| ^~~~~
cc1: some warnings being treated as errors
vim +290 kernel/sched/tune.c
98
99 static inline struct schedtune *task_schedtune(struct task_struct *tsk)
100 {
> 101 return css_st(task_css(tsk, schedtune_cgrp_id));
102 }
103
104 static inline struct schedtune *parent_st(struct schedtune *st)
105 {
106 return css_st(st->css.parent);
107 }
108
109 /*
110 * SchedTune root control group
111 * The root control group is used to defined a system-wide boosting tuning,
112 * which is applied to all tasks in the system.
113 * Task specific boost tuning could be specified by creating and
114 * configuring a child control group under the root one.
115 * By default, system-wide boosting is disabled, i.e. no boosting is applied
116 * to tasks which are not into a child control group.
117 */
118 static struct schedtune
119 root_schedtune = {
120 .boost = 0,
121 .prefer_idle = 0,
122 };
123
124 /*
125 * Maximum number of boost groups to support
126 * When per-task boosting is used we still allow only limited number of
127 * boost groups for two main reasons:
128 * 1. on a real system we usually have only few classes of workloads which
129 * make sense to boost with different values (e.g. background vs foreground
130 * tasks, interactive vs low-priority tasks)
131 * 2. a limited number allows for a simpler and more memory/time efficient
132 * implementation especially for the computation of the per-CPU boost
133 * value
134 */
135 #define BOOSTGROUPS_COUNT 5
136
137 /* Array of configured boostgroups */
138 static struct schedtune *allocated_group[BOOSTGROUPS_COUNT] = {
139 &root_schedtune,
140 NULL,
141 };
142
143 /* SchedTune boost groups
144 * Keep track of all the boost groups which impact on CPU, for example when a
145 * CPU has two RUNNABLE tasks belonging to two different boost groups and thus
146 * likely with different boost values.
147 * Since on each system we expect only a limited number of boost groups, here
148 * we use a simple array to keep track of the metrics required to compute the
149 * maximum per-CPU boosting value.
150 */
151 struct boost_groups {
152 /* Maximum boost value for all RUNNABLE tasks on a CPU */
153 int boost_max;
154 struct {
155 /* True when this boost group maps an actual cgroup */
156 bool valid;
157 /* The boost for tasks on that boost group */
158 int boost;
159 /* Count of RUNNABLE tasks on that boost group */
160 unsigned tasks;
161 } group[BOOSTGROUPS_COUNT];
162 /* CPU's boost group locking */
163 raw_spinlock_t lock;
164 };
165
166 /* Boost groups affecting each CPU in the system */
167 DEFINE_PER_CPU(struct boost_groups, cpu_boost_groups);
168
169 static void
170 schedtune_cpu_update(int cpu)
171 {
172 struct boost_groups *bg = &per_cpu(cpu_boost_groups, cpu);
173 int boost_max;
174 int idx;
175
176 /* The root boost group is always active */
177 boost_max = bg->group[0].boost;
178 for (idx = 1; idx < BOOSTGROUPS_COUNT; ++idx) {
179
180 /* Ignore non boostgroups not mapping a cgroup */
181 if (!bg->group[idx].valid)
182 continue;
183
184 /*
185 * A boost group affects a CPU only if it has
186 * RUNNABLE tasks on that CPU
187 */
188 if (bg->group[idx].tasks == 0)
189 continue;
190
191 boost_max = max(boost_max, bg->group[idx].boost);
192 }
193
194 /* Ensures boost_max is non-negative when all cgroup boost values
195 * are neagtive. Avoids under-accounting of cpu capacity which may cause
196 * task stacking and frequency spikes.*/
197 boost_max = max(boost_max, 0);
198 bg->boost_max = boost_max;
199 }
200
201 static int
202 schedtune_boostgroup_update(int idx, int boost)
203 {
204 struct boost_groups *bg;
205 int cur_boost_max;
206 int old_boost;
207 int cpu;
208
209 /* Update per CPU boost groups */
210 for_each_possible_cpu(cpu) {
211 bg = &per_cpu(cpu_boost_groups, cpu);
212
213 /* CGroups are never associated to non active cgroups */
214 BUG_ON(!bg->group[idx].valid);
215
216 /*
217 * Keep track of current boost values to compute the per CPU
218 * maximum only when it has been affected by the new value of
219 * the updated boost group
220 */
221 cur_boost_max = bg->boost_max;
222 old_boost = bg->group[idx].boost;
223
224 /* Update the boost value of this boost group */
225 bg->group[idx].boost = boost;
226
227 /* Check if this update increase current max */
228 if (boost > cur_boost_max && bg->group[idx].tasks) {
229 bg->boost_max = boost;
230 continue;
231 }
232
233 /* Check if this update has decreased current max */
234 if (cur_boost_max == old_boost && old_boost > boost) {
235 schedtune_cpu_update(cpu);
236 continue;
237 }
238 }
239
240 return 0;
241 }
242
243 #define ENQUEUE_TASK 1
244 #define DEQUEUE_TASK -1
245
246 static inline void
247 schedtune_tasks_update(struct task_struct *p, int cpu, int idx, int task_count)
248 {
249 struct boost_groups *bg = &per_cpu(cpu_boost_groups, cpu);
250 int tasks = bg->group[idx].tasks + task_count;
251
252 /* Update boosted tasks count while avoiding to make it negative */
253 bg->group[idx].tasks = max(0, tasks);
254
255 /* Boost group activation or deactivation on that RQ */
256 if (tasks == 1 || tasks == 0)
257 schedtune_cpu_update(cpu);
258 }
259
260 /*
261 * NOTE: This function must be called while holding the lock on the CPU RQ
262 */
263 void schedtune_enqueue_task(struct task_struct *p, int cpu)
264 {
265 struct boost_groups *bg = &per_cpu(cpu_boost_groups, cpu);
266 unsigned long irq_flags;
267 struct schedtune *st;
268 int idx;
269
270 if (unlikely(!schedtune_initialized))
271 return;
272
273 /*
274 * Boost group accouting is protected by a per-cpu lock and requires
275 * interrupt to be disabled to avoid race conditions for example on
276 * do_exit()::cgroup_exit() and task migration.
277 */
278 raw_spin_lock_irqsave(&bg->lock, irq_flags);
279 rcu_read_lock();
280
281 st = task_schedtune(p);
282 idx = st->idx;
283
284 schedtune_tasks_update(p, cpu, idx, ENQUEUE_TASK);
285
286 rcu_read_unlock();
287 raw_spin_unlock_irqrestore(&bg->lock, irq_flags);
288 }
289
> 290 int schedtune_can_attach(struct cgroup_taskset *tset)
291 {
292 struct task_struct *task;
293 struct cgroup_subsys_state *css;
294 struct boost_groups *bg;
295 struct rq_flags rq_flags;
296 unsigned int cpu;
297 struct rq *rq;
298 int src_bg; /* Source boost group index */
299 int dst_bg; /* Destination boost group index */
300 int tasks;
301
302 if (unlikely(!schedtune_initialized))
303 return 0;
304
305
306 cgroup_taskset_for_each(task, css, tset) {
307
308 /*
309 * Lock the CPU's RQ the task is enqueued to avoid race
310 * conditions with migration code while the task is being
311 * accounted
312 */
313 rq = task_rq_lock(task, &rq_flags);
314
315 if (!task->on_rq) {
316 task_rq_unlock(rq, task, &rq_flags);
317 continue;
318 }
319
320 /*
321 * Boost group accouting is protected by a per-cpu lock and requires
322 * interrupt to be disabled to avoid race conditions on...
323 */
324 cpu = cpu_of(rq);
325 bg = &per_cpu(cpu_boost_groups, cpu);
326 raw_spin_lock(&bg->lock);
327
328 dst_bg = css_st(css)->idx;
329 src_bg = task_schedtune(task)->idx;
330
331 /*
332 * Current task is not changing boostgroup, which can
333 * happen when the new hierarchy is in use.
334 */
335 if (unlikely(dst_bg == src_bg)) {
336 raw_spin_unlock(&bg->lock);
337 task_rq_unlock(rq, task, &rq_flags);
338 continue;
339 }
340
341 /*
342 * This is the case of a RUNNABLE task which is switching its
343 * current boost group.
344 */
345
346 /* Move task from src to dst boost group */
347 tasks = bg->group[src_bg].tasks - 1;
348 bg->group[src_bg].tasks = max(0, tasks);
349 bg->group[dst_bg].tasks += 1;
350
351 raw_spin_unlock(&bg->lock);
352 task_rq_unlock(rq, task, &rq_flags);
353
354 /* Update CPU boost group */
355 if (bg->group[src_bg].tasks == 0 || bg->group[dst_bg].tasks == 1)
356 schedtune_cpu_update(task_cpu(task));
357
358 }
359
360 return 0;
361 }
362
> 363 void schedtune_cancel_attach(struct cgroup_taskset *tset)
364 {
365 /* This can happen only if SchedTune controller is mounted with
366 * other hierarchies ane one of them fails. Since usually SchedTune is
367 * mouted on its own hierarcy, for the time being we do not implement
368 * a proper rollback mechanism */
369 WARN(1, "SchedTune cancel attach not implemented");
370 }
371
372 /*
373 * NOTE: This function must be called while holding the lock on the CPU RQ
374 */
375 void schedtune_dequeue_task(struct task_struct *p, int cpu)
376 {
377 struct boost_groups *bg = &per_cpu(cpu_boost_groups, cpu);
378 unsigned long irq_flags;
379 struct schedtune *st;
380 int idx;
381
382 if (unlikely(!schedtune_initialized))
383 return;
384
385 /*
386 * Boost group accouting is protected by a per-cpu lock and requires
387 * interrupt to be disabled to avoid race conditions on...
388 */
389 raw_spin_lock_irqsave(&bg->lock, irq_flags);
390 rcu_read_lock();
391
392 st = task_schedtune(p);
393 idx = st->idx;
394
395 schedtune_tasks_update(p, cpu, idx, DEQUEUE_TASK);
396
397 rcu_read_unlock();
398 raw_spin_unlock_irqrestore(&bg->lock, irq_flags);
399 }
400
401 int schedtune_cpu_boost(int cpu)
402 {
403 struct boost_groups *bg;
404
405 bg = &per_cpu(cpu_boost_groups, cpu);
406 return bg->boost_max;
407 }
408
409 int schedtune_task_boost(struct task_struct *p)
410 {
411 struct schedtune *st;
412 int task_boost;
413
414 if (unlikely(!schedtune_initialized))
415 return 0;
416
417 /* Get task boost value */
418 rcu_read_lock();
419 st = task_schedtune(p);
420 task_boost = st->boost;
421 rcu_read_unlock();
422
423 return task_boost;
424 }
425
426 int schedtune_prefer_idle(struct task_struct *p)
427 {
428 struct schedtune *st;
429 int prefer_idle;
430
431 if (unlikely(!schedtune_initialized))
432 return 0;
433
434 /* Get prefer_idle value */
435 rcu_read_lock();
436 st = task_schedtune(p);
437 prefer_idle = st->prefer_idle;
438 rcu_read_unlock();
439
440 return prefer_idle;
441 }
442
443 static u64
> 444 prefer_idle_read(struct cgroup_subsys_state *css, struct cftype *cft)
445 {
446 struct schedtune *st = css_st(css);
447
448 return st->prefer_idle;
449 }
450
451 static int
452 prefer_idle_write(struct cgroup_subsys_state *css, struct cftype *cft,
453 u64 prefer_idle)
454 {
455 struct schedtune *st = css_st(css);
456 st->prefer_idle = !!prefer_idle;
457
458 return 0;
459 }
460
461 static s64
462 boost_read(struct cgroup_subsys_state *css, struct cftype *cft)
463 {
464 struct schedtune *st = css_st(css);
465
466 return st->boost;
467 }
468
469 static int
470 boost_write(struct cgroup_subsys_state *css, struct cftype *cft,
471 s64 boost)
472 {
473 struct schedtune *st = css_st(css);
474
475 if (boost < 0 || boost > 100)
476 return -EINVAL;
477
478 st->boost = boost;
479
480 /* Update CPU boost */
481 schedtune_boostgroup_update(st->idx, st->boost);
482
483 return 0;
484 }
485
> 486 static struct cftype files[] = {
487 {
488 .name = "boost",
489 .read_s64 = boost_read,
490 .write_s64 = boost_write,
491 },
492 {
493 .name = "prefer_idle",
494 .read_u64 = prefer_idle_read,
495 .write_u64 = prefer_idle_write,
496 },
497 { } /* terminate */
498 };
499
---
0-DAY CI Kernel Test Service
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
Powered by blists - more mailing lists