| Message ID | 20250218031448.46951-9-inwardvessel@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | cgroup: separate rstat trees | expand |
Hi JP, kernel test robot noticed the following build warnings: [auto build test WARNING on tj-cgroup/for-next] [also build test WARNING on bpf-next/master bpf/master linus/master v6.14-rc3 next-20250219] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in https://git-scm.com/docs/git-format-patch#_base_tree_information] url: https://github.com/intel-lab-lkp/linux/commits/JP-Kobryn/cgroup-move-rstat-pointers-into-struct-of-their-own/20250218-111725 base: https://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git for-next patch link: https://lore.kernel.org/r/20250218031448.46951-9-inwardvessel%40gmail.com patch subject: [PATCH 08/11] cgroup: rstat cpu lock indirection config: arc-randconfig-002-20250219 (https://download.01.org/0day-ci/archive/20250219/202502191619.0t8nOsuQ-lkp@intel.com/config) compiler: arc-elf-gcc (GCC) 13.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250219/202502191619.0t8nOsuQ-lkp@intel.com/reproduce) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202502191619.0t8nOsuQ-lkp@intel.com/ All warnings (new ones prefixed by >>): kernel/cgroup/rstat.c:266: warning: Function parameter or struct member 'ops' not described in 'cgroup_rstat_push_children' kernel/cgroup/rstat.c:326: warning: Function parameter or struct member 'ops' not described in 'cgroup_rstat_updated_list' >> kernel/cgroup/rstat.c:326: warning: Function parameter or struct member 'cpu_lock' not described in 'cgroup_rstat_updated_list' kernel/cgroup/rstat.c:532: warning: Function parameter or struct member 'ops' not described in '__cgroup_rstat_flush_release' >> kernel/cgroup/rstat.c:532: warning: Function parameter or struct member 'lock' not described in '__cgroup_rstat_flush_release' kernel/cgroup/rstat.c:532: warning: expecting prototype for cgroup_rstat_flush_release(). Prototype was for __cgroup_rstat_flush_release() instead vim +326 kernel/cgroup/rstat.c d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 304 d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 305 /** d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 306 * cgroup_rstat_updated_list - return a list of updated cgroups to be flushed d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 307 * @root: root of the cgroup subtree to traverse d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 308 * @cpu: target cpu d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 309 * Return: A singly linked list of cgroups to be flushed d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 310 * d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 311 * Walks the updated rstat_cpu tree on @cpu from @root. During traversal, d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 312 * each returned cgroup is unlinked from the updated tree. 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 313 * 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 314 * The only ordering guarantee is that, for a parent and a child pair d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 315 * covered by a given traversal, the child is before its parent in d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 316 * the list. d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 317 * d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 318 * Note that updated_children is self terminated and points to a list of d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 319 * child cgroups if not empty. Whereas updated_next is like a sibling link d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 320 * within the children list and terminated by the parent cgroup. An exception d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 321 * here is the cgroup root whose updated_next can be self terminated. 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 322 */ 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 323 static struct cgroup_rstat *cgroup_rstat_updated_list( 85c7ff288b9391 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 324 struct cgroup_rstat *root, int cpu, struct cgroup_rstat_ops *ops, 85c7ff288b9391 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 325 raw_spinlock_t *cpu_lock) 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 @326 { 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 327 struct cgroup_rstat_cpu *rstatc = rstat_cpu(root, cpu); 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 328 struct cgroup_rstat *head = NULL, *parent, *child; d67ed623c585f2 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 329 struct cgroup *cgrp; d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 330 unsigned long flags; 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 331 d67ed623c585f2 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 332 cgrp = ops->cgroup_fn(root); 85c7ff288b9391 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 333 flags = _cgroup_rstat_cpu_lock(cpu_lock, cpu, cgrp, false); 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 334 d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 335 /* Return NULL if this subtree is not on-list */ d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 336 if (!rstatc->updated_next) d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 337 goto unlock_ret; 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 338 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 339 /* d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 340 * Unlink @root from its parent. As the updated_children list is 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 341 * singly linked, we have to walk it to find the removal point. 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 342 */ d67ed623c585f2 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 343 parent = ops->parent_fn(root); dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 344 if (parent) { dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 345 struct cgroup_rstat_cpu *prstatc; 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 346 struct cgroup_rstat **nextp; 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 347 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 348 prstatc = rstat_cpu(parent, cpu); c58632b3631cb2 kernel/cgroup/rstat.c Tejun Heo 2018-04-26 349 nextp = &prstatc->updated_children; d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 350 while (*nextp != root) { dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 351 struct cgroup_rstat_cpu *nrstatc; dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 352 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 353 nrstatc = rstat_cpu(*nextp, cpu); 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 354 WARN_ON_ONCE(*nextp == parent); c58632b3631cb2 kernel/cgroup/rstat.c Tejun Heo 2018-04-26 355 nextp = &nrstatc->updated_next; 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 356 } c58632b3631cb2 kernel/cgroup/rstat.c Tejun Heo 2018-04-26 357 *nextp = rstatc->updated_next; dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 358 } 9a9e97b2f1f27e kernel/cgroup/rstat.c Tejun Heo 2018-04-26 359 dc26532aed0ab2 kernel/cgroup/rstat.c Johannes Weiner 2021-04-29 360 rstatc->updated_next = NULL; e76d28bdf9ba53 kernel/cgroup/rstat.c Waiman Long 2023-11-03 361 d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 362 /* Push @root to the list first before pushing the children */ d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 363 head = root; 3d844899ba042a kernel/cgroup/rstat.c JP Kobryn 2025-02-17 364 root->rstat_flush_next = NULL; d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 365 child = rstatc->updated_children; d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 366 rstatc->updated_children = root; d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 367 if (child != root) d67ed623c585f2 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 368 head = cgroup_rstat_push_children(head, child, cpu, ops); d499fd418fa159 kernel/cgroup/rstat.c Waiman Long 2023-11-30 369 unlock_ret: 85c7ff288b9391 kernel/cgroup/rstat.c JP Kobryn 2025-02-17 370 _cgroup_rstat_cpu_unlock(cpu_lock, cpu, cgrp, flags, false); e76d28bdf9ba53 kernel/cgroup/rstat.c Waiman Long 2023-11-03 371 return head; 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 372 } 041cd640b2f3c5 kernel/cgroup/stat.c Tejun Heo 2017-09-25 373
diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c index 4cb0f3ffc1db..9f6da3ea3c8c 100644 --- a/kernel/cgroup/rstat.c +++ b/kernel/cgroup/rstat.c @@ -177,7 +177,7 @@ void _cgroup_rstat_cpu_unlock(raw_spinlock_t *lock, int cpu, } static void __cgroup_rstat_updated(struct cgroup_rstat *rstat, int cpu, - struct cgroup_rstat_ops *ops) + struct cgroup_rstat_ops *ops, raw_spinlock_t *cpu_lock) { struct cgroup *cgrp; unsigned long flags; @@ -194,7 +194,7 @@ static void __cgroup_rstat_updated(struct cgroup_rstat *rstat, int cpu, return; cgrp = ops->cgroup_fn(rstat); - flags = _cgroup_rstat_cpu_lock(&cgroup_rstat_cpu_lock, cpu, cgrp, true); + flags = _cgroup_rstat_cpu_lock(cpu_lock, cpu, cgrp, true); /* put @rstat and all ancestors on the corresponding updated lists */ while (true) { @@ -222,7 +222,7 @@ static void __cgroup_rstat_updated(struct cgroup_rstat *rstat, int cpu, rstat = parent; } - _cgroup_rstat_cpu_unlock(&cgroup_rstat_cpu_lock, cpu, cgrp, flags, true); + _cgroup_rstat_cpu_unlock(cpu_lock, cpu, cgrp, flags, true); } /** @@ -236,13 +236,15 @@ static void __cgroup_rstat_updated(struct cgroup_rstat *rstat, int cpu, */ void cgroup_rstat_updated(struct cgroup_subsys_state *css, int cpu) { - __cgroup_rstat_updated(&css->rstat, cpu, &rstat_css_ops); + __cgroup_rstat_updated(&css->rstat, cpu, &rstat_css_ops, + &cgroup_rstat_cpu_lock); } #ifdef CONFIG_CGROUP_BPF __bpf_kfunc void bpf_cgroup_rstat_updated(struct cgroup *cgroup, int cpu) { - __cgroup_rstat_updated(&(cgroup->bpf.rstat), cpu, &rstat_bpf_ops); + __cgroup_rstat_updated(&(cgroup->bpf.rstat), cpu, &rstat_bpf_ops, + &cgroup_rstat_cpu_lock); } #endif /* CONFIG_CGROUP_BPF */ @@ -319,7 +321,8 @@ static struct cgroup_rstat *cgroup_rstat_push_children( * here is the cgroup root whose updated_next can be self terminated. */ static struct cgroup_rstat *cgroup_rstat_updated_list( - struct cgroup_rstat *root, int cpu, struct cgroup_rstat_ops *ops) + struct cgroup_rstat *root, int cpu, struct cgroup_rstat_ops *ops, + raw_spinlock_t *cpu_lock) { struct cgroup_rstat_cpu *rstatc = rstat_cpu(root, cpu); struct cgroup_rstat *head = NULL, *parent, *child; @@ -327,7 +330,7 @@ static struct cgroup_rstat *cgroup_rstat_updated_list( unsigned long flags; cgrp = ops->cgroup_fn(root); - flags = _cgroup_rstat_cpu_lock(&cgroup_rstat_cpu_lock, cpu, cgrp, false); + flags = _cgroup_rstat_cpu_lock(cpu_lock, cpu, cgrp, false); /* Return NULL if this subtree is not on-list */ if (!rstatc->updated_next) @@ -364,7 +367,7 @@ static struct cgroup_rstat *cgroup_rstat_updated_list( if (child != root) head = cgroup_rstat_push_children(head, child, cpu, ops); unlock_ret: - _cgroup_rstat_cpu_unlock(&cgroup_rstat_cpu_lock, cpu, cgrp, flags, false); + _cgroup_rstat_cpu_unlock(cpu_lock, cpu, cgrp, flags, false); return head; } @@ -422,43 +425,46 @@ static inline void __cgroup_rstat_unlock(spinlock_t *lock, /* see cgroup_rstat_flush() */ static void cgroup_rstat_flush_locked(struct cgroup_rstat *rstat, - struct cgroup_rstat_ops *ops) - __releases(&cgroup_rstat_lock) __acquires(&cgroup_rstat_lock) + struct cgroup_rstat_ops *ops, spinlock_t *lock, + raw_spinlock_t *cpu_lock) + __releases(lock) __acquires(lock) { int cpu; - lockdep_assert_held(&cgroup_rstat_lock); + lockdep_assert_held(lock); for_each_possible_cpu(cpu) { struct cgroup_rstat *pos = cgroup_rstat_updated_list( - rstat, cpu, ops); + rstat, cpu, ops, cpu_lock); for (; pos; pos = pos->rstat_flush_next) ops->flush_fn(pos, cpu); /* play nice and yield if necessary */ - if (need_resched() || spin_needbreak(&cgroup_rstat_lock)) { + if (need_resched() || spin_needbreak(lock)) { struct cgroup *cgrp; cgrp = ops->cgroup_fn(rstat); - __cgroup_rstat_unlock(&cgroup_rstat_lock, cgrp, cpu); + __cgroup_rstat_unlock(lock, cgrp, cpu); if (!cond_resched()) cpu_relax(); - __cgroup_rstat_lock(&cgroup_rstat_lock, cgrp, cpu); + __cgroup_rstat_lock(lock, cgrp, cpu); } } } static void __cgroup_rstat_flush(struct cgroup_rstat *rstat, - struct cgroup_rstat_ops *ops) + struct cgroup_rstat_ops *ops, spinlock_t *lock, + raw_spinlock_t *cpu_lock) + __acquires(lock) __releases(lock) { struct cgroup *cgrp; might_sleep(); cgrp = ops->cgroup_fn(rstat); - __cgroup_rstat_lock(&cgroup_rstat_lock, cgrp, -1); - cgroup_rstat_flush_locked(rstat, ops); - __cgroup_rstat_unlock(&cgroup_rstat_lock, cgrp, -1); + __cgroup_rstat_lock(lock, cgrp, -1); + cgroup_rstat_flush_locked(rstat, ops, lock, cpu_lock); + __cgroup_rstat_unlock(lock, cgrp, -1); } /** @@ -476,26 +482,29 @@ static void __cgroup_rstat_flush(struct cgroup_rstat *rstat, */ void cgroup_rstat_flush(struct cgroup_subsys_state *css) { - __cgroup_rstat_flush(&css->rstat, &rstat_css_ops); + __cgroup_rstat_flush(&css->rstat, &rstat_css_ops, + &cgroup_rstat_lock, &cgroup_rstat_cpu_lock); } #ifdef CONFIG_CGROUP_BPF __bpf_kfunc void bpf_cgroup_rstat_flush(struct cgroup *cgroup) { - __cgroup_rstat_flush(&(cgroup->bpf.rstat), &rstat_bpf_ops); + __cgroup_rstat_flush(&(cgroup->bpf.rstat), &rstat_bpf_ops, + &cgroup_rstat_lock, &cgroup_rstat_cpu_lock); } #endif /* CONFIG_CGROUP_BPF */ static void __cgroup_rstat_flush_hold(struct cgroup_rstat *rstat, - struct cgroup_rstat_ops *ops) - __acquires(&cgroup_rstat_lock) + struct cgroup_rstat_ops *ops, spinlock_t *lock, + raw_spinlock_t *cpu_lock) + __acquires(lock) { struct cgroup *cgrp; might_sleep(); cgrp = ops->cgroup_fn(rstat); - __cgroup_rstat_lock(&cgroup_rstat_lock, cgrp, -1); - cgroup_rstat_flush_locked(rstat, ops); + __cgroup_rstat_lock(lock, cgrp, -1); + cgroup_rstat_flush_locked(rstat, ops, lock, cpu_lock); } /** @@ -509,7 +518,8 @@ static void __cgroup_rstat_flush_hold(struct cgroup_rstat *rstat, */ void cgroup_rstat_flush_hold(struct cgroup_subsys_state *css) { - __cgroup_rstat_flush_hold(&css->rstat, &rstat_css_ops); + __cgroup_rstat_flush_hold(&css->rstat, &rstat_css_ops, + &cgroup_rstat_lock, &cgroup_rstat_cpu_lock); } /** @@ -517,13 +527,13 @@ void cgroup_rstat_flush_hold(struct cgroup_subsys_state *css) * @rstat: rstat node used to find associated cgroup used by tracepoint */ static void __cgroup_rstat_flush_release(struct cgroup_rstat *rstat, - struct cgroup_rstat_ops *ops) - __releases(&cgroup_rstat_lock) + struct cgroup_rstat_ops *ops, spinlock_t *lock) + __releases(lock) { struct cgroup *cgrp; cgrp = ops->cgroup_fn(rstat); - __cgroup_rstat_unlock(&cgroup_rstat_lock, cgrp, -1); + __cgroup_rstat_unlock(lock, cgrp, -1); } /** @@ -532,7 +542,8 @@ static void __cgroup_rstat_flush_release(struct cgroup_rstat *rstat, */ void cgroup_rstat_flush_release(struct cgroup_subsys_state *css) { - __cgroup_rstat_flush_release(&css->rstat, &rstat_css_ops); + __cgroup_rstat_flush_release(&css->rstat, &rstat_css_ops, + &cgroup_rstat_lock); } static void __cgroup_rstat_init(struct cgroup_rstat *rstat) @@ -605,7 +616,8 @@ int bpf_cgroup_rstat_init(struct cgroup_bpf *bpf) void bpf_cgroup_rstat_exit(struct cgroup_bpf *bpf) { - __cgroup_rstat_flush(&bpf->rstat, &rstat_bpf_ops); + __cgroup_rstat_flush(&bpf->rstat, &rstat_bpf_ops, + &cgroup_rstat_lock, &cgroup_rstat_cpu_lock); __cgroup_rstat_exit(&bpf->rstat); } #endif /* CONFIG_CGROUP_BPF */
Where functions access the global per-cpu lock, change their signature to accept the lock instead as a paremeter. Change the code within these functions to only access the parameter. This indirection allows for future code to accept different locks, increasing extensibity. For example, a new lock could be added specifically for the bpf cgroups and it would not contend with the existing lock. Signed-off-by: JP Kobryn <inwardvessel@gmail.com> --- kernel/cgroup/rstat.c | 74 +++++++++++++++++++++++++------------------ 1 file changed, 43 insertions(+), 31 deletions(-)