| 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: <CAKfTPtB9B9jv_uSk0fS7uOTtH5FCwLcAOrPjtXzRuURLUMFnoA@mail.gmail.com>
Date: Mon, 17 May 2021 15:45:29 +0200
From: Vincent Guittot <vincent.guittot@...aro.org>
To: kernel test robot <lkp@...el.com>
Cc: kbuild-all@...ts.01.org, clang-built-linux@...glegroups.com,
linux-kernel <linux-kernel@...r.kernel.org>,
Ingo Molnar <mingo@...nel.org>,
"Peter Zijlstra (Intel)" <peterz@...radead.org>,
Valentin Schneider <valentin.schneider@....com>
Subject: Re: kernel/sched/fair.c:9551:12: warning: stack frame size of 1376
bytes in function 'load_balance'
On Tue, 11 May 2021 at 18:01, kernel test robot <lkp@...el.com> wrote:
>
> tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> head: 1140ab592e2ebf8153d2b322604031a8868ce7a5
> commit: 0826530de3cbdc89e60a89e86def94a5f0fc81ca sched/fair: Remove update of blocked load from newidle_balance
> date: 9 weeks ago
> config: mips-randconfig-r022-20210511 (attached as .config)
> compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project a0fed635fe1701470062495a6ffee1c608f3f1bc)
> 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
> # install mips cross compiling tool for clang build
> # apt-get install binutils-mips-linux-gnu
> # https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0826530de3cbdc89e60a89e86def94a5f0fc81ca
> git remote add linus https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
> git fetch --no-tags linus master
> git checkout 0826530de3cbdc89e60a89e86def94a5f0fc81ca
> # save the attached .config to linux build tree
> COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 ARCH=mips
It was a bit difficult to find the warning among the thousand
>
> 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/fair.c:5381:6: warning: no previous prototype for function 'init_cfs_bandwidth' [-Wmissing-prototypes]
> void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
> ^
> kernel/sched/fair.c:5381:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
> ^
> static
> kernel/sched/fair.c:11166:6: warning: no previous prototype for function 'free_fair_sched_group' [-Wmissing-prototypes]
> void free_fair_sched_group(struct task_group *tg) { }
> ^
> kernel/sched/fair.c:11166:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> void free_fair_sched_group(struct task_group *tg) { }
> ^
> static
> kernel/sched/fair.c:11168:5: warning: no previous prototype for function 'alloc_fair_sched_group' [-Wmissing-prototypes]
> int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
> ^
> kernel/sched/fair.c:11168:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent)
> ^
> static
> kernel/sched/fair.c:11173:6: warning: no previous prototype for function 'online_fair_sched_group' [-Wmissing-prototypes]
> void online_fair_sched_group(struct task_group *tg) { }
> ^
> kernel/sched/fair.c:11173:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> void online_fair_sched_group(struct task_group *tg) { }
> ^
> static
> kernel/sched/fair.c:11175:6: warning: no previous prototype for function 'unregister_fair_sched_group' [-Wmissing-prototypes]
> void unregister_fair_sched_group(struct task_group *tg) { }
> ^
> kernel/sched/fair.c:11175:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> void unregister_fair_sched_group(struct task_group *tg) { }
> ^
> static
> kernel/sched/fair.c:486:20: warning: unused function 'list_del_leaf_cfs_rq' [-Wunused-function]
> static inline void list_del_leaf_cfs_rq(struct cfs_rq *cfs_rq)
> ^
> kernel/sched/fair.c:5362:20: warning: unused function 'sync_throttle' [-Wunused-function]
> static inline void sync_throttle(struct task_group *tg, int cpu) {}
> ^
> kernel/sched/fair.c:5387:37: warning: unused function 'tg_cfs_bandwidth' [-Wunused-function]
> static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
> ^
> kernel/sched/fair.c:5391:20: warning: unused function 'destroy_cfs_bandwidth' [-Wunused-function]
> static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
> ^
> >> kernel/sched/fair.c:9551:12: warning: stack frame size of 1376 bytes in function 'load_balance' [-Wframe-larger-than=]
> static int load_balance(int this_cpu, struct rq *this_rq,
This is triggered by the removal of code in update_sd/sg_lb_stats()
which are in the call path of load_balance.
load_balance
find_busiest_group
update_sd_lb_stats
update_sg_lb_stats
I don't expect the remove of code to increase the use of stack frame
> ^
> 10 warnings generated.
>
>
> vim +/load_balance +9551 kernel/sched/fair.c
>
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9546
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9547 /*
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9548 * Check this_cpu to ensure it is balanced within domain. Attempt to move
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9549 * tasks if there is an imbalance.
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9550 */
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 @9551 static int load_balance(int this_cpu, struct rq *this_rq,
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9552 struct sched_domain *sd, enum cpu_idle_type idle,
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9553 int *continue_balancing)
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9554 {
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9555 int ld_moved, cur_ld_moved, active_balance = 0;
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9556 struct sched_domain *sd_parent = sd->parent;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9557 struct sched_group *group;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9558 struct rq *busiest;
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9559 struct rq_flags rf;
> 4ba2968420fa9d kernel/sched/fair.c Christoph Lameter 2014-08-26 9560 struct cpumask *cpus = this_cpu_cpumask_var_ptr(load_balance_mask);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9561
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9562 struct lb_env env = {
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9563 .sd = sd,
> ddcdf6e7d9919d kernel/sched/fair.c Peter Zijlstra 2012-02-22 9564 .dst_cpu = this_cpu,
> ddcdf6e7d9919d kernel/sched/fair.c Peter Zijlstra 2012-02-22 9565 .dst_rq = this_rq,
> ae4df9d6c93510 kernel/sched/fair.c Peter Zijlstra 2017-05-01 9566 .dst_grpmask = sched_group_span(sd->groups),
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9567 .idle = idle,
> eb95308ee2a694 kernel/sched/fair.c Peter Zijlstra 2012-04-17 9568 .loop_break = sched_nr_migrate_break,
> b9403130a5350f kernel/sched/fair.c Michael Wang 2012-07-12 9569 .cpus = cpus,
> 0ec8aa00f2b4dc kernel/sched/fair.c Peter Zijlstra 2013-10-07 9570 .fbq_type = all,
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9571 .tasks = LIST_HEAD_INIT(env.tasks),
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9572 };
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9573
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9574 cpumask_and(cpus, sched_domain_span(sd), cpu_active_mask);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9575
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9576 schedstat_inc(sd->lb_count[idle]);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9577
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9578 redo:
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9579 if (!should_we_balance(&env)) {
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9580 *continue_balancing = 0;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9581 goto out_balanced;
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9582 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9583
> 23f0d2093c789e kernel/sched/fair.c Joonsoo Kim 2013-08-06 9584 group = find_busiest_group(&env);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9585 if (!group) {
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9586 schedstat_inc(sd->lb_nobusyg[idle]);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9587 goto out_balanced;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9588 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9589
> b9403130a5350f kernel/sched/fair.c Michael Wang 2012-07-12 9590 busiest = find_busiest_queue(&env, group);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9591 if (!busiest) {
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9592 schedstat_inc(sd->lb_nobusyq[idle]);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9593 goto out_balanced;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9594 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9595
> 78feefc512a091 kernel/sched/fair.c Michael Wang 2012-08-06 9596 BUG_ON(busiest == env.dst_rq);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9597
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9598 schedstat_add(sd->lb_imbalance[idle], env.imbalance);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9599
> 1aaf90a4b88aae kernel/sched/fair.c Vincent Guittot 2015-02-27 9600 env.src_cpu = busiest->cpu;
> 1aaf90a4b88aae kernel/sched/fair.c Vincent Guittot 2015-02-27 9601 env.src_rq = busiest;
> 1aaf90a4b88aae kernel/sched/fair.c Vincent Guittot 2015-02-27 9602
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9603 ld_moved = 0;
> 8a41dfcda7a32e kernel/sched/fair.c Vincent Guittot 2021-01-07 9604 /* Clear this flag as soon as we find a pullable task */
> 8a41dfcda7a32e kernel/sched/fair.c Vincent Guittot 2021-01-07 9605 env.flags |= LBF_ALL_PINNED;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9606 if (busiest->nr_running > 1) {
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9607 /*
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9608 * Attempt to move tasks. If find_busiest_group has found
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9609 * an imbalance but busiest->nr_running <= 1, the group is
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9610 * still unbalanced. ld_moved simply stays zero, so it is
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9611 * correctly treated as an imbalance.
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9612 */
> c82513e513556a kernel/sched/fair.c Peter Zijlstra 2012-04-26 9613 env.loop_max = min(sysctl_sched_nr_migrate, busiest->nr_running);
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9614
> 5d6523ebd2f67d kernel/sched/fair.c Peter Zijlstra 2012-03-10 9615 more_balance:
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9616 rq_lock_irqsave(busiest, &rf);
> 3bed5e2166a5e4 kernel/sched/fair.c Peter Zijlstra 2016-10-03 9617 update_rq_clock(busiest);
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9618
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9619 /*
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9620 * cur_ld_moved - load moved in current iteration
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9621 * ld_moved - cumulative load moved across iterations
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9622 */
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9623 cur_ld_moved = detach_tasks(&env);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9624
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9625 /*
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9626 * We've detached some tasks from busiest_rq. Every
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9627 * task is masked "TASK_ON_RQ_MIGRATING", so we can safely
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9628 * unlock busiest->lock, and we are able to be sure
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9629 * that nobody can manipulate the tasks in parallel.
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9630 * See task_rq_lock() family for the details.
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9631 */
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9632
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9633 rq_unlock(busiest, &rf);
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9634
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9635 if (cur_ld_moved) {
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9636 attach_tasks(&env);
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9637 ld_moved += cur_ld_moved;
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9638 }
> 163122b7fcfa28 kernel/sched/fair.c Kirill Tkhai 2014-08-20 9639
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9640 local_irq_restore(rf.flags);
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9641
> f1cd0858100c67 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9642 if (env.flags & LBF_NEED_BREAK) {
> f1cd0858100c67 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9643 env.flags &= ~LBF_NEED_BREAK;
> f1cd0858100c67 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9644 goto more_balance;
> f1cd0858100c67 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9645 }
> f1cd0858100c67 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9646
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9647 /*
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9648 * Revisit (affine) tasks on src_cpu that couldn't be moved to
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9649 * us and move them to an alternate dst_cpu in our sched_group
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9650 * where they can run. The upper limit on how many times we
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9651 * iterate on same src_cpu is dependent on number of CPUs in our
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9652 * sched_group.
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9653 *
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9654 * This changes load balance semantics a bit on who can move
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9655 * load to a given_cpu. In addition to the given_cpu itself
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9656 * (or a ilb_cpu acting on its behalf where given_cpu is
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9657 * nohz-idle), we now have balance_cpu in a position to move
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9658 * load to given_cpu. In rare situations, this may cause
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9659 * conflicts (balance_cpu and given_cpu/ilb_cpu deciding
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9660 * _independently_ and at _same_ time to move some load to
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9661 * given_cpu) causing exceess load to be moved to given_cpu.
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9662 * This however should not happen so much in practice and
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9663 * moreover subsequent load balance cycles should correct the
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9664 * excess load moved.
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9665 */
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9666 if ((env.flags & LBF_DST_PINNED) && env.imbalance > 0) {
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9667
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9668 /* Prevent to re-select dst_cpu via env's CPUs */
> c89d92eddfad11 kernel/sched/fair.c Viresh Kumar 2019-02-12 9669 __cpumask_clear_cpu(env.dst_cpu, env.cpus);
> 7aff2e3a56b724 kernel/sched/fair.c Vladimir Davydov 2013-09-15 9670
> 78feefc512a091 kernel/sched/fair.c Michael Wang 2012-08-06 9671 env.dst_rq = cpu_rq(env.new_dst_cpu);
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9672 env.dst_cpu = env.new_dst_cpu;
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9673 env.flags &= ~LBF_DST_PINNED;
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9674 env.loop = 0;
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9675 env.loop_break = sched_nr_migrate_break;
> e02e60c109ca70 kernel/sched/fair.c Joonsoo Kim 2013-04-23 9676
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9677 /*
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9678 * Go back to "more_balance" rather than "redo" since we
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9679 * need to continue with same src_cpu.
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9680 */
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9681 goto more_balance;
> 88b8dac0a14c51 kernel/sched/fair.c Srivatsa Vaddagiri 2012-06-19 9682 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9683
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9684 /*
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9685 * We failed to reach balance because of affinity.
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9686 */
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9687 if (sd_parent) {
> 63b2ca30bdb3db kernel/sched/fair.c Nicolas Pitre 2014-05-26 9688 int *group_imbalance = &sd_parent->groups->sgc->imbalance;
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9689
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9690 if ((env.flags & LBF_SOME_PINNED) && env.imbalance > 0)
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9691 *group_imbalance = 1;
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9692 }
> 6263322c5e8ffd kernel/sched/fair.c Peter Zijlstra 2013-08-19 9693
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9694 /* All tasks on this runqueue were pinned by CPU affinity */
> 8e45cb545d98bc kernel/sched/fair.c Peter Zijlstra 2012-02-22 9695 if (unlikely(env.flags & LBF_ALL_PINNED)) {
> c89d92eddfad11 kernel/sched/fair.c Viresh Kumar 2019-02-12 9696 __cpumask_clear_cpu(cpu_of(busiest), cpus);
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9697 /*
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9698 * Attempting to continue load balancing at the current
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9699 * sched_domain level only makes sense if there are
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9700 * active CPUs remaining as possible busiest CPUs to
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9701 * pull load from which are not contained within the
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9702 * destination group that is receiving any migrated
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9703 * load.
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9704 */
> 65a4433aebe36c kernel/sched/fair.c Jeffrey Hugo 2017-06-07 9705 if (!cpumask_subset(cpus, env.dst_grpmask)) {
> bbf18b19495942 kernel/sched/fair.c Prashanth Nageshappa 2012-06-19 9706 env.loop = 0;
> bbf18b19495942 kernel/sched/fair.c Prashanth Nageshappa 2012-06-19 9707 env.loop_break = sched_nr_migrate_break;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9708 goto redo;
> bbf18b19495942 kernel/sched/fair.c Prashanth Nageshappa 2012-06-19 9709 }
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9710 goto out_all_pinned;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9711 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9712 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9713
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9714 if (!ld_moved) {
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9715 schedstat_inc(sd->lb_failed[idle]);
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9716 /*
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9717 * Increment the failure counter only on periodic balance.
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9718 * We do not want newidle balance, which can be very
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9719 * frequent, pollute the failure counter causing
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9720 * excessive cache_hot migrations and active balances.
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9721 */
> 58b26c4c025778 kernel/sched_fair.c Venkatesh Pallipadi 2010-09-10 9722 if (idle != CPU_NEWLY_IDLE)
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9723 sd->nr_balance_failed++;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9724
> bd939f45da24e2 kernel/sched/fair.c Peter Zijlstra 2012-05-02 9725 if (need_active_balance(&env)) {
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9726 unsigned long flags;
> 8a8c69c3277886 kernel/sched/fair.c Peter Zijlstra 2016-10-04 9727
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9728 raw_spin_lock_irqsave(&busiest->lock, flags);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9729
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9730 /*
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9731 * Don't kick the active_load_balance_cpu_stop,
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9732 * if the curr task on busiest CPU can't be
> 97fb7a0a8944bd kernel/sched/fair.c Ingo Molnar 2018-03-03 9733 * moved to this_cpu:
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9734 */
> 3bd3706251ee8a kernel/sched/fair.c Sebastian Andrzej Siewior 2019-04-23 9735 if (!cpumask_test_cpu(this_cpu, busiest->curr->cpus_ptr)) {
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9736 raw_spin_unlock_irqrestore(&busiest->lock,
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9737 flags);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9738 goto out_one_pinned;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9739 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9740
> 8a41dfcda7a32e kernel/sched/fair.c Vincent Guittot 2021-01-07 9741 /* Record that we found at least one task that could run on this_cpu */
> 8a41dfcda7a32e kernel/sched/fair.c Vincent Guittot 2021-01-07 9742 env.flags &= ~LBF_ALL_PINNED;
> 8a41dfcda7a32e kernel/sched/fair.c Vincent Guittot 2021-01-07 9743
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9744 /*
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9745 * ->active_balance synchronizes accesses to
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9746 * ->active_balance_work. Once set, it's cleared
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9747 * only after active load balance is finished.
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9748 */
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9749 if (!busiest->active_balance) {
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9750 busiest->active_balance = 1;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9751 busiest->push_cpu = this_cpu;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9752 active_balance = 1;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9753 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9754 raw_spin_unlock_irqrestore(&busiest->lock, flags);
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9755
> bd939f45da24e2 kernel/sched/fair.c Peter Zijlstra 2012-05-02 9756 if (active_balance) {
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9757 stop_one_cpu_nowait(cpu_of(busiest),
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9758 active_load_balance_cpu_stop, busiest,
> 969c79215a35b0 kernel/sched_fair.c Tejun Heo 2010-05-06 9759 &busiest->active_balance_work);
> bd939f45da24e2 kernel/sched/fair.c Peter Zijlstra 2012-05-02 9760 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9761
> d02c071183e1c0 kernel/sched/fair.c Srikar Dronamraju 2016-03-23 9762 /* We've kicked active balancing, force task migration. */
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9763 sd->nr_balance_failed = sd->cache_nice_tries+1;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9764 }
> e9b9734b74656a kernel/sched/fair.c Vincent Guittot 2021-01-07 9765 } else {
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9766 sd->nr_balance_failed = 0;
> e9b9734b74656a kernel/sched/fair.c Vincent Guittot 2021-01-07 9767 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9768
> e9b9734b74656a kernel/sched/fair.c Vincent Guittot 2021-01-07 9769 if (likely(!active_balance) || need_active_balance(&env)) {
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9770 /* We were unbalanced, so reset the balancing interval */
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9771 sd->balance_interval = sd->min_interval;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9772 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9773
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9774 goto out;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9775
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9776 out_balanced:
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9777 /*
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9778 * We reach balance although we may have faced some affinity
> f6cad8df6b30a5 kernel/sched/fair.c Vincent Guittot 2019-07-01 9779 * constraints. Clear the imbalance flag only if other tasks got
> f6cad8df6b30a5 kernel/sched/fair.c Vincent Guittot 2019-07-01 9780 * a chance to move and fix the imbalance.
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9781 */
> f6cad8df6b30a5 kernel/sched/fair.c Vincent Guittot 2019-07-01 9782 if (sd_parent && !(env.flags & LBF_ALL_PINNED)) {
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9783 int *group_imbalance = &sd_parent->groups->sgc->imbalance;
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9784
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9785 if (*group_imbalance)
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9786 *group_imbalance = 0;
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9787 }
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9788
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9789 out_all_pinned:
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9790 /*
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9791 * We reach balance because all tasks are pinned at this level so
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9792 * we can't migrate them. Let the imbalance flag set so parent level
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9793 * can try to migrate them.
> afdeee0510db91 kernel/sched/fair.c Vincent Guittot 2014-08-26 9794 */
> ae92882e5646d8 kernel/sched/fair.c Josh Poimboeuf 2016-06-17 9795 schedstat_inc(sd->lb_balanced[idle]);
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9796
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9797 sd->nr_balance_failed = 0;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9798
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9799 out_one_pinned:
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9800 ld_moved = 0;
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9801
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9802 /*
> 5ba553eff0c3a7 kernel/sched/fair.c Peter Zijlstra 2019-05-29 9803 * newidle_balance() disregards balance intervals, so we could
> 5ba553eff0c3a7 kernel/sched/fair.c Peter Zijlstra 2019-05-29 9804 * repeatedly reach this code, which would lead to balance_interval
> 5ba553eff0c3a7 kernel/sched/fair.c Peter Zijlstra 2019-05-29 9805 * skyrocketting in a short amount of time. Skip the balance_interval
> 5ba553eff0c3a7 kernel/sched/fair.c Peter Zijlstra 2019-05-29 9806 * increase logic to avoid that.
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9807 */
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9808 if (env.idle == CPU_NEWLY_IDLE)
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9809 goto out;
> 3f130a37c442d5 kernel/sched/fair.c Valentin Schneider 2018-09-26 9810
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9811 /* tune up the balancing interval */
> 47b7aee14fd7e4 kernel/sched/fair.c Valentin Schneider 2018-09-26 9812 if ((env.flags & LBF_ALL_PINNED &&
> 5b54b56be5b540 kernel/sched/fair.c Peter Zijlstra 2011-09-22 9813 sd->balance_interval < MAX_PINNED_INTERVAL) ||
> 47b7aee14fd7e4 kernel/sched/fair.c Valentin Schneider 2018-09-26 9814 sd->balance_interval < sd->max_interval)
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9815 sd->balance_interval *= 2;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9816 out:
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9817 return ld_moved;
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9818 }
> 1e3c88bdeb1260 kernel/sched_fair.c Peter Zijlstra 2009-12-17 9819
>
> :::::: The code at line 9551 was first introduced by commit
> :::::: 1e3c88bdeb1260edc341e45c9fb8efd182a5c511 sched: Move load balance code into sched_fair.c
>
> :::::: TO: Peter Zijlstra <a.p.zijlstra@...llo.nl>
> :::::: CC: Ingo Molnar <mingo@...e.hu>
>
> ---
> 0-DAY CI Kernel Test Service, Intel Corporation
> https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
Powered by blists - more mailing lists