Re: [RFC PATCH v3 2/6] sched/uclamp: Track a new util_avg_bias signal

Re: [RFC PATCH v3 2/6] sched/uclamp: Track a new util_avg_bias signal

[kernel test robot]

Hi Hongyan,

[This is a private test report for your RFC patch.]
kernel test robot noticed the following build errors:

[auto build test ERROR on tip/sched/core]
[also build test ERROR on tip/master next-20240509]
[cannot apply to rafael-pm/linux-next rafael-pm/bleeding-edge linus/master tip/auto-latest peterz-queue/sched/core v6.9-rc7]
[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/Hongyan-Xia/Revert-sched-uclamp-Set-max_spare_cap_cpu-even-if-max_spare_cap-is-0/20240507-205300
base:   tip/sched/core
patch link:    https://lore.kernel.org/r/2e43dc6b376ea6d785976a398b5d9ffe823cf35d.1715082714.git.hongyan.xia2%40arm.com
patch subject: [RFC PATCH v3 2/6] sched/uclamp: Track a new util_avg_bias signal
config: arm-allnoconfig (https://download.01.org/0day-ci/archive/20240510/202405100428.Ao1oE7iL-lkp@intel.com/config)
compiler: clang version 19.0.0git (https://github.com/llvm/llvm-project b910bebc300dafb30569cecc3017b446ea8eafa0)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240510/202405100428.Ao1oE7iL-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/202405100428.Ao1oE7iL-lkp@intel.com/

All errors (new ones prefixed by >>):

   In file included from kernel/sched/fair.c:27:
   In file included from include/linux/mm_api.h:1:
   In file included from include/linux/mm.h:2208:
   include/linux/vmstat.h:522:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
     522 |         return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_"
         |                               ~~~~~~~~~~~ ^ ~~~
>> kernel/sched/fair.c:6805:29: error: no member named 'avg' in 'struct cfs_rq'
    6805 |         util_bias_enqueue(&rq->cfs.avg, p);
         |                            ~~~~~~~ ^
   kernel/sched/fair.c:6899:29: error: no member named 'avg' in 'struct cfs_rq'
    6899 |         util_bias_dequeue(&rq->cfs.avg, p);
         |                            ~~~~~~~ ^
   1 warning and 2 errors generated.


vim +6805 kernel/sched/fair.c

  6736	
  6737	/*
  6738	 * The enqueue_task method is called before nr_running is
  6739	 * increased. Here we update the fair scheduling stats and
  6740	 * then put the task into the rbtree:
  6741	 */
  6742	static void
  6743	enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
  6744	{
  6745		struct cfs_rq *cfs_rq;
  6746		struct sched_entity *se = &p->se;
  6747		int idle_h_nr_running = task_has_idle_policy(p);
  6748		int task_new = !(flags & ENQUEUE_WAKEUP);
  6749	
  6750		/*
  6751		 * The code below (indirectly) updates schedutil which looks at
  6752		 * the cfs_rq utilization to select a frequency.
  6753		 * Let's add the task's estimated utilization to the cfs_rq's
  6754		 * estimated utilization, before we update schedutil.
  6755		 */
  6756		util_est_enqueue(&rq->cfs, p);
  6757	
  6758		/*
  6759		 * If in_iowait is set, the code below may not trigger any cpufreq
  6760		 * utilization updates, so do it here explicitly with the IOWAIT flag
  6761		 * passed.
  6762		 */
  6763		if (p->in_iowait)
  6764			cpufreq_update_util(rq, SCHED_CPUFREQ_IOWAIT);
  6765	
  6766		for_each_sched_entity(se) {
  6767			if (se->on_rq)
  6768				break;
  6769			cfs_rq = cfs_rq_of(se);
  6770			enqueue_entity(cfs_rq, se, flags);
  6771	
  6772			cfs_rq->h_nr_running++;
  6773			cfs_rq->idle_h_nr_running += idle_h_nr_running;
  6774	
  6775			if (cfs_rq_is_idle(cfs_rq))
  6776				idle_h_nr_running = 1;
  6777	
  6778			/* end evaluation on encountering a throttled cfs_rq */
  6779			if (cfs_rq_throttled(cfs_rq))
  6780				goto enqueue_throttle;
  6781	
  6782			flags = ENQUEUE_WAKEUP;
  6783		}
  6784	
  6785		for_each_sched_entity(se) {
  6786			cfs_rq = cfs_rq_of(se);
  6787	
  6788			update_load_avg(cfs_rq, se, UPDATE_TG);
  6789			se_update_runnable(se);
  6790			update_cfs_group(se);
  6791	
  6792			cfs_rq->h_nr_running++;
  6793			cfs_rq->idle_h_nr_running += idle_h_nr_running;
  6794	
  6795			if (cfs_rq_is_idle(cfs_rq))
  6796				idle_h_nr_running = 1;
  6797	
  6798			/* end evaluation on encountering a throttled cfs_rq */
  6799			if (cfs_rq_throttled(cfs_rq))
  6800				goto enqueue_throttle;
  6801		}
  6802	
  6803		/* At this point se is NULL and we are at root level*/
  6804		add_nr_running(rq, 1);
> 6805		util_bias_enqueue(&rq->cfs.avg, p);
  6806		/* XXX: We should skip the update above and only do it once here. */
  6807		cpufreq_update_util(rq, 0);
  6808	
  6809		/*
  6810		 * Since new tasks are assigned an initial util_avg equal to
  6811		 * half of the spare capacity of their CPU, tiny tasks have the
  6812		 * ability to cross the overutilized threshold, which will
  6813		 * result in the load balancer ruining all the task placement
  6814		 * done by EAS. As a way to mitigate that effect, do not account
  6815		 * for the first enqueue operation of new tasks during the
  6816		 * overutilized flag detection.
  6817		 *
  6818		 * A better way of solving this problem would be to wait for
  6819		 * the PELT signals of tasks to converge before taking them
  6820		 * into account, but that is not straightforward to implement,
  6821		 * and the following generally works well enough in practice.
  6822		 */
  6823		if (!task_new)
  6824			check_update_overutilized_status(rq);
  6825	
  6826	enqueue_throttle:
  6827		assert_list_leaf_cfs_rq(rq);
  6828	
  6829		hrtick_update(rq);
  6830	}
  6831	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki