sched_ext, sched/core: Don't call scx_group_set_weight() prematurely from sched_creat...

During task_group creation, sched_create_group() calls
scx_group_set_weight() with CGROUP_WEIGHT_DFL to initialize the sched_ext
portion. This is premature and ends up calling ops.cgroup_set_weight() with
an incorrect @cgrp before ops.cgroup_init() is called.

sched_create_group() should just initialize SCX related fields in the new
task_group. Fix it by factoring out scx_tg_init() from sched_init() and
making sched_create_group() call that function instead of
scx_group_set_weight().

v2: Retain CONFIG_EXT_GROUP_SCHED ifdef in sched_init() as removing it leads
    to build failures on !CONFIG_GROUP_SCHED configs.

Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: 819513666966 ("sched_ext: Add cgroup support")
Cc: stable@vger.kernel.org # v6.12+

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index dce50fa57471dffc4311b9d393ae300a43d38d20..8988d38d46a388ba88407b210a5a2e49066f49ea 100644 (file)
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -8545,7 +8545,7 @@ void __init sched_init(void)
                init_cfs_bandwidth(&root_task_group.cfs_bandwidth, NULL);
 #endif /* CONFIG_FAIR_GROUP_SCHED */
 #ifdef CONFIG_EXT_GROUP_SCHED
-               root_task_group.scx_weight = CGROUP_WEIGHT_DFL;
+               scx_tg_init(&root_task_group);
 #endif /* CONFIG_EXT_GROUP_SCHED */
 #ifdef CONFIG_RT_GROUP_SCHED
                root_task_group.rt_se = (struct sched_rt_entity **)ptr;
@@ -8985,7 +8985,7 @@ struct task_group *sched_create_group(struct task_group *parent)
        if (!alloc_rt_sched_group(tg, parent))
                goto err;
 
-       scx_group_set_weight(tg, CGROUP_WEIGHT_DFL);
+       scx_tg_init(tg);
        alloc_uclamp_sched_group(tg, parent);
 
        return tg;

diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index 33a0d8c6ff95fb0059d8f8620c4b6a6b49b622d6..b498d867ba210a94128db5bfb17583ac44eb884a 100644 (file)
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -4092,6 +4092,11 @@ bool scx_can_stop_tick(struct rq *rq)
 DEFINE_STATIC_PERCPU_RWSEM(scx_cgroup_rwsem);
 static bool scx_cgroup_enabled;
 
+void scx_tg_init(struct task_group *tg)
+{
+       tg->scx_weight = CGROUP_WEIGHT_DFL;
+}
+
 int scx_tg_online(struct task_group *tg)
 {
        struct scx_sched *sch = scx_root;

diff --git a/kernel/sched/ext.h b/kernel/sched/ext.h
index 6e5072f577718e08b0ba329c050b0a1fc986e460..a75835c23f15003c35b36556988328f0fc2fbc59 100644 (file)
--- a/kernel/sched/ext.h
+++ b/kernel/sched/ext.h
@@ -79,6 +79,7 @@ static inline void scx_update_idle(struct rq *rq, bool idle, bool do_notify) {}
 
 #ifdef CONFIG_CGROUP_SCHED
 #ifdef CONFIG_EXT_GROUP_SCHED
+void scx_tg_init(struct task_group *tg);
 int scx_tg_online(struct task_group *tg);
 void scx_tg_offline(struct task_group *tg);
 int scx_cgroup_can_attach(struct cgroup_taskset *tset);
@@ -88,6 +89,7 @@ void scx_cgroup_cancel_attach(struct cgroup_taskset *tset);
 void scx_group_set_weight(struct task_group *tg, unsigned long cgrp_weight);
 void scx_group_set_idle(struct task_group *tg, bool idle);
 #else  /* CONFIG_EXT_GROUP_SCHED */
+static inline void scx_tg_init(struct task_group *tg) {}
 static inline int scx_tg_online(struct task_group *tg) { return 0; }
 static inline void scx_tg_offline(struct task_group *tg) {}
 static inline int scx_cgroup_can_attach(struct cgroup_taskset *tset) { return 0; }