]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
sched/core: Reorganize cgroup bandwidth control interface file reads
authorTejun Heo <tj@kernel.org>
Sat, 14 Jun 2025 01:23:29 +0000 (15:23 -1000)
committerPeter Zijlstra <peterz@infradead.org>
Wed, 18 Jun 2025 11:59:57 +0000 (13:59 +0200)
- Update tg_get_cfs_*() to return u64 values. These are now used as the low
  level accessors to the fair's bandwidth configuration parameters.
  Translation to usecs takes place in these functions.

- Add tg_bandwidth() which reads all three bandwidth parameters using
  tg_get_cfs_*().

- Reimplement cgroup interface read functions using tg_bandwidth(). Drop cfs
  from the function names.

This is to prepare for adding bandwidth control support to sched_ext.
tg_bandwidth() will be used as the muxing point similar to tg_weight(). No
functional changes.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20250614012346.2358261-4-tj@kernel.org
kernel/sched/core.c

index dc9668a6592f98c58ba4fc16728dcb21de8557c1..8de93a3bfa2762526f3bf7da2aa7ee74ee35bcee 100644 (file)
@@ -9404,7 +9404,7 @@ static int tg_set_cfs_bandwidth(struct task_group *tg, u64 period, u64 quota,
        return 0;
 }
 
-static long tg_get_cfs_period(struct task_group *tg)
+static u64 tg_get_cfs_period(struct task_group *tg)
 {
        u64 cfs_period_us;
 
@@ -9414,12 +9414,12 @@ static long tg_get_cfs_period(struct task_group *tg)
        return cfs_period_us;
 }
 
-static long tg_get_cfs_quota(struct task_group *tg)
+static u64 tg_get_cfs_quota(struct task_group *tg)
 {
        u64 quota_us;
 
        if (tg->cfs_bandwidth.quota == RUNTIME_INF)
-               return -1;
+               return RUNTIME_INF;
 
        quota_us = tg->cfs_bandwidth.quota;
        do_div(quota_us, NSEC_PER_USEC);
@@ -9427,7 +9427,7 @@ static long tg_get_cfs_quota(struct task_group *tg)
        return quota_us;
 }
 
-static long tg_get_cfs_burst(struct task_group *tg)
+static u64 tg_get_cfs_burst(struct task_group *tg)
 {
        u64 burst_us;
 
@@ -9614,22 +9614,42 @@ static int cpu_cfs_local_stat_show(struct seq_file *sf, void *v)
        return 0;
 }
 
-static u64 cpu_cfs_period_read_u64(struct cgroup_subsys_state *css,
-                                  struct cftype *cft)
+static void tg_bandwidth(struct task_group *tg,
+                        u64 *period_us_p, u64 *quota_us_p, u64 *burst_us_p)
 {
-       return tg_get_cfs_period(css_tg(css));
+       if (period_us_p)
+               *period_us_p = tg_get_cfs_period(tg);
+       if (quota_us_p)
+               *quota_us_p = tg_get_cfs_quota(tg);
+       if (burst_us_p)
+               *burst_us_p = tg_get_cfs_burst(tg);
 }
 
-static s64 cpu_cfs_quota_read_s64(struct cgroup_subsys_state *css,
-                                 struct cftype *cft)
+static u64 cpu_period_read_u64(struct cgroup_subsys_state *css,
+                              struct cftype *cft)
 {
-       return tg_get_cfs_quota(css_tg(css));
+       u64 period_us;
+
+       tg_bandwidth(css_tg(css), &period_us, NULL, NULL);
+       return period_us;
 }
 
-static u64 cpu_cfs_burst_read_u64(struct cgroup_subsys_state *css,
-                                 struct cftype *cft)
+static s64 cpu_quota_read_s64(struct cgroup_subsys_state *css,
+                             struct cftype *cft)
 {
-       return tg_get_cfs_burst(css_tg(css));
+       u64 quota_us;
+
+       tg_bandwidth(css_tg(css), NULL, &quota_us, NULL);
+       return quota_us;        /* (s64)RUNTIME_INF becomes -1 */
+}
+
+static u64 cpu_burst_read_u64(struct cgroup_subsys_state *css,
+                             struct cftype *cft)
+{
+       u64 burst_us;
+
+       tg_bandwidth(css_tg(css), NULL, NULL, &burst_us);
+       return burst_us;
 }
 
 static int cpu_cfs_period_write_u64(struct cgroup_subsys_state *css,
@@ -9712,17 +9732,17 @@ static struct cftype cpu_legacy_files[] = {
 #ifdef CONFIG_CFS_BANDWIDTH
        {
                .name = "cfs_period_us",
-               .read_u64 = cpu_cfs_period_read_u64,
+               .read_u64 = cpu_period_read_u64,
                .write_u64 = cpu_cfs_period_write_u64,
        },
        {
                .name = "cfs_quota_us",
-               .read_s64 = cpu_cfs_quota_read_s64,
+               .read_s64 = cpu_quota_read_s64,
                .write_s64 = cpu_cfs_quota_write_s64,
        },
        {
                .name = "cfs_burst_us",
-               .read_u64 = cpu_cfs_burst_read_u64,
+               .read_u64 = cpu_burst_read_u64,
                .write_u64 = cpu_cfs_burst_write_u64,
        },
        {
@@ -9944,8 +9964,10 @@ static int __maybe_unused cpu_period_quota_parse(char *buf,
 static int cpu_max_show(struct seq_file *sf, void *v)
 {
        struct task_group *tg = css_tg(seq_css(sf));
+       u64 period_us, quota_us;
 
-       cpu_period_quota_print(sf, tg_get_cfs_period(tg), tg_get_cfs_quota(tg));
+       tg_bandwidth(tg, &period_us, &quota_us, NULL);
+       cpu_period_quota_print(sf, period_us, quota_us);
        return 0;
 }
 
@@ -9996,7 +10018,7 @@ static struct cftype cpu_files[] = {
        {
                .name = "max.burst",
                .flags = CFTYPE_NOT_ON_ROOT,
-               .read_u64 = cpu_cfs_burst_read_u64,
+               .read_u64 = cpu_burst_read_u64,
                .write_u64 = cpu_cfs_burst_write_u64,
        },
 #endif /* CONFIG_CFS_BANDWIDTH */