sravan/linux-tkg
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

linux 6.1: Import 6.0 based prjc patchset from ptr1337 until we can find why current upstream is frogged. Enable back pds/bmq and set as unofficial.

Browse Source 
23b7e03df8
This commit is contained in:
Tk-Glitch
2022-12-16 00:52:25 +01:00
parent 4547454e3d
commit 38e73fb059
2 changed files with 394 additions and 305 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
linux-tkg-config/prepare
View File

@@ -282,7 +282,8 @@ _set_cpu_scheduler() {
elif [ "$_kver" = "600" ]; then
_avail_cpu_scheds=("cfs" "pds" "bmq" "tt" "bore")
elif [ "$_kver" = "601" ]; then
_avail_cpu_scheds=("cfs" "tt" "bore")
_avail_cpu_scheds=("cfs" "pds" "bmq" "tt" "bore")
_projectc_unoff=1
else
_avail_cpu_scheds=("cfs")
fi

474
linux-tkg-patches/6.1/0009-prjc_v6.1-r0.patch
View File

@@ -1,8 +1,62 @@
From 4b5e85eec8c30a02e45965aa898d26ed8fdd32be Mon Sep 17 00:00:00 2001
From: Peter Jung <admin@ptr1337.dev>
Date: Mon, 12 Dec 2022 11:28:51 +0100
Subject: [PATCH] prjc-cachy
Signed-off-by: Peter Jung <admin@ptr1337.dev>
---
.../admin-guide/kernel-parameters.txt | 6 +
Documentation/admin-guide/sysctl/kernel.rst | 10 +
Documentation/scheduler/sched-BMQ.txt | 110 +
fs/proc/base.c | 2 +-
include/asm-generic/resource.h | 2 +-
include/linux/sched.h | 33 +-
include/linux/sched/deadline.h | 20 +
include/linux/sched/prio.h | 26 +
include/linux/sched/rt.h | 2 +
include/linux/sched/topology.h | 3 +-
init/Kconfig | 34 +
init/init_task.c | 18 +
kernel/Kconfig.preempt | 2 +-
kernel/cgroup/cpuset.c | 4 +-
kernel/delayacct.c | 2 +-
kernel/exit.c | 4 +-
kernel/locking/rtmutex.c | 16 +-
kernel/sched/Makefile | 5 +
kernel/sched/alt_core.c | 7971 +++++++++++++++++
kernel/sched/alt_debug.c | 31 +
kernel/sched/alt_sched.h | 658 ++
kernel/sched/bmq.h | 110 +
kernel/sched/build_policy.c | 8 +-
kernel/sched/build_utility.c | 3 +-
kernel/sched/cpufreq_schedutil.c | 10 +
kernel/sched/cputime.c | 10 +-
kernel/sched/debug.c | 10 +
kernel/sched/idle.c | 2 +
kernel/sched/pds.h | 127 +
kernel/sched/pelt.c | 4 +-
kernel/sched/pelt.h | 8 +-
kernel/sched/sched.h | 9 +
kernel/sched/stats.c | 4 +
kernel/sched/stats.h | 2 +
kernel/sched/topology.c | 17 +
kernel/sysctl.c | 15 +
kernel/time/hrtimer.c | 2 +
kernel/time/posix-cpu-timers.c | 10 +-
kernel/trace/trace_selftest.c | 5 +
39 files changed, 9292 insertions(+), 23 deletions(-)
create mode 100644 Documentation/scheduler/sched-BMQ.txt
create mode 100644 kernel/sched/alt_core.c
create mode 100644 kernel/sched/alt_debug.c
create mode 100644 kernel/sched/alt_sched.h
create mode 100644 kernel/sched/bmq.h
create mode 100644 kernel/sched/pds.h
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 42af9ca0127e..31747ec54f9d 100644
index 9ffeb6f44966..4dbc3b80f406 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -5406,6 +5406,12 @@
@@ -5415,6 +5415,12 @@
sa1100ir [NET]
See drivers/net/irda/sa1100_ir.c.
@@ -176,7 +230,7 @@ index 8874f681b056..59eb72bf7d5f 100644
[RLIMIT_RTTIME] = { RLIM_INFINITY, RLIM_INFINITY }, \
}
diff --git a/include/linux/sched.h b/include/linux/sched.h
index ffb6eb55cd13..2e730a59caa2 100644
index 5affff14993d..0fe3ce1d81c0 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -762,8 +762,14 @@ struct task_struct {
@@ -232,7 +286,7 @@ index ffb6eb55cd13..2e730a59caa2 100644
#ifdef CONFIG_CGROUP_SCHED
struct task_group *sched_task_group;
@@ -1545,6 +1567,15 @@ struct task_struct {
@@ -1542,6 +1564,15 @@ struct task_struct {
*/
};
@@ -352,10 +406,10 @@ index 816df6cc444e..c8da08e18c91 100644
#else
static inline void rebuild_sched_domains_energy(void)
diff --git a/init/Kconfig b/init/Kconfig
index 94125d3b6893..c87ba766d354 100644
index 5cf5c424fbf1..35d3ec42df0f 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -819,6 +819,7 @@ menu "Scheduler features"
@@ -836,6 +836,7 @@ menu "Scheduler features"
config UCLAMP_TASK
bool "Enable utilization clamping for RT/FAIR tasks"
depends on CPU_FREQ_GOV_SCHEDUTIL
@@ -363,21 +417,21 @@ index 94125d3b6893..c87ba766d354 100644
help
This feature enables the scheduler to track the clamped utilization
of each CPU based on RUNNABLE tasks scheduled on that CPU.
@@ -865,6 +866,35 @@ config UCLAMP_BUCKETS_COUNT
@@ -882,6 +883,35 @@ config UCLAMP_BUCKETS_COUNT
If in doubt, use the default value.
+menuconfig SCHED_ALT
+ bool "Alternative CPU Schedulers"
+ default y
+ default n
+ help
+ This feature enable alternative CPU scheduler"
+ This feature enables the ProjectC alternative CPU schedulers."
+
+if SCHED_ALT
+
+choice
+ prompt "Alternative CPU Scheduler"
+ default SCHED_BMQ
+ prompt "Alternative CPU schedulers"
+ default SCHED_PDS
+
+config SCHED_BMQ
+ bool "BMQ CPU scheduler"
@@ -399,7 +453,7 @@ index 94125d3b6893..c87ba766d354 100644
endmenu
#
@@ -918,6 +948,7 @@ config NUMA_BALANCING
@@ -935,6 +965,7 @@ config NUMA_BALANCING
depends on ARCH_SUPPORTS_NUMA_BALANCING
depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
depends on SMP && NUMA && MIGRATION && !PREEMPT_RT
@@ -407,7 +461,7 @@ index 94125d3b6893..c87ba766d354 100644
help
This option adds support for automatic NUMA aware memory/task placement.
The mechanism is quite primitive and is based on migrating memory when
@@ -1015,6 +1046,7 @@ config FAIR_GROUP_SCHED
@@ -1032,6 +1063,7 @@ config FAIR_GROUP_SCHED
depends on CGROUP_SCHED
default CGROUP_SCHED
@@ -415,7 +469,7 @@ index 94125d3b6893..c87ba766d354 100644
config CFS_BANDWIDTH
bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
depends on FAIR_GROUP_SCHED
@@ -1037,6 +1069,7 @@ config RT_GROUP_SCHED
@@ -1054,6 +1086,7 @@ config RT_GROUP_SCHED
realtime bandwidth for them.
See Documentation/scheduler/sched-rt-group.rst for more information.
@@ -423,7 +477,7 @@ index 94125d3b6893..c87ba766d354 100644
endif #CGROUP_SCHED
config UCLAMP_TASK_GROUP
@@ -1281,6 +1314,7 @@ config CHECKPOINT_RESTORE
@@ -1314,6 +1347,7 @@ config CHECKPOINT_RESTORE
config SCHED_AUTOGROUP
bool "Automatic process group scheduling"
@@ -632,10 +686,10 @@ index 976092b7bd45..31d587c16ec1 100644
obj-y += build_utility.o
diff --git a/kernel/sched/alt_core.c b/kernel/sched/alt_core.c
new file mode 100644
index 000000000000..4bea0c025475
index 000000000000..d3f6f2e1badd
--- /dev/null
+++ b/kernel/sched/alt_core.c
@@ -0,0 +1,7912 @@
@@ -0,0 +1,7971 @@
+/*
+ * kernel/sched/alt_core.c
+ *
@@ -706,7 +760,7 @@ index 000000000000..4bea0c025475
+#define sched_feat(x) (0)
+#endif /* CONFIG_SCHED_DEBUG */
+
+#define ALT_SCHED_VERSION "v6.1-r0"
+#define ALT_SCHED_VERSION "v6.1-r0-CachyOS"
+
+/* rt_prio(prio) defined in include/linux/sched/rt.h */
+#define rt_task(p) rt_prio((p)->prio)
@@ -786,7 +840,91 @@ index 000000000000..4bea0c025475
+#ifdef CONFIG_SCHED_SMT
+static cpumask_t sched_sg_idle_mask ____cacheline_aligned_in_smp;
+#endif
+static cpumask_t sched_rq_watermark[SCHED_QUEUE_BITS] ____cacheline_aligned_in_smp;
+
+#define BITS_PER_ATOMIC_LONG_T BITS_PER_LONG
+typedef struct sched_bitmask {
+ atomic_long_t bits[DIV_ROUND_UP(SCHED_QUEUE_BITS, BITS_PER_ATOMIC_LONG_T)];
+} sched_bitmask_t;
+static sched_bitmask_t sched_rq_watermark[NR_CPUS] ____cacheline_aligned_in_smp;
+
+#define x(p, set, mask) \
+ do { \
+ smp_mb__before_atomic(); \
+ if (set) \
+ atomic_long_or((mask), (p)); \
+ else \
+ atomic_long_and(~(mask), (p)); \
+ smp_mb__after_atomic(); \
+ } while (0)
+
+static __always_inline void sched_rq_watermark_fill_downwards(int cpu, unsigned int end,
+ unsigned int start, bool set)
+{
+ unsigned int start_idx, start_bit;
+ unsigned int end_idx, end_bit;
+ atomic_long_t *p;
+
+ if (end == start) {
+ return;
+ }
+
+ start_idx = start / BITS_PER_ATOMIC_LONG_T;
+ start_bit = start % BITS_PER_ATOMIC_LONG_T;
+ end_idx = (end - 1) / BITS_PER_ATOMIC_LONG_T;
+ end_bit = (end - 1) % BITS_PER_ATOMIC_LONG_T;
+ p = &sched_rq_watermark[cpu].bits[end_idx];
+
+ if (end_idx == start_idx) {
+ x(p, set, (~0UL >> (BITS_PER_ATOMIC_LONG_T - 1 - end_bit)) & (~0UL << start_bit));
+ return;
+ }
+
+ if (end_bit != BITS_PER_ATOMIC_LONG_T - 1) {
+ x(p, set, (~0UL >> (BITS_PER_ATOMIC_LONG_T - 1 - end_bit)));
+ p -= 1;
+ end_idx -= 1;
+ }
+
+ while (end_idx != start_idx) {
+ smp_mb__before_atomic();
+ atomic_long_set(p, set ? ~0UL : 0);
+ smp_mb__after_atomic();
+ p -= 1;
+ end_idx -= 1;
+ }
+
+ x(p, set, ~0UL << start_bit);
+}
+
+#undef x
+
+static __always_inline bool sched_rq_watermark_and(cpumask_t *dstp, const cpumask_t *cpus, int prio, bool not)
+{
+ int cpu;
+ bool ret = false;
+ int idx = prio / BITS_PER_ATOMIC_LONG_T;
+ int bit = prio % BITS_PER_ATOMIC_LONG_T;
+
+ cpumask_clear(dstp);
+ for_each_cpu(cpu, cpus)
+ if (test_bit(bit, (long*)&sched_rq_watermark[cpu].bits[idx].counter) == !not) {
+ __cpumask_set_cpu(cpu, dstp);
+ ret = true;
+ }
+ return ret;
+}
+
+static __always_inline bool sched_rq_watermark_test(const cpumask_t *cpus, int prio, bool not)
+{
+ int cpu;
+ int idx = prio / BITS_PER_ATOMIC_LONG_T;
+ int bit = prio % BITS_PER_ATOMIC_LONG_T;
+
+ for_each_cpu(cpu, cpus)
+ if (test_bit(bit, (long*)&sched_rq_watermark[cpu].bits[idx].counter) == !not)
+ return true;
+ return false;
+}
+
+/* sched_queue related functions */
+static inline void sched_queue_init(struct sched_queue *q)
@@ -815,7 +953,6 @@ index 000000000000..4bea0c025475
+{
+ unsigned long watermark = find_first_bit(rq->queue.bitmap, SCHED_QUEUE_BITS);
+ unsigned long last_wm = rq->watermark;
+ unsigned long i;
+ int cpu;
+
+ if (watermark == last_wm)
@@ -824,28 +961,25 @@ index 000000000000..4bea0c025475
+ rq->watermark = watermark;
+ cpu = cpu_of(rq);
+ if (watermark < last_wm) {
+ for (i = last_wm; i > watermark; i--)
+ cpumask_clear_cpu(cpu, sched_rq_watermark + SCHED_QUEUE_BITS - i);
+ sched_rq_watermark_fill_downwards(cpu, SCHED_QUEUE_BITS - watermark, SCHED_QUEUE_BITS - last_wm, false);
+#ifdef CONFIG_SCHED_SMT
+ if (static_branch_likely(&sched_smt_present) &&
+ IDLE_TASK_SCHED_PRIO == last_wm)
+ unlikely(IDLE_TASK_SCHED_PRIO == last_wm))
+ cpumask_andnot(&sched_sg_idle_mask,
+ &sched_sg_idle_mask, cpu_smt_mask(cpu));
+#endif
+ return;
+ }
+ /* last_wm < watermark */
+ for (i = watermark; i > last_wm; i--)
+ cpumask_set_cpu(cpu, sched_rq_watermark + SCHED_QUEUE_BITS - i);
+ sched_rq_watermark_fill_downwards(cpu, SCHED_QUEUE_BITS - last_wm, SCHED_QUEUE_BITS - watermark, true);
+#ifdef CONFIG_SCHED_SMT
+ if (static_branch_likely(&sched_smt_present) &&
+ IDLE_TASK_SCHED_PRIO == watermark) {
+ cpumask_t tmp;
+ if (static_branch_likely(&sched_smt_present) &&
+ unlikely(IDLE_TASK_SCHED_PRIO == watermark)) {
+ const cpumask_t *smt_mask = cpu_smt_mask(cpu);
+
+ cpumask_and(&tmp, cpu_smt_mask(cpu), sched_rq_watermark);
+ if (cpumask_equal(&tmp, cpu_smt_mask(cpu)))
+ if (!sched_rq_watermark_test(smt_mask, 0, true))
+ cpumask_or(&sched_sg_idle_mask,
+ &sched_sg_idle_mask, cpu_smt_mask(cpu));
+ &sched_sg_idle_mask, smt_mask);
+ }
+#endif
+}
@@ -928,7 +1062,8 @@ index 000000000000..4bea0c025475
+ * p->se.load, p->rt_priority,
+ * p->dl.dl_{runtime, deadline, period, flags, bw, density}
+ * - sched_setnuma(): p->numa_preferred_nid
+ * - sched_move_task(): p->sched_task_group
+ * - sched_move_task()/
+ * cpu_cgroup_fork(): p->sched_task_group
+ * - uclamp_update_active() p->uclamp*
+ *
+ * p->state <- TASK_*:
@@ -1189,7 +1324,6 @@ index 000000000000..4bea0c025475
+
+ rq->prev_irq_time += irq_delta;
+ delta -= irq_delta;
+ psi_account_irqtime(rq->curr, irq_delta);
+#endif
+#ifdef CONFIG_PARAVIRT_TIME_ACCOUNTING
+ if (static_key_false((&paravirt_steal_rq_enabled))) {
@@ -1262,15 +1396,15 @@ index 000000000000..4bea0c025475
+ rq->load_stamp = time;
+}
+
+unsigned long rq_load_util(struct rq *rq, unsigned long max)
+unsigned long rq_load_util(struct rq *rq, int cpu)
+{
+ return RQ_LOAD_HISTORY_TO_UTIL(rq->load_history) * (max >> RQ_UTIL_SHIFT);
+ return RQ_LOAD_HISTORY_TO_UTIL(rq->load_history) * (arch_scale_cpu_capacity(cpu) >> RQ_UTIL_SHIFT);
+}
+
+#ifdef CONFIG_SMP
+unsigned long sched_cpu_util(int cpu)
+{
+ return rq_load_util(cpu_rq(cpu), arch_scale_cpu_capacity(cpu));
+ return rq_load_util(cpu_rq(cpu), cpu);
+}
+#endif /* CONFIG_SMP */
+
@@ -2536,9 +2670,9 @@ index 000000000000..4bea0c025475
+#ifdef CONFIG_SCHED_SMT
+ cpumask_and(&tmp, &chk_mask, &sched_sg_idle_mask) ||
+#endif
+ cpumask_and(&tmp, &chk_mask, sched_rq_watermark) ||
+ cpumask_and(&tmp, &chk_mask,
+ sched_rq_watermark + SCHED_QUEUE_BITS - 1 - task_sched_prio(p)))
+ sched_rq_watermark_and(&tmp, &chk_mask, 0, false) ||
+ sched_rq_watermark_and(&tmp, &chk_mask,
+ SCHED_QUEUE_BITS - 1 - task_sched_prio(p), false))
+ return best_mask_cpu(task_cpu(p), &tmp);
+
+ return best_mask_cpu(task_cpu(p), &chk_mask);
@@ -2979,13 +3113,6 @@ index 000000000000..4bea0c025475
+ if (!llist)
+ return;
+
+ /*
+ * rq::ttwu_pending racy indication of out-standing wakeups.
+ * Races such that false-negatives are possible, since they
+ * are shorter lived that false-positives would be.
+ */
+ WRITE_ONCE(rq->ttwu_pending, 0);
+
+ rq_lock_irqsave(rq, &rf);
+ update_rq_clock(rq);
+
@@ -2999,6 +3126,17 @@ index 000000000000..4bea0c025475
+ ttwu_do_activate(rq, p, p->sched_remote_wakeup ? WF_MIGRATED : 0);
+ }
+
+ /*
+ * Must be after enqueueing at least once task such that
+ * idle_cpu() does not observe a false-negative -- if it does,
+ * it is possible for select_idle_siblings() to stack a number
+ * of tasks on this CPU during that window.
+ *
+ * It is ok to clear ttwu_pending when another task pending.
+ * We will receive IPI after local irq enabled and then enqueue it.
+ * Since now nr_running > 0, idle_cpu() will always get correct result.
+ */
+ WRITE_ONCE(rq->ttwu_pending, 0);
+ rq_unlock_irqrestore(rq, &rf);
+}
+
@@ -3938,8 +4076,7 @@ index 000000000000..4bea0c025475
+ * Claim the task as running, we do this before switching to it
+ * such that any running task will have this set.
+ *
+ * See the smp_load_acquire(&p->on_cpu) case in ttwu() and
+ * its ordering comment.
+ * See the ttwu() WF_ON_CPU case and its ordering comment.
+ */
+ WRITE_ONCE(next->on_cpu, 1);
+}
@@ -4009,7 +4146,7 @@ index 000000000000..4bea0c025475
+ if (likely(!head))
+ return NULL;
+
+ lockdep_assert_rq_held(rq);
+ lockdep_assert_held(&rq->lock);
+ /*
+ * Must not take balance_push_callback off the list when
+ * splice_balance_callbacks() and balance_callbacks() are not
@@ -4678,7 +4815,7 @@ index 000000000000..4bea0c025475
+ * find potential cpus which can migrate the current running task
+ */
+ if (cpumask_test_cpu(cpu, &sched_sg_idle_mask) &&
+ cpumask_andnot(&chk, cpu_online_mask, sched_rq_watermark) &&
+ sched_rq_watermark_and(&chk, cpu_online_mask, 0, true) &&
+ cpumask_andnot(&chk, &chk, &sched_rq_pending_mask)) {
+ int i;
+
@@ -4820,7 +4957,7 @@ index 000000000000..4bea0c025475
+int __init sched_tick_offload_init(void)
+{
+ tick_work_cpu = alloc_percpu(struct tick_work);
+ BUG_ON(!tick_work_cpu);
+ WARN_ON_ONCE(!tick_work_cpu);
+ return 0;
+}
+
@@ -4986,9 +5123,8 @@ index 000000000000..4bea0c025475
+#ifdef ALT_SCHED_DEBUG
+void alt_sched_debug(void)
+{
+ printk(KERN_INFO "sched: pending: 0x%04lx, idle: 0x%04lx, sg_idle: 0x%04lx\n",
+ printk(KERN_INFO "sched: pending: 0x%04lx, sg_idle: 0x%04lx\n",
+ sched_rq_pending_mask.bits[0],
+ sched_rq_watermark[0].bits[0],
+ sched_sg_idle_mask.bits[0]);
+}
+#else
@@ -4997,23 +5133,18 @@ index 000000000000..4bea0c025475
+
+#ifdef CONFIG_SMP
+
+#ifdef CONFIG_PREEMPT_RT
+#define SCHED_NR_MIGRATE_BREAK 8
+#else
+#define SCHED_NR_MIGRATE_BREAK 32
+#endif
+
+const_debug unsigned int sysctl_sched_nr_migrate = SCHED_NR_MIGRATE_BREAK;
+
+#define SCHED_RQ_NR_MIGRATION (32U)
+/*
+ * Migrate pending tasks in @rq to @dest_cpu
+ * Will try to migrate mininal of half of @rq nr_running tasks and
+ * SCHED_RQ_NR_MIGRATION to @dest_cpu
+ */
+static inline int
+migrate_pending_tasks(struct rq *rq, struct rq *dest_rq, const int dest_cpu)
+{
+ struct task_struct *p, *skip = rq->curr;
+ int nr_migrated = 0;
+ int nr_tries = min(rq->nr_running / 2, sysctl_sched_nr_migrate);
+ int nr_tries = min(rq->nr_running / 2, SCHED_RQ_NR_MIGRATION);
+
+ while (skip != rq->idle && nr_tries &&
+ (p = sched_rq_next_task(skip, rq)) != rq->idle) {
@@ -5631,7 +5762,7 @@ index 000000000000..4bea0c025475
+ enum ctx_state prev_state;
+
+ /* Catch callers which need to be fixed */
+ BUG_ON(preempt_count() || !irqs_disabled());
+ WARN_ON_ONCE(preempt_count() || !irqs_disabled());
+
+ prev_state = exception_enter();
+
@@ -5806,29 +5937,17 @@ index 000000000000..4bea0c025475
+EXPORT_SYMBOL(set_user_nice);
+
+/*
+ * is_nice_reduction - check if nice value is an actual reduction
+ *
+ * Similar to can_nice() but does not perform a capability check.
+ *
+ * @p: task
+ * @nice: nice value
+ */
+static bool is_nice_reduction(const struct task_struct *p, const int nice)
+{
+ /* Convert nice value [19,-20] to rlimit style value [1,40]: */
+ int nice_rlim = nice_to_rlimit(nice);
+
+ return (nice_rlim <= task_rlimit(p, RLIMIT_NICE));
+}
+
+/*
+ * can_nice - check if a task can reduce its nice value
+ * @p: task
+ * @nice: nice value
+ */
+int can_nice(const struct task_struct *p, const int nice)
+{
+ return is_nice_reduction(p, nice) || capable(CAP_SYS_NICE);
+ /* Convert nice value [19,-20] to rlimit style value [1,40] */
+ int nice_rlim = nice_to_rlimit(nice);
+
+ return (nice_rlim <= task_rlimit(p, RLIMIT_NICE) ||
+ capable(CAP_SYS_NICE));
+}
+
+#ifdef __ARCH_WANT_SYS_NICE
@@ -5979,45 +6098,6 @@ index 000000000000..4bea0c025475
+ return match;
+}
+
+/*
+ * Allow unprivileged RT tasks to decrease priority.
+ * Only issue a capable test if needed and only once to avoid an audit
+ * event on permitted non-privileged operations:
+ */
+static int user_check_sched_setscheduler(struct task_struct *p,
+ const struct sched_attr *attr,
+ int policy, int reset_on_fork)
+{
+ if (rt_policy(policy)) {
+ unsigned long rlim_rtprio = task_rlimit(p, RLIMIT_RTPRIO);
+
+ /* Can't set/change the rt policy: */
+ if (policy != p->policy && !rlim_rtprio)
+ goto req_priv;
+
+ /* Can't increase priority: */
+ if (attr->sched_priority > p->rt_priority &&
+ attr->sched_priority > rlim_rtprio)
+ goto req_priv;
+ }
+
+ /* Can't change other user's priorities: */
+ if (!check_same_owner(p))
+ goto req_priv;
+
+ /* Normal users shall not reset the sched_reset_on_fork flag: */
+ if (p->sched_reset_on_fork && !reset_on_fork)
+ goto req_priv;
+
+ return 0;
+
+req_priv:
+ if (!capable(CAP_SYS_NICE))
+ return -EPERM;
+
+ return 0;
+}
+
+static int __sched_setscheduler(struct task_struct *p,
+ const struct sched_attr *attr,
+ bool user, bool pi)
@@ -6037,7 +6117,7 @@ index 000000000000..4bea0c025475
+ raw_spinlock_t *lock;
+
+ /* The pi code expects interrupts enabled */
+ BUG_ON(pi && in_interrupt());
+ WARN_ON_ONCE(pi && in_interrupt());
+
+ /*
+ * Alt schedule FW supports SCHED_DEADLINE by squash it as prio 0 SCHED_FIFO
@@ -6074,11 +6154,34 @@ index 000000000000..4bea0c025475
+ (attr->sched_priority != 0))
+ return -EINVAL;
+
+ if (user) {
+ retval = user_check_sched_setscheduler(p, attr, policy, reset_on_fork);
+ if (retval)
+ return retval;
+ /*
+ * Allow unprivileged RT tasks to decrease priority:
+ */
+ if (user && !capable(CAP_SYS_NICE)) {
+ if (SCHED_FIFO == policy || SCHED_RR == policy) {
+ unsigned long rlim_rtprio =
+ task_rlimit(p, RLIMIT_RTPRIO);
+
+ /* Can't set/change the rt policy */
+ if (policy != p->policy && !rlim_rtprio)
+ return -EPERM;
+
+ /* Can't increase priority */
+ if (attr->sched_priority > p->rt_priority &&
+ attr->sched_priority > rlim_rtprio)
+ return -EPERM;
+ }
+
+ /* Can't change other user's priorities */
+ if (!check_same_owner(p))
+ return -EPERM;
+
+ /* Normal users shall not reset the sched_reset_on_fork flag */
+ if (p->sched_reset_on_fork && !reset_on_fork)
+ return -EPERM;
+ }
+
+ if (user) {
+ retval = security_task_setscheduler(p);
+ if (retval)
+ return retval;
@@ -7515,7 +7618,7 @@ index 000000000000..4bea0c025475
+{
+ struct mm_struct *mm = current->active_mm;
+
+ BUG_ON(current != this_rq()->idle);
+ WARN_ON_ONCE(current != this_rq()->idle);
+
+ if (mm != &init_mm) {
+ switch_mm(mm, &init_mm, current);
@@ -8029,8 +8132,17 @@ index 000000000000..4bea0c025475
+ wait_bit_init();
+
+#ifdef CONFIG_SMP
+ for (i = 0; i < SCHED_QUEUE_BITS; i++)
+ cpumask_copy(sched_rq_watermark + i, cpu_present_mask);
+ for (i = 0; i < nr_cpu_ids; i++) {
+ long val = cpumask_test_cpu(i, cpu_present_mask) ? -1L : 0;
+ int j;
+ for (j = 0; j < DIV_ROUND_UP(SCHED_QUEUE_BITS, BITS_PER_ATOMIC_LONG_T); j++)
+ atomic_long_set(&sched_rq_watermark[i].bits[j], val);
+ }
+ for (i = nr_cpu_ids; i < NR_CPUS; i++) {
+ int j;
+ for (j = 0; j < DIV_ROUND_UP(SCHED_QUEUE_BITS, BITS_PER_ATOMIC_LONG_T); j++)
+ atomic_long_set(&sched_rq_watermark[i].bits[j], 0);
+ }
+#endif
+
+#ifdef CONFIG_CGROUP_SCHED
@@ -8456,12 +8568,14 @@ index 000000000000..4bea0c025475
+ sched_unregister_group(tg);
+}
+
+#ifdef CONFIG_RT_GROUP_SCHED
+static void cpu_cgroup_fork(struct task_struct *task)
+{
+}
+
+static int cpu_cgroup_can_attach(struct cgroup_taskset *tset)
+{
+ return 0;
+}
+#endif
+
+static void cpu_cgroup_attach(struct cgroup_taskset *tset)
+{
@@ -8535,9 +8649,8 @@ index 000000000000..4bea0c025475
+ .css_released = cpu_cgroup_css_released,
+ .css_free = cpu_cgroup_css_free,
+ .css_extra_stat_show = cpu_extra_stat_show,
+#ifdef CONFIG_RT_GROUP_SCHED
+ .fork = cpu_cgroup_fork,
+ .can_attach = cpu_cgroup_can_attach,
+#endif
+ .attach = cpu_cgroup_attach,
+ .legacy_cftypes = cpu_files,
+ .legacy_cftypes = cpu_legacy_files,
@@ -8587,10 +8700,10 @@ index 000000000000..1212a031700e
+{}
diff --git a/kernel/sched/alt_sched.h b/kernel/sched/alt_sched.h
new file mode 100644
index 000000000000..7c1cc0cbca0d
index 000000000000..6df234aacdd7
--- /dev/null
+++ b/kernel/sched/alt_sched.h
@@ -0,0 +1,660 @@
@@ -0,0 +1,658 @@
+#ifndef ALT_SCHED_H
+#define ALT_SCHED_H
+
@@ -8673,15 +8786,6 @@ index 000000000000..7c1cc0cbca0d
+#define MAX_SHARES (1UL << 18)
+#endif
+
+/*
+ * Tunables that become constants when CONFIG_SCHED_DEBUG is off:
+ */
+#ifdef CONFIG_SCHED_DEBUG
+# define const_debug __read_mostly
+#else
+# define const_debug const
+#endif
+
+/* task_struct::on_rq states: */
+#define TASK_ON_RQ_QUEUED 1
+#define TASK_ON_RQ_MIGRATING 2
@@ -8838,7 +8942,7 @@ index 000000000000..7c1cc0cbca0d
+#endif /* CONFIG_NO_HZ_COMMON */
+};
+
+extern unsigned long rq_load_util(struct rq *rq, unsigned long max);
+extern unsigned long rq_load_util(struct rq *rq, int cpu);
+
+extern unsigned long calc_load_update;
+extern atomic_long_t calc_load_tasks;
@@ -8990,6 +9094,13 @@ index 000000000000..7c1cc0cbca0d
+}
+
+static inline void
+rq_lock_irq(struct rq *rq, struct rq_flags *rf)
+ __acquires(rq->lock)
+{
+ raw_spin_lock_irq(&rq->lock);
+}
+
+static inline void
+rq_lock(struct rq *rq, struct rq_flags *rf)
+ __acquires(rq->lock)
+{
@@ -9393,10 +9504,18 @@ index d9dc9ab3773f..71a25540d65e 100644
+#include "deadline.c"
+#endif
diff --git a/kernel/sched/build_utility.c b/kernel/sched/build_utility.c
index 99bdd96f454f..23f80a86d2d7 100644
index 99bdd96f454f..bc17d5a6fc41 100644
--- a/kernel/sched/build_utility.c
+++ b/kernel/sched/build_utility.c
@@ -85,7 +85,9 @@
@@ -34,7 +34,6 @@
#include <linux/nospec.h>
#include <linux/proc_fs.h>
#include <linux/psi.h>
-#include <linux/psi.h>
#include <linux/ptrace_api.h>
#include <linux/sched_clock.h>
#include <linux/security.h>
@@ -85,7 +84,9 @@
#ifdef CONFIG_SMP
# include "cpupri.c"
@@ -9407,7 +9526,7 @@ index 99bdd96f454f..23f80a86d2d7 100644
#endif
diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c
index 1207c78f85c1..68812e0756cb 100644
index 1207c78f85c1..f66b715e4287 100644
--- a/kernel/sched/cpufreq_schedutil.c
+++ b/kernel/sched/cpufreq_schedutil.c
@@ -159,9 +159,14 @@ static void sugov_get_util(struct sugov_cpu *sg_cpu)
@@ -9420,7 +9539,7 @@ index 1207c78f85c1..68812e0756cb 100644
FREQUENCY_UTIL, NULL);
+#else
+ sg_cpu->bw_dl = 0;
+ sg_cpu->util = rq_load_util(rq, sg_cpu->max);
+ sg_cpu->util = rq_load_util(rq, sg_cpu->cpu);
+#endif /* CONFIG_SCHED_ALT */
}
@@ -9504,7 +9623,7 @@ index 95fc77853743..b48b3f9ed47f 100644
if (task_cputime(p, &cputime.utime, &cputime.stime))
diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c
index 1637b65ba07a..033c6deeb515 100644
index a8377d0e5ebd..b6e8e386bbfc 100644
--- a/kernel/sched/debug.c
+++ b/kernel/sched/debug.c
@@ -7,6 +7,7 @@
@@ -9555,7 +9674,7 @@ index 1637b65ba07a..033c6deeb515 100644
debugfs_create_u32("latency_ns", 0644, debugfs_sched, &sysctl_sched_latency);
debugfs_create_u32("min_granularity_ns", 0644, debugfs_sched, &sysctl_sched_min_granularity);
debugfs_create_u32("idle_min_granularity_ns", 0644, debugfs_sched, &sysctl_sched_idle_min_granularity);
@@ -337,11 +344,13 @@ static __init int sched_init_debug(void)
@@ -339,11 +346,13 @@ static __init int sched_init_debug(void)
#endif
debugfs_create_file("debug", 0444, debugfs_sched, NULL, &sched_debug_fops);
@@ -9569,7 +9688,7 @@ index 1637b65ba07a..033c6deeb515 100644
#ifdef CONFIG_SMP
static cpumask_var_t sd_sysctl_cpus;
@@ -1068,6 +1077,7 @@ void proc_sched_set_task(struct task_struct *p)
@@ -1070,6 +1079,7 @@ void proc_sched_set_task(struct task_struct *p)
memset(&p->stats, 0, sizeof(p->stats));
#endif
}
@@ -9728,7 +9847,7 @@ index 000000000000..56a649d02e49
+#endif
+static inline void sched_task_deactivate(struct task_struct *p, struct rq *rq) {}
diff --git a/kernel/sched/pelt.c b/kernel/sched/pelt.c
index 0f310768260c..bd38bf738fe9 100644
index 036b0e2cd2b4..a00ed09127bd 100644
--- a/kernel/sched/pelt.c
+++ b/kernel/sched/pelt.c
@@ -266,6 +266,7 @@ ___update_load_avg(struct sched_avg *sa, unsigned long load)
@@ -9751,7 +9870,7 @@ index 0f310768260c..bd38bf738fe9 100644
* thermal:
*
diff --git a/kernel/sched/pelt.h b/kernel/sched/pelt.h
index 3a0e0dc28721..e8a7d84aa5a5 100644
index 9b35b5072bae..6e457b864d66 100644
--- a/kernel/sched/pelt.h
+++ b/kernel/sched/pelt.h
@@ -1,13 +1,15 @@
@@ -9779,7 +9898,7 @@ index 3a0e0dc28721..e8a7d84aa5a5 100644
static inline void cfs_se_util_change(struct sched_avg *avg)
{
unsigned int enqueued;
@@ -180,9 +183,11 @@ static inline u64 cfs_rq_clock_pelt(struct cfs_rq *cfs_rq)
@@ -209,9 +212,11 @@ static inline u64 cfs_rq_clock_pelt(struct cfs_rq *cfs_rq)
return rq_clock_pelt(rq_of(cfs_rq));
}
#endif
@@ -9791,7 +9910,7 @@ index 3a0e0dc28721..e8a7d84aa5a5 100644
static inline int
update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq)
{
@@ -200,6 +205,7 @@ update_dl_rq_load_avg(u64 now, struct rq *rq, int running)
@@ -229,6 +234,7 @@ update_dl_rq_load_avg(u64 now, struct rq *rq, int running)
{
return 0;
}
@@ -9800,7 +9919,7 @@ index 3a0e0dc28721..e8a7d84aa5a5 100644
static inline int
update_thermal_load_avg(u64 now, struct rq *rq, u64 capacity)
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index a4a20046e586..c363693cd869 100644
index 2c89aaa9200f..21d2d4a188fa 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -5,6 +5,10 @@
@@ -9814,7 +9933,7 @@ index a4a20046e586..c363693cd869 100644
#include <linux/sched/affinity.h>
#include <linux/sched/autogroup.h>
#include <linux/sched/cpufreq.h>
@@ -3183,4 +3187,9 @@ static inline void update_current_exec_runtime(struct task_struct *curr,
@@ -3264,4 +3268,9 @@ static inline void update_current_exec_runtime(struct task_struct *curr,
cgroup_account_cputime(curr, delta_exec);
}
@@ -9856,7 +9975,7 @@ index 857f837f52cb..5486c63e4790 100644
}
return 0;
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 84a188913cc9..53934e7ef5db 100644
index 38f3698f5e5b..b9d597394316 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -89,6 +89,7 @@ static inline void rq_sched_info_depart (struct rq *rq, unsigned long long delt
@@ -9876,7 +9995,7 @@ index 84a188913cc9..53934e7ef5db 100644
#ifdef CONFIG_PSI
void psi_task_change(struct task_struct *task, int clear, int set);
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 8739c2a5a54e..d8dd6c15eb47 100644
index dea9fa39e7c0..b401e6423102 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -3,6 +3,7 @@
@@ -9931,7 +10050,7 @@ index 8739c2a5a54e..d8dd6c15eb47 100644
+#endif /* CONFIG_NUMA */
+#endif
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index c6d9dec11b74..2bc42ce8b48e 100644
index 1c7c7c953876..f9dc7d89a6d2 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -93,6 +93,10 @@ EXPORT_SYMBOL_GPL(sysctl_long_vals);
@@ -9945,23 +10064,7 @@ index c6d9dec11b74..2bc42ce8b48e 100644
#ifdef CONFIG_PERF_EVENTS
static const int six_hundred_forty_kb = 640 * 1024;
#endif
@@ -1633,6 +1637,7 @@ int proc_do_static_key(struct ctl_table *table, int write,
}
static struct ctl_table kern_table[] = {
+#ifndef CONFIG_SCHED_ALT
#ifdef CONFIG_NUMA_BALANCING
{
.procname = "numa_balancing",
@@ -1652,6 +1657,7 @@ static struct ctl_table kern_table[] = {
.extra1 = SYSCTL_ZERO,
},
#endif /* CONFIG_NUMA_BALANCING */
+#endif /* !CONFIG_SCHED_ALT */
{
.procname = "panic",
.data = &panic_timeout,
@@ -1953,6 +1959,17 @@ static struct ctl_table kern_table[] = {
@@ -1965,6 +1969,17 @@ static struct ctl_table kern_table[] = {
.proc_handler = proc_dointvec,
},
#endif
@@ -10074,21 +10177,6 @@ index a2d301f58ced..2ccdede8585c 100644
};
struct wakeup_test_data *x = data;
diff --git a/kernel/sched/alt_sched.h b/kernel/sched/alt_sched.h
index 93ff3bddd36f85c67a3130a68ed67e2525513353..a00bc84b93b2c8789d4798a2900299fdd39a2e58 100644
--- a/kernel/sched/alt_sched.h
+++ b/kernel/sched/alt_sched.h
@@ -387,6 +387,13 @@ task_rq_unlock(struct rq *rq, struct task_struct *p, struct rq_flags *rf)
raw_spin_unlock_irqrestore(&p->pi_lock, rf->flags);
}
--
2.39.0.rc2.1.gbd5df96b79
+static inline void
+rq_lock_irq(struct rq *rq, struct rq_flags *rf)
+ __acquires(rq->lock)
+{
+ raw_spin_lock_irq(&rq->lock);
+}
+
static inline void
rq_lock(struct rq *rq, struct rq_flags *rf)
__acquires(rq->lock)