commit: bbd5b42d3ff847b17f85f7ce29fa19f28f88b798 Author: Mike Pagano <mpagano <AT> gentoo <DOT> org> AuthorDate: Mon Jan 13 17:18:55 2025 +0000 Commit: Mike Pagano <mpagano <AT> gentoo <DOT> org> CommitDate: Mon Jan 13 17:18:55 2025 +0000 URL: https://gitweb.gentoo.org/proj/linux-patches.git/commit/?id=bbd5b42d
BMQ(BitMap Queue) Schedule r1 version bump Signed-off-by: Mike Pagano <mpagano <AT> gentoo.org> 0000_README | 2 +- ...=> 5020_BMQ-and-PDS-io-scheduler-v6.12-r1.patch | 252 +++++++++++++++------ 2 files changed, 184 insertions(+), 70 deletions(-) diff --git a/0000_README b/0000_README index 29d9187b..06b9cb3f 100644 --- a/0000_README +++ b/0000_README @@ -127,7 +127,7 @@ Patch: 5010_enable-cpu-optimizations-universal.patch From: https://github.com/graysky2/kernel_compiler_patch Desc: Kernel >= 5.15 patch enables gcc = v11.1+ optimizations for additional CPUs. -Patch: 5020_BMQ-and-PDS-io-scheduler-v6.12-r0.patch +Patch: 5020_BMQ-and-PDS-io-scheduler-v6.12-r1.patch From: https://gitlab.com/alfredchen/projectc Desc: BMQ(BitMap Queue) Scheduler. A new CPU scheduler developed from PDS(incld). Inspired by the scheduler in zircon. diff --git a/5020_BMQ-and-PDS-io-scheduler-v6.12-r0.patch b/5020_BMQ-and-PDS-io-scheduler-v6.12-r1.patch similarity index 98% rename from 5020_BMQ-and-PDS-io-scheduler-v6.12-r0.patch rename to 5020_BMQ-and-PDS-io-scheduler-v6.12-r1.patch index 9eb3139f..532813fd 100644 --- a/5020_BMQ-and-PDS-io-scheduler-v6.12-r0.patch +++ b/5020_BMQ-and-PDS-io-scheduler-v6.12-r1.patch @@ -158,7 +158,7 @@ index 8874f681b056..59eb72bf7d5f 100644 [RLIMIT_RTTIME] = { RLIM_INFINITY, RLIM_INFINITY }, \ } diff --git a/include/linux/sched.h b/include/linux/sched.h -index bb343136ddd0..212d9204e9aa 100644 +index bb343136ddd0..6adfea989b7b 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -804,9 +804,13 @@ struct task_struct { @@ -212,7 +212,34 @@ index bb343136ddd0..212d9204e9aa 100644 #ifdef CONFIG_CGROUP_SCHED struct task_group *sched_task_group; -@@ -1609,6 +1628,15 @@ struct task_struct { +@@ -878,11 +897,15 @@ struct task_struct { + const cpumask_t *cpus_ptr; + cpumask_t *user_cpus_ptr; + cpumask_t cpus_mask; ++#ifndef CONFIG_SCHED_ALT + void *migration_pending; ++#endif + #ifdef CONFIG_SMP + unsigned short migration_disabled; + #endif ++#ifndef CONFIG_SCHED_ALT + unsigned short migration_flags; ++#endif + + #ifdef CONFIG_PREEMPT_RCU + int rcu_read_lock_nesting; +@@ -914,8 +937,10 @@ struct task_struct { + + struct list_head tasks; + #ifdef CONFIG_SMP ++#ifndef CONFIG_SCHED_ALT + struct plist_node pushable_tasks; + struct rb_node pushable_dl_tasks; ++#endif + #endif + + struct mm_struct *mm; +@@ -1609,6 +1634,15 @@ struct task_struct { */ }; @@ -228,7 +255,7 @@ index bb343136ddd0..212d9204e9aa 100644 #define TASK_REPORT_IDLE (TASK_REPORT + 1) #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1) -@@ -2135,7 +2163,11 @@ static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) +@@ -2135,7 +2169,11 @@ static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) static inline bool task_is_runnable(struct task_struct *p) { @@ -341,7 +368,7 @@ index 4237daa5ac7a..3cebd93c49c8 100644 #else static inline void rebuild_sched_domains_energy(void) diff --git a/init/Kconfig b/init/Kconfig -index c521e1421ad4..131a599fcde2 100644 +index c521e1421ad4..4a397b48a453 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -652,6 +652,7 @@ config TASK_IO_ACCOUNTING @@ -352,15 +379,7 @@ index c521e1421ad4..131a599fcde2 100644 select KERNFS help Collect metrics that indicate how overcommitted the CPU, memory, -@@ -817,6 +818,7 @@ menu "Scheduler features" - config UCLAMP_TASK - bool "Enable utilization clamping for RT/FAIR tasks" - depends on CPU_FREQ_GOV_SCHEDUTIL -+ depends on !SCHED_ALT - help - This feature enables the scheduler to track the clamped utilization - of each CPU based on RUNNABLE tasks scheduled on that CPU. -@@ -863,6 +865,35 @@ config UCLAMP_BUCKETS_COUNT +@@ -863,6 +864,35 @@ config UCLAMP_BUCKETS_COUNT If in doubt, use the default value. @@ -396,7 +415,7 @@ index c521e1421ad4..131a599fcde2 100644 endmenu # -@@ -928,6 +959,7 @@ config NUMA_BALANCING +@@ -928,6 +958,7 @@ config NUMA_BALANCING depends on ARCH_SUPPORTS_NUMA_BALANCING depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY depends on SMP && NUMA && MIGRATION && !PREEMPT_RT @@ -404,23 +423,7 @@ index c521e1421ad4..131a599fcde2 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 -@@ -1036,6 +1068,7 @@ menuconfig CGROUP_SCHED - tasks. - - if CGROUP_SCHED -+if !SCHED_ALT - config GROUP_SCHED_WEIGHT - def_bool n - -@@ -1073,6 +1106,7 @@ config EXT_GROUP_SCHED - select GROUP_SCHED_WEIGHT - default y - -+endif #!SCHED_ALT - endif #CGROUP_SCHED - - config SCHED_MM_CID -@@ -1334,6 +1368,7 @@ config CHECKPOINT_RESTORE +@@ -1334,6 +1365,7 @@ config CHECKPOINT_RESTORE config SCHED_AUTOGROUP bool "Automatic process group scheduling" @@ -429,7 +432,7 @@ index c521e1421ad4..131a599fcde2 100644 select CGROUP_SCHED select FAIR_GROUP_SCHED diff --git a/init/init_task.c b/init/init_task.c -index 136a8231355a..03770079619a 100644 +index 136a8231355a..12c01ab8e718 100644 --- a/init/init_task.c +++ b/init/init_task.c @@ -71,9 +71,16 @@ struct task_struct init_task __aligned(L1_CACHE_BYTES) = { @@ -466,14 +469,20 @@ index 136a8231355a..03770079619a 100644 .se = { .group_node = LIST_HEAD_INIT(init_task.se.group_node), }, -@@ -93,6 +110,7 @@ struct task_struct init_task __aligned(L1_CACHE_BYTES) = { +@@ -93,10 +110,13 @@ struct task_struct init_task __aligned(L1_CACHE_BYTES) = { .run_list = LIST_HEAD_INIT(init_task.rt.run_list), .time_slice = RR_TIMESLICE, }, +#endif .tasks = LIST_HEAD_INIT(init_task.tasks), ++#ifndef CONFIG_SCHED_ALT #ifdef CONFIG_SMP .pushable_tasks = PLIST_NODE_INIT(init_task.pushable_tasks, MAX_PRIO), + #endif ++#endif + #ifdef CONFIG_CGROUP_SCHED + .sched_task_group = &root_task_group, + #endif diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt index fe782cd77388..d27d2154d71a 100644 --- a/kernel/Kconfig.preempt @@ -700,10 +709,10 @@ 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..c59691742340 +index 000000000000..0a08bc0176ac --- /dev/null +++ b/kernel/sched/alt_core.c -@@ -0,0 +1,7458 @@ +@@ -0,0 +1,7515 @@ +/* + * kernel/sched/alt_core.c + * @@ -782,7 +791,7 @@ index 000000000000..c59691742340 +#define sched_feat(x) (0) +#endif /* CONFIG_SCHED_DEBUG */ + -+#define ALT_SCHED_VERSION "v6.12-r0" ++#define ALT_SCHED_VERSION "v6.12-r1" + +#define STOP_PRIO (MAX_RT_PRIO - 1) + @@ -2144,8 +2153,6 @@ index 000000000000..c59691742340 + __set_task_cpu(p, new_cpu); +} + -+#define MDF_FORCE_ENABLED 0x80 -+ +static void +__do_set_cpus_ptr(struct task_struct *p, const struct cpumask *new_mask) +{ @@ -2186,8 +2193,6 @@ index 000000000000..c59691742340 + if (cpumask_test_cpu(cpu, &p->cpus_mask)) { + cpu_rq(cpu)->nr_pinned++; + p->migration_disabled = 1; -+ p->migration_flags &= ~MDF_FORCE_ENABLED; -+ + /* + * Violates locking rules! see comment in __do_set_cpus_ptr(). + */ @@ -2237,6 +2242,15 @@ index 000000000000..c59691742340 +} +EXPORT_SYMBOL_GPL(migrate_enable); + ++static void __migrate_force_enable(struct task_struct *p, struct rq *rq) ++{ ++ if (likely(p->cpus_ptr != &p->cpus_mask)) ++ __do_set_cpus_ptr(p, &p->cpus_mask); ++ p->migration_disabled = 0; ++ /* When p is migrate_disabled, rq->lock should be held */ ++ rq->nr_pinned--; ++} ++ +static inline bool rq_has_pinned_tasks(struct rq *rq) +{ + return rq->nr_pinned; @@ -2417,6 +2431,9 @@ index 000000000000..c59691742340 + + __do_set_cpus_allowed(p, &ac); + ++ if (is_migration_disabled(p) && !cpumask_test_cpu(task_cpu(p), &p->cpus_mask)) ++ __migrate_force_enable(p, task_rq(p)); ++ + /* + * Because this is called with p->pi_lock held, it is not possible + * to use kfree() here (when PREEMPT_RT=y), therefore punt to using @@ -2712,14 +2729,8 @@ index 000000000000..c59691742340 +{ + /* Can the task run on the task's current CPU? If so, we're done */ + if (!cpumask_test_cpu(task_cpu(p), &p->cpus_mask)) { -+ if (p->migration_disabled) { -+ if (likely(p->cpus_ptr != &p->cpus_mask)) -+ __do_set_cpus_ptr(p, &p->cpus_mask); -+ p->migration_disabled = 0; -+ p->migration_flags |= MDF_FORCE_ENABLED; -+ /* When p is migrate_disabled, rq->lock should be held */ -+ rq->nr_pinned--; -+ } ++ if (is_migration_disabled(p)) ++ __migrate_force_enable(p, rq); + + if (task_on_cpu(p) || READ_ONCE(p->__state) == TASK_WAKING) { + struct migration_arg arg = { p, dest_cpu }; @@ -7178,9 +7189,6 @@ index 000000000000..c59691742340 + if (preempt_count() > 0) + return; + -+ if (current->migration_flags & MDF_FORCE_ENABLED) -+ return; -+ + if (time_before(jiffies, prev_jiffy + HZ) && prev_jiffy) + return; + prev_jiffy = jiffies; @@ -7374,6 +7382,43 @@ index 000000000000..c59691742340 +{ +} + ++#ifdef CONFIG_GROUP_SCHED_WEIGHT ++static int sched_group_set_shares(struct task_group *tg, unsigned long shares) ++{ ++ return 0; ++} ++ ++static int sched_group_set_idle(struct task_group *tg, long idle) ++{ ++ return 0; ++} ++ ++static int cpu_shares_write_u64(struct cgroup_subsys_state *css, ++ struct cftype *cftype, u64 shareval) ++{ ++ return sched_group_set_shares(css_tg(css), shareval); ++} ++ ++static u64 cpu_shares_read_u64(struct cgroup_subsys_state *css, ++ struct cftype *cft) ++{ ++ return 0; ++} ++ ++static s64 cpu_idle_read_s64(struct cgroup_subsys_state *css, ++ struct cftype *cft) ++{ ++ return 0; ++} ++ ++static int cpu_idle_write_s64(struct cgroup_subsys_state *css, ++ struct cftype *cft, s64 idle) ++{ ++ return sched_group_set_idle(css_tg(css), idle); ++} ++#endif ++ ++#ifdef CONFIG_CFS_BANDWIDTH +static s64 cpu_cfs_quota_read_s64(struct cgroup_subsys_state *css, + struct cftype *cft) +{ @@ -7419,7 +7464,9 @@ index 000000000000..c59691742340 +{ + return 0; +} ++#endif + ++#ifdef CONFIG_RT_GROUP_SCHED +static int cpu_rt_runtime_write(struct cgroup_subsys_state *css, + struct cftype *cft, s64 val) +{ @@ -7443,7 +7490,9 @@ index 000000000000..c59691742340 +{ + return 0; +} ++#endif + ++#ifdef CONFIG_UCLAMP_TASK_GROUP +static int cpu_uclamp_min_show(struct seq_file *sf, void *v) +{ + return 0; @@ -7467,8 +7516,22 @@ index 000000000000..c59691742340 +{ + return nbytes; +} ++#endif + +static struct cftype cpu_legacy_files[] = { ++#ifdef CONFIG_GROUP_SCHED_WEIGHT ++ { ++ .name = "shares", ++ .read_u64 = cpu_shares_read_u64, ++ .write_u64 = cpu_shares_write_u64, ++ }, ++ { ++ .name = "idle", ++ .read_s64 = cpu_idle_read_s64, ++ .write_s64 = cpu_idle_write_s64, ++ }, ++#endif ++#ifdef CONFIG_CFS_BANDWIDTH + { + .name = "cfs_quota_us", + .read_s64 = cpu_cfs_quota_read_s64, @@ -7492,6 +7555,8 @@ index 000000000000..c59691742340 + .name = "stat.local", + .seq_show = cpu_cfs_local_stat_show, + }, ++#endif ++#ifdef CONFIG_RT_GROUP_SCHED + { + .name = "rt_runtime_us", + .read_s64 = cpu_rt_runtime_read, @@ -7502,6 +7567,8 @@ index 000000000000..c59691742340 + .read_u64 = cpu_rt_period_read_uint, + .write_u64 = cpu_rt_period_write_uint, + }, ++#endif ++#ifdef CONFIG_UCLAMP_TASK_GROUP + { + .name = "uclamp.min", + .flags = CFTYPE_NOT_ON_ROOT, @@ -7514,9 +7581,11 @@ index 000000000000..c59691742340 + .seq_show = cpu_uclamp_max_show, + .write = cpu_uclamp_max_write, + }, ++#endif + { } /* Terminate */ +}; + ++#ifdef CONFIG_GROUP_SCHED_WEIGHT +static u64 cpu_weight_read_u64(struct cgroup_subsys_state *css, + struct cftype *cft) +{ @@ -7540,19 +7609,9 @@ index 000000000000..c59691742340 +{ + return 0; +} ++#endif + -+static s64 cpu_idle_read_s64(struct cgroup_subsys_state *css, -+ struct cftype *cft) -+{ -+ return 0; -+} -+ -+static int cpu_idle_write_s64(struct cgroup_subsys_state *css, -+ struct cftype *cft, s64 idle) -+{ -+ return 0; -+} -+ ++#ifdef CONFIG_CFS_BANDWIDTH +static int cpu_max_show(struct seq_file *sf, void *v) +{ + return 0; @@ -7563,8 +7622,10 @@ index 000000000000..c59691742340 +{ + return nbytes; +} ++#endif + +static struct cftype cpu_files[] = { ++#ifdef CONFIG_GROUP_SCHED_WEIGHT + { + .name = "weight", + .flags = CFTYPE_NOT_ON_ROOT, @@ -7583,6 +7644,8 @@ index 000000000000..c59691742340 + .read_s64 = cpu_idle_read_s64, + .write_s64 = cpu_idle_write_s64, + }, ++#endif ++#ifdef CONFIG_CFS_BANDWIDTH + { + .name = "max", + .flags = CFTYPE_NOT_ON_ROOT, @@ -7595,6 +7658,8 @@ index 000000000000..c59691742340 + .read_u64 = cpu_cfs_burst_read_u64, + .write_u64 = cpu_cfs_burst_write_u64, + }, ++#endif ++#ifdef CONFIG_UCLAMP_TASK_GROUP + { + .name = "uclamp.min", + .flags = CFTYPE_NOT_ON_ROOT, @@ -7607,6 +7672,7 @@ index 000000000000..c59691742340 + .seq_show = cpu_uclamp_max_show, + .write = cpu_uclamp_max_write, + }, ++#endif + { } /* terminate */ +}; + @@ -8421,10 +8487,10 @@ index 000000000000..1dbd7eb6a434 +{} diff --git a/kernel/sched/alt_sched.h b/kernel/sched/alt_sched.h new file mode 100644 -index 000000000000..09c9e9f80bf4 +index 000000000000..7fb3433c5c41 --- /dev/null +++ b/kernel/sched/alt_sched.h -@@ -0,0 +1,971 @@ +@@ -0,0 +1,997 @@ +#ifndef _KERNEL_SCHED_ALT_SCHED_H +#define _KERNEL_SCHED_ALT_SCHED_H + @@ -9120,15 +9186,41 @@ index 000000000000..09c9e9f80bf4 + +static inline void nohz_run_idle_balance(int cpu) { } + -+static inline -+unsigned long uclamp_rq_util_with(struct rq *rq, unsigned long util, -+ struct task_struct *p) ++static inline unsigned long ++uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id) +{ -+ return util; ++ if (clamp_id == UCLAMP_MIN) ++ return 0; ++ ++ return SCHED_CAPACITY_SCALE; +} + +static inline bool uclamp_rq_is_capped(struct rq *rq) { return false; } + ++static inline bool uclamp_is_used(void) ++{ ++ return false; ++} ++ ++static inline unsigned long ++uclamp_rq_get(struct rq *rq, enum uclamp_id clamp_id) ++{ ++ if (clamp_id == UCLAMP_MIN) ++ return 0; ++ ++ return SCHED_CAPACITY_SCALE; ++} ++ ++static inline void ++uclamp_rq_set(struct rq *rq, enum uclamp_id clamp_id, unsigned int value) ++{ ++} ++ ++static inline bool uclamp_rq_is_idle(struct rq *rq) ++{ ++ return false; ++} ++ +#ifdef CONFIG_SCHED_MM_CID + +#define SCHED_MM_CID_PERIOD_NS (100ULL * 1000000) /* 100ms */ @@ -11109,6 +11201,28 @@ index 6bcee4704059..cf88205fd4a2 100644 return false; } +diff --git a/kernel/trace/trace_osnoise.c b/kernel/trace/trace_osnoise.c +index a50ed23bee77..be0477666049 100644 +--- a/kernel/trace/trace_osnoise.c ++++ b/kernel/trace/trace_osnoise.c +@@ -1665,6 +1665,9 @@ static void osnoise_sleep(bool skip_period) + */ + static inline int osnoise_migration_pending(void) + { ++#ifdef CONFIG_SCHED_ALT ++ return 0; ++#else + if (!current->migration_pending) + return 0; + +@@ -1686,6 +1689,7 @@ static inline int osnoise_migration_pending(void) + mutex_unlock(&interface_lock); + + return 1; ++#endif + } + + /* diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 1469dd8075fa..803527a0e48a 100644 --- a/kernel/trace/trace_selftest.c
