Scheduler changes for v6.13:

- Core facilities:
 
     - Add the "Lazy preemption" model (CONFIG_PREEMPT_LAZY=y), which optimizes
       fair-class preemption by delaying preemption requests to the
       tick boundary, while working as full preemption for RR/FIFO/DEADLINE
       classes. (Peter Zijlstra)
 
         - x86: Enable Lazy preemption (Peter Zijlstra)
         - riscv: Enable Lazy preemption (Jisheng Zhang)
 
     - Initialize idle tasks only once (Thomas Gleixner)
 
     - sched/ext: Remove sched_fork() hack (Thomas Gleixner)
 
  - Fair scheduler:
     - Optimize the PLACE_LAG when se->vlag is zero (Huang Shijie)
 
  - Idle loop:
       Optimize the generic idle loop by removing unnecessary
       memory barrier (Zhongqiu Han)
 
  - RSEQ:
     - Improve cache locality of RSEQ concurrency IDs for
       intermittent workloads (Mathieu Desnoyers)
 
  - Waitqueues:
     - Make wake_up_{bit,var} less fragile (Neil Brown)
 
  - PSI:
     - Pass enqueue/dequeue flags to psi callbacks directly (Johannes Weiner)
 
  - Preparatory patches for proxy execution:
     - core: Add move_queued_task_locked helper (Connor O'Brien)
     - core: Consolidate pick_*_task to task_is_pushable helper (Connor O'Brien)
     - core: Split out __schedule() deactivate task logic into a helper (John Stultz)
     - core: Split scheduler and execution contexts (Peter Zijlstra)
     - locking/mutex: Make mutex::wait_lock irq safe (Juri Lelli)
     - locking/mutex: Expose __mutex_owner() (Juri Lelli)
     - locking/mutex: Remove wakeups from under mutex::wait_lock (Peter Zijlstra)
 
  - Misc fixes and cleanups:
     - core: Remove unused __HAVE_THREAD_FUNCTIONS hook support (David Disseldorp)
     - core: Update the comment for TIF_NEED_RESCHED_LAZY (Sebastian Andrzej Siewior)
     - wait: Remove unused bit_wait_io_timeout (Dr. David Alan Gilbert)
     - fair: remove the DOUBLE_TICK feature (Huang Shijie)
     - fair: fix the comment for PREEMPT_SHORT (Huang Shijie)
     - uclamp: Fix unnused variable warning (Christian Loehle)
     - rt: No PREEMPT_RT=y for all{yes,mod}config
 
 Signed-off-by: Ingo Molnar <mingo@kernel.org>
 -----BEGIN PGP SIGNATURE-----
 
 iQJFBAABCgAvFiEEBpT5eoXrXCwVQwEKEnMQ0APhK1gFAmc7fnQRHG1pbmdvQGtl
 cm5lbC5vcmcACgkQEnMQ0APhK1hZTBAAozVdWA2m51aNa67HvAZta/olmrIagVbW
 inwbTgqa8b+UfeWEuKOfrZr5khjEh6pLgR3dBTib1uH6xxYj/Okds+qbPWSBPVLh
 yzavlm/zJZM1U1XtxE3eyVfqWik4GrY7DoIMDQQr+YH7rNXonJeJkll38OI2E5MC
 q3Q01qyMo8RJJX8qkf3f8ObOoP/51NsVniTw0Zb2fzEhXz8FjezLlxk6cMfgSkJG
 lg9gfIwUZ7Xg5neRo4kJcc3Ht31KYOhWSiupBJzRD1hss/N/AybvMcTX/Cm8d07w
 HIAdDDAn84o46miFo/a0V/hsJZ72idWbqxVJUCtaezrpOUiFkG+uInRvG/ynr0lF
 5dEI9f+6PUw8Nc7L72IyHkobjPqS2IefSaxYYCBKmxMX2qrenfTor/pKiWzzhBIl
 rX3MZSuUJ8NjV4rNGD/qXRM1IsMJrsDwxDyv+sRec3XdH33x286ds6aAUEPDQ6N7
 96VS0sOKcNUJN8776ErNjlIxRl8HTlpkaO3nZlQIfXgTlXUpRvOuKbEWqP+606lo
 oANgJTKgUhgJPWZnvmdRxDjSiOp93QcImjus9i1tN81FGiEDleONsJUxu2Di1E5+
 s1nCiytjq+cdvzCqFyiOZUh+g6kSZ4yXxNgLg2UvbXzX1zOeUQT3WtyKUhMPXhU8
 esh1TgbUbpE=
 =Zcqj
 -----END PGP SIGNATURE-----

Merge tag 'sched-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler updates from Ingo Molnar:
 "Core facilities:

   - Add the "Lazy preemption" model (CONFIG_PREEMPT_LAZY=y), which
     optimizes fair-class preemption by delaying preemption requests to
     the tick boundary, while working as full preemption for
     RR/FIFO/DEADLINE classes. (Peter Zijlstra)
        - x86: Enable Lazy preemption (Peter Zijlstra)
        - riscv: Enable Lazy preemption (Jisheng Zhang)

   - Initialize idle tasks only once (Thomas Gleixner)

   - sched/ext: Remove sched_fork() hack (Thomas Gleixner)

  Fair scheduler:

   - Optimize the PLACE_LAG when se->vlag is zero (Huang Shijie)

  Idle loop:

   - Optimize the generic idle loop by removing unnecessary memory
     barrier (Zhongqiu Han)

  RSEQ:

   - Improve cache locality of RSEQ concurrency IDs for intermittent
     workloads (Mathieu Desnoyers)

  Waitqueues:

   - Make wake_up_{bit,var} less fragile (Neil Brown)

  PSI:

   - Pass enqueue/dequeue flags to psi callbacks directly (Johannes
     Weiner)

  Preparatory patches for proxy execution:

   - Add move_queued_task_locked helper (Connor O'Brien)

   - Consolidate pick_*_task to task_is_pushable helper (Connor O'Brien)

   - Split out __schedule() deactivate task logic into a helper (John
     Stultz)

   - Split scheduler and execution contexts (Peter Zijlstra)

   - Make mutex::wait_lock irq safe (Juri Lelli)

   - Expose __mutex_owner() (Juri Lelli)

   - Remove wakeups from under mutex::wait_lock (Peter Zijlstra)

  Misc fixes and cleanups:

   - Remove unused __HAVE_THREAD_FUNCTIONS hook support (David
     Disseldorp)

   - Update the comment for TIF_NEED_RESCHED_LAZY (Sebastian Andrzej
     Siewior)

   - Remove unused bit_wait_io_timeout (Dr. David Alan Gilbert)

   - remove the DOUBLE_TICK feature (Huang Shijie)

   - fix the comment for PREEMPT_SHORT (Huang Shijie)

   - Fix unnused variable warning (Christian Loehle)

   - No PREEMPT_RT=y for all{yes,mod}config"

* tag 'sched-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (33 commits)
  sched, x86: Update the comment for TIF_NEED_RESCHED_LAZY.
  sched: No PREEMPT_RT=y for all{yes,mod}config
  riscv: add PREEMPT_LAZY support
  sched, x86: Enable Lazy preemption
  sched: Enable PREEMPT_DYNAMIC for PREEMPT_RT
  sched: Add Lazy preemption model
  sched: Add TIF_NEED_RESCHED_LAZY infrastructure
  sched/ext: Remove sched_fork() hack
  sched: Initialize idle tasks only once
  sched: psi: pass enqueue/dequeue flags to psi callbacks directly
  sched/uclamp: Fix unnused variable warning
  sched: Split scheduler and execution contexts
  sched: Split out __schedule() deactivate task logic into a helper
  sched: Consolidate pick_*_task to task_is_pushable helper
  sched: Add move_queued_task_locked helper
  locking/mutex: Expose __mutex_owner()
  locking/mutex: Make mutex::wait_lock irq safe
  locking/mutex: Remove wakeups from under mutex::wait_lock
  sched: Improve cache locality of RSEQ concurrency IDs for intermittent workloads
  sched: idle: Optimize the generic idle loop by removing needless memory barrier
  ...

arch/riscv/Kconfig

@@ -39,6 +39,7 @@ config RISCV
 	select ARCH_HAS_MMIOWB
 	select ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
 	select ARCH_HAS_PMEM_API
+	select ARCH_HAS_PREEMPT_LAZY
 	select ARCH_HAS_PREPARE_SYNC_CORE_CMD
 	select ARCH_HAS_PTE_DEVMAP if 64BIT && MMU
 	select ARCH_HAS_PTE_SPECIAL

arch/riscv/include/asm/thread_info.h

@@ -107,9 +107,10 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src);
  * - pending work-to-be-done flags are in lowest half-word
  * - other flags in upper half-word(s)
  */
-#define TIF_NOTIFY_RESUME	1	/* callback before returning to user */
+#define TIF_NEED_RESCHED	0	/* rescheduling necessary */
-#define TIF_SIGPENDING		2	/* signal pending */
+#define TIF_NEED_RESCHED_LAZY	1       /* Lazy rescheduling needed */
-#define TIF_NEED_RESCHED	3	/* rescheduling necessary */
+#define TIF_NOTIFY_RESUME	2	/* callback before returning to user */
+#define TIF_SIGPENDING		3	/* signal pending */
 #define TIF_RESTORE_SIGMASK	4	/* restore signal mask in do_signal() */
 #define TIF_MEMDIE		5	/* is terminating due to OOM killer */
 #define TIF_NOTIFY_SIGNAL	9	/* signal notifications exist */
@@ -117,9 +118,10 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src);
 #define TIF_32BIT		11	/* compat-mode 32bit process */
 #define TIF_RISCV_V_DEFER_RESTORE	12 /* restore Vector before returing to user */
 
+#define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
+#define _TIF_NEED_RESCHED_LAZY	(1 << TIF_NEED_RESCHED_LAZY)
 #define _TIF_NOTIFY_RESUME	(1 << TIF_NOTIFY_RESUME)
 #define _TIF_SIGPENDING		(1 << TIF_SIGPENDING)
-#define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
 #define _TIF_NOTIFY_SIGNAL	(1 << TIF_NOTIFY_SIGNAL)
 #define _TIF_UPROBE		(1 << TIF_UPROBE)
 #define _TIF_RISCV_V_DEFER_RESTORE	(1 << TIF_RISCV_V_DEFER_RESTORE)

arch/x86/Kconfig

@@ -93,6 +93,7 @@ config X86
 	select ARCH_HAS_NMI_SAFE_THIS_CPU_OPS
 	select ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
 	select ARCH_HAS_PMEM_API		if X86_64
+	select ARCH_HAS_PREEMPT_LAZY
 	select ARCH_HAS_PTE_DEVMAP		if X86_64
 	select ARCH_HAS_PTE_SPECIAL
 	select ARCH_HAS_HW_PTE_YOUNG

arch/x86/include/asm/thread_info.h

@@ -87,8 +87,9 @@ struct thread_info {
 #define TIF_NOTIFY_RESUME	1	/* callback before returning to user */
 #define TIF_SIGPENDING		2	/* signal pending */
 #define TIF_NEED_RESCHED	3	/* rescheduling necessary */
-#define TIF_SINGLESTEP		4	/* reenable singlestep on user return*/
+#define TIF_NEED_RESCHED_LAZY	4	/* Lazy rescheduling needed */
-#define TIF_SSBD		5	/* Speculative store bypass disable */
+#define TIF_SINGLESTEP		5	/* reenable singlestep on user return*/
+#define TIF_SSBD		6	/* Speculative store bypass disable */
 #define TIF_SPEC_IB		9	/* Indirect branch speculation mitigation */
 #define TIF_SPEC_L1D_FLUSH	10	/* Flush L1D on mm switches (processes) */
 #define TIF_USER_RETURN_NOTIFY	11	/* notify kernel of userspace return */
@@ -110,6 +111,7 @@ struct thread_info {
 #define _TIF_NOTIFY_RESUME	(1 << TIF_NOTIFY_RESUME)
 #define _TIF_SIGPENDING		(1 << TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
+#define _TIF_NEED_RESCHED_LAZY	(1 << TIF_NEED_RESCHED_LAZY)
 #define _TIF_SINGLESTEP		(1 << TIF_SINGLESTEP)
 #define _TIF_SSBD		(1 << TIF_SSBD)
 #define _TIF_SPEC_IB		(1 << TIF_SPEC_IB)

fs/exec.c

@@ -990,7 +990,7 @@ static int exec_mmap(struct mm_struct *mm)
 	active_mm = tsk->active_mm;
 	tsk->active_mm = mm;
 	tsk->mm = mm;
-	mm_init_cid(mm);
+	mm_init_cid(mm, tsk);
 	/*
 	 * This prevents preemption while active_mm is being loaded and
 	 * it and mm are being updated, which could cause problems for

include/linux/entry-common.h

@@ -64,7 +64,8 @@
 
 #define EXIT_TO_USER_MODE_WORK						\
 	(_TIF_SIGPENDING | _TIF_NOTIFY_RESUME | _TIF_UPROBE |		\
-	 _TIF_NEED_RESCHED | _TIF_PATCH_PENDING | _TIF_NOTIFY_SIGNAL |	\
+	 _TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY |			\
+	 _TIF_PATCH_PENDING | _TIF_NOTIFY_SIGNAL |			\
 	 ARCH_EXIT_TO_USER_MODE_WORK)
 
 /**

include/linux/entry-kvm.h

@@ -17,8 +17,9 @@
 #endif
 
 #define XFER_TO_GUEST_MODE_WORK						\
-	(_TIF_NEED_RESCHED | _TIF_SIGPENDING | _TIF_NOTIFY_SIGNAL |	\
+	(_TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY | _TIF_SIGPENDING | \
-	 _TIF_NOTIFY_RESUME | ARCH_XFER_TO_GUEST_MODE_WORK)
+	 _TIF_NOTIFY_SIGNAL | _TIF_NOTIFY_RESUME |			\
+	 ARCH_XFER_TO_GUEST_MODE_WORK)
 
 struct kvm_vcpu;
 

include/linux/mm_types.h

@@ -782,6 +782,7 @@ struct vm_area_struct {
 struct mm_cid {
 	u64 time;
 	int cid;
+	int recent_cid;
 };
 #endif
 
@@ -852,6 +853,27 @@ struct mm_struct {
 		 * When the next mm_cid scan is due (in jiffies).
 		 */
 		unsigned long mm_cid_next_scan;
+		/**
+		 * @nr_cpus_allowed: Number of CPUs allowed for mm.
+		 *
+		 * Number of CPUs allowed in the union of all mm's
+		 * threads allowed CPUs.
+		 */
+		unsigned int nr_cpus_allowed;
+		/**
+		 * @max_nr_cid: Maximum number of concurrency IDs allocated.
+		 *
+		 * Track the highest number of concurrency IDs allocated for the
+		 * mm.
+		 */
+		atomic_t max_nr_cid;
+		/**
+		 * @cpus_allowed_lock: Lock protecting mm cpus_allowed.
+		 *
+		 * Provide mutual exclusion for mm cpus_allowed and
+		 * mm nr_cpus_allowed updates.
+		 */
+		raw_spinlock_t cpus_allowed_lock;
 #endif
 #ifdef CONFIG_MMU
 		atomic_long_t pgtables_bytes;	/* size of all page tables */
@@ -1170,18 +1192,30 @@ static inline int mm_cid_clear_lazy_put(int cid)
 	return cid & ~MM_CID_LAZY_PUT;
 }
 
+/*
+ * mm_cpus_allowed: Union of all mm's threads allowed CPUs.
+ */
+static inline cpumask_t *mm_cpus_allowed(struct mm_struct *mm)
+{
+	unsigned long bitmap = (unsigned long)mm;
+
+	bitmap += offsetof(struct mm_struct, cpu_bitmap);
+	/* Skip cpu_bitmap */
+	bitmap += cpumask_size();
+	return (struct cpumask *)bitmap;
+}
+
 /* Accessor for struct mm_struct's cidmask. */
 static inline cpumask_t *mm_cidmask(struct mm_struct *mm)
 {
-	unsigned long cid_bitmap = (unsigned long)mm;
+	unsigned long cid_bitmap = (unsigned long)mm_cpus_allowed(mm);
 
-	cid_bitmap += offsetof(struct mm_struct, cpu_bitmap);
+	/* Skip mm_cpus_allowed */
-	/* Skip cpu_bitmap */
 	cid_bitmap += cpumask_size();
 	return (struct cpumask *)cid_bitmap;
 }
 
-static inline void mm_init_cid(struct mm_struct *mm)
+static inline void mm_init_cid(struct mm_struct *mm, struct task_struct *p)
 {
 	int i;
 
@@ -1189,17 +1223,22 @@ static inline void mm_init_cid(struct mm_struct *mm)
 		struct mm_cid *pcpu_cid = per_cpu_ptr(mm->pcpu_cid, i);
 
 		pcpu_cid->cid = MM_CID_UNSET;
+		pcpu_cid->recent_cid = MM_CID_UNSET;
 		pcpu_cid->time = 0;
 	}
+	mm->nr_cpus_allowed = p->nr_cpus_allowed;
+	atomic_set(&mm->max_nr_cid, 0);
+	raw_spin_lock_init(&mm->cpus_allowed_lock);
+	cpumask_copy(mm_cpus_allowed(mm), &p->cpus_mask);
 	cpumask_clear(mm_cidmask(mm));
 }
 
-static inline int mm_alloc_cid_noprof(struct mm_struct *mm)
+static inline int mm_alloc_cid_noprof(struct mm_struct *mm, struct task_struct *p)
 {
 	mm->pcpu_cid = alloc_percpu_noprof(struct mm_cid);
 	if (!mm->pcpu_cid)
 		return -ENOMEM;
-	mm_init_cid(mm);
+	mm_init_cid(mm, p);
 	return 0;
 }
 #define mm_alloc_cid(...)	alloc_hooks(mm_alloc_cid_noprof(__VA_ARGS__))
@@ -1212,16 +1251,31 @@ static inline void mm_destroy_cid(struct mm_struct *mm)
 
 static inline unsigned int mm_cid_size(void)
 {
-	return cpumask_size();
+	return 2 * cpumask_size();	/* mm_cpus_allowed(), mm_cidmask(). */
+}
+
+static inline void mm_set_cpus_allowed(struct mm_struct *mm, const struct cpumask *cpumask)
+{
+	struct cpumask *mm_allowed = mm_cpus_allowed(mm);
+
+	if (!mm)
+		return;
+	/* The mm_cpus_allowed is the union of each thread allowed CPUs masks. */
+	raw_spin_lock(&mm->cpus_allowed_lock);
+	cpumask_or(mm_allowed, mm_allowed, cpumask);
+	WRITE_ONCE(mm->nr_cpus_allowed, cpumask_weight(mm_allowed));
+	raw_spin_unlock(&mm->cpus_allowed_lock);
 }
 #else /* CONFIG_SCHED_MM_CID */
-static inline void mm_init_cid(struct mm_struct *mm) { }
+static inline void mm_init_cid(struct mm_struct *mm, struct task_struct *p) { }
-static inline int mm_alloc_cid(struct mm_struct *mm) { return 0; }
+static inline int mm_alloc_cid(struct mm_struct *mm, struct task_struct *p) { return 0; }
 static inline void mm_destroy_cid(struct mm_struct *mm) { }
+
 static inline unsigned int mm_cid_size(void)
 {
 	return 0;
 }
+static inline void mm_set_cpus_allowed(struct mm_struct *mm, const struct cpumask *cpumask) { }
 #endif /* CONFIG_SCHED_MM_CID */
 
 struct mmu_gather;

include/linux/preempt.h

@@ -486,6 +486,7 @@ DEFINE_LOCK_GUARD_0(migrate, migrate_disable(), migrate_enable())
 extern bool preempt_model_none(void);
 extern bool preempt_model_voluntary(void);
 extern bool preempt_model_full(void);
+extern bool preempt_model_lazy(void);
 
 #else
 
@@ -502,6 +503,11 @@ static inline bool preempt_model_full(void)
 	return IS_ENABLED(CONFIG_PREEMPT);
 }
 
+static inline bool preempt_model_lazy(void)
+{
+	return IS_ENABLED(CONFIG_PREEMPT_LAZY);
+}
+
 #endif
 
 static inline bool preempt_model_rt(void)
@@ -519,7 +525,7 @@ static inline bool preempt_model_rt(void)
  */
 static inline bool preempt_model_preemptible(void)
 {
-	return preempt_model_full() || preempt_model_rt();
+	return preempt_model_full() || preempt_model_lazy() || preempt_model_rt();
 }
 
 #endif /* __LINUX_PREEMPT_H */

include/linux/sched.h

@@ -1898,7 +1898,7 @@ extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
 
 #ifdef CONFIG_THREAD_INFO_IN_TASK
 # define task_thread_info(task)	(&(task)->thread_info)
-#elif !defined(__HAVE_THREAD_FUNCTIONS)
+#else
 # define task_thread_info(task)	((struct thread_info *)(task)->stack)
 #endif
 
@@ -2002,7 +2002,8 @@ static inline void set_tsk_need_resched(struct task_struct *tsk)
 
 static inline void clear_tsk_need_resched(struct task_struct *tsk)
 {
-	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
+	atomic_long_andnot(_TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY,
+			   (atomic_long_t *)&task_thread_info(tsk)->flags);
 }
 
 static inline int test_tsk_need_resched(struct task_struct *tsk)

include/linux/sched/ext.h

@@ -199,7 +199,6 @@ struct sched_ext_entity {
 #ifdef CONFIG_EXT_GROUP_SCHED
 	struct cgroup		*cgrp_moving_from;
 #endif
-	/* must be the last field, see init_scx_entity() */
 	struct list_head	tasks_node;
 };
 

include/linux/sched/task_stack.h

@@ -34,7 +34,7 @@ static __always_inline unsigned long *end_of_stack(const struct task_struct *tas
 #endif
 }
 
-#elif !defined(__HAVE_THREAD_FUNCTIONS)
+#else
 
 #define task_stack_page(task)	((void *)(task)->stack)
 

include/linux/thread_info.h

@@ -59,6 +59,14 @@ enum syscall_work_bit {
 
 #include <asm/thread_info.h>
 
+#ifndef TIF_NEED_RESCHED_LAZY
+#ifdef CONFIG_ARCH_HAS_PREEMPT_LAZY
+#error Inconsistent PREEMPT_LAZY
+#endif
+#define TIF_NEED_RESCHED_LAZY TIF_NEED_RESCHED
+#define _TIF_NEED_RESCHED_LAZY _TIF_NEED_RESCHED
+#endif
+
 #ifdef __KERNEL__
 
 #ifndef arch_set_restart_data
@@ -179,22 +187,27 @@ static __always_inline unsigned long read_ti_thread_flags(struct thread_info *ti
 
 #ifdef _ASM_GENERIC_BITOPS_INSTRUMENTED_NON_ATOMIC_H
 
-static __always_inline bool tif_need_resched(void)
+static __always_inline bool tif_test_bit(int bit)
 {
-	return arch_test_bit(TIF_NEED_RESCHED,
+	return arch_test_bit(bit,
 			     (unsigned long *)(&current_thread_info()->flags));
 }
 
 #else
 
-static __always_inline bool tif_need_resched(void)
+static __always_inline bool tif_test_bit(int bit)
 {
-	return test_bit(TIF_NEED_RESCHED,
+	return test_bit(bit,
 			(unsigned long *)(&current_thread_info()->flags));
 }
 
 #endif /* _ASM_GENERIC_BITOPS_INSTRUMENTED_NON_ATOMIC_H */
 
+static __always_inline bool tif_need_resched(void)
+{
+	return tif_test_bit(TIF_NEED_RESCHED);
+}
+
 #ifndef CONFIG_HAVE_ARCH_WITHIN_STACK_FRAMES
 static inline int arch_within_stack_frames(const void * const stack,
 					   const void * const stackend,

include/linux/wait_bit.h

@@ -8,7 +8,7 @@
 #include <linux/wait.h>
 
 struct wait_bit_key {
-	void			*flags;
+	unsigned long		*flags;
 	int			bit_nr;
 	unsigned long		timeout;
 };
@@ -23,14 +23,14 @@ struct wait_bit_queue_entry {
 
 typedef int wait_bit_action_f(struct wait_bit_key *key, int mode);
 
-void __wake_up_bit(struct wait_queue_head *wq_head, void *word, int bit);
+void __wake_up_bit(struct wait_queue_head *wq_head, unsigned long *word, int bit);
 int __wait_on_bit(struct wait_queue_head *wq_head, struct wait_bit_queue_entry *wbq_entry, wait_bit_action_f *action, unsigned int mode);
 int __wait_on_bit_lock(struct wait_queue_head *wq_head, struct wait_bit_queue_entry *wbq_entry, wait_bit_action_f *action, unsigned int mode);
-void wake_up_bit(void *word, int bit);
+void wake_up_bit(unsigned long *word, int bit);
-int out_of_line_wait_on_bit(void *word, int, wait_bit_action_f *action, unsigned int mode);
+int out_of_line_wait_on_bit(unsigned long *word, int, wait_bit_action_f *action, unsigned int mode);
-int out_of_line_wait_on_bit_timeout(void *word, int, wait_bit_action_f *action, unsigned int mode, unsigned long timeout);
+int out_of_line_wait_on_bit_timeout(unsigned long *word, int, wait_bit_action_f *action, unsigned int mode, unsigned long timeout);
-int out_of_line_wait_on_bit_lock(void *word, int, wait_bit_action_f *action, unsigned int mode);
+int out_of_line_wait_on_bit_lock(unsigned long *word, int, wait_bit_action_f *action, unsigned int mode);
-struct wait_queue_head *bit_waitqueue(void *word, int bit);
+struct wait_queue_head *bit_waitqueue(unsigned long *word, int bit);
 extern void __init wait_bit_init(void);
 
 int wake_bit_function(struct wait_queue_entry *wq_entry, unsigned mode, int sync, void *key);
@@ -49,23 +49,24 @@ int wake_bit_function(struct wait_queue_entry *wq_entry, unsigned mode, int sync
 extern int bit_wait(struct wait_bit_key *key, int mode);
 extern int bit_wait_io(struct wait_bit_key *key, int mode);
 extern int bit_wait_timeout(struct wait_bit_key *key, int mode);
-extern int bit_wait_io_timeout(struct wait_bit_key *key, int mode);
 
 /**
  * wait_on_bit - wait for a bit to be cleared
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit at that address being waited on
  * @mode: the task state to sleep in
  *
- * There is a standard hashed waitqueue table for generic use. This
+ * Wait for the given bit in an unsigned long or bitmap (see DECLARE_BITMAP())
- * is the part of the hashtable's accessor API that waits on a bit.
+ * to be cleared.  The clearing of the bit must be signalled with
- * For instance, if one were to have waiters on a bitflag, one would
+ * wake_up_bit(), often as clear_and_wake_up_bit().
- * call wait_on_bit() in threads waiting for the bit to clear.
+ *
- * One uses wait_on_bit() where one is waiting for the bit to clear,
+ * The process will wait on a waitqueue selected by hash from a shared
- * but has no intention of setting it.
+ * pool.  It will only be woken on a wake_up for the target bit, even
- * Returned value will be zero if the bit was cleared, or non-zero
+ * if other processes on the same queue are waiting for other bits.
- * if the process received a signal and the mode permitted wakeup
+ *
- * on that signal.
+ * Returned value will be zero if the bit was cleared in which case the
+ * call has ACQUIRE semantics, or %-EINTR if the process received a
+ * signal and the mode permitted wake up on that signal.
  */
 static inline int
 wait_on_bit(unsigned long *word, int bit, unsigned mode)
@@ -80,17 +81,20 @@ wait_on_bit(unsigned long *word, int bit, unsigned mode)
 
 /**
  * wait_on_bit_io - wait for a bit to be cleared
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit at that address being waited on
  * @mode: the task state to sleep in
  *
- * Use the standard hashed waitqueue table to wait for a bit
+ * Wait for the given bit in an unsigned long or bitmap (see DECLARE_BITMAP())
- * to be cleared.  This is similar to wait_on_bit(), but calls
+ * to be cleared.  The clearing of the bit must be signalled with
- * io_schedule() instead of schedule() for the actual waiting.
+ * wake_up_bit(), often as clear_and_wake_up_bit().
  *
- * Returned value will be zero if the bit was cleared, or non-zero
+ * This is similar to wait_on_bit(), but calls io_schedule() instead of
- * if the process received a signal and the mode permitted wakeup
+ * schedule() for the actual waiting.
- * on that signal.
+ *
+ * Returned value will be zero if the bit was cleared in which case the
+ * call has ACQUIRE semantics, or %-EINTR if the process received a
+ * signal and the mode permitted wake up on that signal.
  */
 static inline int
 wait_on_bit_io(unsigned long *word, int bit, unsigned mode)
@@ -104,19 +108,24 @@ wait_on_bit_io(unsigned long *word, int bit, unsigned mode)
 }
 
 /**
- * wait_on_bit_timeout - wait for a bit to be cleared or a timeout elapses
+ * wait_on_bit_timeout - wait for a bit to be cleared or a timeout to elapse
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit at that address being waited on
  * @mode: the task state to sleep in
  * @timeout: timeout, in jiffies
  *
- * Use the standard hashed waitqueue table to wait for a bit
+ * Wait for the given bit in an unsigned long or bitmap (see
- * to be cleared. This is similar to wait_on_bit(), except also takes a
+ * DECLARE_BITMAP()) to be cleared, or for a timeout to expire.  The
- * timeout parameter.
+ * clearing of the bit must be signalled with wake_up_bit(), often as
+ * clear_and_wake_up_bit().
  *
- * Returned value will be zero if the bit was cleared before the
+ * This is similar to wait_on_bit(), except it also takes a timeout
- * @timeout elapsed, or non-zero if the @timeout elapsed or process
+ * parameter.
- * received a signal and the mode permitted wakeup on that signal.
+ *
+ * Returned value will be zero if the bit was cleared in which case the
+ * call has ACQUIRE semantics, or %-EINTR if the process received a
+ * signal and the mode permitted wake up on that signal, or %-EAGAIN if the
+ * timeout elapsed.
  */
 static inline int
 wait_on_bit_timeout(unsigned long *word, int bit, unsigned mode,
@@ -132,19 +141,21 @@ wait_on_bit_timeout(unsigned long *word, int bit, unsigned mode,
 
 /**
  * wait_on_bit_action - wait for a bit to be cleared
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit at that address being waited on
  * @action: the function used to sleep, which may take special actions
  * @mode: the task state to sleep in
  *
- * Use the standard hashed waitqueue table to wait for a bit
+ * Wait for the given bit in an unsigned long or bitmap (see DECLARE_BITMAP())
- * to be cleared, and allow the waiting action to be specified.
+ * to be cleared.  The clearing of the bit must be signalled with
- * This is like wait_on_bit() but allows fine control of how the waiting
+ * wake_up_bit(), often as clear_and_wake_up_bit().
- * is done.
  *
- * Returned value will be zero if the bit was cleared, or non-zero
+ * This is similar to wait_on_bit(), but calls @action() instead of
- * if the process received a signal and the mode permitted wakeup
+ * schedule() for the actual waiting.
- * on that signal.
+ *
+ * Returned value will be zero if the bit was cleared in which case the
+ * call has ACQUIRE semantics, or the error code returned by @action if
+ * that call returned non-zero.
  */
 static inline int
 wait_on_bit_action(unsigned long *word, int bit, wait_bit_action_f *action,
@@ -157,23 +168,22 @@ wait_on_bit_action(unsigned long *word, int bit, wait_bit_action_f *action,
 }
 
 /**
- * wait_on_bit_lock - wait for a bit to be cleared, when wanting to set it
+ * wait_on_bit_lock - wait for a bit to be cleared, then set it
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit of the word being waited on and set
  * @mode: the task state to sleep in
  *
- * There is a standard hashed waitqueue table for generic use. This
+ * Wait for the given bit in an unsigned long or bitmap (see
- * is the part of the hashtable's accessor API that waits on a bit
+ * DECLARE_BITMAP()) to be cleared.  The clearing of the bit must be
- * when one intends to set it, for instance, trying to lock bitflags.
+ * signalled with wake_up_bit(), often as clear_and_wake_up_bit().  As
- * For instance, if one were to have waiters trying to set bitflag
+ * soon as it is clear, atomically set it and return.
- * and waiting for it to clear before setting it, one would call
- * wait_on_bit() in threads waiting to be able to set the bit.
- * One uses wait_on_bit_lock() where one is waiting for the bit to
- * clear with the intention of setting it, and when done, clearing it.
  *
- * Returns zero if the bit was (eventually) found to be clear and was
+ * This is similar to wait_on_bit(), but sets the bit before returning.
- * set.  Returns non-zero if a signal was delivered to the process and
+ *
- * the @mode allows that signal to wake the process.
+ * Returned value will be zero if the bit was successfully set in which
+ * case the call has the same memory sequencing semantics as
+ * test_and_clear_bit(), or %-EINTR if the process received a signal and
+ * the mode permitted wake up on that signal.
  */
 static inline int
 wait_on_bit_lock(unsigned long *word, int bit, unsigned mode)
@@ -185,15 +195,18 @@ wait_on_bit_lock(unsigned long *word, int bit, unsigned mode)
 }
 
 /**
- * wait_on_bit_lock_io - wait for a bit to be cleared, when wanting to set it
+ * wait_on_bit_lock_io - wait for a bit to be cleared, then set it
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit of the word being waited on and set
  * @mode: the task state to sleep in
  *
- * Use the standard hashed waitqueue table to wait for a bit
+ * Wait for the given bit in an unsigned long or bitmap (see
- * to be cleared and then to atomically set it.  This is similar
+ * DECLARE_BITMAP()) to be cleared.  The clearing of the bit must be
- * to wait_on_bit(), but calls io_schedule() instead of schedule()
+ * signalled with wake_up_bit(), often as clear_and_wake_up_bit().  As
- * for the actual waiting.
+ * soon as it is clear, atomically set it and return.
+ *
+ * This is similar to wait_on_bit_lock(), but calls io_schedule() instead
+ * of schedule().
  *
  * Returns zero if the bit was (eventually) found to be clear and was
  * set.  Returns non-zero if a signal was delivered to the process and
@@ -209,21 +222,19 @@ wait_on_bit_lock_io(unsigned long *word, int bit, unsigned mode)
 }
 
 /**
- * wait_on_bit_lock_action - wait for a bit to be cleared, when wanting to set it
+ * wait_on_bit_lock_action - wait for a bit to be cleared, then set it
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit of the word being waited on and set
  * @action: the function used to sleep, which may take special actions
  * @mode: the task state to sleep in
  *
- * Use the standard hashed waitqueue table to wait for a bit
+ * This is similar to wait_on_bit_lock(), but calls @action() instead of
- * to be cleared and then to set it, and allow the waiting action
+ * schedule() for the actual waiting.
- * to be specified.
- * This is like wait_on_bit() but allows fine control of how the waiting
- * is done.
  *
- * Returns zero if the bit was (eventually) found to be clear and was
+ * Returned value will be zero if the bit was successfully set in which
- * set.  Returns non-zero if a signal was delivered to the process and
+ * case the call has the same memory sequencing semantics as
- * the @mode allows that signal to wake the process.
+ * test_and_clear_bit(), or the error code returned by @action if that
+ * call returned non-zero.
  */
 static inline int
 wait_on_bit_lock_action(unsigned long *word, int bit, wait_bit_action_f *action,
@@ -269,7 +280,26 @@ __out:	__ret;								\
 #define __wait_var_event(var, condition)				\
 	___wait_var_event(var, condition, TASK_UNINTERRUPTIBLE, 0, 0,	\
 			  schedule())
+#define __wait_var_event_io(var, condition)				\
+	___wait_var_event(var, condition, TASK_UNINTERRUPTIBLE, 0, 0,	\
+			  io_schedule())
 
+/**
+ * wait_var_event - wait for a variable to be updated and notified
+ * @var: the address of variable being waited on
+ * @condition: the condition to wait for
+ *
+ * Wait for a @condition to be true, only re-checking when a wake up is
+ * received for the given @var (an arbitrary kernel address which need
+ * not be directly related to the given condition, but usually is).
+ *
+ * The process will wait on a waitqueue selected by hash from a shared
+ * pool.  It will only be woken on a wake_up for the given address.
+ *
+ * The condition should normally use smp_load_acquire() or a similarly
+ * ordered access to ensure that any changes to memory made before the
+ * condition became true will be visible after the wait completes.
+ */
 #define wait_var_event(var, condition)					\
 do {									\
 	might_sleep();							\
@@ -278,10 +308,56 @@ do {									\
 	__wait_var_event(var, condition);				\
 } while (0)
 
+/**
+ * wait_var_event_io - wait for a variable to be updated and notified
+ * @var: the address of variable being waited on
+ * @condition: the condition to wait for
+ *
+ * Wait for an IO related @condition to be true, only re-checking when a
+ * wake up is received for the given @var (an arbitrary kernel address
+ * which need not be directly related to the given condition, but
+ * usually is).
+ *
+ * The process will wait on a waitqueue selected by hash from a shared
+ * pool.  It will only be woken on a wake_up for the given address.
+ *
+ * This is similar to wait_var_event(), but calls io_schedule() instead
+ * of schedule().
+ *
+ * The condition should normally use smp_load_acquire() or a similarly
+ * ordered access to ensure that any changes to memory made before the
+ * condition became true will be visible after the wait completes.
+ */
+#define wait_var_event_io(var, condition)				\
+do {									\
+	might_sleep();							\
+	if (condition)							\
+		break;							\
+	__wait_var_event_io(var, condition);				\
+} while (0)
+
 #define __wait_var_event_killable(var, condition)			\
 	___wait_var_event(var, condition, TASK_KILLABLE, 0, 0,		\
 			  schedule())
 
+/**
+ * wait_var_event_killable - wait for a variable to be updated and notified
+ * @var: the address of variable being waited on
+ * @condition: the condition to wait for
+ *
+ * Wait for a @condition to be true or a fatal signal to be received,
+ * only re-checking the condition when a wake up is received for the given
+ * @var (an arbitrary kernel address which need not be directly related
+ * to the given condition, but usually is).
+ *
+ * This is similar to wait_var_event() but returns a value which is
+ * 0 if the condition became true, or %-ERESTARTSYS if a fatal signal
+ * was received.
+ *
+ * The condition should normally use smp_load_acquire() or a similarly
+ * ordered access to ensure that any changes to memory made before the
+ * condition became true will be visible after the wait completes.
+ */
 #define wait_var_event_killable(var, condition)				\
 ({									\
 	int __ret = 0;							\
@@ -296,6 +372,26 @@ do {									\
 			  TASK_UNINTERRUPTIBLE, 0, timeout,		\
 			  __ret = schedule_timeout(__ret))
 
+/**
+ * wait_var_event_timeout - wait for a variable to be updated or a timeout to expire
+ * @var: the address of variable being waited on
+ * @condition: the condition to wait for
+ * @timeout: maximum time to wait in jiffies
+ *
+ * Wait for a @condition to be true or a timeout to expire, only
+ * re-checking the condition when a wake up is received for the given
+ * @var (an arbitrary kernel address which need not be directly related
+ * to the given condition, but usually is).
+ *
+ * This is similar to wait_var_event() but returns a value which is 0 if
+ * the timeout expired and the condition was still false, or the
+ * remaining time left in the timeout (but at least 1) if the condition
+ * was found to be true.
+ *
+ * The condition should normally use smp_load_acquire() or a similarly
+ * ordered access to ensure that any changes to memory made before the
+ * condition became true will be visible after the wait completes.
+ */
 #define wait_var_event_timeout(var, condition, timeout)			\
 ({									\
 	long __ret = timeout;						\
@@ -309,6 +405,23 @@ do {									\
 	___wait_var_event(var, condition, TASK_INTERRUPTIBLE, 0, 0,	\
 			  schedule())
 
+/**
+ * wait_var_event_killable - wait for a variable to be updated and notified
+ * @var: the address of variable being waited on
+ * @condition: the condition to wait for
+ *
+ * Wait for a @condition to be true or a signal to be received, only
+ * re-checking the condition when a wake up is received for the given
+ * @var (an arbitrary kernel address which need not be directly related
+ * to the given condition, but usually is).
+ *
+ * This is similar to wait_var_event() but returns a value which is 0 if
+ * the condition became true, or %-ERESTARTSYS if a signal was received.
+ *
+ * The condition should normally use smp_load_acquire() or a similarly
+ * ordered access to ensure that any changes to memory made before the
+ * condition became true will be visible after the wait completes.
+ */
 #define wait_var_event_interruptible(var, condition)			\
 ({									\
 	int __ret = 0;							\
@@ -319,15 +432,122 @@ do {									\
 })
 
 /**
- * clear_and_wake_up_bit - clear a bit and wake up anyone waiting on that bit
+ * wait_var_event_any_lock - wait for a variable to be updated under a lock
+ * @var: the address of the variable being waited on
+ * @condition: condition to wait for
+ * @lock: the object that is locked to protect updates to the variable
+ * @type: prefix on lock and unlock operations
+ * @state: waiting state, %TASK_UNINTERRUPTIBLE etc.
  *
- * @bit: the bit of the word being waited on
+ * Wait for a condition which can only be reliably tested while holding
- * @word: the word being waited on, a kernel virtual address
+ * a lock.  The variables assessed in the condition will normal be updated
+ * under the same lock, and the wake up should be signalled with
+ * wake_up_var_locked() under the same lock.
  *
- * You can use this helper if bitflags are manipulated atomically rather than
+ * This is similar to wait_var_event(), but assumes a lock is held
- * non-atomically under a lock.
+ * while calling this function and while updating the variable.
+ *
+ * This must be called while the given lock is held and the lock will be
+ * dropped when schedule() is called to wait for a wake up, and will be
+ * reclaimed before testing the condition again.  The functions used to
+ * unlock and lock the object are constructed by appending _unlock and _lock
+ * to @type.
+ *
+ * Return %-ERESTARTSYS if a signal arrives which is allowed to interrupt
+ * the wait according to @state.
  */
-static inline void clear_and_wake_up_bit(int bit, void *word)
+#define wait_var_event_any_lock(var, condition, lock, type, state)	\
+({									\
+	int __ret = 0;							\
+	if (!(condition))						\
+		__ret = ___wait_var_event(var, condition, state, 0, 0,	\
+					  type ## _unlock(lock);	\
+					  schedule();			\
+					  type ## _lock(lock));		\
+	__ret;								\
+})
+
+/**
+ * wait_var_event_spinlock - wait for a variable to be updated under a spinlock
+ * @var: the address of the variable being waited on
+ * @condition: condition to wait for
+ * @lock: the spinlock which protects updates to the variable
+ *
+ * Wait for a condition which can only be reliably tested while holding
+ * a spinlock.  The variables assessed in the condition will normal be updated
+ * under the same spinlock, and the wake up should be signalled with
+ * wake_up_var_locked() under the same spinlock.
+ *
+ * This is similar to wait_var_event(), but assumes a spinlock is held
+ * while calling this function and while updating the variable.
+ *
+ * This must be called while the given lock is held and the lock will be
+ * dropped when schedule() is called to wait for a wake up, and will be
+ * reclaimed before testing the condition again.
+ */
+#define wait_var_event_spinlock(var, condition, lock)			\
+	wait_var_event_any_lock(var, condition, lock, spin, TASK_UNINTERRUPTIBLE)
+
+/**
+ * wait_var_event_mutex - wait for a variable to be updated under a mutex
+ * @var: the address of the variable being waited on
+ * @condition: condition to wait for
+ * @mutex: the mutex which protects updates to the variable
+ *
+ * Wait for a condition which can only be reliably tested while holding
+ * a mutex.  The variables assessed in the condition will normal be
+ * updated under the same mutex, and the wake up should be signalled
+ * with wake_up_var_locked() under the same mutex.
+ *
+ * This is similar to wait_var_event(), but assumes a mutex is held
+ * while calling this function and while updating the variable.
+ *
+ * This must be called while the given mutex is held and the mutex will be
+ * dropped when schedule() is called to wait for a wake up, and will be
+ * reclaimed before testing the condition again.
+ */
+#define wait_var_event_mutex(var, condition, lock)			\
+	wait_var_event_any_lock(var, condition, lock, mutex, TASK_UNINTERRUPTIBLE)
+
+/**
+ * wake_up_var_protected - wake up waiters for a variable asserting that it is safe
+ * @var: the address of the variable being waited on
+ * @cond: the condition which afirms this is safe
+ *
+ * When waking waiters which use wait_var_event_any_lock() the waker must be
+ * holding the reelvant lock to avoid races.  This version of wake_up_var()
+ * asserts that the relevant lock is held and so no barrier is needed.
+ * The @cond is only tested when CONFIG_LOCKDEP is enabled.
+ */
+#define wake_up_var_protected(var, cond)				\
+do {									\
+	lockdep_assert(cond);						\
+	wake_up_var(var);						\
+} while (0)
+
+/**
+ * wake_up_var_locked - wake up waiters for a variable while holding a spinlock or mutex
+ * @var: the address of the variable being waited on
+ * @lock: The spinlock or mutex what protects the variable
+ *
+ * Send a wake up for the given variable which should be waited for with
+ * wait_var_event_spinlock() or wait_var_event_mutex().  Unlike wake_up_var(),
+ * no extra barriers are needed as the locking provides sufficient sequencing.
+ */
+#define wake_up_var_locked(var, lock)					\
+	wake_up_var_protected(var, lockdep_is_held(lock))
+
+/**
+ * clear_and_wake_up_bit - clear a bit and wake up anyone waiting on that bit
+ * @bit: the bit of the word being waited on
+ * @word: the address containing the bit being waited on
+ *
+ * The designated bit is cleared and any tasks waiting in wait_on_bit()
+ * or similar will be woken.  This call has RELEASE semantics so that
+ * any changes to memory made before this call are guaranteed to be visible
+ * after the corresponding wait_on_bit() completes.
+ */
+static inline void clear_and_wake_up_bit(int bit, unsigned long *word)
 {
 	clear_bit_unlock(bit, word);
 	/* See wake_up_bit() for which memory barrier you need to use. */
@@ -335,4 +555,64 @@ static inline void clear_and_wake_up_bit(int bit, void *word)
 	wake_up_bit(word, bit);
 }
 
+/**
+ * test_and_clear_wake_up_bit - clear a bit if it was set: wake up anyone waiting on that bit
+ * @bit: the bit of the word being waited on
+ * @word: the address of memory containing that bit
+ *
+ * If the bit is set and can be atomically cleared, any tasks waiting in
+ * wait_on_bit() or similar will be woken.  This call has the same
+ * complete ordering semantics as test_and_clear_bit().  Any changes to
+ * memory made before this call are guaranteed to be visible after the
+ * corresponding wait_on_bit() completes.
+ *
+ * Returns %true if the bit was successfully set and the wake up was sent.
+ */
+static inline bool test_and_clear_wake_up_bit(int bit, unsigned long *word)
+{
+	if (!test_and_clear_bit(bit, word))
+		return false;
+	/* no extra barrier required */
+	wake_up_bit(word, bit);
+	return true;
+}
+
+/**
+ * atomic_dec_and_wake_up - decrement an atomic_t and if zero, wake up waiters
+ * @var: the variable to dec and test
+ *
+ * Decrements the atomic variable and if it reaches zero, send a wake_up to any
+ * processes waiting on the variable.
+ *
+ * This function has the same complete ordering semantics as atomic_dec_and_test.
+ *
+ * Returns %true is the variable reaches zero and the wake up was sent.
+ */
+
+static inline bool atomic_dec_and_wake_up(atomic_t *var)
+{
+	if (!atomic_dec_and_test(var))
+		return false;
+	/* No extra barrier required */
+	wake_up_var(var);
+	return true;
+}
+
+/**
+ * store_release_wake_up - update a variable and send a wake_up
+ * @var: the address of the variable to be updated and woken
+ * @val: the value to store in the variable.
+ *
+ * Store the given value in the variable send a wake up to any tasks
+ * waiting on the variable.  All necessary barriers are included to ensure
+ * the task calling wait_var_event() sees the new value and all values
+ * written to memory before this call.
+ */
+#define store_release_wake_up(var, val)					\
+do {									\
+	smp_store_release(var, val);					\
+	smp_mb();							\
+	wake_up_var(var);						\
+} while (0)
+
 #endif /* _LINUX_WAIT_BIT_H */

kernel/Kconfig.preempt

@@ -11,12 +11,16 @@ config PREEMPT_BUILD
 	select PREEMPTION
 	select UNINLINE_SPIN_UNLOCK if !ARCH_INLINE_SPIN_UNLOCK
 
+config ARCH_HAS_PREEMPT_LAZY
+	bool
+
 choice
 	prompt "Preemption Model"
 	default PREEMPT_NONE
 
 config PREEMPT_NONE
 	bool "No Forced Preemption (Server)"
+	depends on !PREEMPT_RT
 	select PREEMPT_NONE_BUILD if !PREEMPT_DYNAMIC
 	help
 	  This is the traditional Linux preemption model, geared towards
@@ -32,6 +36,7 @@ config PREEMPT_NONE
 config PREEMPT_VOLUNTARY
 	bool "Voluntary Kernel Preemption (Desktop)"
 	depends on !ARCH_NO_PREEMPT
+	depends on !PREEMPT_RT
 	select PREEMPT_VOLUNTARY_BUILD if !PREEMPT_DYNAMIC
 	help
 	  This option reduces the latency of the kernel by adding more
@@ -51,7 +56,7 @@ config PREEMPT_VOLUNTARY
 config PREEMPT
 	bool "Preemptible Kernel (Low-Latency Desktop)"
 	depends on !ARCH_NO_PREEMPT
-	select PREEMPT_BUILD
+	select PREEMPT_BUILD if !PREEMPT_DYNAMIC
 	help
 	  This option reduces the latency of the kernel by making
 	  all kernel code (that is not executing in a critical section)
@@ -67,9 +72,23 @@ config PREEMPT
 	  embedded system with latency requirements in the milliseconds
 	  range.
 
+config PREEMPT_LAZY
+	bool "Scheduler controlled preemption model"
+	depends on !ARCH_NO_PREEMPT
+	depends on ARCH_HAS_PREEMPT_LAZY
+	select PREEMPT_BUILD if !PREEMPT_DYNAMIC
+	help
+	  This option provides a scheduler driven preemption model that
+	  is fundamentally similar to full preemption, but is less
+	  eager to preempt SCHED_NORMAL tasks in an attempt to
+	  reduce lock holder preemption and recover some of the performance
+	  gains seen from using Voluntary preemption.
+
+endchoice
+
 config PREEMPT_RT
 	bool "Fully Preemptible Kernel (Real-Time)"
-	depends on EXPERT && ARCH_SUPPORTS_RT
+	depends on EXPERT && ARCH_SUPPORTS_RT && !COMPILE_TEST
 	select PREEMPTION
 	help
 	  This option turns the kernel into a real-time kernel by replacing
@@ -84,8 +103,6 @@ config PREEMPT_RT
 	  Select this if you are building a kernel for systems which
 	  require real-time guarantees.
 
-endchoice
-
 config PREEMPT_COUNT
        bool
 
@@ -95,7 +112,7 @@ config PREEMPTION
 
 config PREEMPT_DYNAMIC
 	bool "Preemption behaviour defined on boot"
-	depends on HAVE_PREEMPT_DYNAMIC && !PREEMPT_RT
+	depends on HAVE_PREEMPT_DYNAMIC
 	select JUMP_LABEL if HAVE_PREEMPT_DYNAMIC_KEY
 	select PREEMPT_BUILD
 	default y if HAVE_PREEMPT_DYNAMIC_CALL

kernel/entry/common.c

@@ -98,7 +98,7 @@ __always_inline unsigned long exit_to_user_mode_loop(struct pt_regs *regs,
 
 		local_irq_enable_exit_to_user(ti_work);
 
-		if (ti_work & _TIF_NEED_RESCHED)
+		if (ti_work & (_TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY))
 			schedule();
 
 		if (ti_work & _TIF_UPROBE)

kernel/entry/kvm.c

@@ -13,7 +13,7 @@ static int xfer_to_guest_mode_work(struct kvm_vcpu *vcpu, unsigned long ti_work)
 			return -EINTR;
 		}
 
-		if (ti_work & _TIF_NEED_RESCHED)
+		if (ti_work & (_TIF_NEED_RESCHED | _TIF_NEED_RESCHED_LAZY))
 			schedule();
 
 		if (ti_work & _TIF_NOTIFY_RESUME)
@@ -24,7 +24,7 @@ static int xfer_to_guest_mode_work(struct kvm_vcpu *vcpu, unsigned long ti_work)
 			return ret;
 
 		ti_work = read_thread_flags();
-	} while (ti_work & XFER_TO_GUEST_MODE_WORK || need_resched());
+	} while (ti_work & XFER_TO_GUEST_MODE_WORK);
 	return 0;
 }
 

kernel/fork.c

@@ -1299,7 +1299,7 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p,
 	if (init_new_context(p, mm))
 		goto fail_nocontext;
 
-	if (mm_alloc_cid(mm))
+	if (mm_alloc_cid(mm, p))
 		goto fail_cid;
 
 	if (percpu_counter_init_many(mm->rss_stat, 0, GFP_KERNEL_ACCOUNT,

kernel/futex/pi.c

@@ -922,6 +922,7 @@ int futex_lock_pi(u32 __user *uaddr, unsigned int flags, ktime_t *time, int tryl
 	struct rt_mutex_waiter rt_waiter;
 	struct futex_hash_bucket *hb;
 	struct futex_q q = futex_q_init;
+	DEFINE_WAKE_Q(wake_q);
 	int res, ret;
 
 	if (!IS_ENABLED(CONFIG_FUTEX_PI))
@@ -1018,8 +1019,11 @@ retry_private:
 	 * such that futex_unlock_pi() is guaranteed to observe the waiter when
 	 * it sees the futex_q::pi_state.
 	 */
-	ret = __rt_mutex_start_proxy_lock(&q.pi_state->pi_mutex, &rt_waiter, current);
+	ret = __rt_mutex_start_proxy_lock(&q.pi_state->pi_mutex, &rt_waiter, current, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irq(&q.pi_state->pi_mutex.wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
 
 	if (ret) {
 		if (ret == 1)

kernel/locking/mutex.c

@@ -56,31 +56,6 @@ __mutex_init(struct mutex *lock, const char *name, struct lock_class_key *key)
 }
 EXPORT_SYMBOL(__mutex_init);
 
-/*
- * @owner: contains: 'struct task_struct *' to the current lock owner,
- * NULL means not owned. Since task_struct pointers are aligned at
- * at least L1_CACHE_BYTES, we have low bits to store extra state.
- *
- * Bit0 indicates a non-empty waiter list; unlock must issue a wakeup.
- * Bit1 indicates unlock needs to hand the lock to the top-waiter
- * Bit2 indicates handoff has been done and we're waiting for pickup.
- */
-#define MUTEX_FLAG_WAITERS	0x01
-#define MUTEX_FLAG_HANDOFF	0x02
-#define MUTEX_FLAG_PICKUP	0x04
-
-#define MUTEX_FLAGS		0x07
-
-/*
- * Internal helper function; C doesn't allow us to hide it :/
- *
- * DO NOT USE (outside of mutex code).
- */
-static inline struct task_struct *__mutex_owner(struct mutex *lock)
-{
-	return (struct task_struct *)(atomic_long_read(&lock->owner) & ~MUTEX_FLAGS);
-}
-
 static inline struct task_struct *__owner_task(unsigned long owner)
 {
 	return (struct task_struct *)(owner & ~MUTEX_FLAGS);
@@ -575,8 +550,10 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		    struct lockdep_map *nest_lock, unsigned long ip,
 		    struct ww_acquire_ctx *ww_ctx, const bool use_ww_ctx)
 {
+	DEFINE_WAKE_Q(wake_q);
 	struct mutex_waiter waiter;
 	struct ww_mutex *ww;
+	unsigned long flags;
 	int ret;
 
 	if (!use_ww_ctx)
@@ -619,13 +596,13 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		return 0;
 	}
 
-	raw_spin_lock(&lock->wait_lock);
+	raw_spin_lock_irqsave(&lock->wait_lock, flags);
 	/*
 	 * After waiting to acquire the wait_lock, try again.
 	 */
 	if (__mutex_trylock(lock)) {
 		if (ww_ctx)
-			__ww_mutex_check_waiters(lock, ww_ctx);
+			__ww_mutex_check_waiters(lock, ww_ctx, &wake_q);
 
 		goto skip_wait;
 	}
@@ -645,7 +622,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		 * Add in stamp order, waking up waiters that must kill
 		 * themselves.
 		 */
-		ret = __ww_mutex_add_waiter(&waiter, lock, ww_ctx);
+		ret = __ww_mutex_add_waiter(&waiter, lock, ww_ctx, &wake_q);
 		if (ret)
 			goto err_early_kill;
 	}
@@ -680,7 +657,11 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 				goto err;
 		}
 
-		raw_spin_unlock(&lock->wait_lock);
+		raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+		/* Make sure we do wakeups before calling schedule */
+		wake_up_q(&wake_q);
+		wake_q_init(&wake_q);
+
 		schedule_preempt_disabled();
 
 		first = __mutex_waiter_is_first(lock, &waiter);
@@ -701,9 +682,9 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 			trace_contention_begin(lock, LCB_F_MUTEX);
 		}
 
-		raw_spin_lock(&lock->wait_lock);
+		raw_spin_lock_irqsave(&lock->wait_lock, flags);
 	}
-	raw_spin_lock(&lock->wait_lock);
+	raw_spin_lock_irqsave(&lock->wait_lock, flags);
 acquired:
 	__set_current_state(TASK_RUNNING);
 
@@ -714,7 +695,7 @@ acquired:
 		 */
 		if (!ww_ctx->is_wait_die &&
 		    !__mutex_waiter_is_first(lock, &waiter))
-			__ww_mutex_check_waiters(lock, ww_ctx);
+			__ww_mutex_check_waiters(lock, ww_ctx, &wake_q);
 	}
 
 	__mutex_remove_waiter(lock, &waiter);
@@ -729,7 +710,8 @@ skip_wait:
 	if (ww_ctx)
 		ww_mutex_lock_acquired(ww, ww_ctx);
 
-	raw_spin_unlock(&lock->wait_lock);
+	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+	wake_up_q(&wake_q);
 	preempt_enable();
 	return 0;
 
@@ -738,9 +720,10 @@ err:
 	__mutex_remove_waiter(lock, &waiter);
 err_early_kill:
 	trace_contention_end(lock, ret);
-	raw_spin_unlock(&lock->wait_lock);
+	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
 	debug_mutex_free_waiter(&waiter);
 	mutex_release(&lock->dep_map, ip);
+	wake_up_q(&wake_q);
 	preempt_enable();
 	return ret;
 }
@@ -908,6 +891,7 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
 	struct task_struct *next = NULL;
 	DEFINE_WAKE_Q(wake_q);
 	unsigned long owner;
+	unsigned long flags;
 
 	mutex_release(&lock->dep_map, ip);
 
@@ -934,7 +918,7 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
 		}
 	}
 
-	raw_spin_lock(&lock->wait_lock);
+	raw_spin_lock_irqsave(&lock->wait_lock, flags);
 	debug_mutex_unlock(lock);
 	if (!list_empty(&lock->wait_list)) {
 		/* get the first entry from the wait-list: */
@@ -951,9 +935,10 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
 	if (owner & MUTEX_FLAG_HANDOFF)
 		__mutex_handoff(lock, next);
 
-	raw_spin_unlock(&lock->wait_lock);
+	preempt_disable();
-
+	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
 	wake_up_q(&wake_q);
+	preempt_enable();
 }
 
 #ifndef CONFIG_DEBUG_LOCK_ALLOC

kernel/locking/mutex.h

@@ -20,6 +20,33 @@ struct mutex_waiter {
 #endif
 };
 
+/*
+ * @owner: contains: 'struct task_struct *' to the current lock owner,
+ * NULL means not owned. Since task_struct pointers are aligned at
+ * at least L1_CACHE_BYTES, we have low bits to store extra state.
+ *
+ * Bit0 indicates a non-empty waiter list; unlock must issue a wakeup.
+ * Bit1 indicates unlock needs to hand the lock to the top-waiter
+ * Bit2 indicates handoff has been done and we're waiting for pickup.
+ */
+#define MUTEX_FLAG_WAITERS	0x01
+#define MUTEX_FLAG_HANDOFF	0x02
+#define MUTEX_FLAG_PICKUP	0x04
+
+#define MUTEX_FLAGS		0x07
+
+/*
+ * Internal helper function; C doesn't allow us to hide it :/
+ *
+ * DO NOT USE (outside of mutex & scheduler code).
+ */
+static inline struct task_struct *__mutex_owner(struct mutex *lock)
+{
+	if (!lock)
+		return NULL;
+	return (struct task_struct *)(atomic_long_read(&lock->owner) & ~MUTEX_FLAGS);
+}
+
 #ifdef CONFIG_DEBUG_MUTEXES
 extern void debug_mutex_lock_common(struct mutex *lock,
 				    struct mutex_waiter *waiter);

kernel/locking/rtmutex.c

@@ -34,13 +34,15 @@
 
 static inline int __ww_mutex_add_waiter(struct rt_mutex_waiter *waiter,
 					struct rt_mutex *lock,
-					struct ww_acquire_ctx *ww_ctx)
+					struct ww_acquire_ctx *ww_ctx,
+					struct wake_q_head *wake_q)
 {
 	return 0;
 }
 
 static inline void __ww_mutex_check_waiters(struct rt_mutex *lock,
-					    struct ww_acquire_ctx *ww_ctx)
+					    struct ww_acquire_ctx *ww_ctx,
+					    struct wake_q_head *wake_q)
 {
 }
 
@@ -1201,7 +1203,8 @@ static int __sched task_blocks_on_rt_mutex(struct rt_mutex_base *lock,
 					   struct rt_mutex_waiter *waiter,
 					   struct task_struct *task,
 					   struct ww_acquire_ctx *ww_ctx,
-					   enum rtmutex_chainwalk chwalk)
+					   enum rtmutex_chainwalk chwalk,
+					   struct wake_q_head *wake_q)
 {
 	struct task_struct *owner = rt_mutex_owner(lock);
 	struct rt_mutex_waiter *top_waiter = waiter;
@@ -1245,7 +1248,10 @@ static int __sched task_blocks_on_rt_mutex(struct rt_mutex_base *lock,
 
 		/* Check whether the waiter should back out immediately */
 		rtm = container_of(lock, struct rt_mutex, rtmutex);
-		res = __ww_mutex_add_waiter(waiter, rtm, ww_ctx);
+		preempt_disable();
+		res = __ww_mutex_add_waiter(waiter, rtm, ww_ctx, wake_q);
+		wake_up_q(wake_q);
+		preempt_enable();
 		if (res) {
 			raw_spin_lock(&task->pi_lock);
 			rt_mutex_dequeue(lock, waiter);
@@ -1675,12 +1681,14 @@ static void __sched rt_mutex_handle_deadlock(int res, int detect_deadlock,
  * @state:	The task state for sleeping
  * @chwalk:	Indicator whether full or partial chainwalk is requested
  * @waiter:	Initializer waiter for blocking
+ * @wake_q:	The wake_q to wake tasks after we release the wait_lock
  */
 static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 				       struct ww_acquire_ctx *ww_ctx,
 				       unsigned int state,
 				       enum rtmutex_chainwalk chwalk,
-				       struct rt_mutex_waiter *waiter)
+				       struct rt_mutex_waiter *waiter,
+				       struct wake_q_head *wake_q)
 {
 	struct rt_mutex *rtm = container_of(lock, struct rt_mutex, rtmutex);
 	struct ww_mutex *ww = ww_container_of(rtm);
@@ -1691,7 +1699,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 	/* Try to acquire the lock again: */
 	if (try_to_take_rt_mutex(lock, current, NULL)) {
 		if (build_ww_mutex() && ww_ctx) {
-			__ww_mutex_check_waiters(rtm, ww_ctx);
+			__ww_mutex_check_waiters(rtm, ww_ctx, wake_q);
 			ww_mutex_lock_acquired(ww, ww_ctx);
 		}
 		return 0;
@@ -1701,7 +1709,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 
 	trace_contention_begin(lock, LCB_F_RT);
 
-	ret = task_blocks_on_rt_mutex(lock, waiter, current, ww_ctx, chwalk);
+	ret = task_blocks_on_rt_mutex(lock, waiter, current, ww_ctx, chwalk, wake_q);
 	if (likely(!ret))
 		ret = rt_mutex_slowlock_block(lock, ww_ctx, state, NULL, waiter);
 
@@ -1709,7 +1717,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 		/* acquired the lock */
 		if (build_ww_mutex() && ww_ctx) {
 			if (!ww_ctx->is_wait_die)
-				__ww_mutex_check_waiters(rtm, ww_ctx);
+				__ww_mutex_check_waiters(rtm, ww_ctx, wake_q);
 			ww_mutex_lock_acquired(ww, ww_ctx);
 		}
 	} else {
@@ -1731,7 +1739,8 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 
 static inline int __rt_mutex_slowlock_locked(struct rt_mutex_base *lock,
 					     struct ww_acquire_ctx *ww_ctx,
-					     unsigned int state)
+					     unsigned int state,
+					     struct wake_q_head *wake_q)
 {
 	struct rt_mutex_waiter waiter;
 	int ret;
@@ -1740,7 +1749,7 @@ static inline int __rt_mutex_slowlock_locked(struct rt_mutex_base *lock,
 	waiter.ww_ctx = ww_ctx;
 
 	ret = __rt_mutex_slowlock(lock, ww_ctx, state, RT_MUTEX_MIN_CHAINWALK,
-				  &waiter);
+				  &waiter, wake_q);
 
 	debug_rt_mutex_free_waiter(&waiter);
 	return ret;
@@ -1756,6 +1765,7 @@ static int __sched rt_mutex_slowlock(struct rt_mutex_base *lock,
 				     struct ww_acquire_ctx *ww_ctx,
 				     unsigned int state)
 {
+	DEFINE_WAKE_Q(wake_q);
 	unsigned long flags;
 	int ret;
 
@@ -1777,8 +1787,11 @@ static int __sched rt_mutex_slowlock(struct rt_mutex_base *lock,
 	 * irqsave/restore variants.
 	 */
 	raw_spin_lock_irqsave(&lock->wait_lock, flags);
-	ret = __rt_mutex_slowlock_locked(lock, ww_ctx, state);
+	ret = __rt_mutex_slowlock_locked(lock, ww_ctx, state, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+	wake_up_q(&wake_q);
+	preempt_enable();
 	rt_mutex_post_schedule();
 
 	return ret;
@@ -1804,8 +1817,10 @@ static __always_inline int __rt_mutex_lock(struct rt_mutex_base *lock,
 /**
  * rtlock_slowlock_locked - Slow path lock acquisition for RT locks
  * @lock:	The underlying RT mutex
+ * @wake_q:	The wake_q to wake tasks after we release the wait_lock
  */
-static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
+static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock,
+					   struct wake_q_head *wake_q)
 	__releases(&lock->wait_lock) __acquires(&lock->wait_lock)
 {
 	struct rt_mutex_waiter waiter;
@@ -1823,7 +1838,7 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 
 	trace_contention_begin(lock, LCB_F_RT);
 
-	task_blocks_on_rt_mutex(lock, &waiter, current, NULL, RT_MUTEX_MIN_CHAINWALK);
+	task_blocks_on_rt_mutex(lock, &waiter, current, NULL, RT_MUTEX_MIN_CHAINWALK, wake_q);
 
 	for (;;) {
 		/* Try to acquire the lock again */
@@ -1834,7 +1849,11 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 			owner = rt_mutex_owner(lock);
 		else
 			owner = NULL;
+		preempt_disable();
 		raw_spin_unlock_irq(&lock->wait_lock);
+		wake_up_q(wake_q);
+		wake_q_init(wake_q);
+		preempt_enable();
 
 		if (!owner || !rtmutex_spin_on_owner(lock, &waiter, owner))
 			schedule_rtlock();
@@ -1859,10 +1878,14 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 static __always_inline void __sched rtlock_slowlock(struct rt_mutex_base *lock)
 {
 	unsigned long flags;
+	DEFINE_WAKE_Q(wake_q);
 
 	raw_spin_lock_irqsave(&lock->wait_lock, flags);
-	rtlock_slowlock_locked(lock);
+	rtlock_slowlock_locked(lock, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+	wake_up_q(&wake_q);
+	preempt_enable();
 }
 
 #endif /* RT_MUTEX_BUILD_SPINLOCKS */

kernel/locking/rtmutex_api.c

@@ -275,6 +275,7 @@ void __sched rt_mutex_proxy_unlock(struct rt_mutex_base *lock)
  * @lock:		the rt_mutex to take
  * @waiter:		the pre-initialized rt_mutex_waiter
  * @task:		the task to prepare
+ * @wake_q:		the wake_q to wake tasks after we release the wait_lock
  *
  * Starts the rt_mutex acquire; it enqueues the @waiter and does deadlock
  * detection. It does not wait, see rt_mutex_wait_proxy_lock() for that.
@@ -291,7 +292,8 @@ void __sched rt_mutex_proxy_unlock(struct rt_mutex_base *lock)
  */
 int __sched __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 					struct rt_mutex_waiter *waiter,
-					struct task_struct *task)
+					struct task_struct *task,
+					struct wake_q_head *wake_q)
 {
 	int ret;
 
@@ -302,7 +304,7 @@ int __sched __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 
 	/* We enforce deadlock detection for futexes */
 	ret = task_blocks_on_rt_mutex(lock, waiter, task, NULL,
-				      RT_MUTEX_FULL_CHAINWALK);
+				      RT_MUTEX_FULL_CHAINWALK, wake_q);
 
 	if (ret && !rt_mutex_owner(lock)) {
 		/*
@@ -341,12 +343,16 @@ int __sched rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				      struct task_struct *task)
 {
 	int ret;
+	DEFINE_WAKE_Q(wake_q);
 
 	raw_spin_lock_irq(&lock->wait_lock);
-	ret = __rt_mutex_start_proxy_lock(lock, waiter, task);
+	ret = __rt_mutex_start_proxy_lock(lock, waiter, task, &wake_q);
 	if (unlikely(ret))
 		remove_waiter(lock, waiter);
+	preempt_disable();
 	raw_spin_unlock_irq(&lock->wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
 
 	return ret;
 }

kernel/locking/rtmutex_common.h

@@ -83,7 +83,8 @@ extern void rt_mutex_init_proxy_locked(struct rt_mutex_base *lock,
 extern void rt_mutex_proxy_unlock(struct rt_mutex_base *lock);
 extern int __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				     struct rt_mutex_waiter *waiter,
-				     struct task_struct *task);
+				     struct task_struct *task,
+				     struct wake_q_head *);
 extern int rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				     struct rt_mutex_waiter *waiter,
 				     struct task_struct *task);

kernel/locking/rwbase_rt.c

@@ -69,6 +69,7 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 				      unsigned int state)
 {
 	struct rt_mutex_base *rtm = &rwb->rtmutex;
+	DEFINE_WAKE_Q(wake_q);
 	int ret;
 
 	rwbase_pre_schedule();
@@ -110,7 +111,7 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 	 * For rwlocks this returns 0 unconditionally, so the below
 	 * !ret conditionals are optimized out.
 	 */
-	ret = rwbase_rtmutex_slowlock_locked(rtm, state);
+	ret = rwbase_rtmutex_slowlock_locked(rtm, state, &wake_q);
 
 	/*
 	 * On success the rtmutex is held, so there can't be a writer
@@ -121,7 +122,12 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 	 */
 	if (!ret)
 		atomic_inc(&rwb->readers);
+
+	preempt_disable();
 	raw_spin_unlock_irq(&rtm->wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
+
 	if (!ret)
 		rwbase_rtmutex_unlock(rtm);
 

kernel/locking/rwsem.c

@@ -1413,8 +1413,8 @@ static inline void __downgrade_write(struct rw_semaphore *sem)
 #define rwbase_rtmutex_lock_state(rtm, state)		\
 	__rt_mutex_lock(rtm, state)
 
-#define rwbase_rtmutex_slowlock_locked(rtm, state)	\
+#define rwbase_rtmutex_slowlock_locked(rtm, state, wq)	\
-	__rt_mutex_slowlock_locked(rtm, NULL, state)
+	__rt_mutex_slowlock_locked(rtm, NULL, state, wq)
 
 #define rwbase_rtmutex_unlock(rtm)			\
 	__rt_mutex_unlock(rtm)

kernel/locking/spinlock_rt.c

@@ -162,9 +162,10 @@ rwbase_rtmutex_lock_state(struct rt_mutex_base *rtm, unsigned int state)
 }
 
 static __always_inline int
-rwbase_rtmutex_slowlock_locked(struct rt_mutex_base *rtm, unsigned int state)
+rwbase_rtmutex_slowlock_locked(struct rt_mutex_base *rtm, unsigned int state,
+			       struct wake_q_head *wake_q)
 {
-	rtlock_slowlock_locked(rtm);
+	rtlock_slowlock_locked(rtm, wake_q);
 	return 0;
 }
 

kernel/locking/ww_mutex.h

@@ -70,14 +70,14 @@ __ww_mutex_has_waiters(struct mutex *lock)
 	return atomic_long_read(&lock->owner) & MUTEX_FLAG_WAITERS;
 }
 
-static inline void lock_wait_lock(struct mutex *lock)
+static inline void lock_wait_lock(struct mutex *lock, unsigned long *flags)
 {
-	raw_spin_lock(&lock->wait_lock);
+	raw_spin_lock_irqsave(&lock->wait_lock, *flags);
 }
 
-static inline void unlock_wait_lock(struct mutex *lock)
+static inline void unlock_wait_lock(struct mutex *lock, unsigned long *flags)
 {
-	raw_spin_unlock(&lock->wait_lock);
+	raw_spin_unlock_irqrestore(&lock->wait_lock, *flags);
 }
 
 static inline void lockdep_assert_wait_lock_held(struct mutex *lock)
@@ -144,14 +144,14 @@ __ww_mutex_has_waiters(struct rt_mutex *lock)
 	return rt_mutex_has_waiters(&lock->rtmutex);
 }
 
-static inline void lock_wait_lock(struct rt_mutex *lock)
+static inline void lock_wait_lock(struct rt_mutex *lock, unsigned long *flags)
 {
-	raw_spin_lock(&lock->rtmutex.wait_lock);
+	raw_spin_lock_irqsave(&lock->rtmutex.wait_lock, *flags);
 }
 
-static inline void unlock_wait_lock(struct rt_mutex *lock)
+static inline void unlock_wait_lock(struct rt_mutex *lock, unsigned long *flags)
 {
-	raw_spin_unlock(&lock->rtmutex.wait_lock);
+	raw_spin_unlock_irqrestore(&lock->rtmutex.wait_lock, *flags);
 }
 
 static inline void lockdep_assert_wait_lock_held(struct rt_mutex *lock)
@@ -275,7 +275,7 @@ __ww_ctx_less(struct ww_acquire_ctx *a, struct ww_acquire_ctx *b)
  */
 static bool
 __ww_mutex_die(struct MUTEX *lock, struct MUTEX_WAITER *waiter,
-	       struct ww_acquire_ctx *ww_ctx)
+	       struct ww_acquire_ctx *ww_ctx, struct wake_q_head *wake_q)
 {
 	if (!ww_ctx->is_wait_die)
 		return false;
@@ -284,7 +284,7 @@ __ww_mutex_die(struct MUTEX *lock, struct MUTEX_WAITER *waiter,
 #ifndef WW_RT
 		debug_mutex_wake_waiter(lock, waiter);
 #endif
-		wake_up_process(waiter->task);
+		wake_q_add(wake_q, waiter->task);
 	}
 
 	return true;
@@ -299,7 +299,8 @@ __ww_mutex_die(struct MUTEX *lock, struct MUTEX_WAITER *waiter,
  */
 static bool __ww_mutex_wound(struct MUTEX *lock,
 			     struct ww_acquire_ctx *ww_ctx,
-			     struct ww_acquire_ctx *hold_ctx)
+			     struct ww_acquire_ctx *hold_ctx,
+			     struct wake_q_head *wake_q)
 {
 	struct task_struct *owner = __ww_mutex_owner(lock);
 
@@ -331,7 +332,7 @@ static bool __ww_mutex_wound(struct MUTEX *lock,
 		 * wakeup pending to re-read the wounded state.
 		 */
 		if (owner != current)
-			wake_up_process(owner);
+			wake_q_add(wake_q, owner);
 
 		return true;
 	}
@@ -352,7 +353,8 @@ static bool __ww_mutex_wound(struct MUTEX *lock,
  * The current task must not be on the wait list.
  */
 static void
-__ww_mutex_check_waiters(struct MUTEX *lock, struct ww_acquire_ctx *ww_ctx)
+__ww_mutex_check_waiters(struct MUTEX *lock, struct ww_acquire_ctx *ww_ctx,
+			 struct wake_q_head *wake_q)
 {
 	struct MUTEX_WAITER *cur;
 
@@ -364,8 +366,8 @@ __ww_mutex_check_waiters(struct MUTEX *lock, struct ww_acquire_ctx *ww_ctx)
 		if (!cur->ww_ctx)
 			continue;
 
-		if (__ww_mutex_die(lock, cur, ww_ctx) ||
+		if (__ww_mutex_die(lock, cur, ww_ctx, wake_q) ||
-		    __ww_mutex_wound(lock, cur->ww_ctx, ww_ctx))
+		    __ww_mutex_wound(lock, cur->ww_ctx, ww_ctx, wake_q))
 			break;
 	}
 }
@@ -377,6 +379,9 @@ __ww_mutex_check_waiters(struct MUTEX *lock, struct ww_acquire_ctx *ww_ctx)
 static __always_inline void
 ww_mutex_set_context_fastpath(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
 {
+	DEFINE_WAKE_Q(wake_q);
+	unsigned long flags;
+
 	ww_mutex_lock_acquired(lock, ctx);
 
 	/*
@@ -404,9 +409,12 @@ ww_mutex_set_context_fastpath(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
 	 * Uh oh, we raced in fastpath, check if any of the waiters need to
 	 * die or wound us.
 	 */
-	lock_wait_lock(&lock->base);
+	lock_wait_lock(&lock->base, &flags);
-	__ww_mutex_check_waiters(&lock->base, ctx);
+	__ww_mutex_check_waiters(&lock->base, ctx, &wake_q);
-	unlock_wait_lock(&lock->base);
+	preempt_disable();
+	unlock_wait_lock(&lock->base, &flags);
+	wake_up_q(&wake_q);
+	preempt_enable();
 }
 
 static __always_inline int
@@ -488,7 +496,8 @@ __ww_mutex_check_kill(struct MUTEX *lock, struct MUTEX_WAITER *waiter,
 static inline int
 __ww_mutex_add_waiter(struct MUTEX_WAITER *waiter,
 		      struct MUTEX *lock,
-		      struct ww_acquire_ctx *ww_ctx)
+		      struct ww_acquire_ctx *ww_ctx,
+		      struct wake_q_head *wake_q)
 {
 	struct MUTEX_WAITER *cur, *pos = NULL;
 	bool is_wait_die;
@@ -532,7 +541,7 @@ __ww_mutex_add_waiter(struct MUTEX_WAITER *waiter,
 		pos = cur;
 
 		/* Wait-Die: ensure younger waiters die. */
-		__ww_mutex_die(lock, cur, ww_ctx);
+		__ww_mutex_die(lock, cur, ww_ctx, wake_q);
 	}
 
 	__ww_waiter_add(lock, waiter, pos);
@@ -550,7 +559,7 @@ __ww_mutex_add_waiter(struct MUTEX_WAITER *waiter,
 		 * such that either we or the fastpath will wound @ww->ctx.
 		 */
 		smp_mb();
-		__ww_mutex_wound(lock, ww_ctx, ww->ctx);
+		__ww_mutex_wound(lock, ww_ctx, ww->ctx, wake_q);
 	}
 
 	return 0;

kernel/sched/core.c

@@ -832,7 +832,7 @@ static enum hrtimer_restart hrtick(struct hrtimer *timer)
 
 	rq_lock(rq, &rf);
 	update_rq_clock(rq);
-	rq->curr->sched_class->task_tick(rq, rq->curr, 1);
+	rq->donor->sched_class->task_tick(rq, rq->curr, 1);
 	rq_unlock(rq, &rf);
 
 	return HRTIMER_NORESTART;
@@ -941,10 +941,9 @@ static inline void hrtick_rq_init(struct rq *rq)
  * this avoids any races wrt polling state changes and thereby avoids
  * spurious IPIs.
  */
-static inline bool set_nr_and_not_polling(struct task_struct *p)
+static inline bool set_nr_and_not_polling(struct thread_info *ti, int tif)
 {
-	struct thread_info *ti = task_thread_info(p);
+	return !(fetch_or(&ti->flags, 1 << tif) & _TIF_POLLING_NRFLAG);
-	return !(fetch_or(&ti->flags, _TIF_NEED_RESCHED) & _TIF_POLLING_NRFLAG);
 }
 
 /*
@@ -969,9 +968,9 @@ static bool set_nr_if_polling(struct task_struct *p)
 }
 
 #else
-static inline bool set_nr_and_not_polling(struct task_struct *p)
+static inline bool set_nr_and_not_polling(struct thread_info *ti, int tif)
 {
-	set_tsk_need_resched(p);
+	set_ti_thread_flag(ti, tif);
 	return true;
 }
 
@@ -1076,28 +1075,70 @@ void wake_up_q(struct wake_q_head *head)
  * might also involve a cross-CPU call to trigger the scheduler on
  * the target CPU.
  */
-void resched_curr(struct rq *rq)
+static void __resched_curr(struct rq *rq, int tif)
 {
 	struct task_struct *curr = rq->curr;
+	struct thread_info *cti = task_thread_info(curr);
 	int cpu;
 
 	lockdep_assert_rq_held(rq);
 
-	if (test_tsk_need_resched(curr))
+	/*
+	 * Always immediately preempt the idle task; no point in delaying doing
+	 * actual work.
+	 */
+	if (is_idle_task(curr) && tif == TIF_NEED_RESCHED_LAZY)
+		tif = TIF_NEED_RESCHED;
+
+	if (cti->flags & ((1 << tif) | _TIF_NEED_RESCHED))
 		return;
 
 	cpu = cpu_of(rq);
 
 	if (cpu == smp_processor_id()) {
-		set_tsk_need_resched(curr);
+		set_ti_thread_flag(cti, tif);
-		set_preempt_need_resched();
+		if (tif == TIF_NEED_RESCHED)
+			set_preempt_need_resched();
 		return;
 	}
 
-	if (set_nr_and_not_polling(curr))
+	if (set_nr_and_not_polling(cti, tif)) {
-		smp_send_reschedule(cpu);
+		if (tif == TIF_NEED_RESCHED)
-	else
+			smp_send_reschedule(cpu);
+	} else {
 		trace_sched_wake_idle_without_ipi(cpu);
+	}
+}
+
+void resched_curr(struct rq *rq)
+{
+	__resched_curr(rq, TIF_NEED_RESCHED);
+}
+
+#ifdef CONFIG_PREEMPT_DYNAMIC
+static DEFINE_STATIC_KEY_FALSE(sk_dynamic_preempt_lazy);
+static __always_inline bool dynamic_preempt_lazy(void)
+{
+	return static_branch_unlikely(&sk_dynamic_preempt_lazy);
+}
+#else
+static __always_inline bool dynamic_preempt_lazy(void)
+{
+	return IS_ENABLED(CONFIG_PREEMPT_LAZY);
+}
+#endif
+
+static __always_inline int get_lazy_tif_bit(void)
+{
+	if (dynamic_preempt_lazy())
+		return TIF_NEED_RESCHED_LAZY;
+
+	return TIF_NEED_RESCHED;
+}
+
+void resched_curr_lazy(struct rq *rq)
+{
+	__resched_curr(rq, get_lazy_tif_bit());
 }
 
 void resched_cpu(int cpu)
@@ -1192,7 +1233,7 @@ static void wake_up_idle_cpu(int cpu)
 	 * and testing of the above solutions didn't appear to report
 	 * much benefits.
 	 */
-	if (set_nr_and_not_polling(rq->idle))
+	if (set_nr_and_not_polling(task_thread_info(rq->idle), TIF_NEED_RESCHED))
 		smp_send_reschedule(cpu);
 	else
 		trace_sched_wake_idle_without_ipi(cpu);
@@ -1399,7 +1440,7 @@ void set_load_weight(struct task_struct *p, bool update_load)
  * requests are serialized using a mutex to reduce the risk of conflicting
  * updates or API abuses.
  */
-static DEFINE_MUTEX(uclamp_mutex);
+static __maybe_unused DEFINE_MUTEX(uclamp_mutex);
 
 /* Max allowed minimum utilization */
 static unsigned int __maybe_unused sysctl_sched_uclamp_util_min = SCHED_CAPACITY_SCALE;
@@ -2024,10 +2065,10 @@ void enqueue_task(struct rq *rq, struct task_struct *p, int flags)
 	 */
 	uclamp_rq_inc(rq, p);
 
-	if (!(flags & ENQUEUE_RESTORE)) {
+	psi_enqueue(p, flags);
+
+	if (!(flags & ENQUEUE_RESTORE))
 		sched_info_enqueue(rq, p);
-		psi_enqueue(p, flags & ENQUEUE_MIGRATED);
-	}
 
 	if (sched_core_enabled(rq))
 		sched_core_enqueue(rq, p);
@@ -2044,10 +2085,10 @@ inline bool dequeue_task(struct rq *rq, struct task_struct *p, int flags)
 	if (!(flags & DEQUEUE_NOCLOCK))
 		update_rq_clock(rq);
 
-	if (!(flags & DEQUEUE_SAVE)) {
+	if (!(flags & DEQUEUE_SAVE))
 		sched_info_dequeue(rq, p);
-		psi_dequeue(p, !(flags & DEQUEUE_SLEEP));
+
-	}
+	psi_dequeue(p, flags);
 
 	/*
 	 * Must be before ->dequeue_task() because ->dequeue_task() can 'fail'
@@ -2135,16 +2176,18 @@ void check_class_changed(struct rq *rq, struct task_struct *p,
 
 void wakeup_preempt(struct rq *rq, struct task_struct *p, int flags)
 {
-	if (p->sched_class == rq->curr->sched_class)
+	struct task_struct *donor = rq->donor;
-		rq->curr->sched_class->wakeup_preempt(rq, p, flags);
+
-	else if (sched_class_above(p->sched_class, rq->curr->sched_class))
+	if (p->sched_class == donor->sched_class)
+		donor->sched_class->wakeup_preempt(rq, p, flags);
+	else if (sched_class_above(p->sched_class, donor->sched_class))
 		resched_curr(rq);
 
 	/*
 	 * A queue event has occurred, and we're going to schedule.  In
 	 * this case, we can save a useless back to back clock update.
 	 */
-	if (task_on_rq_queued(rq->curr) && test_tsk_need_resched(rq->curr))
+	if (task_on_rq_queued(donor) && test_tsk_need_resched(rq->curr))
 		rq_clock_skip_update(rq);
 }
 
@@ -2620,9 +2663,7 @@ int push_cpu_stop(void *arg)
 
 	// XXX validate p is still the highest prio task
 	if (task_rq(p) == rq) {
-		deactivate_task(rq, p, 0);
+		move_queued_task_locked(rq, lowest_rq, p);
-		set_task_cpu(p, lowest_rq->cpu);
-		activate_task(lowest_rq, p, 0);
 		resched_curr(lowest_rq);
 	}
 
@@ -2682,7 +2723,7 @@ __do_set_cpus_allowed(struct task_struct *p, struct affinity_context *ctx)
 		lockdep_assert_held(&p->pi_lock);
 
 	queued = task_on_rq_queued(p);
-	running = task_current(rq, p);
+	running = task_current_donor(rq, p);
 
 	if (queued) {
 		/*
@@ -2696,6 +2737,7 @@ __do_set_cpus_allowed(struct task_struct *p, struct affinity_context *ctx)
 		put_prev_task(rq, p);
 
 	p->sched_class->set_cpus_allowed(p, ctx);
+	mm_set_cpus_allowed(p->mm, ctx->new_mask);
 
 	if (queued)
 		enqueue_task(rq, p, ENQUEUE_RESTORE | ENQUEUE_NOCLOCK);
@@ -3308,9 +3350,7 @@ static void __migrate_swap_task(struct task_struct *p, int cpu)
 		rq_pin_lock(src_rq, &srf);
 		rq_pin_lock(dst_rq, &drf);
 
-		deactivate_task(src_rq, p, 0);
+		move_queued_task_locked(src_rq, dst_rq, p);
-		set_task_cpu(p, cpu);
-		activate_task(dst_rq, p, 0);
 		wakeup_preempt(dst_rq, p, 0);
 
 		rq_unpin_lock(dst_rq, &drf);
@@ -4424,7 +4464,8 @@ int wake_up_state(struct task_struct *p, unsigned int state)
  * Perform scheduler related setup for a newly forked process p.
  * p is forked by current.
  *
- * __sched_fork() is basic setup used by init_idle() too:
+ * __sched_fork() is basic setup which is also used by sched_init() to
+ * initialize the boot CPU's idle task.
  */
 static void __sched_fork(unsigned long clone_flags, struct task_struct *p)
 {
@@ -5517,7 +5558,7 @@ unsigned long long task_sched_runtime(struct task_struct *p)
 	 * project cycles that may never be accounted to this
 	 * thread, breaking clock_gettime().
 	 */
-	if (task_current(rq, p) && task_on_rq_queued(p)) {
+	if (task_current_donor(rq, p) && task_on_rq_queued(p)) {
 		prefetch_curr_exec_start(p);
 		update_rq_clock(rq);
 		p->sched_class->update_curr(rq);
@@ -5585,7 +5626,8 @@ void sched_tick(void)
 {
 	int cpu = smp_processor_id();
 	struct rq *rq = cpu_rq(cpu);
-	struct task_struct *curr;
+	/* accounting goes to the donor task */
+	struct task_struct *donor;
 	struct rq_flags rf;
 	unsigned long hw_pressure;
 	u64 resched_latency;
@@ -5596,19 +5638,23 @@ void sched_tick(void)
 	sched_clock_tick();
 
 	rq_lock(rq, &rf);
+	donor = rq->donor;
 
-	curr = rq->curr;
+	psi_account_irqtime(rq, donor, NULL);
-	psi_account_irqtime(rq, curr, NULL);
 
 	update_rq_clock(rq);
 	hw_pressure = arch_scale_hw_pressure(cpu_of(rq));
 	update_hw_load_avg(rq_clock_task(rq), rq, hw_pressure);
-	curr->sched_class->task_tick(rq, curr, 0);
+
+	if (dynamic_preempt_lazy() && tif_test_bit(TIF_NEED_RESCHED_LAZY))
+		resched_curr(rq);
+
+	donor->sched_class->task_tick(rq, donor, 0);
 	if (sched_feat(LATENCY_WARN))
 		resched_latency = cpu_resched_latency(rq);
 	calc_global_load_tick(rq);
 	sched_core_tick(rq);
-	task_tick_mm_cid(rq, curr);
+	task_tick_mm_cid(rq, donor);
 	scx_tick(rq);
 
 	rq_unlock(rq, &rf);
@@ -5618,8 +5664,8 @@ void sched_tick(void)
 
 	perf_event_task_tick();
 
-	if (curr->flags & PF_WQ_WORKER)
+	if (donor->flags & PF_WQ_WORKER)
-		wq_worker_tick(curr);
+		wq_worker_tick(donor);
 
 #ifdef CONFIG_SMP
 	if (!scx_switched_all()) {
@@ -5686,6 +5732,12 @@ static void sched_tick_remote(struct work_struct *work)
 		struct task_struct *curr = rq->curr;
 
 		if (cpu_online(cpu)) {
+			/*
+			 * Since this is a remote tick for full dynticks mode,
+			 * we are always sure that there is no proxy (only a
+			 * single task is running).
+			 */
+			SCHED_WARN_ON(rq->curr != rq->donor);
 			update_rq_clock(rq);
 
 			if (!is_idle_task(curr)) {
@@ -6309,10 +6361,7 @@ static bool try_steal_cookie(int this, int that)
 		if (sched_task_is_throttled(p, this))
 			goto next;
 
-		deactivate_task(src, p, 0);
+		move_queued_task_locked(src, dst, p);
-		set_task_cpu(p, this);
-		activate_task(dst, p, 0);
-
 		resched_curr(dst);
 
 		success = true;
@@ -6506,6 +6555,45 @@ pick_next_task(struct rq *rq, struct task_struct *prev, struct rq_flags *rf)
 #define SM_PREEMPT		1
 #define SM_RTLOCK_WAIT		2
 
+/*
+ * Helper function for __schedule()
+ *
+ * If a task does not have signals pending, deactivate it
+ * Otherwise marks the task's __state as RUNNING
+ */
+static bool try_to_block_task(struct rq *rq, struct task_struct *p,
+			      unsigned long task_state)
+{
+	int flags = DEQUEUE_NOCLOCK;
+
+	if (signal_pending_state(task_state, p)) {
+		WRITE_ONCE(p->__state, TASK_RUNNING);
+		return false;
+	}
+
+	p->sched_contributes_to_load =
+		(task_state & TASK_UNINTERRUPTIBLE) &&
+		!(task_state & TASK_NOLOAD) &&
+		!(task_state & TASK_FROZEN);
+
+	if (unlikely(is_special_task_state(task_state)))
+		flags |= DEQUEUE_SPECIAL;
+
+	/*
+	 * __schedule()			ttwu()
+	 *   prev_state = prev->state;    if (p->on_rq && ...)
+	 *   if (prev_state)		    goto out;
+	 *     p->on_rq = 0;		  smp_acquire__after_ctrl_dep();
+	 *				  p->state = TASK_WAKING
+	 *
+	 * Where __schedule() and ttwu() have matching control dependencies.
+	 *
+	 * After this, schedule() must not care about p->state any more.
+	 */
+	block_task(rq, p, flags);
+	return true;
+}
+
 /*
  * __schedule() is the main scheduler function.
  *
@@ -6614,37 +6702,12 @@ static void __sched notrace __schedule(int sched_mode)
 			goto picked;
 		}
 	} else if (!preempt && prev_state) {
-		if (signal_pending_state(prev_state, prev)) {
+		block = try_to_block_task(rq, prev, prev_state);
-			WRITE_ONCE(prev->__state, TASK_RUNNING);
-		} else {
-			int flags = DEQUEUE_NOCLOCK;
-
-			prev->sched_contributes_to_load =
-				(prev_state & TASK_UNINTERRUPTIBLE) &&
-				!(prev_state & TASK_NOLOAD) &&
-				!(prev_state & TASK_FROZEN);
-
-			if (unlikely(is_special_task_state(prev_state)))
-				flags |= DEQUEUE_SPECIAL;
-
-			/*
-			 * __schedule()			ttwu()
-			 *   prev_state = prev->state;    if (p->on_rq && ...)
-			 *   if (prev_state)		    goto out;
-			 *     p->on_rq = 0;		  smp_acquire__after_ctrl_dep();
-			 *				  p->state = TASK_WAKING
-			 *
-			 * Where __schedule() and ttwu() have matching control dependencies.
-			 *
-			 * After this, schedule() must not care about p->state any more.
-			 */
-			block_task(rq, prev, flags);
-			block = true;
-		}
 		switch_count = &prev->nvcsw;
 	}
 
 	next = pick_next_task(rq, prev, &rf);
+	rq_set_donor(rq, next);
 picked:
 	clear_tsk_need_resched(prev);
 	clear_preempt_need_resched();
@@ -7151,7 +7214,7 @@ void rt_mutex_setprio(struct task_struct *p, struct task_struct *pi_task)
 		dequeue_task(rq, p, DEQUEUE_SLEEP | DEQUEUE_DELAYED | DEQUEUE_NOCLOCK);
 
 	queued = task_on_rq_queued(p);
-	running = task_current(rq, p);
+	running = task_current_donor(rq, p);
 	if (queued)
 		dequeue_task(rq, p, queue_flag);
 	if (running)
@@ -7351,6 +7414,7 @@ EXPORT_SYMBOL(__cond_resched_rwlock_write);
  *   preempt_schedule           <- NOP
  *   preempt_schedule_notrace   <- NOP
  *   irqentry_exit_cond_resched <- NOP
+ *   dynamic_preempt_lazy       <- false
  *
  * VOLUNTARY:
  *   cond_resched               <- __cond_resched
@@ -7358,6 +7422,7 @@ EXPORT_SYMBOL(__cond_resched_rwlock_write);
  *   preempt_schedule           <- NOP
  *   preempt_schedule_notrace   <- NOP
  *   irqentry_exit_cond_resched <- NOP
+ *   dynamic_preempt_lazy       <- false
  *
  * FULL:
  *   cond_resched               <- RET0
@@ -7365,6 +7430,15 @@ EXPORT_SYMBOL(__cond_resched_rwlock_write);
  *   preempt_schedule           <- preempt_schedule
  *   preempt_schedule_notrace   <- preempt_schedule_notrace
  *   irqentry_exit_cond_resched <- irqentry_exit_cond_resched
+ *   dynamic_preempt_lazy       <- false
+ *
+ * LAZY:
+ *   cond_resched               <- RET0
+ *   might_resched              <- RET0
+ *   preempt_schedule           <- preempt_schedule
+ *   preempt_schedule_notrace   <- preempt_schedule_notrace
+ *   irqentry_exit_cond_resched <- irqentry_exit_cond_resched
+ *   dynamic_preempt_lazy       <- true
  */
 
 enum {
@@ -7372,30 +7446,41 @@ enum {
 	preempt_dynamic_none,
 	preempt_dynamic_voluntary,
 	preempt_dynamic_full,
+	preempt_dynamic_lazy,
 };
 
 int preempt_dynamic_mode = preempt_dynamic_undefined;
 
 int sched_dynamic_mode(const char *str)
 {
+#ifndef CONFIG_PREEMPT_RT
 	if (!strcmp(str, "none"))
 		return preempt_dynamic_none;
 
 	if (!strcmp(str, "voluntary"))
 		return preempt_dynamic_voluntary;
+#endif
 
 	if (!strcmp(str, "full"))
 		return preempt_dynamic_full;
 
+#ifdef CONFIG_ARCH_HAS_PREEMPT_LAZY
+	if (!strcmp(str, "lazy"))
+		return preempt_dynamic_lazy;
+#endif
+
 	return -EINVAL;
 }
 
+#define preempt_dynamic_key_enable(f)	static_key_enable(&sk_dynamic_##f.key)
+#define preempt_dynamic_key_disable(f)	static_key_disable(&sk_dynamic_##f.key)
+
 #if defined(CONFIG_HAVE_PREEMPT_DYNAMIC_CALL)
 #define preempt_dynamic_enable(f)	static_call_update(f, f##_dynamic_enabled)
 #define preempt_dynamic_disable(f)	static_call_update(f, f##_dynamic_disabled)
 #elif defined(CONFIG_HAVE_PREEMPT_DYNAMIC_KEY)
-#define preempt_dynamic_enable(f)	static_key_enable(&sk_dynamic_##f.key)
+#define preempt_dynamic_enable(f)	preempt_dynamic_key_enable(f)
-#define preempt_dynamic_disable(f)	static_key_disable(&sk_dynamic_##f.key)
+#define preempt_dynamic_disable(f)	preempt_dynamic_key_disable(f)
 #else
 #error "Unsupported PREEMPT_DYNAMIC mechanism"
 #endif
@@ -7415,6 +7500,7 @@ static void __sched_dynamic_update(int mode)
 	preempt_dynamic_enable(preempt_schedule);
 	preempt_dynamic_enable(preempt_schedule_notrace);
 	preempt_dynamic_enable(irqentry_exit_cond_resched);
+	preempt_dynamic_key_disable(preempt_lazy);
 
 	switch (mode) {
 	case preempt_dynamic_none:
@@ -7424,6 +7510,7 @@ static void __sched_dynamic_update(int mode)
 		preempt_dynamic_disable(preempt_schedule);
 		preempt_dynamic_disable(preempt_schedule_notrace);
 		preempt_dynamic_disable(irqentry_exit_cond_resched);
+		preempt_dynamic_key_disable(preempt_lazy);
 		if (mode != preempt_dynamic_mode)
 			pr_info("Dynamic Preempt: none\n");
 		break;
@@ -7435,6 +7522,7 @@ static void __sched_dynamic_update(int mode)
 		preempt_dynamic_disable(preempt_schedule);
 		preempt_dynamic_disable(preempt_schedule_notrace);
 		preempt_dynamic_disable(irqentry_exit_cond_resched);
+		preempt_dynamic_key_disable(preempt_lazy);
 		if (mode != preempt_dynamic_mode)
 			pr_info("Dynamic Preempt: voluntary\n");
 		break;
@@ -7446,9 +7534,22 @@ static void __sched_dynamic_update(int mode)
 		preempt_dynamic_enable(preempt_schedule);
 		preempt_dynamic_enable(preempt_schedule_notrace);
 		preempt_dynamic_enable(irqentry_exit_cond_resched);
+		preempt_dynamic_key_disable(preempt_lazy);
 		if (mode != preempt_dynamic_mode)
 			pr_info("Dynamic Preempt: full\n");
 		break;
+
+	case preempt_dynamic_lazy:
+		if (!klp_override)
+			preempt_dynamic_disable(cond_resched);
+		preempt_dynamic_disable(might_resched);
+		preempt_dynamic_enable(preempt_schedule);
+		preempt_dynamic_enable(preempt_schedule_notrace);
+		preempt_dynamic_enable(irqentry_exit_cond_resched);
+		preempt_dynamic_key_enable(preempt_lazy);
+		if (mode != preempt_dynamic_mode)
+			pr_info("Dynamic Preempt: lazy\n");
+		break;
 	}
 
 	preempt_dynamic_mode = mode;
@@ -7511,6 +7612,8 @@ static void __init preempt_dynamic_init(void)
 			sched_dynamic_update(preempt_dynamic_none);
 		} else if (IS_ENABLED(CONFIG_PREEMPT_VOLUNTARY)) {
 			sched_dynamic_update(preempt_dynamic_voluntary);
+		} else if (IS_ENABLED(CONFIG_PREEMPT_LAZY)) {
+			sched_dynamic_update(preempt_dynamic_lazy);
 		} else {
 			/* Default static call setting, nothing to do */
 			WARN_ON_ONCE(!IS_ENABLED(CONFIG_PREEMPT));
@@ -7531,6 +7634,7 @@ static void __init preempt_dynamic_init(void)
 PREEMPT_MODEL_ACCESSOR(none);
 PREEMPT_MODEL_ACCESSOR(voluntary);
 PREEMPT_MODEL_ACCESSOR(full);
+PREEMPT_MODEL_ACCESSOR(lazy);
 
 #else /* !CONFIG_PREEMPT_DYNAMIC: */
 
@@ -7683,8 +7787,6 @@ void __init init_idle(struct task_struct *idle, int cpu)
 	struct rq *rq = cpu_rq(cpu);
 	unsigned long flags;
 
-	__sched_fork(0, idle);
-
 	raw_spin_lock_irqsave(&idle->pi_lock, flags);
 	raw_spin_rq_lock(rq);
 
@@ -7699,10 +7801,8 @@ void __init init_idle(struct task_struct *idle, int cpu)
 
 #ifdef CONFIG_SMP
 	/*
-	 * It's possible that init_idle() gets called multiple times on a task,
+	 * No validation and serialization required at boot time and for
-	 * in that case do_set_cpus_allowed() will not do the right thing.
+	 * setting up the idle tasks of not yet online CPUs.
-	 *
-	 * And since this is boot we can forgo the serialization.
 	 */
 	set_cpus_allowed_common(idle, &ac);
 #endif
@@ -7721,6 +7821,7 @@ void __init init_idle(struct task_struct *idle, int cpu)
 	rcu_read_unlock();
 
 	rq->idle = idle;
+	rq_set_donor(rq, idle);
 	rcu_assign_pointer(rq->curr, idle);
 	idle->on_rq = TASK_ON_RQ_QUEUED;
 #ifdef CONFIG_SMP
@@ -7810,7 +7911,7 @@ void sched_setnuma(struct task_struct *p, int nid)
 
 	rq = task_rq_lock(p, &rf);
 	queued = task_on_rq_queued(p);
-	running = task_current(rq, p);
+	running = task_current_donor(rq, p);
 
 	if (queued)
 		dequeue_task(rq, p, DEQUEUE_SAVE);
@@ -8546,6 +8647,7 @@ void __init sched_init(void)
 	 * but because we are the idle thread, we just pick up running again
 	 * when this runqueue becomes "idle".
 	 */
+	__sched_fork(0, current);
 	init_idle(current, smp_processor_id());
 
 	calc_load_update = jiffies + LOAD_FREQ;
@@ -8960,7 +9062,7 @@ void sched_move_task(struct task_struct *tsk)
 
 	update_rq_clock(rq);
 
-	running = task_current(rq, tsk);
+	running = task_current_donor(rq, tsk);
 	queued = task_on_rq_queued(tsk);
 
 	if (queued)
@@ -10253,6 +10355,7 @@ int __sched_mm_cid_migrate_from_try_steal_cid(struct rq *src_rq,
 	 */
 	if (!try_cmpxchg(&src_pcpu_cid->cid, &lazy_cid, MM_CID_UNSET))
 		return -1;
+	WRITE_ONCE(src_pcpu_cid->recent_cid, MM_CID_UNSET);
 	return src_cid;
 }
 
@@ -10265,7 +10368,8 @@ void sched_mm_cid_migrate_to(struct rq *dst_rq, struct task_struct *t)
 {
 	struct mm_cid *src_pcpu_cid, *dst_pcpu_cid;
 	struct mm_struct *mm = t->mm;
-	int src_cid, dst_cid, src_cpu;
+	int src_cid, src_cpu;
+	bool dst_cid_is_set;
 	struct rq *src_rq;
 
 	lockdep_assert_rq_held(dst_rq);
@@ -10282,9 +10386,9 @@ void sched_mm_cid_migrate_to(struct rq *dst_rq, struct task_struct *t)
 	 * allocation closest to 0 in cases where few threads migrate around
 	 * many CPUs.
 	 *
-	 * If destination cid is already set, we may have to just clear
+	 * If destination cid or recent cid is already set, we may have
-	 * the src cid to ensure compactness in frequent migrations
+	 * to just clear the src cid to ensure compactness in frequent
-	 * scenarios.
+	 * migrations scenarios.
 	 *
 	 * It is not useful to clear the src cid when the number of threads is
 	 * greater or equal to the number of allowed CPUs, because user-space
@@ -10292,9 +10396,9 @@ void sched_mm_cid_migrate_to(struct rq *dst_rq, struct task_struct *t)
 	 * allowed CPUs.
 	 */
 	dst_pcpu_cid = per_cpu_ptr(mm->pcpu_cid, cpu_of(dst_rq));
-	dst_cid = READ_ONCE(dst_pcpu_cid->cid);
+	dst_cid_is_set = !mm_cid_is_unset(READ_ONCE(dst_pcpu_cid->cid)) ||
-	if (!mm_cid_is_unset(dst_cid) &&
+			 !mm_cid_is_unset(READ_ONCE(dst_pcpu_cid->recent_cid));
-	    atomic_read(&mm->mm_users) >= t->nr_cpus_allowed)
+	if (dst_cid_is_set && atomic_read(&mm->mm_users) >= READ_ONCE(mm->nr_cpus_allowed))
 		return;
 	src_pcpu_cid = per_cpu_ptr(mm->pcpu_cid, src_cpu);
 	src_rq = cpu_rq(src_cpu);
@@ -10305,13 +10409,14 @@ void sched_mm_cid_migrate_to(struct rq *dst_rq, struct task_struct *t)
 							    src_cid);
 	if (src_cid == -1)
 		return;
-	if (!mm_cid_is_unset(dst_cid)) {
+	if (dst_cid_is_set) {
 		__mm_cid_put(mm, src_cid);
 		return;
 	}
 	/* Move src_cid to dst cpu. */
 	mm_cid_snapshot_time(dst_rq, mm);
 	WRITE_ONCE(dst_pcpu_cid->cid, src_cid);
+	WRITE_ONCE(dst_pcpu_cid->recent_cid, src_cid);
 }
 
 static void sched_mm_cid_remote_clear(struct mm_struct *mm, struct mm_cid *pcpu_cid,
@@ -10550,7 +10655,7 @@ void sched_mm_cid_after_execve(struct task_struct *t)
 		 * Matches barrier in sched_mm_cid_remote_clear_old().
 		 */
 		smp_mb();
-		t->last_mm_cid = t->mm_cid = mm_cid_get(rq, mm);
+		t->last_mm_cid = t->mm_cid = mm_cid_get(rq, t, mm);
 	}
 	rseq_set_notify_resume(t);
 }

kernel/sched/deadline.c

@@ -1339,7 +1339,7 @@ static enum hrtimer_restart dl_task_timer(struct hrtimer *timer)
 #endif
 
 	enqueue_task_dl(rq, p, ENQUEUE_REPLENISH);
-	if (dl_task(rq->curr))
+	if (dl_task(rq->donor))
 		wakeup_preempt_dl(rq, p, 0);
 	else
 		resched_curr(rq);
@@ -1736,11 +1736,11 @@ int dl_server_apply_params(struct sched_dl_entity *dl_se, u64 runtime, u64 perio
  */
 static void update_curr_dl(struct rq *rq)
 {
-	struct task_struct *curr = rq->curr;
+	struct task_struct *donor = rq->donor;
-	struct sched_dl_entity *dl_se = &curr->dl;
+	struct sched_dl_entity *dl_se = &donor->dl;
 	s64 delta_exec;
 
-	if (!dl_task(curr) || !on_dl_rq(dl_se))
+	if (!dl_task(donor) || !on_dl_rq(dl_se))
 		return;
 
 	/*
@@ -2213,7 +2213,7 @@ static int find_later_rq(struct task_struct *task);
 static int
 select_task_rq_dl(struct task_struct *p, int cpu, int flags)
 {
-	struct task_struct *curr;
+	struct task_struct *curr, *donor;
 	bool select_rq;
 	struct rq *rq;
 
@@ -2224,6 +2224,7 @@ select_task_rq_dl(struct task_struct *p, int cpu, int flags)
 
 	rcu_read_lock();
 	curr = READ_ONCE(rq->curr); /* unlocked access */
+	donor = READ_ONCE(rq->donor);
 
 	/*
 	 * If we are dealing with a -deadline task, we must
@@ -2234,9 +2235,9 @@ select_task_rq_dl(struct task_struct *p, int cpu, int flags)
 	 * other hand, if it has a shorter deadline, we
 	 * try to make it stay here, it might be important.
 	 */
-	select_rq = unlikely(dl_task(curr)) &&
+	select_rq = unlikely(dl_task(donor)) &&
 		    (curr->nr_cpus_allowed < 2 ||
-		     !dl_entity_preempt(&p->dl, &curr->dl)) &&
+		     !dl_entity_preempt(&p->dl, &donor->dl)) &&
 		    p->nr_cpus_allowed > 1;
 
 	/*
@@ -2299,7 +2300,7 @@ static void check_preempt_equal_dl(struct rq *rq, struct task_struct *p)
 	 * let's hope p can move out.
 	 */
 	if (rq->curr->nr_cpus_allowed == 1 ||
-	    !cpudl_find(&rq->rd->cpudl, rq->curr, NULL))
+	    !cpudl_find(&rq->rd->cpudl, rq->donor, NULL))
 		return;
 
 	/*
@@ -2338,7 +2339,7 @@ static int balance_dl(struct rq *rq, struct task_struct *p, struct rq_flags *rf)
 static void wakeup_preempt_dl(struct rq *rq, struct task_struct *p,
 				  int flags)
 {
-	if (dl_entity_preempt(&p->dl, &rq->curr->dl)) {
+	if (dl_entity_preempt(&p->dl, &rq->donor->dl)) {
 		resched_curr(rq);
 		return;
 	}
@@ -2348,7 +2349,7 @@ static void wakeup_preempt_dl(struct rq *rq, struct task_struct *p,
 	 * In the unlikely case current and p have the same deadline
 	 * let us try to decide what's the best thing to do...
 	 */
-	if ((p->dl.deadline == rq->curr->dl.deadline) &&
+	if ((p->dl.deadline == rq->donor->dl.deadline) &&
 	    !test_tsk_need_resched(rq->curr))
 		check_preempt_equal_dl(rq, p);
 #endif /* CONFIG_SMP */
@@ -2380,7 +2381,7 @@ static void set_next_task_dl(struct rq *rq, struct task_struct *p, bool first)
 	if (!first)
 		return;
 
-	if (rq->curr->sched_class != &dl_sched_class)
+	if (rq->donor->sched_class != &dl_sched_class)
 		update_dl_rq_load_avg(rq_clock_pelt(rq), rq, 0);
 
 	deadline_queue_push_tasks(rq);
@@ -2487,14 +2488,6 @@ static void task_fork_dl(struct task_struct *p)
 /* Only try algorithms three times */
 #define DL_MAX_TRIES 3
 
-static int pick_dl_task(struct rq *rq, struct task_struct *p, int cpu)
-{
-	if (!task_on_cpu(rq, p) &&
-	    cpumask_test_cpu(cpu, &p->cpus_mask))
-		return 1;
-	return 0;
-}
-
 /*
  * Return the earliest pushable rq's task, which is suitable to be executed
  * on the CPU, NULL otherwise:
@@ -2513,7 +2506,7 @@ next_node:
 	if (next_node) {
 		p = __node_2_pdl(next_node);
 
-		if (pick_dl_task(rq, p, cpu))
+		if (task_is_pushable(rq, p, cpu))
 			return p;
 
 		next_node = rb_next(next_node);
@@ -2707,8 +2700,8 @@ retry:
 	 * can move away, it makes sense to just reschedule
 	 * without going further in pushing next_task.
 	 */
-	if (dl_task(rq->curr) &&
+	if (dl_task(rq->donor) &&
-	    dl_time_before(next_task->dl.deadline, rq->curr->dl.deadline) &&
+	    dl_time_before(next_task->dl.deadline, rq->donor->dl.deadline) &&
 	    rq->curr->nr_cpus_allowed > 1) {
 		resched_curr(rq);
 		return 0;
@@ -2751,9 +2744,7 @@ retry:
 		goto retry;
 	}
 
-	deactivate_task(rq, next_task, 0);
+	move_queued_task_locked(rq, later_rq, next_task);
-	set_task_cpu(next_task, later_rq->cpu);
-	activate_task(later_rq, next_task, 0);
 	ret = 1;
 
 	resched_curr(later_rq);
@@ -2833,15 +2824,13 @@ static void pull_dl_task(struct rq *this_rq)
 			 * deadline than the current task of its runqueue.
 			 */
 			if (dl_time_before(p->dl.deadline,
-					   src_rq->curr->dl.deadline))
+					   src_rq->donor->dl.deadline))
 				goto skip;
 
 			if (is_migration_disabled(p)) {
 				push_task = get_push_task(src_rq);
 			} else {
-				deactivate_task(src_rq, p, 0);
+				move_queued_task_locked(src_rq, this_rq, p);
-				set_task_cpu(p, this_cpu);
-				activate_task(this_rq, p, 0);
 				dmin = p->dl.deadline;
 				resched = true;
 			}
@@ -2874,9 +2863,9 @@ static void task_woken_dl(struct rq *rq, struct task_struct *p)
 	if (!task_on_cpu(rq, p) &&
 	    !test_tsk_need_resched(rq->curr) &&
 	    p->nr_cpus_allowed > 1 &&
-	    dl_task(rq->curr) &&
+	    dl_task(rq->donor) &&
 	    (rq->curr->nr_cpus_allowed < 2 ||
-	     !dl_entity_preempt(&p->dl, &rq->curr->dl))) {
+	     !dl_entity_preempt(&p->dl, &rq->donor->dl))) {
 		push_dl_tasks(rq);
 	}
 }
@@ -3051,12 +3040,12 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p)
 		return;
 	}
 
-	if (rq->curr != p) {
+	if (rq->donor != p) {
 #ifdef CONFIG_SMP
 		if (p->nr_cpus_allowed > 1 && rq->dl.overloaded)
 			deadline_queue_push_tasks(rq);
 #endif
-		if (dl_task(rq->curr))
+		if (dl_task(rq->donor))
 			wakeup_preempt_dl(rq, p, 0);
 		else
 			resched_curr(rq);
@@ -3085,7 +3074,7 @@ static void prio_changed_dl(struct rq *rq, struct task_struct *p,
 	if (!rq->dl.overloaded)
 		deadline_queue_pull_task(rq);
 
-	if (task_current(rq, p)) {
+	if (task_current_donor(rq, p)) {
 		/*
 		 * If we now have a earlier deadline task than p,
 		 * then reschedule, provided p is still on this

kernel/sched/debug.c

@@ -245,11 +245,12 @@ static ssize_t sched_dynamic_write(struct file *filp, const char __user *ubuf,
 static int sched_dynamic_show(struct seq_file *m, void *v)
 {
 	static const char * preempt_modes[] = {
-		"none", "voluntary", "full"
+		"none", "voluntary", "full", "lazy",
 	};
-	int i;
+	int j = ARRAY_SIZE(preempt_modes) - !IS_ENABLED(CONFIG_ARCH_HAS_PREEMPT_LAZY);
+	int i = IS_ENABLED(CONFIG_PREEMPT_RT) * 2;
 
-	for (i = 0; i < ARRAY_SIZE(preempt_modes); i++) {
+	for (; i < j; i++) {
 		if (preempt_dynamic_mode == i)
 			seq_puts(m, "(");
 		seq_puts(m, preempt_modes[i]);

kernel/sched/ext.c

@@ -3567,12 +3567,7 @@ static void scx_ops_exit_task(struct task_struct *p)
 
 void init_scx_entity(struct sched_ext_entity *scx)
 {
-	/*
+	memset(scx, 0, sizeof(*scx));
-	 * init_idle() calls this function again after fork sequence is
-	 * complete. Don't touch ->tasks_node as it's already linked.
-	 */
-	memset(scx, 0, offsetof(struct sched_ext_entity, tasks_node));
-
 	INIT_LIST_HEAD(&scx->dsq_list.node);
 	RB_CLEAR_NODE(&scx->dsq_priq);
 	scx->sticky_cpu = -1;

kernel/sched/fair.c

@@ -1200,12 +1200,12 @@ static inline bool do_preempt_short(struct cfs_rq *cfs_rq,
  */
 s64 update_curr_common(struct rq *rq)
 {
-	struct task_struct *curr = rq->curr;
+	struct task_struct *donor = rq->donor;
 	s64 delta_exec;
 
-	delta_exec = update_curr_se(rq, &curr->se);
+	delta_exec = update_curr_se(rq, &donor->se);
 	if (likely(delta_exec > 0))
-		update_curr_task(curr, delta_exec);
+		update_curr_task(donor, delta_exec);
 
 	return delta_exec;
 }
@@ -1251,14 +1251,14 @@ static void update_curr(struct cfs_rq *cfs_rq)
 		return;
 
 	if (resched || did_preempt_short(cfs_rq, curr)) {
-		resched_curr(rq);
+		resched_curr_lazy(rq);
 		clear_buddies(cfs_rq, curr);
 	}
 }
 
 static void update_curr_fair(struct rq *rq)
 {
-	update_curr(cfs_rq_of(&rq->curr->se));
+	update_curr(cfs_rq_of(&rq->donor->se));
 }
 
 static inline void
@@ -5280,7 +5280,7 @@ place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
 	 *
 	 * EEVDF: placement strategy #1 / #2
 	 */
-	if (sched_feat(PLACE_LAG) && cfs_rq->nr_running) {
+	if (sched_feat(PLACE_LAG) && cfs_rq->nr_running && se->vlag) {
 		struct sched_entity *curr = cfs_rq->curr;
 		unsigned long load;
 
@@ -5678,15 +5678,9 @@ entity_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr, int queued)
 	 * validating it and just reschedule.
 	 */
 	if (queued) {
-		resched_curr(rq_of(cfs_rq));
+		resched_curr_lazy(rq_of(cfs_rq));
 		return;
 	}
-	/*
-	 * don't let the period tick interfere with the hrtick preemption
-	 */
-	if (!sched_feat(DOUBLE_TICK) &&
-			hrtimer_active(&rq_of(cfs_rq)->hrtick_timer))
-		return;
 #endif
 }
 
@@ -6822,7 +6816,7 @@ static void hrtick_start_fair(struct rq *rq, struct task_struct *p)
 		s64 delta = slice - ran;
 
 		if (delta < 0) {
-			if (task_current(rq, p))
+			if (task_current_donor(rq, p))
 				resched_curr(rq);
 			return;
 		}
@@ -6837,12 +6831,12 @@ static void hrtick_start_fair(struct rq *rq, struct task_struct *p)
  */
 static void hrtick_update(struct rq *rq)
 {
-	struct task_struct *curr = rq->curr;
+	struct task_struct *donor = rq->donor;
 
-	if (!hrtick_enabled_fair(rq) || curr->sched_class != &fair_sched_class)
+	if (!hrtick_enabled_fair(rq) || donor->sched_class != &fair_sched_class)
 		return;
 
-	hrtick_start_fair(rq, curr);
+	hrtick_start_fair(rq, donor);
 }
 #else /* !CONFIG_SCHED_HRTICK */
 static inline void
@@ -8763,9 +8757,9 @@ static void set_next_buddy(struct sched_entity *se)
  */
 static void check_preempt_wakeup_fair(struct rq *rq, struct task_struct *p, int wake_flags)
 {
-	struct task_struct *curr = rq->curr;
+	struct task_struct *donor = rq->donor;
-	struct sched_entity *se = &curr->se, *pse = &p->se;
+	struct sched_entity *se = &donor->se, *pse = &p->se;
-	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
+	struct cfs_rq *cfs_rq = task_cfs_rq(donor);
 	int cse_is_idle, pse_is_idle;
 
 	if (unlikely(se == pse))
@@ -8794,7 +8788,7 @@ static void check_preempt_wakeup_fair(struct rq *rq, struct task_struct *p, int
 	 * prevents us from potentially nominating it as a false LAST_BUDDY
 	 * below.
 	 */
-	if (test_tsk_need_resched(curr))
+	if (test_tsk_need_resched(rq->curr))
 		return;
 
 	if (!sched_feat(WAKEUP_PREEMPTION))
@@ -8842,7 +8836,7 @@ static void check_preempt_wakeup_fair(struct rq *rq, struct task_struct *p, int
 	return;
 
 preempt:
-	resched_curr(rq);
+	resched_curr_lazy(rq);
 }
 
 static struct task_struct *pick_task_fair(struct rq *rq)
@@ -13093,7 +13087,7 @@ prio_changed_fair(struct rq *rq, struct task_struct *p, int oldprio)
 	 * our priority decreased, or if we are not currently running on
 	 * this runqueue and our priority is higher than the current's
 	 */
-	if (task_current(rq, p)) {
+	if (task_current_donor(rq, p)) {
 		if (p->prio > oldprio)
 			resched_curr(rq);
 	} else
@@ -13200,7 +13194,7 @@ static void switched_to_fair(struct rq *rq, struct task_struct *p)
 		 * kick off the schedule if running, otherwise just see
 		 * if we can still preempt the current task.
 		 */
-		if (task_current(rq, p))
+		if (task_current_donor(rq, p))
 			resched_curr(rq);
 		else
 			wakeup_preempt(rq, p, 0);

kernel/sched/features.h

@@ -19,7 +19,7 @@ SCHED_FEAT(PLACE_REL_DEADLINE, true)
  */
 SCHED_FEAT(RUN_TO_PARITY, true)
 /*
- * Allow wakeup of tasks with a shorter slice to cancel RESPECT_SLICE for
+ * Allow wakeup of tasks with a shorter slice to cancel RUN_TO_PARITY for
  * current.
  */
 SCHED_FEAT(PREEMPT_SHORT, true)
@@ -56,7 +56,6 @@ SCHED_FEAT(WAKEUP_PREEMPTION, true)
 
 SCHED_FEAT(HRTICK, false)
 SCHED_FEAT(HRTICK_DL, false)
-SCHED_FEAT(DOUBLE_TICK, false)
 
 /*
  * Decrement CPU capacity based on time not spent running tasks

kernel/sched/idle.c

@@ -271,7 +271,6 @@ static void do_idle(void)
 	tick_nohz_idle_enter();
 
 	while (!need_resched()) {
-		rmb();
 
 		/*
 		 * Interrupts shouldn't be re-enabled from that point on until

kernel/sched/pelt.c

@@ -476,7 +476,7 @@ int update_irq_load_avg(struct rq *rq, u64 running)
 bool update_other_load_avgs(struct rq *rq)
 {
 	u64 now = rq_clock_pelt(rq);
-	const struct sched_class *curr_class = rq->curr->sched_class;
+	const struct sched_class *curr_class = rq->donor->sched_class;
 	unsigned long hw_pressure = arch_scale_hw_pressure(cpu_of(rq));
 
 	lockdep_assert_rq_held(rq);

kernel/sched/rt.c

@@ -528,7 +528,7 @@ static void dequeue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags)
 
 static void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
 {
-	struct task_struct *curr = rq_of_rt_rq(rt_rq)->curr;
+	struct task_struct *donor = rq_of_rt_rq(rt_rq)->donor;
 	struct rq *rq = rq_of_rt_rq(rt_rq);
 	struct sched_rt_entity *rt_se;
 
@@ -542,7 +542,7 @@ static void sched_rt_rq_enqueue(struct rt_rq *rt_rq)
 		else if (!on_rt_rq(rt_se))
 			enqueue_rt_entity(rt_se, 0);
 
-		if (rt_rq->highest_prio.curr < curr->prio)
+		if (rt_rq->highest_prio.curr < donor->prio)
 			resched_curr(rq);
 	}
 }
@@ -988,10 +988,10 @@ static inline int rt_se_prio(struct sched_rt_entity *rt_se)
  */
 static void update_curr_rt(struct rq *rq)
 {
-	struct task_struct *curr = rq->curr;
+	struct task_struct *donor = rq->donor;
 	s64 delta_exec;
 
-	if (curr->sched_class != &rt_sched_class)
+	if (donor->sched_class != &rt_sched_class)
 		return;
 
 	delta_exec = update_curr_common(rq);
@@ -999,7 +999,7 @@ static void update_curr_rt(struct rq *rq)
 		return;
 
 #ifdef CONFIG_RT_GROUP_SCHED
-	struct sched_rt_entity *rt_se = &curr->rt;
+	struct sched_rt_entity *rt_se = &donor->rt;
 
 	if (!rt_bandwidth_enabled())
 		return;
@@ -1535,7 +1535,7 @@ static int find_lowest_rq(struct task_struct *task);
 static int
 select_task_rq_rt(struct task_struct *p, int cpu, int flags)
 {
-	struct task_struct *curr;
+	struct task_struct *curr, *donor;
 	struct rq *rq;
 	bool test;
 
@@ -1547,6 +1547,7 @@ select_task_rq_rt(struct task_struct *p, int cpu, int flags)
 
 	rcu_read_lock();
 	curr = READ_ONCE(rq->curr); /* unlocked access */
+	donor = READ_ONCE(rq->donor);
 
 	/*
 	 * If the current task on @p's runqueue is an RT task, then
@@ -1575,8 +1576,8 @@ select_task_rq_rt(struct task_struct *p, int cpu, int flags)
 	 * systems like big.LITTLE.
 	 */
 	test = curr &&
-	       unlikely(rt_task(curr)) &&
+	       unlikely(rt_task(donor)) &&
-	       (curr->nr_cpus_allowed < 2 || curr->prio <= p->prio);
+	       (curr->nr_cpus_allowed < 2 || donor->prio <= p->prio);
 
 	if (test || !rt_task_fits_capacity(p, cpu)) {
 		int target = find_lowest_rq(p);
@@ -1606,12 +1607,8 @@ out:
 
 static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p)
 {
-	/*
-	 * Current can't be migrated, useless to reschedule,
-	 * let's hope p can move out.
-	 */
 	if (rq->curr->nr_cpus_allowed == 1 ||
-	    !cpupri_find(&rq->rd->cpupri, rq->curr, NULL))
+	    !cpupri_find(&rq->rd->cpupri, rq->donor, NULL))
 		return;
 
 	/*
@@ -1654,7 +1651,9 @@ static int balance_rt(struct rq *rq, struct task_struct *p, struct rq_flags *rf)
  */
 static void wakeup_preempt_rt(struct rq *rq, struct task_struct *p, int flags)
 {
-	if (p->prio < rq->curr->prio) {
+	struct task_struct *donor = rq->donor;
+
+	if (p->prio < donor->prio) {
 		resched_curr(rq);
 		return;
 	}
@@ -1672,7 +1671,7 @@ static void wakeup_preempt_rt(struct rq *rq, struct task_struct *p, int flags)
 	 * to move current somewhere else, making room for our non-migratable
 	 * task.
 	 */
-	if (p->prio == rq->curr->prio && !test_tsk_need_resched(rq->curr))
+	if (p->prio == donor->prio && !test_tsk_need_resched(rq->curr))
 		check_preempt_equal_prio(rq, p);
 #endif
 }
@@ -1697,7 +1696,7 @@ static inline void set_next_task_rt(struct rq *rq, struct task_struct *p, bool f
 	 * utilization. We only care of the case where we start to schedule a
 	 * rt task
 	 */
-	if (rq->curr->sched_class != &rt_sched_class)
+	if (rq->donor->sched_class != &rt_sched_class)
 		update_rt_rq_load_avg(rq_clock_pelt(rq), rq, 0);
 
 	rt_queue_push_tasks(rq);
@@ -1773,15 +1772,6 @@ static void put_prev_task_rt(struct rq *rq, struct task_struct *p, struct task_s
 /* Only try algorithms three times */
 #define RT_MAX_TRIES 3
 
-static int pick_rt_task(struct rq *rq, struct task_struct *p, int cpu)
-{
-	if (!task_on_cpu(rq, p) &&
-	    cpumask_test_cpu(cpu, &p->cpus_mask))
-		return 1;
-
-	return 0;
-}
-
 /*
  * Return the highest pushable rq's task, which is suitable to be executed
  * on the CPU, NULL otherwise
@@ -1795,7 +1785,7 @@ static struct task_struct *pick_highest_pushable_task(struct rq *rq, int cpu)
 		return NULL;
 
 	plist_for_each_entry(p, head, pushable_tasks) {
-		if (pick_rt_task(rq, p, cpu))
+		if (task_is_pushable(rq, p, cpu))
 			return p;
 	}
 
@@ -1968,6 +1958,7 @@ static struct task_struct *pick_next_pushable_task(struct rq *rq)
 
 	BUG_ON(rq->cpu != task_cpu(p));
 	BUG_ON(task_current(rq, p));
+	BUG_ON(task_current_donor(rq, p));
 	BUG_ON(p->nr_cpus_allowed <= 1);
 
 	BUG_ON(!task_on_rq_queued(p));
@@ -2000,7 +1991,7 @@ retry:
 	 * higher priority than current. If that's the case
 	 * just reschedule current.
 	 */
-	if (unlikely(next_task->prio < rq->curr->prio)) {
+	if (unlikely(next_task->prio < rq->donor->prio)) {
 		resched_curr(rq);
 		return 0;
 	}
@@ -2021,7 +2012,7 @@ retry:
 		 * Note that the stoppers are masqueraded as SCHED_FIFO
 		 * (cf. sched_set_stop_task()), so we can't rely on rt_task().
 		 */
-		if (rq->curr->sched_class != &rt_sched_class)
+		if (rq->donor->sched_class != &rt_sched_class)
 			return 0;
 
 		cpu = find_lowest_rq(rq->curr);
@@ -2088,9 +2079,7 @@ retry:
 		goto retry;
 	}
 
-	deactivate_task(rq, next_task, 0);
+	move_queued_task_locked(rq, lowest_rq, next_task);
-	set_task_cpu(next_task, lowest_rq->cpu);
-	activate_task(lowest_rq, next_task, 0);
 	resched_curr(lowest_rq);
 	ret = 1;
 
@@ -2355,15 +2344,13 @@ static void pull_rt_task(struct rq *this_rq)
 			 * p if it is lower in priority than the
 			 * current task on the run queue
 			 */
-			if (p->prio < src_rq->curr->prio)
+			if (p->prio < src_rq->donor->prio)
 				goto skip;
 
 			if (is_migration_disabled(p)) {
 				push_task = get_push_task(src_rq);
 			} else {
-				deactivate_task(src_rq, p, 0);
+				move_queued_task_locked(src_rq, this_rq, p);
-				set_task_cpu(p, this_cpu);
-				activate_task(this_rq, p, 0);
 				resched = true;
 			}
 			/*
@@ -2399,9 +2386,9 @@ static void task_woken_rt(struct rq *rq, struct task_struct *p)
 	bool need_to_push = !task_on_cpu(rq, p) &&
 			    !test_tsk_need_resched(rq->curr) &&
 			    p->nr_cpus_allowed > 1 &&
-			    (dl_task(rq->curr) || rt_task(rq->curr)) &&
+			    (dl_task(rq->donor) || rt_task(rq->donor)) &&
 			    (rq->curr->nr_cpus_allowed < 2 ||
-			     rq->curr->prio <= p->prio);
+			     rq->donor->prio <= p->prio);
 
 	if (need_to_push)
 		push_rt_tasks(rq);
@@ -2485,7 +2472,7 @@ static void switched_to_rt(struct rq *rq, struct task_struct *p)
 		if (p->nr_cpus_allowed > 1 && rq->rt.overloaded)
 			rt_queue_push_tasks(rq);
 #endif /* CONFIG_SMP */
-		if (p->prio < rq->curr->prio && cpu_online(cpu_of(rq)))
+		if (p->prio < rq->donor->prio && cpu_online(cpu_of(rq)))
 			resched_curr(rq);
 	}
 }
@@ -2500,7 +2487,7 @@ prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio)
 	if (!task_on_rq_queued(p))
 		return;
 
-	if (task_current(rq, p)) {
+	if (task_current_donor(rq, p)) {
 #ifdef CONFIG_SMP
 		/*
 		 * If our priority decreases while running, we
@@ -2526,7 +2513,7 @@ prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio)
 		 * greater than the current running task
 		 * then reschedule.
 		 */
-		if (p->prio < rq->curr->prio)
+		if (p->prio < rq->donor->prio)
 			resched_curr(rq);
 	}
 }

kernel/sched/sched.h

@@ -1148,7 +1148,10 @@ struct rq {
 	 */
 	unsigned int		nr_uninterruptible;
 
-	struct task_struct __rcu	*curr;
+	union {
+		struct task_struct __rcu *donor; /* Scheduler context */
+		struct task_struct __rcu *curr;  /* Execution context */
+	};
 	struct sched_dl_entity	*dl_server;
 	struct task_struct	*idle;
 	struct task_struct	*stop;
@@ -1345,6 +1348,11 @@ DECLARE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
 #define cpu_curr(cpu)		(cpu_rq(cpu)->curr)
 #define raw_rq()		raw_cpu_ptr(&runqueues)
 
+static inline void rq_set_donor(struct rq *rq, struct task_struct *t)
+{
+	/* Do nothing */
+}
+
 #ifdef CONFIG_SCHED_CORE
 static inline struct cpumask *sched_group_span(struct sched_group *sg);
 
@@ -2086,34 +2094,6 @@ static inline const struct cpumask *task_user_cpus(struct task_struct *p)
 
 #endif /* CONFIG_SMP */
 
-#include "stats.h"
-
-#if defined(CONFIG_SCHED_CORE) && defined(CONFIG_SCHEDSTATS)
-
-extern void __sched_core_account_forceidle(struct rq *rq);
-
-static inline void sched_core_account_forceidle(struct rq *rq)
-{
-	if (schedstat_enabled())
-		__sched_core_account_forceidle(rq);
-}
-
-extern void __sched_core_tick(struct rq *rq);
-
-static inline void sched_core_tick(struct rq *rq)
-{
-	if (sched_core_enabled(rq) && schedstat_enabled())
-		__sched_core_tick(rq);
-}
-
-#else /* !(CONFIG_SCHED_CORE && CONFIG_SCHEDSTATS): */
-
-static inline void sched_core_account_forceidle(struct rq *rq) { }
-
-static inline void sched_core_tick(struct rq *rq) { }
-
-#endif /* !(CONFIG_SCHED_CORE && CONFIG_SCHEDSTATS) */
-
 #ifdef CONFIG_CGROUP_SCHED
 
 /*
@@ -2261,11 +2241,25 @@ static inline u64 global_rt_runtime(void)
 	return (u64)sysctl_sched_rt_runtime * NSEC_PER_USEC;
 }
 
+/*
+ * Is p the current execution context?
+ */
 static inline int task_current(struct rq *rq, struct task_struct *p)
 {
 	return rq->curr == p;
 }
 
+/*
+ * Is p the current scheduling context?
+ *
+ * Note that it might be the current execution context at the same time if
+ * rq->curr == rq->donor == p.
+ */
+static inline int task_current_donor(struct rq *rq, struct task_struct *p)
+{
+	return rq->donor == p;
+}
+
 static inline int task_on_cpu(struct rq *rq, struct task_struct *p)
 {
 #ifdef CONFIG_SMP
@@ -2452,7 +2446,7 @@ struct sched_class {
 
 static inline void put_prev_task(struct rq *rq, struct task_struct *prev)
 {
-	WARN_ON_ONCE(rq->curr != prev);
+	WARN_ON_ONCE(rq->donor != prev);
 	prev->sched_class->put_prev_task(rq, prev, NULL);
 }
 
@@ -2616,7 +2610,7 @@ static inline cpumask_t *alloc_user_cpus_ptr(int node)
 
 static inline struct task_struct *get_push_task(struct rq *rq)
 {
-	struct task_struct *p = rq->curr;
+	struct task_struct *p = rq->donor;
 
 	lockdep_assert_rq_held(rq);
 
@@ -2696,6 +2690,7 @@ extern void init_sched_rt_class(void);
 extern void init_sched_fair_class(void);
 
 extern void resched_curr(struct rq *rq);
+extern void resched_curr_lazy(struct rq *rq);
 extern void resched_cpu(int cpu);
 
 extern void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime);
@@ -3200,6 +3195,34 @@ extern void nohz_run_idle_balance(int cpu);
 static inline void nohz_run_idle_balance(int cpu) { }
 #endif
 
+#include "stats.h"
+
+#if defined(CONFIG_SCHED_CORE) && defined(CONFIG_SCHEDSTATS)
+
+extern void __sched_core_account_forceidle(struct rq *rq);
+
+static inline void sched_core_account_forceidle(struct rq *rq)
+{
+	if (schedstat_enabled())
+		__sched_core_account_forceidle(rq);
+}
+
+extern void __sched_core_tick(struct rq *rq);
+
+static inline void sched_core_tick(struct rq *rq)
+{
+	if (sched_core_enabled(rq) && schedstat_enabled())
+		__sched_core_tick(rq);
+}
+
+#else /* !(CONFIG_SCHED_CORE && CONFIG_SCHEDSTATS): */
+
+static inline void sched_core_account_forceidle(struct rq *rq) { }
+
+static inline void sched_core_tick(struct rq *rq) { }
+
+#endif /* !(CONFIG_SCHED_CORE && CONFIG_SCHEDSTATS) */
+
 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
 
 struct irqtime {
@@ -3630,24 +3653,41 @@ static inline void mm_cid_put(struct mm_struct *mm)
 	__mm_cid_put(mm, mm_cid_clear_lazy_put(cid));
 }
 
-static inline int __mm_cid_try_get(struct mm_struct *mm)
+static inline int __mm_cid_try_get(struct task_struct *t, struct mm_struct *mm)
 {
-	struct cpumask *cpumask;
+	struct cpumask *cidmask = mm_cidmask(mm);
-	int cid;
+	struct mm_cid __percpu *pcpu_cid = mm->pcpu_cid;
+	int cid = __this_cpu_read(pcpu_cid->recent_cid);
 
-	cpumask = mm_cidmask(mm);
+	/* Try to re-use recent cid. This improves cache locality. */
+	if (!mm_cid_is_unset(cid) && !cpumask_test_and_set_cpu(cid, cidmask))
+		return cid;
 	/*
+	 * Expand cid allocation if the maximum number of concurrency
+	 * IDs allocated (max_nr_cid) is below the number cpus allowed
+	 * and number of threads. Expanding cid allocation as much as
+	 * possible improves cache locality.
+	 */
+	cid = atomic_read(&mm->max_nr_cid);
+	while (cid < READ_ONCE(mm->nr_cpus_allowed) && cid < atomic_read(&mm->mm_users)) {
+		if (!atomic_try_cmpxchg(&mm->max_nr_cid, &cid, cid + 1))
+			continue;
+		if (!cpumask_test_and_set_cpu(cid, cidmask))
+			return cid;
+	}
+	/*
+	 * Find the first available concurrency id.
 	 * Retry finding first zero bit if the mask is temporarily
 	 * filled. This only happens during concurrent remote-clear
 	 * which owns a cid without holding a rq lock.
 	 */
 	for (;;) {
-		cid = cpumask_first_zero(cpumask);
+		cid = cpumask_first_zero(cidmask);
-		if (cid < nr_cpu_ids)
+		if (cid < READ_ONCE(mm->nr_cpus_allowed))
 			break;
 		cpu_relax();
 	}
-	if (cpumask_test_and_set_cpu(cid, cpumask))
+	if (cpumask_test_and_set_cpu(cid, cidmask))
 		return -1;
 
 	return cid;
@@ -3665,7 +3705,8 @@ static inline void mm_cid_snapshot_time(struct rq *rq, struct mm_struct *mm)
 	WRITE_ONCE(pcpu_cid->time, rq->clock);
 }
 
-static inline int __mm_cid_get(struct rq *rq, struct mm_struct *mm)
+static inline int __mm_cid_get(struct rq *rq, struct task_struct *t,
+			       struct mm_struct *mm)
 {
 	int cid;
 
@@ -3675,13 +3716,13 @@ static inline int __mm_cid_get(struct rq *rq, struct mm_struct *mm)
 	 * guarantee forward progress.
 	 */
 	if (!READ_ONCE(use_cid_lock)) {
-		cid = __mm_cid_try_get(mm);
+		cid = __mm_cid_try_get(t, mm);
 		if (cid >= 0)
 			goto end;
 		raw_spin_lock(&cid_lock);
 	} else {
 		raw_spin_lock(&cid_lock);
-		cid = __mm_cid_try_get(mm);
+		cid = __mm_cid_try_get(t, mm);
 		if (cid >= 0)
 			goto unlock;
 	}
@@ -3701,7 +3742,7 @@ static inline int __mm_cid_get(struct rq *rq, struct mm_struct *mm)
 	 * all newcoming allocations observe the use_cid_lock flag set.
 	 */
 	do {
-		cid = __mm_cid_try_get(mm);
+		cid = __mm_cid_try_get(t, mm);
 		cpu_relax();
 	} while (cid < 0);
 	/*
@@ -3718,7 +3759,8 @@ end:
 	return cid;
 }
 
-static inline int mm_cid_get(struct rq *rq, struct mm_struct *mm)
+static inline int mm_cid_get(struct rq *rq, struct task_struct *t,
+			     struct mm_struct *mm)
 {
 	struct mm_cid __percpu *pcpu_cid = mm->pcpu_cid;
 	struct cpumask *cpumask;
@@ -3735,8 +3777,9 @@ static inline int mm_cid_get(struct rq *rq, struct mm_struct *mm)
 		if (try_cmpxchg(&this_cpu_ptr(pcpu_cid)->cid, &cid, MM_CID_UNSET))
 			__mm_cid_put(mm, mm_cid_clear_lazy_put(cid));
 	}
-	cid = __mm_cid_get(rq, mm);
+	cid = __mm_cid_get(rq, t, mm);
 	__this_cpu_write(pcpu_cid->cid, cid);
+	__this_cpu_write(pcpu_cid->recent_cid, cid);
 
 	return cid;
 }
@@ -3789,7 +3832,7 @@ static inline void switch_mm_cid(struct rq *rq,
 		prev->mm_cid = -1;
 	}
 	if (next->mm_cid_active)
-		next->last_mm_cid = next->mm_cid = mm_cid_get(rq, next->mm);
+		next->last_mm_cid = next->mm_cid = mm_cid_get(rq, next, next->mm);
 }
 
 #else /* !CONFIG_SCHED_MM_CID: */
@@ -3802,6 +3845,28 @@ static inline void init_sched_mm_cid(struct task_struct *t) { }
 
 extern u64 avg_vruntime(struct cfs_rq *cfs_rq);
 extern int entity_eligible(struct cfs_rq *cfs_rq, struct sched_entity *se);
+#ifdef CONFIG_SMP
+static inline
+void move_queued_task_locked(struct rq *src_rq, struct rq *dst_rq, struct task_struct *task)
+{
+	lockdep_assert_rq_held(src_rq);
+	lockdep_assert_rq_held(dst_rq);
+
+	deactivate_task(src_rq, task, 0);
+	set_task_cpu(task, dst_rq->cpu);
+	activate_task(dst_rq, task, 0);
+}
+
+static inline
+bool task_is_pushable(struct rq *rq, struct task_struct *p, int cpu)
+{
+	if (!task_on_cpu(rq, p) &&
+	    cpumask_test_cpu(cpu, &p->cpus_mask))
+		return true;
+
+	return false;
+}
+#endif
 
 #ifdef CONFIG_RT_MUTEXES
 

kernel/sched/stats.h

@@ -127,21 +127,25 @@ static inline void psi_account_irqtime(struct rq *rq, struct task_struct *curr,
  * go through migration requeues. In this case, *sleeping* states need
  * to be transferred.
  */
-static inline void psi_enqueue(struct task_struct *p, bool migrate)
+static inline void psi_enqueue(struct task_struct *p, int flags)
 {
 	int clear = 0, set = 0;
 
 	if (static_branch_likely(&psi_disabled))
 		return;
 
+	/* Same runqueue, nothing changed for psi */
+	if (flags & ENQUEUE_RESTORE)
+		return;
+
 	if (p->se.sched_delayed) {
 		/* CPU migration of "sleeping" task */
-		SCHED_WARN_ON(!migrate);
+		SCHED_WARN_ON(!(flags & ENQUEUE_MIGRATED));
 		if (p->in_memstall)
 			set |= TSK_MEMSTALL;
 		if (p->in_iowait)
 			set |= TSK_IOWAIT;
-	} else if (migrate) {
+	} else if (flags & ENQUEUE_MIGRATED) {
 		/* CPU migration of runnable task */
 		set = TSK_RUNNING;
 		if (p->in_memstall)
@@ -158,17 +162,14 @@ static inline void psi_enqueue(struct task_struct *p, bool migrate)
 	psi_task_change(p, clear, set);
 }
 
-static inline void psi_dequeue(struct task_struct *p, bool migrate)
+static inline void psi_dequeue(struct task_struct *p, int flags)
 {
 	if (static_branch_likely(&psi_disabled))
 		return;
 
-	/*
+	/* Same runqueue, nothing changed for psi */
-	 * When migrating a task to another CPU, clear all psi
+	if (flags & DEQUEUE_SAVE)
-	 * state. The enqueue callback above will work it out.
+		return;
-	 */
-	if (migrate)
-		psi_task_change(p, p->psi_flags, 0);
 
 	/*
 	 * A voluntary sleep is a dequeue followed by a task switch. To
@@ -176,6 +177,14 @@ static inline void psi_dequeue(struct task_struct *p, bool migrate)
 	 * TSK_RUNNING and TSK_IOWAIT for us when it moves TSK_ONCPU.
 	 * Do nothing here.
 	 */
+	if (flags & DEQUEUE_SLEEP)
+		return;
+
+	/*
+	 * When migrating a task to another CPU, clear all psi
+	 * state. The enqueue callback above will work it out.
+	 */
+	psi_task_change(p, p->psi_flags, 0);
 }
 
 static inline void psi_ttwu_dequeue(struct task_struct *p)

kernel/sched/syscalls.c

@@ -91,7 +91,7 @@ void set_user_nice(struct task_struct *p, long nice)
 	}
 
 	queued = task_on_rq_queued(p);
-	running = task_current(rq, p);
+	running = task_current_donor(rq, p);
 	if (queued)
 		dequeue_task(rq, p, DEQUEUE_SAVE | DEQUEUE_NOCLOCK);
 	if (running)
@@ -713,7 +713,7 @@ change:
 		dequeue_task(rq, p, DEQUEUE_SLEEP | DEQUEUE_DELAYED | DEQUEUE_NOCLOCK);
 
 	queued = task_on_rq_queued(p);
-	running = task_current(rq, p);
+	running = task_current_donor(rq, p);
 	if (queued)
 		dequeue_task(rq, p, queue_flags);
 	if (running)

kernel/sched/wait_bit.c

@@ -9,7 +9,7 @@
 
 static wait_queue_head_t bit_wait_table[WAIT_TABLE_SIZE] __cacheline_aligned;
 
-wait_queue_head_t *bit_waitqueue(void *word, int bit)
+wait_queue_head_t *bit_waitqueue(unsigned long *word, int bit)
 {
 	const int shift = BITS_PER_LONG == 32 ? 5 : 6;
 	unsigned long val = (unsigned long)word << shift | bit;
@@ -55,7 +55,7 @@ __wait_on_bit(struct wait_queue_head *wq_head, struct wait_bit_queue_entry *wbq_
 }
 EXPORT_SYMBOL(__wait_on_bit);
 
-int __sched out_of_line_wait_on_bit(void *word, int bit,
+int __sched out_of_line_wait_on_bit(unsigned long *word, int bit,
 				    wait_bit_action_f *action, unsigned mode)
 {
 	struct wait_queue_head *wq_head = bit_waitqueue(word, bit);
@@ -66,7 +66,7 @@ int __sched out_of_line_wait_on_bit(void *word, int bit,
 EXPORT_SYMBOL(out_of_line_wait_on_bit);
 
 int __sched out_of_line_wait_on_bit_timeout(
-	void *word, int bit, wait_bit_action_f *action,
+	unsigned long *word, int bit, wait_bit_action_f *action,
 	unsigned mode, unsigned long timeout)
 {
 	struct wait_queue_head *wq_head = bit_waitqueue(word, bit);
@@ -108,7 +108,7 @@ __wait_on_bit_lock(struct wait_queue_head *wq_head, struct wait_bit_queue_entry
 }
 EXPORT_SYMBOL(__wait_on_bit_lock);
 
-int __sched out_of_line_wait_on_bit_lock(void *word, int bit,
+int __sched out_of_line_wait_on_bit_lock(unsigned long *word, int bit,
 					 wait_bit_action_f *action, unsigned mode)
 {
 	struct wait_queue_head *wq_head = bit_waitqueue(word, bit);
@@ -118,7 +118,7 @@ int __sched out_of_line_wait_on_bit_lock(void *word, int bit,
 }
 EXPORT_SYMBOL(out_of_line_wait_on_bit_lock);
 
-void __wake_up_bit(struct wait_queue_head *wq_head, void *word, int bit)
+void __wake_up_bit(struct wait_queue_head *wq_head, unsigned long *word, int bit)
 {
 	struct wait_bit_key key = __WAIT_BIT_KEY_INITIALIZER(word, bit);
 
@@ -128,23 +128,31 @@ void __wake_up_bit(struct wait_queue_head *wq_head, void *word, int bit)
 EXPORT_SYMBOL(__wake_up_bit);
 
 /**
- * wake_up_bit - wake up a waiter on a bit
+ * wake_up_bit - wake up waiters on a bit
- * @word: the word being waited on, a kernel virtual address
+ * @word: the address containing the bit being waited on
- * @bit: the bit of the word being waited on
+ * @bit: the bit at that address being waited on
  *
- * There is a standard hashed waitqueue table for generic use. This
+ * Wake up any process waiting in wait_on_bit() or similar for the
- * is the part of the hash-table's accessor API that wakes up waiters
+ * given bit to be cleared.
- * on a bit. For instance, if one were to have waiters on a bitflag,
- * one would call wake_up_bit() after clearing the bit.
  *
- * In order for this to function properly, as it uses waitqueue_active()
+ * The wake-up is sent to tasks in a waitqueue selected by hash from a
- * internally, some kind of memory barrier must be done prior to calling
+ * shared pool.  Only those tasks on that queue which have requested
- * this. Typically, this will be smp_mb__after_atomic(), but in some
+ * wake_up on this specific address and bit will be woken, and only if the
- * cases where bitflags are manipulated non-atomically under a lock, one
+ * bit is clear.
- * may need to use a less regular barrier, such fs/inode.c's smp_mb(),
+ *
- * because spin_unlock() does not guarantee a memory barrier.
+ * In order for this to function properly there must be a full memory
+ * barrier after the bit is cleared and before this function is called.
+ * If the bit was cleared atomically, such as a by clear_bit() then
+ * smb_mb__after_atomic() can be used, othwewise smb_mb() is needed.
+ * If the bit was cleared with a fully-ordered operation, no further
+ * barrier is required.
+ *
+ * Normally the bit should be cleared by an operation with RELEASE
+ * semantics so that any changes to memory made before the bit is
+ * cleared are guaranteed to be visible after the matching wait_on_bit()
+ * completes.
  */
-void wake_up_bit(void *word, int bit)
+void wake_up_bit(unsigned long *word, int bit)
 {
 	__wake_up_bit(bit_waitqueue(word, bit), word, bit);
 }
@@ -188,6 +196,36 @@ void init_wait_var_entry(struct wait_bit_queue_entry *wbq_entry, void *var, int
 }
 EXPORT_SYMBOL(init_wait_var_entry);
 
+/**
+ * wake_up_var - wake up waiters on a variable (kernel address)
+ * @var: the address of the variable being waited on
+ *
+ * Wake up any process waiting in wait_var_event() or similar for the
+ * given variable to change.  wait_var_event() can be waiting for an
+ * arbitrary condition to be true and associates that condition with an
+ * address.  Calling wake_up_var() suggests that the condition has been
+ * made true, but does not strictly require the condtion to use the
+ * address given.
+ *
+ * The wake-up is sent to tasks in a waitqueue selected by hash from a
+ * shared pool.  Only those tasks on that queue which have requested
+ * wake_up on this specific address will be woken.
+ *
+ * In order for this to function properly there must be a full memory
+ * barrier after the variable is updated (or more accurately, after the
+ * condition waited on has been made to be true) and before this function
+ * is called.  If the variable was updated atomically, such as a by
+ * atomic_dec() then smb_mb__after_atomic() can be used.  If the
+ * variable was updated by a fully ordered operation such as
+ * atomic_dec_and_test() then no extra barrier is required.  Otherwise
+ * smb_mb() is needed.
+ *
+ * Normally the variable should be updated (the condition should be made
+ * to be true) by an operation with RELEASE semantics such as
+ * smp_store_release() so that any changes to memory made before the
+ * variable was updated are guaranteed to be visible after the matching
+ * wait_var_event() completes.
+ */
 void wake_up_var(void *var)
 {
 	__wake_up_bit(__var_waitqueue(var), var, -1);
@@ -228,20 +266,6 @@ __sched int bit_wait_timeout(struct wait_bit_key *word, int mode)
 }
 EXPORT_SYMBOL_GPL(bit_wait_timeout);
 
-__sched int bit_wait_io_timeout(struct wait_bit_key *word, int mode)
-{
-	unsigned long now = READ_ONCE(jiffies);
-
-	if (time_after_eq(now, word->timeout))
-		return -EAGAIN;
-	io_schedule_timeout(word->timeout - now);
-	if (signal_pending_state(mode, current))
-		return -EINTR;
-
-	return 0;
-}
-EXPORT_SYMBOL_GPL(bit_wait_io_timeout);
-
 void __init wait_bit_init(void)
 {
 	int i;

kernel/softirq.c

@@ -748,10 +748,8 @@ EXPORT_SYMBOL(__tasklet_hi_schedule);
 
 static bool tasklet_clear_sched(struct tasklet_struct *t)
 {
-	if (test_and_clear_bit(TASKLET_STATE_SCHED, &t->state)) {
+	if (test_and_clear_wake_up_bit(TASKLET_STATE_SCHED, &t->state))
-		wake_up_var(&t->state);
 		return true;
-	}
 
 	WARN_ONCE(1, "tasklet SCHED state not set: %s %pS\n",
 		  t->use_callback ? "callback" : "func",
@@ -871,8 +869,7 @@ void tasklet_kill(struct tasklet_struct *t)
 	if (in_interrupt())
 		pr_notice("Attempt to kill tasklet from interrupt\n");
 
-	while (test_and_set_bit(TASKLET_STATE_SCHED, &t->state))
+	wait_on_bit_lock(&t->state, TASKLET_STATE_SCHED, TASK_UNINTERRUPTIBLE);
-		wait_var_event(&t->state, !test_bit(TASKLET_STATE_SCHED, &t->state));
 
 	tasklet_unlock_wait(t);
 	tasklet_clear_sched(t);
@@ -882,16 +879,13 @@ EXPORT_SYMBOL(tasklet_kill);
 #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT_RT)
 void tasklet_unlock(struct tasklet_struct *t)
 {
-	smp_mb__before_atomic();
+	clear_and_wake_up_bit(TASKLET_STATE_RUN, &t->state);
-	clear_bit(TASKLET_STATE_RUN, &t->state);
-	smp_mb__after_atomic();
-	wake_up_var(&t->state);
 }
 EXPORT_SYMBOL_GPL(tasklet_unlock);
 
 void tasklet_unlock_wait(struct tasklet_struct *t)
 {
-	wait_var_event(&t->state, !test_bit(TASKLET_STATE_RUN, &t->state));
+	wait_on_bit(&t->state, TASKLET_STATE_RUN, TASK_UNINTERRUPTIBLE);
 }
 EXPORT_SYMBOL_GPL(tasklet_unlock_wait);
 #endif