Linux 5.10.78-rt56-rc3

Dear RT Folks,

This is the RT stable review cycle of patch 5.10.78-rt56-rc3.

Please scream at me if I messed something up. Please test the patches too.

The -rc release will be uploaded to kernel.org and will be deleted when the final release is out. This is just a review release (or release candidate).

The pre-releases will not be pushed to the git repository, only the final release is.

If all goes well, this patch will be converted to the next main release on 11/26/2021.

Enjoy,

-- Steve

To build 5.10.78-rt56-rc3 directly, the following patches should be applied:

http://www.kernel.org/pub/linux/kernel/v5.x/linux-5.10.tar.xz

http://www.kernel.org/pub/linux/kernel/v5.x/patch-5.10.78.xz

http://www.kernel.org/pub/linux/kernel/projects/rt/5.10/patch-5.10.78-rt56-rc3.patch.xz

You can also build from 5.10.78-rt55 by applying the incremental patch:

http://www.kernel.org/pub/linux/kernel/projects/rt/5.10/incr/patch-5.10.78-rt55-rt56-rc3.patch.xz

Changes from 5.10.78-rt55:

---

Mike Galbraith (1):
mm, zsmalloc: Convert zsmalloc_handle.lock to spinlock_t

Sebastian Andrzej Siewior (11):
sched: Fix get_push_task() vs migrate_disable()
sched: Switch wait_task_inactive to HRTIMER_MODE_REL_HARD
preempt: Move preempt_enable_no_resched() to the RT block
mm: Disable NUMA_BALANCING_DEFAULT_ENABLED and TRANSPARENT_HUGEPAGE on PREEMPT_RT
fscache: Use only one fscache_object_cong_wait.
fscache: Use only one fscache_object_cong_wait.
locking: Drop might_resched() from might_sleep_no_state_check() drm/i915/gt: Queue and wait for the irq_work item.
irq_work: Allow irq_work_sync() to sleep if irq_work() no IRQ support.
irq_work: Handle some irq_work in a per-CPU thread on PREEMPT_RT
irq_work: Also rcuwait for !IRQ_WORK_HARD_IRQ on PREEMPT_RT

Steven Rostedt (VMware) (1):

      Linux 5.10.78-rt56-rc3

----

 drivers/gpu/drm/i915/gt/intel_breadcrumbs.c |   5 +-
 fs/fscache/internal.h                       |   1 -
 fs/fscache/main.c                           |   6 --
 fs/fscache/object.c                         |  13 +--
 include/linux/irq_work.h                    |  31 ++++--
 include/linux/kernel.h                      |   2 +-
 include/linux/preempt.h                     |   6 +-
 init/Kconfig                                |   2 +-
 kernel/irq_work.c                           | 143 +++++++++++++++++++++-------
 kernel/sched/core.c                         |   2 +-
 kernel/sched/sched.h                        |   3 +
 kernel/time/timer.c                         |   2 -
 localversion-rt                             |   2 +-
 mm/zsmalloc.c                               |  12 +--
 14 files changed, 155 insertions(+), 75 deletions(-)

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Mike Galbraith <[email protected]>

local_lock_t becoming a synonym of spinlock_t had consequences for the RT mods to zsmalloc, which were taking a mutex while holding a local_lock, inspiring a lockdep "BUG: Invalid wait context" gripe.

Converting zsmalloc_handle.lock to a spinlock_t restored lockdep silence.

---
 mm/zsmalloc.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

index 277d426c881f..3595c1644135 100644
@@ -82,7 +82,7 @@
 
 struct zsmalloc_handle {
 	unsigned long addr;
-	struct mutex lock;
+	spinlock_t lock;
 };
 
 #define ZS_HANDLE_ALLOC_SIZE (sizeof(struct zsmalloc_handle))

@@ -370,7 +370,7 @@ static unsigned long cache_alloc_handle(struct zs_pool *pool, gfp_t gfp)
 	if (p) {
 		struct zsmalloc_handle *zh = p;
 
-		mutex_init(&zh->lock);
+		spin_lock_init(&zh->lock);
 	}
 #endif
 	return (unsigned long)p;

@@ -930,7 +930,7 @@ static inline int testpin_tag(unsigned long handle)
 #ifdef CONFIG_PREEMPT_RT
 	struct zsmalloc_handle *zh = zs_get_pure_handle(handle);
 
-	return mutex_is_locked(&zh->lock);
+	return spin_is_locked(&zh->lock);
 #else
 	return bit_spin_is_locked(HANDLE_PIN_BIT, (unsigned long *)handle);
 #endif

@@ -941,7 +941,7 @@ static inline int trypin_tag(unsigned long handle)
 #ifdef CONFIG_PREEMPT_RT
 	struct zsmalloc_handle *zh = zs_get_pure_handle(handle);
 
-	return mutex_trylock(&zh->lock);
+	return spin_trylock(&zh->lock);
 #else
 	return bit_spin_trylock(HANDLE_PIN_BIT, (unsigned long *)handle);
 #endif

@@ -952,7 +952,7 @@ static void pin_tag(unsigned long handle) __acquires(bitlock)
 #ifdef CONFIG_PREEMPT_RT
 	struct zsmalloc_handle *zh = zs_get_pure_handle(handle);
 
-	return mutex_lock(&zh->lock);
+	return spin_lock(&zh->lock);
 #else
 	bit_spin_lock(HANDLE_PIN_BIT, (unsigned long *)handle);
 #endif

@@ -963,7 +963,7 @@ static void unpin_tag(unsigned long handle) __releases(bitlock)
 #ifdef CONFIG_PREEMPT_RT
 	struct zsmalloc_handle *zh = zs_get_pure_handle(handle);
 
-	return mutex_unlock(&zh->lock);
+	return spin_unlock(&zh->lock);
 #else
 	bit_spin_unlock(HANDLE_PIN_BIT, (unsigned long *)handle);
 #endif

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

push_rt_task() attempts to move the currently running task away if the next runnable task has migration disabled and therefore is pinned on the current CPU.

The current task is retrieved via get_push_task() which only checks for nr_cpus_allowed == 1, but does not check whether the task has migration disabled and therefore cannot be moved either. The consequence is a pointless invocation of the migration thread which correctly observes that the task cannot be moved.

Return NULL if the task has migration disabled and cannot be moved to another CPU.

---
 kernel/sched/sched.h | 3 +++
 1 file changed, 3 insertions(+)

index 826ea17e144d..c2c9c386456d 100644
@@ -1949,6 +1949,9 @@ static inline struct task_struct *get_push_task(struct rq *rq)
 	if (p->nr_cpus_allowed == 1)
 		return NULL;
 
+	if (p->migration_disabled)
+		return NULL;
+
 	rq->push_busy = true;
 	return get_task_struct(p);
 }

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

With PREEMPT_RT enabled all hrtimers callbacks will be invoked in
softirq mode unless they are explicitly marked as HRTIMER_MODE_HARD.
During boot kthread_bind() is used for the creation of per-CPU threads
and then hangs in wait_task_inactive() if the ksoftirqd is not
yet up and running.
The hang disappeared since commit
26c7295be0c5e ("kthread: Do not preempt current task if it is going to call schedule()")

but enabling function trace on boot reliably leads to the freeze on boot behaviour again.
The timer in wait_task_inactive() can not be directly used by an user interface to abuse it and create a mass wake of several tasks at the same time which would to long sections with disabled interrupts. Therefore it is safe to make the timer HRTIMER_MODE_REL_HARD.

Switch the timer to HRTIMER_MODE_REL_HARD.

---
 kernel/sched/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index f638d9420553..54fa3bb1b7c4 100644
@@ -2734,7 +2734,7 @@ unsigned long wait_task_inactive(struct task_struct *p, long match_state)
 			ktime_t to = NSEC_PER_SEC / HZ;
 
 			set_current_state(TASK_UNINTERRUPTIBLE);
-			schedule_hrtimeout(&to, HRTIMER_MODE_REL);
+			schedule_hrtimeout(&to, HRTIMER_MODE_REL_HARD);
 			continue;
 		}
 

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

preempt_enable_no_resched() should point to preempt_enable() on PREEMPT_RT so nobody is playing any preempt tricks and enables preemption without checking for the need-resched flag.

This was misplaced in v3.14.0-rt1 und remained unnoticed until now.

Point preempt_enable_no_resched() and preempt_enable() on RT.

---
 include/linux/preempt.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

index af39859f02ee..7b5b2ed55531 100644
@@ -208,12 +208,12 @@ do { \
 	preempt_count_dec(); \
 } while (0)
 
-#ifdef CONFIG_PREEMPT_RT
+#ifndef CONFIG_PREEMPT_RT
 # define preempt_enable_no_resched() sched_preempt_enable_no_resched()
-# define preempt_check_resched_rt() preempt_check_resched()
+# define preempt_check_resched_rt() barrier();
 #else
 # define preempt_enable_no_resched() preempt_enable()
-# define preempt_check_resched_rt() barrier();
+# define preempt_check_resched_rt() preempt_check_resched()
 #endif
 
 #define preemptible()	(preempt_count() == 0 && !irqs_disabled())

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

TRANSPARENT_HUGEPAGE:
There are potential non-deterministic delays to an RT thread if a critical memory region is not THP-aligned and a non-RT buffer is located in the same hugepage-aligned region. It's also possible for an unrelated thread to migrate pages belonging to an RT task incurring unexpected page faults due to memory defragmentation even if khugepaged is disabled.

Regular HUGEPAGEs are not affected by this can be used.

NUMA_BALANCING:
There is a non-deterministic delay to mark PTEs PROT_NONE to gather NUMA fault samples, increased page faults of regions even if mlocked and non-deterministic delays when migrating pages.

[Mel Gorman worded 99% of the commit description].

---
 init/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 7ba2b602b707..9bfc60e7eead 100644
@@ -861,7 +861,7 @@ config NUMA_BALANCING
 	bool "Memory placement aware NUMA scheduler"
 	depends on ARCH_SUPPORTS_NUMA_BALANCING
 	depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
-	depends on SMP && NUMA && MIGRATION
+	depends on SMP && NUMA && MIGRATION && !PREEMPT_RT
 	help
 	  This option adds support for automatic NUMA aware memory/task placement.
 	  The mechanism is quite primitive and is based on migrating memory when

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

In the commit mentioned below, fscache was converted from slow-work to workqueue. slow_work_enqueue() and slow_work_sleep_till_thread_needed() did not use a per-CPU workqueue. They choose from two global waitqueues depending on the SLOW_WORK_VERY_SLOW bit which was not set so it always one waitqueue.

I can't find out how it is ensured that a waiter on certain CPU is woken up be the other side. My guess is that the timeout in schedule_timeout() ensures that it does not wait forever (or a random wake up).

fscache_object_sleep_till_congested() must be invoked from preemptible context in order for schedule() to work. In this case this_cpu_ptr() should complain with CONFIG_DEBUG_PREEMPT enabled except the thread is bound to one CPU.

wake_up() wakes only one waiter and I'm not sure if it is guaranteed that only one waiter exists.

Replace the per-CPU waitqueue with one global waitqueue.

---
 fs/fscache/internal.h |  1 -
 fs/fscache/main.c     |  6 ------
 fs/fscache/object.c   | 11 +++++------
 3 files changed, 5 insertions(+), 13 deletions(-)

index 64aa552b296d..7dae569dafb9 100644
@@ -95,7 +95,6 @@ extern unsigned fscache_debug;
 extern struct kobject *fscache_root;
 extern struct workqueue_struct *fscache_object_wq;
 extern struct workqueue_struct *fscache_op_wq;
-DECLARE_PER_CPU(wait_queue_head_t, fscache_object_cong_wait);
 
 extern unsigned int fscache_hash(unsigned int salt, unsigned int *data, unsigned int n);
 

index 4207f98e405f..85f8cf3a323d 100644
@@ -41,8 +41,6 @@ struct kobject *fscache_root;
 struct workqueue_struct *fscache_object_wq;
 struct workqueue_struct *fscache_op_wq;
 
-DEFINE_PER_CPU(wait_queue_head_t, fscache_object_cong_wait);
-
 /* these values serve as lower bounds, will be adjusted in fscache_init() */
 static unsigned fscache_object_max_active = 4;
 static unsigned fscache_op_max_active = 2;

@@ -138,7 +136,6 @@ unsigned int fscache_hash(unsigned int salt, unsigned int *data, unsigned int n)
 static int __init fscache_init(void)
 {
 	unsigned int nr_cpus = num_possible_cpus();
-	unsigned int cpu;
 	int ret;
 
 	fscache_object_max_active =

@@ -161,9 +158,6 @@ static int __init fscache_init(void)
 	if (!fscache_op_wq)
 		goto error_op_wq;
 
-	for_each_possible_cpu(cpu)
-		init_waitqueue_head(&per_cpu(fscache_object_cong_wait, cpu));
-
 	ret = fscache_proc_init();
 	if (ret < 0)
 		goto error_proc;

index cb2146e02cd5..55158f30d093 100644
@@ -807,6 +807,8 @@ void fscache_object_destroy(struct fscache_object *object)
 }
 EXPORT_SYMBOL(fscache_object_destroy);
 
+static DECLARE_WAIT_QUEUE_HEAD(fscache_object_cong_wait);
+
 /*
  * enqueue an object for metadata-type processing
  */

@@ -815,12 +817,10 @@ void fscache_enqueue_object(struct fscache_object *object)
 	_enter("{OBJ%x}", object->debug_id);
 
 	if (fscache_get_object(object, fscache_obj_get_queue) >= 0) {
-		wait_queue_head_t *cong_wq =
-			&get_cpu_var(fscache_object_cong_wait);
 
 		if (queue_work(fscache_object_wq, &object->work)) {
 			if (fscache_object_congested())
-				wake_up(cong_wq);
+				wake_up(&fscache_object_cong_wait);
 		} else
 			fscache_put_object(object, fscache_obj_put_queue);
 

@@ -842,16 +842,15 @@ void fscache_enqueue_object(struct fscache_object *object)
  */
 bool fscache_object_sleep_till_congested(signed long *timeoutp)
 {
-	wait_queue_head_t *cong_wq = this_cpu_ptr(&fscache_object_cong_wait);
 	DEFINE_WAIT(wait);
 
 	if (fscache_object_congested())
 		return true;
 
-	add_wait_queue_exclusive(cong_wq, &wait);
+	add_wait_queue_exclusive(&fscache_object_cong_wait, &wait);
 	if (!fscache_object_congested())
 		*timeoutp = schedule_timeout(*timeoutp);
-	finish_wait(cong_wq, &wait);
+	finish_wait(&fscache_object_cong_wait, &wait);
 
 	return fscache_object_congested();
 }

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

This is an update of the original patch, removing put_cpu_var() which was overseen in the initial patch.

---
 fs/fscache/object.c | 2 --
 1 file changed, 2 deletions(-)

index 55158f30d093..fb9794dce721 100644
@@ -823,8 +823,6 @@ void fscache_enqueue_object(struct fscache_object *object)
 				wake_up(&fscache_object_cong_wait);
 		} else
 			fscache_put_object(object, fscache_obj_put_queue);
-
-		put_cpu_var(fscache_object_cong_wait);
 	}
 }
 

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

might_sleep_no_state_check() serves the same purpose as might_sleep() except it is used before sleeping locks are acquired and therefore does not check task_struct::state because the state is preserved.

That state is preserved in the locking slow path so we must not schedule at the begin of the locking function because the state will be lost and not preserved at that time.

Remove might_resched() from might_sleep_no_state_check() to avoid losing the state before it is preserved.

---
 include/linux/kernel.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 2cff7554395d..6eb0ab994f4c 100644
@@ -222,7 +222,7 @@ extern void __cant_migrate(const char *file, int line);
 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
 
 # define might_sleep_no_state_check() \
-	do { ___might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
+	do { ___might_sleep(__FILE__, __LINE__, 0); } while (0)
 
 /**
  * cant_sleep - annotation for functions that cannot sleep

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

Disabling interrupts and invoking the irq_work function directly breaks on PREEMPT_RT.
PREEMPT_RT does not invoke all irq_work from hardirq context because some of the user have spinlock_t locking in the callback function. These locks are then turned into a sleeping locks which can not be acquired with disabled interrupts.

Using irq_work_queue() has the benefit that the irqwork will be invoked in the regular context. In general there is "no" delay between enqueuing the callback and its invocation because the interrupt is raised right away on architectures which support it (which includes x86).

Use irq_work_queue() + irq_work_sync() instead invoking the callback directly.

---
 drivers/gpu/drm/i915/gt/intel_breadcrumbs.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

index 0040b4765a54..3f4f854786f2 100644
@@ -342,10 +342,9 @@ void intel_breadcrumbs_park(struct intel_breadcrumbs *b)
 	/* Kick the work once more to drain the signalers */
 	irq_work_sync(&b->irq_work);
 	while (unlikely(READ_ONCE(b->irq_armed))) {
-		local_irq_disable();
-		signal_irq_work(&b->irq_work);
-		local_irq_enable();
+		irq_work_queue(&b->irq_work);
 		cond_resched();
+		irq_work_sync(&b->irq_work);
 	}
 	GEM_BUG_ON(!list_empty(&b->signalers));
 }

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

irq_work() triggers instantly an interrupt if supported by the architecture. Otherwise the work will be processed on the next timer tick. In worst case irq_work_sync() could spin up to a jiffy.

irq_work_sync() is usually used in tear down context which is fully preemptible. Based on review irq_work_sync() is invoked from preemptible context and there is one waiter at a time. This qualifies it to use rcuwait for synchronisation.

Let irq_work_sync() synchronize with rcuwait if the architecture processes irqwork via the timer tick.

---
 include/linux/irq_work.h | 10 +++++++++-
 kernel/irq_work.c        | 10 ++++++++++
 2 files changed, 19 insertions(+), 1 deletion(-)

index f941f2d7d71c..3c6d3a96bca0 100644
@@ -3,6 +3,7 @@
 #define _LINUX_IRQ_WORK_H
 
 #include <linux/smp_types.h>
+#include <linux/rcuwait.h>
 
 /*
  * An entry can be in one of four states:

@@ -22,6 +23,7 @@ struct irq_work {
 		};
 	};
 	void (*func)(struct irq_work *);
+	struct rcuwait irqwait;
 };
 
 static inline

@@ -29,13 +31,19 @@ void init_irq_work(struct irq_work *work, void (*func)(struct irq_work *))
 {
 	atomic_set(&work->flags, 0);
 	work->func = func;
+	rcuwait_init(&work->irqwait);
 }
 
 #define DEFINE_IRQ_WORK(name, _f) struct irq_work name = {	\
 		.flags = ATOMIC_INIT(0),			\
-		.func  = (_f)					\
+		.func  = (_f),					\
+		.irqwait = __RCUWAIT_INITIALIZER(irqwait),	\
 }
 
+static inline bool irq_work_is_busy(struct irq_work *work)
+{
+	return atomic_read(&work->flags) & IRQ_WORK_BUSY;
+}
 
 bool irq_work_queue(struct irq_work *work);
 bool irq_work_queue_on(struct irq_work *work, int cpu);

index 8183d30e1bb1..8969aff790e2 100644
@@ -165,6 +165,9 @@ void irq_work_single(void *arg)
 	 */
 	flags &= ~IRQ_WORK_PENDING;
 	(void)atomic_cmpxchg(&work->flags, flags, flags & ~IRQ_WORK_BUSY);
+
+	if (!arch_irq_work_has_interrupt())
+		rcuwait_wake_up(&work->irqwait);
 }
 
 static void irq_work_run_list(struct llist_head *list)

@@ -231,6 +234,13 @@ void irq_work_tick_soft(void)
 void irq_work_sync(struct irq_work *work)
 {
 	lockdep_assert_irqs_enabled();
+	might_sleep();
+
+	if (!arch_irq_work_has_interrupt()) {
+		rcuwait_wait_event(&work->irqwait, !irq_work_is_busy(work),
+				   TASK_UNINTERRUPTIBLE);
+		return;
+	}
 
 	while (atomic_read(&work->flags) & IRQ_WORK_BUSY)
 		cpu_relax();

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

The irq_work callback is invoked in hard IRQ context. By default all callbacks are scheduled for invocation right away (given supported by the architecture) except for the ones marked IRQ_WORK_LAZY which are delayed until the next timer-tick.

While looking over the callbacks, some of them may acquire locks (spinlock_t, rwlock_t) which are transformed into sleeping locks on PREEMPT_RT and must not be acquired in hard IRQ context. Changing the locks into locks which could be acquired in this context will lead to other problems such as increased latencies if everything in the chain has IRQ-off locks. This will not solve all the issues as one callback has been noticed which invoked kref_put() and its callback invokes kfree() and this can not be invoked in hardirq context.

Some callbacks are required to be invoked in hardirq context even on PREEMPT_RT to work properly. This includes for instance the NO_HZ callback which needs to be able to observe the idle context.

The callbacks which require to be run in hardirq have already been marked. Use this information to split the callbacks onto the two lists on PREEMPT_RT:
- lazy_list
Work items which are not marked with IRQ_WORK_HARD_IRQ will be added to this list. Callbacks on this list will be invoked from a per-CPU thread.
The handler here may acquire sleeping locks such as spinlock_t and invoke kfree().

- raised_list
Work items which are marked with IRQ_WORK_HARD_IRQ will be added to this list. They will be invoked in hardirq context and must not acquire any sleeping locks.

The wake up of the per-CPU thread occurs from irq_work handler/ hardirq context. The thread runs with lowest RT priority to ensure it runs before any SCHED_OTHER tasks do.

[bigeasy: melt tglx's irq_work_tick_soft() which splits irq_work_tick() into a
hard and soft variant. Collected fixes over time from Steven Rostedt and Mike Galbraith. Move to per-CPU threads instead of softirq as suggested by PeterZ.]

---
 include/linux/irq_work.h |  16 +++--
 kernel/irq_work.c        | 131 ++++++++++++++++++++++++++++-----------
 kernel/time/timer.c      |   2 -
 3 files changed, 106 insertions(+), 43 deletions(-)

index 3c6d3a96bca0..f551ba9c99d4 100644
@@ -40,6 +40,16 @@ void init_irq_work(struct irq_work *work, void (*func)(struct irq_work *))
 		.irqwait = __RCUWAIT_INITIALIZER(irqwait),	\
 }
 
+#define __IRQ_WORK_INIT(_func, _flags) (struct irq_work){	\
+	.flags = ATOMIC_INIT(_flags),				\
+	.func = (_func),					\
+	.irqwait = __RCUWAIT_INITIALIZER(irqwait),		\
+}
+
+#define IRQ_WORK_INIT(_func) __IRQ_WORK_INIT(_func, 0)
+#define IRQ_WORK_INIT_LAZY(_func) __IRQ_WORK_INIT(_func, IRQ_WORK_LAZY)
+#define IRQ_WORK_INIT_HARD(_func) __IRQ_WORK_INIT(_func, IRQ_WORK_HARD_IRQ)
+
 static inline bool irq_work_is_busy(struct irq_work *work)
 {
 	return atomic_read(&work->flags) & IRQ_WORK_BUSY;

@@ -63,10 +73,4 @@ static inline void irq_work_run(void) { }
 static inline void irq_work_single(void *arg) { }
 #endif
 
-#if defined(CONFIG_IRQ_WORK) && defined(CONFIG_PREEMPT_RT)
-void irq_work_tick_soft(void);
-#else
-static inline void irq_work_tick_soft(void) { }
-#endif
-
 #endif /* _LINUX_IRQ_WORK_H */

index af3daf03c917..cd67ee6d2634 100644
@@ -1767,8 +1767,6 @@ static __latent_entropy void run_timer_softirq(struct softirq_action *h)
 {
 	struct timer_base *base = this_cpu_ptr(&timer_bases[BASE_STD]);
 
-	irq_work_tick_soft();
-
 	__run_timers(base);
 	if (IS_ENABLED(CONFIG_NO_HZ_COMMON))
 		__run_timers(this_cpu_ptr(&timer_bases[BASE_DEF]));

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

On PREEMPT_RT most items are processed as LAZY via softirq context. Avoid to spin-wait for them because irq_work_sync() could have higher priority and not allow the irq-work to be completed.

Wait additionally for !IRQ_WORK_HARD_IRQ irq_work items on PREEMPT_RT.

---
 include/linux/irq_work.h | 5 +++++
 kernel/irq_work.c        | 6 ++++--
 2 files changed, 9 insertions(+), 2 deletions(-)

index f551ba9c99d4..2c0059340871 100644
@@ -55,6 +55,11 @@ static inline bool irq_work_is_busy(struct irq_work *work)
 	return atomic_read(&work->flags) & IRQ_WORK_BUSY;
 }
 
+static inline bool irq_work_is_hard(struct irq_work *work)
+{
+	return atomic_read(&work->flags) & IRQ_WORK_HARD_IRQ;
+}
+
 bool irq_work_queue(struct irq_work *work);
 bool irq_work_queue_on(struct irq_work *work, int cpu);
 

index 03d09d779ee1..cbec10c32ead 100644
@@ -211,7 +211,8 @@ void irq_work_single(void *arg)
 	flags &= ~IRQ_WORK_PENDING;
 	(void)atomic_cmpxchg(&work->flags, flags, flags & ~IRQ_WORK_BUSY);
 
-	if (!arch_irq_work_has_interrupt())
+	if ((IS_ENABLED(CONFIG_PREEMPT_RT) && !irq_work_is_hard(work)) ||
+	    !arch_irq_work_has_interrupt())
 		rcuwait_wake_up(&work->irqwait);
 }
 

@@ -271,7 +272,8 @@ void irq_work_sync(struct irq_work *work)
 	lockdep_assert_irqs_enabled();
 	might_sleep();
 
-	if (!arch_irq_work_has_interrupt()) {
+	if ((IS_ENABLED(CONFIG_PREEMPT_RT) && !irq_work_is_hard(work)) ||
+	    !arch_irq_work_has_interrupt()) {
 		rcuwait_wait_event(&work->irqwait, !irq_work_is_busy(work),
 				   TASK_UNINTERRUPTIBLE);
 		return;

2.33.0

5.10.78-rt56-rc3 stable review patch.
If anyone has any objections, please let me know.

------------------

From: "Steven Rostedt (VMware)" <[email protected]>

---
 localversion-rt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 51b05e9abe6f..dfd6a5d63829 100644
@@ -1 +1 @@
--rt55
+-rt56-rc3

2.33.0