[Intel-gfx] [PATCH v3] softirq: Prevent looping on disabled tasklets

[Chris Wilson]

Disabling a tasklet causes it not to run during tasklet_action, but is
put back onto the runnable tasklet list, and a new softirq raised. As
the softirq is raised from within __do_softirq() this causing
__do_softirq() to loop constantly until its timeslice expires and is
transferred to the ksoftirq thread. ksoftirq then permanently spins,
as on each action, the disabled tasklet keeps reraising the softirq.

Break this vicious cycle by moving the softirq from the action to the
final tasklet_enable().

This behaviour appears to be historic (since the first git import).
However, the looping until timeslice duration (to a max of 2ms) was
first introduced in commit c10d73671ad3 ("softirq: reduce latencies"),
with the restart limit restored in commit 34376a50fb1f ("Fix lockup
related to stop_machine being stuck in __do_softirq.")

v2: Export tasklet_enable() to work with modules.
v3: Restore the looping over a failed tasklet_trylock()

Reported-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
Cc: Thomas Gleixner <tglx at linutronix.de>
Cc: Hannes Reinecke <hare at suse.com>
Cc: Jens Axboe <axboe at kernel.dk>
Cc: Bjorn Helgaas <bhelgaas at google.com>
Cc: Alexander Potapenko <glider at google.com>
Cc: Chen Fan <chen.fan.fnst at cn.fujitsu.com>
Cc: Ingo Molnar <mingo at kernel.org>
Cc: "Peter Zijlstra (Intel)" <peterz at infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy at linutronix.de>
Cc: Johannes Thumshirn <jthumshirn at suse.de>
Cc: Emese Revfy <re.emese at gmail.com>
Cc: Sagi Grimberg <sagi at grimberg.me>
Cc: Eric Dumazet <edumazet at google.com>
Cc: Tom Herbert <therbert at google.com>
Cc: Ben Hutchings <bhutchings at solarflare.com>
---
 include/linux/interrupt.h |  7 +------
 kernel/softirq.c          | 20 ++++++++++++++++++--
 2 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 53144e78a369..a1fa88e7e509 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -611,12 +611,7 @@ static inline void tasklet_disable(struct tasklet_struct *t)
 	smp_mb();
 }
 
-static inline void tasklet_enable(struct tasklet_struct *t)
-{
-	smp_mb__before_atomic();
-	atomic_dec(&t->count);
-}
-
+extern void tasklet_enable(struct tasklet_struct *t);
 extern void tasklet_kill(struct tasklet_struct *t);
 extern void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu);
 extern void tasklet_init(struct tasklet_struct *t,
diff --git a/kernel/softirq.c b/kernel/softirq.c
index 080eb57789c4..47c8933d315e 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -516,6 +516,7 @@ static __latent_entropy void tasklet_action(struct softirq_action *a)
 
 	while (list) {
 		struct tasklet_struct *t = list;
+		bool raise_softirq = true;
 
 		list = list->next;
 
@@ -529,13 +530,15 @@ static __latent_entropy void tasklet_action(struct softirq_action *a)
 				continue;
 			}
 			tasklet_unlock(t);
+			raise_softirq = false;
 		}
 
 		local_irq_disable();
 		t->next = NULL;
 		*__this_cpu_read(tasklet_vec.tail) = t;
 		__this_cpu_write(tasklet_vec.tail, &(t->next));
-		__raise_softirq_irqoff(TASKLET_SOFTIRQ);
+		if (raise_softirq)
+			__raise_softirq_irqoff(TASKLET_SOFTIRQ);
 		local_irq_enable();
 	}
 }
@@ -552,6 +555,7 @@ static __latent_entropy void tasklet_hi_action(struct softirq_action *a)
 
 	while (list) {
 		struct tasklet_struct *t = list;
+		bool raise_softirq = true;
 
 		list = list->next;
 
@@ -565,13 +569,15 @@ static __latent_entropy void tasklet_hi_action(struct softirq_action *a)
 				continue;
 			}
 			tasklet_unlock(t);
+			raise_softirq = false;
 		}
 
 		local_irq_disable();
 		t->next = NULL;
 		*__this_cpu_read(tasklet_hi_vec.tail) = t;
 		__this_cpu_write(tasklet_hi_vec.tail, &(t->next));
-		__raise_softirq_irqoff(HI_SOFTIRQ);
+		if (raise_softirq)
+			__raise_softirq_irqoff(HI_SOFTIRQ);
 		local_irq_enable();
 	}
 }
@@ -587,6 +593,16 @@ void tasklet_init(struct tasklet_struct *t,
 }
 EXPORT_SYMBOL(tasklet_init);
 
+void tasklet_enable(struct tasklet_struct *t)
+{
+	if (!atomic_dec_and_test(&t->count))
+		return;
+
+	if (test_bit(TASKLET_STATE_SCHED, &t->state))
+		raise_softirq(HI_SOFTIRQ | TASKLET_SOFTIRQ);
+}
+EXPORT_SYMBOL(tasklet_enable);
+
 void tasklet_kill(struct tasklet_struct *t)
 {
 	if (in_interrupt())
-- 
2.11.0