sched: Debug missed preemption checks

Developers use preempt_enable_no_resched() in places where the code
calls schedule() immediately which is correct. But there are places
where preempt_enable_no_resched() is not followed by schedule().

Add debug infrastructre to find the offending code. The identified
correct users are converted to use __preempt_enable_no_resched().

For the ever repeating "preempt_enable_no_resched(); schedule();"
sequences a onvenience macro preempt_enable_and_schedule() is
introduced.

Based on a previous patch from Ingo Molnar <mingo@elte.hu>

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

1 files changed, 3 insertions, 4 deletions

diff --git a/kernel/softirq.c b/kernel/softirq.c
index 5965d9f48fe9..07d2eba650b1 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -308,7 +308,7 @@ void irq_exit(void)
         if (idle_cpu(smp_processor_id()) && !in_interrupt() && !need_resched())
                 tick_nohz_stop_sched_tick(0);
 #endif
-        preempt_enable_no_resched();
+        __preempt_enable_no_resched();
 }
 
 /*
@@ -642,8 +642,7 @@ static int ksoftirqd(void * __bind_cpu)
         while (!kthread_should_stop()) {
                 preempt_disable();
                 if (!local_softirq_pending()) {
-                        preempt_enable_no_resched();
-                        schedule();
+                        preempt_enable_and_schedule();
                         preempt_disable();
                 }
 
@@ -656,7 +655,7 @@ static int ksoftirqd(void * __bind_cpu)
                         if (cpu_is_offline((long)__bind_cpu))
                                 goto wait_to_die;
                         do_softirq();
-                        preempt_enable_no_resched();
+                        __preempt_enable_no_resched();
                         cond_resched();
                         preempt_disable();
                         rcu_qsctr_inc((long)__bind_cpu);