Re: [patch] remove set_tsk_need_resched() from init_idle()

[Ingo Molnar <mingo@elte.hu>]

* Ingo Molnar <mingo@elte.hu> wrote:

> 
> * Nick Piggin <nickpiggin@yahoo.com.au> wrote:
> 
> > > we could do it in the other direction just as much - i only touched 
> > > 3 architectures. Up to Andrew i guess.
> > 
> > How about just setting need_resched at the start of the cpu_idle 
> > function instead? Rather than changing the structure of the idle loops 
> > themselves. That would suit me best.
> 
> that's fine with me too.

updated patch below.

--

this patch (ontop of the current -mm scheduler patchset) tweaks 
cpu_idle() semantics a bit: it changes the idle loops (that do 
preemption) to call the first schedule() unconditionally.

the advantage is that as a result we dont have to set the idle thread's 
NEED_RESCHED flag in init_idle(), which in turn makes cond_resched() 
even more of an invariant: it can be called even from init code without 
it having any effect. A cond resched in the init codepath hangs 
otherwise.

this patch, while having no negative side-effects, enables wider use of 
cond_resched()s. (which might happen in the stock kernel too, but it's 
particulary important for voluntary-preempt) (note that for now this 
patch only covers architectures that use kernel/Kconfig.preempt, but all 
other architectures will work just fine too.)

Signed-off-by: Ingo Molnar <mingo@elte.hu>

 arch/i386/kernel/process.c   |    2 ++
 arch/ppc64/kernel/idle.c     |    1 +
 arch/x86_64/kernel/process.c |    2 ++
 kernel/sched.c               |   10 +++++++++-
 4 files changed, 14 insertions(+), 1 deletion(-)

--- linux/kernel/sched.c.orig
+++ linux/kernel/sched.c
@@ -4163,6 +4163,15 @@ void show_state(void)
 	read_unlock(&tasklist_lock);
 }
 
+/**
+ * init_idle - set up an idle thread for a given CPU
+ * @idle: task in question
+ * @cpu: cpu the idle task belongs to
+ *
+ * NOTE: this function does not set the idle thread's NEED_RESCHED
+ * flag, to make booting more robust. Architecture-level cpu_idle()
+ * functions should set it explicitly, before entering their idle-loop.
+ */
 void __devinit init_idle(task_t *idle, int cpu)
 {
 	runqueue_t *rq = cpu_rq(cpu);
@@ -4180,7 +4189,6 @@ void __devinit init_idle(task_t *idle, i
 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
 	idle->oncpu = 1;
 #endif
-	set_tsk_need_resched(idle);
 	spin_unlock_irqrestore(&rq->lock, flags);
 
 	/* Set the preempt count _outside_ the spinlocks! */
--- linux/arch/x86_64/kernel/process.c.orig
+++ linux/arch/x86_64/kernel/process.c
@@ -162,6 +162,8 @@ EXPORT_SYMBOL_GPL(cpu_idle_wait);
  */
 void cpu_idle (void)
 {
+	set_tsk_need_resched(current);
+
 	/* endless idle loop with no priority at all */
 	while (1) {
 		while (!need_resched()) {
--- linux/arch/ppc64/kernel/idle.c.orig
+++ linux/arch/ppc64/kernel/idle.c
@@ -305,6 +305,7 @@ static int native_idle(void)
 
 void cpu_idle(void)
 {
+	set_tsk_need_resched(current);
 	idle_loop();
 }
 
--- linux/arch/i386/kernel/process.c.orig
+++ linux/arch/i386/kernel/process.c
@@ -179,6 +179,8 @@ void cpu_idle(void)
 {
 	int cpu = _smp_processor_id();
 
+	set_tsk_need_resched(current);
+
 	/* endless idle loop with no priority at all */
 	while (1) {
 		while (!need_resched()) {