sched/idle: Move cpu/idle.c to sched/idle.c
authorNicolas Pitre <nicolas.pitre@linaro.org>
Mon, 27 Jan 2014 04:42:01 +0000 (23:42 -0500)
committerIngo Molnar <mingo@kernel.org>
Tue, 11 Feb 2014 08:58:30 +0000 (09:58 +0100)
Integration of cpuidle with the scheduler requires that the idle loop be
closely integrated with the scheduler proper. Moving cpu/idle.c into the
sched directory will allow for a smoother integration, and eliminate a
subdirectory which contained only one source file.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/alpine.LFD.2.11.1401301102210.1652@knanqh.ubzr
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/Makefile
kernel/cpu/Makefile [deleted file]
kernel/sched/Makefile
kernel/sched/idle.c [new file with mode: 0644]

index bc010ee272b6cfec39892e6cb04b98cc06995e1a..6f1c7e5cfca1f52fbde869d2a9967500b8fdd4bc 100644 (file)
@@ -22,7 +22,6 @@ obj-y += sched/
 obj-y += locking/
 obj-y += power/
 obj-y += printk/
-obj-y += cpu/
 obj-y += irq/
 obj-y += rcu/
 
diff --git a/kernel/cpu/Makefile b/kernel/cpu/Makefile
deleted file mode 100644 (file)
index 59ab052..0000000
+++ /dev/null
@@ -1 +0,0 @@
-obj-y  = idle.o
index 9a95c8c2af2af0e9fb62c794926df3f1775a9dbe..ab32b7b0db5c6b30b4cc64e61e6a857fb51c87e6 100644 (file)
@@ -13,7 +13,7 @@ endif
 
 obj-y += core.o proc.o clock.o cputime.o
 obj-y += idle_task.o fair.o rt.o deadline.o stop_task.o
-obj-y += wait.o completion.o
+obj-y += wait.o completion.o idle.o
 obj-$(CONFIG_SMP) += cpupri.o cpudeadline.o
 obj-$(CONFIG_SCHED_AUTOGROUP) += auto_group.o
 obj-$(CONFIG_SCHEDSTATS) += stats.o
diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
new file mode 100644 (file)
index 0000000..14ca434
--- /dev/null
@@ -0,0 +1,144 @@
+/*
+ * Generic entry point for the idle threads
+ */
+#include <linux/sched.h>
+#include <linux/cpu.h>
+#include <linux/cpuidle.h>
+#include <linux/tick.h>
+#include <linux/mm.h>
+#include <linux/stackprotector.h>
+
+#include <asm/tlb.h>
+
+#include <trace/events/power.h>
+
+static int __read_mostly cpu_idle_force_poll;
+
+void cpu_idle_poll_ctrl(bool enable)
+{
+       if (enable) {
+               cpu_idle_force_poll++;
+       } else {
+               cpu_idle_force_poll--;
+               WARN_ON_ONCE(cpu_idle_force_poll < 0);
+       }
+}
+
+#ifdef CONFIG_GENERIC_IDLE_POLL_SETUP
+static int __init cpu_idle_poll_setup(char *__unused)
+{
+       cpu_idle_force_poll = 1;
+       return 1;
+}
+__setup("nohlt", cpu_idle_poll_setup);
+
+static int __init cpu_idle_nopoll_setup(char *__unused)
+{
+       cpu_idle_force_poll = 0;
+       return 1;
+}
+__setup("hlt", cpu_idle_nopoll_setup);
+#endif
+
+static inline int cpu_idle_poll(void)
+{
+       rcu_idle_enter();
+       trace_cpu_idle_rcuidle(0, smp_processor_id());
+       local_irq_enable();
+       while (!tif_need_resched())
+               cpu_relax();
+       trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, smp_processor_id());
+       rcu_idle_exit();
+       return 1;
+}
+
+/* Weak implementations for optional arch specific functions */
+void __weak arch_cpu_idle_prepare(void) { }
+void __weak arch_cpu_idle_enter(void) { }
+void __weak arch_cpu_idle_exit(void) { }
+void __weak arch_cpu_idle_dead(void) { }
+void __weak arch_cpu_idle(void)
+{
+       cpu_idle_force_poll = 1;
+       local_irq_enable();
+}
+
+/*
+ * Generic idle loop implementation
+ */
+static void cpu_idle_loop(void)
+{
+       while (1) {
+               tick_nohz_idle_enter();
+
+               while (!need_resched()) {
+                       check_pgt_cache();
+                       rmb();
+
+                       if (cpu_is_offline(smp_processor_id()))
+                               arch_cpu_idle_dead();
+
+                       local_irq_disable();
+                       arch_cpu_idle_enter();
+
+                       /*
+                        * In poll mode we reenable interrupts and spin.
+                        *
+                        * Also if we detected in the wakeup from idle
+                        * path that the tick broadcast device expired
+                        * for us, we don't want to go deep idle as we
+                        * know that the IPI is going to arrive right
+                        * away
+                        */
+                       if (cpu_idle_force_poll || tick_check_broadcast_expired()) {
+                               cpu_idle_poll();
+                       } else {
+                               if (!current_clr_polling_and_test()) {
+                                       stop_critical_timings();
+                                       rcu_idle_enter();
+                                       if (cpuidle_idle_call())
+                                               arch_cpu_idle();
+                                       if (WARN_ON_ONCE(irqs_disabled()))
+                                               local_irq_enable();
+                                       rcu_idle_exit();
+                                       start_critical_timings();
+                               } else {
+                                       local_irq_enable();
+                               }
+                               __current_set_polling();
+                       }
+                       arch_cpu_idle_exit();
+                       /*
+                        * We need to test and propagate the TIF_NEED_RESCHED
+                        * bit here because we might not have send the
+                        * reschedule IPI to idle tasks.
+                        */
+                       if (tif_need_resched())
+                               set_preempt_need_resched();
+               }
+               tick_nohz_idle_exit();
+               schedule_preempt_disabled();
+       }
+}
+
+void cpu_startup_entry(enum cpuhp_state state)
+{
+       /*
+        * This #ifdef needs to die, but it's too late in the cycle to
+        * make this generic (arm and sh have never invoked the canary
+        * init for the non boot cpus!). Will be fixed in 3.11
+        */
+#ifdef CONFIG_X86
+       /*
+        * If we're the non-boot CPU, nothing set the stack canary up
+        * for us. The boot CPU already has it initialized but no harm
+        * in doing it again. This is a good place for updating it, as
+        * we wont ever return from this function (so the invalid
+        * canaries already on the stack wont ever trigger).
+        */
+       boot_init_stack_canary();
+#endif
+       __current_set_polling();
+       arch_cpu_idle_prepare();
+       cpu_idle_loop();
+}