[S390] smp: delay idle task creation

Delay idle task creation until a cpu gets set online instead of
creating them for all possible cpus at system startup.
For one cpu system this should safe more than 1 MB.
On my debug system with lots of debug stuff enabled this saves 2 MB.

Same as on x86.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index a9702df..8e84b5a 100644 (file)
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -23,6 +23,7 @@
 #define KMSG_COMPONENT "cpu"
 #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
 
+#include <linux/workqueue.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/mm.h>
@@ -477,18 +478,20 @@ int __cpuinit start_secondary(void *cpuvoid)
        return 0;
 }
 
-static void __init smp_create_idle(unsigned int cpu)
+struct create_idle {
+       struct work_struct work;
+       struct task_struct *idle;
+       struct completion done;
+       int cpu;
+};
+
+static void __cpuinit smp_fork_idle(struct work_struct *work)
 {
-       struct task_struct *p;
+       struct create_idle *c_idle;
 
-       /*
-        *  don't care about the psw and regs settings since we'll never
-        *  reschedule the forked task.
-        */
-       p = fork_idle(cpu);
-       if (IS_ERR(p))
-               panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
-       current_set[cpu] = p;
+       c_idle = container_of(work, struct create_idle, work);
+       c_idle->idle = fork_idle(c_idle->cpu);
+       complete(&c_idle->done);
 }
 
 static int __cpuinit smp_alloc_lowcore(int cpu)
@@ -552,6 +555,7 @@ static void smp_free_lowcore(int cpu)
 int __cpuinit __cpu_up(unsigned int cpu)
 {
        struct _lowcore *cpu_lowcore;
+       struct create_idle c_idle;
        struct task_struct *idle;
        struct stack_frame *sf;
        u32 lowcore;
@@ -559,6 +563,18 @@ int __cpuinit __cpu_up(unsigned int cpu)
 
        if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED)
                return -EIO;
+       idle = current_set[cpu];
+       if (!idle) {
+               c_idle.done = COMPLETION_INITIALIZER_ONSTACK(c_idle.done);
+               INIT_WORK_ONSTACK(&c_idle.work, smp_fork_idle);
+               c_idle.cpu = cpu;
+               schedule_work(&c_idle.work);
+               wait_for_completion(&c_idle.done);
+               if (IS_ERR(c_idle.idle))
+                       return PTR_ERR(c_idle.idle);
+               idle = c_idle.idle;
+               current_set[cpu] = c_idle.idle;
+       }
        if (smp_alloc_lowcore(cpu))
                return -ENOMEM;
        do {
@@ -573,7 +589,6 @@ int __cpuinit __cpu_up(unsigned int cpu)
        while (sigp_p(lowcore, cpu, sigp_set_prefix) == sigp_busy)
                udelay(10);
 
-       idle = current_set[cpu];
        cpu_lowcore = lowcore_ptr[cpu];
        cpu_lowcore->kernel_stack = (unsigned long)
                task_stack_page(idle) + THREAD_SIZE;
@@ -685,7 +700,6 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 #endif
        unsigned long async_stack, panic_stack;
        struct _lowcore *lowcore;
-       unsigned int cpu;
 
        smp_detect_cpus();
 
@@ -720,9 +734,6 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
        if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore))
                BUG();
 #endif
-       for_each_possible_cpu(cpu)
-               if (cpu != smp_processor_id())
-                       smp_create_idle(cpu);
 }
 
 void __init smp_prepare_boot_cpu(void)