[PATCH] don't call console drivers on non-online CPUs

George Anzinger identified the following problem: when a secondary CPU is
coming up, it calls printk() before it is "online".  It calls the console
drivers before its per-cpu storage has been prepared.  And the vga console
driver does a mod_timer().  This CPU's timers have not yet been initialised;
it is not clear why this doesn't oops - George thinks it is because virtual
address zero is still accessible at that time.

I believe the right way to fix this is to change printk so that a not-online
CPU will not call the console drivers.  Because printk should always be
callable.  If the CPU is not online the message is buffered, so the next
caller to printk who is online will actually display it.

ia64 has been doing exactly this for ages, so we can remove the
arch_consoles_callable() hook and just open-code the cpu_online() test in
printk.



That fixes things up for the secondary CPUs.  But this change causes a
problem for the boot CPU: it is being marked online very late in boot, so the
printk buffer is being displayed much later than we would like.

I believe that the solution to this is to mark the boot CPU online much
earlier.  So in this patch we call the new arch-provided function
smp_prepare_boot_cpu() immediately after the boot CPU's per-cpu areas are set
up.  Its mandate is to (at least) mark the boot CPU "online".

The change has been reviewed by davem and rth.  No comments were received
from the other arch maintainers.

arch/alpha/kernel/smp.c

@@ -607,6 +607,12 @@ smp_prepare_cpus(unsigned int max_cpus)
 	smp_boot_cpus();
 }
 
+void __devinit
+smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_present_mask);
+}
+
 int __devinit
 __cpu_up(unsigned int cpu)
 {

arch/i386/kernel/smpboot.c

@@ -995,11 +995,6 @@ static void __init smp_boot_cpus(unsigned int max_cpus)
 	printk("CPU%d: ", 0);
 	print_cpu_info(&cpu_data[0]);
 
-	/*
-	 * We have the boot CPU online for sure.
-	 */
-	set_bit(0, &cpu_online_map);
-	set_bit(0, &cpu_callout_map);
 	boot_cpu_logical_apicid = logical_smp_processor_id();
 	map_cpu_to_boot_apicid(0, boot_cpu_apicid);
 
@@ -1172,6 +1167,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	smp_boot_cpus(max_cpus);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_callout_map);
+}
+
 int __devinit __cpu_up(unsigned int cpu)
 {
 	/* This only works at boot for x86.  See "rewrite" above. */

arch/i386/mach-voyager/voyager_smp.c

@@ -1942,6 +1942,12 @@ smp_prepare_cpus(unsigned int max_cpus)
 	smp_boot_cpus();
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_callout_map);
+}
+
 int __devinit
 __cpu_up(unsigned int cpu)
 {

arch/ia64/kernel/smpboot.c

@@ -506,6 +506,12 @@ smp_prepare_cpus (unsigned int max_cpus)
 	}
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_callin_map);
+}
+
 void
 smp_cpus_done (unsigned int dummy)
 {

arch/parisc/kernel/smp.c

@@ -706,6 +706,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	smp_boot_cpus();
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_present_mask);
+}
+
 int __devinit __cpu_up(unsigned int cpu)
 {
 	return cpu_online(cpu) ? 0 : -ENOSYS;

arch/ppc/kernel/smp.c

@@ -351,6 +351,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 		smp_ops->space_timers(num_cpus);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_possible_map);
+}
+
 int __init setup_profiling_timer(unsigned int multiplier)
 {
 	return 0;

arch/ppc64/kernel/smp.c

@@ -604,6 +604,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	smp_space_timers(max_cpus);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	/* FIXME: what about cpu_possible()? */
+}
+
 int __devinit __cpu_up(unsigned int cpu)
 {
 	struct pt_regs regs;

arch/s390/kernel/smp.c

@@ -572,6 +572,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	set_prefix((u32) lowcore_ptr[smp_processor_id()]);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_possible_map);
+}
+
 void smp_cpus_done(unsigned int max_cpus)
 {
 }

arch/s390x/kernel/smp.c

@@ -554,6 +554,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	set_prefix((u32)(u64) lowcore_ptr[smp_processor_id()]);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_possible_map);
+}
+
 void smp_cpus_done(unsigned int max_cpis)
 {
 }

arch/sparc64/kernel/smp.c

@@ -1246,6 +1246,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	smp_store_cpu_info(boot_cpu_id);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &phys_cpu_present_map);
+}
+
 int __devinit __cpu_up(unsigned int cpu)
 {
 	int ret = smp_boot_one_cpu(cpu);

arch/um/kernel/smp.c

@@ -186,6 +186,11 @@ void smp_prepare_cpus(unsigned int maxcpus)
 	}
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+}
+
 int __cpu_up(unsigned int cpu)
 {
 	set_bit(cpu, &smp_commenced_mask);

arch/x86_64/kernel/smpboot.c

@@ -962,6 +962,12 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 	smp_boot_cpus(max_cpus);
 }
 
+void __devinit smp_prepare_boot_cpu(void)
+{
+	set_bit(smp_processor_id(), &cpu_online_map);
+	set_bit(smp_processor_id(), &cpu_callout_map);
+}
+
 int __devinit __cpu_up(unsigned int cpu)
 {
 	/* This only works at boot for x86.  See "rewrite" above. */

include/asm-ia64/system.h

@@ -234,9 +234,6 @@ extern void ia64_load_extra (struct task_struct *task);
 
 #ifdef CONFIG_SMP
 
-/* Return true if this CPU can call the console drivers in printk() */
-#define arch_consoles_callable() (cpu_online_map & (1UL << smp_processor_id()))
-
 /*
  * In the SMP case, we save the fph state when context-switching
  * away from a thread that modified fph.  This way, when the thread

include/linux/smp.h

@@ -78,6 +78,13 @@ extern int register_cpu_notifier(struct notifier_block *nb);
 extern void unregister_cpu_notifier(struct notifier_block *nb);
 
 int cpu_up(unsigned int cpu);
+
+/*
+ * Mark the boot cpu "online" so that it can call console drivers in
+ * printk() and can access its per-cpu storage.
+ */
+void smp_prepare_boot_cpu(void);
+
 #else /* !SMP */
 
 /*
@@ -95,6 +102,7 @@ static inline void smp_send_reschedule_all(void) { }
 #define num_online_cpus()			1
 #define num_booting_cpus()			1
 #define cpu_possible(cpu)			({ BUG_ON((cpu) != 0); 1; })
+#define smp_prepare_boot_cpu()			do {} while (0)
 
 struct notifier_block;
 

init/main.c

@@ -377,6 +377,13 @@ asmlinkage void __init start_kernel(void)
 	printk(linux_banner);
 	setup_arch(&command_line);
 	setup_per_cpu_areas();
+
+	/*
+	 * Mark the boot cpu "online" so that it can call console drivers in
+	 * printk() and can access its per-cpu storage.
+	 */
+	smp_prepare_boot_cpu();
+
 	build_all_zonelists();
 	page_alloc_init();
 	printk("Kernel command line: %s\n", saved_command_line);

kernel/printk.c

@@ -27,6 +27,7 @@
 #include <linux/interrupt.h>			/* For in_interrupt() */
 #include <linux/config.h>
 #include <linux/delay.h>
+#include <linux/smp.h>
 
 #include <asm/uaccess.h>
 
@@ -42,10 +43,6 @@
 
 #define LOG_BUF_MASK	(LOG_BUF_LEN-1)
 
-#ifndef arch_consoles_callable
-#define arch_consoles_callable() (1)
-#endif
-
 /* printk's without a loglevel use this.. */
 #define DEFAULT_MESSAGE_LOGLEVEL 4 /* KERN_WARNING */
 
@@ -447,10 +444,12 @@ asmlinkage int printk(const char *fmt, ...)
 			log_level_unknown = 1;
 	}
 
-	if (!arch_consoles_callable()) {
+	if (!cpu_online(smp_processor_id())) {
 		/*
-		 * On some architectures, the consoles are not usable
-		 * on secondary CPUs early in the boot process.
+		 * Some console drivers may assume that per-cpu resources have
+		 * been allocated.  So don't allow them to be called by this
+		 * CPU until it is officially up.  We shouldn't be calling into
+		 * random console drivers on a CPU which doesn't exist yet..
 		 */
 		spin_unlock_irqrestore(&logbuf_lock, flags);
 		goto out;
@@ -638,7 +637,8 @@ void register_console(struct console * console)
 	}
 	if (console->flags & CON_PRINTBUFFER) {
 		/*
-		 * release_console_sem() will print out the buffered messages for us.
+		 * release_console_sem() will print out the buffered messages
+		 * for us.
 		 */
 		spin_lock_irqsave(&logbuf_lock, flags);
 		con_start = log_start;