[PATCH] CPU hotplug: call check_tsc_sync_source() with irqs off

check_tsc_sync_source() depends on being called with irqs disabled (it
checks whether the TSC is coherent across two specific CPUs). This is
incidentally true during bootup, but not during cpu hotplug __cpu_up().
This got found via smp_processor_id() debugging.

disable irqs explicitly and remove the unconditional enabling of
interrupts. Add touch_nmi_watchdog() to the cpu_online_map busy loop.

this bug is present both on i386 and on x86_64.

Reported-by: Michal Piotrowski <michal.k.k.piotrowski@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

2 files changed, 14 insertions, 7 deletions

diff --git a/arch/i386/kernel/smpboot.c b/arch/i386/kernel/smpboot.c
index 9b0dd2744c82..4ff55e675576 100644
--- a/arch/i386/kernel/smpboot.c
+++ b/arch/i386/kernel/smpboot.c
@@ -45,6 +45,7 @@
 #include <linux/notifier.h>
 #include <linux/cpu.h>
 #include <linux/percpu.h>
+#include <linux/nmi.h>
 
 #include <linux/delay.h>
 #include <linux/mc146818rtc.h>
@@ -1278,8 +1279,9 @@ void __cpu_die(unsigned int cpu)
 
 int __cpuinit __cpu_up(unsigned int cpu)
 {
+        unsigned long flags;
 #ifdef CONFIG_HOTPLUG_CPU
-        int ret=0;
+        int ret = 0;
 
         /*
          * We do warm boot only on cpus that had booted earlier
@@ -1297,23 +1299,25 @@ int __cpuinit __cpu_up(unsigned int cpu)
         /* In case one didn't come up */
         if (!cpu_isset(cpu, cpu_callin_map)) {
                 printk(KERN_DEBUG "skipping cpu%d, didn't come online\n", cpu);
-                local_irq_enable();
                 return -EIO;
         }
 
-        local_irq_enable();
-
         per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
         /* Unleash the CPU! */
         cpu_set(cpu, smp_commenced_mask);
 
         /*
-         * Check TSC synchronization with the AP:
+         * Check TSC synchronization with the AP (keep irqs disabled
+         * while doing so):
          */
+        local_irq_save(flags);
         check_tsc_sync_source(cpu);
+        local_irq_restore(flags);
 
-        while (!cpu_isset(cpu, cpu_online_map))
+        while (!cpu_isset(cpu, cpu_online_map)) {
                 cpu_relax();
+                touch_nmi_watchdog();
+        }
 
 #ifdef CONFIG_X86_GENERICARCH
         if (num_online_cpus() > 8 && genapic == &apic_default)
diff --git a/arch/x86_64/kernel/smpboot.c b/arch/x86_64/kernel/smpboot.c
index 35443729aad8..cd4643a37022 100644
--- a/arch/x86_64/kernel/smpboot.c
+++ b/arch/x86_64/kernel/smpboot.c
@@ -923,8 +923,9 @@ void __init smp_prepare_boot_cpu(void)
  */
 int __cpuinit __cpu_up(unsigned int cpu)
 {
-        int err;
         int apicid = cpu_present_to_apicid(cpu);
+        unsigned long flags;
+        int err;
 
         WARN_ON(irqs_disabled());
 
@@ -958,7 +959,9 @@ int __cpuinit __cpu_up(unsigned int cpu)
         /*
          * Make sure and check TSC sync:
          */
+        local_irq_save(flags);
         check_tsc_sync_source(cpu);
+        local_irq_restore(flags);
 
         while (!cpu_isset(cpu, cpu_online_map))
                 cpu_relax();