tick/broadcast: Use for_each_cpu() specially on UP kernels
authorDexuan Cui <decui@microsoft.com>
Tue, 15 May 2018 19:52:50 +0000 (19:52 +0000)
committerThomas Gleixner <tglx@linutronix.de>
Tue, 15 May 2018 20:45:54 +0000 (22:45 +0200)
for_each_cpu() unintuitively reports CPU0 as set independent of the actual
cpumask content on UP kernels. This causes an unexpected PIT interrupt
storm on a UP kernel running in an SMP virtual machine on Hyper-V, and as
a result, the virtual machine can suffer from a strange random delay of 1~20
minutes during boot-up, and sometimes it can hang forever.

Protect if by checking whether the cpumask is empty before entering the
for_each_cpu() loop.

[ tglx: Use !IS_ENABLED(CONFIG_SMP) instead of #ifdeffery ]

Signed-off-by: Dexuan Cui <decui@microsoft.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Josh Poulson <jopoulso@microsoft.com>
Cc: "Michael Kelley (EOSG)" <Michael.H.Kelley@microsoft.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: stable@vger.kernel.org
Cc: Rakib Mullick <rakib.mullick@gmail.com>
Cc: Jork Loeser <Jork.Loeser@microsoft.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: KY Srinivasan <kys@microsoft.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Link: https://lkml.kernel.org/r/KL1P15301MB000678289FE55BA365B3279ABF990@KL1P15301MB0006.APCP153.PROD.OUTLOOK.COM
Link: https://lkml.kernel.org/r/KL1P15301MB0006FA63BC22BEB64902EAA0BF930@KL1P15301MB0006.APCP153.PROD.OUTLOOK.COM
kernel/time/tick-broadcast.c

index b398c2ea69b290cdaec1769b7d11cbc501646652..aa2094d5dd275372f999d051887aec22bfab9f19 100644 (file)
@@ -612,6 +612,14 @@ static void tick_handle_oneshot_broadcast(struct clock_event_device *dev)
        now = ktime_get();
        /* Find all expired events */
        for_each_cpu(cpu, tick_broadcast_oneshot_mask) {
+               /*
+                * Required for !SMP because for_each_cpu() reports
+                * unconditionally CPU0 as set on UP kernels.
+                */
+               if (!IS_ENABLED(CONFIG_SMP) &&
+                   cpumask_empty(tick_broadcast_oneshot_mask))
+                       break;
+
                td = &per_cpu(tick_cpu_device, cpu);
                if (td->evtdev->next_event <= now) {
                        cpumask_set_cpu(cpu, tmpmask);