[PATCH v2] sched/cputime: add steal clock warp handling

[Wanpeng Li]

From: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>

I observed that sometimes st is 100% instantaneous, then idle is 100% 
even if there is a cpu hog on the guest cpu after the cpu hotplug comes 
back(N.B. this can not always be readily reproduced). I add trace to 
capture it as below:

cpuhp/1-12    [001] d.h1   167.461657: account_process_tick: steal = 1291385514, prev_steal_time = 0         
cpuhp/1-12    [001] d.h1   167.461659: account_process_tick: steal_jiffies = 1291          
<idle>-0     [001] d.h1   167.462663: account_process_tick: steal = 18732255, prev_steal_time = 1291000000          
<idle>-0     [001] d.h1   167.462664: account_process_tick: steal_jiffies = 18446744072437

The steal clock warp and then steal_jiffies overflow.

Rik also pointed out to me:
 
| I have seen stuff like that with live migration too, in the past 

This patch adds steal clock warp handling by a safe threshold to only 
apply steal times that are positive and smaller than one second (as 
long as nohz_full has the one second timer tick left), ignoring intervals 
that are negative or longer than a second, and using those to sync up 
the guest with the host.

Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
Cc: Radim <rkrcmar@xxxxxxxxxx>
Signed-off-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
---
v1 -> v2:
 * update patch subject, description and comments
 * deal with the case where steal time suddenly increases by a ludicrous amount

 kernel/sched/cputime.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c
index f51c98c..751798a 100644
--- a/kernel/sched/cputime.c
+++ b/kernel/sched/cputime.c
@@ -262,17 +262,28 @@ static __always_inline unsigned long steal_account_process_tick(void)
 #ifdef CONFIG_PARAVIRT
 	if (static_key_false(&paravirt_steal_enabled)) {
 		u64 steal;
+		s64 delta;
 		unsigned long steal_jiffies;
 
 		steal = paravirt_steal_clock(smp_processor_id());
-		steal -= this_rq()->prev_steal_time;
+		delta = steal - this_rq()->prev_steal_time;
+		/*
+		 * Ignore this steal time difference if the guest and the host got
+		 * out of sync. This can happen due to events like live migration,
+		 * or CPU hotplug. The upper threshold is set to one second to match
+		 * the one second timer tick with nohz_full.
+		 */
+		if (unlikely(delta < 0 || delta > NSEC_PER_SEC)) {
+			this_rq()->prev_steal_time = steal;
+			return 0;
+		}
 
 		/*
 		 * steal is in nsecs but our caller is expecting steal
 		 * time in jiffies. Lets cast the result to jiffies
 		 * granularity and account the rest on the next rounds.
 		 */
-		steal_jiffies = nsecs_to_jiffies(steal);
+		steal_jiffies = nsecs_to_jiffies(delta);
 		this_rq()->prev_steal_time += jiffies_to_nsecs(steal_jiffies);
 
 		account_steal_time(jiffies_to_cputime(steal_jiffies));
-- 
1.9.1