[PATCH 4.7 074/143] sched/cputime: Fix NO_HZ_FULL getrusage() monotonicity regression

From: Greg Kroah-Hartman
Date: Mon Sep 05 2016 - 12:52:26 EST


4.7-stable review patch. If anyone has any objections, please let me know.

------------------

From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>

commit 173be9a14f7b2e901cf77c18b1aafd4d672e9d9e upstream.

Mike reports:

Roughly 10% of the time, ltp testcase getrusage04 fails:
getrusage04 0 TINFO : Expected timers granularity is 4000 us
getrusage04 0 TINFO : Using 1 as multiply factor for max [us]time increment (1000+4000us)!
getrusage04 0 TINFO : utime: 0us; stime: 179us
getrusage04 0 TINFO : utime: 3751us; stime: 0us
getrusage04 1 TFAIL : getrusage04.c:133: stime increased > 5000us:

And tracked it down to the case where the task simply doesn't get
_any_ [us]time ticks.

Update the code to assume all rtime is utime when we lack information,
thus ensuring a task that elides the tick gets time accounted.

Reported-by: Mike Galbraith <umgwanakikbuti@xxxxxxxxx>
Tested-by: Mike Galbraith <umgwanakikbuti@xxxxxxxxx>
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Cc: Fredrik Markstrom <fredrik.markstrom@xxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Radim <rkrcmar@xxxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxx>
Cc: Stephane Eranian <eranian@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Vince Weaver <vincent.weaver@xxxxxxxxx>
Cc: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
Fixes: 9d7fb0427648 ("sched/cputime: Guarantee stime + utime == rtime")
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>

---
kernel/sched/cputime.c | 15 ++++++++++-----
1 file changed, 10 insertions(+), 5 deletions(-)

--- a/kernel/sched/cputime.c
+++ b/kernel/sched/cputime.c
@@ -603,19 +603,25 @@ static void cputime_adjust(struct task_c
stime = curr->stime;
utime = curr->utime;

- if (utime == 0) {
- stime = rtime;
+ /*
+ * If either stime or both stime and utime are 0, assume all runtime is
+ * userspace. Once a task gets some ticks, the monotonicy code at
+ * 'update' will ensure things converge to the observed ratio.
+ */
+ if (stime == 0) {
+ utime = rtime;
goto update;
}

- if (stime == 0) {
- utime = rtime;
+ if (utime == 0) {
+ stime = rtime;
goto update;
}

stime = scale_stime((__force u64)stime, (__force u64)rtime,
(__force u64)(stime + utime));

+update:
/*
* Make sure stime doesn't go backwards; this preserves monotonicity
* for utime because rtime is monotonic.
@@ -638,7 +644,6 @@ static void cputime_adjust(struct task_c
stime = rtime - utime;
}

-update:
prev->stime = stime;
prev->utime = utime;
out: