Similarly to what happened whith autosuspend, a deadlock has been raised
with runtime accounting in the sequence:

change_clocksource
    ...
    write_seqcount_begin
    ...
    timekeeping_update
        ...
        sh_cmt_clocksource_enable
            ...
            rpm_resume
                update_pm_runtime_accounting
                    ktime_get
                        do
                            read_seqcount_begin
                        while read_seqcount_retry
    ....
    write_seqcount_end

Move runtime accounting on ktime_get_mono_fast_ns()
With ktime_get_mono_fast_ns, the timestamp is not guaranteed to be
monotonic across an update of timekeeping and as a result time can goes
backward. Add a test to skip accounting for such situation which should
stay exceptional.

Fixes: a08c2a5a3194 ("PM-runtime: Replace jiffies-based accounting with 
ktime-based accounting")
Reported-by: Biju Das <[email protected]>
Signed-off-by: Vincent Guittot <[email protected]>
---
 drivers/base/power/runtime.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
index 1caa394..50740da 100644
--- a/drivers/base/power/runtime.c
+++ b/drivers/base/power/runtime.c
@@ -66,13 +66,22 @@ static int rpm_suspend(struct device *dev, int rpmflags);
  */
 void update_pm_runtime_accounting(struct device *dev)
 {
-       u64 now = ktime_to_ns(ktime_get());
+       u64 now = ktime_get_mono_fast_ns();
+       u64 last = dev->power.accounting_timestamp;
        u64 delta;
 
-       delta = now - dev->power.accounting_timestamp;
-
        dev->power.accounting_timestamp = now;
 
+       /*
+        * Because ktime_get_mono_fast_ns() is not monotonic during
+        * timekeeping update, we must ensure that now is after the last saved
+        * timesptamp
+        */
+       if (now < last)
+               return;
+
+       delta = now - last;
+
        if (dev->power.disable_depth > 0)
                return;
 
@@ -1312,7 +1321,7 @@ void pm_runtime_enable(struct device *dev)
 
                /* About to enable runtime pm, set accounting_timestamp to now 
*/
                if (!dev->power.disable_depth)
-                       dev->power.accounting_timestamp = 
ktime_to_ns(ktime_get());
+                       dev->power.accounting_timestamp = 
ktime_get_mono_fast_ns();
        } else {
                dev_warn(dev, "Unbalanced %s!\n", __func__);
        }
-- 
2.7.4

Reply via email to