queue-4.19/timekeeping-repair-ktime_get_coarse-granularity.patch

   1 From e3ff9c3678b4d80e22d2557b68726174578eaf52 Mon Sep 17 00:00:00 2001
   2 From: Thomas Gleixner <tglx@linutronix.de>
   3 Date: Thu, 13 Jun 2019 21:40:45 +0200
   4 Subject: timekeeping: Repair ktime_get_coarse*() granularity
   5
   6 From: Thomas Gleixner <tglx@linutronix.de>
   7
   8 commit e3ff9c3678b4d80e22d2557b68726174578eaf52 upstream.
   9
  10 Jason reported that the coarse ktime based time getters advance only once
  11 per second and not once per tick as advertised.
  12
  13 The code reads only the monotonic base time, which advances once per
  14 second. The nanoseconds are accumulated on every tick in xtime_nsec up to
  15 a second and the regular time getters take this nanoseconds offset into
  16 account, but the ktime_get_coarse*() implementation fails to do so.
  17
  18 Add the accumulated xtime_nsec value to the monotonic base time to get the
  19 proper per tick advancing coarse tinme.
  20
  21 Fixes: b9ff604cff11 ("timekeeping: Add ktime_get_coarse_with_offset")
  22 Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
  23 Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
  24 Tested-by: Jason A. Donenfeld <Jason@zx2c4.com>
  25 Cc: Arnd Bergmann <arnd@arndb.de>
  26 Cc: Peter Zijlstra <peterz@infradead.org>
  27 Cc: Clemens Ladisch <clemens@ladisch.de>
  28 Cc: Sultan Alsawaf <sultan@kerneltoast.com>
  29 Cc: Waiman Long <longman@redhat.com>
  30 Cc: stable@vger.kernel.org
  31 Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1906132136280.1791@nanos.tec.linutronix.de
  32 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  33
  34 ---
  35  kernel/time/timekeeping.c |    5 +++--
  36  1 file changed, 3 insertions(+), 2 deletions(-)
  37
  38 --- a/kernel/time/timekeeping.c
  39 +++ b/kernel/time/timekeeping.c
  40 @@ -812,17 +812,18 @@ ktime_t ktime_get_coarse_with_offset(enu
  41         struct timekeeper *tk = &tk_core.timekeeper;
  42         unsigned int seq;
  43         ktime_t base, *offset = offsets[offs];
  44 +       u64 nsecs;
  45
  46         WARN_ON(timekeeping_suspended);
  47
  48         do {
  49                 seq = read_seqcount_begin(&tk_core.seq);
  50                 base = ktime_add(tk->tkr_mono.base, *offset);
  51 +               nsecs = tk->tkr_mono.xtime_nsec >> tk->tkr_mono.shift;
  52
  53         } while (read_seqcount_retry(&tk_core.seq, seq));
  54
  55 -       return base;
  56 -
  57 +       return base + nsecs;
  58  }
  59  EXPORT_SYMBOL_GPL(ktime_get_coarse_with_offset);
  60