arch/x86/kernel/pvclock.c

   1 /*  paravirtual clock -- common code used by kvm/xen
   2
   3     This program is free software; you can redistribute it and/or modify
   4     it under the terms of the GNU General Public License as published by
   5     the Free Software Foundation; either version 2 of the License, or
   6     (at your option) any later version.
   7
   8     This program is distributed in the hope that it will be useful,
   9     but WITHOUT ANY WARRANTY; without even the implied warranty of
  10     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  11     GNU General Public License for more details.
  12
  13     You should have received a copy of the GNU General Public License
  14     along with this program; if not, write to the Free Software
  15     Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  16 */
  17
  18 #include <linux/kernel.h>
  19 #include <linux/percpu.h>
  20 #include <linux/notifier.h>
  21 #include <linux/sched.h>
  22 #include <linux/gfp.h>
  23 #include <linux/bootmem.h>
  24 #include <linux/nmi.h>
  25
  26 #include <asm/fixmap.h>
  27 #include <asm/pvclock.h>
  28
  29 static u8 valid_flags __read_mostly = 0;
  30
  31 void pvclock_set_flags(u8 flags)
  32 {
  33         valid_flags = flags;
  34 }
  35
  36 unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src)
  37 {
  38         u64 pv_tsc_khz = 1000000ULL << 32;
  39
  40         do_div(pv_tsc_khz, src->tsc_to_system_mul);
  41         if (src->tsc_shift < 0)
  42                 pv_tsc_khz <<= -src->tsc_shift;
  43         else
  44                 pv_tsc_khz >>= src->tsc_shift;
  45         return pv_tsc_khz;
  46 }
  47
  48 void pvclock_touch_watchdogs(void)
  49 {
  50         touch_softlockup_watchdog_sync();
  51         clocksource_touch_watchdog();
  52         rcu_cpu_stall_reset();
  53         reset_hung_task_detector();
  54 }
  55
  56 static atomic64_t last_value = ATOMIC64_INIT(0);
  57
  58 void pvclock_resume(void)
  59 {
  60         atomic64_set(&last_value, 0);
  61 }
  62
  63 u8 pvclock_read_flags(struct pvclock_vcpu_time_info *src)
  64 {
  65         unsigned version;
  66         u8 flags;
  67
  68         do {
  69                 version = pvclock_read_begin(src);
  70                 flags = src->flags;
  71         } while (pvclock_read_retry(src, version));
  72
  73         return flags & valid_flags;
  74 }
  75
  76 u64 pvclock_clocksource_read(struct pvclock_vcpu_time_info *src)
  77 {
  78         unsigned version;
  79         u64 ret;
  80         u64 last;
  81         u8 flags;
  82
  83         do {
  84                 version = pvclock_read_begin(src);
  85                 ret = __pvclock_read_cycles(src, rdtsc_ordered());
  86                 flags = src->flags;
  87         } while (pvclock_read_retry(src, version));
  88
  89         if (unlikely((flags & PVCLOCK_GUEST_STOPPED) != 0)) {
  90                 src->flags &= ~PVCLOCK_GUEST_STOPPED;
  91                 pvclock_touch_watchdogs();
  92         }
  93
  94         if ((valid_flags & PVCLOCK_TSC_STABLE_BIT) &&
  95                 (flags & PVCLOCK_TSC_STABLE_BIT))
  96                 return ret;
  97
  98         /*
  99          * Assumption here is that last_value, a global accumulator, always goes
 100          * forward. If we are less than that, we should not be much smaller.
 101          * We assume there is an error marging we're inside, and then the correction
 102          * does not sacrifice accuracy.
 103          *
 104          * For reads: global may have changed between test and return,
 105          * but this means someone else updated poked the clock at a later time.
 106          * We just need to make sure we are not seeing a backwards event.
 107          *
 108          * For updates: last_value = ret is not enough, since two vcpus could be
 109          * updating at the same time, and one of them could be slightly behind,
 110          * making the assumption that last_value always go forward fail to hold.
 111          */
 112         last = atomic64_read(&last_value);
 113         do {
 114                 if (ret < last)
 115                         return last;
 116                 last = atomic64_cmpxchg(&last_value, last, ret);
 117         } while (unlikely(last != ret));
 118
 119         return ret;
 120 }
 121
 122 void pvclock_read_wallclock(struct pvclock_wall_clock *wall_clock,
 123                             struct pvclock_vcpu_time_info *vcpu_time,
 124                             struct timespec *ts)
 125 {
 126         u32 version;
 127         u64 delta;
 128         struct timespec now;
 129
 130         /* get wallclock at system boot */
 131         do {
 132                 version = wall_clock->version;
 133                 rmb();          /* fetch version before time */
 134                 now.tv_sec  = wall_clock->sec;
 135                 now.tv_nsec = wall_clock->nsec;
 136                 rmb();          /* fetch time before checking version */
 137         } while ((wall_clock->version & 1) || (version != wall_clock->version));
 138
 139         delta = pvclock_clocksource_read(vcpu_time);    /* time since system boot */
 140         delta += now.tv_sec * (u64)NSEC_PER_SEC + now.tv_nsec;
 141
 142         now.tv_nsec = do_div(delta, NSEC_PER_SEC);
 143         now.tv_sec = delta;
 144
 145         set_normalized_timespec(ts, now.tv_sec, now.tv_nsec);
 146 }