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 <asm/fixmap.h> 25 #include <asm/pvclock.h> 26 27 static u8 valid_flags __read_mostly = 0; 28 29 void pvclock_set_flags(u8 flags) 30 { 31 valid_flags = flags; 32 } 33 34 unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src) 35 { 36 u64 pv_tsc_khz = 1000000ULL << 32; 37 38 do_div(pv_tsc_khz, src->tsc_to_system_mul); 39 if (src->tsc_shift < 0) 40 pv_tsc_khz <<= -src->tsc_shift; 41 else 42 pv_tsc_khz >>= src->tsc_shift; 43 return pv_tsc_khz; 44 } 45 46 void pvclock_touch_watchdogs(void) 47 { 48 touch_softlockup_watchdog_sync(); 49 clocksource_touch_watchdog(); 50 rcu_cpu_stall_reset(); 51 reset_hung_task_detector(); 52 } 53 54 static atomic64_t last_value = ATOMIC64_INIT(0); 55 56 void pvclock_resume(void) 57 { 58 atomic64_set(&last_value, 0); 59 } 60 61 u8 pvclock_read_flags(struct pvclock_vcpu_time_info *src) 62 { 63 unsigned version; 64 u8 flags; 65 66 do { 67 version = pvclock_read_begin(src); 68 flags = src->flags; 69 } while (pvclock_read_retry(src, version)); 70 71 return flags & valid_flags; 72 } 73 74 cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src) 75 { 76 unsigned version; 77 cycle_t ret; 78 u64 last; 79 u8 flags; 80 81 do { 82 version = pvclock_read_begin(src); 83 ret = __pvclock_read_cycles(src); 84 flags = src->flags; 85 } while (pvclock_read_retry(src, version)); 86 87 if (unlikely((flags & PVCLOCK_GUEST_STOPPED) != 0)) { 88 src->flags &= ~PVCLOCK_GUEST_STOPPED; 89 pvclock_touch_watchdogs(); 90 } 91 92 if ((valid_flags & PVCLOCK_TSC_STABLE_BIT) && 93 (flags & PVCLOCK_TSC_STABLE_BIT)) 94 return ret; 95 96 /* 97 * Assumption here is that last_value, a global accumulator, always goes 98 * forward. If we are less than that, we should not be much smaller. 99 * We assume there is an error marging we're inside, and then the correction 100 * does not sacrifice accuracy. 101 * 102 * For reads: global may have changed between test and return, 103 * but this means someone else updated poked the clock at a later time. 104 * We just need to make sure we are not seeing a backwards event. 105 * 106 * For updates: last_value = ret is not enough, since two vcpus could be 107 * updating at the same time, and one of them could be slightly behind, 108 * making the assumption that last_value always go forward fail to hold. 109 */ 110 last = atomic64_read(&last_value); 111 do { 112 if (ret < last) 113 return last; 114 last = atomic64_cmpxchg(&last_value, last, ret); 115 } while (unlikely(last != ret)); 116 117 return ret; 118 } 119 120 void pvclock_read_wallclock(struct pvclock_wall_clock *wall_clock, 121 struct pvclock_vcpu_time_info *vcpu_time, 122 struct timespec *ts) 123 { 124 u32 version; 125 u64 delta; 126 struct timespec now; 127 128 /* get wallclock at system boot */ 129 do { 130 version = wall_clock->version; 131 rmb(); /* fetch version before time */ 132 now.tv_sec = wall_clock->sec; 133 now.tv_nsec = wall_clock->nsec; 134 rmb(); /* fetch time before checking version */ 135 } while ((wall_clock->version & 1) || (version != wall_clock->version)); 136 137 delta = pvclock_clocksource_read(vcpu_time); /* time since system boot */ 138 delta += now.tv_sec * (u64)NSEC_PER_SEC + now.tv_nsec; 139 140 now.tv_nsec = do_div(delta, NSEC_PER_SEC); 141 now.tv_sec = delta; 142 143 set_normalized_timespec(ts, now.tv_sec, now.tv_nsec); 144 } 145