time.c

xen 3.2.2 源码

            rc = init_pmtimer(pts);

        if ( rc <= 0 )
            printk("WARNING: %s clocksource '%s'.\n",
                   (rc == 0) ? "Could not initialise" : "Unrecognised",
                   opt_clocksource);
    }

    if ( (rc <= 0) &&
         !init_cyclone(pts) &&
         !init_hpet(pts) &&
         !init_pmtimer(pts) )
        init_pit(pts);

    plt_mask = (u32)~0u >> (32 - pts->counter_bits);

    set_time_scale(&plt_scale, pts->frequency);

    overflow_period = scale_delta(1ull << (pts->counter_bits-1), &plt_scale);
    do_div(overflow_period, MILLISECS(1000/HZ));
    plt_overflow_period = overflow_period;
    plt_overflow();
    printk("Platform timer overflows in %d jiffies.\n", plt_overflow_period);

    platform_timer_stamp = plt_count64;

    printk("Platform timer is %s %s\n",
           freq_string(pts->frequency), pts->name);
}


/***************************************************************************
 * CMOS Timer functions
 ***************************************************************************/

/* Converts Gregorian date to seconds since 1970-01-01 00:00:00.
 * Assumes input in normal date format, i.e. 1980-12-31 23:59:59
 * => year=1980, mon=12, day=31, hour=23, min=59, sec=59.
 *
 * [For the Julian calendar (which was used in Russia before 1917,
 * Britain & colonies before 1752, anywhere else before 1582,
 * and is still in use by some communities) leave out the
 * -year/100+year/400 terms, and add 10.]
 *
 * This algorithm was first published by Gauss (I think).
 *
 * WARNING: this function will overflow on 2106-02-07 06:28:16 on
 * machines were long is 32-bit! (However, as time_t is signed, we
 * will already get problems at other places on 2038-01-19 03:14:08)
 */
unsigned long
mktime (unsigned int year, unsigned int mon,
        unsigned int day, unsigned int hour,
        unsigned int min, unsigned int sec)
{
    /* 1..12 -> 11,12,1..10: put Feb last since it has a leap day. */
    if ( 0 >= (int) (mon -= 2) )
    {
        mon += 12;
        year -= 1;
    }

    return ((((unsigned long)(year/4 - year/100 + year/400 + 367*mon/12 + day)+
              year*365 - 719499
        )*24 + hour /* now have hours */
        )*60 + min  /* now have minutes */
        )*60 + sec; /* finally seconds */
}

static unsigned long __get_cmos_time(void)
{
    unsigned int year, mon, day, hour, min, sec;

    sec  = CMOS_READ(RTC_SECONDS);
    min  = CMOS_READ(RTC_MINUTES);
    hour = CMOS_READ(RTC_HOURS);
    day  = CMOS_READ(RTC_DAY_OF_MONTH);
    mon  = CMOS_READ(RTC_MONTH);
    year = CMOS_READ(RTC_YEAR);
    
    if ( !(CMOS_READ(RTC_CONTROL) & RTC_DM_BINARY) || RTC_ALWAYS_BCD )
    {
        BCD_TO_BIN(sec);
        BCD_TO_BIN(min);
        BCD_TO_BIN(hour);
        BCD_TO_BIN(day);
        BCD_TO_BIN(mon);
        BCD_TO_BIN(year);
    }

    if ( (year += 1900) < 1970 )
        year += 100;

    return mktime(year, mon, day, hour, min, sec);
}

static unsigned long get_cmos_time(void)
{
    unsigned long res, flags;
    int i;

    spin_lock_irqsave(&rtc_lock, flags);

    /* read RTC exactly on falling edge of update flag */
    for ( i = 0 ; i < 1000000 ; i++ ) /* may take up to 1 second... */
        if ( (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
            break;
    for ( i = 0 ; i < 1000000 ; i++ ) /* must try at least 2.228 ms */
        if ( !(CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
            break;

    res = __get_cmos_time();

    spin_unlock_irqrestore(&rtc_lock, flags);
    return res;
}

/***************************************************************************
 * System Time
 ***************************************************************************/

s_time_t get_s_time(void)
{
    struct cpu_time *t = &this_cpu(cpu_time);
    u64 tsc, delta;
    s_time_t now;

    rdtscll(tsc);
    delta = tsc - t->local_tsc_stamp;
    now = t->stime_local_stamp + scale_delta(delta, &t->tsc_scale);

    return now;
}

static inline void version_update_begin(u32 *version)
{
    /* Explicitly OR with 1 just in case version number gets out of sync. */
    *version = (*version + 1) | 1;
    wmb();
}

static inline void version_update_end(u32 *version)
{
    wmb();
    (*version)++;
}

void update_vcpu_system_time(struct vcpu *v)
{
    struct cpu_time       *t;
    struct vcpu_time_info *u;

    if ( v->vcpu_info == NULL )
        return;

    t = &this_cpu(cpu_time);
    u = &vcpu_info(v, time);

    if ( u->tsc_timestamp == t->local_tsc_stamp )
        return;

    version_update_begin(&u->version);

    u->tsc_timestamp     = t->local_tsc_stamp;
    u->system_time       = t->stime_local_stamp;
    u->tsc_to_system_mul = t->tsc_scale.mul_frac;
    u->tsc_shift         = (s8)t->tsc_scale.shift;

    version_update_end(&u->version);
}

void update_domain_wallclock_time(struct domain *d)
{
    spin_lock(&wc_lock);
    version_update_begin(&shared_info(d, wc_version));
    shared_info(d, wc_sec)  = wc_sec + d->time_offset_seconds;
    shared_info(d, wc_nsec) = wc_nsec;
    version_update_end(&shared_info(d, wc_version));
    spin_unlock(&wc_lock);
}

int cpu_frequency_change(u64 freq)
{
    struct cpu_time *t = &this_cpu(cpu_time);
    u64 curr_tsc;

    /* Sanity check: CPU frequency allegedly dropping below 1MHz? */
    if ( freq < 1000000u )
    {
        gdprintk(XENLOG_WARNING, "Rejecting CPU frequency change "
                 "to %"PRIu64" Hz.\n", freq);
        return -EINVAL;
    }

    local_irq_disable();
    /* Platform time /first/, as we may be delayed by platform_timer_lock. */
    t->stime_master_stamp = read_platform_stime();
    /* TSC-extrapolated time may be bogus after frequency change. */
    /*t->stime_local_stamp = get_s_time();*/
    t->stime_local_stamp = t->stime_master_stamp;
    rdtscll(curr_tsc);
    t->local_tsc_stamp = curr_tsc;
    set_time_scale(&t->tsc_scale, freq);
    local_irq_enable();

    update_vcpu_system_time(current);

    /* A full epoch should pass before we check for deviation. */
    set_timer(&t->calibration_timer, NOW() + EPOCH);
    if ( smp_processor_id() == 0 )
        platform_time_calibration();

    return 0;
}

/* Set clock to <secs,usecs> after 00:00:00 UTC, 1 January, 1970. */
void do_settime(unsigned long secs, unsigned long nsecs, u64 system_time_base)
{
    u64 x;
    u32 y, _wc_sec, _wc_nsec;
    struct domain *d;

    x = (secs * 1000000000ULL) + (u64)nsecs - system_time_base;
    y = do_div(x, 1000000000);

    spin_lock(&wc_lock);
    wc_sec  = _wc_sec  = (u32)x;
    wc_nsec = _wc_nsec = (u32)y;
    spin_unlock(&wc_lock);

    rcu_read_lock(&domlist_read_lock);
    for_each_domain ( d )
        update_domain_wallclock_time(d);
    rcu_read_unlock(&domlist_read_lock);
}

static void local_time_calibration(void *unused)
{
    struct cpu_time *t = &this_cpu(cpu_time);

    /*
     * System timestamps, extrapolated from local and master oscillators,
     * taken during this calibration and the previous calibration.
     */
    s_time_t prev_local_stime, curr_local_stime;
    s_time_t prev_master_stime, curr_master_stime;

    /* TSC timestamps taken during this calibration and prev calibration. */
    u64 prev_tsc, curr_tsc;

    /*
     * System time and TSC ticks elapsed during the previous calibration
     * 'epoch'. These values are down-shifted to fit in 32 bits.
     */
    u64 stime_elapsed64, tsc_elapsed64;
    u32 stime_elapsed32, tsc_elapsed32;

    /* The accumulated error in the local estimate. */
    u64 local_stime_err;

    /* Error correction to slow down a fast local clock. */
    u32 error_factor = 0;

    /* Calculated TSC shift to ensure 32-bit scale multiplier. */
    int tsc_shift = 0;

    /* The overall calibration scale multiplier. */
    u32 calibration_mul_frac;

    prev_tsc          = t->local_tsc_stamp;
    prev_local_stime  = t->stime_local_stamp;
    prev_master_stime = t->stime_master_stamp;

    /*
     * Disable IRQs to get 'instantaneous' current timestamps. We read platform
     * time first, as we may be delayed when acquiring platform_timer_lock.
     */
    local_irq_disable();
    curr_master_stime = read_platform_stime();
    curr_local_stime  = get_s_time();
    rdtscll(curr_tsc);
    local_irq_enable();

#if 0
    printk("PRE%d: tsc=%"PRIu64" stime=%"PRIu64" master=%"PRIu64"\n",
           smp_processor_id(), prev_tsc, prev_local_stime, prev_master_stime);
    printk("CUR%d: tsc=%"PRIu64" stime=%"PRIu64" master=%"PRIu64
           " -> %"PRId64"\n",
           smp_processor_id(), curr_tsc, curr_local_stime, curr_master_stime,
           curr_master_stime - curr_local_stime);
#endif

    /* Local time warps forward if it lags behind master time. */
    if ( curr_local_stime < curr_master_stime )
        curr_local_stime = curr_master_stime;

    stime_elapsed64 = curr_master_stime - prev_master_stime;
    tsc_elapsed64   = curr_tsc - prev_tsc;

    /*
     * Weirdness can happen if we lose sync with the platform timer.
     * We could be smarter here: resync platform timer with local timer?
     */
    if ( ((s64)stime_elapsed64 < (EPOCH / 2)) )
        goto out;

    /*
     * Calculate error-correction factor. This only slows down a fast local
     * clock (slow clocks are warped forwards). The scale factor is clamped
     * to >= 0.5.
     */
    if ( curr_local_stime != curr_master_stime )
    {
        local_stime_err = curr_local_stime - curr_master_stime;
        if ( local_stime_err > EPOCH )
            local_stime_err = EPOCH;
        error_factor = div_frac(EPOCH, EPOCH + (u32)local_stime_err);
    }

    /*
     * We require 0 < stime_elapsed < 2^31.
     * This allows us to binary shift a 32-bit tsc_elapsed such that:
     * stime_elapsed < tsc_elapsed <= 2*stime_elapsed
     */
    while ( ((u32)stime_elapsed64 != stime_elapsed64) ||
            ((s32)stime_elapsed64 < 0) )
    {
        stime_elapsed64 >>= 1;
        tsc_elapsed64   >>= 1;
    }

    /* stime_master_diff now fits in a 32-bit word. */
    stime_elapsed32 = (u32)stime_elapsed64;

    /* tsc_elapsed <= 2*stime_elapsed */
    while ( tsc_elapsed64 > (stime_elapsed32 * 2) )
    {
        tsc_elapsed64 >>= 1;
        tsc_shift--;
    }

    /* Local difference must now fit in 32 bits. */
    ASSERT((u32)tsc_elapsed64 == tsc_elapsed64);
    tsc_elapsed32 = (u32)tsc_elapsed64;

    /* tsc_elapsed > stime_elapsed */
    ASSERT(tsc_elapsed32 != 0);
    while ( tsc_elapsed32 <= stime_elapsed32 )
    {
        tsc_elapsed32 <<= 1;
        tsc_shift++;
    }

    calibration_mul_frac = div_frac(stime_elapsed32, tsc_elapsed32);
    if ( error_factor != 0 )
        calibration_mul_frac = mul_frac(calibration_mul_frac, error_factor);

#if 0
    printk("---%d: %08x %08x %d\n", smp_processor_id(),
           error_factor, calibration_mul_frac, tsc_shift);
#endif

    /* Record new timestamp information, atomically w.r.t. interrupts. */
    local_irq_disable();
    t->tsc_scale.mul_frac = calibration_mul_frac;
    t->tsc_scale.shift    = tsc_shift;
    t->local_tsc_stamp    = curr_tsc;
    t->stime_local_stamp  = curr_local_stime;
    t->stime_master_stamp = curr_master_stime;
    local_irq_enable();

    update_vcpu_system_time(current);

 out:
    set_timer(&t->calibration_timer, NOW() + EPOCH);

    if ( smp_processor_id() == 0 )
        platform_time_calibration();
}

void init_percpu_time(void)
{
    struct cpu_time *t = &this_cpu(cpu_time);
    unsigned long flags;
    s_time_t now;

    local_irq_save(flags);
    rdtscll(t->local_tsc_stamp);
    now = !plt_src.read_counter ? 0 : read_platform_stime();
    local_irq_restore(flags);

    t->stime_master_stamp = now;
    t->stime_local_stamp  = now;

    init_timer(&t->calibration_timer, local_time_calibration,
               NULL, smp_processor_id());
    set_timer(&t->calibration_timer, NOW() + EPOCH);
}

/* Late init function (after all CPUs are booted). */
int __init init_xen_time(void)
{
    wc_sec = get_cmos_time();

    local_irq_disable();

    init_percpu_time();

    stime_platform_stamp = 0;
    init_platform_timer();

    local_irq_enable();

    return 0;
}


/* Early init function. */
void __init early_time_init(void)
{
    u64 tmp = init_pit_and_calibrate_tsc();

    set_time_scale(&this_cpu(cpu_time).tsc_scale, tmp);

    do_div(tmp, 1000);
    cpu_khz = (unsigned long)tmp;
    printk("Detected %lu.%03lu MHz processor.\n", 
           cpu_khz / 1000, cpu_khz % 1000);

    setup_irq(0, &irq0);
}

void send_timer_event(struct vcpu *v)
{
    send_guest_vcpu_virq(v, VIRQ_TIMER);
}

/* Return secs after 00:00:00 localtime, 1 January, 1970. */
unsigned long get_localtime(struct domain *d)
{
    return wc_sec + (wc_nsec + NOW()) / 1000000000ULL 
        + d->time_offset_seconds;
}

/* "cmos_utc_offset" is the difference between UTC time and CMOS time. */
static long cmos_utc_offset; /* in seconds */

int time_suspend(void)
{
    if ( smp_processor_id() == 0 )
    {
        cmos_utc_offset = -get_cmos_time();
        cmos_utc_offset += (wc_sec + (wc_nsec + NOW()) / 1000000000ULL);
    }

    /* Better to cancel calibration timer for accuracy. */
    kill_timer(&this_cpu(cpu_time).calibration_timer);

    return 0;
}

int time_resume(void)
{
    u64 tmp = init_pit_and_calibrate_tsc();

    set_time_scale(&this_cpu(cpu_time).tsc_scale, tmp);

    resume_platform_timer();

    do_settime(get_cmos_time() + cmos_utc_offset, 0, read_platform_stime());

    init_percpu_time();

    if ( !is_idle_vcpu(current) )
        update_vcpu_system_time(current);

    return 0;
}

int dom0_pit_access(struct ioreq *ioreq)
{
    /* Is Xen using Channel 2? Then disallow direct dom0 access. */
    if ( plt_src.read_counter == read_pit_count )
        return 0;

    switch ( ioreq->addr )
    {
    case PIT_CH2:
        if ( ioreq->dir == IOREQ_READ )
            ioreq->data = inb(PIT_CH2);
        else
            outb(ioreq->data, PIT_CH2);
        return 1;

    case PIT_MODE:
        if ( ioreq->dir == IOREQ_READ )
            return 0; /* urk! */
        switch ( ioreq->data & 0xc0 )
        {
        case 0xc0: /* Read Back */
            if ( ioreq->data & 0x08 )    /* Select Channel 2? */
                outb(ioreq->data & 0xf8, PIT_MODE);
            if ( !(ioreq->data & 0x06) ) /* Select Channel 0/1? */
                return 1; /* no - we're done */
            /* Filter Channel 2 and reserved bit 0. */
            ioreq->data &= ~0x09;
            return 0; /* emulate ch0/1 readback */
        case 0x80: /* Select Counter 2 */
            outb(ioreq->data, PIT_MODE);
            return 1;
        }

    case 0x61:
        if ( ioreq->dir == IOREQ_READ )
            ioreq->data = inb(0x61);
        else
            outb((inb(0x61) & ~3) | (ioreq->data & 3), 0x61);
        return 1;
    }

    return 0;
}

struct tm wallclock_time(void)
{
    uint64_t seconds;

    if ( !wc_sec )
        return (struct tm) { 0 };

    seconds = NOW() + (wc_sec * 1000000000ull) + wc_nsec;
    do_div(seconds, 1000000000);
    return gmtime(seconds);
}

/*
 * Local variables:
 * mode: C
 * c-set-style: "BSD"
 * c-basic-offset: 4
 * tab-width: 4
 * indent-tabs-mode: nil
 * End:
 */