/*      $NetBSD: subr_time_arith.c,v 1.3 2025/04/01 23:14:23 riastradh Exp $    */

/*-
* Copyright (c) 2000, 2004, 2005, 2007, 2008, 2009, 2020
*     The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Christopher G. Demetriou, by Andrew Doran, and by Jason R. Thorpe.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
*    notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
*    notice, this list of conditions and the following disclaimer in the
*    documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/

/*
* Copyright (c) 1982, 1986, 1989, 1993
*      The Regents of the University of California.  All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
*    notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
*    notice, this list of conditions and the following disclaimer in the
*    documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
*    may be used to endorse or promote products derived from this software
*    without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*      @(#)kern_clock.c        8.5 (Berkeley) 1/21/94
*      @(#)kern_time.c 8.4 (Berkeley) 5/26/95
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: subr_time_arith.c,v 1.3 2025/04/01 23:14:23 riastradh Exp $");

#include <sys/types.h>

#include <sys/errno.h>
#include <sys/time.h>
#include <sys/timearith.h>

#if defined(_KERNEL)

#include <sys/kernel.h>
#include <sys/systm.h>

#include <machine/limits.h>

#elif defined(_TIME_TESTING)

#include <assert.h>
#include <limits.h>
#include <stdbool.h>

extern int hz;
extern int tick;

#define KASSERT         assert
#define MIN(X, Y)       ((X) < (Y) ? (X) : (Y))

#endif

/*
* Compute number of ticks in the specified amount of time.
*/
int
tvtohz(const struct timeval *tv)
{
       unsigned long ticks;
       long sec, usec;

       /*
        * If the number of usecs in the whole seconds part of the time
        * difference fits in a long, then the total number of usecs will
        * fit in an unsigned long.  Compute the total and convert it to
        * ticks, rounding up and adding 1 to allow for the current tick
        * to expire.  Rounding also depends on unsigned long arithmetic
        * to avoid overflow.
        *
        * Otherwise, if the number of ticks in the whole seconds part of
        * the time difference fits in a long, then convert the parts to
        * ticks separately and add, using similar rounding methods and
        * overflow avoidance.  This method would work in the previous
        * case, but it is slightly slower and assumes that hz is integral.
        *
        * Otherwise, round the time difference down to the maximum
        * representable value.
        *
        * If ints are 32-bit, then the maximum value for any timeout in
        * 10ms ticks is 248 days.
        */
       sec = tv->tv_sec;
       usec = tv->tv_usec;

       KASSERT(usec >= 0);
       KASSERT(usec < 1000000);

       /* catch overflows in conversion time_t->int */
       if (tv->tv_sec > INT_MAX)
               return INT_MAX;
       if (tv->tv_sec < 0)
               return 0;

       if (sec < 0 || (sec == 0 && usec == 0)) {
               /*
                * Would expire now or in the past.  Return 0 ticks.
                * This is different from the legacy tvhzto() interface,
                * and callers need to check for it.
                */
               ticks = 0;
       } else if (sec <= (LONG_MAX / 1000000))
               ticks = (((sec * 1000000) + (unsigned long)usec + (tick - 1))
                   / tick) + 1;
       else if (sec <= (LONG_MAX / hz))
               ticks = (sec * hz) +
                   (((unsigned long)usec + (tick - 1)) / tick) + 1;
       else
               ticks = LONG_MAX;

       if (ticks > INT_MAX)
               ticks = INT_MAX;

       return ((int)ticks);
}

/*
* Check that a proposed value to load into the .it_value or
* .it_interval part of an interval timer is acceptable, and
* fix it to have at least minimal value (i.e. if it is less
* than the resolution of the clock, round it up.). We don't
* timeout the 0,0 value because this means to disable the
* timer or the interval.
*/
int
itimerfix(struct timeval *tv)
{

       if (tv->tv_usec < 0 || tv->tv_usec >= 1000000)
               return EINVAL;
       if (tv->tv_sec < 0)
               return ETIMEDOUT;
       if (tv->tv_sec == 0 && tv->tv_usec != 0 && tv->tv_usec < tick)
               tv->tv_usec = tick;
       return 0;
}

int
itimespecfix(struct timespec *ts)
{

       if (ts->tv_nsec < 0 || ts->tv_nsec >= 1000000000)
               return EINVAL;
       if (ts->tv_sec < 0)
               return ETIMEDOUT;
       if (ts->tv_sec == 0 && ts->tv_nsec != 0 && ts->tv_nsec < tick * 1000)
               ts->tv_nsec = tick * 1000;
       return 0;
}

/*
* timespecaddok(tsp, usp)
*
*      True if tsp + usp can be computed without overflow, i.e., if it
*      is OK to do timespecadd(tsp, usp, ...).
*/
bool
timespecaddok(const struct timespec *tsp, const struct timespec *usp)
{
       enum { TIME_MIN = __type_min(time_t), TIME_MAX = __type_max(time_t) };
       time_t a = tsp->tv_sec;
       time_t b = usp->tv_sec;
       bool carry;

       /*
        * Caller is responsible for guaranteeing valid timespec
        * inputs.  Any user-controlled inputs must be validated or
        * adjusted.
        */
       KASSERT(tsp->tv_nsec >= 0);
       KASSERT(usp->tv_nsec >= 0);
       KASSERT(tsp->tv_nsec < 1000000000L);
       KASSERT(usp->tv_nsec < 1000000000L);
       __CTASSERT(1000000000L <= __type_max(long) - 1000000000L);

       /*
        * Fail if a + b + carry overflows TIME_MAX, or if a + b
        * overflows TIME_MIN because timespecadd adds the carry after
        * computing a + b.
        *
        * Break it into two mutually exclusive and exhaustive cases:
        * I. a >= 0
        * II. a < 0
        */
       carry = (tsp->tv_nsec + usp->tv_nsec >= 1000000000L);
       if (a >= 0) {
               /*
                * Case I: a >= 0.  If b < 0, then b + 1 <= 0, so
                *
                *      a + b + 1 <= a + 0 <= TIME_MAX,
                *
                * and
                *
                *      a + b >= 0 + b = b >= TIME_MIN,
                *
                * so this can't overflow.
                *
                * If b >= 0, then a + b + carry >= a + b >= 0, so
                * negative results and thus results below TIME_MIN are
                * impossible; we need only avoid
                *
                *      a + b + carry > TIME_MAX,
                *
                * which we will do by rejecting if
                *
                *      b > TIME_MAX - a - carry,
                *
                * which in turn is incidentally always false if b < 0
                * so we don't need extra logic to discriminate on the
                * b >= 0 and b < 0 cases.
                *
                * Since 0 <= a <= TIME_MAX, we know
                *
                *      0 <= TIME_MAX - a <= TIME_MAX,
                *
                * and hence
                *
                *      -1 <= TIME_MAX - a - 1 < TIME_MAX.
                *
                * So we can compute TIME_MAX - a - carry (i.e., either
                * TIME_MAX - a or TIME_MAX - a - 1) safely without
                * overflow.
                */
               if (b > TIME_MAX - a - carry)
                       return false;
       } else {
               /*
                * Case II: a < 0.  If b >= 0, then since a + 1 <= 0,
                * we have
                *
                *      a + b + 1 <= b <= TIME_MAX,
                *
                * and
                *
                *      a + b >= a >= TIME_MIN,
                *
                * so this can't overflow.
                *
                * If b < 0, then the intermediate a + b is negative
                * and the outcome a + b + 1 is nonpositive, so we need
                * only avoid
                *
                *      a + b < TIME_MIN,
                *
                * which we will do by rejecting if
                *
                *      a < TIME_MIN - b.
                *
                * (Reminder: The carry is added afterward in
                * timespecadd, so to avoid overflow it is not enough
                * to merely reject a + b + carry < TIME_MIN.)
                *
                * It is safe to compute the difference TIME_MIN - b
                * because b is negative, so the result lies in
                * (TIME_MIN, 0].
                */
               if (b < 0 && a < TIME_MIN - b)
                       return false;
       }

       return true;
}

/*
* timespecsubok(tsp, usp)
*
*      True if tsp - usp can be computed without overflow, i.e., if it
*      is OK to do timespecsub(tsp, usp, ...).
*/
bool
timespecsubok(const struct timespec *tsp, const struct timespec *usp)
{
       enum { TIME_MIN = __type_min(time_t), TIME_MAX = __type_max(time_t) };
       time_t a = tsp->tv_sec, b = usp->tv_sec;
       bool borrow;

       /*
        * Caller is responsible for guaranteeing valid timespec
        * inputs.  Any user-controlled inputs must be validated or
        * adjusted.
        */
       KASSERT(tsp->tv_nsec >= 0);
       KASSERT(usp->tv_nsec >= 0);
       KASSERT(tsp->tv_nsec < 1000000000L);
       KASSERT(usp->tv_nsec < 1000000000L);
       __CTASSERT(1000000000L <= __type_max(long) - 1000000000L);

       /*
        * Fail if a - b - borrow overflows TIME_MIN, or if a - b
        * overflows TIME_MAX because timespecsub subtracts the borrow
        * after computing a - b.
        *
        * Break it into two mutually exclusive and exhaustive cases:
        * I. a < 0
        * II. a >= 0
        */
       borrow = (tsp->tv_nsec - usp->tv_nsec < 0);
       if (a < 0) {
               /*
                * Case I: a < 0.  If b < 0, then -b - 1 >= 0, so
                *
                *      a - b - 1 >= a + 0 >= TIME_MIN,
                *
                * and, since a <= -1, provided that TIME_MIN <=
                * -TIME_MAX - 1 so that TIME_MAX <= -TIME_MIN - 1 (in
                * fact, equality holds, under the assumption of
                * two's-complement arithmetic),
                *
                *      a - b <= -1 - b = -b - 1 <= TIME_MAX,
                *
                * so this can't overflow.
                */
               __CTASSERT(TIME_MIN <= -TIME_MAX - 1);

               /*
                * If b >= 0, then a - b - borrow <= a - b < 0, so
                * positive results and thus results above TIME_MAX are
                * impossible; we need only avoid
                *
                *      a - b - borrow < TIME_MIN,
                *
                * which we will do by rejecting if
                *
                *      a < TIME_MIN + b + borrow.
                *
                * The right-hand side is safe to evaluate for any
                * values of b and borrow as long as TIME_MIN +
                * TIME_MAX + 1 <= TIME_MAX, i.e., TIME_MIN <= -1.
                * (Note: If time_t were unsigned, this would fail!)
                *
                * Note: Unlike Case I in timespecaddok, this criterion
                * does not work for b < 0, nor can the roles of a and
                * b in the inequality be reversed (e.g., -b < TIME_MIN
                * - a + borrow) without extra cases like checking for
                * b = TEST_MIN.
                */
               __CTASSERT(TIME_MIN < -1);
               if (b >= 0 && a < TIME_MIN + b + borrow)
                       return false;
       } else {
               /*
                * Case II: a >= 0.  If b >= 0, then
                *
                *      a - b <= a <= TIME_MAX,
                *
                * and, provided TIME_MIN <= -TIME_MAX - 1 (in fact,
                * equality holds, under the assumption of
                * two's-complement arithmetic)
                *
                *      a - b - 1 >= -b - 1 >= -TIME_MAX - 1 >= TIME_MIN,
                *
                * so this can't overflow.
                */
               __CTASSERT(TIME_MIN <= -TIME_MAX - 1);

               /*
                * If b < 0, then a - b >= a >= 0, so negative results
                * and thus results below TIME_MIN are impossible; we
                * need only avoid
                *
                *      a - b > TIME_MAX,
                *
                * which we will do by rejecting if
                *
                *      a > TIME_MAX + b.
                *
                * (Reminder: The borrow is subtracted afterward in
                * timespecsub, so to avoid overflow it is not enough
                * to merely reject a - b - borrow > TIME_MAX.)
                *
                * It is safe to compute the sum TIME_MAX + b because b
                * is negative, so the result lies in [0, TIME_MAX).
                */
               if (b < 0 && a > TIME_MAX + b)
                       return false;
       }

       return true;
}

static bool
timespec2nsok(const struct timespec *ts)
{

       return ts->tv_sec < INT64_MAX/1000000000 ||
           (ts->tv_sec == INT64_MAX/1000000000 &&
               ts->tv_nsec <= INT64_MAX - (INT64_MAX/1000000000)*1000000000);
}

/*
* itimer_transition(it, now, next, &overruns)
*
*      Given:
*
*      - it: the current state of an itimer (it_value = last expiry
*        time, it_interval = periodic rescheduling interval), and
*
*      - now: the current time on the itimer's clock;
*
*      compute:
*
*      - next: the next time the itimer should be scheduled for, and
*      - overruns: the number of overruns if we're firing late.
*
*      XXX This should maybe also say whether the itimer should expire
*      at all.
*/
void
itimer_transition(const struct itimerspec *restrict it,
   const struct timespec *restrict now,
   struct timespec *restrict next,
   int *restrict overrunsp)
{
       int64_t last_val, next_val, interval, remainder, now_ns;
       int backwards;

       /*
        * Zero the outputs so we can test assertions in userland
        * without undefined behaviour.
        */
       timespecclear(next);
       *overrunsp = 0;

       /*
        * Paranoia: Caller should guarantee this.
        */
       if (!timespecisset(&it->it_interval)) {
               timespecclear(next);
               return;
       }

       /* Did the clock wind backwards? */
       backwards = (timespeccmp(&it->it_value, now, >));

       /* Valid value and interval guaranteed by itimerfix. */
       KASSERT(it->it_value.tv_sec >= 0);
       KASSERT(it->it_value.tv_nsec < 1000000000);
       KASSERT(it->it_interval.tv_sec >= 0);
       KASSERT(it->it_interval.tv_nsec < 1000000000);

       /* Nonnegative interval guaranteed by itimerfix.  */
       KASSERT(it->it_interval.tv_sec >= 0);
       KASSERT(it->it_interval.tv_nsec >= 0);

       /* Handle the easy case of non-overflown timers first. */
       if (__predict_true(!backwards)) {
               if (__predict_false(!timespecaddok(&it->it_value,
                           &it->it_interval)))
                       goto overflow;
               timespecadd(&it->it_value, &it->it_interval, next);
               if (__predict_true(timespeccmp(now, next, <)))
                       return;
       }

       /*
        * If we can't represent the input as a number of nanoseconds,
        * bail.  This is good up to the year 2262, if we start
        * counting from 1970 (2^63 nanoseconds ~ 292 years).
        */
       if (__predict_false(!timespec2nsok(now)) ||
           __predict_false(!timespec2nsok(&it->it_value)) ||
           __predict_false(!timespec2nsok(&it->it_interval)))
               goto overflow;

       now_ns = timespec2ns(now);
       last_val = timespec2ns(&it->it_value);
       interval = timespec2ns(&it->it_interval);

       KASSERT(now_ns >= 0);
       KASSERT(last_val >= 0);
       KASSERT(interval >= 0);

       /*
        *            now [backwards]         overruns    now [forwards]
        *           |                      v    v    v  |
        * |--+----+-*--x----+----+----|----+----+----+--*-x----+-->
        *            \/               |               \/
        *         remainder        last_val        remainder
        *     (zero or negative)                (zero or positive)
        *
        * Set next_val to last_value + k*interval for some k.
        *
        * The interval is always positive, and division in C
        * truncates, so dividing a positive duration by the interval
        * always gives zero or a positive remainder, and dividing a
        * negative duration by the interval always gives zero or a
        * negative remainder.  Hence:
        *
        * - If now_ns < last_val -- which happens iff backwards, i.e.,
        *   the clock was wound backwards -- then remainder is zero or
        *   negative, so subtracting it stays in place or moves
        *   forward in time, and thus this finds the _earliest_ value
        *   that is not earlier than now_ns.  We will advance this by
        *   one more interval if we are already firing exactly on the
        *   interval to find the earliest value _after_ now_ns.
        *
        * - If now_ns > last_val -- which happens iff !backwards,
        *   i.e., the clock ran fast -- then remainder is zero or
        *   positive positive, so this finds the _latest_ value not
        *   later than now_ns.  We will always advance this by one
        *   more interval to find the earliest value _after_ now_ns.
        *   We will also count overflows.
        *
        * (now_ns == last_val is not possible at this point because it
        * only happens if the addition of struct timespec would
        * overflow, and that is only possible when timespec2ns would
        * also overflow for at least one of the inputs.)
        */
       KASSERT(last_val != now_ns);
       remainder = (now_ns - last_val) % interval;
       next_val = now_ns - remainder;
       KASSERT((last_val - next_val) % interval == 0);
       if (backwards) {
               /*
                * If the clock was wound back to an exact multiple of
                * the interval, so next_val = now_ns, don't demand to
                * fire again in the same instant -- advance to the
                * next interval.  Overflow is not possible; proof is
                * asserted.
                */
               if (remainder == 0) {
                       KASSERT(now_ns < last_val);
                       KASSERT(next_val == now_ns);
                       KASSERT(last_val - next_val >= interval);
                       KASSERT(interval <= last_val - next_val);
                       KASSERT(next_val <= last_val - interval);
                       KASSERT(next_val <= INT64_MAX - interval);
                       next_val += interval;
               }
       } else {
               /*
                * next_val is the largest integer multiple of interval
                * not later than now_ns.  Count the number of full
                * intervals that were skipped (division should be
                * exact here), not counting any partial interval
                * between next_val and now_ns, as the number of
                * overruns.  Advance by one interval -- unless that
                * would overflow.
                */
               *overrunsp += MIN(INT_MAX - *overrunsp,
                   (next_val - last_val) / interval);
               if (__predict_false(next_val > INT64_MAX - interval))
                       goto overflow;
               next_val += interval;
       }

       next->tv_sec = next_val / 1000000000;
       next->tv_nsec = next_val % 1000000000;
       return;

overflow:
       next->tv_sec = 0;
       next->tv_nsec = 0;
}