/*      $NetBSD: intr.c,v 1.57 2025/04/06 01:58:22 riastradh Exp $      */

/*
* Copyright (c) 2008-2010, 2015 Antti Kantee.  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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: intr.c,v 1.57 2025/04/06 01:58:22 riastradh Exp $");

#include <sys/param.h>
#include <sys/atomic.h>
#include <sys/cpu.h>
#include <sys/kernel.h>
#include <sys/kmem.h>
#include <sys/kthread.h>
#include <sys/malloc.h>
#include <sys/intr.h>
#include <sys/timetc.h>

#include <rump-sys/kern.h>

#include <rump/rumpuser.h>

/*
* Interrupt simulator.  It executes hardclock() and softintrs.
*/

#define SI_MPSAFE 0x01
#define SI_KILLME 0x02

struct softint_percpu;
struct softint {
       void (*si_func)(void *);
       void *si_arg;
       int si_flags;
       int si_level;

       struct softint_percpu *si_entry; /* [0,ncpu-1] */
};

struct softint_percpu {
       struct softint *sip_parent;
       bool sip_onlist;
       bool sip_onlist_cpu;

       TAILQ_ENTRY(softint_percpu) sip_entries;        /* scheduled */
       TAILQ_ENTRY(softint_percpu) sip_entries_cpu;    /* to be scheduled */
};

struct softint_lev {
       struct rumpuser_cv *si_cv;
       TAILQ_HEAD(, softint_percpu) si_pending;
};

static TAILQ_HEAD(, softint_percpu) sicpupending \
   = TAILQ_HEAD_INITIALIZER(sicpupending);
static struct rumpuser_mtx *sicpumtx;
static struct rumpuser_cv *sicpucv;

kcondvar_t lbolt; /* Oh Kath Ra */

static int ncpu_final;

void noclock(void); void noclock(void) {return;}
__strong_alias(sched_schedclock,noclock);
__strong_alias(cpu_initclocks,noclock);
__strong_alias(addupc_intr,noclock);
__strong_alias(sched_tick,noclock);
__strong_alias(setstatclockrate,noclock);

/*
* clock "interrupt"
*/
static void
doclock(void *noarg)
{
       struct timespec thetick, curclock;
       struct clockframe *clkframe;
       int64_t sec;
       long nsec;
       int error;
       struct cpu_info *ci = curcpu();

       error = rumpuser_clock_gettime(RUMPUSER_CLOCK_ABSMONO, &sec, &nsec);
       if (error)
               panic("clock: cannot get monotonic time");

       curclock.tv_sec = sec;
       curclock.tv_nsec = nsec;
       thetick.tv_sec = 0;
       thetick.tv_nsec = 1000000000/hz;

       /* generate dummy clockframe for hardclock to consume */
       clkframe = rump_cpu_makeclockframe();

       for (;;) {
               int lbolt_ticks = 0;

               hardclock(clkframe);
               if (CPU_IS_PRIMARY(ci)) {
                       if (++lbolt_ticks >= hz) {
                               lbolt_ticks = 0;
                               cv_broadcast(&lbolt);
                       }
               }

               error = rumpuser_clock_sleep(RUMPUSER_CLOCK_ABSMONO,
                   curclock.tv_sec, curclock.tv_nsec);
               if (error) {
                       panic("rumpuser_clock_sleep failed with error %d",
                           error);
               }
               timespecadd(&curclock, &thetick, &curclock);
       }
}

/*
* Soft interrupt execution thread.  This thread is pinned to the
* same CPU that scheduled the interrupt, so we don't need to do
* lock against si_lvl.
*/
static void
sithread(void *arg)
{
       struct softint_percpu *sip;
       struct softint *si;
       void (*func)(void *) = NULL;
       void *funarg;
       bool mpsafe;
       int mylevel = (uintptr_t)arg;
       struct softint_lev *si_lvlp, *si_lvl;
       struct cpu_data *cd = &curcpu()->ci_data;

       si_lvlp = cd->cpu_softcpu;
       si_lvl = &si_lvlp[mylevel];

       for (;;) {
               if (!TAILQ_EMPTY(&si_lvl->si_pending)) {
                       sip = TAILQ_FIRST(&si_lvl->si_pending);
                       si = sip->sip_parent;

                       func = si->si_func;
                       funarg = si->si_arg;
                       mpsafe = si->si_flags & SI_MPSAFE;

                       sip->sip_onlist = false;
                       TAILQ_REMOVE(&si_lvl->si_pending, sip, sip_entries);
                       if (si->si_flags & SI_KILLME) {
                               softint_disestablish(si);
                               continue;
                       }
               } else {
                       rump_schedlock_cv_wait(si_lvl->si_cv);
                       continue;
               }

               if (!mpsafe)
                       KERNEL_LOCK(1, curlwp);
               func(funarg);
               if (!mpsafe)
                       KERNEL_UNLOCK_ONE(curlwp);
       }

       panic("sithread unreachable");
}

/*
* Helper for softint_schedule_cpu()
*/
static void
sithread_cpu_bouncer(void *arg)
{
       struct lwp *me;

       me = curlwp;
       me->l_pflag |= LP_BOUND;

       rump_unschedule();
       for (;;) {
               struct softint_percpu *sip;
               struct softint *si;
               struct cpu_info *ci;
               unsigned int cidx;

               rumpuser_mutex_enter_nowrap(sicpumtx);
               while (TAILQ_EMPTY(&sicpupending)) {
                       rumpuser_cv_wait_nowrap(sicpucv, sicpumtx);
               }
               sip = TAILQ_FIRST(&sicpupending);
               TAILQ_REMOVE(&sicpupending, sip, sip_entries_cpu);
               sip->sip_onlist_cpu = false;
               rumpuser_mutex_exit(sicpumtx);

               /*
                * ok, now figure out which cpu we need the softint to
                * be handled on
                */
               si = sip->sip_parent;
               cidx = sip - si->si_entry;
               ci = cpu_lookup(cidx);
               me->l_target_cpu = ci;

               /* schedule ourselves there, and then schedule the softint */
               rump_schedule();
               KASSERT(curcpu() == ci);
               softint_schedule(si);
               rump_unschedule();
       }
       panic("sithread_cpu_bouncer unreasonable");
}

static kmutex_t sithr_emtx;
static unsigned int sithr_est;
static int sithr_canest;

/*
* Create softint handler threads when the softint for each respective
* level is established for the first time.  Most rump kernels don't
* need at least half of the softint levels, so on-demand saves bootstrap
* time and memory resources.  Note, though, that this routine may be
* called before it's possible to call kthread_create().  Creation of
* those softints (SOFTINT_CLOCK, as of writing this) will be deferred
* to until softint_init() is called for the main CPU.
*/
static void
sithread_establish(int level)
{
       int docreate, rv;
       int lvlbit = 1<<level;
       int i;

       KASSERT((level & ~SOFTINT_LVLMASK) == 0);
       if (__predict_true(sithr_est & lvlbit))
               return;

       mutex_enter(&sithr_emtx);
       docreate = (sithr_est & lvlbit) == 0 && sithr_canest;
       sithr_est |= lvlbit;
       mutex_exit(&sithr_emtx);

       if (docreate) {
               for (i = 0; i < ncpu_final; i++) {
                       if ((rv = kthread_create(PRI_NONE,
                           KTHREAD_MPSAFE | KTHREAD_INTR,
                           cpu_lookup(i), sithread, (void *)(uintptr_t)level,
                           NULL, "rsi%d/%d", i, level)) != 0)
                               panic("softint thread create failed: %d", rv);
               }
       }
}

void
rump_intr_init(int numcpu)
{

       cv_init(&lbolt, "oh kath ra");
       mutex_init(&sithr_emtx, MUTEX_DEFAULT, IPL_NONE);
       ncpu_final = numcpu;
}

void
softint_init(struct cpu_info *ci)
{
       struct cpu_data *cd = &ci->ci_data;
       struct softint_lev *slev;
       int rv, i;

       if (!rump_threads)
               return;

       slev = kmem_alloc(sizeof(struct softint_lev) * SOFTINT_COUNT, KM_SLEEP);
       for (i = 0; i < SOFTINT_COUNT; i++) {
               rumpuser_cv_init(&slev[i].si_cv);
               TAILQ_INIT(&slev[i].si_pending);
       }
       cd->cpu_softcpu = slev;

       /* overloaded global init ... */
       /* XXX: should be done the last time we are called */
       if (ci->ci_index == 0) {
               int sithr_swap;

               /* pretend that we have our own for these */
               stathz = 1;
               schedhz = 1;
               profhz = 1;

               initclocks();

               /* create deferred softint threads */
               mutex_enter(&sithr_emtx);
               sithr_swap = sithr_est;
               sithr_est = 0;
               sithr_canest = 1;
               mutex_exit(&sithr_emtx);
               for (i = 0; i < SOFTINT_COUNT; i++) {
                       if (sithr_swap & (1<<i))
                               sithread_establish(i);
               }
       }

       /* well, not really a "soft" interrupt ... */
       if ((rv = kthread_create(PRI_NONE, KTHREAD_MPSAFE,
           ci, doclock, NULL, NULL, "rumpclk%d", ci->ci_index)) != 0)
               panic("clock thread creation failed: %d", rv);

       /* not one either, but at least a softint helper */
       if (CPU_IS_PRIMARY(ci)) {
               rumpuser_mutex_init(&sicpumtx, RUMPUSER_MTX_SPIN);
               rumpuser_cv_init(&sicpucv);
               if ((rv = kthread_create(PRI_NONE, KTHREAD_MPSAFE,
                   NULL, sithread_cpu_bouncer, NULL, NULL, "sipbnc")) != 0)
                       panic("softint cpu bouncer creation failed: %d", rv);
       }
}

void *
softint_establish(u_int flags, void (*func)(void *), void *arg)
{
       struct softint *si;
       struct softint_percpu *sip;
       int level = flags & SOFTINT_LVLMASK;
       int i;

       si = malloc(sizeof(*si), M_TEMP, M_WAITOK);
       si->si_func = func;
       si->si_arg = arg;
       si->si_flags = flags & SOFTINT_MPSAFE ? SI_MPSAFE : 0;
       si->si_level = level;
       KASSERT(si->si_level < SOFTINT_COUNT);
       si->si_entry = malloc(sizeof(*si->si_entry) * ncpu_final,
           M_TEMP, M_WAITOK | M_ZERO);
       for (i = 0; i < ncpu_final; i++) {
               sip = &si->si_entry[i];
               sip->sip_parent = si;
       }
       sithread_establish(level);

       return si;
}

static struct softint_percpu *
sitosip(struct softint *si, struct cpu_info *ci)
{

       return &si->si_entry[ci->ci_index];
}

/*
* Soft interrupts bring two choices.  If we are running with thread
* support enabled, defer execution, otherwise execute in place.
*/

void
softint_schedule(void *arg)
{
       struct softint *si = arg;
       struct cpu_info *ci = curcpu();
       struct softint_percpu *sip = sitosip(si, ci);
       struct cpu_data *cd = &ci->ci_data;
       struct softint_lev *si_lvl = cd->cpu_softcpu;

       if (!rump_threads) {
               si->si_func(si->si_arg);
       } else {
               if (!sip->sip_onlist) {
                       TAILQ_INSERT_TAIL(&si_lvl[si->si_level].si_pending,
                           sip, sip_entries);
                       sip->sip_onlist = true;
               }
       }
}

/*
* Like softint_schedule(), except schedule softint to be handled on
* the core designated by ci_tgt instead of the core the call is made on.
*
* Unlike softint_schedule(), the performance is not important
* (unless ci_tgt == curcpu): high-performance rump kernel I/O stacks
* should arrange data to already be on the right core at the driver
* layer.
*/
void
softint_schedule_cpu(void *arg, struct cpu_info *ci_tgt)
{
       struct softint *si = arg;
       struct cpu_info *ci_cur = curcpu();
       struct softint_percpu *sip;

       KASSERT(rump_threads);

       /* preferred case (which can be optimized some day) */
       if (ci_cur == ci_tgt) {
               softint_schedule(si);
               return;
       }

       /*
        * no?  then it's softint turtles all the way down
        */

       sip = sitosip(si, ci_tgt);
       rumpuser_mutex_enter_nowrap(sicpumtx);
       if (sip->sip_onlist_cpu) {
               rumpuser_mutex_exit(sicpumtx);
               return;
       }
       TAILQ_INSERT_TAIL(&sicpupending, sip, sip_entries_cpu);
       sip->sip_onlist_cpu = true;
       rumpuser_cv_signal(sicpucv);
       rumpuser_mutex_exit(sicpumtx);
}

/*
* flimsy disestablish: should wait for softints to finish.
*/
void
softint_disestablish(void *cook)
{
       struct softint *si = cook;
       int i;

       for (i = 0; i < ncpu_final; i++) {
               struct softint_percpu *sip;

               sip = &si->si_entry[i];
               if (sip->sip_onlist) {
                       si->si_flags |= SI_KILLME;
                       return;
               }
       }
       free(si->si_entry, M_TEMP);
       free(si, M_TEMP);
}

void
rump_softint_run(struct cpu_info *ci)
{
       struct cpu_data *cd = &ci->ci_data;
       struct softint_lev *si_lvl = cd->cpu_softcpu;
       int i;

       if (!rump_threads)
               return;

       for (i = 0; i < SOFTINT_COUNT; i++) {
               if (!TAILQ_EMPTY(&si_lvl[i].si_pending))
                       rump_schedlock_cv_signal(ci, si_lvl[i].si_cv);
       }
}

bool
cpu_intr_p(void)
{

       return false;
}