/*      $NetBSD: subr_cpu.c,v 1.23 2025/06/23 22:50:23 ad Exp $ */

/*-
* Copyright (c) 2007, 2008, 2009, 2010, 2012, 2019, 2020
*     The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Andrew Doran.
*
* 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)2007 YAMAMOTO Takashi,
* 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 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 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.
*/

/*
* CPU related routines shared with rump.
*/

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: subr_cpu.c,v 1.23 2025/06/23 22:50:23 ad Exp $");

#include <sys/param.h>
#include <sys/atomic.h>
#include <sys/systm.h>
#include <sys/sched.h>
#include <sys/conf.h>
#include <sys/cpu.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/kmem.h>

static void     cpu_topology_fake1(struct cpu_info *);

kmutex_t        cpu_lock                __cacheline_aligned;
int             ncpu                    __read_mostly;
int             ncpuonline              __read_mostly;
bool            mp_online               __read_mostly;
static bool     cpu_topology_present    __read_mostly;
static bool     cpu_topology_haveslow   __read_mostly;
int64_t         cpu_counts[CPU_COUNT_MAX];

/* An array of CPUs.  There are ncpu entries. */
struct cpu_info **cpu_infos             __read_mostly;

/* Note: set on mi_cpu_attach() and idle_loop(). */
kcpuset_t *     kcpuset_attached        __read_mostly   = NULL;
kcpuset_t *     kcpuset_running         __read_mostly   = NULL;

static char cpu_model[128];

/*
* mi_cpu_init: early initialisation of MI CPU related structures.
*
* Note: may not block and memory allocator is not yet available.
*/
void
mi_cpu_init(void)
{
       struct cpu_info *ci;

       mutex_init(&cpu_lock, MUTEX_DEFAULT, IPL_NONE);

       kcpuset_create(&kcpuset_attached, true);
       kcpuset_create(&kcpuset_running, true);
       kcpuset_set(kcpuset_running, 0);

       ci = curcpu();
       cpu_topology_fake1(ci);
}

int
cpu_setmodel(const char *fmt, ...)
{
       int len;
       va_list ap;

       va_start(ap, fmt);
       len = vsnprintf(cpu_model, sizeof(cpu_model), fmt, ap);
       va_end(ap);
       return len;
}

const char *
cpu_getmodel(void)
{
       return cpu_model;
}

bool
cpu_softintr_p(void)
{

       return (curlwp->l_pflag & LP_INTR) != 0;
}

bool
curcpu_stable(void)
{
       struct lwp *const l = curlwp;
       const int pflag = l->l_pflag;
       const int nopreempt = l->l_nopreempt;

       /*
        * - Softints (LP_INTR) never migrate between CPUs.
        * - Bound lwps (LP_BOUND), either kthreads created bound to
        *   a CPU or any lwps bound with curlwp_bind, never migrate.
        * - If kpreemption is disabled, the lwp can't migrate.
        * - If we're in interrupt context, preemption is blocked.
        *
        * We combine the LP_INTR, LP_BOUND, and l_nopreempt test into
        * a single predicted-true branch so this is cheap to assert in
        * most contexts where it will be used, then fall back to
        * calling the full kpreempt_disabled() and cpu_intr_p() as
        * subroutines.
        *
        * XXX Is cpu_intr_p redundant with kpreempt_disabled?
        */
       return __predict_true(((pflag & (LP_INTR|LP_BOUND)) | nopreempt)
               != 0) ||
           kpreempt_disabled() ||
           cpu_intr_p();
}

/*
* Collect CPU topology information as each CPU is attached.  This can be
* called early during boot, so we need to be careful what we do.
*/
void
cpu_topology_set(struct cpu_info *ci, u_int package_id, u_int core_id,
   u_int smt_id, u_int numa_id)
{
       enum cpu_rel rel;

       cpu_topology_present = true;
       ci->ci_package_id = package_id;
       ci->ci_core_id = core_id;
       ci->ci_smt_id = smt_id;
       ci->ci_numa_id = numa_id;
       for (rel = 0; rel < __arraycount(ci->ci_sibling); rel++) {
               ci->ci_sibling[rel] = ci;
               ci->ci_nsibling[rel] = 1;
       }
}

/*
* Collect CPU relative speed
*/
void
cpu_topology_setspeed(struct cpu_info *ci, bool slow)
{

       cpu_topology_haveslow |= slow;
       ci->ci_is_slow = slow;
}

/*
* Link a CPU into the given circular list.
*/
static void
cpu_topology_link(struct cpu_info *ci, struct cpu_info *ci2, enum cpu_rel rel)
{
       struct cpu_info *ci3;

       /* Walk to the end of the existing circular list and append. */
       for (ci3 = ci2;; ci3 = ci3->ci_sibling[rel]) {
               ci3->ci_nsibling[rel]++;
               if (ci3->ci_sibling[rel] == ci2) {
                       break;
               }
       }
       ci->ci_sibling[rel] = ci2;
       ci3->ci_sibling[rel] = ci;
       ci->ci_nsibling[rel] = ci3->ci_nsibling[rel];
}

/*
* Print out the topology lists.
*/
static void
cpu_topology_dump(void)
{
       CPU_INFO_ITERATOR cii;
       struct cpu_info *ci, *ci2;
       const char *names[] = { "core", "pkg", "1st" };
       enum cpu_rel rel;
       int i;

       CTASSERT(__arraycount(names) >= __arraycount(ci->ci_sibling));
       if (ncpu == 1) {
               return;
       }

       for (CPU_INFO_FOREACH(cii, ci)) {
               if (cpu_topology_haveslow)
                       aprint_debug("%s ", ci->ci_is_slow ? "slow" : "fast");
               for (rel = 0; rel < __arraycount(ci->ci_sibling); rel++) {
                       aprint_debug("%s has %d %s siblings:", cpu_name(ci),
                           ci->ci_nsibling[rel], names[rel]);
                       ci2 = ci->ci_sibling[rel];
                       i = 0;
                       do {
                               aprint_debug(" %s", cpu_name(ci2));
                               ci2 = ci2->ci_sibling[rel];
                       } while (++i < 64 && ci2 != ci->ci_sibling[rel]);
                       if (i == 64) {
                               aprint_debug(" GAVE UP");
                       }
                       aprint_debug("\n");
               }
               aprint_debug("%s first in package: %s\n", cpu_name(ci),
                   cpu_name(ci->ci_package1st));
       }
}

/*
* Fake up topology info if we have none, or if what we got was bogus.
* Used early in boot, and by cpu_topology_fake().
*/
static void
cpu_topology_fake1(struct cpu_info *ci)
{
       enum cpu_rel rel;

       for (rel = 0; rel < __arraycount(ci->ci_sibling); rel++) {
               ci->ci_sibling[rel] = ci;
               ci->ci_nsibling[rel] = 1;
       }
       if (!cpu_topology_present) {
               ci->ci_package_id = cpu_index(ci);
       }
       ci->ci_schedstate.spc_flags |=
           (SPCF_CORE1ST | SPCF_PACKAGE1ST | SPCF_1STCLASS);
       ci->ci_package1st = ci;
       if (!cpu_topology_haveslow) {
               ci->ci_is_slow = false;
       }
}

/*
* Fake up topology info if we have none, or if what we got was bogus.
* Don't override ci_package_id, etc, if cpu_topology_present is set.
* MD code also uses these.
*/
static void
cpu_topology_fake(void)
{
       CPU_INFO_ITERATOR cii;
       struct cpu_info *ci;

       for (CPU_INFO_FOREACH(cii, ci)) {
               cpu_topology_fake1(ci);
               /* Undo (early boot) flag set so everything links OK. */
               ci->ci_schedstate.spc_flags &=
                   ~(SPCF_CORE1ST | SPCF_PACKAGE1ST | SPCF_1STCLASS);
       }
}

/*
* Fix up basic CPU topology info.  Right now that means attach each CPU to
* circular lists of its siblings in the same core, and in the same package.
*/
void
cpu_topology_init(void)
{
       CPU_INFO_ITERATOR cii, cii2;
       struct cpu_info *ci, *ci2, *ci3;
       u_int minsmt, mincore;

       if (!cpu_topology_present) {
               cpu_topology_fake();
               goto linkit;
       }

       /* Find siblings in same core and package. */
       for (CPU_INFO_FOREACH(cii, ci)) {
               ci->ci_schedstate.spc_flags &=
                   ~(SPCF_CORE1ST | SPCF_PACKAGE1ST | SPCF_1STCLASS);
               for (CPU_INFO_FOREACH(cii2, ci2)) {
                       /* Avoid bad things happening. */
                       if (ci2->ci_package_id == ci->ci_package_id &&
                           ci2->ci_core_id == ci->ci_core_id &&
                           ci2->ci_smt_id == ci->ci_smt_id &&
                           ci2 != ci) {
#ifdef DEBUG
                               printf("cpu%u %p pkg %u core %u smt %u same as "
                                      "cpu%u %p pkg %u core %u smt %u\n",
                                      cpu_index(ci), ci, ci->ci_package_id,
                                      ci->ci_core_id, ci->ci_smt_id,
                                      cpu_index(ci2), ci2, ci2->ci_package_id,
                                      ci2->ci_core_id, ci2->ci_smt_id);
#endif
                               printf("cpu_topology_init: info bogus, "
                                   "faking it\n");
                               cpu_topology_fake();
                               goto linkit;
                       }
                       if (ci2 == ci ||
                           ci2->ci_package_id != ci->ci_package_id) {
                               continue;
                       }
                       /* Find CPUs in the same core. */
                       if (ci->ci_nsibling[CPUREL_CORE] == 1 &&
                           ci->ci_core_id == ci2->ci_core_id) {
                               cpu_topology_link(ci, ci2, CPUREL_CORE);
                       }
                       /* Find CPUs in the same package. */
                       if (ci->ci_nsibling[CPUREL_PACKAGE] == 1) {
                               cpu_topology_link(ci, ci2, CPUREL_PACKAGE);
                       }
                       if (ci->ci_nsibling[CPUREL_CORE] > 1 &&
                           ci->ci_nsibling[CPUREL_PACKAGE] > 1) {
                               break;
                       }
               }
       }

linkit:
       /* Identify lowest numbered SMT in each core. */
       for (CPU_INFO_FOREACH(cii, ci)) {
               ci2 = ci3 = ci;
               minsmt = ci->ci_smt_id;
               do {
                       if (ci2->ci_smt_id < minsmt) {
                               ci3 = ci2;
                               minsmt = ci2->ci_smt_id;
                       }
                       ci2 = ci2->ci_sibling[CPUREL_CORE];
               } while (ci2 != ci);
               ci3->ci_schedstate.spc_flags |= SPCF_CORE1ST;
       }

       /* Identify lowest numbered SMT in each package. */
       ci3 = NULL;
       for (CPU_INFO_FOREACH(cii, ci)) {
               if ((ci->ci_schedstate.spc_flags & SPCF_CORE1ST) == 0) {
                       continue;
               }
               ci2 = ci3 = ci;
               mincore = ci->ci_core_id;
               do {
                       if ((ci2->ci_schedstate.spc_flags &
                           SPCF_CORE1ST) != 0 &&
                           ci2->ci_core_id < mincore) {
                               ci3 = ci2;
                               mincore = ci2->ci_core_id;
                       }
                       ci2 = ci2->ci_sibling[CPUREL_PACKAGE];
               } while (ci2 != ci);

               if ((ci3->ci_schedstate.spc_flags & SPCF_PACKAGE1ST) != 0) {
                       /* Already identified - nothing more to do. */
                       continue;
               }
               ci3->ci_schedstate.spc_flags |= SPCF_PACKAGE1ST;

               /* Walk through all CPUs in package and point to first. */
               ci2 = ci3;
               do {
                       ci2->ci_package1st = ci3;
                       ci2->ci_sibling[CPUREL_PACKAGE1ST] = ci3;
                       ci2 = ci2->ci_sibling[CPUREL_PACKAGE];
               } while (ci2 != ci3);

               /* Now look for somebody else to link to. */
               for (CPU_INFO_FOREACH(cii2, ci2)) {
                       if ((ci2->ci_schedstate.spc_flags & SPCF_PACKAGE1ST)
                           != 0 && ci2 != ci3) {
                               cpu_topology_link(ci3, ci2, CPUREL_PACKAGE1ST);
                               break;
                       }
               }
       }

       /* Walk through all packages, starting with value of ci3 from above. */
       KASSERT(ci3 != NULL);
       ci = ci3;
       do {
               /* Walk through CPUs in the package and copy in PACKAGE1ST. */
               ci2 = ci;
               do {
                       ci2->ci_sibling[CPUREL_PACKAGE1ST] =
                           ci->ci_sibling[CPUREL_PACKAGE1ST];
                       ci2->ci_nsibling[CPUREL_PACKAGE1ST] =
                           ci->ci_nsibling[CPUREL_PACKAGE1ST];
                       ci2 = ci2->ci_sibling[CPUREL_PACKAGE];
               } while (ci2 != ci);
               ci = ci->ci_sibling[CPUREL_PACKAGE1ST];
       } while (ci != ci3);

       if (cpu_topology_haveslow) {
               /*
                * For asymmetric systems where some CPUs are slower than
                * others, mark first class CPUs for the scheduler.  This
                * conflicts with SMT right now so whinge if observed.
                */
               if (curcpu()->ci_nsibling[CPUREL_CORE] > 1) {
                       printf("cpu_topology_init: asymmetric & SMT??\n");
               }
               for (CPU_INFO_FOREACH(cii, ci)) {
                       if (!ci->ci_is_slow) {
                               ci->ci_schedstate.spc_flags |= SPCF_1STCLASS;
                       }
               }
       } else {
               /*
                * For any other configuration mark the 1st CPU in each
                * core as a first class CPU.
                */
               for (CPU_INFO_FOREACH(cii, ci)) {
                       if ((ci->ci_schedstate.spc_flags & SPCF_CORE1ST) != 0) {
                               ci->ci_schedstate.spc_flags |= SPCF_1STCLASS;
                       }
               }
       }

       cpu_topology_dump();
}

/*
* Adjust one count, for a counter that's NOT updated from interrupt
* context.  Hardly worth making an inline due to preemption stuff.
*/
void
cpu_count(enum cpu_count idx, int64_t delta)
{
       lwp_t *l = curlwp;
       KPREEMPT_DISABLE(l);
       l->l_cpu->ci_counts[idx] += delta;
       KPREEMPT_ENABLE(l);
}

/*
* Fetch fresh sum total for all counts.  Expensive - don't call often.
*
* If poll is true, the caller is okay with less recent values (but
* no more than 1/hz seconds old).  Where this is called very often that
* should be the case.
*
* This should be reasonably quick so that any value collected isn't totally
* out of whack.  It can also be called from interrupt context, so go to
* splvm() while summing the counters.  It's tempting to use a spin mutex
* here but this routine is called from DDB.
*/
void
cpu_count_sync(bool poll)
{
       CPU_INFO_ITERATOR cii;
       struct cpu_info *ci;
       int64_t sum[CPU_COUNT_MAX], *ptr;
       static int lasttick;
       int curtick, s;
       enum cpu_count i;

       KASSERT(sizeof(ci->ci_counts) == sizeof(cpu_counts));

       if (__predict_false(!mp_online)) {
               memcpy(cpu_counts, curcpu()->ci_counts, sizeof(cpu_counts));
               return;
       }

       s = splvm();
       curtick = getticks();
       if (poll && atomic_load_acquire(&lasttick) == curtick) {
               splx(s);
               return;
       }
       memset(sum, 0, sizeof(sum));
       curcpu()->ci_counts[CPU_COUNT_SYNC]++;
       for (CPU_INFO_FOREACH(cii, ci)) {
               ptr = ci->ci_counts;
               for (i = 0; i < CPU_COUNT_MAX; i += 8) {
                       sum[i+0] += ptr[i+0];
                       sum[i+1] += ptr[i+1];
                       sum[i+2] += ptr[i+2];
                       sum[i+3] += ptr[i+3];
                       sum[i+4] += ptr[i+4];
                       sum[i+5] += ptr[i+5];
                       sum[i+6] += ptr[i+6];
                       sum[i+7] += ptr[i+7];
               }
               KASSERT(i == CPU_COUNT_MAX);
       }
       memcpy(cpu_counts, sum, sizeof(cpu_counts));
       atomic_store_release(&lasttick, curtick);
       splx(s);
}

You are viewing proxied material from ftp.icm.edu.pl. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.