/*      $NetBSD: pfil.c,v 1.42 2022/08/16 04:35:57 knakahara Exp $      */

/*
* Copyright (c) 2013 Mindaugas Rasiukevicius <rmind at NetBSD org>
* Copyright (c) 1996 Matthew R. Green
* 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 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: pfil.c,v 1.42 2022/08/16 04:35:57 knakahara Exp $");

#if defined(_KERNEL_OPT)
#include "opt_net_mpsafe.h"
#endif

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/queue.h>
#include <sys/kmem.h>
#include <sys/psref.h>
#include <sys/cpu.h>

#include <net/if.h>
#include <net/pfil.h>

#define MAX_HOOKS       8

/* Func is either pfil_func_t or pfil_ifunc_t. */
typedef void            (*pfil_polyfunc_t)(void);

typedef struct {
       pfil_polyfunc_t pfil_func;
       void *          pfil_arg;
} pfil_hook_t;

typedef struct {
       pfil_hook_t     hooks[MAX_HOOKS];
       u_int           nhooks;
       struct psref_target psref;
} pfil_list_t;

typedef struct {
       pfil_list_t     *active;        /* lists[0] or lists[1] */
       pfil_list_t     lists[2];
} pfil_listset_t;

CTASSERT(PFIL_IN == 1);
CTASSERT(PFIL_OUT == 2);

struct pfil_head {
       pfil_listset_t  ph_in;
       pfil_listset_t  ph_out;
       pfil_listset_t  ph_ifaddr;
       pfil_listset_t  ph_ifevent;
       int             ph_type;
       void *          ph_key;
       LIST_ENTRY(pfil_head) ph_list;
};

static const int pfil_flag_cases[] = {
       PFIL_IN, PFIL_OUT
};

static LIST_HEAD(, pfil_head) pfil_head_list __read_mostly =
   LIST_HEAD_INITIALIZER(&pfil_head_list);

static kmutex_t pfil_mtx __cacheline_aligned;
static struct psref_class *pfil_psref_class __read_mostly;
#ifdef NET_MPSAFE
static pserialize_t pfil_psz;
#endif

void
pfil_init(void)
{
       mutex_init(&pfil_mtx, MUTEX_DEFAULT, IPL_NONE);
#ifdef NET_MPSAFE
       pfil_psz = pserialize_create();
#endif
       pfil_psref_class = psref_class_create("pfil", IPL_SOFTNET);
}

static inline void
pfil_listset_init(pfil_listset_t *pflistset)
{
       pflistset->active = &pflistset->lists[0];
       psref_target_init(&pflistset->active->psref, pfil_psref_class);
}

/*
* pfil_head_create: create and register a packet filter head.
*/
pfil_head_t *
pfil_head_create(int type, void *key)
{
       pfil_head_t *ph;

       if (pfil_head_get(type, key)) {
               return NULL;
       }
       ph = kmem_zalloc(sizeof(pfil_head_t), KM_SLEEP);
       ph->ph_type = type;
       ph->ph_key = key;

       pfil_listset_init(&ph->ph_in);
       pfil_listset_init(&ph->ph_out);
       pfil_listset_init(&ph->ph_ifaddr);
       pfil_listset_init(&ph->ph_ifevent);

       LIST_INSERT_HEAD(&pfil_head_list, ph, ph_list);
       return ph;
}

/*
* pfil_head_destroy: remove and destroy a packet filter head.
*/
void
pfil_head_destroy(pfil_head_t *pfh)
{
       LIST_REMOVE(pfh, ph_list);

       psref_target_destroy(&pfh->ph_in.active->psref, pfil_psref_class);
       psref_target_destroy(&pfh->ph_out.active->psref, pfil_psref_class);
       psref_target_destroy(&pfh->ph_ifaddr.active->psref, pfil_psref_class);
       psref_target_destroy(&pfh->ph_ifevent.active->psref, pfil_psref_class);

       kmem_free(pfh, sizeof(pfil_head_t));
}

/*
* pfil_head_get: returns the packer filter head for a given key.
*/
pfil_head_t *
pfil_head_get(int type, void *key)
{
       pfil_head_t *ph;

       LIST_FOREACH(ph, &pfil_head_list, ph_list) {
               if (ph->ph_type == type && ph->ph_key == key)
                       break;
       }
       return ph;
}

static pfil_listset_t *
pfil_hook_get(int dir, pfil_head_t *ph)
{
       switch (dir) {
       case PFIL_IN:
               return &ph->ph_in;
       case PFIL_OUT:
               return &ph->ph_out;
       case PFIL_IFADDR:
               return &ph->ph_ifaddr;
       case PFIL_IFNET:
               return &ph->ph_ifevent;
       }
       return NULL;
}

static int
pfil_list_add(pfil_listset_t *phlistset, pfil_polyfunc_t func, void *arg,
             int flags)
{
       u_int nhooks;
       pfil_list_t *newlist, *oldlist;
       pfil_hook_t *pfh;

       mutex_enter(&pfil_mtx);

       /* Check if we have a free slot. */
       nhooks = phlistset->active->nhooks;
       if (nhooks == MAX_HOOKS) {
               mutex_exit(&pfil_mtx);
               return ENOSPC;
       }
       KASSERT(nhooks < MAX_HOOKS);

       if (phlistset->active == &phlistset->lists[0]) {
               oldlist = &phlistset->lists[0];
               newlist = &phlistset->lists[1];
       } else{
               oldlist = &phlistset->lists[1];
               newlist = &phlistset->lists[0];
       }

       /* Make sure the hook is not already added. */
       for (u_int i = 0; i < nhooks; i++) {
               pfh = &oldlist->hooks[i];
               if (pfh->pfil_func == func && pfh->pfil_arg == arg) {
                       mutex_exit(&pfil_mtx);
                       return EEXIST;
               }
       }

       /* create new pfil_list_t copied from old */
       memcpy(newlist, oldlist, sizeof(pfil_list_t));
       psref_target_init(&newlist->psref, pfil_psref_class);

       /*
        * Finally, add the hook.  Note: for PFIL_IN we insert the hooks in
        * reverse order of the PFIL_OUT so that the same path is followed
        * in or out of the kernel.
        */
       if (flags & PFIL_IN) {
               /* XXX: May want to revisit this later; */
               size_t len = sizeof(pfil_hook_t) * nhooks;
               pfh = &newlist->hooks[0];
               memmove(&newlist->hooks[1], pfh, len);
       } else {
               pfh = &newlist->hooks[nhooks];
       }
       newlist->nhooks++;

       pfh->pfil_func = func;
       pfh->pfil_arg  = arg;

       /* switch from oldlist to newlist */
       atomic_store_release(&phlistset->active, newlist);
#ifdef NET_MPSAFE
       pserialize_perform(pfil_psz);
#endif
       mutex_exit(&pfil_mtx);

       /* Wait for all readers */
#ifdef NET_MPSAFE
       psref_target_destroy(&oldlist->psref, pfil_psref_class);
#endif

       return 0;
}

/*
* pfil_add_hook: add a function (hook) to the packet filter head.
* The possible flags are:
*
*      PFIL_IN         call on incoming packets
*      PFIL_OUT        call on outgoing packets
*      PFIL_ALL        call on all of the above
*/
int
pfil_add_hook(pfil_func_t func, void *arg, int flags, pfil_head_t *ph)
{
       int error = 0;

       KASSERT(func != NULL);
       KASSERT((flags & ~PFIL_ALL) == 0);

       ASSERT_SLEEPABLE();

       for (u_int i = 0; i < __arraycount(pfil_flag_cases); i++) {
               const int fcase = pfil_flag_cases[i];
               pfil_listset_t *phlistset;

               if ((flags & fcase) == 0) {
                       continue;
               }
               phlistset = pfil_hook_get(fcase, ph);
               error = pfil_list_add(phlistset, (pfil_polyfunc_t)func, arg,
                   flags);
               if (error && (error != EEXIST))
                       break;
       }
       if (error && (error != EEXIST)) {
               pfil_remove_hook(func, arg, flags, ph);
       }
       return error;
}

/*
* pfil_add_ihook: add an interface-event function (hook) to the packet
* filter head.  The possible flags are:
*
*      PFIL_IFADDR     call on interface reconfig (cmd is ioctl #)
*      PFIL_IFNET      call on interface attach/detach (cmd is PFIL_IFNET_*)
*/
int
pfil_add_ihook(pfil_ifunc_t func, void *arg, int flags, pfil_head_t *ph)
{
       pfil_listset_t *phlistset;

       KASSERT(func != NULL);
       KASSERT(flags == PFIL_IFADDR || flags == PFIL_IFNET);

       ASSERT_SLEEPABLE();

       phlistset = pfil_hook_get(flags, ph);
       return pfil_list_add(phlistset, (pfil_polyfunc_t)func, arg, flags);
}

/*
* pfil_list_remove: remove the hook from a specified list.
*/
static int
pfil_list_remove(pfil_listset_t *phlistset, pfil_polyfunc_t func, void *arg)
{
       u_int nhooks;
       pfil_list_t *oldlist, *newlist;

       mutex_enter(&pfil_mtx);

       /* create new pfil_list_t copied from old */
       if (phlistset->active == &phlistset->lists[0]) {
               oldlist = &phlistset->lists[0];
               newlist = &phlistset->lists[1];
       } else{
               oldlist = &phlistset->lists[1];
               newlist = &phlistset->lists[0];
       }
       memcpy(newlist, oldlist, sizeof(*newlist));
       psref_target_init(&newlist->psref, pfil_psref_class);

       nhooks = newlist->nhooks;
       for (u_int i = 0; i < nhooks; i++) {
               pfil_hook_t *last, *pfh = &newlist->hooks[i];

               if (pfh->pfil_func != func || pfh->pfil_arg != arg) {
                       continue;
               }
               if ((last = &newlist->hooks[nhooks - 1]) != pfh) {
                       memcpy(pfh, last, sizeof(pfil_hook_t));
               }
               newlist->nhooks--;

               /* switch from oldlist to newlist */
               atomic_store_release(&phlistset->active, newlist);
#ifdef NET_MPSAFE
               pserialize_perform(pfil_psz);
#endif
               mutex_exit(&pfil_mtx);

               /* Wait for all readers */
#ifdef NET_MPSAFE
               psref_target_destroy(&oldlist->psref, pfil_psref_class);
#endif

               return 0;
       }
       mutex_exit(&pfil_mtx);
       return ENOENT;
}

/*
* pfil_remove_hook: remove the hook from the packet filter head.
*/
int
pfil_remove_hook(pfil_func_t func, void *arg, int flags, pfil_head_t *ph)
{
       KASSERT((flags & ~PFIL_ALL) == 0);

       ASSERT_SLEEPABLE();

       for (u_int i = 0; i < __arraycount(pfil_flag_cases); i++) {
               const int fcase = pfil_flag_cases[i];
               pfil_listset_t *pflistset;

               if ((flags & fcase) == 0) {
                       continue;
               }
               pflistset = pfil_hook_get(fcase, ph);
               (void)pfil_list_remove(pflistset, (pfil_polyfunc_t)func, arg);
       }
       return 0;
}

int
pfil_remove_ihook(pfil_ifunc_t func, void *arg, int flags, pfil_head_t *ph)
{
       pfil_listset_t *pflistset;

       KASSERT(flags == PFIL_IFADDR || flags == PFIL_IFNET);

       ASSERT_SLEEPABLE();

       pflistset = pfil_hook_get(flags, ph);
       (void)pfil_list_remove(pflistset, (pfil_polyfunc_t)func, arg);
       return 0;
}

/*
* pfil_run_hooks: run the specified packet filter hooks.
*/
int
pfil_run_hooks(pfil_head_t *ph, struct mbuf **mp, ifnet_t *ifp, int dir)
{
       struct mbuf *m = mp ? *mp : NULL;
       pfil_listset_t *phlistset;
       pfil_list_t *phlist;
       struct psref psref;
       int s, bound;
       int ret = 0;

       KASSERT(dir == PFIL_IN || dir == PFIL_OUT);
       KASSERT(!cpu_intr_p());

       if (ph == NULL) {
               return ret;
       }

       if (__predict_false((phlistset = pfil_hook_get(dir, ph)) == NULL)) {
               return ret;
       }

       bound = curlwp_bind();
       s = pserialize_read_enter();
       phlist = atomic_load_consume(&phlistset->active);
       if (phlist->nhooks == 0) {
               pserialize_read_exit(s);
               curlwp_bindx(bound);
               return ret;
       }
       psref_acquire(&psref, &phlist->psref, pfil_psref_class);
       pserialize_read_exit(s);
       for (u_int i = 0; i < phlist->nhooks; i++) {
               pfil_hook_t *pfh = &phlist->hooks[i];
               pfil_func_t func = (pfil_func_t)pfh->pfil_func;

               ret = (*func)(pfh->pfil_arg, &m, ifp, dir);
               if (m == NULL || ret)
                       break;
       }
       psref_release(&psref, &phlist->psref, pfil_psref_class);
       curlwp_bindx(bound);

       if (mp) {
               *mp = m;
       }
       return ret;
}

static void
pfil_run_arg(pfil_listset_t *phlistset, u_long cmd, void *arg)
{
       pfil_list_t *phlist;
       struct psref psref;
       int s, bound;

       KASSERT(!cpu_intr_p());

       bound = curlwp_bind();
       s = pserialize_read_enter();
       phlist = atomic_load_consume(&phlistset->active);
       psref_acquire(&psref, &phlist->psref, pfil_psref_class);
       pserialize_read_exit(s);
       for (u_int i = 0; i < phlist->nhooks; i++) {
               pfil_hook_t *pfh = &phlist->hooks[i];
               pfil_ifunc_t func = (pfil_ifunc_t)pfh->pfil_func;
               (*func)(pfh->pfil_arg, cmd, arg);
       }
       psref_release(&psref, &phlist->psref, pfil_psref_class);
       curlwp_bindx(bound);
}

void
pfil_run_addrhooks(pfil_head_t *ph, u_long cmd, struct ifaddr *ifa)
{
       pfil_run_arg(&ph->ph_ifaddr, cmd, ifa);
}

void
pfil_run_ifhooks(pfil_head_t *ph, u_long cmd, struct ifnet *ifp)
{
       pfil_run_arg(&ph->ph_ifevent, cmd, ifp);
}