/*      $NetBSD: vfs_dirhash.c,v 1.16 2024/12/07 02:11:42 riastradh Exp $       */

/*
* Copyright (c) 2008 Reinoud Zandijk
* 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: vfs_dirhash.c,v 1.16 2024/12/07 02:11:42 riastradh Exp $");

/* CLEAN UP! */
#include <sys/param.h>
#include <sys/types.h>

#include <sys/buf.h>
#include <sys/dirent.h>
#include <sys/dirhash.h>
#include <sys/hash.h>
#include <sys/kernel.h>
#include <sys/mutex.h>
#include <sys/pool.h>
#include <sys/queue.h>
#include <sys/sysctl.h>
#include <sys/vnode.h>

#if 1
#       define DPRINTF(a) __nothing
#else
#       define DPRINTF(a) printf a
#endif

/*
* The locking protocol of the dirhash structures is fairly simple:
*
* The global dirhash_queue is protected by the dirhashmutex. This lock is
* internal only and is FS/mountpoint/vnode independent. On exit of the
* exported functions this mutex is not held.
*
* The dirhash structure is considered part of the vnode/inode structure and
* will thus use the lock that protects that vnode/inode.
*
* The dirhash entries are considered part of the dirhash structure and thus
* are on the same lock.
*/

static struct sysctllog *sysctl_log;
static struct pool dirhash_pool;
static struct pool dirhash_entry_pool;

static kmutex_t dirhashmutex;
static uint32_t maxdirhashsize = DIRHASH_SIZE;
static uint32_t dirhashsize    = 0;
static TAILQ_HEAD(_dirhash, dirhash) dirhash_queue;


void
dirhash_init(void)
{
       const struct sysctlnode *rnode, *cnode;
       size_t sz;
       uint32_t max_entries;

       /* initialise dirhash queue */
       TAILQ_INIT(&dirhash_queue);

       /* init dirhash pools */
       sz = sizeof(struct dirhash);
       pool_init(&dirhash_pool, sz, 0, 0, 0,
           "dirhpl", NULL, IPL_NONE);

       sz = sizeof(struct dirhash_entry);
       pool_init(&dirhash_entry_pool, sz, 0, 0, 0,
           "dirhepl", NULL, IPL_NONE);

       mutex_init(&dirhashmutex, MUTEX_DEFAULT, IPL_NONE);
       max_entries = maxdirhashsize / sz;
       pool_sethiwat(&dirhash_entry_pool, max_entries);
       dirhashsize = 0;

       /* create sysctl knobs and dials */
       sysctl_log = NULL;
       sysctl_createv(&sysctl_log, 0, NULL, &rnode,
           CTLFLAG_PERMANENT,
           CTLTYPE_NODE, "dirhash", NULL,
           NULL, 0, NULL, 0,
           CTL_VFS, VFS_GENERIC, CTL_CREATE, CTL_EOL);
       sysctl_createv(&sysctl_log, 0, &rnode, &cnode,
           CTLFLAG_PERMANENT,
           CTLTYPE_INT, "memused",
           SYSCTL_DESCR("current dirhash memory usage"),
           NULL, 0, &dirhashsize, 0,
           CTL_CREATE, CTL_EOL);
       sysctl_createv(&sysctl_log, 0, &rnode, &cnode,
           CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
           CTLTYPE_INT, "maxmem",
           SYSCTL_DESCR("maximum dirhash memory usage"),
           NULL, 0, &maxdirhashsize, 0,
           CTL_CREATE, CTL_EOL);
}


#if 0
void
dirhash_finish(void)
{
       pool_destroy(&dirhash_pool);
       pool_destroy(&dirhash_entry_pool);

       mutex_destroy(&dirhashmutex);

       /* sysctl_teardown(&sysctl_log); */
}
#endif

/*
* generic dirhash implementation
*/

void
dirhash_purge_entries(struct dirhash *dirh)
{
       struct dirhash_entry *dirh_e;
       uint32_t hashline;

       if (dirh == NULL)
               return;

       if (dirh->size == 0)
               return;

       for (hashline = 0; hashline < DIRHASH_HASHSIZE; hashline++) {
               while ((dirh_e = LIST_FIRST(&dirh->entries[hashline]))
                   != NULL) {
                       LIST_REMOVE(dirh_e, next);
                       pool_put(&dirhash_entry_pool, dirh_e);
               }
       }

       while ((dirh_e = LIST_FIRST(&dirh->free_entries)) != NULL) {
               LIST_REMOVE(dirh_e, next);
               pool_put(&dirhash_entry_pool, dirh_e);
       }

       dirh->flags &= ~DIRH_COMPLETE;
       dirh->flags |=  DIRH_PURGED;
       dirh->num_files = 0;

       dirhashsize -= dirh->size;
       dirh->size = 0;
}

void
dirhash_purge(struct dirhash **dirhp)
{
       struct dirhash *dirh = *dirhp;

       if (dirh == NULL)
               return;

       /* purge its entries */
       dirhash_purge_entries(dirh);

       /* recycle */
       mutex_enter(&dirhashmutex);
       TAILQ_REMOVE(&dirhash_queue, dirh, next);
       mutex_exit(&dirhashmutex);

       pool_put(&dirhash_pool, dirh);
       *dirhp = NULL;
}

void
dirhash_get(struct dirhash **dirhp)
{
       struct dirhash *dirh;
       uint32_t hashline;

       /* if no dirhash was given, allocate one */
       dirh = *dirhp;
       if (dirh == NULL) {
               dirh = pool_get(&dirhash_pool, PR_WAITOK | PR_ZERO);
               for (hashline = 0; hashline < DIRHASH_HASHSIZE; hashline++) {
                       LIST_INIT(&dirh->entries[hashline]);
               }
       }

       /* implement LRU on the dirhash queue */
       mutex_enter(&dirhashmutex);
       if (*dirhp) {
               /* remove from queue to be requeued */
               TAILQ_REMOVE(&dirhash_queue, dirh, next);
       }
       dirh->refcnt++;
       TAILQ_INSERT_HEAD(&dirhash_queue, dirh, next);
       mutex_exit(&dirhashmutex);

       *dirhp = dirh;
}

void
dirhash_put(struct dirhash *dirh)
{

       mutex_enter(&dirhashmutex);
       dirh->refcnt--;
       mutex_exit(&dirhashmutex);
}

void
dirhash_enter(struct dirhash *dirh,
   struct dirent *dirent, uint64_t offset, uint32_t entry_size, int new_p)
{
       struct dirhash *del_dirh, *prev_dirh;
       struct dirhash_entry *dirh_e;
       uint32_t hashvalue, hashline;
       int entrysize;

       /* make sure we have a dirhash to work on */
       KASSERT(dirh);
       KASSERT(dirh->refcnt > 0);

       /* are we trying to re-enter an entry? */
       if (!new_p && (dirh->flags & DIRH_COMPLETE))
               return;

       /* calculate our hash */
       hashvalue = hash32_strn(dirent->d_name, dirent->d_namlen,
           HASH32_STR_INIT);
       hashline  = hashvalue & DIRHASH_HASHMASK;

       /* lookup and insert entry if not there yet */
       LIST_FOREACH(dirh_e, &dirh->entries[hashline], next) {
               /* check for hash collision */
               if (dirh_e->hashvalue != hashvalue)
                       continue;
               if (dirh_e->offset != offset)
                       continue;
               /* got it already */
               KASSERT(dirh_e->d_namlen == dirent->d_namlen);
               KASSERT(dirh_e->entry_size == entry_size);
               return;
       }

       DPRINTF(("dirhash enter %"PRIu64", %d, %d for `%*.*s`\n",
               offset, entry_size, dirent->d_namlen,
               dirent->d_namlen, dirent->d_namlen, dirent->d_name));

       /* check if entry is in free space list */
       LIST_FOREACH(dirh_e, &dirh->free_entries, next) {
               if (dirh_e->offset == offset) {
                       DPRINTF(("\tremoving free entry\n"));
                       LIST_REMOVE(dirh_e, next);
                       pool_put(&dirhash_entry_pool, dirh_e);
                       break;
               }
       }

       /* ensure we are not passing the dirhash limit */
       entrysize = sizeof(struct dirhash_entry);
       if (dirhashsize + entrysize > maxdirhashsize) {
               /* we walk the dirhash_queue, so need to lock it */
               mutex_enter(&dirhashmutex);
               del_dirh = TAILQ_LAST(&dirhash_queue, _dirhash);
               KASSERT(del_dirh);
               while (dirhashsize + entrysize > maxdirhashsize) {
                       /* no use trying to delete myself */
                       if (del_dirh == dirh)
                               break;
                       prev_dirh = TAILQ_PREV(del_dirh, _dirhash, next);
                       if (del_dirh->refcnt == 0)
                               dirhash_purge_entries(del_dirh);
                       del_dirh = prev_dirh;
               }
               mutex_exit(&dirhashmutex);
       }

       /* add to the hashline */
       dirh_e = pool_get(&dirhash_entry_pool, PR_WAITOK | PR_ZERO);

       dirh_e->hashvalue = hashvalue;
       dirh_e->offset    = offset;
       dirh_e->d_namlen  = dirent->d_namlen;
       dirh_e->entry_size  = entry_size;

       dirh->size  += sizeof(struct dirhash_entry);
       dirh->num_files++;
       dirhashsize += sizeof(struct dirhash_entry);
       LIST_INSERT_HEAD(&dirh->entries[hashline], dirh_e, next);
}

void
dirhash_enter_freed(struct dirhash *dirh, uint64_t offset, uint32_t entry_size)
{
       struct dirhash_entry *dirh_e;

       /* make sure we have a dirhash to work on */
       KASSERT(dirh);
       KASSERT(dirh->refcnt > 0);

       /* check for double entry of free space */
       LIST_FOREACH(dirh_e, &dirh->free_entries, next) {
               KASSERT(dirh_e->offset != offset);
       }

       DPRINTF(("dirhash enter FREED %"PRIu64", %d\n",
               offset, entry_size));
       dirh_e = pool_get(&dirhash_entry_pool, PR_WAITOK | PR_ZERO);

       dirh_e->hashvalue = 0;          /* not relevant */
       dirh_e->offset    = offset;
       dirh_e->d_namlen  = 0;          /* not relevant */
       dirh_e->entry_size  = entry_size;

       /* XXX it might be preferable to append them at the tail */
       LIST_INSERT_HEAD(&dirh->free_entries, dirh_e, next);
       dirh->size  += sizeof(struct dirhash_entry);
       dirhashsize += sizeof(struct dirhash_entry);
}

void
dirhash_remove(struct dirhash *dirh, struct dirent *dirent,
   uint64_t offset, uint32_t entry_size)
{
       struct dirhash_entry *dirh_e;
       uint32_t hashvalue, hashline;

       DPRINTF(("dirhash remove %"PRIu64", %d for `%*.*s`\n",
               offset, entry_size,
               dirent->d_namlen, dirent->d_namlen, dirent->d_name));

       /* make sure we have a dirhash to work on */
       KASSERT(dirh);
       KASSERT(dirh->refcnt > 0);

       /* calculate our hash */
       hashvalue = hash32_strn(dirent->d_name, dirent->d_namlen,
           HASH32_STR_INIT);
       hashline  = hashvalue & DIRHASH_HASHMASK;

       /* lookup entry */
       LIST_FOREACH(dirh_e, &dirh->entries[hashline], next) {
               /* check for hash collision */
               if (dirh_e->hashvalue != hashvalue)
                       continue;
               if (dirh_e->offset != offset)
                       continue;

               /* got it! */
               KASSERT(dirh_e->d_namlen == dirent->d_namlen);
               KASSERT(dirh_e->entry_size == entry_size);
               LIST_REMOVE(dirh_e, next);
               dirh->size -= sizeof(struct dirhash_entry);
               KASSERT(dirh->num_files > 0);
               dirh->num_files--;
               dirhashsize -= sizeof(struct dirhash_entry);

               dirhash_enter_freed(dirh, offset, entry_size);
               return;
       }

       /* not found! */
       panic("dirhash_remove couldn't find entry in hash table\n");
}

/*
* BUGALERT: don't use result longer than needed, never past the node lock.
* Call with NULL *result initially and it will return nonzero if again.
*/
int
dirhash_lookup(struct dirhash *dirh, const char *d_name, int d_namlen,
   struct dirhash_entry **result)
{
       struct dirhash_entry *dirh_e;
       uint32_t hashvalue, hashline;

       /* make sure we have a dirhash to work on */
       KASSERT(dirh);
       KASSERT(dirh->refcnt > 0);

       /* start where we were */
       if (*result) {
               dirh_e = *result;

               /* retrieve information to avoid recalculation and advance */
               hashvalue = dirh_e->hashvalue;
               dirh_e = LIST_NEXT(*result, next);
       } else {
               /* calculate our hash and lookup all entries in hashline */
               hashvalue = hash32_strn(d_name, d_namlen, HASH32_STR_INIT);
               hashline  = hashvalue & DIRHASH_HASHMASK;
               dirh_e = LIST_FIRST(&dirh->entries[hashline]);
       }

       for (; dirh_e; dirh_e = LIST_NEXT(dirh_e, next)) {
               /* check for hash collision */
               if (dirh_e->hashvalue != hashvalue)
                       continue;
               if (dirh_e->d_namlen != d_namlen)
                       continue;
               /* might have an entry in the cache */
               *result = dirh_e;
               return 1;
       }

       *result = NULL;
       return 0;
}

/*
* BUGALERT: don't use result longer than needed, never past the node lock.
* Call with NULL *result initially and it will return nonzero if again.
*/
int
dirhash_lookup_freed(struct dirhash *dirh, uint32_t min_entrysize,
   struct dirhash_entry **result)
{
       struct dirhash_entry *dirh_e;

       /* make sure we have a dirhash to work on */
       KASSERT(dirh);
       KASSERT(dirh->refcnt > 0);

       /* start where we were */
       if (*result) {
               dirh_e = LIST_NEXT(*result, next);
       } else {
               /* lookup all entries that match */
               dirh_e = LIST_FIRST(&dirh->free_entries);
       }

       for (; dirh_e; dirh_e = LIST_NEXT(dirh_e, next)) {
               /* check for minimum size */
               if (dirh_e->entry_size < min_entrysize)
                       continue;
               /* might be a candidate */
               *result = dirh_e;
               return 1;
       }

       *result = NULL;
       return 0;
}

bool
dirhash_dir_isempty(struct dirhash *dirh)
{
#ifdef DEBUG
       struct dirhash_entry *dirh_e;
       int hashline, num;

       num = 0;
       for (hashline = 0; hashline < DIRHASH_HASHSIZE; hashline++) {
               LIST_FOREACH(dirh_e, &dirh->entries[hashline], next) {
                       num++;
               }
       }

       if (dirh->num_files != num) {
               printf("dirhash_dir_isempy: dirhash_counter failed: "
                   "dirh->num_files = %d, counted %d\n",
                   dirh->num_files, num);
               assert(dirh->num_files == num);
       }
#endif
       /* assert the directory hash info is valid */
       KASSERT(dirh->flags & DIRH_COMPLETE);

       /* the directory is empty when only '..' lifes in it or is absent */
       return (dirh->num_files <= 1);
}