source: mainline/kernel/generic/src/mm/slab.c@ b60615bd

/*
 * Copyright (c) 2006 Ondrej Palkovsky
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 * - Redistributions of source code must retain the above copyright
 *   notice, this list of conditions and the following disclaimer.
 * - 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.
 * - The name of the author may not be used to endorse or promote products
 *   derived from this software without specific prior written permission.
 *
 * 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.
 */

/** @addtogroup kernel_generic_mm
 * @{
 */

/**
 * @file
 * @brief Slab allocator.
 *
 * The slab allocator is closely modelled after OpenSolaris slab allocator.
 * @see http://www.usenix.org/events/usenix01/full_papers/bonwick/bonwick_html/
 *
 * with the following exceptions:
 * @li empty slabs are deallocated immediately
 *     (in Linux they are kept in linked list, in Solaris ???)
 * @li empty magazines are deallocated when not needed
 *     (in Solaris they are held in linked list in slab cache)
 *
 * Following features are not currently supported but would be easy to do:
 * @li cache coloring
 * @li dynamic magazine growing (different magazine sizes are already
 *     supported, but we would need to adjust allocation strategy)
 *
 * The slab allocator supports per-CPU caches ('magazines') to facilitate
 * good SMP scaling.
 *
 * When a new object is being allocated, it is first checked, if it is
 * available in a CPU-bound magazine. If it is not found there, it is
 * allocated from a CPU-shared slab - if a partially full one is found,
 * it is used, otherwise a new one is allocated.
 *
 * When an object is being deallocated, it is put to a CPU-bound magazine.
 * If there is no such magazine, a new one is allocated (if this fails,
 * the object is deallocated into slab). If the magazine is full, it is
 * put into cpu-shared list of magazines and a new one is allocated.
 *
 * The CPU-bound magazine is actually a pair of magazines in order to avoid
 * thrashing when somebody is allocating/deallocating 1 item at the magazine
 * size boundary. LIFO order is enforced, which should avoid fragmentation
 * as much as possible.
 *
 * Every cache contains list of full slabs and list of partially full slabs.
 * Empty slabs are immediately freed (thrashing will be avoided because
 * of magazines).
 *
 * The slab information structure is kept inside the data area, if possible.
 * The cache can be marked that it should not use magazines. This is used
 * only for slab related caches to avoid deadlocks and infinite recursion
 * (the slab allocator uses itself for allocating all it's control structures).
 *
 * The slab allocator allocates a lot of space and does not free it. When
 * the frame allocator fails to allocate a frame, it calls slab_reclaim().
 * It tries 'light reclaim' first, then brutal reclaim. The light reclaim
 * releases slabs from cpu-shared magazine-list, until at least 1 slab
 * is deallocated in each cache (this algorithm should probably change).
 * The brutal reclaim removes all cached objects, even from CPU-bound
 * magazines.
 *
 * @todo
 * For better CPU-scaling the magazine allocation strategy should
 * be extended. Currently, if the cache does not have magazine, it asks
 * for non-cpu cached magazine cache to provide one. It might be feasible
 * to add cpu-cached magazine cache (which would allocate it's magazines
 * from non-cpu-cached mag. cache). This would provide a nice per-cpu
 * buffer. The other possibility is to use the per-cache
 * 'empty-magazine-list', which decreases competing for 1 per-system
 * magazine cache.
 *
 * @todo
 * It might be good to add granularity of locks even to slab level,
 * we could then try_spinlock over all partial slabs and thus improve
 * scalability even on slab level.
 *
 */

#include <assert.h>
#include <errno.h>
#include <synch/spinlock.h>
#include <mm/slab.h>
#include <adt/list.h>
#include <mem.h>
#include <align.h>
#include <mm/frame.h>
#include <config.h>
#include <stdio.h>
#include <arch.h>
#include <panic.h>
#include <bitops.h>
#include <macros.h>
#include <cpu.h>
#include <stdlib.h>

IRQ_SPINLOCK_STATIC_INITIALIZE(slab_cache_lock);
static LIST_INITIALIZE(slab_cache_list);

/** Magazine cache */
static slab_cache_t mag_cache;

/** Cache for cache descriptors */
static slab_cache_t slab_cache_cache;

/** Cache for per-CPU magazines of caches */
static slab_cache_t slab_mag_cache;

/** Cache for external slab descriptors
 * This time we want per-cpu cache, so do not make it static
 * - using slab for internal slab structures will not deadlock,
 *   as all slab structures are 'small' - control structures of
 *   their caches do not require further allocation
 */
static slab_cache_t *slab_extern_cache;

/** Slab descriptor */
typedef struct {
        slab_cache_t *cache;  /**< Pointer to parent cache. */
        link_t link;          /**< List of full/partial slabs. */
        void *start;          /**< Start address of first available item. */
        size_t available;     /**< Count of available items in this slab. */
        size_t nextavail;     /**< The index of next available item. */
} slab_t;

#ifdef CONFIG_DEBUG
static unsigned int _slab_initialized = 0;
#endif

/**************************************/
/* Slab allocation functions          */
/**************************************/

/** Allocate frames for slab space and initialize
 *
 */
NO_TRACE static slab_t *slab_space_alloc(slab_cache_t *cache,
    unsigned int flags)
{
        size_t zone = 0;

        uintptr_t data_phys =
            frame_alloc_generic(cache->frames, FRAME_LOWMEM | flags, 0, &zone);
        if (!data_phys)
                return NULL;

        void *data = (void *) PA2KA(data_phys);

        slab_t *slab;
        size_t fsize;

        if (!(cache->flags & SLAB_CACHE_SLINSIDE)) {
                slab = slab_alloc(slab_extern_cache, flags);
                if (!slab) {
                        frame_free(KA2PA(data), cache->frames);
                        return NULL;
                }
        } else {
                fsize = FRAMES2SIZE(cache->frames);
                slab = data + fsize - sizeof(*slab);
        }

        /* Fill in slab structures */
        size_t i;
        for (i = 0; i < cache->frames; i++)
                frame_set_parent(ADDR2PFN(KA2PA(data)) + i, slab, zone);

        slab->start = data;
        slab->available = cache->objects;
        slab->nextavail = 0;
        slab->cache = cache;

        for (i = 0; i < cache->objects; i++)
                *((size_t *) (slab->start + i * cache->size)) = i + 1;

        atomic_inc(&cache->allocated_slabs);
        return slab;
}

/** Deallocate space associated with slab
 *
 * @return number of freed frames
 *
 */
NO_TRACE static size_t slab_space_free(slab_cache_t *cache, slab_t *slab)
{
        frame_free(KA2PA(slab->start), slab->cache->frames);
        if (!(cache->flags & SLAB_CACHE_SLINSIDE))
                slab_free(slab_extern_cache, slab);

        atomic_dec(&cache->allocated_slabs);

        return cache->frames;
}

/** Map object to slab structure */
NO_TRACE static slab_t *obj2slab(void *obj)
{
        return (slab_t *) frame_get_parent(ADDR2PFN(KA2PA(obj)), 0);
}

/******************/
/* Slab functions */
/******************/

/** Return object to slab and call a destructor
 *
 * @param slab If the caller knows directly slab of the object, otherwise NULL
 *
 * @return Number of freed pages
 *
 */
NO_TRACE static size_t slab_obj_destroy(slab_cache_t *cache, void *obj,
    slab_t *slab)
{
        if (!slab)
                slab = obj2slab(obj);

        assert(slab->cache == cache);

        size_t freed = 0;

        if (cache->destructor)
                freed = cache->destructor(obj);

        irq_spinlock_lock(&cache->slablock, true);
        assert(slab->available < cache->objects);

        *((size_t *) obj) = slab->nextavail;
        slab->nextavail = (obj - slab->start) / cache->size;
        slab->available++;

        /* Move it to correct list */
        if (slab->available == cache->objects) {
                /* Free associated memory */
                list_remove(&slab->link);
                irq_spinlock_unlock(&cache->slablock, true);

                return freed + slab_space_free(cache, slab);
        } else if (slab->available == 1) {
                /* It was in full, move to partial */
                list_remove(&slab->link);
                list_prepend(&slab->link, &cache->partial_slabs);
        }

        irq_spinlock_unlock(&cache->slablock, true);
        return freed;
}

/** Take new object from slab or create new if needed
 *
 * @return Object address or null
 *
 */
NO_TRACE static void *slab_obj_create(slab_cache_t *cache, unsigned int flags)
{
        irq_spinlock_lock(&cache->slablock, true);

        slab_t *slab;

        if (list_empty(&cache->partial_slabs)) {
                /*
                 * Allow recursion and reclaiming
                 * - this should work, as the slab control structures
                 *   are small and do not need to allocate with anything
                 *   other than frame_alloc when they are allocating,
                 *   that's why we should get recursion at most 1-level deep
                 *
                 */
                irq_spinlock_unlock(&cache->slablock, true);
                slab = slab_space_alloc(cache, flags);
                if (!slab)
                        return NULL;

                irq_spinlock_lock(&cache->slablock, true);
        } else {
                slab = list_get_instance(list_first(&cache->partial_slabs),
                    slab_t, link);
                list_remove(&slab->link);
        }

        void *obj = slab->start + slab->nextavail * cache->size;
        slab->nextavail = *((size_t *) obj);
        slab->available--;

        if (!slab->available)
                list_prepend(&slab->link, &cache->full_slabs);
        else
                list_prepend(&slab->link, &cache->partial_slabs);

        irq_spinlock_unlock(&cache->slablock, true);

        if ((cache->constructor) && (cache->constructor(obj, flags) != EOK)) {
                /* Bad, bad, construction failed */
                slab_obj_destroy(cache, obj, slab);
                return NULL;
        }

        return obj;
}

/****************************/
/* CPU-Cache slab functions */
/****************************/

/** Find a full magazine in cache, take it from list and return it
 *
 * @param first If true, return first, else last mag.
 *
 */
NO_TRACE static slab_magazine_t *get_mag_from_cache(slab_cache_t *cache,
    bool first)
{
        slab_magazine_t *mag = NULL;
        link_t *cur;

        irq_spinlock_lock(&cache->maglock, true);
        if (!list_empty(&cache->magazines)) {
                if (first)
                        cur = list_first(&cache->magazines);
                else
                        cur = list_last(&cache->magazines);

                mag = list_get_instance(cur, slab_magazine_t, link);
                list_remove(&mag->link);
                atomic_dec(&cache->magazine_counter);
        }
        irq_spinlock_unlock(&cache->maglock, true);

        return mag;
}

/** Prepend magazine to magazine list in cache
 *
 */
NO_TRACE static void put_mag_to_cache(slab_cache_t *cache,
    slab_magazine_t *mag)
{
        irq_spinlock_lock(&cache->maglock, true);

        list_prepend(&mag->link, &cache->magazines);
        atomic_inc(&cache->magazine_counter);

        irq_spinlock_unlock(&cache->maglock, true);
}

/** Free all objects in magazine and free memory associated with magazine
 *
 * @return Number of freed pages
 *
 */
NO_TRACE static size_t magazine_destroy(slab_cache_t *cache,
    slab_magazine_t *mag)
{
        size_t i;
        size_t frames = 0;

        for (i = 0; i < mag->busy; i++) {
                frames += slab_obj_destroy(cache, mag->objs[i], NULL);
                atomic_dec(&cache->cached_objs);
        }

        slab_free(&mag_cache, mag);

        return frames;
}

/** Find full magazine, set it as current and return it
 *
 */
NO_TRACE static slab_magazine_t *get_full_current_mag(slab_cache_t *cache)
{
        slab_magazine_t *cmag = cache->mag_cache[CPU->id].current;
        slab_magazine_t *lastmag = cache->mag_cache[CPU->id].last;

        assert(irq_spinlock_locked(&cache->mag_cache[CPU->id].lock));

        if (cmag) { /* First try local CPU magazines */
                if (cmag->busy)
                        return cmag;

                if ((lastmag) && (lastmag->busy)) {
                        cache->mag_cache[CPU->id].current = lastmag;
                        cache->mag_cache[CPU->id].last = cmag;
                        return lastmag;
                }
        }

        /* Local magazines are empty, import one from magazine list */
        slab_magazine_t *newmag = get_mag_from_cache(cache, 1);
        if (!newmag)
                return NULL;

        if (lastmag)
                magazine_destroy(cache, lastmag);

        cache->mag_cache[CPU->id].last = cmag;
        cache->mag_cache[CPU->id].current = newmag;

        return newmag;
}

/** Try to find object in CPU-cache magazines
 *
 * @return Pointer to object or NULL if not available
 *
 */
NO_TRACE static void *magazine_obj_get(slab_cache_t *cache)
{
        if (!CPU)
                return NULL;

        irq_spinlock_lock(&cache->mag_cache[CPU->id].lock, true);

        slab_magazine_t *mag = get_full_current_mag(cache);
        if (!mag) {
                irq_spinlock_unlock(&cache->mag_cache[CPU->id].lock, true);
                return NULL;
        }

        void *obj = mag->objs[--mag->busy];
        irq_spinlock_unlock(&cache->mag_cache[CPU->id].lock, true);

        atomic_dec(&cache->cached_objs);

        return obj;
}

/** Assure that the current magazine is empty, return pointer to it,
 * or NULL if no empty magazine is available and cannot be allocated
 *
 * We have 2 magazines bound to processor.
 * First try the current.
 * If full, try the last.
 * If full, put to magazines list.
 *
 */
NO_TRACE static slab_magazine_t *make_empty_current_mag(slab_cache_t *cache)
{
        slab_magazine_t *cmag = cache->mag_cache[CPU->id].current;
        slab_magazine_t *lastmag = cache->mag_cache[CPU->id].last;

        assert(irq_spinlock_locked(&cache->mag_cache[CPU->id].lock));

        if (cmag) {
                if (cmag->busy < cmag->size)
                        return cmag;

                if ((lastmag) && (lastmag->busy < lastmag->size)) {
                        cache->mag_cache[CPU->id].last = cmag;
                        cache->mag_cache[CPU->id].current = lastmag;
                        return lastmag;
                }
        }

        /* current | last are full | nonexistent, allocate new */

        /*
         * We do not want to sleep just because of caching,
         * especially we do not want reclaiming to start, as
         * this would deadlock.
         *
         */
        slab_magazine_t *newmag = slab_alloc(&mag_cache,
            FRAME_ATOMIC | FRAME_NO_RECLAIM);
        if (!newmag)
                return NULL;

        newmag->size = SLAB_MAG_SIZE;
        newmag->busy = 0;

        /* Flush last to magazine list */
        if (lastmag)
                put_mag_to_cache(cache, lastmag);

        /* Move current as last, save new as current */
        cache->mag_cache[CPU->id].last = cmag;
        cache->mag_cache[CPU->id].current = newmag;

        return newmag;
}

/** Put object into CPU-cache magazine
 *
 * @return 0 on success, -1 on no memory
 *
 */
NO_TRACE static int magazine_obj_put(slab_cache_t *cache, void *obj)
{
        if (!CPU)
                return -1;

        irq_spinlock_lock(&cache->mag_cache[CPU->id].lock, true);

        slab_magazine_t *mag = make_empty_current_mag(cache);
        if (!mag) {
                irq_spinlock_unlock(&cache->mag_cache[CPU->id].lock, true);
                return -1;
        }

        mag->objs[mag->busy++] = obj;

        irq_spinlock_unlock(&cache->mag_cache[CPU->id].lock, true);

        atomic_inc(&cache->cached_objs);

        return 0;
}

/************************/
/* Slab cache functions */
/************************/

/** Return number of objects that fit in certain cache size
 *
 */
NO_TRACE static size_t comp_objects(slab_cache_t *cache)
{
        if (cache->flags & SLAB_CACHE_SLINSIDE)
                return (FRAMES2SIZE(cache->frames) - sizeof(slab_t)) /
                    cache->size;
        else
                return FRAMES2SIZE(cache->frames) / cache->size;
}

/** Return wasted space in slab
 *
 */
NO_TRACE static size_t badness(slab_cache_t *cache)
{
        size_t objects = comp_objects(cache);
        size_t ssize = FRAMES2SIZE(cache->frames);

        if (cache->flags & SLAB_CACHE_SLINSIDE)
                ssize -= sizeof(slab_t);

        return ssize - objects * cache->size;
}

/** Initialize mag_cache structure in slab cache
 *
 */
NO_TRACE static bool make_magcache(slab_cache_t *cache)
{
        assert(_slab_initialized >= 2);

        cache->mag_cache = slab_alloc(&slab_mag_cache, FRAME_ATOMIC);
        if (!cache->mag_cache)
                return false;

        size_t i;
        for (i = 0; i < config.cpu_count; i++) {
                memsetb(&cache->mag_cache[i], sizeof(cache->mag_cache[i]), 0);
                irq_spinlock_initialize(&cache->mag_cache[i].lock,
                    "slab.cache.mag_cache[].lock");
        }

        return true;
}

/** Initialize allocated memory as a slab cache
 *
 */
NO_TRACE static void _slab_cache_create(slab_cache_t *cache, const char *name,
    size_t size, size_t align, errno_t (*constructor)(void *obj,
    unsigned int kmflag), size_t (*destructor)(void *obj), unsigned int flags)
{
        assert(size > 0);

        memsetb(cache, sizeof(*cache), 0);
        cache->name = name;

        if (align < sizeof(sysarg_t))
                align = sizeof(sysarg_t);

        size = ALIGN_UP(size, align);

        cache->size = size;
        cache->constructor = constructor;
        cache->destructor = destructor;
        cache->flags = flags;

        list_initialize(&cache->full_slabs);
        list_initialize(&cache->partial_slabs);
        list_initialize(&cache->magazines);

        irq_spinlock_initialize(&cache->slablock, "slab.cache.slablock");
        irq_spinlock_initialize(&cache->maglock, "slab.cache.maglock");

        if (!(cache->flags & SLAB_CACHE_NOMAGAZINE))
                (void) make_magcache(cache);

        /* Compute slab sizes, object counts in slabs etc. */
        if (cache->size < SLAB_INSIDE_SIZE)
                cache->flags |= SLAB_CACHE_SLINSIDE;

        /* Minimum slab frames */
        cache->frames = SIZE2FRAMES(cache->size);

        while (badness(cache) > SLAB_MAX_BADNESS(cache))
                cache->frames <<= 1;

        cache->objects = comp_objects(cache);

        /* If info fits in, put it inside */
        if (badness(cache) > sizeof(slab_t))
                cache->flags |= SLAB_CACHE_SLINSIDE;

        /* Add cache to cache list */
        irq_spinlock_lock(&slab_cache_lock, true);
        list_append(&cache->link, &slab_cache_list);
        irq_spinlock_unlock(&slab_cache_lock, true);
}

/** Create slab cache
 *
 */
slab_cache_t *slab_cache_create(const char *name, size_t size, size_t align,
    errno_t (*constructor)(void *obj, unsigned int kmflag),
    size_t (*destructor)(void *obj), unsigned int flags)
{
        slab_cache_t *cache = slab_alloc(&slab_cache_cache, FRAME_ATOMIC);
        if (!cache)
                panic("Not enough memory to allocate slab cache %s.", name);

        _slab_cache_create(cache, name, size, align, constructor, destructor,
            flags);

        return cache;
}

/** Reclaim space occupied by objects that are already free
 *
 * @param flags If contains SLAB_RECLAIM_ALL, do aggressive freeing
 *
 * @return Number of freed pages
 *
 */
NO_TRACE static size_t _slab_reclaim(slab_cache_t *cache, unsigned int flags)
{
        if (cache->flags & SLAB_CACHE_NOMAGAZINE)
                return 0; /* Nothing to do */

        /*
         * We count up to original magazine count to avoid
         * endless loop
         */
        size_t magcount = atomic_load(&cache->magazine_counter);

        slab_magazine_t *mag;
        size_t frames = 0;

        while ((magcount--) && (mag = get_mag_from_cache(cache, 0))) {
                frames += magazine_destroy(cache, mag);
                if ((!(flags & SLAB_RECLAIM_ALL)) && (frames))
                        break;
        }

        if (flags & SLAB_RECLAIM_ALL) {
                /* Free cpu-bound magazines */
                /* Destroy CPU magazines */
                size_t i;
                for (i = 0; i < config.cpu_count; i++) {
                        irq_spinlock_lock(&cache->mag_cache[i].lock, true);

                        mag = cache->mag_cache[i].current;
                        if (mag)
                                frames += magazine_destroy(cache, mag);
                        cache->mag_cache[i].current = NULL;

                        mag = cache->mag_cache[i].last;
                        if (mag)
                                frames += magazine_destroy(cache, mag);
                        cache->mag_cache[i].last = NULL;

                        irq_spinlock_unlock(&cache->mag_cache[i].lock, true);
                }
        }

        return frames;
}

/** Return object to cache, use slab if known
 *
 */
NO_TRACE static void _slab_free(slab_cache_t *cache, void *obj, slab_t *slab)
{
        if (!obj)
                return;

        ipl_t ipl = interrupts_disable();

        if ((cache->flags & SLAB_CACHE_NOMAGAZINE) ||
            (magazine_obj_put(cache, obj)))
                slab_obj_destroy(cache, obj, slab);

        interrupts_restore(ipl);
        atomic_dec(&cache->allocated_objs);
}

/** Check that there are no slabs and remove cache from system
 *
 */
void slab_cache_destroy(slab_cache_t *cache)
{
        /*
         * First remove cache from link, so that we don't need
         * to disable interrupts later
         *
         */
        irq_spinlock_lock(&slab_cache_lock, true);
        list_remove(&cache->link);
        irq_spinlock_unlock(&slab_cache_lock, true);

        /*
         * Do not lock anything, we assume the software is correct and
         * does not touch the cache when it decides to destroy it
         *
         */

        /* Destroy all magazines */
        _slab_reclaim(cache, SLAB_RECLAIM_ALL);

        /* All slabs must be empty */
        if ((!list_empty(&cache->full_slabs)) ||
            (!list_empty(&cache->partial_slabs)))
                panic("Destroying cache that is not empty.");

        if (!(cache->flags & SLAB_CACHE_NOMAGAZINE) && cache->mag_cache) {
                slab_free(&slab_mag_cache, cache->mag_cache);
        }

        slab_free(&slab_cache_cache, cache);
}

/** Allocate new object from cache - if no flags given, always returns memory
 *
 */
void *slab_alloc(slab_cache_t *cache, unsigned int flags)
{
        /* Disable interrupts to avoid deadlocks with interrupt handlers */
        ipl_t ipl = interrupts_disable();

        void *result = NULL;

        if (!(cache->flags & SLAB_CACHE_NOMAGAZINE))
                result = magazine_obj_get(cache);

        if (!result)
                result = slab_obj_create(cache, flags);

        interrupts_restore(ipl);

        if (result)
                atomic_inc(&cache->allocated_objs);

        return result;
}

/** Return slab object to cache
 *
 */
void slab_free(slab_cache_t *cache, void *obj)
{
        _slab_free(cache, obj, NULL);
}

/** Go through all caches and reclaim what is possible */
size_t slab_reclaim(unsigned int flags)
{
        irq_spinlock_lock(&slab_cache_lock, true);

        size_t frames = 0;
        list_foreach(slab_cache_list, link, slab_cache_t, cache) {
                frames += _slab_reclaim(cache, flags);
        }

        irq_spinlock_unlock(&slab_cache_lock, true);

        return frames;
}

/* Print list of caches */
void slab_print_list(void)
{
        printf("[cache name      ] [size  ] [pages ] [obj/pg] [slabs ]"
            " [cached] [alloc ] [ctl]\n");

        size_t skip = 0;
        while (true) {
                /*
                 * We must not hold the slab_cache_lock spinlock when printing
                 * the statistics. Otherwise we can easily deadlock if the print
                 * needs to allocate memory.
                 *
                 * Therefore, we walk through the slab cache list, skipping some
                 * amount of already processed caches during each iteration and
                 * gathering statistics about the first unprocessed cache. For
                 * the sake of printing the statistics, we realese the
                 * slab_cache_lock and reacquire it afterwards. Then the walk
                 * starts again.
                 *
                 * This limits both the efficiency and also accuracy of the
                 * obtained statistics. The efficiency is decreased because the
                 * time complexity of the algorithm is quadratic instead of
                 * linear. The accuracy is impacted because we drop the lock
                 * after processing one cache. If there is someone else
                 * manipulating the cache list, we might omit an arbitrary
                 * number of caches or process one cache multiple times.
                 * However, we don't bleed for this algorithm for it is only
                 * statistics.
                 */

                irq_spinlock_lock(&slab_cache_lock, true);

                link_t *cur = slab_cache_list.head.next;
                size_t i = 0;
                while (i < skip && cur != &slab_cache_list.head) {
                        i++;
                        cur = cur->next;
                }

                if (cur == &slab_cache_list.head) {
                        irq_spinlock_unlock(&slab_cache_lock, true);
                        break;
                }

                skip++;

                slab_cache_t *cache = list_get_instance(cur, slab_cache_t, link);

                const char *name = cache->name;
                size_t frames = cache->frames;
                size_t size = cache->size;
                size_t objects = cache->objects;
                long allocated_slabs = atomic_load(&cache->allocated_slabs);
                long cached_objs = atomic_load(&cache->cached_objs);
                long allocated_objs = atomic_load(&cache->allocated_objs);
                unsigned int flags = cache->flags;

                irq_spinlock_unlock(&slab_cache_lock, true);

                printf("%-18s %8zu %8zu %8zu %8ld %8ld %8ld %-5s\n",
                    name, size, frames, objects, allocated_slabs,
                    cached_objs, allocated_objs,
                    flags & SLAB_CACHE_SLINSIDE ? "in" : "out");
        }
}

void slab_cache_init(void)
{
        /* Initialize magazine cache */
        _slab_cache_create(&mag_cache, "slab_magazine_t",
            sizeof(slab_magazine_t) + SLAB_MAG_SIZE * sizeof(void *),
            sizeof(uintptr_t), NULL, NULL, SLAB_CACHE_NOMAGAZINE |
            SLAB_CACHE_SLINSIDE);

        /* Initialize slab_cache cache */
        _slab_cache_create(&slab_cache_cache, "slab_cache_cache",
            sizeof(slab_cache_cache), sizeof(uintptr_t), NULL, NULL,
            SLAB_CACHE_NOMAGAZINE | SLAB_CACHE_SLINSIDE);

        /* Initialize external slab cache */
        slab_extern_cache = slab_cache_create("slab_t", sizeof(slab_t), 0,
            NULL, NULL, SLAB_CACHE_SLINSIDE | SLAB_CACHE_MAGDEFERRED);

#ifdef CONFIG_DEBUG
        _slab_initialized = 1;
#endif
}

/** Enable cpu_cache
 *
 * Kernel calls this function, when it knows the real number of
 * processors. Allocate slab for cpucache and enable it on all
 * existing slabs that are SLAB_CACHE_MAGDEFERRED
 *
 */
void slab_enable_cpucache(void)
{
#ifdef CONFIG_DEBUG
        _slab_initialized = 2;
#endif

        _slab_cache_create(&slab_mag_cache, "slab_mag_cache",
            sizeof(slab_mag_cache_t) * config.cpu_count, sizeof(uintptr_t),
            NULL, NULL, SLAB_CACHE_NOMAGAZINE | SLAB_CACHE_SLINSIDE);

        irq_spinlock_lock(&slab_cache_lock, false);

        list_foreach(slab_cache_list, link, slab_cache_t, slab) {
                if ((slab->flags & SLAB_CACHE_MAGDEFERRED) !=
                    SLAB_CACHE_MAGDEFERRED)
                        continue;

                (void) make_magcache(slab);
                slab->flags &= ~SLAB_CACHE_MAGDEFERRED;
        }

        irq_spinlock_unlock(&slab_cache_lock, false);
}

/** @}
 */