freebsd-nq/sys/netinet/ipfw/dn_heap.c

551 lines
14 KiB
C
Raw Normal View History

Bring in the most recent version of ipfw and dummynet, developed and tested over the past two months in the ipfw3-head branch. This also happens to be the same code available in the Linux and Windows ports of ipfw and dummynet. The major enhancement is a completely restructured version of dummynet, with support for different packet scheduling algorithms (loadable at runtime), faster queue/pipe lookup, and a much cleaner internal architecture and kernel/userland ABI which simplifies future extensions. In addition to the existing schedulers (FIFO and WF2Q+), we include a Deficit Round Robin (DRR or RR for brevity) scheduler, and a new, very fast version of WF2Q+ called QFQ. Some test code is also present (in sys/netinet/ipfw/test) that lets you build and test schedulers in userland. Also, we have added a compatibility layer that understands requests from the RELENG_7 and RELENG_8 versions of the /sbin/ipfw binaries, and replies correctly (at least, it does its best; sometimes you just cannot tell who sent the request and how to answer). The compatibility layer should make it possible to MFC this code in a relatively short time. Some minor glitches (e.g. handling of ipfw set enable/disable, and a workaround for a bug in RELENG_7's /sbin/ipfw) will be fixed with separate commits. CREDITS: This work has been partly supported by the ONELAB2 project, and mostly developed by Riccardo Panicucci and myself. The code for the qfq scheduler is mostly from Fabio Checconi, and Marta Carbone and Francesco Magno have helped with testing, debugging and some bug fixes.
2010-03-02 17:40:48 +00:00
/*-
* Copyright (c) 1998-2002,2010 Luigi Rizzo, Universita` di Pisa
* 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.
*/
/*
* Binary heap and hash tables, used in dummynet
*
* $FreeBSD$
*/
#include <sys/cdefs.h>
#include <sys/param.h>
#ifdef _KERNEL
__FBSDID("$FreeBSD$");
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <netinet/ipfw/dn_heap.h>
#ifndef log
#define log(x, arg...)
#endif
#else /* !_KERNEL */
#include <stdio.h>
#include <dn_test.h>
#include <strings.h>
#include <stdlib.h>
#include "dn_heap.h"
#define log(x, arg...) fprintf(stderr, ## arg)
#define panic(x...) fprintf(stderr, ## x), exit(1)
#define MALLOC_DEFINE(a, b, c)
static void *my_malloc(int s) { return malloc(s); }
static void my_free(void *p) { free(p); }
#define malloc(s, t, w) my_malloc(s)
#define free(p, t) my_free(p)
#endif /* !_KERNEL */
MALLOC_DEFINE(M_DN_HEAP, "dummynet", "dummynet heap");
/*
* Heap management functions.
*
* In the heap, first node is element 0. Children of i are 2i+1 and 2i+2.
* Some macros help finding parent/children so we can optimize them.
*
* heap_init() is called to expand the heap when needed.
* Increment size in blocks of 16 entries.
* Returns 1 on error, 0 on success
*/
#define HEAP_FATHER(x) ( ( (x) - 1 ) / 2 )
#define HEAP_LEFT(x) ( (x)+(x) + 1 )
#define HEAP_SWAP(a, b, buffer) { buffer = a ; a = b ; b = buffer ; }
#define HEAP_INCREMENT 15
static int
heap_resize(struct dn_heap *h, unsigned int new_size)
{
struct dn_heap_entry *p;
if (h->size >= new_size ) /* have enough room */
return 0;
#if 1 /* round to the next power of 2 */
new_size |= new_size >> 1;
new_size |= new_size >> 2;
new_size |= new_size >> 4;
new_size |= new_size >> 8;
new_size |= new_size >> 16;
#else
new_size = (new_size + HEAP_INCREMENT ) & ~HEAP_INCREMENT;
#endif
p = malloc(new_size * sizeof(*p), M_DN_HEAP, M_NOWAIT);
if (p == NULL) {
printf("--- %s, resize %d failed\n", __func__, new_size );
return 1; /* error */
}
if (h->size > 0) {
bcopy(h->p, p, h->size * sizeof(*p) );
free(h->p, M_DN_HEAP);
}
h->p = p;
h->size = new_size;
return 0;
}
int
heap_init(struct dn_heap *h, int size, int ofs)
{
if (heap_resize(h, size))
return 1;
h->elements = 0;
h->ofs = ofs;
return 0;
}
/*
* Insert element in heap. Normally, p != NULL, we insert p in
* a new position and bubble up. If p == NULL, then the element is
* already in place, and key is the position where to start the
* bubble-up.
* Returns 1 on failure (cannot allocate new heap entry)
*
* If ofs > 0 the position (index, int) of the element in the heap is
* also stored in the element itself at the given offset in bytes.
*/
#define SET_OFFSET(h, i) do { \
if (h->ofs > 0) \
*((int32_t *)((char *)(h->p[i].object) + h->ofs)) = i; \
} while (0)
/*
* RESET_OFFSET is used for sanity checks. It sets ofs
* to an invalid value.
*/
#define RESET_OFFSET(h, i) do { \
if (h->ofs > 0) \
*((int32_t *)((char *)(h->p[i].object) + h->ofs)) = -16; \
} while (0)
int
heap_insert(struct dn_heap *h, uint64_t key1, void *p)
{
int son = h->elements;
//log("%s key %llu p %p\n", __FUNCTION__, key1, p);
if (p == NULL) { /* data already there, set starting point */
son = key1;
} else { /* insert new element at the end, possibly resize */
son = h->elements;
if (son == h->size) /* need resize... */
// XXX expand by 16 or so
if (heap_resize(h, h->elements+16) )
return 1; /* failure... */
h->p[son].object = p;
h->p[son].key = key1;
h->elements++;
}
/* make sure that son >= father along the path */
while (son > 0) {
int father = HEAP_FATHER(son);
struct dn_heap_entry tmp;
if (DN_KEY_LT( h->p[father].key, h->p[son].key ) )
break; /* found right position */
/* son smaller than father, swap and repeat */
HEAP_SWAP(h->p[son], h->p[father], tmp);
SET_OFFSET(h, son);
son = father;
}
SET_OFFSET(h, son);
return 0;
}
/*
* remove top element from heap, or obj if obj != NULL
*/
void
heap_extract(struct dn_heap *h, void *obj)
{
int child, father, max = h->elements - 1;
if (max < 0) {
printf("--- %s: empty heap 0x%p\n", __FUNCTION__, h);
return;
}
if (obj == NULL)
father = 0; /* default: move up smallest child */
else { /* extract specific element, index is at offset */
if (h->ofs <= 0)
panic("%s: extract from middle not set on %p\n",
__FUNCTION__, h);
father = *((int *)((char *)obj + h->ofs));
if (father < 0 || father >= h->elements) {
panic("%s: father %d out of bound 0..%d\n",
__FUNCTION__, father, h->elements);
}
}
/*
* below, father is the index of the empty element, which
* we replace at each step with the smallest child until we
* reach the bottom level.
*/
// XXX why removing RESET_OFFSET increases runtime by 10% ?
RESET_OFFSET(h, father);
while ( (child = HEAP_LEFT(father)) <= max ) {
if (child != max &&
DN_KEY_LT(h->p[child+1].key, h->p[child].key) )
child++; /* take right child, otherwise left */
h->p[father] = h->p[child];
SET_OFFSET(h, father);
father = child;
}
h->elements--;
if (father != max) {
/*
* Fill hole with last entry and bubble up,
* reusing the insert code
*/
h->p[father] = h->p[max];
heap_insert(h, father, NULL);
}
}
#if 0
/*
* change object position and update references
* XXX this one is never used!
*/
static void
heap_move(struct dn_heap *h, uint64_t new_key, void *object)
{
int temp, i, max = h->elements-1;
struct dn_heap_entry *p, buf;
if (h->ofs <= 0)
panic("cannot move items on this heap");
p = h->p; /* shortcut */
i = *((int *)((char *)object + h->ofs));
if (DN_KEY_LT(new_key, p[i].key) ) { /* must move up */
p[i].key = new_key;
for (; i>0 &&
DN_KEY_LT(new_key, p[(temp = HEAP_FATHER(i))].key);
i = temp ) { /* bubble up */
HEAP_SWAP(p[i], p[temp], buf);
SET_OFFSET(h, i);
}
} else { /* must move down */
p[i].key = new_key;
while ( (temp = HEAP_LEFT(i)) <= max ) {
/* found left child */
if (temp != max &&
DN_KEY_LT(p[temp+1].key, p[temp].key))
temp++; /* select child with min key */
if (DN_KEY_LT(>p[temp].key, new_key)) {
/* go down */
HEAP_SWAP(p[i], p[temp], buf);
SET_OFFSET(h, i);
} else
break;
i = temp;
}
}
SET_OFFSET(h, i);
}
#endif /* heap_move, unused */
/*
* heapify() will reorganize data inside an array to maintain the
* heap property. It is needed when we delete a bunch of entries.
*/
static void
heapify(struct dn_heap *h)
{
int i;
for (i = 0; i < h->elements; i++ )
heap_insert(h, i , NULL);
}
int
heap_scan(struct dn_heap *h, int (*fn)(void *, uintptr_t),
uintptr_t arg)
{
int i, ret, found;
for (i = found = 0 ; i < h->elements ;) {
ret = fn(h->p[i].object, arg);
if (ret & HEAP_SCAN_DEL) {
h->elements-- ;
h->p[i] = h->p[h->elements] ;
found++ ;
} else
i++ ;
if (ret & HEAP_SCAN_END)
break;
}
if (found)
heapify(h);
return found;
}
/*
* cleanup the heap and free data structure
*/
void
heap_free(struct dn_heap *h)
{
if (h->size >0 )
free(h->p, M_DN_HEAP);
bzero(h, sizeof(*h) );
}
/*
* hash table support.
*/
struct dn_ht {
int buckets; /* how many buckets, really buckets - 1*/
int entries; /* how many entries */
int ofs; /* offset of link field */
uint32_t (*hash)(uintptr_t, int, void *arg);
int (*match)(void *_el, uintptr_t key, int, void *);
void *(*new)(uintptr_t, int, void *);
void **ht; /* bucket heads */
};
/*
* Initialize, allocating bucket pointers inline.
* Recycle previous record if possible.
* If the 'new' function is not supplied, we assume that the
* key passed to ht_find is the same object to be stored in.
*/
struct dn_ht *
dn_ht_init(struct dn_ht *ht, int buckets, int ofs,
uint32_t (*h)(uintptr_t, int, void *),
int (*match)(void *, uintptr_t, int, void *),
void *(*new)(uintptr_t, int, void *))
{
int l;
/*
* Notes about rounding bucket size to a power of two.
* Given the original bucket size, we compute the nearest lower and
* higher power of two, minus 1 (respectively b_min and b_max) because
* this value will be used to do an AND with the index returned
* by hash function.
* To choice between these two values, the original bucket size is
* compared with b_min. If the original size is greater than 4/3 b_min,
* we round the bucket size to b_max, else to b_min.
* This ratio try to round to the nearest power of two, advantaging
* the greater size if the different between two power is relatively
* big.
* Rounding the bucket size to a power of two avoid the use of
* module when calculating the correct bucket.
* The ht->buckets variable store the bucket size - 1 to simply
* do an AND between the index returned by hash function and ht->bucket
* instead of a module.
*/
int b_min; /* min buckets */
int b_max; /* max buckets */
int b_ori; /* original buckets */
if (h == NULL || match == NULL) {
printf("--- missing hash or match function");
return NULL;
}
if (buckets < 1 || buckets > 65536)
return NULL;
b_ori = buckets;
/* calculate next power of 2, - 1*/
buckets |= buckets >> 1;
buckets |= buckets >> 2;
buckets |= buckets >> 4;
buckets |= buckets >> 8;
buckets |= buckets >> 16;
b_max = buckets; /* Next power */
b_min = buckets >> 1; /* Previous power */
/* Calculate the 'nearest' bucket size */
if (b_min * 4000 / 3000 < b_ori)
buckets = b_max;
else
buckets = b_min;
if (ht) { /* see if we can reuse */
if (buckets <= ht->buckets) {
ht->buckets = buckets;
} else {
/* free pointers if not allocated inline */
if (ht->ht != (void *)(ht + 1))
free(ht->ht, M_DN_HEAP);
free(ht, M_DN_HEAP);
ht = NULL;
}
}
if (ht == NULL) {
/* Allocate buckets + 1 entries because buckets is use to
* do the AND with the index returned by hash function
*/
l = sizeof(*ht) + (buckets + 1) * sizeof(void **);
ht = malloc(l, M_DN_HEAP, M_NOWAIT | M_ZERO);
}
if (ht) {
ht->ht = (void **)(ht + 1);
ht->buckets = buckets;
ht->ofs = ofs;
ht->hash = h;
ht->match = match;
ht->new = new;
}
return ht;
}
/* dummy callback for dn_ht_free to unlink all */
static int
do_del(void *obj, void *arg)
{
return DNHT_SCAN_DEL;
}
void
dn_ht_free(struct dn_ht *ht, int flags)
{
if (ht == NULL)
return;
if (flags & DNHT_REMOVE) {
(void)dn_ht_scan(ht, do_del, NULL);
} else {
if (ht->ht && ht->ht != (void *)(ht + 1))
free(ht->ht, M_DN_HEAP);
free(ht, M_DN_HEAP);
}
}
int
dn_ht_entries(struct dn_ht *ht)
{
return ht ? ht->entries : 0;
}
/* lookup and optionally create or delete element */
void *
dn_ht_find(struct dn_ht *ht, uintptr_t key, int flags, void *arg)
{
int i;
void **pp, *p;
if (ht == NULL) /* easy on an empty hash */
return NULL;
i = (ht->buckets == 1) ? 0 :
(ht->hash(key, flags, arg) & ht->buckets);
for (pp = &ht->ht[i]; (p = *pp); pp = (void **)((char *)p + ht->ofs)) {
if (flags & DNHT_MATCH_PTR) {
if (key == (uintptr_t)p)
break;
} else if (ht->match(p, key, flags, arg)) /* found match */
break;
}
if (p) {
if (flags & DNHT_REMOVE) {
/* link in the next element */
*pp = *(void **)((char *)p + ht->ofs);
*(void **)((char *)p + ht->ofs) = NULL;
ht->entries--;
}
} else if (flags & DNHT_INSERT) {
// printf("%s before calling new, bucket %d ofs %d\n",
// __FUNCTION__, i, ht->ofs);
p = ht->new ? ht->new(key, flags, arg) : (void *)key;
// printf("%s new returns %p\n", __FUNCTION__, p);
if (p) {
ht->entries++;
*(void **)((char *)p + ht->ofs) = ht->ht[i];
ht->ht[i] = p;
}
}
return p;
}
/*
* do a scan with the option to delete the object. Extract next before
* running the callback because the element may be destroyed there.
*/
int
dn_ht_scan(struct dn_ht *ht, int (*fn)(void *, void *), void *arg)
{
int i, ret, found = 0;
void **curp, *cur, *next;
if (ht == NULL || fn == NULL)
return 0;
for (i = 0; i <= ht->buckets; i++) {
curp = &ht->ht[i];
while ( (cur = *curp) != NULL) {
next = *(void **)((char *)cur + ht->ofs);
ret = fn(cur, arg);
if (ret & DNHT_SCAN_DEL) {
found++;
ht->entries--;
*curp = next;
} else {
curp = (void **)((char *)cur + ht->ofs);
}
if (ret & DNHT_SCAN_END)
return found;
}
}
return found;
}
/*
* Similar to dn_ht_scan(), except thah the scan is performed only
* in the bucket 'bucket'. The function returns a correct bucket number if
* the original is invalid
*/
int
dn_ht_scan_bucket(struct dn_ht *ht, int *bucket, int (*fn)(void *, void *),
void *arg)
{
int i, ret, found = 0;
void **curp, *cur, *next;
if (ht == NULL || fn == NULL)
return 0;
if (*bucket > ht->buckets)
*bucket = 0;
i = *bucket;
curp = &ht->ht[i];
while ( (cur = *curp) != NULL) {
next = *(void **)((char *)cur + ht->ofs);
ret = fn(cur, arg);
if (ret & DNHT_SCAN_DEL) {
found++;
ht->entries--;
*curp = next;
} else {
curp = (void **)((char *)cur + ht->ofs);
}
if (ret & DNHT_SCAN_END)
return found;
}
return found;
}