* [PATCH 2.4] backport neighbour cache redesign
@ 2004-09-30 15:47 Harald Welte
2004-10-04 19:17 ` David S. Miller
0 siblings, 1 reply; 6+ messages in thread
From: Harald Welte @ 2004-09-30 15:47 UTC (permalink / raw)
To: David Miller; +Cc: Linux Netdev List
[-- Attachment #1.1: Type: text/plain, Size: 805 bytes --]
Hi Dave!
I've now finished the 2.4.x backport of the neighbour cache redesign.
The patch applies to 2.4.28-pre3-bk5.
It compiles cleanly, and ipv4/ipv6 are working as expected. I don't
have a decnet or atm test setup, so I leave testing to somebody else ;)
Let's hope I didn't introduce new bugs, the merging got a bit ugly since
percpu is missing and (more important) 2.4.x networking didn't yet use
seq_file. So in some cases like ATM, clip is now using seq_file, while
everything else uses the old get_info() proc methods.
Cheers,
Harald
--
- Harald Welte <laforge@gnumonks.org> http://www.gnumonks.org/
============================================================================
Programming is like sex: One mistake and you have to support it your lifetime
[-- Attachment #1.2: 2.4.28-pre3-bk5-neigh26.patch --]
[-- Type: text/plain, Size: 60604 bytes --]
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/drivers/net/bonding/bond_main.c linux-2.4.28-pre3-bk5-neigh/drivers/net/bonding/bond_main.c
--- linux-2.4.28-pre3-bk5-plain/drivers/net/bonding/bond_main.c 2004-04-14 15:05:30.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/drivers/net/bonding/bond_main.c 2004-09-30 14:16:12.000000000 +0200
@@ -3086,8 +3086,6 @@
#ifdef CONFIG_PROC_FS
-#define SEQ_START_TOKEN ((void *)1)
-
static void *bond_info_seq_start(struct seq_file *seq, loff_t *pos)
{
struct bonding *bond = seq->private;
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/fs/proc/root.c linux-2.4.28-pre3-bk5-neigh/fs/proc/root.c
--- linux-2.4.28-pre3-bk5-plain/fs/proc/root.c 2002-08-03 02:39:45.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/fs/proc/root.c 2004-09-30 11:35:52.000000000 +0200
@@ -17,7 +17,7 @@
#include <linux/module.h>
#include <asm/bitops.h>
-struct proc_dir_entry *proc_net, *proc_bus, *proc_root_fs, *proc_root_driver;
+struct proc_dir_entry *proc_net, *proc_net_stat, *proc_bus, *proc_root_fs, *proc_root_driver;
#ifdef CONFIG_SYSCTL
struct proc_dir_entry *proc_sys_root;
@@ -38,6 +38,8 @@
}
proc_misc_init();
proc_net = proc_mkdir("net", 0);
+ proc_net_stat = proc_mkdir("net/stat", NULL);
+
#ifdef CONFIG_SYSVIPC
proc_mkdir("sysvipc", 0);
#endif
@@ -143,5 +145,6 @@
EXPORT_SYMBOL(proc_root);
EXPORT_SYMBOL(proc_root_fs);
EXPORT_SYMBOL(proc_net);
+EXPORT_SYMBOL(proc_net_stat);
EXPORT_SYMBOL(proc_bus);
EXPORT_SYMBOL(proc_root_driver);
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/include/linux/proc_fs.h linux-2.4.28-pre3-bk5-neigh/include/linux/proc_fs.h
--- linux-2.4.28-pre3-bk5-plain/include/linux/proc_fs.h 2003-11-28 19:26:21.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/include/linux/proc_fs.h 2004-09-30 16:34:36.000000000 +0200
@@ -79,6 +79,7 @@
extern struct proc_dir_entry proc_root;
extern struct proc_dir_entry *proc_root_fs;
extern struct proc_dir_entry *proc_net;
+extern struct proc_dir_entry *proc_net_stat;
extern struct proc_dir_entry *proc_bus;
extern struct proc_dir_entry *proc_root_driver;
extern struct proc_dir_entry *proc_root_kcore;
@@ -170,6 +171,16 @@
return create_proc_info_entry(name,mode,proc_net,get_info);
}
+static inline struct proc_dir_entry *proc_net_fops_create(const char *name,
+ mode_t mode, struct file_operations *fops)
+{
+ struct proc_dir_entry *res = create_proc_entry(name, mode, proc_net);
+
+ if (res)
+ res->proc_fops = fops;
+ return res;
+}
+
static inline void proc_net_remove(const char *name)
{
remove_proc_entry(name,proc_net);
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/include/linux/seq_file.h linux-2.4.28-pre3-bk5-neigh/include/linux/seq_file.h
--- linux-2.4.28-pre3-bk5-plain/include/linux/seq_file.h 2004-09-30 14:08:53.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/include/linux/seq_file.h 2004-09-30 14:15:35.000000000 +0200
@@ -65,5 +65,8 @@
int single_open(struct file *, int (*)(struct seq_file *, void *), void *);
int single_release(struct inode *, struct file *);
int seq_release_private(struct inode *, struct file *);
+
+#define SEQ_START_TOKEN ((void *)1)
+
#endif
#endif
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/include/net/dn_neigh.h linux-2.4.28-pre3-bk5-neigh/include/net/dn_neigh.h
--- linux-2.4.28-pre3-bk5-plain/include/net/dn_neigh.h 2000-03-02 19:13:16.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/include/net/dn_neigh.h 2004-09-30 11:36:35.000000000 +0200
@@ -18,7 +18,6 @@
extern void dn_neigh_init(void);
extern void dn_neigh_cleanup(void);
-extern struct neighbour *dn_neigh_lookup(struct neigh_table *tbl, void *ptr);
extern int dn_neigh_router_hello(struct sk_buff *skb);
extern int dn_neigh_endnode_hello(struct sk_buff *skb);
extern void dn_neigh_pointopoint_hello(struct sk_buff *skb);
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/include/net/neighbour.h linux-2.4.28-pre3-bk5-neigh/include/net/neighbour.h
--- linux-2.4.28-pre3-bk5-plain/include/net/neighbour.h 2004-09-30 11:31:17.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/include/net/neighbour.h 2004-09-30 14:16:33.000000000 +0200
@@ -7,6 +7,11 @@
* Authors:
* Pedro Roque <pedro_m@yahoo.com>
* Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
+ *
+ * Changes:
+ *
+ * Harald Welte: <laforge@gnumonks.org>
+ * - Add neighbour cache statistics like rtstat
*/
/* The following flags & states are exported to user space,
@@ -45,6 +50,7 @@
#include <asm/atomic.h>
#include <linux/skbuff.h>
+#include <linux/seq_file.h>
#define NUD_IN_TIMER (NUD_INCOMPLETE|NUD_DELAY|NUD_PROBE)
#define NUD_VALID (NUD_PERMANENT|NUD_NOARP|NUD_REACHABLE|NUD_PROBE|NUD_STALE|NUD_DELAY)
@@ -78,12 +84,25 @@
struct neigh_statistics
{
- unsigned long allocs;
- unsigned long res_failed;
- unsigned long rcv_probes_mcast;
- unsigned long rcv_probes_ucast;
+ unsigned long allocs; /* number of allocated neighs */
+ unsigned long destroys; /* number of destroyed neighs */
+ unsigned long hash_grows; /* number of hash resizes */
+
+ unsigned long res_failed; /* nomber of failed resolutions */
+
+ unsigned long lookups; /* number of lookups */
+ unsigned long hits; /* number of hits (among lookups) */
+
+ unsigned long rcv_probes_mcast; /* number of received mcast ipv6 */
+ unsigned long rcv_probes_ucast; /* number of received ucast ipv6 */
+
+ unsigned long periodic_gc_runs; /* number of periodic GC runs */
+ unsigned long forced_gc_runs; /* number of forced GC runs */
};
+#define NEIGH_CACHE_STAT_INC(tbl, field) \
+ ((tbl)->stats[smp_processor_id()].field++)
+
struct neighbour
{
struct neighbour *next;
@@ -128,9 +147,6 @@
u8 key[0];
};
-#define NEIGH_HASHMASK 0x1F
-#define PNEIGH_HASHMASK 0xF
-
/*
* neighbour table manipulation
*/
@@ -164,9 +180,15 @@
struct neigh_parms *parms_list;
kmem_cache_t *kmem_cachep;
struct tasklet_struct gc_task;
- struct neigh_statistics stats;
- struct neighbour *hash_buckets[NEIGH_HASHMASK+1];
- struct pneigh_entry *phash_buckets[PNEIGH_HASHMASK+1];
+ struct neigh_statistics stats[NR_CPUS];
+ struct neighbour **hash_buckets;
+ unsigned int hash_mask;
+ __u32 hash_rnd;
+ unsigned int hash_chain_gc;
+ struct pneigh_entry **phash_buckets;
+#ifdef CONFIG_PROC_FS
+ struct proc_dir_entry *pde;
+#endif
};
extern void neigh_table_init(struct neigh_table *tbl);
@@ -174,6 +196,8 @@
extern struct neighbour * neigh_lookup(struct neigh_table *tbl,
const void *pkey,
struct net_device *dev);
+extern struct neighbour * neigh_lookup_nodev(struct neigh_table *tbl,
+ const void *pkey);
extern struct neighbour * neigh_create(struct neigh_table *tbl,
const void *pkey,
struct net_device *dev);
@@ -205,6 +229,24 @@
extern int neigh_delete(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg);
extern void neigh_app_ns(struct neighbour *n);
+extern void neigh_for_each(struct neigh_table *tbl, void (*cb)(struct neighbour *, void *), void *cookie);
+extern void __neigh_for_each_release(struct neigh_table *tbl, int (*cb)(struct neighbour *));
+extern void pneigh_for_each(struct neigh_table *tbl, void (*cb)(struct pneigh_entry *));
+
+struct neigh_seq_state {
+ struct neigh_table *tbl;
+ void *(*neigh_sub_iter)(struct neigh_seq_state *state,
+ struct neighbour *n, loff_t *pos);
+ unsigned int bucket;
+ unsigned int flags;
+#define NEIGH_SEQ_NEIGH_ONLY 0x00000001
+#define NEIGH_SEQ_IS_PNEIGH 0x00000002
+#define NEIGH_SEQ_SKIP_NOARP 0x00000004
+};
+extern void *neigh_seq_start(struct seq_file *, loff_t *, struct neigh_table *, unsigned int);
+extern void *neigh_seq_next(struct seq_file *, void *, loff_t *);
+extern void neigh_seq_stop(struct seq_file *, void *);
+
extern int neigh_sysctl_register(struct net_device *dev, struct neigh_parms *p,
int p_id, int pdev_id, char *p_name);
extern void neigh_sysctl_unregister(struct neigh_parms *p);
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/atm/clip.c linux-2.4.28-pre3-bk5-neigh/net/atm/clip.c
--- linux-2.4.28-pre3-bk5-plain/net/atm/clip.c 2004-02-18 14:36:32.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/net/atm/clip.c 2004-09-30 12:50:09.000000000 +0200
@@ -1,6 +1,10 @@
/* net/atm/clip.c - RFC1577 Classical IP over ATM */
-/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA */
+/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA
+ *
+ * Changes:
+ * Harald Welte <laforge@gnumonks.org>:
+ * - backport DaveM's generalized neighbour cache from 2.6.9-rcX */
#include <linux/config.h>
@@ -24,6 +28,7 @@
#include <linux/if.h> /* for IFF_UP */
#include <linux/inetdevice.h>
#include <linux/bitops.h>
+#include <linux/jhash.h>
#include <net/route.h> /* for struct rtable and routing */
#include <net/icmp.h> /* icmp_send */
#include <asm/param.h> /* for HZ */
@@ -119,64 +124,49 @@
spin_unlock_bh(&entry->neigh->dev->xmit_lock);
}
-
-static void idle_timer_check(unsigned long dummy)
+/* The neighbour entry n->lock is held. */
+static int neigh_check_cb(struct neighbour *n)
{
- int i;
+ struct atmarp_entry *entry = NEIGH2ENTRY(n);
+ struct clip_vcc *cv;
- /*DPRINTK("idle_timer_check\n");*/
- write_lock(&clip_tbl.lock);
- for (i = 0; i <= NEIGH_HASHMASK; i++) {
- struct neighbour **np;
+ for (cv = entry->vccs; cv; cv = cv->next) {
+ unsigned long exp = cv->last_use + cv->idle_timeout;
- for (np = &clip_tbl.hash_buckets[i]; *np;) {
- struct neighbour *n = *np;
- struct atmarp_entry *entry = NEIGH2ENTRY(n);
- struct clip_vcc *clip_vcc;
-
- write_lock(&n->lock);
-
- for (clip_vcc = entry->vccs; clip_vcc;
- clip_vcc = clip_vcc->next)
- if (clip_vcc->idle_timeout &&
- time_after(jiffies, clip_vcc->last_use+
- clip_vcc->idle_timeout)) {
- DPRINTK("releasing vcc %p->%p of "
- "entry %p\n",clip_vcc,clip_vcc->vcc,
- entry);
- vcc_release_async(clip_vcc->vcc,
- -ETIMEDOUT);
- }
- if (entry->vccs ||
- time_before(jiffies, entry->expires)) {
- np = &n->next;
- write_unlock(&n->lock);
- continue;
- }
- if (atomic_read(&n->refcnt) > 1) {
- struct sk_buff *skb;
-
- DPRINTK("destruction postponed with ref %d\n",
- atomic_read(&n->refcnt));
- while ((skb = skb_dequeue(&n->arp_queue)) !=
- NULL)
- dev_kfree_skb(skb);
- np = &n->next;
- write_unlock(&n->lock);
- continue;
- }
- *np = n->next;
- DPRINTK("expired neigh %p\n",n);
- n->dead = 1;
- write_unlock(&n->lock);
- neigh_release(n);
+ if (cv->idle_timeout && time_after(jiffies, exp)) {
+ DPRINTK("releasing vcc %p->%p of entry %p\n",
+ cv, cv->vcc, entry);
+ vcc_release_async(cv->vcc, -ETIMEDOUT);
}
}
+
+ if (entry->vccs || time_before(jiffies, entry->expires))
+ return 0;
+
+ if (atomic_read(&n->refcnt) > 1) {
+ struct sk_buff *skb;
+
+ DPRINTK("destruction postponed with ref %d\n",
+ atomic_read(&n->refcnt));
+
+ while ((skb = skb_dequeue(&n->arp_queue)) != NULL)
+ dev_kfree_skb(skb);
+
+ return 0;
+ }
+
+ DPRINTK("expired neigh %p\n",n);
+ return 1;
+}
+
+static void idle_timer_check(unsigned long dummy)
+{
+ write_lock(&clip_tbl.lock);
+ __neigh_for_each_release(&clip_tbl, neigh_check_cb);
mod_timer(&idle_timer, jiffies+CLIP_CHECK_INTERVAL*HZ);
write_unlock(&clip_tbl.lock);
}
-
static int clip_arp_rcv(struct sk_buff *skb)
{
struct atm_vcc *vcc;
@@ -320,15 +310,7 @@
static u32 clip_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(u32*)pkey;
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
-
- return hash_val;
+ return jhash_2words(*(u32 *)pkey, dev->ifindex, clip_tbl.hash_rnd);
}
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/atm/proc.c linux-2.4.28-pre3-bk5-neigh/net/atm/proc.c
--- linux-2.4.28-pre3-bk5-plain/net/atm/proc.c 2003-11-28 19:26:21.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/net/atm/proc.c 2004-09-30 15:01:36.000000000 +0200
@@ -90,74 +90,202 @@
#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
+#define SEQ_NO_VCC_TOKEN ((void *) 2)
-static int svc_addr(char *buf,struct sockaddr_atmsvc *addr)
+static void svc_addr(struct seq_file *seq, struct sockaddr_atmsvc *addr)
{
static int code[] = { 1,2,10,6,1,0 };
static int e164[] = { 1,8,4,6,1,0 };
- int *fields;
- int len,i,j,pos;
- len = 0;
if (*addr->sas_addr.pub) {
- strcpy(buf,addr->sas_addr.pub);
- len = strlen(addr->sas_addr.pub);
- buf += len;
- if (*addr->sas_addr.prv) {
- *buf++ = '+';
- len++;
- }
+ seq_printf(seq, "%s", addr->sas_addr.pub);
+ if (*addr->sas_addr.prv)
+ seq_putc(seq, '+');
+ } else if (!*addr->sas_addr.prv) {
+ seq_printf(seq, "%s", "(none)");
+ return;
}
- else if (!*addr->sas_addr.prv) {
- strcpy(buf,"(none)");
- return strlen(buf);
- }
if (*addr->sas_addr.prv) {
- len += 44;
- pos = 0;
- fields = *addr->sas_addr.prv == ATM_AFI_E164 ? e164 : code;
+ unsigned char *prv = addr->sas_addr.prv;
+ int *fields;
+ int i, j;
+
+ fields = *prv == ATM_AFI_E164 ? e164 : code;
for (i = 0; fields[i]; i++) {
- for (j = fields[i]; j; j--) {
- sprintf(buf,"%02X",addr->sas_addr.prv[pos++]);
- buf += 2;
- }
- if (fields[i+1]) *buf++ = '.';
+ for (j = fields[i]; j; j--)
+ seq_printf(seq, "%02X", *prv++);
+ if (fields[i+1])
+ seq_putc(seq, '.');
}
}
- return len;
}
-static void atmarp_info(struct net_device *dev,struct atmarp_entry *entry,
- struct clip_vcc *clip_vcc,char *buf)
-{
- unsigned char *ip;
- int svc,off,ip_len;
+static void atmarp_info(struct seq_file *seq, struct net_device *dev,struct
+ atmarp_entry *entry, struct clip_vcc *clip_vcc) {
+ unsigned long exp;
+ char buf[17];
+ int svc, llc, off;
+
+ svc = ((clip_vcc == SEQ_NO_VCC_TOKEN) ||
+ (clip_vcc->vcc->sk->family == AF_ATMSVC));
+
+ llc = ((clip_vcc == SEQ_NO_VCC_TOKEN) ||
+ (clip_vcc->encap));
+
+ if (clip_vcc == SEQ_NO_VCC_TOKEN)
+ exp = entry->neigh->used;
+ else
+ exp = clip_vcc->last_use;
+
+ exp = (jiffies - exp) / HZ;
+
+ seq_printf(seq, "%-6s%-4s%-4s%5ld ",
+ dev->name,
+ svc ? "SVC" : "PVC",
+ llc ? "LLC" : "NULL",
+ exp);
- svc = !clip_vcc || clip_vcc->vcc->sk->family == AF_ATMSVC;
- off = sprintf(buf,"%-6s%-4s%-4s%5ld ",dev->name,svc ? "SVC" : "PVC",
- !clip_vcc || clip_vcc->encap ? "LLC" : "NULL",
- (jiffies-(clip_vcc ? clip_vcc->last_use : entry->neigh->used))/
- HZ);
- ip = (unsigned char *) &entry->ip;
- ip_len = sprintf(buf+off,"%d.%d.%d.%d",ip[0],ip[1],ip[2],ip[3]);
- off += ip_len;
- while (ip_len++ < 16) buf[off++] = ' ';
- if (!clip_vcc)
+ off = snprintf(buf, sizeof(buf)-1, "%d.%d.%d.%d", NIPQUAD(entry->ip));
+ while (off < 16)
+ buf[off++] = ' ';
+ buf[off] = '\0';
+ seq_printf(seq, "%s", buf);
+
+ if (clip_vcc == SEQ_NO_VCC_TOKEN) {
if (time_before(jiffies, entry->expires))
- strcpy(buf+off,"(resolving)\n");
- else sprintf(buf+off,"(expired, ref %d)\n",
- atomic_read(&entry->neigh->refcnt));
- else if (!svc)
- sprintf(buf+off,"%d.%d.%d\n",clip_vcc->vcc->dev->number,
- clip_vcc->vcc->vpi,clip_vcc->vcc->vci);
- else {
- off += svc_addr(buf+off,&clip_vcc->vcc->remote);
- strcpy(buf+off,"\n");
+ seq_printf(seq, "(resolving)\n");
+ else
+ seq_printf(seq, "(expired, ref %d)\n",
+ atomic_read(&entry->neigh->refcnt));
+ } else if (!svc) {
+ seq_printf(seq, "%d.%d.%d\n",
+ clip_vcc->vcc->dev->number,
+ clip_vcc->vcc->vpi,
+ clip_vcc->vcc->vci);
+ } else {
+ svc_addr(seq, &clip_vcc->vcc->remote);
+ seq_putc(seq, '\n');
+ }
+}
+
+struct clip_seq_state {
+ /* This member must be first. */
+ struct neigh_seq_state ns;
+
+ /* Local to clip specific iteration. */
+ struct clip_vcc *vcc;
+};
+
+static struct clip_vcc *clip_seq_next_vcc(struct atmarp_entry *e,
+ struct clip_vcc *curr)
+{
+ if (!curr) {
+ curr = e->vccs;
+ if (!curr)
+ return SEQ_NO_VCC_TOKEN;
+ return curr;
+ }
+
+ if (curr == SEQ_NO_VCC_TOKEN)
+ return NULL;
+
+ curr = curr->next;
+
+ return curr;
+}
+
+static void *clip_seq_vcc_walk(struct clip_seq_state *state,
+ struct atmarp_entry *e, loff_t *pos)
+{
+ struct clip_vcc *vcc = state->vcc;
+
+ vcc = clip_seq_next_vcc(e, vcc);
+ if (vcc && pos != NULL) {
+ while (*pos) {
+ vcc = clip_seq_next_vcc(e, vcc);
+ if (!vcc)
+ break;
+ --(*pos);
}
+ }
+ state->vcc = vcc;
+
+ return vcc;
}
+static void *clip_seq_sub_iter(struct neigh_seq_state *_state,
+ struct neighbour *n, loff_t *pos)
+{
+ struct clip_seq_state *state = (struct clip_seq_state *) _state;
+ return clip_seq_vcc_walk(state, NEIGH2ENTRY(n), pos);
+}
+
+static void *clip_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ return neigh_seq_start(seq, pos, clip_tbl_hook, NEIGH_SEQ_NEIGH_ONLY);
+}
+
+static int clip_seq_show(struct seq_file *seq, void *v)
+{
+ static char atm_arp_banner[] =
+ "IPitf TypeEncp Idle IP address ATM address\n";
+
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, atm_arp_banner);
+ } else {
+ struct clip_seq_state *state = seq->private;
+ struct neighbour *n = v;
+ struct clip_vcc *vcc = state->vcc;
+
+ atmarp_info(seq, n->dev, NEIGH2ENTRY(n), vcc);
+ }
+ return 0;
+}
+
+static struct seq_operations arp_seq_ops = {
+ .start = clip_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = clip_seq_show,
+};
+
+static int arp_seq_open(struct inode *inode, struct file *file)
+{
+ struct clip_seq_state *state;
+ struct seq_file *seq;
+ int rc = -EAGAIN;
+
+ state = kmalloc(sizeof(*state), GFP_KERNEL);
+ if (!state) {
+ rc = -ENOMEM;
+ goto out_kfree;
+ }
+ memset(state, 0, sizeof(*state));
+ state->ns.neigh_sub_iter = clip_seq_sub_iter;
+
+ rc = seq_open(file, &arp_seq_ops);
+ if (rc)
+ goto out_kfree;
+
+ seq = file->private_data;
+ seq->private = state;
+out:
+ return rc;
+
+out_kfree:
+ kfree(state);
+ goto out;
+}
+
+static struct file_operations arp_seq_fops = {
+ .open = arp_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+ .owner = THIS_MODULE,
+};
#endif
@@ -416,6 +544,7 @@
return 0;
}
+#if 0
#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
static int atm_arp_info(loff_t pos,char *buf)
{
@@ -459,6 +588,7 @@
return 0;
}
#endif
+#endif
#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
static int atm_lec_info(loff_t pos,char *buf)
@@ -666,7 +796,10 @@
CREATE_ENTRY(svc);
CREATE_ENTRY(vc);
#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
- CREATE_ENTRY(arp);
+ arp = create_proc_entry("arp", S_IRUGO, atm_proc_root);
+ if (!arp)
+ goto cleanup;
+ arp->proc_fops = &arp_seq_fops;
#endif
#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
CREATE_ENTRY(lec);
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/core/neighbour.c linux-2.4.28-pre3-bk5-neigh/net/core/neighbour.c
--- linux-2.4.28-pre3-bk5-plain/net/core/neighbour.c 2004-09-30 11:31:17.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/net/core/neighbour.c 2004-09-30 14:21:06.000000000 +0200
@@ -12,6 +12,8 @@
*
* Fixes:
* Vitaly E. Lavrov releasing NULL neighbor in neigh_add.
+ * Harald Welte Add neighbour cache statistics like rtstat
+ * Harald Welte port neighbour cache rework from 2.6.9-rcX
*/
#include <linux/config.h>
@@ -20,6 +22,7 @@
#include <linux/socket.h>
#include <linux/sched.h>
#include <linux/netdevice.h>
+#include <linux/proc_fs.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#endif
@@ -27,6 +30,7 @@
#include <net/dst.h>
#include <net/sock.h>
#include <linux/rtnetlink.h>
+#include <linux/random.h>
#define NEIGH_DEBUG 1
@@ -45,6 +49,8 @@
#define NEIGH_PRINTK2 NEIGH_PRINTK
#endif
+#define PNEIGH_HASHMASK 0xF
+
static void neigh_timer_handler(unsigned long arg);
#ifdef CONFIG_ARPD
static void neigh_app_notify(struct neighbour *n);
@@ -54,6 +60,7 @@
static int neigh_glbl_allocs;
static struct neigh_table *neigh_tables;
+static struct file_operations neigh_stat_seq_fops;
/*
Neighbour hash table buckets are protected with rwlock tbl->lock.
@@ -111,27 +118,21 @@
int shrunk = 0;
int i;
- for (i=0; i<=NEIGH_HASHMASK; i++) {
+ NEIGH_CACHE_STAT_INC(tbl, forced_gc_runs);
+
+ write_lock_bh(&tbl->lock);
+ for (i = 0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
- write_lock_bh(&tbl->lock);
while ((n = *np) != NULL) {
/* Neighbour record may be discarded if:
- - nobody refers to it.
- - it is not permanent
- - (NEW and probably wrong)
- INCOMPLETE entries are kept at least for
- n->parms->retrans_time, otherwise we could
- flood network with resolution requests.
- It is not clear, what is better table overflow
- or flooding.
+ * - nobody refers to it.
+ * - it is not permanent
*/
write_lock(&n->lock);
if (atomic_read(&n->refcnt) == 1 &&
- !(n->nud_state&NUD_PERMANENT) &&
- (n->nud_state != NUD_INCOMPLETE ||
- jiffies - n->used > n->parms->retrans_time)) {
+ !(n->nud_state&NUD_PERMANENT)) {
*np = n->next;
n->dead = 1;
shrunk = 1;
@@ -142,10 +143,12 @@
write_unlock(&n->lock);
np = &n->next;
}
- write_unlock_bh(&tbl->lock);
}
tbl->last_flush = jiffies;
+
+ write_unlock_bh(&tbl->lock);
+
return shrunk;
}
@@ -176,7 +179,7 @@
write_lock_bh(&tbl->lock);
- for (i=0; i <= NEIGH_HASHMASK; i++) {
+ for (i=0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
@@ -203,7 +206,7 @@
write_lock_bh(&tbl->lock);
- for (i=0; i<=NEIGH_HASHMASK; i++) {
+ for (i = 0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
@@ -277,7 +280,7 @@
init_timer(&n->timer);
n->timer.function = neigh_timer_handler;
n->timer.data = (unsigned long)n;
- tbl->stats.allocs++;
+ NEIGH_CACHE_STAT_INC(tbl, allocs);
neigh_glbl_allocs++;
tbl->entries++;
n->tbl = tbl;
@@ -286,20 +289,103 @@
return n;
}
+static struct neighbour **neigh_hash_alloc(unsigned int entries)
+{
+ unsigned long size = entries * sizeof(struct neighbour *);
+ struct neighbour **ret;
+
+ if (size <= PAGE_SIZE) {
+ ret = kmalloc(size, GFP_ATOMIC);
+ } else {
+ ret = (struct neighbour **)
+ __get_free_pages(GFP_ATOMIC, get_order(size));
+ }
+ if (ret)
+ memset(ret, 0, size);
+
+ return ret;
+}
+
+static void neigh_hash_free(struct neighbour **hash, unsigned int entries)
+{
+ unsigned long size = entries * sizeof(struct neighbour *);
+
+ if (size <= PAGE_SIZE)
+ kfree(hash);
+ else
+ free_pages((unsigned long)hash, get_order(size));
+}
+
+static void neigh_hash_grow(struct neigh_table *tbl, unsigned long new_entries)
+{
+ struct neighbour **new_hash, **old_hash;
+ unsigned int i, new_hash_mask, old_entries;
+
+ NEIGH_CACHE_STAT_INC(tbl, hash_grows);
+
+ BUG_ON(new_entries & (new_entries - 1));
+ new_hash = neigh_hash_alloc(new_entries);
+ if (!new_hash)
+ return;
+
+ old_entries = tbl->hash_mask + 1;
+ new_hash_mask = new_entries - 1;
+ old_hash = tbl->hash_buckets;
+
+ for (i = 0; i < old_entries; i++) {
+ struct neighbour *n, *next;
+
+ for (n = old_hash[i]; n; n = next) {
+ unsigned int hash_val = tbl->hash(n->primary_key, n->dev);
+
+ hash_val &= new_hash_mask;
+ next = n->next;
+
+ n->next = new_hash[hash_val];
+ new_hash[hash_val] = n;
+ }
+ }
+ tbl->hash_buckets = new_hash;
+ tbl->hash_mask = new_hash_mask;
+
+ neigh_hash_free(old_hash, old_entries);
+}
+
struct neighbour *neigh_lookup(struct neigh_table *tbl, const void *pkey,
struct net_device *dev)
{
struct neighbour *n;
- u32 hash_val;
int key_len = tbl->key_len;
+ u32 hash_val = tbl->hash(pkey, dev) & tbl->hash_mask;
- hash_val = tbl->hash(pkey, dev);
+ NEIGH_CACHE_STAT_INC(tbl, lookups);
read_lock_bh(&tbl->lock);
for (n = tbl->hash_buckets[hash_val]; n; n = n->next) {
if (dev == n->dev &&
memcmp(n->primary_key, pkey, key_len) == 0) {
neigh_hold(n);
+ NEIGH_CACHE_STAT_INC(tbl, hits);
+ break;
+ }
+ }
+ read_unlock_bh(&tbl->lock);
+ return n;
+}
+
+struct neighbour *neigh_lookup_nodev(struct neigh_table *tbl, const void *pkey)
+{
+ struct neighbour *n;
+ int key_len = tbl->key_len;
+ u32 hash_val = tbl->hash(pkey, NULL) & tbl->hash_mask;
+
+ NEIGH_CACHE_STAT_INC(tbl, lookups);
+
+ read_lock_bh(&tbl->lock);
+ for (n = tbl->hash_buckets[hash_val]; n; n = n->next) {
+ if (!memcmp(n->primary_key, pkey, key_len)) {
+ neigh_hold(n);
+ NEIGH_CACHE_STAT_INC(tbl, hits);
break;
}
}
@@ -319,6 +405,12 @@
if (n == NULL)
return ERR_PTR(-ENOBUFS);
+ if (tbl->entries > (tbl->hash_mask + 1)) {
+ write_lock_bh(&tbl->lock);
+ neigh_hash_grow(tbl, (tbl->hash_mask + 1) << 1);
+ write_unlock_bh(&tbl->lock);
+ }
+
memcpy(n->primary_key, pkey, key_len);
n->dev = dev;
dev_hold(dev);
@@ -338,7 +430,7 @@
n->confirmed = jiffies - (n->parms->base_reachable_time<<1);
- hash_val = tbl->hash(pkey, dev);
+ hash_val = tbl->hash(pkey, dev) & tbl->hash_mask;
write_lock_bh(&tbl->lock);
for (n1 = tbl->hash_buckets[hash_val]; n1; n1 = n1->next) {
@@ -418,9 +510,9 @@
hash_val ^= hash_val>>4;
hash_val &= PNEIGH_HASHMASK;
+ write_lock_bh(&tbl->lock);
for (np = &tbl->phash_buckets[hash_val]; (n=*np) != NULL; np = &n->next) {
if (memcmp(n->key, pkey, key_len) == 0 && n->dev == dev) {
- write_lock_bh(&tbl->lock);
*np = n->next;
write_unlock_bh(&tbl->lock);
if (tbl->pdestructor)
@@ -429,6 +521,7 @@
return 0;
}
}
+ write_unlock_bh(&tbl->lock);
return -ENOENT;
}
@@ -462,6 +555,8 @@
{
struct hh_cache *hh;
+ NEIGH_CACHE_STAT_INC(neigh->tbl, destroys);
+
if (!neigh->dead) {
printk("Destroying alive neighbour %p\n", neigh);
dump_stack();
@@ -566,9 +661,10 @@
static void SMP_TIMER_NAME(neigh_periodic_timer)(unsigned long arg)
{
struct neigh_table *tbl = (struct neigh_table*)arg;
- unsigned long now = jiffies;
- int i;
+ struct neighbour *n, **np;
+ unsigned long expire, now = jiffies;
+ NEIGH_CACHE_STAT_INC(tbl, periodic_gc_runs);
write_lock(&tbl->lock);
@@ -583,46 +679,49 @@
p->reachable_time = neigh_rand_reach_time(p->base_reachable_time);
}
- for (i=0; i <= NEIGH_HASHMASK; i++) {
- struct neighbour *n, **np;
-
- np = &tbl->hash_buckets[i];
- while ((n = *np) != NULL) {
- unsigned state;
-
- write_lock(&n->lock);
+ np = &tbl->hash_buckets[tbl->hash_chain_gc];
+ tbl->hash_chain_gc = ((tbl->hash_chain_gc + 1) & tbl->hash_mask);
- state = n->nud_state;
- if (state&(NUD_PERMANENT|NUD_IN_TIMER)) {
- write_unlock(&n->lock);
- goto next_elt;
- }
+ while ((n = *np) != NULL) {
+ unsigned int state;
- if ((long)(n->used - n->confirmed) < 0)
- n->used = n->confirmed;
+ write_lock(&n->lock);
+
+ state = n->nud_state;
+ if (state & (NUD_PERMANENT | NUD_IN_TIMER)) {
+ write_unlock(&n->lock);
+ goto next_elt;
+ }
- if (atomic_read(&n->refcnt) == 1 &&
- (state == NUD_FAILED || now - n->used > n->parms->gc_staletime)) {
- *np = n->next;
- n->dead = 1;
- write_unlock(&n->lock);
- neigh_release(n);
- continue;
- }
+ if (time_before(n->used, n->confirmed))
+ n->used = n->confirmed;
- if (n->nud_state&NUD_REACHABLE &&
- now - n->confirmed > n->parms->reachable_time) {
- n->nud_state = NUD_STALE;
- neigh_suspect(n);
- }
+ if (atomic_read(&n->refcnt) == 1 &&
+ (state == NUD_FAILED ||
+ time_after(now, n->used + n->parms->gc_staletime))) {
+ *np = n->next;
+ n->dead = 1;
write_unlock(&n->lock);
+ neigh_release(n);
+ continue;
+ }
+ write_unlock(&n->lock);
next_elt:
- np = &n->next;
- }
+ np = &n->next;
}
+
+ /* Cycle through all hash buckets every base_reachable_time/2 ticks.
+ * ARP entry timeouts range from 1/2 base_reachable_time to 3/2
+ * base_reachable_time.
+ */
+ expire = tbl->parms.base_reachable_time >> 1;
+ expire /= (tbl->hash_mask + 1);
+ if (!expire)
+ expire = 1;
+
+ mod_timer(&tbl->gc_timer, now + expire);
- mod_timer(&tbl->gc_timer, now + tbl->gc_interval);
write_unlock(&tbl->lock);
}
@@ -680,7 +779,7 @@
neigh->nud_state = NUD_FAILED;
notify = 1;
- neigh->tbl->stats.res_failed++;
+ NEIGH_CACHE_STAT_INC(neigh->tbl, res_failed);
NEIGH_PRINTK2("neigh %p is failed.\n", neigh);
/* It is very thin place. report_unreachable is very complicated
@@ -1132,6 +1231,7 @@
void neigh_table_init(struct neigh_table *tbl)
{
unsigned long now = jiffies;
+ unsigned long phsize;
tbl->parms.reachable_time = neigh_rand_reach_time(tbl->parms.base_reachable_time);
@@ -1141,6 +1241,30 @@
0, SLAB_HWCACHE_ALIGN,
NULL, NULL);
+ if (!tbl->kmem_cachep)
+ panic("cannot create neighbour cache");
+
+#ifdef CONFIG_PROC_FS
+ tbl->pde = create_proc_entry(tbl->id, 0, proc_net_stat);
+ if (!tbl->pde)
+ panic("cannot create neighbour proc dir entry");
+ tbl->pde->proc_fops = &neigh_stat_seq_fops;
+ tbl->pde->data = tbl;
+#endif
+
+ tbl->hash_mask = 0x1f;
+ tbl->hash_buckets = neigh_hash_alloc(tbl->hash_mask + 1);
+
+ phsize = (PNEIGH_HASHMASK + 1) * sizeof(struct pneigh_entry *);
+ tbl->phash_buckets = kmalloc(phsize, GFP_KERNEL);
+
+ if (!tbl->hash_buckets || !tbl->phash_buckets)
+ panic("cannot allocate neighbour cache hashes");
+
+ memset(tbl->phash_buckets, 0, phsize);
+
+ get_random_bytes(&tbl->hash_rnd, sizeof(tbl->hash_rnd));
+
#ifdef CONFIG_SMP
tasklet_init(&tbl->gc_task, SMP_TIMER_NAME(neigh_periodic_timer), (unsigned long)tbl);
#endif
@@ -1148,7 +1272,7 @@
tbl->lock = RW_LOCK_UNLOCKED;
tbl->gc_timer.data = (unsigned long)tbl;
tbl->gc_timer.function = neigh_periodic_timer;
- tbl->gc_timer.expires = now + tbl->gc_interval + tbl->parms.reachable_time;
+ tbl->gc_timer.expires = now + 1;
add_timer(&tbl->gc_timer);
init_timer(&tbl->proxy_timer);
@@ -1184,6 +1308,13 @@
}
}
write_unlock(&neigh_tbl_lock);
+
+ neigh_hash_free(tbl->hash_buckets, tbl->hash_mask + 1);
+ tbl->hash_buckets = NULL;
+
+ kfree(tbl->phash_buckets);
+ tbl->phash_buckets = NULL;
+
#ifdef CONFIG_SYSCTL
neigh_sysctl_unregister(&tbl->parms);
#endif
@@ -1364,7 +1495,7 @@
s_h = cb->args[1];
s_idx = idx = cb->args[2];
- for (h=0; h <= NEIGH_HASHMASK; h++) {
+ for (h=0; h <= tbl->hash_mask; h++) {
if (h < s_h) continue;
if (h > s_h)
s_idx = 0;
@@ -1415,6 +1546,359 @@
return skb->len;
}
+void neigh_for_each(struct neigh_table *tbl, void (*cb)(struct neighbour *, void *), void *cookie)
+{
+ int chain;
+
+ read_lock_bh(&tbl->lock);
+ for (chain = 0; chain <= tbl->hash_mask; chain++) {
+ struct neighbour *n;
+
+ for (n = tbl->hash_buckets[chain]; n; n = n->next)
+ cb(n, cookie);
+ }
+ read_unlock_bh(&tbl->lock);
+}
+
+/* The tbl->lock must be held as a writer and BH disabled. */
+void __neigh_for_each_release(struct neigh_table *tbl,
+ int (*cb)(struct neighbour *))
+{
+ int chain;
+
+ for (chain = 0; chain <= tbl->hash_mask; chain++) {
+ struct neighbour *n, **np;
+
+ np = &tbl->hash_buckets[chain];
+ while ((n = *np) != NULL) {
+ int release;
+
+ write_lock(&n->lock);
+ release = cb(n);
+ if (release) {
+ *np = n->next;
+ n->dead = 1;
+ } else
+ np = &n->next;
+ write_unlock(&n->lock);
+ if (release)
+ neigh_release(n);
+ }
+ }
+}
+
+#ifdef CONFIG_PROC_FS
+
+static struct neighbour *neigh_get_first(struct seq_file *seq)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+ struct neighbour *n = NULL;
+ int bucket = state->bucket;
+
+ state->flags &= ~NEIGH_SEQ_IS_PNEIGH;
+ for (bucket = 0; bucket <= tbl->hash_mask; bucket++) {
+ n = tbl->hash_buckets[bucket];
+
+ while (n) {
+ if (state->neigh_sub_iter) {
+ loff_t fakep = 0;
+ void *v;
+
+ v = state->neigh_sub_iter(state, n, &fakep);
+ if (!v)
+ goto next;
+ }
+ if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
+ break;
+ if (n->nud_state & ~NUD_NOARP)
+ break;
+ next:
+ n = n->next;
+ }
+
+ if (n)
+ break;
+ }
+ state->bucket = bucket;
+
+ return n;
+}
+
+static struct neighbour *neigh_get_next(struct seq_file *seq,
+ struct neighbour *n,
+ loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ if (state->neigh_sub_iter) {
+ void *v = state->neigh_sub_iter(state, n, pos);
+ if (v)
+ return n;
+ }
+ n = n->next;
+
+ while (1) {
+ while (n) {
+ if (state->neigh_sub_iter) {
+ void *v = state->neigh_sub_iter(state, n, pos);
+ if (v)
+ return n;
+ goto next;
+ }
+ if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
+ break;
+
+ if (n->nud_state & ~NUD_NOARP)
+ break;
+ next:
+ n = n->next;
+ }
+
+ if (n)
+ break;
+
+ if (++state->bucket > tbl->hash_mask)
+ break;
+
+ n = tbl->hash_buckets[state->bucket];
+ }
+
+ if (n && pos)
+ --(*pos);
+ return n;
+}
+
+static struct neighbour *neigh_get_idx(struct seq_file *seq, loff_t *pos)
+{
+ struct neighbour *n = neigh_get_first(seq);
+
+ if (n) {
+ while (*pos) {
+ n = neigh_get_next(seq, n, pos);
+ if (!n)
+ break;
+ }
+ }
+ return *pos ? NULL : n;
+}
+
+static struct pneigh_entry *pneigh_get_first(struct seq_file *seq)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+ struct pneigh_entry *pn = NULL;
+ int bucket = state->bucket;
+
+ state->flags |= NEIGH_SEQ_IS_PNEIGH;
+ for (bucket = 0; bucket <= PNEIGH_HASHMASK; bucket++) {
+ pn = tbl->phash_buckets[bucket];
+ if (pn)
+ break;
+ }
+ state->bucket = bucket;
+
+ return pn;
+}
+
+static struct pneigh_entry *pneigh_get_next(struct seq_file *seq,
+ struct pneigh_entry *pn,
+ loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ pn = pn->next;
+ while (!pn) {
+ if (++state->bucket > PNEIGH_HASHMASK)
+ break;
+ pn = tbl->phash_buckets[state->bucket];
+ if (pn)
+ break;
+ }
+
+ if (pn && pos)
+ --(*pos);
+
+ return pn;
+}
+
+static struct pneigh_entry *pneigh_get_idx(struct seq_file *seq, loff_t *pos)
+{
+ struct pneigh_entry *pn = pneigh_get_first(seq);
+
+ if (pn) {
+ while (*pos) {
+ pn = pneigh_get_next(seq, pn, pos);
+ if (!pn)
+ break;
+ }
+ }
+ return *pos ? NULL : pn;
+}
+
+static void *neigh_get_idx_any(struct seq_file *seq, loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ void *rc;
+
+ rc = neigh_get_idx(seq, pos);
+ if (!rc && !(state->flags & NEIGH_SEQ_NEIGH_ONLY))
+ rc = pneigh_get_idx(seq, pos);
+
+ return rc;
+}
+
+void *neigh_seq_start(struct seq_file *seq, loff_t *pos, struct neigh_table *tbl, unsigned int neigh_seq_flags)
+{
+ struct neigh_seq_state *state = seq->private;
+ loff_t pos_minus_one;
+
+ state->tbl = tbl;
+ state->bucket = 0;
+ state->flags = (neigh_seq_flags & ~NEIGH_SEQ_IS_PNEIGH);
+
+ read_lock_bh(&tbl->lock);
+
+ pos_minus_one = *pos - 1;
+ return *pos ? neigh_get_idx_any(seq, &pos_minus_one) : SEQ_START_TOKEN;
+}
+
+void *neigh_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+ struct neigh_seq_state *state;
+ void *rc;
+
+ if (v == SEQ_START_TOKEN) {
+ rc = neigh_get_idx(seq, pos);
+ goto out;
+ }
+
+ state = seq->private;
+ if (!(state->flags & NEIGH_SEQ_IS_PNEIGH)) {
+ rc = neigh_get_next(seq, v, NULL);
+ if (rc)
+ goto out;
+ if (!(state->flags & NEIGH_SEQ_NEIGH_ONLY))
+ rc = pneigh_get_first(seq);
+ } else {
+ BUG_ON(state->flags & NEIGH_SEQ_NEIGH_ONLY);
+ rc = pneigh_get_next(seq, v, NULL);
+ }
+out:
+ ++(*pos);
+ return rc;
+}
+
+void neigh_seq_stop(struct seq_file *seq, void *v)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ read_unlock_bh(&tbl->lock);
+}
+
+/* statistics via seq_file */
+
+static void *neigh_stat_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ int lcpu;
+
+ if (*pos == 0)
+ return SEQ_START_TOKEN;
+
+ for (lcpu = *pos-1; lcpu < smp_num_cpus; ++lcpu) {
+ int i = cpu_logical_map(lcpu);
+ *pos = lcpu+1;
+ return &tbl->stats[i];
+ }
+ return NULL;
+}
+
+static void *neigh_stat_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ int lcpu;
+
+ for (lcpu = *pos; lcpu < smp_num_cpus; ++lcpu) {
+ int i = cpu_logical_map(lcpu);
+ *pos = lcpu+1;
+ return &tbl->stats[i];
+ }
+ return NULL;
+}
+
+static void neigh_stat_seq_stop(struct seq_file *seq, void *v)
+{
+
+}
+
+static int neigh_stat_seq_show(struct seq_file *seq, void *v)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ struct neigh_statistics *st = v;
+
+ if (v == SEQ_START_TOKEN) {
+ seq_printf(seq, "entries allocs destroys hash_grows lookups hits res_failed rcv_probes_mcast rcv_probes_ucast periodic_gc_runs forced_gc_runs forced_gc_goal_miss\n");
+ return 0;
+ }
+
+ seq_printf(seq, "%08x %08lx %08lx %08lx %08lx %08lx %08lx "
+ "%08lx %08lx %08lx %08lx\n",
+ tbl->entries,
+
+ st->allocs,
+ st->destroys,
+ st->hash_grows,
+
+ st->lookups,
+ st->hits,
+
+ st->res_failed,
+
+ st->rcv_probes_mcast,
+ st->rcv_probes_ucast,
+
+ st->periodic_gc_runs,
+ st->forced_gc_runs
+ );
+
+ return 0;
+}
+
+static struct seq_operations neigh_stat_seq_ops = {
+ .start = neigh_stat_seq_start,
+ .next = neigh_stat_seq_next,
+ .stop = neigh_stat_seq_stop,
+ .show = neigh_stat_seq_show,
+};
+
+static int neigh_stat_seq_open(struct inode *inode, struct file *file)
+{
+ int ret = seq_open(file, &neigh_stat_seq_ops);
+
+ if (!ret) {
+ struct seq_file *sf = file->private_data;
+ sf->private = PDE(inode);
+ }
+ return ret;
+};
+
+static struct file_operations neigh_stat_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = neigh_stat_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release,
+};
+
+#endif /* CONFIG_PROC_FS */
+
#ifdef CONFIG_ARPD
void neigh_app_ns(struct neighbour *n)
{
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/decnet/dn_neigh.c linux-2.4.28-pre3-bk5-neigh/net/decnet/dn_neigh.c
--- linux-2.4.28-pre3-bk5-plain/net/decnet/dn_neigh.c 2004-02-18 14:36:32.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/net/decnet/dn_neigh.c 2004-09-30 15:29:46.000000000 +0200
@@ -20,6 +20,7 @@
* Steve Whitehouse : Fixed neighbour states (for now anyway).
* Steve Whitehouse : Made error_report functions dummies. This
* is not the right place to return skbs.
+ * Harald Welte : Port to DaveM's generalized ncache from 2.6.x
*
*/
@@ -33,6 +34,8 @@
#include <linux/string.h>
#include <linux/netfilter_decnet.h>
#include <linux/spinlock.h>
+#include <linux/seq_file.h>
+#include <linux/jhash.h>
#include <asm/atomic.h>
#include <net/neighbour.h>
#include <net/dst.h>
@@ -118,13 +121,7 @@
static u32 dn_neigh_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(dn_address *)pkey;
- hash_val ^= (hash_val >> 10);
- hash_val ^= (hash_val >> 3);
-
- return hash_val & NEIGH_HASHMASK;
+ return jhash_2words(*(dn_address *)pkey, 0, dn_neigh_table.hash_rnd);
}
static int dn_neigh_construct(struct neighbour *neigh)
@@ -322,33 +319,6 @@
}
/*
- * Unfortunately, the neighbour code uses the device in its hash
- * function, so we don't get any advantage from it. This function
- * basically does a neigh_lookup(), but without comparing the device
- * field. This is required for the On-Ethernet cache
- */
-struct neighbour *dn_neigh_lookup(struct neigh_table *tbl, void *ptr)
-{
- struct neighbour *neigh;
- u32 hash_val;
-
- hash_val = tbl->hash(ptr, NULL);
-
- read_lock_bh(&tbl->lock);
- for(neigh = tbl->hash_buckets[hash_val]; neigh != NULL; neigh = neigh->next) {
- if (memcmp(neigh->primary_key, ptr, tbl->key_len) == 0) {
- atomic_inc(&neigh->refcnt);
- read_unlock_bh(&tbl->lock);
- return neigh;
- }
- }
- read_unlock_bh(&tbl->lock);
-
- return NULL;
-}
-
-
-/*
* Any traffic on a pointopoint link causes the timer to be reset
* for the entry in the neighbour table.
*/
@@ -484,113 +454,146 @@
return (*min < priority) ? (min - 6) : NULL;
}
-int dn_neigh_elist(struct net_device *dev, unsigned char *ptr, int n)
+struct elist_cb_state {
+ struct net_device *dev;
+ unsigned char *ptr;
+ unsigned char *rs;
+ int t, n;
+};
+
+static void neigh_elist_cb(struct neighbour *neigh, void *_info)
{
- int t = 0;
- int i;
- struct neighbour *neigh;
+ struct elist_cb_state *s = _info;
+ struct dn_dev *dn_db;
struct dn_neigh *dn;
- struct neigh_table *tbl = &dn_neigh_table;
- unsigned char *rs = ptr;
- struct dn_dev *dn_db = (struct dn_dev *)dev->dn_ptr;
- read_lock_bh(&tbl->lock);
+ if (neigh->dev != s->dev)
+ return;
- for(i = 0; i < NEIGH_HASHMASK; i++) {
- for(neigh = tbl->hash_buckets[i]; neigh != NULL; neigh = neigh->next) {
- if (neigh->dev != dev)
- continue;
- dn = (struct dn_neigh *)neigh;
- if (!(dn->flags & (DN_NDFLAG_R1|DN_NDFLAG_R2)))
- continue;
- if (dn_db->parms.forwarding == 1 && (dn->flags & DN_NDFLAG_R2))
- continue;
- if (t == n)
- rs = dn_find_slot(ptr, n, dn->priority);
- else
- t++;
- if (rs == NULL)
- continue;
- dn_dn2eth(rs, dn->addr);
- rs += 6;
- *rs = neigh->nud_state & NUD_CONNECTED ? 0x80 : 0x0;
- *rs |= dn->priority;
- rs++;
- }
- }
+ dn = (struct dn_neigh *) neigh;
+ if (!(dn->flags & (DN_NDFLAG_R1|DN_NDFLAG_R2)))
+ return;
+
+ dn_db = (struct dn_dev *) s->dev->dn_ptr;
+ if (dn_db->parms.forwarding == 1 && (dn->flags & DN_NDFLAG_R2))
+ return;
- read_unlock_bh(&tbl->lock);
+ if (s->t == s->n)
+ s->rs = dn_find_slot(s->ptr, s->n, dn->priority);
+ else
+ s->t++;
+ if (s->rs == NULL)
+ return;
+
+ dn_dn2eth(s->rs, dn->addr);
+ s->rs += 6;
+ *(s->rs) = neigh->nud_state & NUD_CONNECTED ? 0x80 : 0x0;
+ *(s->rs) |= dn->priority;
+ s->rs++;
+}
+
+int dn_neigh_elist(struct net_device *dev, unsigned char *ptr, int n)
+{
+ struct elist_cb_state state;
+
+ state.dev = dev;
+ state.t = 0;
+ state.n = n;
+ state.ptr = ptr;
+ state.rs = ptr;
+
+ neigh_for_each(&dn_neigh_table, neigh_elist_cb, &state);
- return t;
+ return state.t;
}
+
#endif /* CONFIG_DECNET_ROUTER */
#ifdef CONFIG_PROC_FS
-static int dn_neigh_get_info(char *buffer, char **start, off_t offset, int length)
+
+static inline void dn_neigh_format_entry(struct seq_file *seq,
+ struct neighbour *n)
{
- int len = 0;
- off_t pos = 0;
- off_t begin = 0;
- struct neighbour *n;
- int i;
+ struct dn_neigh *dn = (struct dn_neigh *) n;
char buf[DN_ASCBUF_LEN];
- len += sprintf(buffer + len, "Addr Flags State Use Blksize Dev\n");
-
- for(i=0;i <= NEIGH_HASHMASK; i++) {
- read_lock_bh(&dn_neigh_table.lock);
- n = dn_neigh_table.hash_buckets[i];
- for(; n != NULL; n = n->next) {
- struct dn_neigh *dn = (struct dn_neigh *)n;
-
- read_lock(&n->lock);
- len += sprintf(buffer+len, "%-7s %s%s%s %02x %02d %07ld %-8s\n",
- dn_addr2asc(dn_ntohs(dn->addr), buf),
- (dn->flags&DN_NDFLAG_R1) ? "1" : "-",
- (dn->flags&DN_NDFLAG_R2) ? "2" : "-",
- (dn->flags&DN_NDFLAG_P3) ? "3" : "-",
- dn->n.nud_state,
- atomic_read(&dn->n.refcnt),
- dn->blksize,
- (dn->n.dev) ? dn->n.dev->name : "?");
- read_unlock(&n->lock);
-
- pos = begin + len;
-
- if (pos < offset) {
- len = 0;
- begin = pos;
- }
-
- if (pos > offset + length) {
- read_unlock_bh(&dn_neigh_table.lock);
- goto done;
- }
- }
- read_unlock_bh(&dn_neigh_table.lock);
+ read_lock(&n->lock);
+ seq_printf(seq, "%-7s %s%s%s %02x %02d %07ld %-8s\n",
+ dn_addr2asc(dn_ntohs(dn->addr), buf),
+ (dn->flags&DN_NDFLAG_R1) ? "1" : "-",
+ (dn->flags&DN_NDFLAG_R2) ? "2" : "-",
+ (dn->flags&DN_NDFLAG_P3) ? "3" : "-",
+ dn->n.nud_state,
+ atomic_read(&dn->n.refcnt),
+ dn->blksize,
+ (dn->n.dev) ? dn->n.dev->name : "?");
+ read_unlock(&n->lock);
+}
+
+static int dn_neigh_seq_show(struct seq_file *seq, void *v)
+{
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, "Addr Flags State Use Blksize Dev\n");
+ } else {
+ dn_neigh_format_entry(seq, v);
}
-done:
+ return 0;
+}
- *start = buffer + (offset - begin);
- len -= offset - begin;
+static void *dn_neigh_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ return neigh_seq_start(seq, pos, &dn_neigh_table,
+ NEIGH_SEQ_NEIGH_ONLY);
+}
- if (len > length) len = length;
+static struct seq_operations dn_neigh_seq_ops = {
+ .start = dn_neigh_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = dn_neigh_seq_show,
+};
- return len;
-}
+static int dn_neigh_seq_open(struct inode *inode, struct file *file)
+{
+ struct seq_file *seq;
+ int rc = -ENOMEM;
+ struct neigh_seq_state *s = kmalloc(sizeof(*s), GFP_KERNEL);
+
+ if (!s)
+ goto out;
+
+ memset(s, 0, sizeof(*s));
+ rc = seq_open(file, &dn_neigh_seq_ops);
+ if (rc)
+ goto out_kfree;
+
+ seq = file->private_data;
+ seq->private = s;
+ memset(s, 0, sizeof(*s));
+out:
+ return rc;
+out_kfree:
+ kfree(s);
+ goto out;
+}
+
+static struct file_operations dn_neigh_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = dn_neigh_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+};
#endif
void __init dn_neigh_init(void)
{
neigh_table_init(&dn_neigh_table);
-
-#ifdef CONFIG_PROC_FS
- proc_net_create("decnet_neigh",0,dn_neigh_get_info);
-#endif /* CONFIG_PROC_FS */
+ proc_net_fops_create("decnet_neigh", S_IRUGO, &dn_neigh_seq_fops);
}
void __exit dn_neigh_cleanup(void)
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/decnet/dn_route.c linux-2.4.28-pre3-bk5-neigh/net/decnet/dn_route.c
--- linux-2.4.28-pre3-bk5-plain/net/decnet/dn_route.c 2002-11-29 00:53:15.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/net/decnet/dn_route.c 2004-09-30 12:13:03.000000000 +0200
@@ -761,7 +761,7 @@
/* Look in On-Ethernet cache first */
if (!(flags & MSG_TRYHARD)) {
- if ((neigh = dn_neigh_lookup(&dn_neigh_table, &dst)) != NULL)
+ if ((neigh = neigh_lookup_nodev(&dn_neigh_table, &dst)) != NULL)
goto got_route;
}
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/ipv4/arp.c linux-2.4.28-pre3-bk5-neigh/net/ipv4/arp.c
--- linux-2.4.28-pre3-bk5-plain/net/ipv4/arp.c 2004-04-14 15:05:41.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/net/ipv4/arp.c 2004-09-30 17:07:25.000000000 +0200
@@ -70,6 +70,7 @@
* arp_xmit so intermediate drivers like
* bonding can change the skb before
* sending (e.g. insert 8021q tag).
+ * Harald Welte : convert to make use of jenkins hash
*/
#include <linux/types.h>
@@ -92,6 +93,7 @@
#include <linux/proc_fs.h>
#include <linux/stat.h>
#include <linux/init.h>
+#include <linux/jhash.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#endif
@@ -218,15 +220,7 @@
static u32 arp_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(u32*)pkey;
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
-
- return hash_val;
+ return jhash_2words(*(u32 *)pkey, dev->ifindex, arp_tbl.hash_rnd);
}
static int arp_constructor(struct neighbour *neigh)
@@ -1185,129 +1179,155 @@
return err;
}
+#ifdef CONFIG_PROC_FS
+#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
+
+/* ------------------------------------------------------------------------ */
/*
- * Write the contents of the ARP cache to a PROCfs file.
+ * ax25 -> ASCII conversion
*/
-#ifndef CONFIG_PROC_FS
-static int arp_get_info(char *buffer, char **start, off_t offset, int length) { return 0; }
-#else
-#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
-static char *ax2asc2(ax25_address *a, char *buf);
-#endif
+static char *ax2asc2(ax25_address *a, char *buf)
+{
+ char c, *s;
+ int n;
+
+ for (n = 0, s = buf; n < 6; n++) {
+ c = (a->ax25_call[n] >> 1) & 0x7F;
+
+ if (c != ' ') *s++ = c;
+ }
+
+ *s++ = '-';
+
+ if ((n = ((a->ax25_call[6] >> 1) & 0x0F)) > 9) {
+ *s++ = '1';
+ n -= 10;
+ }
+
+ *s++ = n + '0';
+ *s++ = '\0';
+
+ if (*buf == '\0' || *buf == '-')
+ return "*";
+
+ return buf;
+
+}
+#endif /* CONFIG_AX25 */
+
#define HBUFFERLEN 30
-static int arp_get_info(char *buffer, char **start, off_t offset, int length)
+static void arp_format_neigh_entry(struct seq_file *seq,
+ struct neighbour *n)
{
- int len=0;
- off_t pos=0;
- int size;
char hbuffer[HBUFFERLEN];
- int i,j,k;
const char hexbuf[] = "0123456789ABCDEF";
+ int k, j;
+ char tbuf[16];
+ struct net_device *dev = n->dev;
+ int hatype = dev->type;
- size = sprintf(buffer,"IP address HW type Flags HW address Mask Device\n");
-
- pos+=size;
- len+=size;
+ read_lock(&n->lock);
- for(i=0; i<=NEIGH_HASHMASK; i++) {
- struct neighbour *n;
- read_lock_bh(&arp_tbl.lock);
- for (n=arp_tbl.hash_buckets[i]; n; n=n->next) {
- struct net_device *dev = n->dev;
- int hatype = dev->type;
-
- /* Do not confuse users "arp -a" with magic entries */
- if (!(n->nud_state&~NUD_NOARP))
- continue;
-
- read_lock(&n->lock);
-
-/*
- * Convert hardware address to XX:XX:XX:XX ... form.
- */
+ /* Convert hardware address to XX:XX:XX:XX ... form. */
#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
- if (hatype == ARPHRD_AX25 || hatype == ARPHRD_NETROM)
- ax2asc2((ax25_address *)n->ha, hbuffer);
- else {
-#endif
- for (k=0,j=0;k<HBUFFERLEN-3 && j<dev->addr_len;j++) {
- hbuffer[k++]=hexbuf[(n->ha[j]>>4)&15 ];
- hbuffer[k++]=hexbuf[n->ha[j]&15 ];
- hbuffer[k++]=':';
- }
- hbuffer[--k]=0;
-
+ if (hatype == ARPHRD_AX25 || hatype == ARPHRD_NETROM)
+ ax2asc2((ax25_address *)n->ha, hbuffer);
+ else {
+#endif
+ for (k=0,j=0;k<HBUFFERLEN-3 && j<dev->addr_len;j++) {
+ hbuffer[k++]=hexbuf[(n->ha[j]>>4)&15 ];
+ hbuffer[k++]=hexbuf[n->ha[j]&15 ];
+ hbuffer[k++]=':';
+ }
+ hbuffer[--k]=0;
#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
- }
+ }
#endif
+ sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->primary_key));
+ seq_printf(seq, "%-16s 0x%-10x0x%-10x%s * %s\n",
+ tbuf, hatype, arp_state_to_flags(n), hbuffer, dev->name);
+ read_unlock(&n->lock);
+}
- {
- char tbuf[16];
- sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->primary_key));
- size = sprintf(buffer+len, "%-16s 0x%-10x0x%-10x%s"
- " * %s\n",
- tbuf,
- hatype,
- arp_state_to_flags(n),
- hbuffer,
- dev->name);
- }
+static void arp_format_pneigh_entry(struct seq_file *seq,
+ struct pneigh_entry *n)
+{
+ struct net_device *dev = n->dev;
+ int hatype = dev ? dev->type : 0;
+ char tbuf[16];
- read_unlock(&n->lock);
+ sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->key));
+ seq_printf(seq, "%-16s 0x%-10x0x%-10x%s * %s\n",
+ tbuf, hatype, ATF_PUBL | ATF_PERM, "00:00:00:00:00:00",
+ dev ? dev->name : "*");
+}
- len += size;
- pos += size;
-
- if (pos <= offset)
- len=0;
- if (pos >= offset+length) {
- read_unlock_bh(&arp_tbl.lock);
- goto done;
- }
- }
- read_unlock_bh(&arp_tbl.lock);
+static int arp_seq_show(struct seq_file *seq, void *v)
+{
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, "IP address HW type Flags "
+ "HW address Mask Device\n");
+ } else {
+ struct neigh_seq_state *state = seq->private;
+
+ if (state->flags & NEIGH_SEQ_IS_PNEIGH)
+ arp_format_pneigh_entry(seq, v);
+ else
+ arp_format_neigh_entry(seq, v);
}
- for (i=0; i<=PNEIGH_HASHMASK; i++) {
- struct pneigh_entry *n;
- for (n=arp_tbl.phash_buckets[i]; n; n=n->next) {
- struct net_device *dev = n->dev;
- int hatype = dev ? dev->type : 0;
-
- {
- char tbuf[16];
- sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->key));
- size = sprintf(buffer+len, "%-16s 0x%-10x0x%-10x%s"
- " * %s\n",
- tbuf,
- hatype,
- ATF_PUBL|ATF_PERM,
- "00:00:00:00:00:00",
- dev ? dev->name : "*");
- }
+ return 0;
+}
- len += size;
- pos += size;
-
- if (pos <= offset)
- len=0;
- if (pos >= offset+length)
- goto done;
- }
- }
+static void *arp_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ /* Don't want to confuse "arp -a" w/ magic entries,
+ * so we tell the generic iterator to skip NUD_NOARP.
+ */
+ return neigh_seq_start(seq, pos, &arp_tbl, NEIGH_SEQ_SKIP_NOARP);
+}
+
+/* ------------------------------------------------------------------------ */
+
+static struct seq_operations arp_seq_ops = {
+ .start = arp_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = arp_seq_show,
+};
+
+static int arp_seq_open(struct inode *inode, struct file *file)
+{
+ struct seq_file *seq;
+ int rc = -ENOMEM;
+ struct neigh_seq_state *s = kmalloc(sizeof(*s), GFP_KERNEL);
+
+ if (!s)
+ goto out;
+
+ memset(s, 0, sizeof(*s));
+ rc = seq_open(file, &arp_seq_ops);
+ if (rc)
+ goto out_kfree;
-done:
-
- *start = buffer+len-(pos-offset); /* Start of wanted data */
- len = pos-offset; /* Start slop */
- if (len>length)
- len = length; /* Ending slop */
- if (len<0)
- len = 0;
- return len;
+ seq = file->private_data;
+ seq->private = s;
+out:
+ return rc;
+out_kfree:
+ kfree(s);
+ goto out;
}
-#endif
+
+static struct file_operations arp_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = arp_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+};
+#endif /* CONFIG_PROC_FS */
static int arp_netdev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
@@ -1355,8 +1375,10 @@
dev_add_pack(&arp_packet_type);
- proc_net_create ("arp", 0, arp_get_info);
-
+#ifdef CONFIG_PROC_FS
+ if (!proc_net_fops_create("arp", S_IRUGO, &arp_seq_fops))
+ panic("unable to create arp proc entry");
+#endif
#ifdef CONFIG_SYSCTL
neigh_sysctl_register(NULL, &arp_tbl.parms, NET_IPV4, NET_IPV4_NEIGH, "ipv4");
#endif
@@ -1364,39 +1386,3 @@
}
-#ifdef CONFIG_PROC_FS
-#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
-
-/*
- * ax25 -> ASCII conversion
- */
-char *ax2asc2(ax25_address *a, char *buf)
-{
- char c, *s;
- int n;
-
- for (n = 0, s = buf; n < 6; n++) {
- c = (a->ax25_call[n] >> 1) & 0x7F;
-
- if (c != ' ') *s++ = c;
- }
-
- *s++ = '-';
-
- if ((n = ((a->ax25_call[6] >> 1) & 0x0F)) > 9) {
- *s++ = '1';
- n -= 10;
- }
-
- *s++ = n + '0';
- *s++ = '\0';
-
- if (*buf == '\0' || *buf == '-')
- return "*";
-
- return buf;
-
-}
-
-#endif
-#endif
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/ipv4/route.c linux-2.4.28-pre3-bk5-neigh/net/ipv4/route.c
--- linux-2.4.28-pre3-bk5-plain/net/ipv4/route.c 2003-11-28 19:26:21.000000000 +0100
+++ linux-2.4.28-pre3-bk5-neigh/net/ipv4/route.c 2004-09-30 16:33:07.000000000 +0200
@@ -284,6 +284,7 @@
int i, lcpu;
int len = 0;
+ len += sprintf(buffer+len, "entries in_hit in_slow_tot in_slow_mc in_no_route in_brd in_martian_dst in_martian_src out_hit out_slow_tot out_slow_mc gc_total gc_ignored gc_goal_miss gc_dst_overflow in_hlist_search out_hlist_search\n");
for (lcpu = 0; lcpu < smp_num_cpus; lcpu++) {
i = cpu_logical_map(lcpu);
@@ -2625,7 +2626,8 @@
add_timer(&rt_secret_timer);
proc_net_create ("rt_cache", 0, rt_cache_get_info);
- proc_net_create ("rt_cache_stat", 0, rt_cache_stat_get_info);
+ create_proc_info_entry ("rt_cache", 0, proc_net_stat,
+ rt_cache_stat_get_info);
#ifdef CONFIG_NET_CLS_ROUTE
create_proc_read_entry("net/rt_acct", 0, 0, ip_rt_acct_read, NULL);
#endif
diff -Nru --exclude-from=/sunbeam/home/laforge/scripts/dontdiff linux-2.4.28-pre3-bk5-plain/net/ipv6/ndisc.c linux-2.4.28-pre3-bk5-neigh/net/ipv6/ndisc.c
--- linux-2.4.28-pre3-bk5-plain/net/ipv6/ndisc.c 2004-09-30 11:31:18.000000000 +0200
+++ linux-2.4.28-pre3-bk5-neigh/net/ipv6/ndisc.c 2004-09-30 16:12:47.000000000 +0200
@@ -60,6 +60,7 @@
#include <linux/if_arp.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
+#include <linux/jhash.h>
#include <net/sock.h>
#include <net/snmp.h>
@@ -240,15 +241,14 @@
static u32 ndisc_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
+ const u32 *p32 = pkey;
+ u32 addr_hash, i;
- hash_val = *(u32*)(pkey + sizeof(struct in6_addr) - 4);
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
+ addr_hash = 0;
+ for (i = 0; i < (sizeof(struct in6_addr) / sizeof(u32)); i++)
+ addr_hash ^= *p32++;
- return hash_val;
+ return jhash_2words(addr_hash, dev->ifindex, nd_tbl.hash_rnd);
}
static int ndisc_constructor(struct neighbour *neigh)
@@ -696,9 +696,9 @@
int inc = ipv6_addr_type(daddr)&IPV6_ADDR_MULTICAST;
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_ucast);
/*
* update / create cache entry
@@ -741,9 +741,9 @@
if (addr_type & IPV6_ADDR_UNICAST) {
int inc = ipv6_addr_type(daddr)&IPV6_ADDR_MULTICAST;
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_ucast);
/*
* update / create cache entry
@@ -775,9 +775,11 @@
inc == 0 ||
in6_dev->nd_parms->proxy_delay == 0) {
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl,
+ rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl,
+ rcv_probes_ucast);
neigh = neigh_event_ns(&nd_tbl, lladdr, saddr, dev);
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2.4] backport neighbour cache redesign
2004-09-30 15:47 [PATCH 2.4] backport neighbour cache redesign Harald Welte
@ 2004-10-04 19:17 ` David S. Miller
2004-10-04 20:15 ` Harald Welte
0 siblings, 1 reply; 6+ messages in thread
From: David S. Miller @ 2004-10-04 19:17 UTC (permalink / raw)
To: Harald Welte; +Cc: netdev
Ok I started playing with this, here is what I have currently.
Changes:
1) Merge in missing fixes from 2.6.x port. In particular:
- missing "recalculate hash rand on hash grow"
- missing ARP clip fixes (was initializing neigh
table by hand, needed to use net/core/neighbour.c
interfaces for that). This one wasn't Harald's fault
as this fix was made after he created this 2.4.x backport
2) Missing symbol exports for the new interfaces modules can use,
namely:
neigh_lookup_nodev
neigh_seq_start
neigh_seq_next
neigh_seq_stop
3) Delete instead of "#if 0" out the now totally unused code
inside of net/atm/clip.c
We need to test and verify this patch a lot before I can toss
such a big thing over to Marcelo, but I definitely intend to
send this to him.
===== drivers/net/bonding/bond_main.c 1.86 vs edited =====
--- 1.86/drivers/net/bonding/bond_main.c 2004-01-21 08:55:17 -08:00
+++ edited/drivers/net/bonding/bond_main.c 2004-10-03 15:12:55 -07:00
@@ -3086,8 +3086,6 @@ out:
#ifdef CONFIG_PROC_FS
-#define SEQ_START_TOKEN ((void *)1)
-
static void *bond_info_seq_start(struct seq_file *seq, loff_t *pos)
{
struct bonding *bond = seq->private;
===== fs/proc/root.c 1.8 vs edited =====
--- 1.8/fs/proc/root.c 2003-11-15 08:23:40 -08:00
+++ edited/fs/proc/root.c 2004-10-03 15:12:55 -07:00
@@ -17,7 +17,7 @@
#include <linux/module.h>
#include <asm/bitops.h>
-struct proc_dir_entry *proc_net, *proc_bus, *proc_root_fs, *proc_root_driver;
+struct proc_dir_entry *proc_net, *proc_net_stat, *proc_bus, *proc_root_fs, *proc_root_driver;
#ifdef CONFIG_SYSCTL
struct proc_dir_entry *proc_sys_root;
@@ -38,6 +38,8 @@ void __init proc_root_init(void)
}
proc_misc_init();
proc_net = proc_mkdir("net", 0);
+ proc_net_stat = proc_mkdir("net/stat", NULL);
+
#ifdef CONFIG_SYSVIPC
proc_mkdir("sysvipc", 0);
#endif
@@ -143,5 +145,6 @@ EXPORT_SYMBOL(remove_proc_entry);
EXPORT_SYMBOL(proc_root);
EXPORT_SYMBOL(proc_root_fs);
EXPORT_SYMBOL(proc_net);
+EXPORT_SYMBOL(proc_net_stat);
EXPORT_SYMBOL(proc_bus);
EXPORT_SYMBOL(proc_root_driver);
===== include/linux/proc_fs.h 1.9 vs edited =====
--- 1.9/include/linux/proc_fs.h 2003-09-15 01:39:45 -07:00
+++ edited/include/linux/proc_fs.h 2004-10-03 15:12:55 -07:00
@@ -79,6 +79,7 @@ struct proc_dir_entry {
extern struct proc_dir_entry proc_root;
extern struct proc_dir_entry *proc_root_fs;
extern struct proc_dir_entry *proc_net;
+extern struct proc_dir_entry *proc_net_stat;
extern struct proc_dir_entry *proc_bus;
extern struct proc_dir_entry *proc_root_driver;
extern struct proc_dir_entry *proc_root_kcore;
@@ -168,6 +169,16 @@ static inline struct proc_dir_entry *pro
mode_t mode, get_info_t *get_info)
{
return create_proc_info_entry(name,mode,proc_net,get_info);
+}
+
+static inline struct proc_dir_entry *proc_net_fops_create(const char *name,
+ mode_t mode, struct file_operations *fops)
+{
+ struct proc_dir_entry *res = create_proc_entry(name, mode, proc_net);
+
+ if (res)
+ res->proc_fops = fops;
+ return res;
}
static inline void proc_net_remove(const char *name)
===== include/linux/seq_file.h 1.3 vs edited =====
--- 1.3/include/linux/seq_file.h 2003-08-25 09:34:39 -07:00
+++ edited/include/linux/seq_file.h 2004-10-03 15:12:55 -07:00
@@ -65,5 +65,8 @@ int seq_path(struct seq_file *, struct v
int single_open(struct file *, int (*)(struct seq_file *, void *), void *);
int single_release(struct inode *, struct file *);
int seq_release_private(struct inode *, struct file *);
+
+#define SEQ_START_TOKEN ((void *)1)
+
#endif
#endif
===== include/net/dn_neigh.h 1.1 vs edited =====
--- 1.1/include/net/dn_neigh.h 2002-02-05 09:39:49 -08:00
+++ edited/include/net/dn_neigh.h 2004-10-03 15:12:55 -07:00
@@ -18,7 +18,6 @@ struct dn_neigh {
extern void dn_neigh_init(void);
extern void dn_neigh_cleanup(void);
-extern struct neighbour *dn_neigh_lookup(struct neigh_table *tbl, void *ptr);
extern int dn_neigh_router_hello(struct sk_buff *skb);
extern int dn_neigh_endnode_hello(struct sk_buff *skb);
extern void dn_neigh_pointopoint_hello(struct sk_buff *skb);
===== include/net/neighbour.h 1.3 vs edited =====
--- 1.3/include/net/neighbour.h 2004-07-15 13:33:03 -07:00
+++ edited/include/net/neighbour.h 2004-10-03 15:12:55 -07:00
@@ -7,6 +7,11 @@
* Authors:
* Pedro Roque <pedro_m@yahoo.com>
* Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
+ *
+ * Changes:
+ *
+ * Harald Welte: <laforge@gnumonks.org>
+ * - Add neighbour cache statistics like rtstat
*/
/* The following flags & states are exported to user space,
@@ -45,6 +50,7 @@
#include <asm/atomic.h>
#include <linux/skbuff.h>
+#include <linux/seq_file.h>
#define NUD_IN_TIMER (NUD_INCOMPLETE|NUD_DELAY|NUD_PROBE)
#define NUD_VALID (NUD_PERMANENT|NUD_NOARP|NUD_REACHABLE|NUD_PROBE|NUD_STALE|NUD_DELAY)
@@ -78,12 +84,25 @@ struct neigh_parms
struct neigh_statistics
{
- unsigned long allocs;
- unsigned long res_failed;
- unsigned long rcv_probes_mcast;
- unsigned long rcv_probes_ucast;
+ unsigned long allocs; /* number of allocated neighs */
+ unsigned long destroys; /* number of destroyed neighs */
+ unsigned long hash_grows; /* number of hash resizes */
+
+ unsigned long res_failed; /* nomber of failed resolutions */
+
+ unsigned long lookups; /* number of lookups */
+ unsigned long hits; /* number of hits (among lookups) */
+
+ unsigned long rcv_probes_mcast; /* number of received mcast ipv6 */
+ unsigned long rcv_probes_ucast; /* number of received ucast ipv6 */
+
+ unsigned long periodic_gc_runs; /* number of periodic GC runs */
+ unsigned long forced_gc_runs; /* number of forced GC runs */
};
+#define NEIGH_CACHE_STAT_INC(tbl, field) \
+ ((tbl)->stats[smp_processor_id()].field++)
+
struct neighbour
{
struct neighbour *next;
@@ -128,9 +147,6 @@ struct pneigh_entry
u8 key[0];
};
-#define NEIGH_HASHMASK 0x1F
-#define PNEIGH_HASHMASK 0xF
-
/*
* neighbour table manipulation
*/
@@ -164,9 +180,15 @@ struct neigh_table
struct neigh_parms *parms_list;
kmem_cache_t *kmem_cachep;
struct tasklet_struct gc_task;
- struct neigh_statistics stats;
- struct neighbour *hash_buckets[NEIGH_HASHMASK+1];
- struct pneigh_entry *phash_buckets[PNEIGH_HASHMASK+1];
+ struct neigh_statistics stats[NR_CPUS];
+ struct neighbour **hash_buckets;
+ unsigned int hash_mask;
+ __u32 hash_rnd;
+ unsigned int hash_chain_gc;
+ struct pneigh_entry **phash_buckets;
+#ifdef CONFIG_PROC_FS
+ struct proc_dir_entry *pde;
+#endif
};
extern void neigh_table_init(struct neigh_table *tbl);
@@ -174,6 +196,8 @@ extern int neigh_table_clear(struct ne
extern struct neighbour * neigh_lookup(struct neigh_table *tbl,
const void *pkey,
struct net_device *dev);
+extern struct neighbour * neigh_lookup_nodev(struct neigh_table *tbl,
+ const void *pkey);
extern struct neighbour * neigh_create(struct neigh_table *tbl,
const void *pkey,
struct net_device *dev);
@@ -204,6 +228,24 @@ extern int neigh_dump_info(struct sk_buf
extern int neigh_add(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg);
extern int neigh_delete(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg);
extern void neigh_app_ns(struct neighbour *n);
+
+extern void neigh_for_each(struct neigh_table *tbl, void (*cb)(struct neighbour *, void *), void *cookie);
+extern void __neigh_for_each_release(struct neigh_table *tbl, int (*cb)(struct neighbour *));
+extern void pneigh_for_each(struct neigh_table *tbl, void (*cb)(struct pneigh_entry *));
+
+struct neigh_seq_state {
+ struct neigh_table *tbl;
+ void *(*neigh_sub_iter)(struct neigh_seq_state *state,
+ struct neighbour *n, loff_t *pos);
+ unsigned int bucket;
+ unsigned int flags;
+#define NEIGH_SEQ_NEIGH_ONLY 0x00000001
+#define NEIGH_SEQ_IS_PNEIGH 0x00000002
+#define NEIGH_SEQ_SKIP_NOARP 0x00000004
+};
+extern void *neigh_seq_start(struct seq_file *, loff_t *, struct neigh_table *, unsigned int);
+extern void *neigh_seq_next(struct seq_file *, void *, loff_t *);
+extern void neigh_seq_stop(struct seq_file *, void *);
extern int neigh_sysctl_register(struct net_device *dev, struct neigh_parms *p,
int p_id, int pdev_id, char *p_name);
===== net/netsyms.c 1.49 vs edited =====
--- 1.49/net/netsyms.c 2004-08-29 21:11:49 -07:00
+++ edited/net/netsyms.c 2004-10-03 19:14:10 -07:00
@@ -179,9 +179,13 @@ EXPORT_SYMBOL(neigh_connected_output);
EXPORT_SYMBOL(neigh_update);
EXPORT_SYMBOL(neigh_create);
EXPORT_SYMBOL(neigh_lookup);
+EXPORT_SYMBOL(neigh_lookup_nodev);
EXPORT_SYMBOL(__neigh_event_send);
EXPORT_SYMBOL(neigh_event_ns);
EXPORT_SYMBOL(neigh_ifdown);
+EXPORT_SYMBOL(neigh_seq_start);
+EXPORT_SYMBOL(neigh_seq_next);
+EXPORT_SYMBOL(neigh_seq_stop);
#ifdef CONFIG_ARPD
EXPORT_SYMBOL(neigh_app_ns);
#endif
===== net/atm/clip.c 1.13 vs edited =====
--- 1.13/net/atm/clip.c 2004-02-04 23:20:52 -08:00
+++ edited/net/atm/clip.c 2004-10-03 15:19:49 -07:00
@@ -1,6 +1,10 @@
/* net/atm/clip.c - RFC1577 Classical IP over ATM */
-/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA */
+/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA
+ *
+ * Changes:
+ * Harald Welte <laforge@gnumonks.org>:
+ * - backport DaveM's generalized neighbour cache from 2.6.9-rcX */
#include <linux/config.h>
@@ -24,6 +28,7 @@
#include <linux/if.h> /* for IFF_UP */
#include <linux/inetdevice.h>
#include <linux/bitops.h>
+#include <linux/jhash.h>
#include <net/route.h> /* for struct rtable and routing */
#include <net/icmp.h> /* icmp_send */
#include <asm/param.h> /* for HZ */
@@ -119,64 +124,49 @@ out:
spin_unlock_bh(&entry->neigh->dev->xmit_lock);
}
-
-static void idle_timer_check(unsigned long dummy)
+/* The neighbour entry n->lock is held. */
+static int neigh_check_cb(struct neighbour *n)
{
- int i;
+ struct atmarp_entry *entry = NEIGH2ENTRY(n);
+ struct clip_vcc *cv;
- /*DPRINTK("idle_timer_check\n");*/
- write_lock(&clip_tbl.lock);
- for (i = 0; i <= NEIGH_HASHMASK; i++) {
- struct neighbour **np;
+ for (cv = entry->vccs; cv; cv = cv->next) {
+ unsigned long exp = cv->last_use + cv->idle_timeout;
- for (np = &clip_tbl.hash_buckets[i]; *np;) {
- struct neighbour *n = *np;
- struct atmarp_entry *entry = NEIGH2ENTRY(n);
- struct clip_vcc *clip_vcc;
-
- write_lock(&n->lock);
-
- for (clip_vcc = entry->vccs; clip_vcc;
- clip_vcc = clip_vcc->next)
- if (clip_vcc->idle_timeout &&
- time_after(jiffies, clip_vcc->last_use+
- clip_vcc->idle_timeout)) {
- DPRINTK("releasing vcc %p->%p of "
- "entry %p\n",clip_vcc,clip_vcc->vcc,
- entry);
- vcc_release_async(clip_vcc->vcc,
- -ETIMEDOUT);
- }
- if (entry->vccs ||
- time_before(jiffies, entry->expires)) {
- np = &n->next;
- write_unlock(&n->lock);
- continue;
- }
- if (atomic_read(&n->refcnt) > 1) {
- struct sk_buff *skb;
-
- DPRINTK("destruction postponed with ref %d\n",
- atomic_read(&n->refcnt));
- while ((skb = skb_dequeue(&n->arp_queue)) !=
- NULL)
- dev_kfree_skb(skb);
- np = &n->next;
- write_unlock(&n->lock);
- continue;
- }
- *np = n->next;
- DPRINTK("expired neigh %p\n",n);
- n->dead = 1;
- write_unlock(&n->lock);
- neigh_release(n);
+ if (cv->idle_timeout && time_after(jiffies, exp)) {
+ DPRINTK("releasing vcc %p->%p of entry %p\n",
+ cv, cv->vcc, entry);
+ vcc_release_async(cv->vcc, -ETIMEDOUT);
}
}
+
+ if (entry->vccs || time_before(jiffies, entry->expires))
+ return 0;
+
+ if (atomic_read(&n->refcnt) > 1) {
+ struct sk_buff *skb;
+
+ DPRINTK("destruction postponed with ref %d\n",
+ atomic_read(&n->refcnt));
+
+ while ((skb = skb_dequeue(&n->arp_queue)) != NULL)
+ dev_kfree_skb(skb);
+
+ return 0;
+ }
+
+ DPRINTK("expired neigh %p\n",n);
+ return 1;
+}
+
+static void idle_timer_check(unsigned long dummy)
+{
+ write_lock(&clip_tbl.lock);
+ __neigh_for_each_release(&clip_tbl, neigh_check_cb);
mod_timer(&idle_timer, jiffies+CLIP_CHECK_INTERVAL*HZ);
write_unlock(&clip_tbl.lock);
}
-
static int clip_arp_rcv(struct sk_buff *skb)
{
struct atm_vcc *vcc;
@@ -320,15 +310,7 @@ static int clip_constructor(struct neigh
static u32 clip_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(u32*)pkey;
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
-
- return hash_val;
+ return jhash_2words(*(u32 *)pkey, dev->ifindex, clip_tbl.hash_rnd);
}
@@ -768,19 +750,7 @@ static struct atm_clip_ops __atm_clip_op
static int __init atm_clip_init(void)
{
- /* we should use neigh_table_init() */
- clip_tbl.lock = RW_LOCK_UNLOCKED;
- clip_tbl.kmem_cachep = kmem_cache_create(clip_tbl.id,
- clip_tbl.entry_size, 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
-
- if (!clip_tbl.kmem_cachep)
- return -ENOMEM;
-
- /* so neigh_ifdown() doesn't complain */
- clip_tbl.proxy_timer.data = 0;
- clip_tbl.proxy_timer.function = 0;
- init_timer(&clip_tbl.proxy_timer);
- skb_queue_head_init(&clip_tbl.proxy_queue);
+ neigh_table_init(&clip_tbl);
clip_tbl_hook = &clip_tbl;
atm_clip_ops_set(&__atm_clip_ops);
@@ -794,7 +764,18 @@ static void __exit atm_clip_exit(void)
atm_clip_ops_set(NULL);
+ /* First, stop the idle timer, so it stops banging
+ * on the table.
+ */
+ if (start_timer == 0)
+ del_timer(&idle_timer);
+
+ /* Next, purge the table, so that the device
+ * unregister loop below does not hang due to
+ * device references remaining in the table.
+ */
neigh_ifdown(&clip_tbl, NULL);
+
dev = clip_devs;
while (dev) {
next = PRIV(dev)->next;
@@ -802,9 +783,9 @@ static void __exit atm_clip_exit(void)
kfree(dev);
dev = next;
}
- if (start_timer == 0) del_timer(&idle_timer);
- kmem_cache_destroy(clip_tbl.kmem_cachep);
+ /* Now it is safe to fully shutdown whole table. */
+ neigh_table_clear(&clip_tbl);
clip_tbl_hook = NULL;
}
===== net/atm/proc.c 1.11 vs edited =====
--- 1.11/net/atm/proc.c 2003-10-14 14:01:53 -07:00
+++ edited/net/atm/proc.c 2004-10-03 15:30:31 -07:00
@@ -44,11 +44,6 @@
#include "ipcommon.h"
#endif
-#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
-#include "lec.h"
-#include "lec_arpc.h"
-#endif
-
static ssize_t proc_dev_atm_read(struct file *file,char *buf,size_t count,
loff_t *pos);
static ssize_t proc_spec_atm_read(struct file *file,char *buf,size_t count,
@@ -90,74 +85,202 @@ static void dev_info(const struct atm_de
#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
+#define SEQ_NO_VCC_TOKEN ((void *) 2)
-static int svc_addr(char *buf,struct sockaddr_atmsvc *addr)
+static void svc_addr(struct seq_file *seq, struct sockaddr_atmsvc *addr)
{
static int code[] = { 1,2,10,6,1,0 };
static int e164[] = { 1,8,4,6,1,0 };
- int *fields;
- int len,i,j,pos;
- len = 0;
if (*addr->sas_addr.pub) {
- strcpy(buf,addr->sas_addr.pub);
- len = strlen(addr->sas_addr.pub);
- buf += len;
- if (*addr->sas_addr.prv) {
- *buf++ = '+';
- len++;
- }
+ seq_printf(seq, "%s", addr->sas_addr.pub);
+ if (*addr->sas_addr.prv)
+ seq_putc(seq, '+');
+ } else if (!*addr->sas_addr.prv) {
+ seq_printf(seq, "%s", "(none)");
+ return;
}
- else if (!*addr->sas_addr.prv) {
- strcpy(buf,"(none)");
- return strlen(buf);
- }
if (*addr->sas_addr.prv) {
- len += 44;
- pos = 0;
- fields = *addr->sas_addr.prv == ATM_AFI_E164 ? e164 : code;
+ unsigned char *prv = addr->sas_addr.prv;
+ int *fields;
+ int i, j;
+
+ fields = *prv == ATM_AFI_E164 ? e164 : code;
for (i = 0; fields[i]; i++) {
- for (j = fields[i]; j; j--) {
- sprintf(buf,"%02X",addr->sas_addr.prv[pos++]);
- buf += 2;
- }
- if (fields[i+1]) *buf++ = '.';
+ for (j = fields[i]; j; j--)
+ seq_printf(seq, "%02X", *prv++);
+ if (fields[i+1])
+ seq_putc(seq, '.');
}
}
- return len;
}
-static void atmarp_info(struct net_device *dev,struct atmarp_entry *entry,
- struct clip_vcc *clip_vcc,char *buf)
-{
- unsigned char *ip;
- int svc,off,ip_len;
-
- svc = !clip_vcc || clip_vcc->vcc->sk->family == AF_ATMSVC;
- off = sprintf(buf,"%-6s%-4s%-4s%5ld ",dev->name,svc ? "SVC" : "PVC",
- !clip_vcc || clip_vcc->encap ? "LLC" : "NULL",
- (jiffies-(clip_vcc ? clip_vcc->last_use : entry->neigh->used))/
- HZ);
- ip = (unsigned char *) &entry->ip;
- ip_len = sprintf(buf+off,"%d.%d.%d.%d",ip[0],ip[1],ip[2],ip[3]);
- off += ip_len;
- while (ip_len++ < 16) buf[off++] = ' ';
- if (!clip_vcc)
+static void atmarp_info(struct seq_file *seq, struct net_device *dev,struct
+ atmarp_entry *entry, struct clip_vcc *clip_vcc) {
+ unsigned long exp;
+ char buf[17];
+ int svc, llc, off;
+
+ svc = ((clip_vcc == SEQ_NO_VCC_TOKEN) ||
+ (clip_vcc->vcc->sk->family == AF_ATMSVC));
+
+ llc = ((clip_vcc == SEQ_NO_VCC_TOKEN) ||
+ (clip_vcc->encap));
+
+ if (clip_vcc == SEQ_NO_VCC_TOKEN)
+ exp = entry->neigh->used;
+ else
+ exp = clip_vcc->last_use;
+
+ exp = (jiffies - exp) / HZ;
+
+ seq_printf(seq, "%-6s%-4s%-4s%5ld ",
+ dev->name,
+ svc ? "SVC" : "PVC",
+ llc ? "LLC" : "NULL",
+ exp);
+
+ off = snprintf(buf, sizeof(buf)-1, "%d.%d.%d.%d", NIPQUAD(entry->ip));
+ while (off < 16)
+ buf[off++] = ' ';
+ buf[off] = '\0';
+ seq_printf(seq, "%s", buf);
+
+ if (clip_vcc == SEQ_NO_VCC_TOKEN) {
if (time_before(jiffies, entry->expires))
- strcpy(buf+off,"(resolving)\n");
- else sprintf(buf+off,"(expired, ref %d)\n",
- atomic_read(&entry->neigh->refcnt));
- else if (!svc)
- sprintf(buf+off,"%d.%d.%d\n",clip_vcc->vcc->dev->number,
- clip_vcc->vcc->vpi,clip_vcc->vcc->vci);
- else {
- off += svc_addr(buf+off,&clip_vcc->vcc->remote);
- strcpy(buf+off,"\n");
+ seq_printf(seq, "(resolving)\n");
+ else
+ seq_printf(seq, "(expired, ref %d)\n",
+ atomic_read(&entry->neigh->refcnt));
+ } else if (!svc) {
+ seq_printf(seq, "%d.%d.%d\n",
+ clip_vcc->vcc->dev->number,
+ clip_vcc->vcc->vpi,
+ clip_vcc->vcc->vci);
+ } else {
+ svc_addr(seq, &clip_vcc->vcc->remote);
+ seq_putc(seq, '\n');
+ }
+}
+
+struct clip_seq_state {
+ /* This member must be first. */
+ struct neigh_seq_state ns;
+
+ /* Local to clip specific iteration. */
+ struct clip_vcc *vcc;
+};
+
+static struct clip_vcc *clip_seq_next_vcc(struct atmarp_entry *e,
+ struct clip_vcc *curr)
+{
+ if (!curr) {
+ curr = e->vccs;
+ if (!curr)
+ return SEQ_NO_VCC_TOKEN;
+ return curr;
+ }
+
+ if (curr == SEQ_NO_VCC_TOKEN)
+ return NULL;
+
+ curr = curr->next;
+
+ return curr;
+}
+
+static void *clip_seq_vcc_walk(struct clip_seq_state *state,
+ struct atmarp_entry *e, loff_t *pos)
+{
+ struct clip_vcc *vcc = state->vcc;
+
+ vcc = clip_seq_next_vcc(e, vcc);
+ if (vcc && pos != NULL) {
+ while (*pos) {
+ vcc = clip_seq_next_vcc(e, vcc);
+ if (!vcc)
+ break;
+ --(*pos);
}
+ }
+ state->vcc = vcc;
+
+ return vcc;
+}
+
+static void *clip_seq_sub_iter(struct neigh_seq_state *_state,
+ struct neighbour *n, loff_t *pos)
+{
+ struct clip_seq_state *state = (struct clip_seq_state *) _state;
+
+ return clip_seq_vcc_walk(state, NEIGH2ENTRY(n), pos);
+}
+
+static void *clip_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ return neigh_seq_start(seq, pos, clip_tbl_hook, NEIGH_SEQ_NEIGH_ONLY);
+}
+
+static int clip_seq_show(struct seq_file *seq, void *v)
+{
+ static char atm_arp_banner[] =
+ "IPitf TypeEncp Idle IP address ATM address\n";
+
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, atm_arp_banner);
+ } else {
+ struct clip_seq_state *state = seq->private;
+ struct neighbour *n = v;
+ struct clip_vcc *vcc = state->vcc;
+
+ atmarp_info(seq, n->dev, NEIGH2ENTRY(n), vcc);
+ }
+ return 0;
}
+static struct seq_operations arp_seq_ops = {
+ .start = clip_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = clip_seq_show,
+};
+static int arp_seq_open(struct inode *inode, struct file *file)
+{
+ struct clip_seq_state *state;
+ struct seq_file *seq;
+ int rc = -EAGAIN;
+
+ state = kmalloc(sizeof(*state), GFP_KERNEL);
+ if (!state) {
+ rc = -ENOMEM;
+ goto out_kfree;
+ }
+ memset(state, 0, sizeof(*state));
+ state->ns.neigh_sub_iter = clip_seq_sub_iter;
+
+ rc = seq_open(file, &arp_seq_ops);
+ if (rc)
+ goto out_kfree;
+
+ seq = file->private_data;
+ seq->private = state;
+out:
+ return rc;
+
+out_kfree:
+ kfree(state);
+ goto out;
+}
+
+static struct file_operations arp_seq_fops = {
+ .open = arp_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+ .owner = THIS_MODULE,
+};
#endif
@@ -250,57 +373,6 @@ static void svc_info(struct atm_vcc *vcc
}
-#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
-
-static char*
-lec_arp_get_status_string(unsigned char status)
-{
- switch(status) {
- case ESI_UNKNOWN:
- return "ESI_UNKNOWN ";
- case ESI_ARP_PENDING:
- return "ESI_ARP_PENDING ";
- case ESI_VC_PENDING:
- return "ESI_VC_PENDING ";
- case ESI_FLUSH_PENDING:
- return "ESI_FLUSH_PENDING ";
- case ESI_FORWARD_DIRECT:
- return "ESI_FORWARD_DIRECT";
- default:
- return "<Unknown> ";
- }
-}
-
-static void
-lec_info(struct lec_arp_table *entry, char *buf)
-{
- int j, offset=0;
-
- for(j=0;j<ETH_ALEN;j++) {
- offset+=sprintf(buf+offset,"%2.2x",0xff&entry->mac_addr[j]);
- }
- offset+=sprintf(buf+offset, " ");
- for(j=0;j<ATM_ESA_LEN;j++) {
- offset+=sprintf(buf+offset,"%2.2x",0xff&entry->atm_addr[j]);
- }
- offset+=sprintf(buf+offset, " %s %4.4x",
- lec_arp_get_status_string(entry->status),
- entry->flags&0xffff);
- if (entry->vcc) {
- offset+=sprintf(buf+offset, "%3d %3d ", entry->vcc->vpi,
- entry->vcc->vci);
- } else
- offset+=sprintf(buf+offset, " ");
- if (entry->recv_vcc) {
- offset+=sprintf(buf+offset, " %3d %3d",
- entry->recv_vcc->vpi, entry->recv_vcc->vci);
- }
-
- sprintf(buf+offset,"\n");
-}
-
-#endif
-
static int atm_devices_info(loff_t pos,char *buf)
{
struct atm_dev *dev;
@@ -416,129 +488,6 @@ static int atm_svc_info(loff_t pos,char
return 0;
}
-#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
-static int atm_arp_info(loff_t pos,char *buf)
-{
- struct neighbour *n;
- int i,count;
-
- if (!pos) {
- return sprintf(buf,"IPitf TypeEncp Idle IP address "
- "ATM address\n");
- }
- if (!try_atm_clip_ops())
- return 0;
- count = pos;
- read_lock_bh(&clip_tbl_hook->lock);
- for (i = 0; i <= NEIGH_HASHMASK; i++)
- for (n = clip_tbl_hook->hash_buckets[i]; n; n = n->next) {
- struct atmarp_entry *entry = NEIGH2ENTRY(n);
- struct clip_vcc *vcc;
-
- if (!entry->vccs) {
- if (--count) continue;
- atmarp_info(n->dev,entry,NULL,buf);
- read_unlock_bh(&clip_tbl_hook->lock);
- if (atm_clip_ops->owner)
- __MOD_DEC_USE_COUNT(atm_clip_ops->owner);
- return strlen(buf);
- }
- for (vcc = entry->vccs; vcc;
- vcc = vcc->next) {
- if (--count) continue;
- atmarp_info(n->dev,entry,vcc,buf);
- read_unlock_bh(&clip_tbl_hook->lock);
- if (atm_clip_ops->owner)
- __MOD_DEC_USE_COUNT(atm_clip_ops->owner);
- return strlen(buf);
- }
- }
- read_unlock_bh(&clip_tbl_hook->lock);
- if (atm_clip_ops->owner)
- __MOD_DEC_USE_COUNT(atm_clip_ops->owner);
- return 0;
-}
-#endif
-
-#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
-static int atm_lec_info(loff_t pos,char *buf)
-{
- unsigned long flags;
- struct lec_priv *priv;
- struct lec_arp_table *entry;
- int i, count, d, e;
- struct net_device *dev;
-
- if (!pos) {
- return sprintf(buf,"Itf MAC ATM destination"
- " Status Flags "
- "VPI/VCI Recv VPI/VCI\n");
- }
- if (!try_atm_lane_ops())
- return 0; /* the lane module is not there yet */
-
- count = pos;
- for(d = 0; d < MAX_LEC_ITF; d++) {
- dev = atm_lane_ops->get_lec(d);
- if (!dev || !(priv = (struct lec_priv *) dev->priv))
- continue;
- spin_lock_irqsave(&priv->lec_arp_lock, flags);
- for(i = 0; i < LEC_ARP_TABLE_SIZE; i++) {
- for(entry = priv->lec_arp_tables[i]; entry; entry = entry->next) {
- if (--count)
- continue;
- e = sprintf(buf,"%s ", dev->name);
- lec_info(entry, buf+e);
- spin_unlock_irqrestore(&priv->lec_arp_lock, flags);
- dev_put(dev);
- if (atm_lane_ops->owner)
- __MOD_DEC_USE_COUNT(atm_lane_ops->owner);
- return strlen(buf);
- }
- }
- for(entry = priv->lec_arp_empty_ones; entry; entry = entry->next) {
- if (--count)
- continue;
- e = sprintf(buf,"%s ", dev->name);
- lec_info(entry, buf+e);
- spin_unlock_irqrestore(&priv->lec_arp_lock, flags);
- dev_put(dev);
- if (atm_lane_ops->owner)
- __MOD_DEC_USE_COUNT(atm_lane_ops->owner);
- return strlen(buf);
- }
- for(entry = priv->lec_no_forward; entry; entry=entry->next) {
- if (--count)
- continue;
- e = sprintf(buf,"%s ", dev->name);
- lec_info(entry, buf+e);
- spin_unlock_irqrestore(&priv->lec_arp_lock, flags);
- dev_put(dev);
- if (atm_lane_ops->owner)
- __MOD_DEC_USE_COUNT(atm_lane_ops->owner);
- return strlen(buf);
- }
- for(entry = priv->mcast_fwds; entry; entry = entry->next) {
- if (--count)
- continue;
- e = sprintf(buf,"%s ", dev->name);
- lec_info(entry, buf+e);
- spin_unlock_irqrestore(&priv->lec_arp_lock, flags);
- dev_put(dev);
- if (atm_lane_ops->owner)
- __MOD_DEC_USE_COUNT(atm_lane_ops->owner);
- return strlen(buf);
- }
- spin_unlock_irqrestore(&priv->lec_arp_lock, flags);
- dev_put(dev);
- }
- if (atm_lane_ops->owner)
- __MOD_DEC_USE_COUNT(atm_lane_ops->owner);
- return 0;
-}
-#endif
-
-
static ssize_t proc_dev_atm_read(struct file *file,char *buf,size_t count,
loff_t *pos)
{
@@ -666,7 +615,10 @@ int atm_proc_init(void)
CREATE_ENTRY(svc);
CREATE_ENTRY(vc);
#if defined(CONFIG_ATM_CLIP) || defined(CONFIG_ATM_CLIP_MODULE)
- CREATE_ENTRY(arp);
+ arp = create_proc_entry("arp", S_IRUGO, atm_proc_root);
+ if (!arp)
+ goto cleanup;
+ arp->proc_fops = &arp_seq_fops;
#endif
#if defined(CONFIG_ATM_LANE) || defined(CONFIG_ATM_LANE_MODULE)
CREATE_ENTRY(lec);
===== net/core/neighbour.c 1.12 vs edited =====
--- 1.12/net/core/neighbour.c 2004-07-15 13:33:03 -07:00
+++ edited/net/core/neighbour.c 2004-10-03 15:43:21 -07:00
@@ -12,6 +12,8 @@
*
* Fixes:
* Vitaly E. Lavrov releasing NULL neighbor in neigh_add.
+ * Harald Welte Add neighbour cache statistics like rtstat
+ * Harald Welte port neighbour cache rework from 2.6.9-rcX
*/
#include <linux/config.h>
@@ -20,6 +22,7 @@
#include <linux/socket.h>
#include <linux/sched.h>
#include <linux/netdevice.h>
+#include <linux/proc_fs.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#endif
@@ -27,6 +30,7 @@
#include <net/dst.h>
#include <net/sock.h>
#include <linux/rtnetlink.h>
+#include <linux/random.h>
#define NEIGH_DEBUG 1
@@ -45,6 +49,8 @@
#define NEIGH_PRINTK2 NEIGH_PRINTK
#endif
+#define PNEIGH_HASHMASK 0xF
+
static void neigh_timer_handler(unsigned long arg);
#ifdef CONFIG_ARPD
static void neigh_app_notify(struct neighbour *n);
@@ -54,6 +60,7 @@ void neigh_changeaddr(struct neigh_table
static int neigh_glbl_allocs;
static struct neigh_table *neigh_tables;
+static struct file_operations neigh_stat_seq_fops;
/*
Neighbour hash table buckets are protected with rwlock tbl->lock.
@@ -111,27 +118,21 @@ static int neigh_forced_gc(struct neigh_
int shrunk = 0;
int i;
- for (i=0; i<=NEIGH_HASHMASK; i++) {
+ NEIGH_CACHE_STAT_INC(tbl, forced_gc_runs);
+
+ write_lock_bh(&tbl->lock);
+ for (i = 0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
- write_lock_bh(&tbl->lock);
while ((n = *np) != NULL) {
/* Neighbour record may be discarded if:
- - nobody refers to it.
- - it is not permanent
- - (NEW and probably wrong)
- INCOMPLETE entries are kept at least for
- n->parms->retrans_time, otherwise we could
- flood network with resolution requests.
- It is not clear, what is better table overflow
- or flooding.
+ * - nobody refers to it.
+ * - it is not permanent
*/
write_lock(&n->lock);
if (atomic_read(&n->refcnt) == 1 &&
- !(n->nud_state&NUD_PERMANENT) &&
- (n->nud_state != NUD_INCOMPLETE ||
- jiffies - n->used > n->parms->retrans_time)) {
+ !(n->nud_state&NUD_PERMANENT)) {
*np = n->next;
n->dead = 1;
shrunk = 1;
@@ -142,10 +143,12 @@ static int neigh_forced_gc(struct neigh_
write_unlock(&n->lock);
np = &n->next;
}
- write_unlock_bh(&tbl->lock);
}
tbl->last_flush = jiffies;
+
+ write_unlock_bh(&tbl->lock);
+
return shrunk;
}
@@ -176,7 +179,7 @@ void neigh_changeaddr(struct neigh_table
write_lock_bh(&tbl->lock);
- for (i=0; i <= NEIGH_HASHMASK; i++) {
+ for (i=0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
@@ -203,7 +206,7 @@ int neigh_ifdown(struct neigh_table *tbl
write_lock_bh(&tbl->lock);
- for (i=0; i<=NEIGH_HASHMASK; i++) {
+ for (i = 0; i <= tbl->hash_mask; i++) {
struct neighbour *n, **np;
np = &tbl->hash_buckets[i];
@@ -277,7 +280,7 @@ static struct neighbour *neigh_alloc(str
init_timer(&n->timer);
n->timer.function = neigh_timer_handler;
n->timer.data = (unsigned long)n;
- tbl->stats.allocs++;
+ NEIGH_CACHE_STAT_INC(tbl, allocs);
neigh_glbl_allocs++;
tbl->entries++;
n->tbl = tbl;
@@ -286,20 +289,104 @@ static struct neighbour *neigh_alloc(str
return n;
}
+static struct neighbour **neigh_hash_alloc(unsigned int entries)
+{
+ unsigned long size = entries * sizeof(struct neighbour *);
+ struct neighbour **ret;
+
+ if (size <= PAGE_SIZE) {
+ ret = kmalloc(size, GFP_ATOMIC);
+ } else {
+ ret = (struct neighbour **)
+ __get_free_pages(GFP_ATOMIC, get_order(size));
+ }
+ if (ret)
+ memset(ret, 0, size);
+
+ return ret;
+}
+
+static void neigh_hash_free(struct neighbour **hash, unsigned int entries)
+{
+ unsigned long size = entries * sizeof(struct neighbour *);
+
+ if (size <= PAGE_SIZE)
+ kfree(hash);
+ else
+ free_pages((unsigned long)hash, get_order(size));
+}
+
+static void neigh_hash_grow(struct neigh_table *tbl, unsigned long new_entries)
+{
+ struct neighbour **new_hash, **old_hash;
+ unsigned int i, new_hash_mask, old_entries;
+
+ NEIGH_CACHE_STAT_INC(tbl, hash_grows);
+
+ BUG_ON(new_entries & (new_entries - 1));
+ new_hash = neigh_hash_alloc(new_entries);
+ if (!new_hash)
+ return;
+
+ old_entries = tbl->hash_mask + 1;
+ new_hash_mask = new_entries - 1;
+ old_hash = tbl->hash_buckets;
+
+ get_random_bytes(&tbl->hash_rnd, sizeof(tbl->hash_rnd));
+ for (i = 0; i < old_entries; i++) {
+ struct neighbour *n, *next;
+
+ for (n = old_hash[i]; n; n = next) {
+ unsigned int hash_val = tbl->hash(n->primary_key, n->dev);
+
+ hash_val &= new_hash_mask;
+ next = n->next;
+
+ n->next = new_hash[hash_val];
+ new_hash[hash_val] = n;
+ }
+ }
+ tbl->hash_buckets = new_hash;
+ tbl->hash_mask = new_hash_mask;
+
+ neigh_hash_free(old_hash, old_entries);
+}
+
struct neighbour *neigh_lookup(struct neigh_table *tbl, const void *pkey,
struct net_device *dev)
{
struct neighbour *n;
- u32 hash_val;
int key_len = tbl->key_len;
+ u32 hash_val = tbl->hash(pkey, dev) & tbl->hash_mask;
- hash_val = tbl->hash(pkey, dev);
+ NEIGH_CACHE_STAT_INC(tbl, lookups);
read_lock_bh(&tbl->lock);
for (n = tbl->hash_buckets[hash_val]; n; n = n->next) {
if (dev == n->dev &&
memcmp(n->primary_key, pkey, key_len) == 0) {
neigh_hold(n);
+ NEIGH_CACHE_STAT_INC(tbl, hits);
+ break;
+ }
+ }
+ read_unlock_bh(&tbl->lock);
+ return n;
+}
+
+struct neighbour *neigh_lookup_nodev(struct neigh_table *tbl, const void *pkey)
+{
+ struct neighbour *n;
+ int key_len = tbl->key_len;
+ u32 hash_val = tbl->hash(pkey, NULL) & tbl->hash_mask;
+
+ NEIGH_CACHE_STAT_INC(tbl, lookups);
+
+ read_lock_bh(&tbl->lock);
+ for (n = tbl->hash_buckets[hash_val]; n; n = n->next) {
+ if (!memcmp(n->primary_key, pkey, key_len)) {
+ neigh_hold(n);
+ NEIGH_CACHE_STAT_INC(tbl, hits);
break;
}
}
@@ -338,9 +425,11 @@ struct neighbour * neigh_create(struct n
n->confirmed = jiffies - (n->parms->base_reachable_time<<1);
- hash_val = tbl->hash(pkey, dev);
+ hash_val = tbl->hash(pkey, dev) & tbl->hash_mask;
write_lock_bh(&tbl->lock);
+ if (tbl->entries > (tbl->hash_mask + 1))
+ neigh_hash_grow(tbl, (tbl->hash_mask + 1) << 1);
for (n1 = tbl->hash_buckets[hash_val]; n1; n1 = n1->next) {
if (dev == n1->dev &&
memcmp(n1->primary_key, pkey, key_len) == 0) {
@@ -418,9 +507,9 @@ int pneigh_delete(struct neigh_table *tb
hash_val ^= hash_val>>4;
hash_val &= PNEIGH_HASHMASK;
+ write_lock_bh(&tbl->lock);
for (np = &tbl->phash_buckets[hash_val]; (n=*np) != NULL; np = &n->next) {
if (memcmp(n->key, pkey, key_len) == 0 && n->dev == dev) {
- write_lock_bh(&tbl->lock);
*np = n->next;
write_unlock_bh(&tbl->lock);
if (tbl->pdestructor)
@@ -429,6 +518,7 @@ int pneigh_delete(struct neigh_table *tb
return 0;
}
}
+ write_unlock_bh(&tbl->lock);
return -ENOENT;
}
@@ -462,6 +552,8 @@ void neigh_destroy(struct neighbour *nei
{
struct hh_cache *hh;
+ NEIGH_CACHE_STAT_INC(neigh->tbl, destroys);
+
if (!neigh->dead) {
printk("Destroying alive neighbour %p\n", neigh);
dump_stack();
@@ -566,9 +658,10 @@ static void neigh_sync(struct neighbour
static void SMP_TIMER_NAME(neigh_periodic_timer)(unsigned long arg)
{
struct neigh_table *tbl = (struct neigh_table*)arg;
- unsigned long now = jiffies;
- int i;
+ struct neighbour *n, **np;
+ unsigned long expire, now = jiffies;
+ NEIGH_CACHE_STAT_INC(tbl, periodic_gc_runs);
write_lock(&tbl->lock);
@@ -583,46 +676,49 @@ static void SMP_TIMER_NAME(neigh_periodi
p->reachable_time = neigh_rand_reach_time(p->base_reachable_time);
}
- for (i=0; i <= NEIGH_HASHMASK; i++) {
- struct neighbour *n, **np;
-
- np = &tbl->hash_buckets[i];
- while ((n = *np) != NULL) {
- unsigned state;
-
- write_lock(&n->lock);
+ np = &tbl->hash_buckets[tbl->hash_chain_gc];
+ tbl->hash_chain_gc = ((tbl->hash_chain_gc + 1) & tbl->hash_mask);
- state = n->nud_state;
- if (state&(NUD_PERMANENT|NUD_IN_TIMER)) {
- write_unlock(&n->lock);
- goto next_elt;
- }
+ while ((n = *np) != NULL) {
+ unsigned int state;
- if ((long)(n->used - n->confirmed) < 0)
- n->used = n->confirmed;
+ write_lock(&n->lock);
+
+ state = n->nud_state;
+ if (state & (NUD_PERMANENT | NUD_IN_TIMER)) {
+ write_unlock(&n->lock);
+ goto next_elt;
+ }
- if (atomic_read(&n->refcnt) == 1 &&
- (state == NUD_FAILED || now - n->used > n->parms->gc_staletime)) {
- *np = n->next;
- n->dead = 1;
- write_unlock(&n->lock);
- neigh_release(n);
- continue;
- }
+ if (time_before(n->used, n->confirmed))
+ n->used = n->confirmed;
- if (n->nud_state&NUD_REACHABLE &&
- now - n->confirmed > n->parms->reachable_time) {
- n->nud_state = NUD_STALE;
- neigh_suspect(n);
- }
+ if (atomic_read(&n->refcnt) == 1 &&
+ (state == NUD_FAILED ||
+ time_after(now, n->used + n->parms->gc_staletime))) {
+ *np = n->next;
+ n->dead = 1;
write_unlock(&n->lock);
+ neigh_release(n);
+ continue;
+ }
+ write_unlock(&n->lock);
next_elt:
- np = &n->next;
- }
+ np = &n->next;
}
+
+ /* Cycle through all hash buckets every base_reachable_time/2 ticks.
+ * ARP entry timeouts range from 1/2 base_reachable_time to 3/2
+ * base_reachable_time.
+ */
+ expire = tbl->parms.base_reachable_time >> 1;
+ expire /= (tbl->hash_mask + 1);
+ if (!expire)
+ expire = 1;
+
+ mod_timer(&tbl->gc_timer, now + expire);
- mod_timer(&tbl->gc_timer, now + tbl->gc_interval);
write_unlock(&tbl->lock);
}
@@ -680,7 +776,7 @@ static void neigh_timer_handler(unsigned
neigh->nud_state = NUD_FAILED;
notify = 1;
- neigh->tbl->stats.res_failed++;
+ NEIGH_CACHE_STAT_INC(neigh->tbl, res_failed);
NEIGH_PRINTK2("neigh %p is failed.\n", neigh);
/* It is very thin place. report_unreachable is very complicated
@@ -1132,6 +1228,7 @@ void neigh_parms_release(struct neigh_ta
void neigh_table_init(struct neigh_table *tbl)
{
unsigned long now = jiffies;
+ unsigned long phsize;
tbl->parms.reachable_time = neigh_rand_reach_time(tbl->parms.base_reachable_time);
@@ -1141,6 +1238,30 @@ void neigh_table_init(struct neigh_table
0, SLAB_HWCACHE_ALIGN,
NULL, NULL);
+ if (!tbl->kmem_cachep)
+ panic("cannot create neighbour cache");
+
+#ifdef CONFIG_PROC_FS
+ tbl->pde = create_proc_entry(tbl->id, 0, proc_net_stat);
+ if (!tbl->pde)
+ panic("cannot create neighbour proc dir entry");
+ tbl->pde->proc_fops = &neigh_stat_seq_fops;
+ tbl->pde->data = tbl;
+#endif
+
+ tbl->hash_mask = 1;
+ tbl->hash_buckets = neigh_hash_alloc(tbl->hash_mask + 1);
+
+ phsize = (PNEIGH_HASHMASK + 1) * sizeof(struct pneigh_entry *);
+ tbl->phash_buckets = kmalloc(phsize, GFP_KERNEL);
+
+ if (!tbl->hash_buckets || !tbl->phash_buckets)
+ panic("cannot allocate neighbour cache hashes");
+
+ memset(tbl->phash_buckets, 0, phsize);
+
+ get_random_bytes(&tbl->hash_rnd, sizeof(tbl->hash_rnd));
+
#ifdef CONFIG_SMP
tasklet_init(&tbl->gc_task, SMP_TIMER_NAME(neigh_periodic_timer), (unsigned long)tbl);
#endif
@@ -1148,7 +1269,7 @@ void neigh_table_init(struct neigh_table
tbl->lock = RW_LOCK_UNLOCKED;
tbl->gc_timer.data = (unsigned long)tbl;
tbl->gc_timer.function = neigh_periodic_timer;
- tbl->gc_timer.expires = now + tbl->gc_interval + tbl->parms.reachable_time;
+ tbl->gc_timer.expires = now + 1;
add_timer(&tbl->gc_timer);
init_timer(&tbl->proxy_timer);
@@ -1184,6 +1305,13 @@ int neigh_table_clear(struct neigh_table
}
}
write_unlock(&neigh_tbl_lock);
+
+ neigh_hash_free(tbl->hash_buckets, tbl->hash_mask + 1);
+ tbl->hash_buckets = NULL;
+
+ kfree(tbl->phash_buckets);
+ tbl->phash_buckets = NULL;
+
#ifdef CONFIG_SYSCTL
neigh_sysctl_unregister(&tbl->parms);
#endif
@@ -1364,7 +1492,7 @@ static int neigh_dump_table(struct neigh
s_h = cb->args[1];
s_idx = idx = cb->args[2];
- for (h=0; h <= NEIGH_HASHMASK; h++) {
+ for (h=0; h <= tbl->hash_mask; h++) {
if (h < s_h) continue;
if (h > s_h)
s_idx = 0;
@@ -1414,6 +1542,359 @@ int neigh_dump_info(struct sk_buff *skb,
return skb->len;
}
+
+void neigh_for_each(struct neigh_table *tbl, void (*cb)(struct neighbour *, void *), void *cookie)
+{
+ int chain;
+
+ read_lock_bh(&tbl->lock);
+ for (chain = 0; chain <= tbl->hash_mask; chain++) {
+ struct neighbour *n;
+
+ for (n = tbl->hash_buckets[chain]; n; n = n->next)
+ cb(n, cookie);
+ }
+ read_unlock_bh(&tbl->lock);
+}
+
+/* The tbl->lock must be held as a writer and BH disabled. */
+void __neigh_for_each_release(struct neigh_table *tbl,
+ int (*cb)(struct neighbour *))
+{
+ int chain;
+
+ for (chain = 0; chain <= tbl->hash_mask; chain++) {
+ struct neighbour *n, **np;
+
+ np = &tbl->hash_buckets[chain];
+ while ((n = *np) != NULL) {
+ int release;
+
+ write_lock(&n->lock);
+ release = cb(n);
+ if (release) {
+ *np = n->next;
+ n->dead = 1;
+ } else
+ np = &n->next;
+ write_unlock(&n->lock);
+ if (release)
+ neigh_release(n);
+ }
+ }
+}
+
+#ifdef CONFIG_PROC_FS
+
+static struct neighbour *neigh_get_first(struct seq_file *seq)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+ struct neighbour *n = NULL;
+ int bucket = state->bucket;
+
+ state->flags &= ~NEIGH_SEQ_IS_PNEIGH;
+ for (bucket = 0; bucket <= tbl->hash_mask; bucket++) {
+ n = tbl->hash_buckets[bucket];
+
+ while (n) {
+ if (state->neigh_sub_iter) {
+ loff_t fakep = 0;
+ void *v;
+
+ v = state->neigh_sub_iter(state, n, &fakep);
+ if (!v)
+ goto next;
+ }
+ if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
+ break;
+ if (n->nud_state & ~NUD_NOARP)
+ break;
+ next:
+ n = n->next;
+ }
+
+ if (n)
+ break;
+ }
+ state->bucket = bucket;
+
+ return n;
+}
+
+static struct neighbour *neigh_get_next(struct seq_file *seq,
+ struct neighbour *n,
+ loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ if (state->neigh_sub_iter) {
+ void *v = state->neigh_sub_iter(state, n, pos);
+ if (v)
+ return n;
+ }
+ n = n->next;
+
+ while (1) {
+ while (n) {
+ if (state->neigh_sub_iter) {
+ void *v = state->neigh_sub_iter(state, n, pos);
+ if (v)
+ return n;
+ goto next;
+ }
+ if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
+ break;
+
+ if (n->nud_state & ~NUD_NOARP)
+ break;
+ next:
+ n = n->next;
+ }
+
+ if (n)
+ break;
+
+ if (++state->bucket > tbl->hash_mask)
+ break;
+
+ n = tbl->hash_buckets[state->bucket];
+ }
+
+ if (n && pos)
+ --(*pos);
+ return n;
+}
+
+static struct neighbour *neigh_get_idx(struct seq_file *seq, loff_t *pos)
+{
+ struct neighbour *n = neigh_get_first(seq);
+
+ if (n) {
+ while (*pos) {
+ n = neigh_get_next(seq, n, pos);
+ if (!n)
+ break;
+ }
+ }
+ return *pos ? NULL : n;
+}
+
+static struct pneigh_entry *pneigh_get_first(struct seq_file *seq)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+ struct pneigh_entry *pn = NULL;
+ int bucket = state->bucket;
+
+ state->flags |= NEIGH_SEQ_IS_PNEIGH;
+ for (bucket = 0; bucket <= PNEIGH_HASHMASK; bucket++) {
+ pn = tbl->phash_buckets[bucket];
+ if (pn)
+ break;
+ }
+ state->bucket = bucket;
+
+ return pn;
+}
+
+static struct pneigh_entry *pneigh_get_next(struct seq_file *seq,
+ struct pneigh_entry *pn,
+ loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ pn = pn->next;
+ while (!pn) {
+ if (++state->bucket > PNEIGH_HASHMASK)
+ break;
+ pn = tbl->phash_buckets[state->bucket];
+ if (pn)
+ break;
+ }
+
+ if (pn && pos)
+ --(*pos);
+
+ return pn;
+}
+
+static struct pneigh_entry *pneigh_get_idx(struct seq_file *seq, loff_t *pos)
+{
+ struct pneigh_entry *pn = pneigh_get_first(seq);
+
+ if (pn) {
+ while (*pos) {
+ pn = pneigh_get_next(seq, pn, pos);
+ if (!pn)
+ break;
+ }
+ }
+ return *pos ? NULL : pn;
+}
+
+static void *neigh_get_idx_any(struct seq_file *seq, loff_t *pos)
+{
+ struct neigh_seq_state *state = seq->private;
+ void *rc;
+
+ rc = neigh_get_idx(seq, pos);
+ if (!rc && !(state->flags & NEIGH_SEQ_NEIGH_ONLY))
+ rc = pneigh_get_idx(seq, pos);
+
+ return rc;
+}
+
+void *neigh_seq_start(struct seq_file *seq, loff_t *pos, struct neigh_table *tbl, unsigned int neigh_seq_flags)
+{
+ struct neigh_seq_state *state = seq->private;
+ loff_t pos_minus_one;
+
+ state->tbl = tbl;
+ state->bucket = 0;
+ state->flags = (neigh_seq_flags & ~NEIGH_SEQ_IS_PNEIGH);
+
+ read_lock_bh(&tbl->lock);
+
+ pos_minus_one = *pos - 1;
+ return *pos ? neigh_get_idx_any(seq, &pos_minus_one) : SEQ_START_TOKEN;
+}
+
+void *neigh_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+ struct neigh_seq_state *state;
+ void *rc;
+
+ if (v == SEQ_START_TOKEN) {
+ rc = neigh_get_idx(seq, pos);
+ goto out;
+ }
+
+ state = seq->private;
+ if (!(state->flags & NEIGH_SEQ_IS_PNEIGH)) {
+ rc = neigh_get_next(seq, v, NULL);
+ if (rc)
+ goto out;
+ if (!(state->flags & NEIGH_SEQ_NEIGH_ONLY))
+ rc = pneigh_get_first(seq);
+ } else {
+ BUG_ON(state->flags & NEIGH_SEQ_NEIGH_ONLY);
+ rc = pneigh_get_next(seq, v, NULL);
+ }
+out:
+ ++(*pos);
+ return rc;
+}
+
+void neigh_seq_stop(struct seq_file *seq, void *v)
+{
+ struct neigh_seq_state *state = seq->private;
+ struct neigh_table *tbl = state->tbl;
+
+ read_unlock_bh(&tbl->lock);
+}
+
+/* statistics via seq_file */
+
+static void *neigh_stat_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ int lcpu;
+
+ if (*pos == 0)
+ return SEQ_START_TOKEN;
+
+ for (lcpu = *pos-1; lcpu < smp_num_cpus; ++lcpu) {
+ int i = cpu_logical_map(lcpu);
+ *pos = lcpu+1;
+ return &tbl->stats[i];
+ }
+ return NULL;
+}
+
+static void *neigh_stat_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ int lcpu;
+
+ for (lcpu = *pos; lcpu < smp_num_cpus; ++lcpu) {
+ int i = cpu_logical_map(lcpu);
+ *pos = lcpu+1;
+ return &tbl->stats[i];
+ }
+ return NULL;
+}
+
+static void neigh_stat_seq_stop(struct seq_file *seq, void *v)
+{
+
+}
+
+static int neigh_stat_seq_show(struct seq_file *seq, void *v)
+{
+ struct proc_dir_entry *pde = seq->private;
+ struct neigh_table *tbl = pde->data;
+ struct neigh_statistics *st = v;
+
+ if (v == SEQ_START_TOKEN) {
+ seq_printf(seq, "entries allocs destroys hash_grows lookups hits res_failed rcv_probes_mcast rcv_probes_ucast periodic_gc_runs forced_gc_runs forced_gc_goal_miss\n");
+ return 0;
+ }
+
+ seq_printf(seq, "%08x %08lx %08lx %08lx %08lx %08lx %08lx "
+ "%08lx %08lx %08lx %08lx\n",
+ tbl->entries,
+
+ st->allocs,
+ st->destroys,
+ st->hash_grows,
+
+ st->lookups,
+ st->hits,
+
+ st->res_failed,
+
+ st->rcv_probes_mcast,
+ st->rcv_probes_ucast,
+
+ st->periodic_gc_runs,
+ st->forced_gc_runs
+ );
+
+ return 0;
+}
+
+static struct seq_operations neigh_stat_seq_ops = {
+ .start = neigh_stat_seq_start,
+ .next = neigh_stat_seq_next,
+ .stop = neigh_stat_seq_stop,
+ .show = neigh_stat_seq_show,
+};
+
+static int neigh_stat_seq_open(struct inode *inode, struct file *file)
+{
+ int ret = seq_open(file, &neigh_stat_seq_ops);
+
+ if (!ret) {
+ struct seq_file *sf = file->private_data;
+ sf->private = PDE(inode);
+ }
+ return ret;
+};
+
+static struct file_operations neigh_stat_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = neigh_stat_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release,
+};
+
+#endif /* CONFIG_PROC_FS */
#ifdef CONFIG_ARPD
void neigh_app_ns(struct neighbour *n)
===== net/decnet/dn_neigh.c 1.3 vs edited =====
--- 1.3/net/decnet/dn_neigh.c 2004-01-25 21:21:33 -08:00
+++ edited/net/decnet/dn_neigh.c 2004-10-03 15:12:55 -07:00
@@ -20,6 +20,7 @@
* Steve Whitehouse : Fixed neighbour states (for now anyway).
* Steve Whitehouse : Made error_report functions dummies. This
* is not the right place to return skbs.
+ * Harald Welte : Port to DaveM's generalized ncache from 2.6.x
*
*/
@@ -33,6 +34,8 @@
#include <linux/string.h>
#include <linux/netfilter_decnet.h>
#include <linux/spinlock.h>
+#include <linux/seq_file.h>
+#include <linux/jhash.h>
#include <asm/atomic.h>
#include <net/neighbour.h>
#include <net/dst.h>
@@ -118,13 +121,7 @@ struct neigh_table dn_neigh_table = {
static u32 dn_neigh_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(dn_address *)pkey;
- hash_val ^= (hash_val >> 10);
- hash_val ^= (hash_val >> 3);
-
- return hash_val & NEIGH_HASHMASK;
+ return jhash_2words(*(dn_address *)pkey, 0, dn_neigh_table.hash_rnd);
}
static int dn_neigh_construct(struct neighbour *neigh)
@@ -322,33 +319,6 @@ static int dn_phase3_output(struct sk_bu
}
/*
- * Unfortunately, the neighbour code uses the device in its hash
- * function, so we don't get any advantage from it. This function
- * basically does a neigh_lookup(), but without comparing the device
- * field. This is required for the On-Ethernet cache
- */
-struct neighbour *dn_neigh_lookup(struct neigh_table *tbl, void *ptr)
-{
- struct neighbour *neigh;
- u32 hash_val;
-
- hash_val = tbl->hash(ptr, NULL);
-
- read_lock_bh(&tbl->lock);
- for(neigh = tbl->hash_buckets[hash_val]; neigh != NULL; neigh = neigh->next) {
- if (memcmp(neigh->primary_key, ptr, tbl->key_len) == 0) {
- atomic_inc(&neigh->refcnt);
- read_unlock_bh(&tbl->lock);
- return neigh;
- }
- }
- read_unlock_bh(&tbl->lock);
-
- return NULL;
-}
-
-
-/*
* Any traffic on a pointopoint link causes the timer to be reset
* for the entry in the neighbour table.
*/
@@ -484,113 +454,146 @@ static char *dn_find_slot(char *base, in
return (*min < priority) ? (min - 6) : NULL;
}
-int dn_neigh_elist(struct net_device *dev, unsigned char *ptr, int n)
+struct elist_cb_state {
+ struct net_device *dev;
+ unsigned char *ptr;
+ unsigned char *rs;
+ int t, n;
+};
+
+static void neigh_elist_cb(struct neighbour *neigh, void *_info)
{
- int t = 0;
- int i;
- struct neighbour *neigh;
+ struct elist_cb_state *s = _info;
+ struct dn_dev *dn_db;
struct dn_neigh *dn;
- struct neigh_table *tbl = &dn_neigh_table;
- unsigned char *rs = ptr;
- struct dn_dev *dn_db = (struct dn_dev *)dev->dn_ptr;
- read_lock_bh(&tbl->lock);
+ if (neigh->dev != s->dev)
+ return;
- for(i = 0; i < NEIGH_HASHMASK; i++) {
- for(neigh = tbl->hash_buckets[i]; neigh != NULL; neigh = neigh->next) {
- if (neigh->dev != dev)
- continue;
- dn = (struct dn_neigh *)neigh;
- if (!(dn->flags & (DN_NDFLAG_R1|DN_NDFLAG_R2)))
- continue;
- if (dn_db->parms.forwarding == 1 && (dn->flags & DN_NDFLAG_R2))
- continue;
- if (t == n)
- rs = dn_find_slot(ptr, n, dn->priority);
- else
- t++;
- if (rs == NULL)
- continue;
- dn_dn2eth(rs, dn->addr);
- rs += 6;
- *rs = neigh->nud_state & NUD_CONNECTED ? 0x80 : 0x0;
- *rs |= dn->priority;
- rs++;
- }
- }
+ dn = (struct dn_neigh *) neigh;
+ if (!(dn->flags & (DN_NDFLAG_R1|DN_NDFLAG_R2)))
+ return;
+
+ dn_db = (struct dn_dev *) s->dev->dn_ptr;
+ if (dn_db->parms.forwarding == 1 && (dn->flags & DN_NDFLAG_R2))
+ return;
- read_unlock_bh(&tbl->lock);
+ if (s->t == s->n)
+ s->rs = dn_find_slot(s->ptr, s->n, dn->priority);
+ else
+ s->t++;
+ if (s->rs == NULL)
+ return;
+
+ dn_dn2eth(s->rs, dn->addr);
+ s->rs += 6;
+ *(s->rs) = neigh->nud_state & NUD_CONNECTED ? 0x80 : 0x0;
+ *(s->rs) |= dn->priority;
+ s->rs++;
+}
+
+int dn_neigh_elist(struct net_device *dev, unsigned char *ptr, int n)
+{
+ struct elist_cb_state state;
+
+ state.dev = dev;
+ state.t = 0;
+ state.n = n;
+ state.ptr = ptr;
+ state.rs = ptr;
+
+ neigh_for_each(&dn_neigh_table, neigh_elist_cb, &state);
- return t;
+ return state.t;
}
+
#endif /* CONFIG_DECNET_ROUTER */
#ifdef CONFIG_PROC_FS
-static int dn_neigh_get_info(char *buffer, char **start, off_t offset, int length)
+
+static inline void dn_neigh_format_entry(struct seq_file *seq,
+ struct neighbour *n)
{
- int len = 0;
- off_t pos = 0;
- off_t begin = 0;
- struct neighbour *n;
- int i;
+ struct dn_neigh *dn = (struct dn_neigh *) n;
char buf[DN_ASCBUF_LEN];
- len += sprintf(buffer + len, "Addr Flags State Use Blksize Dev\n");
-
- for(i=0;i <= NEIGH_HASHMASK; i++) {
- read_lock_bh(&dn_neigh_table.lock);
- n = dn_neigh_table.hash_buckets[i];
- for(; n != NULL; n = n->next) {
- struct dn_neigh *dn = (struct dn_neigh *)n;
-
- read_lock(&n->lock);
- len += sprintf(buffer+len, "%-7s %s%s%s %02x %02d %07ld %-8s\n",
- dn_addr2asc(dn_ntohs(dn->addr), buf),
- (dn->flags&DN_NDFLAG_R1) ? "1" : "-",
- (dn->flags&DN_NDFLAG_R2) ? "2" : "-",
- (dn->flags&DN_NDFLAG_P3) ? "3" : "-",
- dn->n.nud_state,
- atomic_read(&dn->n.refcnt),
- dn->blksize,
- (dn->n.dev) ? dn->n.dev->name : "?");
- read_unlock(&n->lock);
-
- pos = begin + len;
-
- if (pos < offset) {
- len = 0;
- begin = pos;
- }
-
- if (pos > offset + length) {
- read_unlock_bh(&dn_neigh_table.lock);
- goto done;
- }
- }
- read_unlock_bh(&dn_neigh_table.lock);
+ read_lock(&n->lock);
+ seq_printf(seq, "%-7s %s%s%s %02x %02d %07ld %-8s\n",
+ dn_addr2asc(dn_ntohs(dn->addr), buf),
+ (dn->flags&DN_NDFLAG_R1) ? "1" : "-",
+ (dn->flags&DN_NDFLAG_R2) ? "2" : "-",
+ (dn->flags&DN_NDFLAG_P3) ? "3" : "-",
+ dn->n.nud_state,
+ atomic_read(&dn->n.refcnt),
+ dn->blksize,
+ (dn->n.dev) ? dn->n.dev->name : "?");
+ read_unlock(&n->lock);
+}
+
+static int dn_neigh_seq_show(struct seq_file *seq, void *v)
+{
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, "Addr Flags State Use Blksize Dev\n");
+ } else {
+ dn_neigh_format_entry(seq, v);
}
-done:
+ return 0;
+}
- *start = buffer + (offset - begin);
- len -= offset - begin;
+static void *dn_neigh_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ return neigh_seq_start(seq, pos, &dn_neigh_table,
+ NEIGH_SEQ_NEIGH_ONLY);
+}
- if (len > length) len = length;
+static struct seq_operations dn_neigh_seq_ops = {
+ .start = dn_neigh_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = dn_neigh_seq_show,
+};
- return len;
-}
+static int dn_neigh_seq_open(struct inode *inode, struct file *file)
+{
+ struct seq_file *seq;
+ int rc = -ENOMEM;
+ struct neigh_seq_state *s = kmalloc(sizeof(*s), GFP_KERNEL);
+
+ if (!s)
+ goto out;
+
+ memset(s, 0, sizeof(*s));
+ rc = seq_open(file, &dn_neigh_seq_ops);
+ if (rc)
+ goto out_kfree;
+
+ seq = file->private_data;
+ seq->private = s;
+ memset(s, 0, sizeof(*s));
+out:
+ return rc;
+out_kfree:
+ kfree(s);
+ goto out;
+}
+
+static struct file_operations dn_neigh_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = dn_neigh_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+};
#endif
void __init dn_neigh_init(void)
{
neigh_table_init(&dn_neigh_table);
-
-#ifdef CONFIG_PROC_FS
- proc_net_create("decnet_neigh",0,dn_neigh_get_info);
-#endif /* CONFIG_PROC_FS */
+ proc_net_fops_create("decnet_neigh", S_IRUGO, &dn_neigh_seq_fops);
}
void __exit dn_neigh_cleanup(void)
===== net/decnet/dn_route.c 1.6 vs edited =====
--- 1.6/net/decnet/dn_route.c 2002-08-12 13:43:21 -07:00
+++ edited/net/decnet/dn_route.c 2004-10-03 15:12:55 -07:00
@@ -761,7 +761,7 @@ static int dn_route_output_slow(struct d
/* Look in On-Ethernet cache first */
if (!(flags & MSG_TRYHARD)) {
- if ((neigh = dn_neigh_lookup(&dn_neigh_table, &dst)) != NULL)
+ if ((neigh = neigh_lookup_nodev(&dn_neigh_table, &dst)) != NULL)
goto got_route;
}
===== net/ipv4/arp.c 1.16 vs edited =====
--- 1.16/net/ipv4/arp.c 2004-02-19 10:13:39 -08:00
+++ edited/net/ipv4/arp.c 2004-10-03 15:12:55 -07:00
@@ -70,6 +70,7 @@
* arp_xmit so intermediate drivers like
* bonding can change the skb before
* sending (e.g. insert 8021q tag).
+ * Harald Welte : convert to make use of jenkins hash
*/
#include <linux/types.h>
@@ -92,6 +93,7 @@
#include <linux/proc_fs.h>
#include <linux/stat.h>
#include <linux/init.h>
+#include <linux/jhash.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#endif
@@ -218,15 +220,7 @@ int arp_mc_map(u32 addr, u8 *haddr, stru
static u32 arp_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
-
- hash_val = *(u32*)pkey;
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
-
- return hash_val;
+ return jhash_2words(*(u32 *)pkey, dev->ifindex, arp_tbl.hash_rnd);
}
static int arp_constructor(struct neighbour *neigh)
@@ -1185,129 +1179,155 @@ out:
return err;
}
+#ifdef CONFIG_PROC_FS
+#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
+
+/* ------------------------------------------------------------------------ */
/*
- * Write the contents of the ARP cache to a PROCfs file.
+ * ax25 -> ASCII conversion
*/
-#ifndef CONFIG_PROC_FS
-static int arp_get_info(char *buffer, char **start, off_t offset, int length) { return 0; }
-#else
-#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
-static char *ax2asc2(ax25_address *a, char *buf);
-#endif
+static char *ax2asc2(ax25_address *a, char *buf)
+{
+ char c, *s;
+ int n;
+
+ for (n = 0, s = buf; n < 6; n++) {
+ c = (a->ax25_call[n] >> 1) & 0x7F;
+
+ if (c != ' ') *s++ = c;
+ }
+
+ *s++ = '-';
+
+ if ((n = ((a->ax25_call[6] >> 1) & 0x0F)) > 9) {
+ *s++ = '1';
+ n -= 10;
+ }
+
+ *s++ = n + '0';
+ *s++ = '\0';
+
+ if (*buf == '\0' || *buf == '-')
+ return "*";
+
+ return buf;
+
+}
+#endif /* CONFIG_AX25 */
+
#define HBUFFERLEN 30
-static int arp_get_info(char *buffer, char **start, off_t offset, int length)
+static void arp_format_neigh_entry(struct seq_file *seq,
+ struct neighbour *n)
{
- int len=0;
- off_t pos=0;
- int size;
char hbuffer[HBUFFERLEN];
- int i,j,k;
const char hexbuf[] = "0123456789ABCDEF";
+ int k, j;
+ char tbuf[16];
+ struct net_device *dev = n->dev;
+ int hatype = dev->type;
- size = sprintf(buffer,"IP address HW type Flags HW address Mask Device\n");
-
- pos+=size;
- len+=size;
+ read_lock(&n->lock);
- for(i=0; i<=NEIGH_HASHMASK; i++) {
- struct neighbour *n;
- read_lock_bh(&arp_tbl.lock);
- for (n=arp_tbl.hash_buckets[i]; n; n=n->next) {
- struct net_device *dev = n->dev;
- int hatype = dev->type;
-
- /* Do not confuse users "arp -a" with magic entries */
- if (!(n->nud_state&~NUD_NOARP))
- continue;
-
- read_lock(&n->lock);
-
-/*
- * Convert hardware address to XX:XX:XX:XX ... form.
- */
+ /* Convert hardware address to XX:XX:XX:XX ... form. */
#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
- if (hatype == ARPHRD_AX25 || hatype == ARPHRD_NETROM)
- ax2asc2((ax25_address *)n->ha, hbuffer);
- else {
-#endif
- for (k=0,j=0;k<HBUFFERLEN-3 && j<dev->addr_len;j++) {
- hbuffer[k++]=hexbuf[(n->ha[j]>>4)&15 ];
- hbuffer[k++]=hexbuf[n->ha[j]&15 ];
- hbuffer[k++]=':';
- }
- hbuffer[--k]=0;
-
+ if (hatype == ARPHRD_AX25 || hatype == ARPHRD_NETROM)
+ ax2asc2((ax25_address *)n->ha, hbuffer);
+ else {
+#endif
+ for (k=0,j=0;k<HBUFFERLEN-3 && j<dev->addr_len;j++) {
+ hbuffer[k++]=hexbuf[(n->ha[j]>>4)&15 ];
+ hbuffer[k++]=hexbuf[n->ha[j]&15 ];
+ hbuffer[k++]=':';
+ }
+ hbuffer[--k]=0;
#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
- }
+ }
#endif
+ sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->primary_key));
+ seq_printf(seq, "%-16s 0x%-10x0x%-10x%s * %s\n",
+ tbuf, hatype, arp_state_to_flags(n), hbuffer, dev->name);
+ read_unlock(&n->lock);
+}
- {
- char tbuf[16];
- sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->primary_key));
- size = sprintf(buffer+len, "%-16s 0x%-10x0x%-10x%s"
- " * %s\n",
- tbuf,
- hatype,
- arp_state_to_flags(n),
- hbuffer,
- dev->name);
- }
+static void arp_format_pneigh_entry(struct seq_file *seq,
+ struct pneigh_entry *n)
+{
+ struct net_device *dev = n->dev;
+ int hatype = dev ? dev->type : 0;
+ char tbuf[16];
- read_unlock(&n->lock);
+ sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->key));
+ seq_printf(seq, "%-16s 0x%-10x0x%-10x%s * %s\n",
+ tbuf, hatype, ATF_PUBL | ATF_PERM, "00:00:00:00:00:00",
+ dev ? dev->name : "*");
+}
- len += size;
- pos += size;
-
- if (pos <= offset)
- len=0;
- if (pos >= offset+length) {
- read_unlock_bh(&arp_tbl.lock);
- goto done;
- }
- }
- read_unlock_bh(&arp_tbl.lock);
+static int arp_seq_show(struct seq_file *seq, void *v)
+{
+ if (v == SEQ_START_TOKEN) {
+ seq_puts(seq, "IP address HW type Flags "
+ "HW address Mask Device\n");
+ } else {
+ struct neigh_seq_state *state = seq->private;
+
+ if (state->flags & NEIGH_SEQ_IS_PNEIGH)
+ arp_format_pneigh_entry(seq, v);
+ else
+ arp_format_neigh_entry(seq, v);
}
- for (i=0; i<=PNEIGH_HASHMASK; i++) {
- struct pneigh_entry *n;
- for (n=arp_tbl.phash_buckets[i]; n; n=n->next) {
- struct net_device *dev = n->dev;
- int hatype = dev ? dev->type : 0;
-
- {
- char tbuf[16];
- sprintf(tbuf, "%u.%u.%u.%u", NIPQUAD(*(u32*)n->key));
- size = sprintf(buffer+len, "%-16s 0x%-10x0x%-10x%s"
- " * %s\n",
- tbuf,
- hatype,
- ATF_PUBL|ATF_PERM,
- "00:00:00:00:00:00",
- dev ? dev->name : "*");
- }
+ return 0;
+}
- len += size;
- pos += size;
-
- if (pos <= offset)
- len=0;
- if (pos >= offset+length)
- goto done;
- }
- }
+static void *arp_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ /* Don't want to confuse "arp -a" w/ magic entries,
+ * so we tell the generic iterator to skip NUD_NOARP.
+ */
+ return neigh_seq_start(seq, pos, &arp_tbl, NEIGH_SEQ_SKIP_NOARP);
+}
+
+/* ------------------------------------------------------------------------ */
+
+static struct seq_operations arp_seq_ops = {
+ .start = arp_seq_start,
+ .next = neigh_seq_next,
+ .stop = neigh_seq_stop,
+ .show = arp_seq_show,
+};
+
+static int arp_seq_open(struct inode *inode, struct file *file)
+{
+ struct seq_file *seq;
+ int rc = -ENOMEM;
+ struct neigh_seq_state *s = kmalloc(sizeof(*s), GFP_KERNEL);
+
+ if (!s)
+ goto out;
+
+ memset(s, 0, sizeof(*s));
+ rc = seq_open(file, &arp_seq_ops);
+ if (rc)
+ goto out_kfree;
-done:
-
- *start = buffer+len-(pos-offset); /* Start of wanted data */
- len = pos-offset; /* Start slop */
- if (len>length)
- len = length; /* Ending slop */
- if (len<0)
- len = 0;
- return len;
+ seq = file->private_data;
+ seq->private = s;
+out:
+ return rc;
+out_kfree:
+ kfree(s);
+ goto out;
}
-#endif
+
+static struct file_operations arp_seq_fops = {
+ .owner = THIS_MODULE,
+ .open = arp_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_private,
+};
+#endif /* CONFIG_PROC_FS */
static int arp_netdev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
@@ -1355,8 +1375,10 @@ void __init arp_init (void)
dev_add_pack(&arp_packet_type);
- proc_net_create ("arp", 0, arp_get_info);
-
+#ifdef CONFIG_PROC_FS
+ if (!proc_net_fops_create("arp", S_IRUGO, &arp_seq_fops))
+ panic("unable to create arp proc entry");
+#endif
#ifdef CONFIG_SYSCTL
neigh_sysctl_register(NULL, &arp_tbl.parms, NET_IPV4, NET_IPV4_NEIGH, "ipv4");
#endif
@@ -1364,39 +1386,3 @@ void __init arp_init (void)
}
-#ifdef CONFIG_PROC_FS
-#if defined(CONFIG_AX25) || defined(CONFIG_AX25_MODULE)
-
-/*
- * ax25 -> ASCII conversion
- */
-char *ax2asc2(ax25_address *a, char *buf)
-{
- char c, *s;
- int n;
-
- for (n = 0, s = buf; n < 6; n++) {
- c = (a->ax25_call[n] >> 1) & 0x7F;
-
- if (c != ' ') *s++ = c;
- }
-
- *s++ = '-';
-
- if ((n = ((a->ax25_call[6] >> 1) & 0x0F)) > 9) {
- *s++ = '1';
- n -= 10;
- }
-
- *s++ = n + '0';
- *s++ = '\0';
-
- if (*buf == '\0' || *buf == '-')
- return "*";
-
- return buf;
-
-}
-
-#endif
-#endif
===== net/ipv4/route.c 1.24 vs edited =====
--- 1.24/net/ipv4/route.c 2003-08-16 12:10:15 -07:00
+++ edited/net/ipv4/route.c 2004-10-03 15:12:55 -07:00
@@ -284,6 +284,7 @@ static int rt_cache_stat_get_info(char *
int i, lcpu;
int len = 0;
+ len += sprintf(buffer+len, "entries in_hit in_slow_tot in_slow_mc in_no_route in_brd in_martian_dst in_martian_src out_hit out_slow_tot out_slow_mc gc_total gc_ignored gc_goal_miss gc_dst_overflow in_hlist_search out_hlist_search\n");
for (lcpu = 0; lcpu < smp_num_cpus; lcpu++) {
i = cpu_logical_map(lcpu);
@@ -2625,7 +2626,8 @@ void __init ip_rt_init(void)
add_timer(&rt_secret_timer);
proc_net_create ("rt_cache", 0, rt_cache_get_info);
- proc_net_create ("rt_cache_stat", 0, rt_cache_stat_get_info);
+ create_proc_info_entry ("rt_cache", 0, proc_net_stat,
+ rt_cache_stat_get_info);
#ifdef CONFIG_NET_CLS_ROUTE
create_proc_read_entry("net/rt_acct", 0, 0, ip_rt_acct_read, NULL);
#endif
===== net/ipv6/ndisc.c 1.26 vs edited =====
--- 1.26/net/ipv6/ndisc.c 2004-07-15 13:33:03 -07:00
+++ edited/net/ipv6/ndisc.c 2004-10-03 15:12:55 -07:00
@@ -60,6 +60,7 @@
#include <linux/if_arp.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
+#include <linux/jhash.h>
#include <net/sock.h>
#include <net/snmp.h>
@@ -240,15 +241,14 @@ int ndisc_mc_map(struct in6_addr *addr,
static u32 ndisc_hash(const void *pkey, const struct net_device *dev)
{
- u32 hash_val;
+ const u32 *p32 = pkey;
+ u32 addr_hash, i;
- hash_val = *(u32*)(pkey + sizeof(struct in6_addr) - 4);
- hash_val ^= (hash_val>>16);
- hash_val ^= hash_val>>8;
- hash_val ^= hash_val>>3;
- hash_val = (hash_val^dev->ifindex)&NEIGH_HASHMASK;
+ addr_hash = 0;
+ for (i = 0; i < (sizeof(struct in6_addr) / sizeof(u32)); i++)
+ addr_hash ^= *p32++;
- return hash_val;
+ return jhash_2words(addr_hash, dev->ifindex, nd_tbl.hash_rnd);
}
static int ndisc_constructor(struct neighbour *neigh)
@@ -696,9 +696,9 @@ void ndisc_recv_ns(struct sk_buff *skb)
int inc = ipv6_addr_type(daddr)&IPV6_ADDR_MULTICAST;
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_ucast);
/*
* update / create cache entry
@@ -741,9 +741,9 @@ void ndisc_recv_ns(struct sk_buff *skb)
if (addr_type & IPV6_ADDR_UNICAST) {
int inc = ipv6_addr_type(daddr)&IPV6_ADDR_MULTICAST;
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_ucast);
/*
* update / create cache entry
@@ -775,9 +775,11 @@ void ndisc_recv_ns(struct sk_buff *skb)
inc == 0 ||
in6_dev->nd_parms->proxy_delay == 0) {
if (inc)
- nd_tbl.stats.rcv_probes_mcast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl,
+ rcv_probes_mcast);
else
- nd_tbl.stats.rcv_probes_ucast++;
+ NEIGH_CACHE_STAT_INC(&nd_tbl,
+ rcv_probes_ucast);
neigh = neigh_event_ns(&nd_tbl, lladdr, saddr, dev);
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2.4] backport neighbour cache redesign
2004-10-04 19:17 ` David S. Miller
@ 2004-10-04 20:15 ` Harald Welte
2004-10-04 20:20 ` David S. Miller
0 siblings, 1 reply; 6+ messages in thread
From: Harald Welte @ 2004-10-04 20:15 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
[-- Attachment #1: Type: text/plain, Size: 2024 bytes --]
On Mon, Oct 04, 2004 at 12:17:59PM -0700, David S. Miller wrote:
>
> Ok I started playing with this, here is what I have currently.
> Changes:
> 2) Missing symbol exports for the new interfaces modules can use,
> namely:
> neigh_lookup_nodev
> neigh_seq_start
> neigh_seq_next
> neigh_seq_stop
oh, sorry. I just linked atm,decnet and ipv6 static so far :(
> 3) Delete instead of "#if 0" out the now totally unused code
> inside of net/atm/clip.c
oops, that slipped through. embarrassing.
> We need to test and verify this patch a lot before I can toss
> such a big thing over to Marcelo, but I definitely intend to
> send this to him.
sure, I am well aware of that. If it helps, at least the IPv4 part will
see some serious deployment after it has passed the Astaro QA :)
I'll keep you posted about that.
One additional note:
If we go back to the initial reason for all those modifications, it was
deployment in large networks. The gc_thresh3 default of 1k is very
small, even if you have only 4 interfaces with /24, you will already max
out this default.
Now that the hash distribution is better, and the table resized
dynamically, we scale better with large neighbour caches.
I was thinking of registering with some notifiers and tuning gc_thresh3
automatically to be at least as large as the theoretical number of
immediate neigbhours. At least for multiple /24 and /16 I think this is
still reasonable.
As soon as we go further up (/8 or ipv6) the limit basically becomes
'unlimited', since we won't be able to receive 2^24 arp-causing packets
per second anyway.
Or be even less conservative and say: there is no limit for neighbour
cache entries? You mentioned that *BSD didn't have a limit, IIRC.
--
- Harald Welte <laforge@gnumonks.org> http://www.gnumonks.org/
============================================================================
Programming is like sex: One mistake and you have to support it your lifetime
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2.4] backport neighbour cache redesign
2004-10-04 20:15 ` Harald Welte
@ 2004-10-04 20:20 ` David S. Miller
2004-10-04 21:11 ` Harald Welte
0 siblings, 1 reply; 6+ messages in thread
From: David S. Miller @ 2004-10-04 20:20 UTC (permalink / raw)
To: Harald Welte; +Cc: netdev
On Mon, 4 Oct 2004 22:15:25 +0200
Harald Welte <laforge@gnumonks.org> wrote:
> If we go back to the initial reason for all those modifications, it was
> deployment in large networks. The gc_thresh3 default of 1k is very
> small, even if you have only 4 interfaces with /24, you will already max
> out this default.
>
> Now that the hash distribution is better, and the table resized
> dynamically, we scale better with large neighbour caches.
>
> I was thinking of registering with some notifiers and tuning gc_thresh3
> automatically to be at least as large as the theoretical number of
> immediate neigbhours. At least for multiple /24 and /16 I think this is
> still reasonable.
>
> As soon as we go further up (/8 or ipv6) the limit basically becomes
> 'unlimited', since we won't be able to receive 2^24 arp-causing packets
> per second anyway.
>
> Or be even less conservative and say: there is no limit for neighbour
> cache entries? You mentioned that *BSD didn't have a limit, IIRC.
I think tieing it to the subnet size is wrong, because the real
cap is the routing cache size. Remember the email where I was
talking about that?
I'm nearly ambivalent about a bound-less neighbour cache. I hate
to think about having tons of crap sitting in the neighbour cache
unused and sucking up memory. BSD's scheme works because when routing
entries die so do the neighbour entries they point to, so they have no
need for garbage collection like we do.
So right now I think a better idea is to have the routing cache tell
the neigh table how big it could get, or something like that.
It is a thorny issue, since usually routing cache --> neighbour cache
assosciations are VERY many to one except in these weird odd cases.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2.4] backport neighbour cache redesign
2004-10-04 20:20 ` David S. Miller
@ 2004-10-04 21:11 ` Harald Welte
2004-10-04 21:16 ` David S. Miller
0 siblings, 1 reply; 6+ messages in thread
From: Harald Welte @ 2004-10-04 21:11 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
[-- Attachment #1: Type: text/plain, Size: 1550 bytes --]
On Mon, Oct 04, 2004 at 01:20:34PM -0700, David S. Miller wrote:
> I think tieing it to the subnet size is wrong, because the real
> cap is the routing cache size. Remember the email where I was
> talking about that?
yes, I remember that email. I didn't think further about it, since I
got the feeling that there is significant movement to get rid of the
neighbour cache at some point (if/when there is some fast lookup
algorithm implemented) - and if we look at ipv6, there is no routing
cache.
> I'm nearly ambivalent about a bound-less neighbour cache. I hate
> to think about having tons of crap sitting in the neighbour cache
> unused and sucking up memory. BSD's scheme works because when routing
> entries die so do the neighbour entries they point to, so they have no
> need for garbage collection like we do.
mh, I see. I understood your email related to that subject that you
actually think this is not that stupid a thing to do...
anyway, I'll drop my prefix-based sizing idea. And I have to admit that
I also dislike the 'routing cache tell neighbour cache' approach (since
it's only applicable to ipv4)... so as for me, we'll just stick with
that hard limit. No more fiddling with that code, I'll get back to flow
accounting and pkttables for some time ;)
--
- Harald Welte <laforge@gnumonks.org> http://www.gnumonks.org/
============================================================================
Programming is like sex: One mistake and you have to support it your lifetime
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 2.4] backport neighbour cache redesign
2004-10-04 21:11 ` Harald Welte
@ 2004-10-04 21:16 ` David S. Miller
0 siblings, 0 replies; 6+ messages in thread
From: David S. Miller @ 2004-10-04 21:16 UTC (permalink / raw)
To: Harald Welte; +Cc: netdev
On Mon, 4 Oct 2004 23:11:58 +0200
Harald Welte <laforge@gnumonks.org> wrote:
> yes, I remember that email. I didn't think further about it, since I
> got the feeling that there is significant movement to get rid of the
> neighbour cache at some point (if/when there is some fast lookup
> algorithm implemented) - and if we look at ipv6, there is no routing
> cache.
Another idea is to dynamically grow the thresholds based upon usage
just as we do for the hash tables. For example, if we find say %90
of neighbour entries in-use at forced GC time, we increment the
thresholds by some factor.
Note that the "in-use" part is very important, it prevents a DoS
spam from growing the neighbour cache since such traffic causes
neighbour entries to quickly become unused.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-10-04 21:16 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-09-30 15:47 [PATCH 2.4] backport neighbour cache redesign Harald Welte
2004-10-04 19:17 ` David S. Miller
2004-10-04 20:15 ` Harald Welte
2004-10-04 20:20 ` David S. Miller
2004-10-04 21:11 ` Harald Welte
2004-10-04 21:16 ` David S. Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).