2
0
mirror of https://github.com/edk2-porting/linux-next.git synced 2024-12-15 00:34:10 +08:00
linux-next/net/ipv4/route.c

3464 lines
84 KiB
C
Raw Normal View History

/*
* INET An implementation of the TCP/IP protocol suite for the LINUX
* operating system. INET is implemented using the BSD Socket
* interface as the means of communication with the user level.
*
* ROUTE - implementation of the IP router.
*
* Authors: Ross Biro
* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
* Alan Cox, <gw4pts@gw4pts.ampr.org>
* Linus Torvalds, <Linus.Torvalds@helsinki.fi>
* Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
*
* Fixes:
* Alan Cox : Verify area fixes.
* Alan Cox : cli() protects routing changes
* Rui Oliveira : ICMP routing table updates
* (rco@di.uminho.pt) Routing table insertion and update
* Linus Torvalds : Rewrote bits to be sensible
* Alan Cox : Added BSD route gw semantics
* Alan Cox : Super /proc >4K
* Alan Cox : MTU in route table
* Alan Cox : MSS actually. Also added the window
* clamper.
* Sam Lantinga : Fixed route matching in rt_del()
* Alan Cox : Routing cache support.
* Alan Cox : Removed compatibility cruft.
* Alan Cox : RTF_REJECT support.
* Alan Cox : TCP irtt support.
* Jonathan Naylor : Added Metric support.
* Miquel van Smoorenburg : BSD API fixes.
* Miquel van Smoorenburg : Metrics.
* Alan Cox : Use __u32 properly
* Alan Cox : Aligned routing errors more closely with BSD
* our system is still very different.
* Alan Cox : Faster /proc handling
* Alexey Kuznetsov : Massive rework to support tree based routing,
* routing caches and better behaviour.
*
* Olaf Erb : irtt wasn't being copied right.
* Bjorn Ekwall : Kerneld route support.
* Alan Cox : Multicast fixed (I hope)
* Pavel Krauz : Limited broadcast fixed
* Mike McLagan : Routing by source
* Alexey Kuznetsov : End of old history. Split to fib.c and
* route.c and rewritten from scratch.
* Andi Kleen : Load-limit warning messages.
* Vitaly E. Lavrov : Transparent proxy revived after year coma.
* Vitaly E. Lavrov : Race condition in ip_route_input_slow.
* Tobias Ringstrom : Uninitialized res.type in ip_route_output_slow.
* Vladimir V. Ivanov : IP rule info (flowid) is really useful.
* Marc Boucher : routing by fwmark
* Robert Olsson : Added rt_cache statistics
* Arnaldo C. Melo : Convert proc stuff to seq_file
* Eric Dumazet : hashed spinlocks and rt_check_expire() fixes.
* Ilia Sotnikov : Ignore TOS on PMTUD and Redirect
* Ilia Sotnikov : Removed TOS from hash calculations
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#define pr_fmt(fmt) "IPv4: " fmt
#include <linux/module.h>
#include <asm/uaccess.h>
#include <linux/bitops.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/bootmem.h>
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/errno.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/proc_fs.h>
#include <linux/init.h>
#include <linux/workqueue.h>
#include <linux/skbuff.h>
#include <linux/inetdevice.h>
#include <linux/igmp.h>
#include <linux/pkt_sched.h>
#include <linux/mroute.h>
#include <linux/netfilter_ipv4.h>
#include <linux/random.h>
#include <linux/jhash.h>
#include <linux/rcupdate.h>
#include <linux/times.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
#include <linux/prefetch.h>
#include <net/dst.h>
#include <net/net_namespace.h>
#include <net/protocol.h>
#include <net/ip.h>
#include <net/route.h>
#include <net/inetpeer.h>
#include <net/sock.h>
#include <net/ip_fib.h>
#include <net/arp.h>
#include <net/tcp.h>
#include <net/icmp.h>
#include <net/xfrm.h>
#include <net/netevent.h>
#include <net/rtnetlink.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#include <linux/kmemleak.h>
#endif
#include <net/secure_seq.h>
#define RT_FL_TOS(oldflp4) \
((oldflp4)->flowi4_tos & (IPTOS_RT_MASK | RTO_ONLINK))
#define IP_MAX_MTU 0xFFF0
#define RT_GC_TIMEOUT (300*HZ)
static int ip_rt_max_size;
static int ip_rt_gc_timeout __read_mostly = RT_GC_TIMEOUT;
static int ip_rt_gc_interval __read_mostly = 60 * HZ;
static int ip_rt_gc_min_interval __read_mostly = HZ / 2;
static int ip_rt_redirect_number __read_mostly = 9;
static int ip_rt_redirect_load __read_mostly = HZ / 50;
static int ip_rt_redirect_silence __read_mostly = ((HZ / 50) << (9 + 1));
static int ip_rt_error_cost __read_mostly = HZ;
static int ip_rt_error_burst __read_mostly = 5 * HZ;
static int ip_rt_gc_elasticity __read_mostly = 8;
static int ip_rt_mtu_expires __read_mostly = 10 * 60 * HZ;
static int ip_rt_min_pmtu __read_mostly = 512 + 20 + 20;
static int ip_rt_min_advmss __read_mostly = 256;
static int rt_chain_length_max __read_mostly = 20;
static struct delayed_work expires_work;
static unsigned long expires_ljiffies;
/*
* Interface to generic destination cache.
*/
static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie);
static unsigned int ipv4_default_advmss(const struct dst_entry *dst);
static unsigned int ipv4_mtu(const struct dst_entry *dst);
static void ipv4_dst_destroy(struct dst_entry *dst);
static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst);
static void ipv4_link_failure(struct sk_buff *skb);
static void ip_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
struct sk_buff *skb, u32 mtu);
static void ip_do_redirect(struct dst_entry *dst, struct sock *sk,
struct sk_buff *skb);
static int rt_garbage_collect(struct dst_ops *ops);
static void ipv4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
int how)
{
}
net: Implement read-only protection and COW'ing of metrics. Routing metrics are now copy-on-write. Initially a route entry points it's metrics at a read-only location. If a routing table entry exists, it will point there. Else it will point at the all zero metric place-holder called 'dst_default_metrics'. The writeability state of the metrics is stored in the low bits of the metrics pointer, we have two bits left to spare if we want to store more states. For the initial implementation, COW is implemented simply via kmalloc. However future enhancements will change this to place the writable metrics somewhere else, in order to increase sharing. Very likely this "somewhere else" will be the inetpeer cache. Note also that this means that metrics updates may transiently fail if we cannot COW the metrics successfully. But even by itself, this patch should decrease memory usage and increase cache locality especially for routing workloads. In those cases the read-only metric copies stay in place and never get written to. TCP workloads where metrics get updated, and those rare cases where PMTU triggers occur, will take a very slight performance hit. But that hit will be alleviated when the long-term writable metrics move to a more sharable location. Since the metrics storage went from a u32 array of RTAX_MAX entries to what is essentially a pointer, some retooling of the dst_entry layout was necessary. Most importantly, we need to preserve the alignment of the reference count so that it doesn't share cache lines with the read-mostly state, as per Eric Dumazet's alignment assertion checks. The only non-trivial bit here is the move of the 'flags' member into the writeable cacheline. This is OK since we are always accessing the flags around the same moment when we made a modification to the reference count. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-01-27 12:51:05 +08:00
static u32 *ipv4_cow_metrics(struct dst_entry *dst, unsigned long old)
{
WARN_ON(1);
return NULL;
net: Implement read-only protection and COW'ing of metrics. Routing metrics are now copy-on-write. Initially a route entry points it's metrics at a read-only location. If a routing table entry exists, it will point there. Else it will point at the all zero metric place-holder called 'dst_default_metrics'. The writeability state of the metrics is stored in the low bits of the metrics pointer, we have two bits left to spare if we want to store more states. For the initial implementation, COW is implemented simply via kmalloc. However future enhancements will change this to place the writable metrics somewhere else, in order to increase sharing. Very likely this "somewhere else" will be the inetpeer cache. Note also that this means that metrics updates may transiently fail if we cannot COW the metrics successfully. But even by itself, this patch should decrease memory usage and increase cache locality especially for routing workloads. In those cases the read-only metric copies stay in place and never get written to. TCP workloads where metrics get updated, and those rare cases where PMTU triggers occur, will take a very slight performance hit. But that hit will be alleviated when the long-term writable metrics move to a more sharable location. Since the metrics storage went from a u32 array of RTAX_MAX entries to what is essentially a pointer, some retooling of the dst_entry layout was necessary. Most importantly, we need to preserve the alignment of the reference count so that it doesn't share cache lines with the read-mostly state, as per Eric Dumazet's alignment assertion checks. The only non-trivial bit here is the move of the 'flags' member into the writeable cacheline. This is OK since we are always accessing the flags around the same moment when we made a modification to the reference count. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-01-27 12:51:05 +08:00
}
static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst,
struct sk_buff *skb,
const void *daddr);
static struct dst_ops ipv4_dst_ops = {
.family = AF_INET,
.protocol = cpu_to_be16(ETH_P_IP),
.gc = rt_garbage_collect,
.check = ipv4_dst_check,
.default_advmss = ipv4_default_advmss,
.mtu = ipv4_mtu,
net: Implement read-only protection and COW'ing of metrics. Routing metrics are now copy-on-write. Initially a route entry points it's metrics at a read-only location. If a routing table entry exists, it will point there. Else it will point at the all zero metric place-holder called 'dst_default_metrics'. The writeability state of the metrics is stored in the low bits of the metrics pointer, we have two bits left to spare if we want to store more states. For the initial implementation, COW is implemented simply via kmalloc. However future enhancements will change this to place the writable metrics somewhere else, in order to increase sharing. Very likely this "somewhere else" will be the inetpeer cache. Note also that this means that metrics updates may transiently fail if we cannot COW the metrics successfully. But even by itself, this patch should decrease memory usage and increase cache locality especially for routing workloads. In those cases the read-only metric copies stay in place and never get written to. TCP workloads where metrics get updated, and those rare cases where PMTU triggers occur, will take a very slight performance hit. But that hit will be alleviated when the long-term writable metrics move to a more sharable location. Since the metrics storage went from a u32 array of RTAX_MAX entries to what is essentially a pointer, some retooling of the dst_entry layout was necessary. Most importantly, we need to preserve the alignment of the reference count so that it doesn't share cache lines with the read-mostly state, as per Eric Dumazet's alignment assertion checks. The only non-trivial bit here is the move of the 'flags' member into the writeable cacheline. This is OK since we are always accessing the flags around the same moment when we made a modification to the reference count. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-01-27 12:51:05 +08:00
.cow_metrics = ipv4_cow_metrics,
.destroy = ipv4_dst_destroy,
.ifdown = ipv4_dst_ifdown,
.negative_advice = ipv4_negative_advice,
.link_failure = ipv4_link_failure,
.update_pmtu = ip_rt_update_pmtu,
.redirect = ip_do_redirect,
.local_out = __ip_local_out,
.neigh_lookup = ipv4_neigh_lookup,
};
#define ECN_OR_COST(class) TC_PRIO_##class
const __u8 ip_tos2prio[16] = {
TC_PRIO_BESTEFFORT,
ECN_OR_COST(BESTEFFORT),
TC_PRIO_BESTEFFORT,
ECN_OR_COST(BESTEFFORT),
TC_PRIO_BULK,
ECN_OR_COST(BULK),
TC_PRIO_BULK,
ECN_OR_COST(BULK),
TC_PRIO_INTERACTIVE,
ECN_OR_COST(INTERACTIVE),
TC_PRIO_INTERACTIVE,
ECN_OR_COST(INTERACTIVE),
TC_PRIO_INTERACTIVE_BULK,
ECN_OR_COST(INTERACTIVE_BULK),
TC_PRIO_INTERACTIVE_BULK,
ECN_OR_COST(INTERACTIVE_BULK)
};
EXPORT_SYMBOL(ip_tos2prio);
/*
* Route cache.
*/
/* The locking scheme is rather straight forward:
*
* 1) Read-Copy Update protects the buckets of the central route hash.
* 2) Only writers remove entries, and they hold the lock
* as they look at rtable reference counts.
* 3) Only readers acquire references to rtable entries,
* they do so with atomic increments and with the
* lock held.
*/
struct rt_hash_bucket {
struct rtable __rcu *chain;
};
#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK) || \
defined(CONFIG_PROVE_LOCKING)
/*
* Instead of using one spinlock for each rt_hash_bucket, we use a table of spinlocks
* The size of this table is a power of two and depends on the number of CPUS.
* (on lockdep we have a quite big spinlock_t, so keep the size down there)
*/
#ifdef CONFIG_LOCKDEP
# define RT_HASH_LOCK_SZ 256
#else
# if NR_CPUS >= 32
# define RT_HASH_LOCK_SZ 4096
# elif NR_CPUS >= 16
# define RT_HASH_LOCK_SZ 2048
# elif NR_CPUS >= 8
# define RT_HASH_LOCK_SZ 1024
# elif NR_CPUS >= 4
# define RT_HASH_LOCK_SZ 512
# else
# define RT_HASH_LOCK_SZ 256
# endif
#endif
static spinlock_t *rt_hash_locks;
# define rt_hash_lock_addr(slot) &rt_hash_locks[(slot) & (RT_HASH_LOCK_SZ - 1)]
static __init void rt_hash_lock_init(void)
{
int i;
rt_hash_locks = kmalloc(sizeof(spinlock_t) * RT_HASH_LOCK_SZ,
GFP_KERNEL);
if (!rt_hash_locks)
panic("IP: failed to allocate rt_hash_locks\n");
for (i = 0; i < RT_HASH_LOCK_SZ; i++)
spin_lock_init(&rt_hash_locks[i]);
}
#else
# define rt_hash_lock_addr(slot) NULL
static inline void rt_hash_lock_init(void)
{
}
#endif
static struct rt_hash_bucket *rt_hash_table __read_mostly;
static unsigned int rt_hash_mask __read_mostly;
static unsigned int rt_hash_log __read_mostly;
static DEFINE_PER_CPU(struct rt_cache_stat, rt_cache_stat);
#define RT_CACHE_STAT_INC(field) __this_cpu_inc(rt_cache_stat.field)
static inline unsigned int rt_hash(__be32 daddr, __be32 saddr, int idx,
int genid)
{
return jhash_3words((__force u32)daddr, (__force u32)saddr,
idx, genid)
& rt_hash_mask;
}
static inline int rt_genid(struct net *net)
{
return atomic_read(&net->ipv4.rt_genid);
}
#ifdef CONFIG_PROC_FS
struct rt_cache_iter_state {
struct seq_net_private p;
int bucket;
int genid;
};
static struct rtable *rt_cache_get_first(struct seq_file *seq)
{
struct rt_cache_iter_state *st = seq->private;
struct rtable *r = NULL;
for (st->bucket = rt_hash_mask; st->bucket >= 0; --st->bucket) {
if (!rcu_access_pointer(rt_hash_table[st->bucket].chain))
continue;
rcu_read_lock_bh();
r = rcu_dereference_bh(rt_hash_table[st->bucket].chain);
while (r) {
if (dev_net(r->dst.dev) == seq_file_net(seq) &&
r->rt_genid == st->genid)
return r;
r = rcu_dereference_bh(r->dst.rt_next);
}
rcu_read_unlock_bh();
}
return r;
}
static struct rtable *__rt_cache_get_next(struct seq_file *seq,
struct rtable *r)
{
struct rt_cache_iter_state *st = seq->private;
r = rcu_dereference_bh(r->dst.rt_next);
while (!r) {
rcu_read_unlock_bh();
do {
if (--st->bucket < 0)
return NULL;
} while (!rcu_access_pointer(rt_hash_table[st->bucket].chain));
rcu_read_lock_bh();
r = rcu_dereference_bh(rt_hash_table[st->bucket].chain);
}
return r;
}
static struct rtable *rt_cache_get_next(struct seq_file *seq,
struct rtable *r)
{
struct rt_cache_iter_state *st = seq->private;
while ((r = __rt_cache_get_next(seq, r)) != NULL) {
if (dev_net(r->dst.dev) != seq_file_net(seq))
continue;
if (r->rt_genid == st->genid)
break;
}
return r;
}
static struct rtable *rt_cache_get_idx(struct seq_file *seq, loff_t pos)
{
struct rtable *r = rt_cache_get_first(seq);
if (r)
while (pos && (r = rt_cache_get_next(seq, r)))
--pos;
return pos ? NULL : r;
}
static void *rt_cache_seq_start(struct seq_file *seq, loff_t *pos)
{
struct rt_cache_iter_state *st = seq->private;
if (*pos)
return rt_cache_get_idx(seq, *pos - 1);
st->genid = rt_genid(seq_file_net(seq));
return SEQ_START_TOKEN;
}
static void *rt_cache_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct rtable *r;
if (v == SEQ_START_TOKEN)
r = rt_cache_get_first(seq);
else
r = rt_cache_get_next(seq, v);
++*pos;
return r;
}
static void rt_cache_seq_stop(struct seq_file *seq, void *v)
{
if (v && v != SEQ_START_TOKEN)
rcu_read_unlock_bh();
}
static int rt_cache_seq_show(struct seq_file *seq, void *v)
{
if (v == SEQ_START_TOKEN)
seq_printf(seq, "%-127s\n",
"Iface\tDestination\tGateway \tFlags\t\tRefCnt\tUse\t"
"Metric\tSource\t\tMTU\tWindow\tIRTT\tTOS\tHHRef\t"
"HHUptod\tSpecDst");
else {
struct rtable *r = v;
int len;
seq_printf(seq, "%s\t%08X\t%08X\t%8X\t%d\t%u\t%d\t"
"%08X\t%d\t%u\t%u\t%02X\t%d\t%1d\t%08X%n",
r->dst.dev ? r->dst.dev->name : "*",
(__force u32)r->rt_dst,
(__force u32)r->rt_gateway,
r->rt_flags, atomic_read(&r->dst.__refcnt),
r->dst.__use, 0, (__force u32)r->rt_src,
dst_metric_advmss(&r->dst) + 40,
dst_metric(&r->dst, RTAX_WINDOW), 0,
r->rt_key_tos,
-1, 0, 0, &len);
seq_printf(seq, "%*s\n", 127 - len, "");
}
return 0;
}
static const struct seq_operations rt_cache_seq_ops = {
.start = rt_cache_seq_start,
.next = rt_cache_seq_next,
.stop = rt_cache_seq_stop,
.show = rt_cache_seq_show,
};
static int rt_cache_seq_open(struct inode *inode, struct file *file)
{
return seq_open_net(inode, file, &rt_cache_seq_ops,
sizeof(struct rt_cache_iter_state));
}
static const struct file_operations rt_cache_seq_fops = {
.owner = THIS_MODULE,
.open = rt_cache_seq_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release_net,
};
static void *rt_cpu_seq_start(struct seq_file *seq, loff_t *pos)
{
int cpu;
if (*pos == 0)
return SEQ_START_TOKEN;
for (cpu = *pos-1; cpu < nr_cpu_ids; ++cpu) {
if (!cpu_possible(cpu))
continue;
*pos = cpu+1;
return &per_cpu(rt_cache_stat, cpu);
}
return NULL;
}
static void *rt_cpu_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
int cpu;
for (cpu = *pos; cpu < nr_cpu_ids; ++cpu) {
if (!cpu_possible(cpu))
continue;
*pos = cpu+1;
return &per_cpu(rt_cache_stat, cpu);
}
return NULL;
}
static void rt_cpu_seq_stop(struct seq_file *seq, void *v)
{
}
static int rt_cpu_seq_show(struct seq_file *seq, void *v)
{
struct rt_cache_stat *st = v;
if (v == SEQ_START_TOKEN) {
seq_printf(seq, "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");
return 0;
}
seq_printf(seq,"%08x %08x %08x %08x %08x %08x %08x %08x "
" %08x %08x %08x %08x %08x %08x %08x %08x %08x \n",
dst_entries_get_slow(&ipv4_dst_ops),
st->in_hit,
st->in_slow_tot,
st->in_slow_mc,
st->in_no_route,
st->in_brd,
st->in_martian_dst,
st->in_martian_src,
st->out_hit,
st->out_slow_tot,
st->out_slow_mc,
st->gc_total,
st->gc_ignored,
st->gc_goal_miss,
st->gc_dst_overflow,
st->in_hlist_search,
st->out_hlist_search
);
return 0;
}
static const struct seq_operations rt_cpu_seq_ops = {
.start = rt_cpu_seq_start,
.next = rt_cpu_seq_next,
.stop = rt_cpu_seq_stop,
.show = rt_cpu_seq_show,
};
static int rt_cpu_seq_open(struct inode *inode, struct file *file)
{
return seq_open(file, &rt_cpu_seq_ops);
}
static const struct file_operations rt_cpu_seq_fops = {
.owner = THIS_MODULE,
.open = rt_cpu_seq_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release,
};
#ifdef CONFIG_IP_ROUTE_CLASSID
static int rt_acct_proc_show(struct seq_file *m, void *v)
{
struct ip_rt_acct *dst, *src;
unsigned int i, j;
dst = kcalloc(256, sizeof(struct ip_rt_acct), GFP_KERNEL);
if (!dst)
return -ENOMEM;
for_each_possible_cpu(i) {
src = (struct ip_rt_acct *)per_cpu_ptr(ip_rt_acct, i);
for (j = 0; j < 256; j++) {
dst[j].o_bytes += src[j].o_bytes;
dst[j].o_packets += src[j].o_packets;
dst[j].i_bytes += src[j].i_bytes;
dst[j].i_packets += src[j].i_packets;
}
}
seq_write(m, dst, 256 * sizeof(struct ip_rt_acct));
kfree(dst);
return 0;
}
static int rt_acct_proc_open(struct inode *inode, struct file *file)
{
return single_open(file, rt_acct_proc_show, NULL);
}
static const struct file_operations rt_acct_proc_fops = {
.owner = THIS_MODULE,
.open = rt_acct_proc_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
#endif
static int __net_init ip_rt_do_proc_init(struct net *net)
{
struct proc_dir_entry *pde;
pde = proc_net_fops_create(net, "rt_cache", S_IRUGO,
&rt_cache_seq_fops);
if (!pde)
goto err1;
pde = proc_create("rt_cache", S_IRUGO,
net->proc_net_stat, &rt_cpu_seq_fops);
if (!pde)
goto err2;
#ifdef CONFIG_IP_ROUTE_CLASSID
pde = proc_create("rt_acct", 0, net->proc_net, &rt_acct_proc_fops);
if (!pde)
goto err3;
#endif
return 0;
#ifdef CONFIG_IP_ROUTE_CLASSID
err3:
remove_proc_entry("rt_cache", net->proc_net_stat);
#endif
err2:
remove_proc_entry("rt_cache", net->proc_net);
err1:
return -ENOMEM;
}
static void __net_exit ip_rt_do_proc_exit(struct net *net)
{
remove_proc_entry("rt_cache", net->proc_net_stat);
remove_proc_entry("rt_cache", net->proc_net);
#ifdef CONFIG_IP_ROUTE_CLASSID
remove_proc_entry("rt_acct", net->proc_net);
#endif
}
static struct pernet_operations ip_rt_proc_ops __net_initdata = {
.init = ip_rt_do_proc_init,
.exit = ip_rt_do_proc_exit,
};
static int __init ip_rt_proc_init(void)
{
return register_pernet_subsys(&ip_rt_proc_ops);
}
#else
static inline int ip_rt_proc_init(void)
{
return 0;
}
#endif /* CONFIG_PROC_FS */
static inline void rt_free(struct rtable *rt)
{
call_rcu_bh(&rt->dst.rcu_head, dst_rcu_free);
}
static inline void rt_drop(struct rtable *rt)
{
ip_rt_put(rt);
call_rcu_bh(&rt->dst.rcu_head, dst_rcu_free);
}
static inline int rt_fast_clean(struct rtable *rth)
{
/* Kill broadcast/multicast entries very aggresively, if they
collide in hash table with more useful entries */
return (rth->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST)) &&
rt_is_input_route(rth) && rth->dst.rt_next;
}
static inline int rt_valuable(struct rtable *rth)
{
return (rth->rt_flags & (RTCF_REDIRECTED | RTCF_NOTIFY)) ||
rth->dst.expires;
}
static int rt_may_expire(struct rtable *rth, unsigned long tmo1, unsigned long tmo2)
{
unsigned long age;
int ret = 0;
if (atomic_read(&rth->dst.__refcnt))
goto out;
age = jiffies - rth->dst.lastuse;
if ((age <= tmo1 && !rt_fast_clean(rth)) ||
(age <= tmo2 && rt_valuable(rth)))
goto out;
ret = 1;
out: return ret;
}
/* Bits of score are:
* 31: very valuable
* 30: not quite useless
* 29..0: usage counter
*/
static inline u32 rt_score(struct rtable *rt)
{
u32 score = jiffies - rt->dst.lastuse;
score = ~score & ~(3<<30);
if (rt_valuable(rt))
score |= (1<<31);
if (rt_is_output_route(rt) ||
!(rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST|RTCF_LOCAL)))
score |= (1<<30);
return score;
}
static inline bool rt_caching(const struct net *net)
{
return net->ipv4.current_rt_cache_rebuild_count <=
net->ipv4.sysctl_rt_cache_rebuild_count;
}
static inline bool compare_hash_inputs(const struct rtable *rt1,
const struct rtable *rt2)
{
return ((((__force u32)rt1->rt_key_dst ^ (__force u32)rt2->rt_key_dst) |
((__force u32)rt1->rt_key_src ^ (__force u32)rt2->rt_key_src) |
(rt1->rt_route_iif ^ rt2->rt_route_iif)) == 0);
}
static inline int compare_keys(struct rtable *rt1, struct rtable *rt2)
{
return (((__force u32)rt1->rt_key_dst ^ (__force u32)rt2->rt_key_dst) |
((__force u32)rt1->rt_key_src ^ (__force u32)rt2->rt_key_src) |
(rt1->rt_mark ^ rt2->rt_mark) |
(rt1->rt_key_tos ^ rt2->rt_key_tos) |
(rt1->rt_route_iif ^ rt2->rt_route_iif) |
(rt1->rt_oif ^ rt2->rt_oif)) == 0;
}
static inline int compare_netns(struct rtable *rt1, struct rtable *rt2)
{
return net_eq(dev_net(rt1->dst.dev), dev_net(rt2->dst.dev));
}
static inline int rt_is_expired(struct rtable *rth)
{
return rth->rt_genid != rt_genid(dev_net(rth->dst.dev));
}
/*
* Perform a full scan of hash table and free all entries.
* Can be called by a softirq or a process.
* In the later case, we want to be reschedule if necessary
*/
static void rt_do_flush(struct net *net, int process_context)
{
unsigned int i;
struct rtable *rth, *next;
for (i = 0; i <= rt_hash_mask; i++) {
struct rtable __rcu **pprev;
struct rtable *list;
if (process_context && need_resched())
cond_resched();
rth = rcu_access_pointer(rt_hash_table[i].chain);
if (!rth)
continue;
spin_lock_bh(rt_hash_lock_addr(i));
list = NULL;
pprev = &rt_hash_table[i].chain;
rth = rcu_dereference_protected(*pprev,
lockdep_is_held(rt_hash_lock_addr(i)));
while (rth) {
next = rcu_dereference_protected(rth->dst.rt_next,
lockdep_is_held(rt_hash_lock_addr(i)));
if (!net ||
net_eq(dev_net(rth->dst.dev), net)) {
rcu_assign_pointer(*pprev, next);
rcu_assign_pointer(rth->dst.rt_next, list);
list = rth;
} else {
pprev = &rth->dst.rt_next;
}
rth = next;
}
spin_unlock_bh(rt_hash_lock_addr(i));
for (; list; list = next) {
next = rcu_dereference_protected(list->dst.rt_next, 1);
rt_free(list);
}
}
}
/*
* While freeing expired entries, we compute average chain length
* and standard deviation, using fixed-point arithmetic.
* This to have an estimation of rt_chain_length_max
* rt_chain_length_max = max(elasticity, AVG + 4*SD)
* We use 3 bits for frational part, and 29 (or 61) for magnitude.
*/
#define FRACT_BITS 3
#define ONE (1UL << FRACT_BITS)
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
/*
* Given a hash chain and an item in this hash chain,
* find if a previous entry has the same hash_inputs
* (but differs on tos, mark or oif)
* Returns 0 if an alias is found.
* Returns ONE if rth has no alias before itself.
*/
static int has_noalias(const struct rtable *head, const struct rtable *rth)
{
const struct rtable *aux = head;
while (aux != rth) {
if (compare_hash_inputs(aux, rth))
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
return 0;
aux = rcu_dereference_protected(aux->dst.rt_next, 1);
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
}
return ONE;
}
static void rt_check_expire(void)
{
static unsigned int rover;
unsigned int i = rover, goal;
struct rtable *rth;
struct rtable __rcu **rthp;
unsigned long samples = 0;
unsigned long sum = 0, sum2 = 0;
unsigned long delta;
u64 mult;
delta = jiffies - expires_ljiffies;
expires_ljiffies = jiffies;
mult = ((u64)delta) << rt_hash_log;
if (ip_rt_gc_timeout > 1)
do_div(mult, ip_rt_gc_timeout);
goal = (unsigned int)mult;
if (goal > rt_hash_mask)
goal = rt_hash_mask + 1;
for (; goal > 0; goal--) {
unsigned long tmo = ip_rt_gc_timeout;
unsigned long length;
i = (i + 1) & rt_hash_mask;
rthp = &rt_hash_table[i].chain;
if (need_resched())
cond_resched();
samples++;
if (rcu_dereference_raw(*rthp) == NULL)
continue;
length = 0;
spin_lock_bh(rt_hash_lock_addr(i));
while ((rth = rcu_dereference_protected(*rthp,
lockdep_is_held(rt_hash_lock_addr(i)))) != NULL) {
prefetch(rth->dst.rt_next);
if (rt_is_expired(rth) ||
rt_may_expire(rth, tmo, ip_rt_gc_timeout)) {
*rthp = rth->dst.rt_next;
rt_free(rth);
continue;
}
/* We only count entries on a chain with equal
* hash inputs once so that entries for
* different QOS levels, and other non-hash
* input attributes don't unfairly skew the
* length computation
*/
tmo >>= 1;
rthp = &rth->dst.rt_next;
length += has_noalias(rt_hash_table[i].chain, rth);
}
spin_unlock_bh(rt_hash_lock_addr(i));
sum += length;
sum2 += length*length;
}
if (samples) {
unsigned long avg = sum / samples;
unsigned long sd = int_sqrt(sum2 / samples - avg*avg);
rt_chain_length_max = max_t(unsigned long,
ip_rt_gc_elasticity,
(avg + 4*sd) >> FRACT_BITS);
}
rover = i;
}
/*
* rt_worker_func() is run in process context.
* we call rt_check_expire() to scan part of the hash table
*/
static void rt_worker_func(struct work_struct *work)
{
rt_check_expire();
schedule_delayed_work(&expires_work, ip_rt_gc_interval);
}
/*
* Perturbation of rt_genid by a small quantity [1..256]
* Using 8 bits of shuffling ensure we can call rt_cache_invalidate()
* many times (2^24) without giving recent rt_genid.
* Jenkins hash is strong enough that litle changes of rt_genid are OK.
*/
static void rt_cache_invalidate(struct net *net)
{
unsigned char shuffle;
get_random_bytes(&shuffle, sizeof(shuffle));
atomic_add(shuffle + 1U, &net->ipv4.rt_genid);
}
/*
* delay < 0 : invalidate cache (fast : entries will be deleted later)
* delay >= 0 : invalidate & flush cache (can be long)
*/
void rt_cache_flush(struct net *net, int delay)
{
rt_cache_invalidate(net);
if (delay >= 0)
rt_do_flush(net, !in_softirq());
}
/* Flush previous cache invalidated entries from the cache */
void rt_cache_flush_batch(struct net *net)
{
rt_do_flush(net, !in_softirq());
}
static void rt_emergency_hash_rebuild(struct net *net)
{
net_warn_ratelimited("Route hash chain too long!\n");
rt_cache_invalidate(net);
}
/*
Short description of GC goals.
We want to build algorithm, which will keep routing cache
at some equilibrium point, when number of aged off entries
is kept approximately equal to newly generated ones.
Current expiration strength is variable "expire".
We try to adjust it dynamically, so that if networking
is idle expires is large enough to keep enough of warm entries,
and when load increases it reduces to limit cache size.
*/
static int rt_garbage_collect(struct dst_ops *ops)
{
static unsigned long expire = RT_GC_TIMEOUT;
static unsigned long last_gc;
static int rover;
static int equilibrium;
struct rtable *rth;
struct rtable __rcu **rthp;
unsigned long now = jiffies;
int goal;
int entries = dst_entries_get_fast(&ipv4_dst_ops);
/*
* Garbage collection is pretty expensive,
* do not make it too frequently.
*/
RT_CACHE_STAT_INC(gc_total);
if (now - last_gc < ip_rt_gc_min_interval &&
entries < ip_rt_max_size) {
RT_CACHE_STAT_INC(gc_ignored);
goto out;
}
entries = dst_entries_get_slow(&ipv4_dst_ops);
/* Calculate number of entries, which we want to expire now. */
goal = entries - (ip_rt_gc_elasticity << rt_hash_log);
if (goal <= 0) {
if (equilibrium < ipv4_dst_ops.gc_thresh)
equilibrium = ipv4_dst_ops.gc_thresh;
goal = entries - equilibrium;
if (goal > 0) {
equilibrium += min_t(unsigned int, goal >> 1, rt_hash_mask + 1);
goal = entries - equilibrium;
}
} else {
/* We are in dangerous area. Try to reduce cache really
* aggressively.
*/
goal = max_t(unsigned int, goal >> 1, rt_hash_mask + 1);
equilibrium = entries - goal;
}
if (now - last_gc >= ip_rt_gc_min_interval)
last_gc = now;
if (goal <= 0) {
equilibrium += goal;
goto work_done;
}
do {
int i, k;
for (i = rt_hash_mask, k = rover; i >= 0; i--) {
unsigned long tmo = expire;
k = (k + 1) & rt_hash_mask;
rthp = &rt_hash_table[k].chain;
spin_lock_bh(rt_hash_lock_addr(k));
while ((rth = rcu_dereference_protected(*rthp,
lockdep_is_held(rt_hash_lock_addr(k)))) != NULL) {
if (!rt_is_expired(rth) &&
!rt_may_expire(rth, tmo, expire)) {
tmo >>= 1;
rthp = &rth->dst.rt_next;
continue;
}
*rthp = rth->dst.rt_next;
rt_free(rth);
goal--;
}
spin_unlock_bh(rt_hash_lock_addr(k));
if (goal <= 0)
break;
}
rover = k;
if (goal <= 0)
goto work_done;
/* Goal is not achieved. We stop process if:
- if expire reduced to zero. Otherwise, expire is halfed.
- if table is not full.
- if we are called from interrupt.
- jiffies check is just fallback/debug loop breaker.
We will not spin here for long time in any case.
*/
RT_CACHE_STAT_INC(gc_goal_miss);
if (expire == 0)
break;
expire >>= 1;
if (dst_entries_get_fast(&ipv4_dst_ops) < ip_rt_max_size)
goto out;
} while (!in_softirq() && time_before_eq(jiffies, now));
if (dst_entries_get_fast(&ipv4_dst_ops) < ip_rt_max_size)
goto out;
if (dst_entries_get_slow(&ipv4_dst_ops) < ip_rt_max_size)
goto out;
net_warn_ratelimited("dst cache overflow\n");
RT_CACHE_STAT_INC(gc_dst_overflow);
return 1;
work_done:
expire += ip_rt_gc_min_interval;
if (expire > ip_rt_gc_timeout ||
dst_entries_get_fast(&ipv4_dst_ops) < ipv4_dst_ops.gc_thresh ||
dst_entries_get_slow(&ipv4_dst_ops) < ipv4_dst_ops.gc_thresh)
expire = ip_rt_gc_timeout;
out: return 0;
}
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
/*
* Returns number of entries in a hash chain that have different hash_inputs
*/
static int slow_chain_length(const struct rtable *head)
{
int length = 0;
const struct rtable *rth = head;
while (rth) {
length += has_noalias(head, rth);
rth = rcu_dereference_protected(rth->dst.rt_next, 1);
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
}
return length >> FRACT_BITS;
}
static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst,
struct sk_buff *skb,
const void *daddr)
{
struct net_device *dev = dst->dev;
const __be32 *pkey = daddr;
const struct rtable *rt;
struct neighbour *n;
rt = (const struct rtable *) dst;
if (rt->rt_gateway)
pkey = (const __be32 *) &rt->rt_gateway;
else if (skb)
pkey = &ip_hdr(skb)->daddr;
n = __ipv4_neigh_lookup(dev, *(__force u32 *)pkey);
if (n)
return n;
return neigh_create(&arp_tbl, pkey, dev);
}
static struct rtable *rt_intern_hash(unsigned int hash, struct rtable *rt,
struct sk_buff *skb, int ifindex)
{
struct rtable *rth, *cand;
struct rtable __rcu **rthp, **candp;
unsigned long now;
u32 min_score;
int chain_length;
restart:
chain_length = 0;
min_score = ~(u32)0;
cand = NULL;
candp = NULL;
now = jiffies;
if (!rt_caching(dev_net(rt->dst.dev)) || (rt->dst.flags & DST_NOCACHE)) {
/*
* If we're not caching, just tell the caller we
* were successful and don't touch the route. The
* caller hold the sole reference to the cache entry, and
* it will be released when the caller is done with it.
* If we drop it here, the callers have no way to resolve routes
* when we're not caching. Instead, just point *rp at rt, so
* the caller gets a single use out of the route
ipv4 routing: Ensure that route cache entries are usable and reclaimable with caching is off When route caching is disabled (rt_caching returns false), We still use route cache entries that are created and passed into rt_intern_hash once. These routes need to be made usable for the one call path that holds a reference to them, and they need to be reclaimed when they're finished with their use. To be made usable, they need to be associated with a neighbor table entry (which they currently are not), otherwise iproute_finish2 just discards the packet, since we don't know which L2 peer to send the packet to. To do this binding, we need to follow the path a bit higher up in rt_intern_hash, which calls arp_bind_neighbour, but not assign the route entry to the hash table. Currently, if caching is off, we simply assign the route to the rp pointer and are reutrn success. This patch associates us with a neighbor entry first. Secondly, we need to make sure that any single use routes like this are known to the garbage collector when caching is off. If caching is off, and we try to hash in a route, it will leak when its refcount reaches zero. To avoid this, this patch calls rt_free on the route cache entry passed into rt_intern_hash. This places us on the gc list for the route cache garbage collector, so that when its refcount reaches zero, it will be reclaimed (Thanks to Alexey for this suggestion). I've tested this on a local system here, and with these patches in place, I'm able to maintain routed connectivity to remote systems, even if I set /proc/sys/net/ipv4/rt_cache_rebuild_count to -1, which forces rt_caching to return false. Signed-off-by: Neil Horman <nhorman@redhat.com> Reported-by: Jarek Poplawski <jarkao2@gmail.com> Reported-by: Maxime Bizon <mbizon@freebox.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-22 18:18:53 +08:00
* Note that we do rt_free on this new route entry, so that
* once its refcount hits zero, we are still able to reap it
* (Thanks Alexey)
* Note: To avoid expensive rcu stuff for this uncached dst,
* we set DST_NOCACHE so that dst_release() can free dst without
* waiting a grace period.
*/
ipv4 routing: Ensure that route cache entries are usable and reclaimable with caching is off When route caching is disabled (rt_caching returns false), We still use route cache entries that are created and passed into rt_intern_hash once. These routes need to be made usable for the one call path that holds a reference to them, and they need to be reclaimed when they're finished with their use. To be made usable, they need to be associated with a neighbor table entry (which they currently are not), otherwise iproute_finish2 just discards the packet, since we don't know which L2 peer to send the packet to. To do this binding, we need to follow the path a bit higher up in rt_intern_hash, which calls arp_bind_neighbour, but not assign the route entry to the hash table. Currently, if caching is off, we simply assign the route to the rp pointer and are reutrn success. This patch associates us with a neighbor entry first. Secondly, we need to make sure that any single use routes like this are known to the garbage collector when caching is off. If caching is off, and we try to hash in a route, it will leak when its refcount reaches zero. To avoid this, this patch calls rt_free on the route cache entry passed into rt_intern_hash. This places us on the gc list for the route cache garbage collector, so that when its refcount reaches zero, it will be reclaimed (Thanks to Alexey for this suggestion). I've tested this on a local system here, and with these patches in place, I'm able to maintain routed connectivity to remote systems, even if I set /proc/sys/net/ipv4/rt_cache_rebuild_count to -1, which forces rt_caching to return false. Signed-off-by: Neil Horman <nhorman@redhat.com> Reported-by: Jarek Poplawski <jarkao2@gmail.com> Reported-by: Maxime Bizon <mbizon@freebox.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-22 18:18:53 +08:00
rt->dst.flags |= DST_NOCACHE;
ipv4 routing: Ensure that route cache entries are usable and reclaimable with caching is off When route caching is disabled (rt_caching returns false), We still use route cache entries that are created and passed into rt_intern_hash once. These routes need to be made usable for the one call path that holds a reference to them, and they need to be reclaimed when they're finished with their use. To be made usable, they need to be associated with a neighbor table entry (which they currently are not), otherwise iproute_finish2 just discards the packet, since we don't know which L2 peer to send the packet to. To do this binding, we need to follow the path a bit higher up in rt_intern_hash, which calls arp_bind_neighbour, but not assign the route entry to the hash table. Currently, if caching is off, we simply assign the route to the rp pointer and are reutrn success. This patch associates us with a neighbor entry first. Secondly, we need to make sure that any single use routes like this are known to the garbage collector when caching is off. If caching is off, and we try to hash in a route, it will leak when its refcount reaches zero. To avoid this, this patch calls rt_free on the route cache entry passed into rt_intern_hash. This places us on the gc list for the route cache garbage collector, so that when its refcount reaches zero, it will be reclaimed (Thanks to Alexey for this suggestion). I've tested this on a local system here, and with these patches in place, I'm able to maintain routed connectivity to remote systems, even if I set /proc/sys/net/ipv4/rt_cache_rebuild_count to -1, which forces rt_caching to return false. Signed-off-by: Neil Horman <nhorman@redhat.com> Reported-by: Jarek Poplawski <jarkao2@gmail.com> Reported-by: Maxime Bizon <mbizon@freebox.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-22 18:18:53 +08:00
goto skip_hashing;
}
rthp = &rt_hash_table[hash].chain;
spin_lock_bh(rt_hash_lock_addr(hash));
while ((rth = rcu_dereference_protected(*rthp,
lockdep_is_held(rt_hash_lock_addr(hash)))) != NULL) {
if (rt_is_expired(rth)) {
*rthp = rth->dst.rt_next;
rt_free(rth);
continue;
}
if (compare_keys(rth, rt) && compare_netns(rth, rt)) {
/* Put it first */
*rthp = rth->dst.rt_next;
/*
* Since lookup is lockfree, the deletion
* must be visible to another weakly ordered CPU before
* the insertion at the start of the hash chain.
*/
rcu_assign_pointer(rth->dst.rt_next,
rt_hash_table[hash].chain);
/*
* Since lookup is lockfree, the update writes
* must be ordered for consistency on SMP.
*/
rcu_assign_pointer(rt_hash_table[hash].chain, rth);
dst_use(&rth->dst, now);
spin_unlock_bh(rt_hash_lock_addr(hash));
rt_drop(rt);
if (skb)
skb_dst_set(skb, &rth->dst);
return rth;
}
if (!atomic_read(&rth->dst.__refcnt)) {
u32 score = rt_score(rth);
if (score <= min_score) {
cand = rth;
candp = rthp;
min_score = score;
}
}
chain_length++;
rthp = &rth->dst.rt_next;
}
if (cand) {
/* ip_rt_gc_elasticity used to be average length of chain
* length, when exceeded gc becomes really aggressive.
*
* The second limit is less certain. At the moment it allows
* only 2 entries per bucket. We will see.
*/
if (chain_length > ip_rt_gc_elasticity) {
*candp = cand->dst.rt_next;
rt_free(cand);
}
} else {
net: fix route cache rebuilds We added an automatic route cache rebuilding in commit 1080d709fb9d8cd43 but had to correct few bugs. One of the assumption of original patch, was that entries where kept sorted in a given way. This assumption is known to be wrong (commit 1ddbcb005c395518 gave an explanation of this and corrected a leak) and expensive to respect. Paweł Staszewski reported to me one of his machine got its routing cache disabled after few messages like : [ 2677.850065] Route hash chain too long! [ 2677.850080] Adjust your secret_interval! [82839.662993] Route hash chain too long! [82839.662996] Adjust your secret_interval! [155843.731650] Route hash chain too long! [155843.731664] Adjust your secret_interval! [155843.811881] Route hash chain too long! [155843.811891] Adjust your secret_interval! [155843.858209] vlan0811: 5 rebuilds is over limit, route caching disabled [155843.858212] Route hash chain too long! [155843.858213] Adjust your secret_interval! This is because rt_intern_hash() might be fooled when computing a chain length, because multiple entries with same keys can differ because of TOS (or mark/oif) bits. In the rare case the fast algorithm see a too long chain, and before taking expensive path, we call a helper function in order to not count duplicates of same routes, that only differ with tos/mark/oif bits. This helper works with data already in cpu cache and is not be very expensive, despite its O(N^2) implementation. Paweł Staszewski sucessfully tested this patch on his loaded router. Reported-and-tested-by: Paweł Staszewski <pstaszewski@itcare.pl> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-08 11:20:00 +08:00
if (chain_length > rt_chain_length_max &&
slow_chain_length(rt_hash_table[hash].chain) > rt_chain_length_max) {
struct net *net = dev_net(rt->dst.dev);
int num = ++net->ipv4.current_rt_cache_rebuild_count;
if (!rt_caching(net)) {
pr_warn("%s: %d rebuilds is over limit, route caching disabled\n",
rt->dst.dev->name, num);
}
rt_emergency_hash_rebuild(net);
spin_unlock_bh(rt_hash_lock_addr(hash));
hash = rt_hash(rt->rt_key_dst, rt->rt_key_src,
ifindex, rt_genid(net));
goto restart;
}
}
rt->dst.rt_next = rt_hash_table[hash].chain;
/*
* Since lookup is lockfree, we must make sure
* previous writes to rt are committed to memory
* before making rt visible to other CPUS.
*/
net: fix rtable leak in net/ipv4/route.c Alexander V. Lukyanov found a regression in 2.6.29 and made a complete analysis found in http://bugzilla.kernel.org/show_bug.cgi?id=13339 Quoted here because its a perfect one : begin_of_quotation 2.6.29 patch has introduced flexible route cache rebuilding. Unfortunately the patch has at least one critical flaw, and another problem. rt_intern_hash calculates rthi pointer, which is later used for new entry insertion. The same loop calculates cand pointer which is used to clean the list. If the pointers are the same, rtable leak occurs, as first the cand is removed then the new entry is appended to it. This leak leads to unregister_netdevice problem (usage count > 0). Another problem of the patch is that it tries to insert the entries in certain order, to facilitate counting of entries distinct by all but QoS parameters. Unfortunately, referencing an existing rtable entry moves it to list beginning, to speed up further lookups, so the carefully built order is destroyed. For the first problem the simplest patch it to set rthi=0 when rthi==cand, but it will also destroy the ordering. end_of_quotation Problematic commit is 1080d709fb9d8cd4392f93476ee46a9d6ea05a5b (net: implement emergency route cache rebulds when gc_elasticity is exceeded) Trying to keep dst_entries ordered is too complex and breaks the fact that order should depend on the frequency of use for garbage collection. A possible fix is to make rt_intern_hash() simpler, and only makes rt_check_expire() a litle bit smarter, being able to cope with an arbitrary entries order. The added loop is running on cache hot data, while cpu is prefetching next object, so should be unnoticied. Reported-and-analyzed-by: Alexander V. Lukyanov <lav@yar.ru> Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-05-20 04:14:28 +08:00
rcu_assign_pointer(rt_hash_table[hash].chain, rt);
spin_unlock_bh(rt_hash_lock_addr(hash));
ipv4 routing: Ensure that route cache entries are usable and reclaimable with caching is off When route caching is disabled (rt_caching returns false), We still use route cache entries that are created and passed into rt_intern_hash once. These routes need to be made usable for the one call path that holds a reference to them, and they need to be reclaimed when they're finished with their use. To be made usable, they need to be associated with a neighbor table entry (which they currently are not), otherwise iproute_finish2 just discards the packet, since we don't know which L2 peer to send the packet to. To do this binding, we need to follow the path a bit higher up in rt_intern_hash, which calls arp_bind_neighbour, but not assign the route entry to the hash table. Currently, if caching is off, we simply assign the route to the rp pointer and are reutrn success. This patch associates us with a neighbor entry first. Secondly, we need to make sure that any single use routes like this are known to the garbage collector when caching is off. If caching is off, and we try to hash in a route, it will leak when its refcount reaches zero. To avoid this, this patch calls rt_free on the route cache entry passed into rt_intern_hash. This places us on the gc list for the route cache garbage collector, so that when its refcount reaches zero, it will be reclaimed (Thanks to Alexey for this suggestion). I've tested this on a local system here, and with these patches in place, I'm able to maintain routed connectivity to remote systems, even if I set /proc/sys/net/ipv4/rt_cache_rebuild_count to -1, which forces rt_caching to return false. Signed-off-by: Neil Horman <nhorman@redhat.com> Reported-by: Jarek Poplawski <jarkao2@gmail.com> Reported-by: Maxime Bizon <mbizon@freebox.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-22 18:18:53 +08:00
skip_hashing:
if (skb)
skb_dst_set(skb, &rt->dst);
return rt;
}
/*
* Peer allocation may fail only in serious out-of-memory conditions. However
* we still can generate some output.
* Random ID selection looks a bit dangerous because we have no chances to
* select ID being unique in a reasonable period of time.
* But broken packet identifier may be better than no packet at all.
*/
static void ip_select_fb_ident(struct iphdr *iph)
{
static DEFINE_SPINLOCK(ip_fb_id_lock);
static u32 ip_fallback_id;
u32 salt;
spin_lock_bh(&ip_fb_id_lock);
salt = secure_ip_id((__force __be32)ip_fallback_id ^ iph->daddr);
iph->id = htons(salt & 0xFFFF);
ip_fallback_id = salt;
spin_unlock_bh(&ip_fb_id_lock);
}
void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more)
{
struct net *net = dev_net(dst->dev);
struct inet_peer *peer;
peer = inet_getpeer_v4(net->ipv4.peers, iph->daddr, 1);
if (peer) {
iph->id = htons(inet_getid(peer, more));
inet_putpeer(peer);
return;
}
ip_select_fb_ident(iph);
}
EXPORT_SYMBOL(__ip_select_ident);
static void rt_del(unsigned int hash, struct rtable *rt)
{
struct rtable __rcu **rthp;
struct rtable *aux;
rthp = &rt_hash_table[hash].chain;
spin_lock_bh(rt_hash_lock_addr(hash));
ip_rt_put(rt);
while ((aux = rcu_dereference_protected(*rthp,
lockdep_is_held(rt_hash_lock_addr(hash)))) != NULL) {
if (aux == rt || rt_is_expired(aux)) {
*rthp = aux->dst.rt_next;
rt_free(aux);
continue;
}
rthp = &aux->dst.rt_next;
}
spin_unlock_bh(rt_hash_lock_addr(hash));
}
static void __build_flow_key(struct flowi4 *fl4, const struct sock *sk,
const struct iphdr *iph,
int oif, u8 tos,
u8 prot, u32 mark, int flow_flags)
{
if (sk) {
const struct inet_sock *inet = inet_sk(sk);
oif = sk->sk_bound_dev_if;
mark = sk->sk_mark;
tos = RT_CONN_FLAGS(sk);
prot = inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol;
}
flowi4_init_output(fl4, oif, mark, tos,
RT_SCOPE_UNIVERSE, prot,
flow_flags,
iph->daddr, iph->saddr, 0, 0);
}
static void build_skb_flow_key(struct flowi4 *fl4, const struct sk_buff *skb,
const struct sock *sk)
{
const struct iphdr *iph = ip_hdr(skb);
int oif = skb->dev->ifindex;
u8 tos = RT_TOS(iph->tos);
u8 prot = iph->protocol;
u32 mark = skb->mark;
__build_flow_key(fl4, sk, iph, oif, tos, prot, mark, 0);
}
static void build_sk_flow_key(struct flowi4 *fl4, const struct sock *sk)
{
const struct inet_sock *inet = inet_sk(sk);
const struct ip_options_rcu *inet_opt;
__be32 daddr = inet->inet_daddr;
rcu_read_lock();
inet_opt = rcu_dereference(inet->inet_opt);
if (inet_opt && inet_opt->opt.srr)
daddr = inet_opt->opt.faddr;
flowi4_init_output(fl4, sk->sk_bound_dev_if, sk->sk_mark,
RT_CONN_FLAGS(sk), RT_SCOPE_UNIVERSE,
inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
inet_sk_flowi_flags(sk),
daddr, inet->inet_saddr, 0, 0);
rcu_read_unlock();
}
static void ip_rt_build_flow_key(struct flowi4 *fl4, const struct sock *sk,
const struct sk_buff *skb)
{
if (skb)
build_skb_flow_key(fl4, skb, sk);
else
build_sk_flow_key(fl4, sk);
}
static DEFINE_SEQLOCK(fnhe_seqlock);
static struct fib_nh_exception *fnhe_oldest(struct fnhe_hash_bucket *hash)
{
struct fib_nh_exception *fnhe, *oldest;
oldest = rcu_dereference(hash->chain);
for (fnhe = rcu_dereference(oldest->fnhe_next); fnhe;
fnhe = rcu_dereference(fnhe->fnhe_next)) {
if (time_before(fnhe->fnhe_stamp, oldest->fnhe_stamp))
oldest = fnhe;
}
return oldest;
}
static inline u32 fnhe_hashfun(__be32 daddr)
{
u32 hval;
hval = (__force u32) daddr;
hval ^= (hval >> 11) ^ (hval >> 22);
return hval & (FNHE_HASH_SIZE - 1);
}
static void update_or_create_fnhe(struct fib_nh *nh, __be32 daddr, __be32 gw,
u32 pmtu, unsigned long expires)
{
struct fnhe_hash_bucket *hash;
struct fib_nh_exception *fnhe;
int depth;
u32 hval = fnhe_hashfun(daddr);
write_seqlock_bh(&fnhe_seqlock);
hash = nh->nh_exceptions;
if (!hash) {
hash = kzalloc(FNHE_HASH_SIZE * sizeof(*hash), GFP_ATOMIC);
if (!hash)
goto out_unlock;
nh->nh_exceptions = hash;
}
hash += hval;
depth = 0;
for (fnhe = rcu_dereference(hash->chain); fnhe;
fnhe = rcu_dereference(fnhe->fnhe_next)) {
if (fnhe->fnhe_daddr == daddr)
break;
depth++;
}
if (fnhe) {
if (gw)
fnhe->fnhe_gw = gw;
if (pmtu) {
fnhe->fnhe_pmtu = pmtu;
fnhe->fnhe_expires = expires;
}
} else {
if (depth > FNHE_RECLAIM_DEPTH)
fnhe = fnhe_oldest(hash);
else {
fnhe = kzalloc(sizeof(*fnhe), GFP_ATOMIC);
if (!fnhe)
goto out_unlock;
fnhe->fnhe_next = hash->chain;
rcu_assign_pointer(hash->chain, fnhe);
}
fnhe->fnhe_daddr = daddr;
fnhe->fnhe_gw = gw;
fnhe->fnhe_pmtu = pmtu;
fnhe->fnhe_expires = expires;
}
fnhe->fnhe_stamp = jiffies;
out_unlock:
write_sequnlock_bh(&fnhe_seqlock);
return;
}
static void __ip_do_redirect(struct rtable *rt, struct sk_buff *skb, struct flowi4 *fl4)
{
__be32 new_gw = icmp_hdr(skb)->un.gateway;
__be32 old_gw = ip_hdr(skb)->saddr;
struct net_device *dev = skb->dev;
struct in_device *in_dev;
struct fib_result res;
struct neighbour *n;
struct net *net;
switch (icmp_hdr(skb)->code & 7) {
case ICMP_REDIR_NET:
case ICMP_REDIR_NETTOS:
case ICMP_REDIR_HOST:
case ICMP_REDIR_HOSTTOS:
break;
default:
return;
}
if (rt->rt_gateway != old_gw)
return;
in_dev = __in_dev_get_rcu(dev);
if (!in_dev)
return;
net = dev_net(dev);
if (new_gw == old_gw || !IN_DEV_RX_REDIRECTS(in_dev) ||
ipv4_is_multicast(new_gw) || ipv4_is_lbcast(new_gw) ||
ipv4_is_zeronet(new_gw))
goto reject_redirect;
if (!IN_DEV_SHARED_MEDIA(in_dev)) {
if (!inet_addr_onlink(in_dev, new_gw, old_gw))
goto reject_redirect;
if (IN_DEV_SEC_REDIRECTS(in_dev) && ip_fib_check_default(new_gw, dev))
goto reject_redirect;
} else {
if (inet_addr_type(net, new_gw) != RTN_UNICAST)
goto reject_redirect;
}
n = ipv4_neigh_lookup(&rt->dst, NULL, &new_gw);
if (n) {
if (!(n->nud_state & NUD_VALID)) {
neigh_event_send(n, NULL);
} else {
if (fib_lookup(net, fl4, &res) == 0) {
struct fib_nh *nh = &FIB_RES_NH(res);
update_or_create_fnhe(nh, fl4->daddr, new_gw,
0, 0);
}
rt->rt_gateway = new_gw;
rt->rt_flags |= RTCF_REDIRECTED;
call_netevent_notifiers(NETEVENT_NEIGH_UPDATE, n);
}
neigh_release(n);
}
return;
reject_redirect:
#ifdef CONFIG_IP_ROUTE_VERBOSE
if (IN_DEV_LOG_MARTIANS(in_dev)) {
const struct iphdr *iph = (const struct iphdr *) skb->data;
__be32 daddr = iph->daddr;
__be32 saddr = iph->saddr;
net_info_ratelimited("Redirect from %pI4 on %s about %pI4 ignored\n"
" Advised path = %pI4 -> %pI4\n",
&old_gw, dev->name, &new_gw,
&saddr, &daddr);
}
#endif
;
}
static void ip_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
{
struct rtable *rt;
struct flowi4 fl4;
rt = (struct rtable *) dst;
ip_rt_build_flow_key(&fl4, sk, skb);
__ip_do_redirect(rt, skb, &fl4);
}
static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst)
{
struct rtable *rt = (struct rtable *)dst;
struct dst_entry *ret = dst;
if (rt) {
if (dst->obsolete > 0) {
ip_rt_put(rt);
ret = NULL;
} else if ((rt->rt_flags & RTCF_REDIRECTED) ||
rt->dst.expires) {
unsigned int hash = rt_hash(rt->rt_key_dst, rt->rt_key_src,
rt->rt_oif,
rt_genid(dev_net(dst->dev)));
rt_del(hash, rt);
ret = NULL;
}
}
return ret;
}
/*
* Algorithm:
* 1. The first ip_rt_redirect_number redirects are sent
* with exponential backoff, then we stop sending them at all,
* assuming that the host ignores our redirects.
* 2. If we did not see packets requiring redirects
* during ip_rt_redirect_silence, we assume that the host
* forgot redirected route and start to send redirects again.
*
* This algorithm is much cheaper and more intelligent than dumb load limiting
* in icmp.c.
*
* NOTE. Do not forget to inhibit load limiting for redirects (redundant)
* and "frag. need" (breaks PMTU discovery) in icmp.c.
*/
void ip_rt_send_redirect(struct sk_buff *skb)
{
struct rtable *rt = skb_rtable(skb);
struct in_device *in_dev;
struct inet_peer *peer;
struct net *net;
int log_martians;
rcu_read_lock();
in_dev = __in_dev_get_rcu(rt->dst.dev);
if (!in_dev || !IN_DEV_TX_REDIRECTS(in_dev)) {
rcu_read_unlock();
return;
}
log_martians = IN_DEV_LOG_MARTIANS(in_dev);
rcu_read_unlock();
net = dev_net(rt->dst.dev);
peer = inet_getpeer_v4(net->ipv4.peers, ip_hdr(skb)->saddr, 1);
if (!peer) {
icmp_send(skb, ICMP_REDIRECT, ICMP_REDIR_HOST, rt->rt_gateway);
return;
}
/* No redirected packets during ip_rt_redirect_silence;
* reset the algorithm.
*/
if (time_after(jiffies, peer->rate_last + ip_rt_redirect_silence))
peer->rate_tokens = 0;
/* Too many ignored redirects; do not send anything
* set dst.rate_last to the last seen redirected packet.
*/
if (peer->rate_tokens >= ip_rt_redirect_number) {
peer->rate_last = jiffies;
goto out_put_peer;
}
/* Check for load limit; set rate_last to the latest sent
* redirect.
*/
if (peer->rate_tokens == 0 ||
time_after(jiffies,
(peer->rate_last +
(ip_rt_redirect_load << peer->rate_tokens)))) {
icmp_send(skb, ICMP_REDIRECT, ICMP_REDIR_HOST, rt->rt_gateway);
peer->rate_last = jiffies;
++peer->rate_tokens;
#ifdef CONFIG_IP_ROUTE_VERBOSE
if (log_martians &&
peer->rate_tokens == ip_rt_redirect_number)
net_warn_ratelimited("host %pI4/if%d ignores redirects for %pI4 to %pI4\n",
&ip_hdr(skb)->saddr, rt->rt_iif,
&rt->rt_dst, &rt->rt_gateway);
#endif
}
out_put_peer:
inet_putpeer(peer);
}
static int ip_error(struct sk_buff *skb)
{
struct in_device *in_dev = __in_dev_get_rcu(skb->dev);
struct rtable *rt = skb_rtable(skb);
struct inet_peer *peer;
unsigned long now;
struct net *net;
bool send;
int code;
net = dev_net(rt->dst.dev);
if (!IN_DEV_FORWARD(in_dev)) {
switch (rt->dst.error) {
case EHOSTUNREACH:
IP_INC_STATS_BH(net, IPSTATS_MIB_INADDRERRORS);
break;
case ENETUNREACH:
IP_INC_STATS_BH(net, IPSTATS_MIB_INNOROUTES);
break;
}
goto out;
}
switch (rt->dst.error) {
case EINVAL:
default:
goto out;
case EHOSTUNREACH:
code = ICMP_HOST_UNREACH;
break;
case ENETUNREACH:
code = ICMP_NET_UNREACH;
IP_INC_STATS_BH(net, IPSTATS_MIB_INNOROUTES);
break;
case EACCES:
code = ICMP_PKT_FILTERED;
break;
}
peer = inet_getpeer_v4(net->ipv4.peers, ip_hdr(skb)->saddr, 1);
send = true;
if (peer) {
now = jiffies;
peer->rate_tokens += now - peer->rate_last;
if (peer->rate_tokens > ip_rt_error_burst)
peer->rate_tokens = ip_rt_error_burst;
peer->rate_last = now;
if (peer->rate_tokens >= ip_rt_error_cost)
peer->rate_tokens -= ip_rt_error_cost;
else
send = false;
inet_putpeer(peer);
}
if (send)
icmp_send(skb, ICMP_DEST_UNREACH, code, 0);
out: kfree_skb(skb);
return 0;
}
static void __ip_rt_update_pmtu(struct rtable *rt, struct flowi4 *fl4, u32 mtu)
{
struct fib_result res;
if (mtu < ip_rt_min_pmtu)
mtu = ip_rt_min_pmtu;
if (fib_lookup(dev_net(rt->dst.dev), fl4, &res) == 0) {
struct fib_nh *nh = &FIB_RES_NH(res);
update_or_create_fnhe(nh, fl4->daddr, 0, mtu,
jiffies + ip_rt_mtu_expires);
}
rt->rt_pmtu = mtu;
dst_set_expires(&rt->dst, ip_rt_mtu_expires);
}
static void ip_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
struct sk_buff *skb, u32 mtu)
{
struct rtable *rt = (struct rtable *) dst;
struct flowi4 fl4;
ip_rt_build_flow_key(&fl4, sk, skb);
__ip_rt_update_pmtu(rt, &fl4, mtu);
}
void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu,
int oif, u32 mark, u8 protocol, int flow_flags)
{
const struct iphdr *iph = (const struct iphdr *) skb->data;
struct flowi4 fl4;
struct rtable *rt;
__build_flow_key(&fl4, NULL, iph, oif,
RT_TOS(iph->tos), protocol, mark, flow_flags);
rt = __ip_route_output_key(net, &fl4);
if (!IS_ERR(rt)) {
__ip_rt_update_pmtu(rt, &fl4, mtu);
ip_rt_put(rt);
}
}
EXPORT_SYMBOL_GPL(ipv4_update_pmtu);
void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu)
{
const struct iphdr *iph = (const struct iphdr *) skb->data;
struct flowi4 fl4;
struct rtable *rt;
__build_flow_key(&fl4, sk, iph, 0, 0, 0, 0, 0);
rt = __ip_route_output_key(sock_net(sk), &fl4);
if (!IS_ERR(rt)) {
__ip_rt_update_pmtu(rt, &fl4, mtu);
ip_rt_put(rt);
}
}
EXPORT_SYMBOL_GPL(ipv4_sk_update_pmtu);
void ipv4_redirect(struct sk_buff *skb, struct net *net,
int oif, u32 mark, u8 protocol, int flow_flags)
{
const struct iphdr *iph = (const struct iphdr *) skb->data;
struct flowi4 fl4;
struct rtable *rt;
__build_flow_key(&fl4, NULL, iph, oif,
RT_TOS(iph->tos), protocol, mark, flow_flags);
rt = __ip_route_output_key(net, &fl4);
if (!IS_ERR(rt)) {
__ip_do_redirect(rt, skb, &fl4);
ip_rt_put(rt);
}
}
EXPORT_SYMBOL_GPL(ipv4_redirect);
void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk)
{
const struct iphdr *iph = (const struct iphdr *) skb->data;
struct flowi4 fl4;
struct rtable *rt;
__build_flow_key(&fl4, sk, iph, 0, 0, 0, 0, 0);
rt = __ip_route_output_key(sock_net(sk), &fl4);
if (!IS_ERR(rt)) {
__ip_do_redirect(rt, skb, &fl4);
ip_rt_put(rt);
}
}
EXPORT_SYMBOL_GPL(ipv4_sk_redirect);
static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie)
{
struct rtable *rt = (struct rtable *) dst;
if (rt_is_expired(rt))
return NULL;
return dst;
}
static void ipv4_dst_destroy(struct dst_entry *dst)
{
struct rtable *rt = (struct rtable *) dst;
net: Implement read-only protection and COW'ing of metrics. Routing metrics are now copy-on-write. Initially a route entry points it's metrics at a read-only location. If a routing table entry exists, it will point there. Else it will point at the all zero metric place-holder called 'dst_default_metrics'. The writeability state of the metrics is stored in the low bits of the metrics pointer, we have two bits left to spare if we want to store more states. For the initial implementation, COW is implemented simply via kmalloc. However future enhancements will change this to place the writable metrics somewhere else, in order to increase sharing. Very likely this "somewhere else" will be the inetpeer cache. Note also that this means that metrics updates may transiently fail if we cannot COW the metrics successfully. But even by itself, this patch should decrease memory usage and increase cache locality especially for routing workloads. In those cases the read-only metric copies stay in place and never get written to. TCP workloads where metrics get updated, and those rare cases where PMTU triggers occur, will take a very slight performance hit. But that hit will be alleviated when the long-term writable metrics move to a more sharable location. Since the metrics storage went from a u32 array of RTAX_MAX entries to what is essentially a pointer, some retooling of the dst_entry layout was necessary. Most importantly, we need to preserve the alignment of the reference count so that it doesn't share cache lines with the read-mostly state, as per Eric Dumazet's alignment assertion checks. The only non-trivial bit here is the move of the 'flags' member into the writeable cacheline. This is OK since we are always accessing the flags around the same moment when we made a modification to the reference count. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-01-27 12:51:05 +08:00
if (rt->fi) {
fib_info_put(rt->fi);
rt->fi = NULL;
}
}
static void ipv4_link_failure(struct sk_buff *skb)
{
struct rtable *rt;
icmp_send(skb, ICMP_DEST_UNREACH, ICMP_HOST_UNREACH, 0);
rt = skb_rtable(skb);
if (rt)
dst_set_expires(&rt->dst, 0);
}
static int ip_rt_bug(struct sk_buff *skb)
{
pr_debug("%s: %pI4 -> %pI4, %s\n",
__func__, &ip_hdr(skb)->saddr, &ip_hdr(skb)->daddr,
skb->dev ? skb->dev->name : "?");
kfree_skb(skb);
WARN_ON(1);
return 0;
}
/*
We do not cache source address of outgoing interface,
because it is used only by IP RR, TS and SRR options,
so that it out of fast path.
BTW remember: "addr" is allowed to be not aligned
in IP options!
*/
void ip_rt_get_source(u8 *addr, struct sk_buff *skb, struct rtable *rt)
{
__be32 src;
if (rt_is_output_route(rt))
src = ip_hdr(skb)->saddr;
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
else {
struct fib_result res;
struct flowi4 fl4;
struct iphdr *iph;
iph = ip_hdr(skb);
memset(&fl4, 0, sizeof(fl4));
fl4.daddr = iph->daddr;
fl4.saddr = iph->saddr;
fl4.flowi4_tos = RT_TOS(iph->tos);
fl4.flowi4_oif = rt->dst.dev->ifindex;
fl4.flowi4_iif = skb->dev->ifindex;
fl4.flowi4_mark = skb->mark;
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
rcu_read_lock();
if (fib_lookup(dev_net(rt->dst.dev), &fl4, &res) == 0)
src = FIB_RES_PREFSRC(dev_net(rt->dst.dev), res);
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
else
src = inet_select_addr(rt->dst.dev, rt->rt_gateway,
RT_SCOPE_UNIVERSE);
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
rcu_read_unlock();
}
memcpy(addr, &src, 4);
}
#ifdef CONFIG_IP_ROUTE_CLASSID
static void set_class_tag(struct rtable *rt, u32 tag)
{
if (!(rt->dst.tclassid & 0xFFFF))
rt->dst.tclassid |= tag & 0xFFFF;
if (!(rt->dst.tclassid & 0xFFFF0000))
rt->dst.tclassid |= tag & 0xFFFF0000;
}
#endif
static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
{
unsigned int advmss = dst_metric_raw(dst, RTAX_ADVMSS);
if (advmss == 0) {
advmss = max_t(unsigned int, dst->dev->mtu - 40,
ip_rt_min_advmss);
if (advmss > 65535 - 40)
advmss = 65535 - 40;
}
return advmss;
}
static unsigned int ipv4_mtu(const struct dst_entry *dst)
{
const struct rtable *rt = (const struct rtable *) dst;
unsigned int mtu = rt->rt_pmtu;
if (mtu && time_after_eq(jiffies, rt->dst.expires))
mtu = 0;
if (!mtu)
mtu = dst_metric_raw(dst, RTAX_MTU);
if (mtu && rt_is_output_route(rt))
return mtu;
mtu = dst->dev->mtu;
if (unlikely(dst_metric_locked(dst, RTAX_MTU))) {
if (rt->rt_gateway != rt->rt_dst && mtu > 576)
mtu = 576;
}
if (mtu > IP_MAX_MTU)
mtu = IP_MAX_MTU;
return mtu;
}
static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4,
struct fib_info *fi)
{
if (fi->fib_metrics != (u32 *) dst_default_metrics) {
rt->fi = fi;
atomic_inc(&fi->fib_clntref);
}
dst_init_metrics(&rt->dst, fi->fib_metrics, true);
}
static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr)
{
struct fnhe_hash_bucket *hash = nh->nh_exceptions;
struct fib_nh_exception *fnhe;
u32 hval;
hval = fnhe_hashfun(daddr);
restart:
for (fnhe = rcu_dereference(hash[hval].chain); fnhe;
fnhe = rcu_dereference(fnhe->fnhe_next)) {
__be32 fnhe_daddr, gw;
unsigned long expires;
unsigned int seq;
u32 pmtu;
seq = read_seqbegin(&fnhe_seqlock);
fnhe_daddr = fnhe->fnhe_daddr;
gw = fnhe->fnhe_gw;
pmtu = fnhe->fnhe_pmtu;
expires = fnhe->fnhe_expires;
if (read_seqretry(&fnhe_seqlock, seq))
goto restart;
if (daddr != fnhe_daddr)
continue;
if (pmtu) {
unsigned long diff = jiffies - expires;
if (time_before(jiffies, expires)) {
rt->rt_pmtu = pmtu;
dst_set_expires(&rt->dst, diff);
}
}
if (gw)
rt->rt_gateway = gw;
fnhe->fnhe_stamp = jiffies;
break;
}
}
static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4,
const struct fib_result *res,
struct fib_info *fi, u16 type, u32 itag)
{
if (fi) {
struct fib_nh *nh = &FIB_RES_NH(*res);
if (nh->nh_gw && nh->nh_scope == RT_SCOPE_LINK)
rt->rt_gateway = nh->nh_gw;
if (unlikely(nh->nh_exceptions))
rt_bind_exception(rt, nh, fl4->daddr);
rt_init_metrics(rt, fl4, fi);
#ifdef CONFIG_IP_ROUTE_CLASSID
rt->dst.tclassid = FIB_RES_NH(*res).nh_tclassid;
#endif
}
#ifdef CONFIG_IP_ROUTE_CLASSID
#ifdef CONFIG_IP_MULTIPLE_TABLES
set_class_tag(rt, res->tclassid);
#endif
set_class_tag(rt, itag);
#endif
}
static struct rtable *rt_dst_alloc(struct net_device *dev,
bool nopolicy, bool noxfrm)
{
return dst_alloc(&ipv4_dst_ops, dev, 1, -1,
DST_HOST |
(nopolicy ? DST_NOPOLICY : 0) |
(noxfrm ? DST_NOXFRM : 0));
}
/* called in rcu_read_lock() section */
static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
u8 tos, struct net_device *dev, int our)
{
unsigned int hash;
struct rtable *rth;
struct in_device *in_dev = __in_dev_get_rcu(dev);
u32 itag = 0;
int err;
/* Primary sanity checks. */
if (in_dev == NULL)
return -EINVAL;
if (ipv4_is_multicast(saddr) || ipv4_is_lbcast(saddr) ||
skb->protocol != htons(ETH_P_IP))
goto e_inval;
if (likely(!IN_DEV_ROUTE_LOCALNET(in_dev)))
if (ipv4_is_loopback(saddr))
goto e_inval;
if (ipv4_is_zeronet(saddr)) {
if (!ipv4_is_local_multicast(daddr))
goto e_inval;
} else {
err = fib_validate_source(skb, saddr, 0, tos, 0, dev,
in_dev, &itag);
if (err < 0)
goto e_err;
}
rth = rt_dst_alloc(dev_net(dev)->loopback_dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
if (!rth)
goto e_nobufs;
#ifdef CONFIG_IP_ROUTE_CLASSID
rth->dst.tclassid = itag;
#endif
rth->dst.output = ip_rt_bug;
rth->rt_key_dst = daddr;
rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(dev_net(dev));
rth->rt_flags = RTCF_MULTICAST;
rth->rt_type = RTN_MULTICAST;
rth->rt_key_tos = tos;
rth->rt_dst = daddr;
rth->rt_src = saddr;
rth->rt_route_iif = dev->ifindex;
rth->rt_iif = dev->ifindex;
rth->rt_oif = 0;
rth->rt_mark = skb->mark;
rth->rt_pmtu = 0;
rth->rt_gateway = daddr;
rth->fi = NULL;
if (our) {
rth->dst.input= ip_local_deliver;
rth->rt_flags |= RTCF_LOCAL;
}
#ifdef CONFIG_IP_MROUTE
if (!ipv4_is_local_multicast(daddr) && IN_DEV_MFORWARD(in_dev))
rth->dst.input = ip_mr_input;
#endif
RT_CACHE_STAT_INC(in_slow_mc);
hash = rt_hash(daddr, saddr, dev->ifindex, rt_genid(dev_net(dev)));
rth = rt_intern_hash(hash, rth, skb, dev->ifindex);
return IS_ERR(rth) ? PTR_ERR(rth) : 0;
e_nobufs:
return -ENOBUFS;
e_inval:
return -EINVAL;
e_err:
return err;
}
static void ip_handle_martian_source(struct net_device *dev,
struct in_device *in_dev,
struct sk_buff *skb,
__be32 daddr,
__be32 saddr)
{
RT_CACHE_STAT_INC(in_martian_src);
#ifdef CONFIG_IP_ROUTE_VERBOSE
if (IN_DEV_LOG_MARTIANS(in_dev) && net_ratelimit()) {
/*
* RFC1812 recommendation, if source is martian,
* the only hint is MAC header.
*/
pr_warn("martian source %pI4 from %pI4, on dev %s\n",
&daddr, &saddr, dev->name);
if (dev->hard_header_len && skb_mac_header_was_set(skb)) {
print_hex_dump(KERN_WARNING, "ll header: ",
DUMP_PREFIX_OFFSET, 16, 1,
skb_mac_header(skb),
dev->hard_header_len, true);
}
}
#endif
}
/* called in rcu_read_lock() section */
static int __mkroute_input(struct sk_buff *skb,
const struct fib_result *res,
struct in_device *in_dev,
__be32 daddr, __be32 saddr, u32 tos,
struct rtable **result)
{
struct rtable *rth;
int err;
struct in_device *out_dev;
unsigned int flags = 0;
u32 itag;
/* get a working reference to the output device */
out_dev = __in_dev_get_rcu(FIB_RES_DEV(*res));
if (out_dev == NULL) {
net_crit_ratelimited("Bug in ip_route_input_slow(). Please report.\n");
return -EINVAL;
}
err = fib_validate_source(skb, saddr, daddr, tos, FIB_RES_OIF(*res),
in_dev->dev, in_dev, &itag);
if (err < 0) {
ip_handle_martian_source(in_dev->dev, in_dev, skb, daddr,
saddr);
goto cleanup;
}
if (err)
flags |= RTCF_DIRECTSRC;
if (out_dev == in_dev && err &&
(IN_DEV_SHARED_MEDIA(out_dev) ||
inet_addr_onlink(out_dev, saddr, FIB_RES_GW(*res))))
flags |= RTCF_DOREDIRECT;
if (skb->protocol != htons(ETH_P_IP)) {
/* Not IP (i.e. ARP). Do not create route, if it is
* invalid for proxy arp. DNAT routes are always valid.
*
* Proxy arp feature have been extended to allow, ARP
* replies back to the same interface, to support
* Private VLAN switch technologies. See arp.c.
*/
if (out_dev == in_dev &&
IN_DEV_PROXY_ARP_PVLAN(in_dev) == 0) {
err = -EINVAL;
goto cleanup;
}
}
rth = rt_dst_alloc(out_dev->dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(out_dev, NOXFRM));
if (!rth) {
err = -ENOBUFS;
goto cleanup;
}
rth->rt_key_dst = daddr;
rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(dev_net(rth->dst.dev));
rth->rt_flags = flags;
rth->rt_type = res->type;
rth->rt_key_tos = tos;
rth->rt_dst = daddr;
rth->rt_src = saddr;
rth->rt_route_iif = in_dev->dev->ifindex;
rth->rt_iif = in_dev->dev->ifindex;
rth->rt_oif = 0;
rth->rt_mark = skb->mark;
rth->rt_pmtu = 0;
rth->rt_gateway = daddr;
rth->fi = NULL;
rth->dst.input = ip_forward;
rth->dst.output = ip_output;
rt_set_nexthop(rth, NULL, res, res->fi, res->type, itag);
*result = rth;
err = 0;
cleanup:
return err;
}
static int ip_mkroute_input(struct sk_buff *skb,
struct fib_result *res,
const struct flowi4 *fl4,
struct in_device *in_dev,
__be32 daddr, __be32 saddr, u32 tos)
{
struct rtable *rth = NULL;
int err;
unsigned int hash;
#ifdef CONFIG_IP_ROUTE_MULTIPATH
if (res->fi && res->fi->fib_nhs > 1)
fib_select_multipath(res);
#endif
/* create a routing cache entry */
err = __mkroute_input(skb, res, in_dev, daddr, saddr, tos, &rth);
if (err)
return err;
/* put it into the cache */
hash = rt_hash(daddr, saddr, fl4->flowi4_iif,
rt_genid(dev_net(rth->dst.dev)));
rth = rt_intern_hash(hash, rth, skb, fl4->flowi4_iif);
if (IS_ERR(rth))
return PTR_ERR(rth);
return 0;
}
/*
* NOTE. We drop all the packets that has local source
* addresses, because every properly looped back packet
* must have correct destination already attached by output routine.
*
* Such approach solves two big problems:
* 1. Not simplex devices are handled properly.
* 2. IP spoofing attempts are filtered with 100% of guarantee.
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
* called with rcu_read_lock()
*/
static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
u8 tos, struct net_device *dev)
{
struct fib_result res;
struct in_device *in_dev = __in_dev_get_rcu(dev);
struct flowi4 fl4;
unsigned int flags = 0;
u32 itag = 0;
struct rtable *rth;
unsigned int hash;
int err = -EINVAL;
struct net *net = dev_net(dev);
/* IP on this device is disabled. */
if (!in_dev)
goto out;
/* Check for the most weird martians, which can be not detected
by fib_lookup.
*/
if (ipv4_is_multicast(saddr) || ipv4_is_lbcast(saddr))
goto martian_source;
if (ipv4_is_lbcast(daddr) || (saddr == 0 && daddr == 0))
goto brd_input;
/* Accept zero addresses only to limited broadcast;
* I even do not know to fix it or not. Waiting for complains :-)
*/
if (ipv4_is_zeronet(saddr))
goto martian_source;
if (ipv4_is_zeronet(daddr))
goto martian_destination;
if (likely(!IN_DEV_ROUTE_LOCALNET(in_dev))) {
if (ipv4_is_loopback(daddr))
goto martian_destination;
if (ipv4_is_loopback(saddr))
goto martian_source;
}
/*
* Now we are ready to route packet.
*/
fl4.flowi4_oif = 0;
fl4.flowi4_iif = dev->ifindex;
fl4.flowi4_mark = skb->mark;
fl4.flowi4_tos = tos;
fl4.flowi4_scope = RT_SCOPE_UNIVERSE;
fl4.daddr = daddr;
fl4.saddr = saddr;
err = fib_lookup(net, &fl4, &res);
if (err != 0)
goto no_route;
RT_CACHE_STAT_INC(in_slow_tot);
if (res.type == RTN_BROADCAST)
goto brd_input;
if (res.type == RTN_LOCAL) {
err = fib_validate_source(skb, saddr, daddr, tos,
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
net->loopback_dev->ifindex,
dev, in_dev, &itag);
if (err < 0)
goto martian_source_keep_err;
if (err)
flags |= RTCF_DIRECTSRC;
goto local_input;
}
if (!IN_DEV_FORWARD(in_dev))
goto no_route;
if (res.type != RTN_UNICAST)
goto martian_destination;
err = ip_mkroute_input(skb, &res, &fl4, in_dev, daddr, saddr, tos);
out: return err;
brd_input:
if (skb->protocol != htons(ETH_P_IP))
goto e_inval;
if (!ipv4_is_zeronet(saddr)) {
err = fib_validate_source(skb, saddr, 0, tos, 0, dev,
in_dev, &itag);
if (err < 0)
goto martian_source_keep_err;
if (err)
flags |= RTCF_DIRECTSRC;
}
flags |= RTCF_BROADCAST;
res.type = RTN_BROADCAST;
RT_CACHE_STAT_INC(in_brd);
local_input:
rth = rt_dst_alloc(net->loopback_dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
if (!rth)
goto e_nobufs;
rth->dst.input= ip_local_deliver;
rth->dst.output= ip_rt_bug;
#ifdef CONFIG_IP_ROUTE_CLASSID
rth->dst.tclassid = itag;
#endif
rth->rt_key_dst = daddr;
rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(net);
rth->rt_flags = flags|RTCF_LOCAL;
rth->rt_type = res.type;
rth->rt_key_tos = tos;
rth->rt_dst = daddr;
rth->rt_src = saddr;
rth->rt_route_iif = dev->ifindex;
rth->rt_iif = dev->ifindex;
rth->rt_oif = 0;
rth->rt_mark = skb->mark;
rth->rt_pmtu = 0;
rth->rt_gateway = daddr;
rth->fi = NULL;
if (res.type == RTN_UNREACHABLE) {
rth->dst.input= ip_error;
rth->dst.error= -err;
rth->rt_flags &= ~RTCF_LOCAL;
}
hash = rt_hash(daddr, saddr, fl4.flowi4_iif, rt_genid(net));
rth = rt_intern_hash(hash, rth, skb, fl4.flowi4_iif);
err = 0;
if (IS_ERR(rth))
err = PTR_ERR(rth);
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
goto out;
no_route:
RT_CACHE_STAT_INC(in_no_route);
res.type = RTN_UNREACHABLE;
if (err == -ESRCH)
err = -ENETUNREACH;
goto local_input;
/*
* Do not cache martian addresses: they should be logged (RFC1812)
*/
martian_destination:
RT_CACHE_STAT_INC(in_martian_dst);
#ifdef CONFIG_IP_ROUTE_VERBOSE
if (IN_DEV_LOG_MARTIANS(in_dev))
net_warn_ratelimited("martian destination %pI4 from %pI4, dev %s\n",
&daddr, &saddr, dev->name);
#endif
e_inval:
err = -EINVAL;
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
goto out;
e_nobufs:
err = -ENOBUFS;
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
goto out;
martian_source:
err = -EINVAL;
martian_source_keep_err:
ip_handle_martian_source(dev, in_dev, skb, daddr, saddr);
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
goto out;
}
int ip_route_input_common(struct sk_buff *skb, __be32 daddr, __be32 saddr,
u8 tos, struct net_device *dev, bool noref)
{
struct rtable *rth;
unsigned int hash;
int iif = dev->ifindex;
struct net *net;
int res;
net = dev_net(dev);
rcu_read_lock();
if (!rt_caching(net))
goto skip_cache;
tos &= IPTOS_RT_MASK;
hash = rt_hash(daddr, saddr, iif, rt_genid(net));
for (rth = rcu_dereference(rt_hash_table[hash].chain); rth;
rth = rcu_dereference(rth->dst.rt_next)) {
if ((((__force u32)rth->rt_key_dst ^ (__force u32)daddr) |
((__force u32)rth->rt_key_src ^ (__force u32)saddr) |
(rth->rt_route_iif ^ iif) |
(rth->rt_key_tos ^ tos)) == 0 &&
rth->rt_mark == skb->mark &&
net_eq(dev_net(rth->dst.dev), net) &&
!rt_is_expired(rth)) {
if (noref) {
dst_use_noref(&rth->dst, jiffies);
skb_dst_set_noref(skb, &rth->dst);
} else {
dst_use(&rth->dst, jiffies);
skb_dst_set(skb, &rth->dst);
}
RT_CACHE_STAT_INC(in_hit);
rcu_read_unlock();
return 0;
}
RT_CACHE_STAT_INC(in_hlist_search);
}
skip_cache:
/* Multicast recognition logic is moved from route cache to here.
The problem was that too many Ethernet cards have broken/missing
hardware multicast filters :-( As result the host on multicasting
network acquires a lot of useless route cache entries, sort of
SDR messages from all the world. Now we try to get rid of them.
Really, provided software IP multicast filter is organized
reasonably (at least, hashed), it does not result in a slowdown
comparing with route cache reject entries.
Note, that multicast routers are not affected, because
route cache entry is created eventually.
*/
if (ipv4_is_multicast(daddr)) {
struct in_device *in_dev = __in_dev_get_rcu(dev);
if (in_dev) {
int our = ip_check_mc_rcu(in_dev, daddr, saddr,
ip_hdr(skb)->protocol);
if (our
#ifdef CONFIG_IP_MROUTE
||
(!ipv4_is_local_multicast(daddr) &&
IN_DEV_MFORWARD(in_dev))
#endif
) {
int res = ip_route_input_mc(skb, daddr, saddr,
tos, dev, our);
rcu_read_unlock();
return res;
}
}
rcu_read_unlock();
return -EINVAL;
}
res = ip_route_input_slow(skb, daddr, saddr, tos, dev);
rcu_read_unlock();
return res;
}
EXPORT_SYMBOL(ip_route_input_common);
fib: RCU conversion of fib_lookup() fib_lookup() converted to be called in RCU protected context, no reference taken and released on a contended cache line (fib_clntref) fib_table_lookup() and fib_semantic_match() get an additional parameter. struct fib_info gets an rcu_head field, and is freed after an rcu grace period. Stress test : (Sending 160.000.000 UDP frames on same neighbour, IP route cache disabled, dual E5540 @2.53GHz, 32bit kernel, FIB_HASH) (about same results for FIB_TRIE) Before patch : real 1m31.199s user 0m13.761s sys 23m24.780s After patch: real 1m5.375s user 0m14.997s sys 15m50.115s Before patch Profile : 13044.00 15.4% __ip_route_output_key vmlinux 8438.00 10.0% dst_destroy vmlinux 5983.00 7.1% fib_semantic_match vmlinux 5410.00 6.4% fib_rules_lookup vmlinux 4803.00 5.7% neigh_lookup vmlinux 4420.00 5.2% _raw_spin_lock vmlinux 3883.00 4.6% rt_set_nexthop vmlinux 3261.00 3.9% _raw_read_lock vmlinux 2794.00 3.3% fib_table_lookup vmlinux 2374.00 2.8% neigh_resolve_output vmlinux 2153.00 2.5% dst_alloc vmlinux 1502.00 1.8% _raw_read_lock_bh vmlinux 1484.00 1.8% kmem_cache_alloc vmlinux 1407.00 1.7% eth_header vmlinux 1406.00 1.7% ipv4_dst_destroy vmlinux 1298.00 1.5% __copy_from_user_ll vmlinux 1174.00 1.4% dev_queue_xmit vmlinux 1000.00 1.2% ip_output vmlinux After patch Profile : 13712.00 15.8% dst_destroy vmlinux 8548.00 9.9% __ip_route_output_key vmlinux 7017.00 8.1% neigh_lookup vmlinux 4554.00 5.3% fib_semantic_match vmlinux 4067.00 4.7% _raw_read_lock vmlinux 3491.00 4.0% dst_alloc vmlinux 3186.00 3.7% neigh_resolve_output vmlinux 3103.00 3.6% fib_table_lookup vmlinux 2098.00 2.4% _raw_read_lock_bh vmlinux 2081.00 2.4% kmem_cache_alloc vmlinux 2013.00 2.3% _raw_spin_lock vmlinux 1763.00 2.0% __copy_from_user_ll vmlinux 1763.00 2.0% ip_output vmlinux 1761.00 2.0% ipv4_dst_destroy vmlinux 1631.00 1.9% eth_header vmlinux 1440.00 1.7% _raw_read_unlock_bh vmlinux Reference results, if IP route cache is enabled : real 0m29.718s user 0m10.845s sys 7m37.341s 25213.00 29.5% __ip_route_output_key vmlinux 9011.00 10.5% dst_release vmlinux 4817.00 5.6% ip_push_pending_frames vmlinux 4232.00 5.0% ip_finish_output vmlinux 3940.00 4.6% udp_sendmsg vmlinux 3730.00 4.4% __copy_from_user_ll vmlinux 3716.00 4.4% ip_route_output_flow vmlinux 2451.00 2.9% __xfrm_lookup vmlinux 2221.00 2.6% ip_append_data vmlinux 1718.00 2.0% _raw_spin_lock_bh vmlinux 1655.00 1.9% __alloc_skb vmlinux 1572.00 1.8% sock_wfree vmlinux 1345.00 1.6% kfree vmlinux Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-10-05 18:41:36 +08:00
/* called with rcu_read_lock() */
static struct rtable *__mkroute_output(const struct fib_result *res,
const struct flowi4 *fl4,
__be32 orig_daddr, __be32 orig_saddr,
int orig_oif, __u8 orig_rtos,
struct net_device *dev_out,
unsigned int flags)
{
struct fib_info *fi = res->fi;
struct in_device *in_dev;
u16 type = res->type;
struct rtable *rth;
in_dev = __in_dev_get_rcu(dev_out);
if (!in_dev)
return ERR_PTR(-EINVAL);
if (likely(!IN_DEV_ROUTE_LOCALNET(in_dev)))
if (ipv4_is_loopback(fl4->saddr) && !(dev_out->flags & IFF_LOOPBACK))
return ERR_PTR(-EINVAL);
if (ipv4_is_lbcast(fl4->daddr))
type = RTN_BROADCAST;
else if (ipv4_is_multicast(fl4->daddr))
type = RTN_MULTICAST;
else if (ipv4_is_zeronet(fl4->daddr))
return ERR_PTR(-EINVAL);
if (dev_out->flags & IFF_LOOPBACK)
flags |= RTCF_LOCAL;
if (type == RTN_BROADCAST) {
flags |= RTCF_BROADCAST | RTCF_LOCAL;
fi = NULL;
} else if (type == RTN_MULTICAST) {
flags |= RTCF_MULTICAST | RTCF_LOCAL;
if (!ip_check_mc_rcu(in_dev, fl4->daddr, fl4->saddr,
fl4->flowi4_proto))
flags &= ~RTCF_LOCAL;
/* If multicast route do not exist use
* default one, but do not gateway in this case.
* Yes, it is hack.
*/
if (fi && res->prefixlen < 4)
fi = NULL;
}
rth = rt_dst_alloc(dev_out,
IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(in_dev, NOXFRM));
if (!rth)
return ERR_PTR(-ENOBUFS);
rth->dst.output = ip_output;
rth->rt_key_dst = orig_daddr;
rth->rt_key_src = orig_saddr;
rth->rt_genid = rt_genid(dev_net(dev_out));
rth->rt_flags = flags;
rth->rt_type = type;
rth->rt_key_tos = orig_rtos;
rth->rt_dst = fl4->daddr;
rth->rt_src = fl4->saddr;
rth->rt_route_iif = 0;
rth->rt_iif = orig_oif ? : dev_out->ifindex;
rth->rt_oif = orig_oif;
rth->rt_mark = fl4->flowi4_mark;
rth->rt_pmtu = 0;
rth->rt_gateway = fl4->daddr;
rth->fi = NULL;
RT_CACHE_STAT_INC(out_slow_tot);
if (flags & RTCF_LOCAL)
rth->dst.input = ip_local_deliver;
if (flags & (RTCF_BROADCAST | RTCF_MULTICAST)) {
if (flags & RTCF_LOCAL &&
!(dev_out->flags & IFF_LOOPBACK)) {
rth->dst.output = ip_mc_output;
RT_CACHE_STAT_INC(out_slow_mc);
}
#ifdef CONFIG_IP_MROUTE
if (type == RTN_MULTICAST) {
if (IN_DEV_MFORWARD(in_dev) &&
!ipv4_is_local_multicast(fl4->daddr)) {
rth->dst.input = ip_mr_input;
rth->dst.output = ip_mc_output;
}
}
#endif
}
rt_set_nexthop(rth, fl4, res, fi, type, 0);
if (fl4->flowi4_flags & FLOWI_FLAG_RT_NOCACHE)
rth->dst.flags |= DST_NOCACHE;
return rth;
}
/*
* Major route resolver routine.
* called with rcu_read_lock();
*/
static struct rtable *ip_route_output_slow(struct net *net, struct flowi4 *fl4)
{
struct net_device *dev_out = NULL;
__u8 tos = RT_FL_TOS(fl4);
unsigned int flags = 0;
struct fib_result res;
struct rtable *rth;
__be32 orig_daddr;
__be32 orig_saddr;
int orig_oif;
res.tclassid = 0;
res.fi = NULL;
res.table = NULL;
orig_daddr = fl4->daddr;
orig_saddr = fl4->saddr;
orig_oif = fl4->flowi4_oif;
fl4->flowi4_iif = net->loopback_dev->ifindex;
fl4->flowi4_tos = tos & IPTOS_RT_MASK;
fl4->flowi4_scope = ((tos & RTO_ONLINK) ?
RT_SCOPE_LINK : RT_SCOPE_UNIVERSE);
rcu_read_lock();
if (fl4->saddr) {
rth = ERR_PTR(-EINVAL);
if (ipv4_is_multicast(fl4->saddr) ||
ipv4_is_lbcast(fl4->saddr) ||
ipv4_is_zeronet(fl4->saddr))
goto out;
/* I removed check for oif == dev_out->oif here.
It was wrong for two reasons:
1. ip_dev_find(net, saddr) can return wrong iface, if saddr
is assigned to multiple interfaces.
2. Moreover, we are allowed to send packets with saddr
of another iface. --ANK
*/
if (fl4->flowi4_oif == 0 &&
(ipv4_is_multicast(fl4->daddr) ||
ipv4_is_lbcast(fl4->daddr))) {
/* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */
dev_out = __ip_dev_find(net, fl4->saddr, false);
if (dev_out == NULL)
goto out;
/* Special hack: user can direct multicasts
and limited broadcast via necessary interface
without fiddling with IP_MULTICAST_IF or IP_PKTINFO.
This hack is not just for fun, it allows
vic,vat and friends to work.
They bind socket to loopback, set ttl to zero
and expect that it will work.
From the viewpoint of routing cache they are broken,
because we are not allowed to build multicast path
with loopback source addr (look, routing cache
cannot know, that ttl is zero, so that packet
will not leave this host and route is valid).
Luckily, this hack is good workaround.
*/
fl4->flowi4_oif = dev_out->ifindex;
goto make_route;
}
if (!(fl4->flowi4_flags & FLOWI_FLAG_ANYSRC)) {
/* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */
if (!__ip_dev_find(net, fl4->saddr, false))
goto out;
}
}
if (fl4->flowi4_oif) {
dev_out = dev_get_by_index_rcu(net, fl4->flowi4_oif);
rth = ERR_PTR(-ENODEV);
if (dev_out == NULL)
goto out;
/* RACE: Check return value of inet_select_addr instead. */
if (!(dev_out->flags & IFF_UP) || !__in_dev_get_rcu(dev_out)) {
rth = ERR_PTR(-ENETUNREACH);
goto out;
}
if (ipv4_is_local_multicast(fl4->daddr) ||
ipv4_is_lbcast(fl4->daddr)) {
if (!fl4->saddr)
fl4->saddr = inet_select_addr(dev_out, 0,
RT_SCOPE_LINK);
goto make_route;
}
if (fl4->saddr) {
if (ipv4_is_multicast(fl4->daddr))
fl4->saddr = inet_select_addr(dev_out, 0,
fl4->flowi4_scope);
else if (!fl4->daddr)
fl4->saddr = inet_select_addr(dev_out, 0,
RT_SCOPE_HOST);
}
}
if (!fl4->daddr) {
fl4->daddr = fl4->saddr;
if (!fl4->daddr)
fl4->daddr = fl4->saddr = htonl(INADDR_LOOPBACK);
dev_out = net->loopback_dev;
fl4->flowi4_oif = net->loopback_dev->ifindex;
res.type = RTN_LOCAL;
flags |= RTCF_LOCAL;
goto make_route;
}
if (fib_lookup(net, fl4, &res)) {
res.fi = NULL;
res.table = NULL;
if (fl4->flowi4_oif) {
/* Apparently, routing tables are wrong. Assume,
that the destination is on link.
WHY? DW.
Because we are allowed to send to iface
even if it has NO routes and NO assigned
addresses. When oif is specified, routing
tables are looked up with only one purpose:
to catch if destination is gatewayed, rather than
direct. Moreover, if MSG_DONTROUTE is set,
we send packet, ignoring both routing tables
and ifaddr state. --ANK
We could make it even if oif is unknown,
likely IPv6, but we do not.
*/
if (fl4->saddr == 0)
fl4->saddr = inet_select_addr(dev_out, 0,
RT_SCOPE_LINK);
res.type = RTN_UNICAST;
goto make_route;
}
rth = ERR_PTR(-ENETUNREACH);
goto out;
}
if (res.type == RTN_LOCAL) {
if (!fl4->saddr) {
if (res.fi->fib_prefsrc)
fl4->saddr = res.fi->fib_prefsrc;
else
fl4->saddr = fl4->daddr;
}
dev_out = net->loopback_dev;
fl4->flowi4_oif = dev_out->ifindex;
res.fi = NULL;
flags |= RTCF_LOCAL;
goto make_route;
}
#ifdef CONFIG_IP_ROUTE_MULTIPATH
if (res.fi->fib_nhs > 1 && fl4->flowi4_oif == 0)
fib_select_multipath(&res);
else
#endif
if (!res.prefixlen &&
res.table->tb_num_default > 1 &&
res.type == RTN_UNICAST && !fl4->flowi4_oif)
fib_select_default(&res);
if (!fl4->saddr)
fl4->saddr = FIB_RES_PREFSRC(net, res);
dev_out = FIB_RES_DEV(res);
fl4->flowi4_oif = dev_out->ifindex;
make_route:
rth = __mkroute_output(&res, fl4, orig_daddr, orig_saddr, orig_oif,
tos, dev_out, flags);
if (!IS_ERR(rth)) {
unsigned int hash;
hash = rt_hash(orig_daddr, orig_saddr, orig_oif,
rt_genid(dev_net(dev_out)));
rth = rt_intern_hash(hash, rth, NULL, orig_oif);
}
out:
rcu_read_unlock();
return rth;
}
struct rtable *__ip_route_output_key(struct net *net, struct flowi4 *flp4)
{
struct rtable *rth;
unsigned int hash;
if (!rt_caching(net))
goto slow_output;
hash = rt_hash(flp4->daddr, flp4->saddr, flp4->flowi4_oif, rt_genid(net));
rcu_read_lock_bh();
for (rth = rcu_dereference_bh(rt_hash_table[hash].chain); rth;
rth = rcu_dereference_bh(rth->dst.rt_next)) {
if (rth->rt_key_dst == flp4->daddr &&
rth->rt_key_src == flp4->saddr &&
rt_is_output_route(rth) &&
rth->rt_oif == flp4->flowi4_oif &&
rth->rt_mark == flp4->flowi4_mark &&
!((rth->rt_key_tos ^ flp4->flowi4_tos) &
(IPTOS_RT_MASK | RTO_ONLINK)) &&
net_eq(dev_net(rth->dst.dev), net) &&
!rt_is_expired(rth)) {
dst_use(&rth->dst, jiffies);
RT_CACHE_STAT_INC(out_hit);
rcu_read_unlock_bh();
if (!flp4->saddr)
flp4->saddr = rth->rt_src;
if (!flp4->daddr)
flp4->daddr = rth->rt_dst;
return rth;
}
RT_CACHE_STAT_INC(out_hlist_search);
}
rcu_read_unlock_bh();
slow_output:
return ip_route_output_slow(net, flp4);
}
EXPORT_SYMBOL_GPL(__ip_route_output_key);
static struct dst_entry *ipv4_blackhole_dst_check(struct dst_entry *dst, u32 cookie)
{
return NULL;
}
static unsigned int ipv4_blackhole_mtu(const struct dst_entry *dst)
{
unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
return mtu ? : dst->dev->mtu;
}
static void ipv4_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
struct sk_buff *skb, u32 mtu)
{
}
static void ipv4_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
struct sk_buff *skb)
{
}
net: provide cow_metrics() methods to blackhole dst_ops Since commit 62fa8a846d7d (net: Implement read-only protection and COW'ing of metrics.) the kernel throws an oops. [ 101.620985] BUG: unable to handle kernel NULL pointer dereference at (null) [ 101.621050] IP: [< (null)>] (null) [ 101.621084] PGD 6e53c067 PUD 3dd6a067 PMD 0 [ 101.621122] Oops: 0010 [#1] SMP [ 101.621153] last sysfs file: /sys/devices/virtual/ppp/ppp/uevent [ 101.621192] CPU 2 [ 101.621206] Modules linked in: l2tp_ppp pppox ppp_generic slhc l2tp_netlink l2tp_core deflate zlib_deflate twofish_x86_64 twofish_common des_generic cbc ecb sha1_generic hmac af_key iptable_filter snd_pcm_oss snd_mixer_oss snd_seq snd_seq_device loop snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_pcm snd_timer snd i2c_i801 iTCO_wdt psmouse soundcore snd_page_alloc evdev uhci_hcd ehci_hcd thermal [ 101.621552] [ 101.621567] Pid: 5129, comm: openl2tpd Not tainted 2.6.39-rc4-Quad #3 Gigabyte Technology Co., Ltd. G33-DS3R/G33-DS3R [ 101.621637] RIP: 0010:[<0000000000000000>] [< (null)>] (null) [ 101.621684] RSP: 0018:ffff88003ddeba60 EFLAGS: 00010202 [ 101.621716] RAX: ffff88003ddb5600 RBX: ffff88003ddb5600 RCX: 0000000000000020 [ 101.621758] RDX: ffffffff81a69a00 RSI: ffffffff81b7ee61 RDI: ffff88003ddb5600 [ 101.621800] RBP: ffff8800537cd900 R08: 0000000000000000 R09: ffff88003ddb5600 [ 101.621840] R10: 0000000000000005 R11: 0000000000014b38 R12: ffff88003ddb5600 [ 101.621881] R13: ffffffff81b7e480 R14: ffffffff81b7e8b8 R15: ffff88003ddebad8 [ 101.621924] FS: 00007f06e4182700(0000) GS:ffff88007fd00000(0000) knlGS:0000000000000000 [ 101.621971] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 101.622005] CR2: 0000000000000000 CR3: 0000000045274000 CR4: 00000000000006e0 [ 101.622046] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 101.622087] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 101.622129] Process openl2tpd (pid: 5129, threadinfo ffff88003ddea000, task ffff88003de9a280) [ 101.622177] Stack: [ 101.622191] ffffffff81447efa ffff88007d3ded80 ffff88003de9a280 ffff88007d3ded80 [ 101.622245] 0000000000000001 ffff88003ddebbb8 ffffffff8148d5a7 0000000000000212 [ 101.622299] ffff88003dcea000 ffff88003dcea188 ffffffff00000001 ffffffff81b7e480 [ 101.622353] Call Trace: [ 101.622374] [<ffffffff81447efa>] ? ipv4_blackhole_route+0x1ba/0x210 [ 101.622415] [<ffffffff8148d5a7>] ? xfrm_lookup+0x417/0x510 [ 101.622450] [<ffffffff8127672a>] ? extract_buf+0x9a/0x140 [ 101.622485] [<ffffffff8144c6a0>] ? __ip_flush_pending_frames+0x70/0x70 [ 101.622526] [<ffffffff8146fbbf>] ? udp_sendmsg+0x62f/0x810 [ 101.622562] [<ffffffff813f98a6>] ? sock_sendmsg+0x116/0x130 [ 101.622599] [<ffffffff8109df58>] ? find_get_page+0x18/0x90 [ 101.622633] [<ffffffff8109fd6a>] ? filemap_fault+0x12a/0x4b0 [ 101.622668] [<ffffffff813fb5c4>] ? move_addr_to_kernel+0x64/0x90 [ 101.622706] [<ffffffff81405d5a>] ? verify_iovec+0x7a/0xf0 [ 101.622739] [<ffffffff813fc772>] ? sys_sendmsg+0x292/0x420 [ 101.622774] [<ffffffff810b994a>] ? handle_pte_fault+0x8a/0x7c0 [ 101.622810] [<ffffffff810b76fe>] ? __pte_alloc+0xae/0x130 [ 101.622844] [<ffffffff810ba2f8>] ? handle_mm_fault+0x138/0x380 [ 101.622880] [<ffffffff81024af9>] ? do_page_fault+0x189/0x410 [ 101.622915] [<ffffffff813fbe03>] ? sys_getsockname+0xf3/0x110 [ 101.622952] [<ffffffff81450c4d>] ? ip_setsockopt+0x4d/0xa0 [ 101.622986] [<ffffffff813f9932>] ? sockfd_lookup_light+0x22/0x90 [ 101.623024] [<ffffffff814b61fb>] ? system_call_fastpath+0x16/0x1b [ 101.623060] Code: Bad RIP value. [ 101.623090] RIP [< (null)>] (null) [ 101.623125] RSP <ffff88003ddeba60> [ 101.623146] CR2: 0000000000000000 [ 101.650871] ---[ end trace ca3856a7d8e8dad4 ]--- [ 101.651011] __sk_free: optmem leakage (160 bytes) detected. The oops happens in dst_metrics_write_ptr() include/net/dst.h:124: return dst->ops->cow_metrics(dst, p); dst->ops->cow_metrics is NULL and causes the oops. Provide cow_metrics() methods, like we did in commit 214f45c91bb (net: provide default_advmss() methods to blackhole dst_ops) Signed-off-by: Held Bernhard <berny156@gmx.de> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-04-25 06:07:32 +08:00
static u32 *ipv4_rt_blackhole_cow_metrics(struct dst_entry *dst,
unsigned long old)
{
return NULL;
}
static struct dst_ops ipv4_dst_blackhole_ops = {
.family = AF_INET,
.protocol = cpu_to_be16(ETH_P_IP),
.destroy = ipv4_dst_destroy,
.check = ipv4_blackhole_dst_check,
.mtu = ipv4_blackhole_mtu,
.default_advmss = ipv4_default_advmss,
.update_pmtu = ipv4_rt_blackhole_update_pmtu,
.redirect = ipv4_rt_blackhole_redirect,
net: provide cow_metrics() methods to blackhole dst_ops Since commit 62fa8a846d7d (net: Implement read-only protection and COW'ing of metrics.) the kernel throws an oops. [ 101.620985] BUG: unable to handle kernel NULL pointer dereference at (null) [ 101.621050] IP: [< (null)>] (null) [ 101.621084] PGD 6e53c067 PUD 3dd6a067 PMD 0 [ 101.621122] Oops: 0010 [#1] SMP [ 101.621153] last sysfs file: /sys/devices/virtual/ppp/ppp/uevent [ 101.621192] CPU 2 [ 101.621206] Modules linked in: l2tp_ppp pppox ppp_generic slhc l2tp_netlink l2tp_core deflate zlib_deflate twofish_x86_64 twofish_common des_generic cbc ecb sha1_generic hmac af_key iptable_filter snd_pcm_oss snd_mixer_oss snd_seq snd_seq_device loop snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_pcm snd_timer snd i2c_i801 iTCO_wdt psmouse soundcore snd_page_alloc evdev uhci_hcd ehci_hcd thermal [ 101.621552] [ 101.621567] Pid: 5129, comm: openl2tpd Not tainted 2.6.39-rc4-Quad #3 Gigabyte Technology Co., Ltd. G33-DS3R/G33-DS3R [ 101.621637] RIP: 0010:[<0000000000000000>] [< (null)>] (null) [ 101.621684] RSP: 0018:ffff88003ddeba60 EFLAGS: 00010202 [ 101.621716] RAX: ffff88003ddb5600 RBX: ffff88003ddb5600 RCX: 0000000000000020 [ 101.621758] RDX: ffffffff81a69a00 RSI: ffffffff81b7ee61 RDI: ffff88003ddb5600 [ 101.621800] RBP: ffff8800537cd900 R08: 0000000000000000 R09: ffff88003ddb5600 [ 101.621840] R10: 0000000000000005 R11: 0000000000014b38 R12: ffff88003ddb5600 [ 101.621881] R13: ffffffff81b7e480 R14: ffffffff81b7e8b8 R15: ffff88003ddebad8 [ 101.621924] FS: 00007f06e4182700(0000) GS:ffff88007fd00000(0000) knlGS:0000000000000000 [ 101.621971] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 101.622005] CR2: 0000000000000000 CR3: 0000000045274000 CR4: 00000000000006e0 [ 101.622046] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 101.622087] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 101.622129] Process openl2tpd (pid: 5129, threadinfo ffff88003ddea000, task ffff88003de9a280) [ 101.622177] Stack: [ 101.622191] ffffffff81447efa ffff88007d3ded80 ffff88003de9a280 ffff88007d3ded80 [ 101.622245] 0000000000000001 ffff88003ddebbb8 ffffffff8148d5a7 0000000000000212 [ 101.622299] ffff88003dcea000 ffff88003dcea188 ffffffff00000001 ffffffff81b7e480 [ 101.622353] Call Trace: [ 101.622374] [<ffffffff81447efa>] ? ipv4_blackhole_route+0x1ba/0x210 [ 101.622415] [<ffffffff8148d5a7>] ? xfrm_lookup+0x417/0x510 [ 101.622450] [<ffffffff8127672a>] ? extract_buf+0x9a/0x140 [ 101.622485] [<ffffffff8144c6a0>] ? __ip_flush_pending_frames+0x70/0x70 [ 101.622526] [<ffffffff8146fbbf>] ? udp_sendmsg+0x62f/0x810 [ 101.622562] [<ffffffff813f98a6>] ? sock_sendmsg+0x116/0x130 [ 101.622599] [<ffffffff8109df58>] ? find_get_page+0x18/0x90 [ 101.622633] [<ffffffff8109fd6a>] ? filemap_fault+0x12a/0x4b0 [ 101.622668] [<ffffffff813fb5c4>] ? move_addr_to_kernel+0x64/0x90 [ 101.622706] [<ffffffff81405d5a>] ? verify_iovec+0x7a/0xf0 [ 101.622739] [<ffffffff813fc772>] ? sys_sendmsg+0x292/0x420 [ 101.622774] [<ffffffff810b994a>] ? handle_pte_fault+0x8a/0x7c0 [ 101.622810] [<ffffffff810b76fe>] ? __pte_alloc+0xae/0x130 [ 101.622844] [<ffffffff810ba2f8>] ? handle_mm_fault+0x138/0x380 [ 101.622880] [<ffffffff81024af9>] ? do_page_fault+0x189/0x410 [ 101.622915] [<ffffffff813fbe03>] ? sys_getsockname+0xf3/0x110 [ 101.622952] [<ffffffff81450c4d>] ? ip_setsockopt+0x4d/0xa0 [ 101.622986] [<ffffffff813f9932>] ? sockfd_lookup_light+0x22/0x90 [ 101.623024] [<ffffffff814b61fb>] ? system_call_fastpath+0x16/0x1b [ 101.623060] Code: Bad RIP value. [ 101.623090] RIP [< (null)>] (null) [ 101.623125] RSP <ffff88003ddeba60> [ 101.623146] CR2: 0000000000000000 [ 101.650871] ---[ end trace ca3856a7d8e8dad4 ]--- [ 101.651011] __sk_free: optmem leakage (160 bytes) detected. The oops happens in dst_metrics_write_ptr() include/net/dst.h:124: return dst->ops->cow_metrics(dst, p); dst->ops->cow_metrics is NULL and causes the oops. Provide cow_metrics() methods, like we did in commit 214f45c91bb (net: provide default_advmss() methods to blackhole dst_ops) Signed-off-by: Held Bernhard <berny156@gmx.de> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-04-25 06:07:32 +08:00
.cow_metrics = ipv4_rt_blackhole_cow_metrics,
.neigh_lookup = ipv4_neigh_lookup,
};
struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig)
{
struct rtable *rt = dst_alloc(&ipv4_dst_blackhole_ops, NULL, 1, 0, 0);
struct rtable *ort = (struct rtable *) dst_orig;
if (rt) {
struct dst_entry *new = &rt->dst;
new->__use = 1;
new->input = dst_discard;
new->output = dst_discard;
new->dev = ort->dst.dev;
if (new->dev)
dev_hold(new->dev);
rt->rt_key_dst = ort->rt_key_dst;
rt->rt_key_src = ort->rt_key_src;
rt->rt_key_tos = ort->rt_key_tos;
rt->rt_route_iif = ort->rt_route_iif;
rt->rt_iif = ort->rt_iif;
rt->rt_oif = ort->rt_oif;
rt->rt_mark = ort->rt_mark;
rt->rt_pmtu = ort->rt_pmtu;
rt->rt_genid = rt_genid(net);
rt->rt_flags = ort->rt_flags;
rt->rt_type = ort->rt_type;
rt->rt_dst = ort->rt_dst;
rt->rt_src = ort->rt_src;
rt->rt_gateway = ort->rt_gateway;
net: Implement read-only protection and COW'ing of metrics. Routing metrics are now copy-on-write. Initially a route entry points it's metrics at a read-only location. If a routing table entry exists, it will point there. Else it will point at the all zero metric place-holder called 'dst_default_metrics'. The writeability state of the metrics is stored in the low bits of the metrics pointer, we have two bits left to spare if we want to store more states. For the initial implementation, COW is implemented simply via kmalloc. However future enhancements will change this to place the writable metrics somewhere else, in order to increase sharing. Very likely this "somewhere else" will be the inetpeer cache. Note also that this means that metrics updates may transiently fail if we cannot COW the metrics successfully. But even by itself, this patch should decrease memory usage and increase cache locality especially for routing workloads. In those cases the read-only metric copies stay in place and never get written to. TCP workloads where metrics get updated, and those rare cases where PMTU triggers occur, will take a very slight performance hit. But that hit will be alleviated when the long-term writable metrics move to a more sharable location. Since the metrics storage went from a u32 array of RTAX_MAX entries to what is essentially a pointer, some retooling of the dst_entry layout was necessary. Most importantly, we need to preserve the alignment of the reference count so that it doesn't share cache lines with the read-mostly state, as per Eric Dumazet's alignment assertion checks. The only non-trivial bit here is the move of the 'flags' member into the writeable cacheline. This is OK since we are always accessing the flags around the same moment when we made a modification to the reference count. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-01-27 12:51:05 +08:00
rt->fi = ort->fi;
if (rt->fi)
atomic_inc(&rt->fi->fib_clntref);
dst_free(new);
}
dst_release(dst_orig);
return rt ? &rt->dst : ERR_PTR(-ENOMEM);
}
struct rtable *ip_route_output_flow(struct net *net, struct flowi4 *flp4,
struct sock *sk)
{
struct rtable *rt = __ip_route_output_key(net, flp4);
if (IS_ERR(rt))
return rt;
if (flp4->flowi4_proto)
rt = (struct rtable *) xfrm_lookup(net, &rt->dst,
flowi4_to_flowi(flp4),
sk, 0);
return rt;
}
EXPORT_SYMBOL_GPL(ip_route_output_flow);
static int rt_fill_info(struct net *net,
struct sk_buff *skb, u32 pid, u32 seq, int event,
int nowait, unsigned int flags)
{
struct rtable *rt = skb_rtable(skb);
struct rtmsg *r;
struct nlmsghdr *nlh;
unsigned long expires = 0;
u32 error;
nlh = nlmsg_put(skb, pid, seq, event, sizeof(*r), flags);
if (nlh == NULL)
return -EMSGSIZE;
r = nlmsg_data(nlh);
r->rtm_family = AF_INET;
r->rtm_dst_len = 32;
r->rtm_src_len = 0;
r->rtm_tos = rt->rt_key_tos;
r->rtm_table = RT_TABLE_MAIN;
if (nla_put_u32(skb, RTA_TABLE, RT_TABLE_MAIN))
goto nla_put_failure;
r->rtm_type = rt->rt_type;
r->rtm_scope = RT_SCOPE_UNIVERSE;
r->rtm_protocol = RTPROT_UNSPEC;
r->rtm_flags = (rt->rt_flags & ~0xFFFF) | RTM_F_CLONED;
if (rt->rt_flags & RTCF_NOTIFY)
r->rtm_flags |= RTM_F_NOTIFY;
if (nla_put_be32(skb, RTA_DST, rt->rt_dst))
goto nla_put_failure;
if (rt->rt_key_src) {
r->rtm_src_len = 32;
if (nla_put_be32(skb, RTA_SRC, rt->rt_key_src))
goto nla_put_failure;
}
if (rt->dst.dev &&
nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
goto nla_put_failure;
#ifdef CONFIG_IP_ROUTE_CLASSID
if (rt->dst.tclassid &&
nla_put_u32(skb, RTA_FLOW, rt->dst.tclassid))
goto nla_put_failure;
#endif
if (!rt_is_input_route(rt) &&
rt->rt_src != rt->rt_key_src) {
if (nla_put_be32(skb, RTA_PREFSRC, rt->rt_src))
goto nla_put_failure;
}
if (rt->rt_dst != rt->rt_gateway &&
nla_put_be32(skb, RTA_GATEWAY, rt->rt_gateway))
goto nla_put_failure;
if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
goto nla_put_failure;
if (rt->rt_mark &&
nla_put_be32(skb, RTA_MARK, rt->rt_mark))
goto nla_put_failure;
error = rt->dst.error;
expires = rt->dst.expires;
if (expires) {
if (time_before(jiffies, expires))
expires -= jiffies;
else
expires = 0;
}
if (rt_is_input_route(rt)) {
#ifdef CONFIG_IP_MROUTE
__be32 dst = rt->rt_dst;
if (ipv4_is_multicast(dst) && !ipv4_is_local_multicast(dst) &&
IPV4_DEVCONF_ALL(net, MC_FORWARDING)) {
int err = ipmr_get_route(net, skb,
rt->rt_src, rt->rt_dst,
r, nowait);
if (err <= 0) {
if (!nowait) {
if (err == 0)
return 0;
goto nla_put_failure;
} else {
if (err == -EMSGSIZE)
goto nla_put_failure;
error = err;
}
}
} else
#endif
if (nla_put_u32(skb, RTA_IIF, rt->rt_iif))
goto nla_put_failure;
}
if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, error) < 0)
goto nla_put_failure;
return nlmsg_end(skb, nlh);
nla_put_failure:
nlmsg_cancel(skb, nlh);
return -EMSGSIZE;
}
static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, void *arg)
{
struct net *net = sock_net(in_skb->sk);
struct rtmsg *rtm;
struct nlattr *tb[RTA_MAX+1];
struct rtable *rt = NULL;
__be32 dst = 0;
__be32 src = 0;
u32 iif;
int err;
int mark;
struct sk_buff *skb;
err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv4_policy);
if (err < 0)
goto errout;
rtm = nlmsg_data(nlh);
skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
if (skb == NULL) {
err = -ENOBUFS;
goto errout;
}
/* Reserve room for dummy headers, this skb can pass
through good chunk of routing engine.
*/
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
/* Bugfix: need to give ip_route_input enough of an IP header to not gag. */
ip_hdr(skb)->protocol = IPPROTO_ICMP;
skb_reserve(skb, MAX_HEADER + sizeof(struct iphdr));
src = tb[RTA_SRC] ? nla_get_be32(tb[RTA_SRC]) : 0;
dst = tb[RTA_DST] ? nla_get_be32(tb[RTA_DST]) : 0;
iif = tb[RTA_IIF] ? nla_get_u32(tb[RTA_IIF]) : 0;
mark = tb[RTA_MARK] ? nla_get_u32(tb[RTA_MARK]) : 0;
if (iif) {
struct net_device *dev;
dev = __dev_get_by_index(net, iif);
if (dev == NULL) {
err = -ENODEV;
goto errout_free;
}
skb->protocol = htons(ETH_P_IP);
skb->dev = dev;
skb->mark = mark;
local_bh_disable();
err = ip_route_input(skb, dst, src, rtm->rtm_tos, dev);
local_bh_enable();
rt = skb_rtable(skb);
if (err == 0 && rt->dst.error)
err = -rt->dst.error;
} else {
struct flowi4 fl4 = {
.daddr = dst,
.saddr = src,
.flowi4_tos = rtm->rtm_tos,
.flowi4_oif = tb[RTA_OIF] ? nla_get_u32(tb[RTA_OIF]) : 0,
.flowi4_mark = mark,
};
rt = ip_route_output_key(net, &fl4);
err = 0;
if (IS_ERR(rt))
err = PTR_ERR(rt);
}
if (err)
goto errout_free;
skb_dst_set(skb, &rt->dst);
if (rtm->rtm_flags & RTM_F_NOTIFY)
rt->rt_flags |= RTCF_NOTIFY;
err = rt_fill_info(net, skb, NETLINK_CB(in_skb).pid, nlh->nlmsg_seq,
RTM_NEWROUTE, 0, 0);
if (err <= 0)
goto errout_free;
err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
errout:
return err;
errout_free:
kfree_skb(skb);
goto errout;
}
int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb)
{
struct rtable *rt;
int h, s_h;
int idx, s_idx;
struct net *net;
net = sock_net(skb->sk);
s_h = cb->args[0];
[IPV4] ROUTE: ip_rt_dump() is unecessary slow I noticed "ip route list cache x.y.z.t" can be *very* slow. While strace-ing -T it I also noticed that first part of route cache is fetched quite fast : recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202 GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3772 <0.000047> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\ 202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3736 <0.000042> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\ 202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3740 <0.000055> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\ 202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3712 <0.000043> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\ 202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3732 <0.000053> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202 GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3708 <0.000052> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202 GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3680 <0.000041> while the part at the end of the table is more expensive: recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3656 <0.003857> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\204\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3772 <0.003891> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3712 <0.003765> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3700 <0.003879> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3676 <0.003797> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"p\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\2\0\2\0"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3724 <0.003856> recvmsg(3, {msg_name(12)={sa_family=AF_NETLINK, pid=0, groups=00000000}, msg_iov(1)=[{"\234\0\0\0\30\0\2\0\254i\202GXm\0\0\2 \0\376\0\0\1\0\2"..., 16384}], msg_controllen=0, msg_flags=0}, 0) = 3736 <0.003848> The following patch corrects this performance/latency problem, removing quadratic behavior. Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-08 13:52:14 +08:00
if (s_h < 0)
s_h = 0;
s_idx = idx = cb->args[1];
for (h = s_h; h <= rt_hash_mask; h++, s_idx = 0) {
if (!rt_hash_table[h].chain)
continue;
rcu_read_lock_bh();
for (rt = rcu_dereference_bh(rt_hash_table[h].chain), idx = 0; rt;
rt = rcu_dereference_bh(rt->dst.rt_next), idx++) {
if (!net_eq(dev_net(rt->dst.dev), net) || idx < s_idx)
continue;
if (rt_is_expired(rt))
continue;
skb_dst_set_noref(skb, &rt->dst);
if (rt_fill_info(net, skb, NETLINK_CB(cb->skb).pid,
cb->nlh->nlmsg_seq, RTM_NEWROUTE,
1, NLM_F_MULTI) <= 0) {
skb_dst_drop(skb);
rcu_read_unlock_bh();
goto done;
}
skb_dst_drop(skb);
}
rcu_read_unlock_bh();
}
done:
cb->args[0] = h;
cb->args[1] = idx;
return skb->len;
}
void ip_rt_multicast_event(struct in_device *in_dev)
{
rt_cache_flush(dev_net(in_dev->dev), 0);
}
#ifdef CONFIG_SYSCTL
static int ipv4_sysctl_rtcache_flush(ctl_table *__ctl, int write,
void __user *buffer,
size_t *lenp, loff_t *ppos)
{
if (write) {
int flush_delay;
ctl_table ctl;
struct net *net;
memcpy(&ctl, __ctl, sizeof(ctl));
ctl.data = &flush_delay;
proc_dointvec(&ctl, write, buffer, lenp, ppos);
net = (struct net *)__ctl->extra1;
rt_cache_flush(net, flush_delay);
return 0;
}
return -EINVAL;
}
static ctl_table ipv4_route_table[] = {
{
.procname = "gc_thresh",
.data = &ipv4_dst_ops.gc_thresh,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "max_size",
.data = &ip_rt_max_size,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
/* Deprecated. Use gc_min_interval_ms */
.procname = "gc_min_interval",
.data = &ip_rt_gc_min_interval,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
},
{
.procname = "gc_min_interval_ms",
.data = &ip_rt_gc_min_interval,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_ms_jiffies,
},
{
.procname = "gc_timeout",
.data = &ip_rt_gc_timeout,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
},
{
.procname = "gc_interval",
.data = &ip_rt_gc_interval,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
},
{
.procname = "redirect_load",
.data = &ip_rt_redirect_load,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "redirect_number",
.data = &ip_rt_redirect_number,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "redirect_silence",
.data = &ip_rt_redirect_silence,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "error_cost",
.data = &ip_rt_error_cost,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "error_burst",
.data = &ip_rt_error_burst,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "gc_elasticity",
.data = &ip_rt_gc_elasticity,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "mtu_expires",
.data = &ip_rt_mtu_expires,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
},
{
.procname = "min_pmtu",
.data = &ip_rt_min_pmtu,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{
.procname = "min_adv_mss",
.data = &ip_rt_min_advmss,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{ }
};
static struct ctl_table ipv4_route_flush_table[] = {
{
.procname = "flush",
.maxlen = sizeof(int),
.mode = 0200,
.proc_handler = ipv4_sysctl_rtcache_flush,
},
{ },
};
static __net_init int sysctl_route_net_init(struct net *net)
{
struct ctl_table *tbl;
tbl = ipv4_route_flush_table;
if (!net_eq(net, &init_net)) {
tbl = kmemdup(tbl, sizeof(ipv4_route_flush_table), GFP_KERNEL);
if (tbl == NULL)
goto err_dup;
}
tbl[0].extra1 = net;
net->ipv4.route_hdr = register_net_sysctl(net, "net/ipv4/route", tbl);
if (net->ipv4.route_hdr == NULL)
goto err_reg;
return 0;
err_reg:
if (tbl != ipv4_route_flush_table)
kfree(tbl);
err_dup:
return -ENOMEM;
}
static __net_exit void sysctl_route_net_exit(struct net *net)
{
struct ctl_table *tbl;
tbl = net->ipv4.route_hdr->ctl_table_arg;
unregister_net_sysctl_table(net->ipv4.route_hdr);
BUG_ON(tbl == ipv4_route_flush_table);
kfree(tbl);
}
static __net_initdata struct pernet_operations sysctl_route_ops = {
.init = sysctl_route_net_init,
.exit = sysctl_route_net_exit,
};
#endif
static __net_init int rt_genid_init(struct net *net)
{
get_random_bytes(&net->ipv4.rt_genid,
sizeof(net->ipv4.rt_genid));
get_random_bytes(&net->ipv4.dev_addr_genid,
sizeof(net->ipv4.dev_addr_genid));
return 0;
}
static __net_initdata struct pernet_operations rt_genid_ops = {
.init = rt_genid_init,
};
static int __net_init ipv4_inetpeer_init(struct net *net)
{
struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
if (!bp)
return -ENOMEM;
inet_peer_base_init(bp);
net->ipv4.peers = bp;
return 0;
}
static void __net_exit ipv4_inetpeer_exit(struct net *net)
{
struct inet_peer_base *bp = net->ipv4.peers;
net->ipv4.peers = NULL;
inetpeer_invalidate_tree(bp);
kfree(bp);
}
static __net_initdata struct pernet_operations ipv4_inetpeer_ops = {
.init = ipv4_inetpeer_init,
.exit = ipv4_inetpeer_exit,
};
#ifdef CONFIG_IP_ROUTE_CLASSID
struct ip_rt_acct __percpu *ip_rt_acct __read_mostly;
#endif /* CONFIG_IP_ROUTE_CLASSID */
static __initdata unsigned long rhash_entries;
static int __init set_rhash_entries(char *str)
{
ssize_t ret;
if (!str)
return 0;
ret = kstrtoul(str, 0, &rhash_entries);
if (ret)
return 0;
return 1;
}
__setup("rhash_entries=", set_rhash_entries);
int __init ip_rt_init(void)
{
int rc = 0;
#ifdef CONFIG_IP_ROUTE_CLASSID
ip_rt_acct = __alloc_percpu(256 * sizeof(struct ip_rt_acct), __alignof__(struct ip_rt_acct));
if (!ip_rt_acct)
panic("IP: failed to allocate ip_rt_acct\n");
#endif
ipv4_dst_ops.kmem_cachep =
kmem_cache_create("ip_dst_cache", sizeof(struct rtable), 0,
SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
ipv4_dst_blackhole_ops.kmem_cachep = ipv4_dst_ops.kmem_cachep;
if (dst_entries_init(&ipv4_dst_ops) < 0)
panic("IP: failed to allocate ipv4_dst_ops counter\n");
if (dst_entries_init(&ipv4_dst_blackhole_ops) < 0)
panic("IP: failed to allocate ipv4_dst_blackhole_ops counter\n");
rt_hash_table = (struct rt_hash_bucket *)
alloc_large_system_hash("IP route cache",
sizeof(struct rt_hash_bucket),
rhash_entries,
(totalram_pages >= 128 * 1024) ?
[IPV4] tcp/route: Another look at hash table sizes The tcp_ehash hash table gets too big on systems with really big memory. It is worse on systems with pages larger than 4KB. It wastes memory that could be better used. It also makes the netstat command slow because reading /proc/net/tcp and /proc/net/tcp6 needs to go through the full hash table. The default value should not be larger for larger page sizes. It seems that the effect of page size is an unintended error dating back a long time. I also wonder if the default value really should be a larger fraction of memory for systems with more memory. While systems with really big ram can afford more space for hash tables, it is not clear to me that they benefit from increasing the allocation ratio for this table. The amount of memory allocated is determined by net/ipv4/tcp.c:tcp_init and mm/page_alloc.c:alloc_large_system_hash. tcp_init calls alloc_large_system_hash passing parameters- bucketsize=sizeof(struct tcp_ehash_bucket) numentries=thash_entries scale=(num_physpages >= 128 * 1024) ? (25-PAGE_SHIFT) : (27-PAGE_SHIFT) limit=0 On i386, PAGE_SHIFT is 12 for a page size of 4K On ia64, PAGE_SHIFT defaults to 14 for a page size of 16K The num_physpages test above makes the allocation take a larger fraction of the total memory on systems with larger memory. The threshold size for a i386 system is 512MB. For an ia64 system with 16KB pages the threshold is 2GB. For smaller memory systems- On i386, scale = (27 - 12) = 15 On ia64, scale = (27 - 14) = 13 For larger memory systems- On i386, scale = (25 - 12) = 13 On ia64, scale = (25 - 14) = 11 For the rest of this discussion, I'll just track the larger memory case. The default behavior has numentries=thash_entries=0, so the allocated size is determined by either scale or by the default limit of 1/16 of total memory. In alloc_large_system_hash- | numentries = (flags & HASH_HIGHMEM) ? nr_all_pages : nr_kernel_pages; | numentries += (1UL << (20 - PAGE_SHIFT)) - 1; | numentries >>= 20 - PAGE_SHIFT; | numentries <<= 20 - PAGE_SHIFT; At this point, numentries is pages for all of memory, rounded up to the nearest megabyte boundary. | /* limit to 1 bucket per 2^scale bytes of low memory */ | if (scale > PAGE_SHIFT) | numentries >>= (scale - PAGE_SHIFT); | else | numentries <<= (PAGE_SHIFT - scale); On i386, numentries >>= (13 - 12), so numentries is 1/8196 of bytes of total memory. On ia64, numentries <<= (14 - 11), so numentries is 1/2048 of bytes of total memory. | log2qty = long_log2(numentries); | | do { | size = bucketsize << log2qty; bucketsize is 16, so size is 16 times numentries, rounded down to a power of two. On i386, size is 1/512 of bytes of total memory. On ia64, size is 1/128 of bytes of total memory. For smaller systems the results are On i386, size is 1/2048 of bytes of total memory. On ia64, size is 1/512 of bytes of total memory. The large page effect can be removed by just replacing the use of PAGE_SHIFT with a constant of 12 in the calls to alloc_large_system_hash. That makes them more like the other uses of that function from fs/inode.c and fs/dcache.c Signed-off-by: David S. Miller <davem@davemloft.net>
2005-11-30 08:12:55 +08:00
15 : 17,
0,
&rt_hash_log,
&rt_hash_mask,
0,
rhash_entries ? 0 : 512 * 1024);
memset(rt_hash_table, 0, (rt_hash_mask + 1) * sizeof(struct rt_hash_bucket));
rt_hash_lock_init();
ipv4_dst_ops.gc_thresh = (rt_hash_mask + 1);
ip_rt_max_size = (rt_hash_mask + 1) * 16;
devinet_init();
ip_fib_init();
INIT_DELAYED_WORK_DEFERRABLE(&expires_work, rt_worker_func);
expires_ljiffies = jiffies;
schedule_delayed_work(&expires_work,
net_random() % ip_rt_gc_interval + ip_rt_gc_interval);
if (ip_rt_proc_init())
pr_err("Unable to create route proc files\n");
#ifdef CONFIG_XFRM
xfrm_init();
xfrm4_init(ip_rt_max_size);
#endif
rtnl_register(PF_INET, RTM_GETROUTE, inet_rtm_getroute, NULL, NULL);
#ifdef CONFIG_SYSCTL
register_pernet_subsys(&sysctl_route_ops);
#endif
register_pernet_subsys(&rt_genid_ops);
register_pernet_subsys(&ipv4_inetpeer_ops);
return rc;
}
#ifdef CONFIG_SYSCTL
/*
* We really need to sanitize the damn ipv4 init order, then all
* this nonsense will go away.
*/
void __init ip_static_sysctl_init(void)
{
register_net_sysctl(&init_net, "net/ipv4/route", ipv4_route_table);
}
#endif