2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* Linux INET6 implementation
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Pedro Roque <roque@di.fc.ul.pt>
|
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef _IP6_FIB_H
|
|
|
|
#define _IP6_FIB_H
|
|
|
|
|
|
|
|
#include <linux/ipv6_route.h>
|
|
|
|
#include <linux/rtnetlink.h>
|
|
|
|
#include <linux/spinlock.h>
|
2017-08-03 18:28:16 +07:00
|
|
|
#include <linux/notifier.h>
|
2006-08-22 14:01:08 +07:00
|
|
|
#include <net/dst.h>
|
|
|
|
#include <net/flow.h>
|
|
|
|
#include <net/netlink.h>
|
2010-12-01 03:27:11 +07:00
|
|
|
#include <net/inetpeer.h>
|
2017-08-03 18:28:16 +07:00
|
|
|
#include <net/fib_notifier.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2009-07-31 08:52:15 +07:00
|
|
|
#ifdef CONFIG_IPV6_MULTIPLE_TABLES
|
|
|
|
#define FIB6_TABLE_HASHSZ 256
|
|
|
|
#else
|
|
|
|
#define FIB6_TABLE_HASHSZ 1
|
|
|
|
#endif
|
|
|
|
|
2017-10-07 02:06:01 +07:00
|
|
|
#define RT6_DEBUG 2
|
|
|
|
|
|
|
|
#if RT6_DEBUG >= 3
|
|
|
|
#define RT6_TRACE(x...) pr_debug(x)
|
|
|
|
#else
|
|
|
|
#define RT6_TRACE(x...) do { ; } while (0)
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
struct rt6_info;
|
2018-04-18 07:33:24 +07:00
|
|
|
struct fib6_info;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2009-11-03 10:26:03 +07:00
|
|
|
struct fib6_config {
|
2006-08-22 14:01:08 +07:00
|
|
|
u32 fc_table;
|
|
|
|
u32 fc_metric;
|
|
|
|
int fc_dst_len;
|
|
|
|
int fc_src_len;
|
|
|
|
int fc_ifindex;
|
|
|
|
u32 fc_flags;
|
|
|
|
u32 fc_protocol;
|
2017-02-03 03:37:08 +07:00
|
|
|
u16 fc_type; /* only 8 bits are used */
|
|
|
|
u16 fc_delete_all_nh : 1,
|
|
|
|
__unused : 15;
|
2006-08-22 14:01:08 +07:00
|
|
|
|
|
|
|
struct in6_addr fc_dst;
|
|
|
|
struct in6_addr fc_src;
|
2011-04-14 04:10:57 +07:00
|
|
|
struct in6_addr fc_prefsrc;
|
2006-08-22 14:01:08 +07:00
|
|
|
struct in6_addr fc_gateway;
|
|
|
|
|
|
|
|
unsigned long fc_expires;
|
|
|
|
struct nlattr *fc_mx;
|
|
|
|
int fc_mx_len;
|
2012-10-22 10:42:09 +07:00
|
|
|
int fc_mp_len;
|
|
|
|
struct nlattr *fc_mp;
|
2006-08-22 14:01:08 +07:00
|
|
|
|
|
|
|
struct nl_info fc_nlinfo;
|
2015-07-21 15:43:48 +07:00
|
|
|
struct nlattr *fc_encap;
|
|
|
|
u16 fc_encap_type;
|
2006-08-22 14:01:08 +07:00
|
|
|
};
|
|
|
|
|
2009-11-03 10:26:03 +07:00
|
|
|
struct fib6_node {
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
struct fib6_node __rcu *parent;
|
|
|
|
struct fib6_node __rcu *left;
|
|
|
|
struct fib6_node __rcu *right;
|
2006-12-14 07:38:29 +07:00
|
|
|
#ifdef CONFIG_IPV6_SUBTREES
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
struct fib6_node __rcu *subtree;
|
2006-12-14 07:38:29 +07:00
|
|
|
#endif
|
2018-04-18 07:33:26 +07:00
|
|
|
struct fib6_info __rcu *leaf;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
__u16 fn_bit; /* bit key */
|
|
|
|
__u16 fn_flags;
|
2014-10-07 00:58:35 +07:00
|
|
|
int fn_sernum;
|
2018-04-18 07:33:26 +07:00
|
|
|
struct fib6_info __rcu *rr_ptr;
|
2017-08-21 23:47:10 +07:00
|
|
|
struct rcu_head rcu;
|
2005-04-17 05:20:36 +07:00
|
|
|
};
|
|
|
|
|
2017-10-07 02:06:01 +07:00
|
|
|
struct fib6_gc_args {
|
|
|
|
int timeout;
|
|
|
|
int more;
|
|
|
|
};
|
|
|
|
|
2006-08-24 07:22:24 +07:00
|
|
|
#ifndef CONFIG_IPV6_SUBTREES
|
|
|
|
#define FIB6_SUBTREE(fn) NULL
|
|
|
|
#else
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
#define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
|
2006-08-24 07:22:24 +07:00
|
|
|
#endif
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* routing information
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2009-11-03 10:26:03 +07:00
|
|
|
struct rt6key {
|
2005-04-17 05:20:36 +07:00
|
|
|
struct in6_addr addr;
|
|
|
|
int plen;
|
|
|
|
};
|
|
|
|
|
2006-08-05 13:20:06 +07:00
|
|
|
struct fib6_table;
|
|
|
|
|
2017-10-07 02:05:57 +07:00
|
|
|
struct rt6_exception_bucket {
|
|
|
|
struct hlist_head chain;
|
|
|
|
int depth;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct rt6_exception {
|
|
|
|
struct hlist_node hlist;
|
|
|
|
struct rt6_info *rt6i;
|
|
|
|
unsigned long stamp;
|
|
|
|
struct rcu_head rcu;
|
|
|
|
};
|
|
|
|
|
|
|
|
#define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
|
|
|
|
#define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
|
|
|
|
#define FIB6_MAX_DEPTH 5
|
|
|
|
|
2018-04-18 07:33:14 +07:00
|
|
|
struct fib6_nh {
|
|
|
|
struct in6_addr nh_gw;
|
|
|
|
struct net_device *nh_dev;
|
|
|
|
struct lwtunnel_state *nh_lwtstate;
|
|
|
|
|
|
|
|
unsigned int nh_flags;
|
|
|
|
atomic_t nh_upper_bound;
|
|
|
|
int nh_weight;
|
|
|
|
};
|
|
|
|
|
2018-04-18 07:33:24 +07:00
|
|
|
struct fib6_info {
|
2018-04-19 05:38:59 +07:00
|
|
|
struct fib6_table *fib6_table;
|
2018-04-18 07:33:24 +07:00
|
|
|
struct fib6_info __rcu *rt6_next;
|
2018-04-19 05:38:59 +07:00
|
|
|
struct fib6_node __rcu *fib6_node;
|
2018-04-18 07:33:24 +07:00
|
|
|
|
|
|
|
/* Multipath routes:
|
|
|
|
* siblings is a list of fib6_info that have the the same metric/weight,
|
|
|
|
* destination, but not the same gateway. nsiblings is just a cache
|
|
|
|
* to speed up lookup.
|
|
|
|
*/
|
2018-04-19 05:38:59 +07:00
|
|
|
struct list_head fib6_siblings;
|
|
|
|
unsigned int fib6_nsiblings;
|
2018-04-18 07:33:24 +07:00
|
|
|
|
2018-04-19 05:38:59 +07:00
|
|
|
atomic_t fib6_ref;
|
2018-04-18 07:33:24 +07:00
|
|
|
unsigned long expires;
|
|
|
|
struct dst_metrics *fib6_metrics;
|
|
|
|
#define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1]
|
|
|
|
|
2018-04-19 05:38:59 +07:00
|
|
|
struct rt6key fib6_dst;
|
|
|
|
u32 fib6_flags;
|
|
|
|
struct rt6key fib6_src;
|
|
|
|
struct rt6key fib6_prefsrc;
|
2018-04-18 07:33:24 +07:00
|
|
|
|
|
|
|
struct rt6_info * __percpu *rt6i_pcpu;
|
|
|
|
struct rt6_exception_bucket __rcu *rt6i_exception_bucket;
|
|
|
|
|
2018-04-19 05:38:59 +07:00
|
|
|
u32 fib6_metric;
|
|
|
|
u8 fib6_protocol;
|
2018-04-18 07:33:24 +07:00
|
|
|
u8 fib6_type;
|
|
|
|
u8 exception_bucket_flushed:1,
|
|
|
|
should_flush:1,
|
|
|
|
dst_nocount:1,
|
|
|
|
dst_nopolicy:1,
|
|
|
|
dst_host:1,
|
|
|
|
unused:3;
|
|
|
|
|
|
|
|
struct fib6_nh fib6_nh;
|
|
|
|
};
|
|
|
|
|
2009-11-03 10:26:03 +07:00
|
|
|
struct rt6_info {
|
2010-06-11 13:31:35 +07:00
|
|
|
struct dst_entry dst;
|
2018-04-21 05:38:02 +07:00
|
|
|
struct fib6_info __rcu *from;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2018-04-18 07:33:27 +07:00
|
|
|
struct rt6key rt6i_dst;
|
|
|
|
struct rt6key rt6i_src;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct in6_addr rt6i_gateway;
|
2018-04-18 07:33:27 +07:00
|
|
|
struct inet6_dev *rt6i_idev;
|
2010-04-01 05:24:22 +07:00
|
|
|
u32 rt6i_flags;
|
2011-04-14 04:10:57 +07:00
|
|
|
struct rt6key rt6i_prefsrc;
|
2007-11-14 12:33:32 +07:00
|
|
|
|
2015-05-23 10:56:04 +07:00
|
|
|
struct list_head rt6i_uncached;
|
|
|
|
struct uncached_list *rt6i_uncached_list;
|
|
|
|
|
2010-04-01 05:24:22 +07:00
|
|
|
/* more non-fragment space at head required */
|
|
|
|
unsigned short rt6i_nfheader_len;
|
2005-04-17 05:20:36 +07:00
|
|
|
};
|
|
|
|
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
#define for_each_fib6_node_rt_rcu(fn) \
|
|
|
|
for (rt = rcu_dereference((fn)->leaf); rt; \
|
2017-11-29 03:40:15 +07:00
|
|
|
rt = rcu_dereference(rt->rt6_next))
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
|
|
|
|
#define for_each_fib6_walker_rt(w) \
|
|
|
|
for (rt = (w)->leaf; rt; \
|
2017-11-29 03:40:15 +07:00
|
|
|
rt = rcu_dereference_protected(rt->rt6_next, 1))
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
|
2006-10-13 14:17:25 +07:00
|
|
|
static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst)
|
|
|
|
{
|
|
|
|
return ((struct rt6_info *)dst)->rt6i_idev;
|
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
static inline void fib6_clean_expires(struct fib6_info *f6i)
|
2018-04-18 07:33:17 +07:00
|
|
|
{
|
2018-04-19 05:38:59 +07:00
|
|
|
f6i->fib6_flags &= ~RTF_EXPIRES;
|
2018-04-18 07:33:17 +07:00
|
|
|
f6i->expires = 0;
|
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
static inline void fib6_set_expires(struct fib6_info *f6i,
|
2018-04-18 07:33:17 +07:00
|
|
|
unsigned long expires)
|
|
|
|
{
|
|
|
|
f6i->expires = expires;
|
2018-04-19 05:38:59 +07:00
|
|
|
f6i->fib6_flags |= RTF_EXPIRES;
|
2018-04-18 07:33:17 +07:00
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
static inline bool fib6_check_expired(const struct fib6_info *f6i)
|
2018-04-18 07:33:17 +07:00
|
|
|
{
|
2018-04-19 05:38:59 +07:00
|
|
|
if (f6i->fib6_flags & RTF_EXPIRES)
|
2018-04-18 07:33:17 +07:00
|
|
|
return time_after(jiffies, f6i->expires);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2017-08-21 23:47:10 +07:00
|
|
|
/* Function to safely get fn->sernum for passed in rt
|
|
|
|
* and store result in passed in cookie.
|
|
|
|
* Return true if we can get cookie safely
|
|
|
|
* Return false if not
|
|
|
|
*/
|
2018-04-21 05:37:58 +07:00
|
|
|
static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i,
|
|
|
|
u32 *cookie)
|
2017-08-21 23:47:10 +07:00
|
|
|
{
|
|
|
|
struct fib6_node *fn;
|
|
|
|
bool status = false;
|
|
|
|
|
2018-04-19 05:38:59 +07:00
|
|
|
fn = rcu_dereference(f6i->fib6_node);
|
2017-08-21 23:47:10 +07:00
|
|
|
|
|
|
|
if (fn) {
|
|
|
|
*cookie = fn->fn_sernum;
|
2017-10-07 02:06:07 +07:00
|
|
|
/* pairs with smp_wmb() in fib6_update_sernum_upto_root() */
|
|
|
|
smp_rmb();
|
2017-08-21 23:47:10 +07:00
|
|
|
status = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
2015-05-23 10:56:01 +07:00
|
|
|
static inline u32 rt6_get_cookie(const struct rt6_info *rt)
|
|
|
|
{
|
2018-04-21 05:38:02 +07:00
|
|
|
struct fib6_info *from;
|
2017-08-21 23:47:10 +07:00
|
|
|
u32 cookie = 0;
|
|
|
|
|
2018-04-21 05:38:00 +07:00
|
|
|
rcu_read_lock();
|
|
|
|
|
2018-04-21 05:38:02 +07:00
|
|
|
from = rcu_dereference(rt->from);
|
|
|
|
if (from && (rt->rt6i_flags & RTF_PCPU ||
|
|
|
|
unlikely(!list_empty(&rt->rt6i_uncached))))
|
|
|
|
fib6_get_cookie_safe(from, &cookie);
|
2017-08-21 23:47:10 +07:00
|
|
|
|
2018-04-21 05:38:00 +07:00
|
|
|
rcu_read_unlock();
|
|
|
|
|
2017-08-21 23:47:10 +07:00
|
|
|
return cookie;
|
2015-05-23 10:56:01 +07:00
|
|
|
}
|
2012-04-06 07:13:10 +07:00
|
|
|
|
2012-10-29 07:13:19 +07:00
|
|
|
static inline void ip6_rt_put(struct rt6_info *rt)
|
|
|
|
{
|
|
|
|
/* dst_release() accepts a NULL parameter.
|
|
|
|
* We rely on dst being first structure in struct rt6_info
|
|
|
|
*/
|
|
|
|
BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0);
|
|
|
|
dst_release(&rt->dst);
|
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
struct fib6_info *fib6_info_alloc(gfp_t gfp_flags);
|
|
|
|
void fib6_info_destroy(struct fib6_info *f6i);
|
2018-04-18 07:33:24 +07:00
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
static inline void fib6_info_hold(struct fib6_info *f6i)
|
2018-04-18 07:33:24 +07:00
|
|
|
{
|
2018-04-19 05:38:59 +07:00
|
|
|
atomic_inc(&f6i->fib6_ref);
|
2018-04-18 07:33:24 +07:00
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
static inline void fib6_info_release(struct fib6_info *f6i)
|
2018-04-18 07:33:24 +07:00
|
|
|
{
|
2018-04-19 05:38:59 +07:00
|
|
|
if (f6i && atomic_dec_and_test(&f6i->fib6_ref))
|
2018-04-18 07:33:24 +07:00
|
|
|
fib6_info_destroy(f6i);
|
|
|
|
}
|
|
|
|
|
2014-10-07 00:58:34 +07:00
|
|
|
enum fib6_walk_state {
|
|
|
|
#ifdef CONFIG_IPV6_SUBTREES
|
|
|
|
FWS_S,
|
|
|
|
#endif
|
|
|
|
FWS_L,
|
|
|
|
FWS_R,
|
|
|
|
FWS_C,
|
|
|
|
FWS_U
|
|
|
|
};
|
|
|
|
|
|
|
|
struct fib6_walker {
|
2010-02-18 15:13:30 +07:00
|
|
|
struct list_head lh;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct fib6_node *root, *node;
|
2018-04-18 07:33:26 +07:00
|
|
|
struct fib6_info *leaf;
|
2014-10-07 00:58:34 +07:00
|
|
|
enum fib6_walk_state state;
|
2010-02-08 12:19:03 +07:00
|
|
|
unsigned int skip;
|
|
|
|
unsigned int count;
|
2014-10-07 00:58:34 +07:00
|
|
|
int (*func)(struct fib6_walker *);
|
2005-04-17 05:20:36 +07:00
|
|
|
void *args;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct rt6_statistics {
|
2017-10-07 02:06:11 +07:00
|
|
|
__u32 fib_nodes; /* all fib6 nodes */
|
|
|
|
__u32 fib_route_nodes; /* intermediate nodes */
|
|
|
|
__u32 fib_rt_entries; /* rt entries in fib table */
|
|
|
|
__u32 fib_rt_cache; /* cached rt entries in exception table */
|
|
|
|
__u32 fib_discarded_routes; /* total number of routes delete */
|
|
|
|
|
|
|
|
/* The following stats are not protected by any lock */
|
|
|
|
atomic_t fib_rt_alloc; /* total number of routes alloced */
|
|
|
|
atomic_t fib_rt_uncache; /* rt entries in uncached list */
|
2005-04-17 05:20:36 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
#define RTN_TL_ROOT 0x0001
|
|
|
|
#define RTN_ROOT 0x0002 /* tree root node */
|
|
|
|
#define RTN_RTINFO 0x0004 /* node with valid routing info */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* priority levels (or metrics)
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
2006-08-05 13:20:06 +07:00
|
|
|
struct fib6_table {
|
|
|
|
struct hlist_node tb6_hlist;
|
|
|
|
u32 tb6_id;
|
ipv6: replace rwlock with rcu and spinlock in fib6_table
With all the preparation work before, we are now ready to replace rwlock
with rcu and spinlock in fib6_table.
That means now all fib6_node in fib6_table are protected by rcu. And
when freeing fib6_node, call_rcu() is used to wait for the rcu grace
period before releasing the memory.
When accessing fib6_node, corresponding rcu APIs need to be used.
And all previous sessions protected by the write lock will now be
protected by the spin lock per table.
All previous sessions protected by read lock will now be protected by
rcu_read_lock().
A couple of things to note here:
1. As part of the work of replacing rwlock with rcu, the linked list of
fn->leaf now has to be rcu protected as well. So both fn->leaf and
rt->dst.rt6_next are now __rcu tagged and corresponding rcu APIs are
used when manipulating them.
2. For fn->rr_ptr, first of all, it also needs to be rcu protected now
and is tagged with __rcu and rcu APIs are used in corresponding places.
Secondly, fn->rr_ptr is changed in rt6_select() which is a reader
thread. This makes the issue a bit complicated. We think a valid
solution for it is to let rt6_select() grab the tb6_lock if it decides
to change it. As it is not in the normal operation and only happens when
there is no valid neighbor cache for the route, we think the performance
impact should be low.
3. fib6_walk_continue() has to be called with tb6_lock held even in the
route dumping related functions, e.g. inet6_dump_fib(),
fib6_tables_dump() and ipv6_route_seq_ops. It is because
fib6_walk_continue() makes modifications to the walker structure, and so
are fib6_repair_tree() and fib6_del_route(). In order to do proper
syncing between them, we need to let fib6_walk_continue() hold the lock.
We may be able to do further improvement on the way we do the tree walk
to get rid of the need for holding the spin lock. But not for now.
4. When fib6_del_route() removes a route from the tree, we no longer
mark rt->dst.rt6_next to NULL to make simultaneous reader be able to
further traverse the list with rcu. However, rt->dst.rt6_next is only
valid within this same rcu period. No one should access it later.
5. All the operation of atomic_inc(rt->rt6i_ref) is changed to be
performed before we publish this route (either by linking it to fn->leaf
or insert it in the list pointed by fn->leaf) just to be safe because as
soon as we publish the route, some read thread will be able to access it.
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-07 02:06:10 +07:00
|
|
|
spinlock_t tb6_lock;
|
2006-08-05 13:20:06 +07:00
|
|
|
struct fib6_node tb6_root;
|
2012-06-11 14:01:52 +07:00
|
|
|
struct inet_peer_base tb6_peers;
|
2016-10-25 00:52:35 +07:00
|
|
|
unsigned int flags;
|
2017-08-03 18:28:19 +07:00
|
|
|
unsigned int fib_seq;
|
2016-10-25 00:52:35 +07:00
|
|
|
#define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
|
2006-08-05 13:20:06 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
#define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
|
|
|
|
#define RT6_TABLE_MAIN RT_TABLE_MAIN
|
|
|
|
#define RT6_TABLE_DFLT RT6_TABLE_MAIN
|
|
|
|
#define RT6_TABLE_INFO RT6_TABLE_MAIN
|
|
|
|
#define RT6_TABLE_PREFIX RT6_TABLE_MAIN
|
|
|
|
|
|
|
|
#ifdef CONFIG_IPV6_MULTIPLE_TABLES
|
|
|
|
#define FIB6_TABLE_MIN 1
|
|
|
|
#define FIB6_TABLE_MAX RT_TABLE_MAX
|
2006-08-04 17:39:02 +07:00
|
|
|
#define RT6_TABLE_LOCAL RT_TABLE_LOCAL
|
2006-08-05 13:20:06 +07:00
|
|
|
#else
|
|
|
|
#define FIB6_TABLE_MIN RT_TABLE_MAIN
|
|
|
|
#define FIB6_TABLE_MAX FIB6_TABLE_MIN
|
2006-08-04 17:39:02 +07:00
|
|
|
#define RT6_TABLE_LOCAL RT6_TABLE_MAIN
|
2006-08-05 13:20:06 +07:00
|
|
|
#endif
|
|
|
|
|
2008-03-05 04:48:30 +07:00
|
|
|
typedef struct rt6_info *(*pol_lookup_t)(struct net *,
|
|
|
|
struct fib6_table *,
|
2018-03-02 23:32:17 +07:00
|
|
|
struct flowi6 *,
|
|
|
|
const struct sk_buff *, int);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2017-08-03 18:28:17 +07:00
|
|
|
struct fib6_entry_notifier_info {
|
|
|
|
struct fib_notifier_info info; /* must be first */
|
2018-04-18 07:33:26 +07:00
|
|
|
struct fib6_info *rt;
|
2017-08-03 18:28:17 +07:00
|
|
|
};
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* exported functions
|
|
|
|
*/
|
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
struct fib6_table *fib6_get_table(struct net *net, u32 id);
|
|
|
|
struct fib6_table *fib6_new_table(struct net *net, u32 id);
|
|
|
|
struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6,
|
2018-03-02 23:32:17 +07:00
|
|
|
const struct sk_buff *skb,
|
2013-09-22 00:22:42 +07:00
|
|
|
int flags, pol_lookup_t lookup);
|
2006-08-05 13:20:06 +07:00
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
struct fib6_node *fib6_lookup(struct fib6_node *root,
|
|
|
|
const struct in6_addr *daddr,
|
|
|
|
const struct in6_addr *saddr);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
struct fib6_node *fib6_locate(struct fib6_node *root,
|
|
|
|
const struct in6_addr *daddr, int dst_len,
|
2017-10-07 02:06:02 +07:00
|
|
|
const struct in6_addr *saddr, int src_len,
|
|
|
|
bool exact_match);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg),
|
2013-12-27 15:32:38 +07:00
|
|
|
void *arg);
|
2006-08-05 13:20:06 +07:00
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
int fib6_add(struct fib6_node *root, struct fib6_info *rt,
|
2018-04-18 07:33:16 +07:00
|
|
|
struct nl_info *info, struct netlink_ext_ack *extack);
|
2018-04-18 07:33:26 +07:00
|
|
|
int fib6_del(struct fib6_info *rt, struct nl_info *info);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2018-04-19 05:39:01 +07:00
|
|
|
static inline struct net_device *fib6_info_nh_dev(const struct fib6_info *f6i)
|
|
|
|
{
|
|
|
|
return f6i->fib6_nh.nh_dev;
|
|
|
|
}
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info,
|
2015-09-14 00:18:33 +07:00
|
|
|
unsigned int flags);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
void fib6_run_gc(unsigned long expires, struct net *net, bool force);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
void fib6_gc_cleanup(void);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2013-09-22 00:22:42 +07:00
|
|
|
int fib6_init(void);
|
2006-08-04 17:39:02 +07:00
|
|
|
|
2013-09-21 21:55:59 +07:00
|
|
|
int ipv6_route_open(struct inode *inode, struct file *file);
|
|
|
|
|
2017-08-03 18:28:16 +07:00
|
|
|
int call_fib6_notifier(struct notifier_block *nb, struct net *net,
|
|
|
|
enum fib_event_type event_type,
|
|
|
|
struct fib_notifier_info *info);
|
|
|
|
int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
|
|
|
|
struct fib_notifier_info *info);
|
|
|
|
|
|
|
|
int __net_init fib6_notifier_init(struct net *net);
|
|
|
|
void __net_exit fib6_notifier_exit(struct net *net);
|
|
|
|
|
2017-08-03 18:28:19 +07:00
|
|
|
unsigned int fib6_tables_seq_read(struct net *net);
|
|
|
|
int fib6_tables_dump(struct net *net, struct notifier_block *nb);
|
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
void fib6_update_sernum(struct net *net, struct fib6_info *rt);
|
|
|
|
void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt);
|
2017-10-07 02:05:56 +07:00
|
|
|
|
2018-04-18 07:33:26 +07:00
|
|
|
void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val);
|
|
|
|
static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric)
|
2018-04-18 07:33:16 +07:00
|
|
|
{
|
|
|
|
return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric));
|
|
|
|
}
|
|
|
|
|
2007-12-08 15:14:54 +07:00
|
|
|
#ifdef CONFIG_IPV6_MULTIPLE_TABLES
|
2013-09-22 00:22:42 +07:00
|
|
|
int fib6_rules_init(void);
|
|
|
|
void fib6_rules_cleanup(void);
|
2017-08-03 18:28:15 +07:00
|
|
|
bool fib6_rule_default(const struct fib_rule *rule);
|
2017-08-03 18:28:18 +07:00
|
|
|
int fib6_rules_dump(struct net *net, struct notifier_block *nb);
|
|
|
|
unsigned int fib6_rules_seq_read(struct net *net);
|
2018-03-01 10:43:22 +07:00
|
|
|
|
|
|
|
static inline bool fib6_rules_early_flow_dissect(struct net *net,
|
|
|
|
struct sk_buff *skb,
|
|
|
|
struct flowi6 *fl6,
|
|
|
|
struct flow_keys *flkeys)
|
|
|
|
{
|
|
|
|
unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
|
|
|
|
|
|
|
|
if (!net->ipv6.fib6_rules_require_fldissect)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
skb_flow_dissect_flow_keys(skb, flkeys, flag);
|
|
|
|
fl6->fl6_sport = flkeys->ports.src;
|
|
|
|
fl6->fl6_dport = flkeys->ports.dst;
|
|
|
|
fl6->flowi6_proto = flkeys->basic.ip_proto;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
2007-12-08 15:14:54 +07:00
|
|
|
#else
|
|
|
|
static inline int fib6_rules_init(void)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
static inline void fib6_rules_cleanup(void)
|
|
|
|
{
|
|
|
|
return ;
|
|
|
|
}
|
2017-08-03 18:28:15 +07:00
|
|
|
static inline bool fib6_rule_default(const struct fib_rule *rule)
|
|
|
|
{
|
|
|
|
return true;
|
|
|
|
}
|
2017-08-03 18:28:18 +07:00
|
|
|
static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
static inline unsigned int fib6_rules_seq_read(struct net *net)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2018-03-01 10:43:22 +07:00
|
|
|
static inline bool fib6_rules_early_flow_dissect(struct net *net,
|
|
|
|
struct sk_buff *skb,
|
|
|
|
struct flowi6 *fl6,
|
|
|
|
struct flow_keys *flkeys)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
2007-12-08 15:14:54 +07:00
|
|
|
#endif
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif
|