David S. Miller
f11e6659ce
[IPV6]: Fix routing round-robin locking. ...
As per RFC2461, section 6.3.6, item #2 , when no routers on the
matching list are known to be reachable or probably reachable we
do round robin on those available routes so that we make sure
to probe as many of them as possible to detect when one becomes
reachable faster.
Each routing table has a rwlock protecting the tree and the linked
list of routes at each leaf. The round robin code executes during
lookup and thus with the rwlock taken as a reader. A small local
spinlock tries to provide protection but this does not work at all
for two reasons:
1) The round-robin list manipulation, as coded, goes like this (with
read lock held):
walk routes finding head and tail
spin_lock();
rotate list using head and tail
spin_unlock();
While one thread is rotating the list, another thread can
end up with stale values of head and tail and then proceed
to corrupt the list when it gets the lock. This ends up causing
the OOPS in fib6_add() later onthat many people have been hitting.
2) All the other code paths that run with the rwlock held as
a reader do not expect the list to change on them, they
expect it to remain completely fixed while they hold the
lock in that way.
So, simply stated, it is impossible to implement this correctly using
a manipulation of the list without violating the rwlock locking
semantics.
Reimplement using a per-fib6_node round-robin pointer. This way we
don't need to manipulate the list at all, and since the round-robin
pointer can only ever point to real existing entries we don't need
to perform any locking on the changing of the round-robin pointer
itself. We only need to reset the round-robin pointer to NULL when
the entry it is pointing to is removed.
The idea is from Thomas Graf and it is very similar to how this
was implemented before the advanced router selection code when in.
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-25 18:48:05 -07:00
[PATCH] hci endianness annotations
2006-12-13 09:05:52 -08:00
[IRDA] net/irda/: proper prototypes
2007-02-26 11:42:43 -08:00
[S390]: Add AF_IUCV socket support
2007-02-08 13:51:54 -08:00
[NETFILTER]: conntrack: fix {nf,ip}_ct_iterate_cleanup endless loops
2007-03-05 13:25:18 -08:00
[SCTP]: Reset some transport and association variables on restart
2007-03-20 00:09:45 -07:00
[PKT_SCHED]: Kill pkt_act.h inlining.
2006-09-22 14:55:10 -07:00
[TIPC]: endianness annotations
2006-12-02 21:21:08 -08:00
[PKT_SCHED]: Kill pkt_act.h inlining.
2006-09-22 14:55:10 -07:00
[IPV6]: Misc endianness annotations.
2006-12-02 21:22:52 -08:00
[IPSEC]: Use HMAC template and hash interface
2006-09-21 11:46:18 +10:00
[IPV6]: Assorted trivial endianness annotations.
2006-12-02 21:22:50 -08:00
[ATM]: Annotations.
2006-12-02 21:22:55 -08:00
[PATCH] mark struct file_operations const 1
2007-02-12 09:48:44 -08:00
[NET]: Make mangling a checksum (0 -> 0xffff on the wire) explicit.
2006-12-02 21:23:39 -08:00
NetLabel: use the correct CIPSOv4 MLS label limits
2006-12-02 21:24:12 -08:00
[NET]: Reduce sizeof(struct flowi) by 20 bytes.
2006-10-21 20:24:01 -07:00
[DECNET]: cleanups
2006-09-22 14:54:40 -07:00
[DECNET]: Convert decnet route to use the new dst_entry 'next' pointer
2007-02-10 23:20:43 -08:00
[NET]: IP header modifier helpers annotations.
2006-12-02 21:23:40 -08:00
[NET]: Reorder fields of struct dst_entry
2007-02-10 23:20:45 -08:00
[IPSEC] esp: Defer output IV initialization to first use.
2006-09-22 15:17:35 -07:00
[NET]: Fix fib_rules compatibility breakage
2007-03-25 18:48:00 -07:00
[NET]: Rethink mark field in struct flowi
2006-12-02 21:21:39 -08:00
[GENETLINK]: Add cmd dump completion.
2006-12-02 21:32:09 -08:00
[IPV4]: icmp_send() annotation
2006-09-28 18:01:06 -07:00
[PATCH] ieee80211: WLAN_GET_SEQ_SEQ fix (select correct region)
2007-01-02 20:56:26 -05:00
WorkStruct: make allyesconfig
2006-11-22 14:57:56 +00:00
[IPV6]: Per-interface statistics support.
2006-12-02 21:22:08 -08:00
[TCP]: Restore SKB socket owner setting in tcp_transmit_skb().
2007-01-26 01:04:55 -08:00
[IPV6]: annotate inet6_hashtables
2006-12-02 21:21:10 -08:00
[TCP]: Restore SKB socket owner setting in tcp_transmit_skb().
2007-01-26 01:04:55 -08:00
[NET]: IP header modifier helpers annotations.
2006-12-02 21:23:40 -08:00
[NET]: change layout of ehash table
2007-02-08 14:16:46 -08:00
[IPV4]: annotate inet_lookup() and friends
2006-09-28 18:02:26 -07:00
[INET]: twcal_jiffie should be unsigned long, not int
2007-03-05 13:32:48 -08:00
[IPV4] inet_peer: Group together avl_left, avl_right, v4daddr to speedup lookups on some CPUS
2006-10-20 00:28:35 -07:00
[TCP]: Restore SKB socket owner setting in tcp_transmit_skb().
2007-01-26 01:04:55 -08:00
[IPV6]: Dumb typo in generic csum_ipv6_magic()
2006-12-22 11:12:07 -08:00
[IPV6]: Fix routing round-robin locking.
2007-03-25 18:48:05 -07:00
[IPV6]: Misc endianness annotations.
2006-12-02 21:22:52 -08:00
[IPV4] nl_fib_lookup: Rename fl_fwmark to fl_mark
2006-12-02 21:21:40 -08:00
[NET]: Rethink mark field in struct flowi
2006-12-02 21:21:39 -08:00
[NET]: ipvs checksum annotations.
2006-12-02 21:23:41 -08:00
[CRYPTO] users: Use crypto_comp and crypto_has_*
2006-09-21 11:46:22 +10:00
[NET]: ipconfig and nfsroot annotations
2006-12-02 21:21:09 -08:00
[IPV6] net/ipv6/sit.c: make 2 functions static
2006-12-02 21:26:15 -08:00
[IPV6]: __ipv6_addr_diff() annotations and cleanup.
2006-12-02 21:22:53 -08:00
[IPX]: Annotate and fix IPX checksum
2006-11-05 14:11:25 -08:00
[LLC]: anotations
2006-12-02 21:21:23 -08:00
[IPV6] MIP6: Add receiving mobility header functions through raw socket.
2006-09-22 15:07:01 -07:00
[IPV6]: Misc endianness annotations.
2006-12-02 21:22:52 -08:00
[NET]: Fix neighbour destructor handling.
2007-03-25 18:48:01 -07:00
Remove all inclusions of <linux/config.h>
2006-10-04 03:38:54 -04:00
NetLabel: convert to an extensibile/sparse category bitmap
2006-12-02 21:31:36 -08:00
[PATCH] severing skbuff.h -> mm.h
2006-12-04 02:00:34 -05:00
[PATCH] mark struct file_operations const 1
2007-02-12 09:48:44 -08:00
[IPv4]: FIB configuration using struct fib_config
2006-09-22 14:55:04 -07:00
[INET]: Change protocol field in struct inet_protosw to u16
2006-12-02 21:30:55 -08:00
[IPV6]: 'info' argument of ipv6 ->err_handler() is net-endian
2006-12-02 21:21:12 -08:00
[PATCH] slab: remove kmem_cache_t
2006-12-07 08:39:25 -08:00
[PATCH] mark struct file_operations const 1
2007-02-12 09:48:44 -08:00
[IPV4]: Convert ipv4 route to use the new dst_entry 'next' pointer
2007-02-10 23:20:38 -08:00
[NET_SCHED]: Fix endless loops caused by inaccurate qlen counters (part 1)
2006-12-02 21:31:42 -08:00
[SCTP]: Extend /proc/net/sctp/snmp to provide more statistics.
2006-09-22 14:55:16 -07:00
[NET]: Revert incorrect accept queue backlog changes.
2007-03-06 11:21:05 -08:00
[TCP]: remove tcp header from tcp_v4_check (take #2 )
2007-02-08 12:38:44 -08:00
[PATCH] slab: remove kmem_cache_t
2006-12-07 08:39:25 -08:00
[NET]: Supporting UDP-Lite (RFC 3828) in Linux
2006-12-02 21:22:46 -08:00
[PATCH] severing skbuff.h -> poll.h
2006-12-04 02:00:31 -05:00
[NET]: Fix assorted misannotations (from md5 and udplite merges).
2006-12-02 21:27:16 -08:00
[X.25]: Adds /proc/sys/net/x25/x25_forward to control forwarding.
2007-02-08 13:34:36 -08:00
[IPSEC]: xfrm_policy delete security check misplaced
2007-03-07 16:08:09 -08:00
f11e6659ce