1
0
Fork 0
alistair23-linux/net/ipv4
Eric Dumazet dbca9b2750 [NET]: change layout of ehash table
ehash table layout is currently this one :

First half of this table is used by sockets not in TIME_WAIT state
Second half of it is used by sockets in TIME_WAIT state.

This is non optimal because of for a given hash or socket, the two chain heads 
are located in separate cache lines.
Moreover the locks of the second half are never used.

If instead of this halving, we use two list heads in inet_ehash_bucket instead 
of only one, we probably can avoid one cache miss, and reduce ram usage, 
particularly if sizeof(rwlock_t) is big (various CONFIG_DEBUG_SPINLOCK, 
CONFIG_DEBUG_LOCK_ALLOC settings). So we still halves the table but we keep 
together related chains to speedup lookups and socket state change.

In this patch I did not try to align struct inet_ehash_bucket, but a future 
patch could try to make this structure have a convenient size (a power of two 
or a multiple of L1_CACHE_SIZE).
I guess rwlock will just vanish as soon as RCU is plugged into ehash :) , so 
maybe we dont need to scratch our heads to align the bucket...

Note : In case struct inet_ehash_bucket is not a power of two, we could 
probably change alloc_large_system_hash() (in case it use __get_free_pages()) 
to free the unused space. It currently allocates a big zone, but the last 
quarter of it could be freed. Again, this should be a temporary 'problem'.

Patch tested on ipv4 tcp only, but should be OK for IPV6 and DCCP.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-08 14:16:46 -08:00
..
ipvs [IPVS]: Make ip_vs_sync.c <= 80col wide. 2006-12-11 14:35:03 -08:00
netfilter [NETFILTER]: {ip,ip6}_tables: use struct xt_table instead of redefined structure names 2007-02-08 12:39:20 -08:00
Kconfig [TCP] MD5SIG: Kill CONFIG_TCP_MD5SIG_DEBUG. 2006-12-02 21:31:47 -08:00
Makefile [NET]: Supporting UDP-Lite (RFC 3828) in Linux 2006-12-02 21:22:46 -08:00
af_inet.c [IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts. 2007-02-08 12:38:45 -08:00
ah4.c [IPV4]: encapsulation annotations 2006-12-02 21:21:17 -08:00
arp.c [IPV6]: Assorted trivial endianness annotations. 2006-12-02 21:22:50 -08:00
cipso_ipv4.c NetLabel: add the ranged tag to the CIPSOv4 protocol 2006-12-02 21:31:38 -08:00
datagram.c [IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts. 2007-02-08 12:38:45 -08:00
devinet.c [NETLINK]: Don't BUG on undersized allocations 2007-02-08 12:38:41 -08:00
esp4.c [IPV4]: encapsulation annotations 2006-12-02 21:21:17 -08:00
fib_frontend.c [NETLINK]: Remove unused dst_pid field in netlink_skb_parms 2006-12-02 21:30:43 -08:00
fib_hash.c [PATCH] slab: remove kmem_cache_t 2006-12-07 08:39:25 -08:00
fib_lookup.h [IPV4]: net/ipv4/fib annotations 2006-09-28 18:02:23 -07:00
fib_rules.c [NETLINK]: Do precise netlink message allocations where possible 2006-12-02 21:22:11 -08:00
fib_semantics.c [NETLINK]: Don't BUG on undersized allocations 2007-02-08 12:38:41 -08:00
fib_trie.c [IPV4]: Fix single-entry /proc/net/fib_trie output. 2007-01-26 19:06:01 -08:00
icmp.c [NET]: Annotate callers of the reset of checksum.h stuff. 2006-12-02 21:23:34 -08:00
igmp.c [IPV4/IPV6] multicast: Check add_grhead() return value 2007-02-08 12:39:04 -08:00
inet_connection_sock.c [NET]: Size listen hash tables using backlog hint 2006-12-02 21:21:44 -08:00
inet_diag.c [NET]: change layout of ehash table 2007-02-08 14:16:46 -08:00
inet_hashtables.c [NET]: change layout of ehash table 2007-02-08 14:16:46 -08:00
inet_timewait_sock.c [NET]: change layout of ehash table 2007-02-08 14:16:46 -08:00
inetpeer.c [PATCH] slab: remove kmem_cache_t 2006-12-07 08:39:25 -08:00
ip_forward.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
ip_fragment.c fix typo in net/ipv4/ip_fragment.c 2006-12-12 19:48:59 +01:00
ip_gre.c [NET]: unregister_netdevice as void 2007-02-08 12:39:06 -08:00
ip_input.c [IPV4]: Clear the whole IPCB, this clears also IPCB(skb)->flags. 2006-07-24 23:45:16 -07:00
ip_options.c [NetLabel]: protect the CIPSOv4 socket option from setsockopt() 2006-10-30 15:24:49 -08:00
ip_output.c [TCP]: Restore SKB socket owner setting in tcp_transmit_skb(). 2007-01-26 01:04:55 -08:00
ip_sockglue.c [IPV6]: Assorted trivial endianness annotations. 2006-12-02 21:22:50 -08:00
ipcomp.c [XFRM]: BEET mode 2006-10-04 00:31:09 -07:00
ipconfig.c [NET]: ipconfig and nfsroot annotations 2006-12-02 21:21:09 -08:00
ipip.c [NET]: unregister_netdevice as void 2007-02-08 12:39:06 -08:00
ipmr.c [PATCH] slab: remove kmem_cache_t 2006-12-07 08:39:25 -08:00
multipath.c Linux-2.6.12-rc2 2005-04-16 15:20:36 -07:00
multipath_drr.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
multipath_random.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
multipath_rr.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
multipath_wrandom.c [IPV4] bug: broken open-coded inet_make_mask() (multipath_wrandom) 2006-09-28 18:01:17 -07:00
netfilter.c [NETFILTER]: Fix routing of REJECT target generated packets in output chain 2007-01-04 12:15:34 -08:00
proc.c [NET]: Supporting UDP-Lite (RFC 3828) in Linux 2006-12-02 21:22:46 -08:00
protocol.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
raw.c [IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts. 2007-02-08 12:38:45 -08:00
route.c [NETLINK]: Don't BUG on undersized allocations 2007-02-08 12:38:41 -08:00
syncookies.c [IPV6]: Assorted trivial endianness annotations. 2006-12-02 21:22:50 -08:00
sysctl_net_ipv4.c [PATCH] sysctl: remove unused "context" param 2006-12-10 09:55:41 -08:00
tcp.c [NET]: change layout of ehash table 2007-02-08 14:16:46 -08:00
tcp_bic.c [TCP] tcp_bic: use BUILD_BUG_ON 2006-09-22 15:18:04 -07:00
tcp_cong.c [TCP]: Allow autoloading of congestion control via setsockopt. 2006-12-02 21:21:50 -08:00
tcp_cubic.c [TCP] cubic: scaling error 2006-10-25 23:04:12 -07:00
tcp_diag.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
tcp_highspeed.c [TCP] Congestion control (modulo lp, bic): use BUILD_BUG_ON 2006-09-22 15:18:13 -07:00
tcp_htcp.c [TCP] htcp: Better packing of struct htcp. 2006-12-02 21:22:14 -08:00
tcp_hybla.c [TCP] Congestion control (modulo lp, bic): use BUILD_BUG_ON 2006-09-22 15:18:13 -07:00
tcp_input.c [TCP]: Check num sacks in SACK fast path 2007-02-08 12:38:50 -08:00
tcp_ipv4.c [NET]: change layout of ehash table 2007-02-08 14:16:46 -08:00
tcp_lp.c [TCP] tcp-lp: prevent chance for oops 2006-09-28 18:03:07 -07:00
tcp_minisocks.c Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6 2006-12-05 14:37:56 +00:00
tcp_output.c [TCP]: Don't apply FIN exception to full TSO segments. 2007-02-08 12:38:51 -08:00
tcp_probe.c [PATCH] email change for shemminger@osdl.org 2007-01-23 14:18:49 -08:00
tcp_scalable.c Remove obsolete #include <linux/config.h> 2006-06-30 19:25:36 +02:00
tcp_timer.c [IPV6]: Assorted trivial endianness annotations. 2006-12-02 21:22:50 -08:00
tcp_vegas.c [TCP] Vegas: Increase default alpha to 2 and beta to 4. 2006-12-02 21:31:03 -08:00
tcp_veno.c [TCP] Congestion control (modulo lp, bic): use BUILD_BUG_ON 2006-09-22 15:18:13 -07:00
tcp_westwood.c [TCP] Congestion control (modulo lp, bic): use BUILD_BUG_ON 2006-09-22 15:18:13 -07:00
tunnel4.c [INET]: Move no-tunnel ICMP error to tunnel4/tunnel6 2006-04-09 22:25:25 -07:00
udp.c [IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts. 2007-02-08 12:38:45 -08:00
udp_impl.h [NET]: Supporting UDP-Lite (RFC 3828) in Linux 2006-12-02 21:22:46 -08:00
udplite.c [NET]: Possible cleanups. 2006-12-02 21:31:51 -08:00
xfrm4_input.c [XFRM]: xfrm_parse_spi() annotations 2006-09-28 18:02:39 -07:00
xfrm4_mode_beet.c [XFRM]: BEET mode 2006-10-04 00:31:09 -07:00
xfrm4_mode_transport.c [IPSEC]: output mode to take an xfrm state as input param 2006-09-22 15:18:48 -07:00
xfrm4_mode_tunnel.c [IPSEC]: IPv6 over IPv4 IPsec tunnel 2007-02-08 12:39:01 -08:00
xfrm4_output.c [IPSEC]: output mode to take an xfrm state as input param 2006-09-22 15:18:48 -07:00
xfrm4_policy.c [XFRM]: Fix missed error setting in xfrm4_policy.c 2007-02-08 12:39:03 -08:00
xfrm4_state.c [IPSEC]: exporting xfrm_state_afinfo 2007-02-08 12:39:00 -08:00
xfrm4_tunnel.c [XFRM]: Add XFRM_MODE_xxx for future use. 2006-09-22 15:05:15 -07:00