Now TCP & DCCP use RCU lookups, we can convert ehash rwlocks to spinlocks.
/proc/net/tcp and other seq_file 'readers' can safely be converted to 'writers'.
This should speedup writers, since spin_lock()/spin_unlock()
only use one atomic operation instead of two for write_lock()/write_unlock()
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
RCU was added to UDP lookups, using a fast infrastructure :
- sockets kmem_cache use SLAB_DESTROY_BY_RCU and dont pay the
price of call_rcu() at freeing time.
- hlist_nulls permits to use few memory barriers.
This patch uses same infrastructure for TCP/DCCP established
and timewait sockets.
Thanks to SLAB_DESTROY_BY_RCU, no slowdown for applications
using short lived TCP connections. A followup patch, converting
rwlocks to spinlocks will even speedup this case.
__inet_lookup_established() is pretty fast now we dont have to
dirty a contended cache line (read_lock/read_unlock)
Only established and timewait hashtable are converted to RCU
(bind table and listen table are still using traditional locking)
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces the IP_TRANSPARENT socket option: enabling that
will make the IPv4 routing omit the non-local source address check on
output. Setting IP_TRANSPARENT requires NET_ADMIN capability.
Signed-off-by: KOVACS Krisztian <hidden@sch.bme.hu>
Signed-off-by: David S. Miller <davem@davemloft.net>
How to reproduce ?
- create a network namespace
- use tcp protocol and get timewait socket
- exit the network namespace
- after a moment (when the timewait socket is destroyed), the kernel
panics.
# BUG: unable to handle kernel NULL pointer dereference at
0000000000000007
IP: [<ffffffff821e394d>] inet_twdr_do_twkill_work+0x6e/0xb8
PGD 119985067 PUD 11c5c0067 PMD 0
Oops: 0000 [1] SMP
CPU 1
Modules linked in: ipv6 button battery ac loop dm_mod tg3 libphy ext3 jbd
edd fan thermal processor thermal_sys sg sata_svw libata dock serverworks
sd_mod scsi_mod ide_disk ide_core [last unloaded: freq_table]
Pid: 0, comm: swapper Not tainted 2.6.27-rc2 #3
RIP: 0010:[<ffffffff821e394d>] [<ffffffff821e394d>]
inet_twdr_do_twkill_work+0x6e/0xb8
RSP: 0018:ffff88011ff7fed0 EFLAGS: 00010246
RAX: ffffffffffffffff RBX: ffffffff82339420 RCX: ffff88011ff7ff30
RDX: 0000000000000001 RSI: ffff88011a4d03c0 RDI: ffff88011ac2fc00
RBP: ffffffff823392e0 R08: 0000000000000000 R09: ffff88002802a200
R10: ffff8800a5c4b000 R11: ffffffff823e4080 R12: ffff88011ac2fc00
R13: 0000000000000001 R14: 0000000000000001 R15: 0000000000000000
FS: 0000000041cbd940(0000) GS:ffff8800bff839c0(0000)
knlGS:0000000000000000
CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b
CR2: 0000000000000007 CR3: 00000000bd87c000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process swapper (pid: 0, threadinfo ffff8800bff9e000, task
ffff88011ff76690)
Stack: ffffffff823392e0 0000000000000100 ffffffff821e3a3a
0000000000000008
0000000000000000 ffffffff821e3a61 ffff8800bff7c000 ffffffff8203c7e7
ffff88011ff7ff10 ffff88011ff7ff10 0000000000000021 ffffffff82351108
Call Trace:
<IRQ> [<ffffffff821e3a3a>] ? inet_twdr_hangman+0x0/0x9e
[<ffffffff821e3a61>] ? inet_twdr_hangman+0x27/0x9e
[<ffffffff8203c7e7>] ? run_timer_softirq+0x12c/0x193
[<ffffffff820390d1>] ? __do_softirq+0x5e/0xcd
[<ffffffff8200d08c>] ? call_softirq+0x1c/0x28
[<ffffffff8200e611>] ? do_softirq+0x2c/0x68
[<ffffffff8201a055>] ? smp_apic_timer_interrupt+0x8e/0xa9
[<ffffffff8200cad6>] ? apic_timer_interrupt+0x66/0x70
<EOI> [<ffffffff82011f4c>] ? default_idle+0x27/0x3b
[<ffffffff8200abbd>] ? cpu_idle+0x5f/0x7d
Code: e8 01 00 00 4c 89 e7 41 ff c5 e8 8d fd ff ff 49 8b 44 24 38 4c 89 e7
65 8b 14 25 24 00 00 00 89 d2 48 8b 80 e8 00 00 00 48 f7 d0 <48> 8b 04 d0
48 ff 40 58 e8 fc fc ff ff 48 89 df e8 c0 5f 04 00
RIP [<ffffffff821e394d>] inet_twdr_do_twkill_work+0x6e/0xb8
RSP <ffff88011ff7fed0>
CR2: 0000000000000007
This patch provides a function to purge all timewait sockets related
to a network namespace. The timewait sockets life cycle is not tied with
the network namespace, that means the timewait sockets stay alive while
the network namespace dies. The timewait sockets are for avoiding to
receive a duplicate packet from the network, if the network namespace is
freed, the network stack is removed, so no chance to receive any packets
from the outside world. Furthermore, having a pending destruction timer
on these sockets with a network namespace freed is not safe and will lead
to an oops if the timer callback which try to access data belonging to
the namespace like for example in:
inet_twdr_do_twkill_work
-> NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
Purging the timewait sockets at the network namespace destruction will:
1) speed up memory freeing for the namespace
2) fix kernel panic on asynchronous timewait destruction
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Acked-by: Denis V. Lunev <den@openvz.org>
Acked-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Removes legacy reinvent-the-wheel type thing. The generic
machinery integrates much better to automated debugging aids
such as kerneloops.org (and others), and is unambiguous due to
better naming. Non-intuively BUG_TRAP() is actually equal to
WARN_ON() rather than BUG_ON() though some might actually be
promoted to BUG_ON() but I left that to future.
I could make at least one BUILD_BUG_ON conversion.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This one is tricky.
The thing is that this macro is only used when killing tw buckets,
but since this killer is promiscuous wrt to which net each particular
tw belongs to, I have to use it only when NET_NS is off. When the net
namespaces are on, I use the INET_INC_STATS_BH for each bucket.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Binding to some port in many namespaces may create too long
chains in bhash-es, so prepare the hashfn to take struct net
into account.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This counter is about to become per-proto-and-per-net, so we'll need
two arguments to determine which cell in this "table" to work with.
All the places, but proc already pass proper net to it - proc will be
tuned a bit later.
Some indentation with spaces in proc files is done to keep the file
coding style consistent.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduce per-sock inlines: sock_net(), sock_net_set()
and per-inet_timewait_sock inlines: twsk_net(), twsk_net_set().
Without CONFIG_NET_NS, no namespace other than &init_net exists.
Let's explicitly define them to help compiler optimizations.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Copy the network namespace from the socket to the timewait socket.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1) Cleanups (all functions are prefixed by sock_prot_inuse)
sock_prot_inc_use(prot) -> sock_prot_inuse_add(prot,-1)
sock_prot_dec_use(prot) -> sock_prot_inuse_add(prot,-1)
sock_prot_inuse() -> sock_prot_inuse_get()
New functions :
sock_prot_inuse_init() and sock_prot_inuse_free() to abstract pcounter use.
2) if CONFIG_PROC_FS=n, we can zap 'inuse' member from "struct proto",
since nobody wants to read the inuse value.
This saves 1372 bytes on i386/SMP and some cpu cycles.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make the INET_TWDR_TWKILL_SLOTS vs sizeof(twdr->thread_slots)
check nicer.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
As done two years ago on IP route cache table (commit
22c047ccbc) , we can avoid using one
lock per hash bucket for the huge TCP/DCCP hash tables.
On a typical x86_64 platform, this saves about 2MB or 4MB of ram, for
litle performance differences. (we hit a different cache line for the
rwlock, but then the bucket cache line have a better sharing factor
among cpus, since we dirty it less often). For netstat or ss commands
that want a full scan of hash table, we perform fewer memory accesses.
Using a 'small' table of hashed rwlocks should be more than enough to
provide correct SMP concurrency between different buckets, without
using too much memory. Sizing of this table depends on
num_possible_cpus() and various CONFIG settings.
This patch provides some locking abstraction that may ease a future
work using a different model for TCP/DCCP table.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Hopefully captured all single statement cases under net/. I'm
not too sure if there is some policy about #includes that are
"guaranteed" (ie., in the current tree) to be available through
some other #included header, so I just added linux/kernel.h to
each changed file that didn't #include it previously.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch makes the needlessly global __inet_twsk_kill() static.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
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>
SLAB_ATOMIC is an alias of GFP_ATOMIC
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
As per Ralf Baechle's observations, the schedule_work() call
should give enough of a memory barrier, so the explicit one
here is totally unnecessary.
Signed-off-by: David S. Miller <davem@davemloft.net>
I believe all the below memory barriers only matter on SMP so
therefore the smp_* variant of the barrier should be used.
I'm wondering if the barrier in net/ipv4/inet_timewait_sock.c should be
dropped entirely. schedule_work's implementation currently implies a
memory barrier and I think sane semantics of schedule_work() should imply
a memory barrier, as needed so the caller shouldn't have to worry.
It's not quite obvious why the barrier in net/packet/af_packet.c is
needed; maybe it should be implied through flush_dcache_page?
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pass the work_struct pointer to the work function rather than context data.
The work function can use container_of() to work out the data.
For the cases where the container of the work_struct may go away the moment the
pending bit is cleared, it is made possible to defer the release of the
structure by deferring the clearing of the pending bit.
To make this work, an extra flag is introduced into the management side of the
work_struct. This governs auto-release of the structure upon execution.
Ordinarily, the work queue executor would release the work_struct for further
scheduling or deallocation by clearing the pending bit prior to jumping to the
work function. This means that, unless the driver makes some guarantee itself
that the work_struct won't go away, the work function may not access anything
else in the work_struct or its container lest they be deallocated.. This is a
problem if the auxiliary data is taken away (as done by the last patch).
However, if the pending bit is *not* cleared before jumping to the work
function, then the work function *may* access the work_struct and its container
with no problems. But then the work function must itself release the
work_struct by calling work_release().
In most cases, automatic release is fine, so this is the default. Special
initiators exist for the non-auto-release case (ending in _NAR).
Signed-Off-By: David Howells <dhowells@redhat.com>
So that we can share several timewait sockets related functions and
make the timewait mini sockets infrastructure closer to the request
mini sockets one.
Next changesets will take advantage of this, moving more code out of
TCP and DCCP v4 and v6 to common infrastructure.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is required to avoid unloading a module that has active timewait
sockets, such as DCCP.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Arnaldo and I agreed it could be applied now, because I have other
pending patches depending on this one (Thank you Arnaldo)
(The other important patch moves skc_refcnt in a separate cache line,
so that the SMP/NUMA performance doesnt suffer from cache line ping pongs)
1) First some performance data :
--------------------------------
tcp_v4_rcv() wastes a *lot* of time in __inet_lookup_established()
The most time critical code is :
sk_for_each(sk, node, &head->chain) {
if (INET_MATCH(sk, acookie, saddr, daddr, ports, dif))
goto hit; /* You sunk my battleship! */
}
The sk_for_each() does use prefetch() hints but only the begining of
"struct sock" is prefetched.
As INET_MATCH first comparison uses inet_sk(__sk)->daddr, wich is far
away from the begining of "struct sock", it has to bring into CPU
cache cold cache line. Each iteration has to use at least 2 cache
lines.
This can be problematic if some chains are very long.
2) The goal
-----------
The idea I had is to change things so that INET_MATCH() may return
FALSE in 99% of cases only using the data already in the CPU cache,
using one cache line per iteration.
3) Description of the patch
---------------------------
Adds a new 'unsigned int skc_hash' field in 'struct sock_common',
filling a 32 bits hole on 64 bits platform.
struct sock_common {
unsigned short skc_family;
volatile unsigned char skc_state;
unsigned char skc_reuse;
int skc_bound_dev_if;
struct hlist_node skc_node;
struct hlist_node skc_bind_node;
atomic_t skc_refcnt;
+ unsigned int skc_hash;
struct proto *skc_prot;
};
Store in this 32 bits field the full hash, not masked by (ehash_size -
1) Using this full hash as the first comparison done in INET_MATCH
permits us immediatly skip the element without touching a second cache
line in case of a miss.
Suppress the sk_hashent/tw_hashent fields since skc_hash (aliased to
sk_hash and tw_hash) already contains the slot number if we mask with
(ehash_size - 1)
File include/net/inet_hashtables.h
64 bits platforms :
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash))
((*((__u64 *)&(inet_sk(__sk)->daddr)))== (__cookie)) && \
((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
32bits platforms:
#define TCP_IPV4_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash)) && \
(inet_sk(__sk)->daddr == (__saddr)) && \
(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
- Adds a prefetch(head->chain.first) in
__inet_lookup_established()/__tcp_v4_check_established() and
__inet6_lookup_established()/__tcp_v6_check_established() and
__dccp_v4_check_established() to bring into cache the first element of the
list, before the {read|write}_lock(&head->lock);
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Also export the ones that will be used in the next changeset, when
DCCP uses this infrastructure.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This creates struct inet_connection_sock, moving members out of struct
tcp_sock that are shareable with other INET connection oriented
protocols, such as DCCP, that in my private tree already uses most of
these members.
The functions that operate on these members were renamed, using a
inet_csk_ prefix while not being moved yet to a new file, so as to
ease the review of these changes.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the parts of tcp_time_wait that are not TCP specific, tcp_time_wait uses
it and so will dccp_time_wait.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
And also some TIME_WAIT functions.
[acme@toy net-2.6.14]$ grep built-in /tmp/before.size /tmp/after.size
/tmp/before.size: 282955 13122 9312 305389 4a8ed net/ipv4/built-in.o
/tmp/after.size: 281566 13122 9312 304000 4a380 net/ipv4/built-in.o
[acme@toy net-2.6.14]$
I kept them still inlined, will uninline at some point to see what
would be the performance difference.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet