eth_type_trans() & get_rps_cpus() currently need two 64bytes cache
lines in packet to compute rxhash.
Increasing NET_SKB_PAD from 32 to 64 reduces the need to one cache
line only, and makes RPS faster.
NET_IP_ALIGN(2) + ethernet_header(14) + IP_header(20/40) + ports(8)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This whole patchset is for adding netpoll support to bridge and bonding
devices. I already tested it for bridge, bonding, bridge over bonding,
and bonding over bridge. It looks fine now.
To make bridge and bonding support netpoll, we need to adjust
some netpoll generic code. This patch does the following things:
1) introduce two new priv_flags for struct net_device:
IFF_IN_NETPOLL which identifies we are processing a netpoll;
IFF_DISABLE_NETPOLL is used to disable netpoll support for a device
at run-time;
2) introduce one new method for netdev_ops:
->ndo_netpoll_cleanup() is used to clean up netpoll when a device is
removed.
3) introduce netpoll_poll_dev() which takes a struct net_device * parameter;
export netpoll_send_skb() and netpoll_poll_dev() which will be used later;
4) hide a pointer to struct netpoll in struct netpoll_info, ditto.
5) introduce ->real_dev for struct netpoll.
6) introduce a new status NETDEV_BONDING_DESLAE, which is used to disable
netconsole before releasing a slave, to avoid deadlocks.
Cc: David Miller <davem@davemloft.net>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With following patch I can reach maximum rate of my pktgen+udpsink
simulator :
- 'old' machine : dual quad core E5450 @3.00GHz
- 64 UDP rx flows (only differ by destination port)
- RPS enabled, NIC interrupts serviced on cpu0
- rps dispatched on 7 other cores. (~130.000 IPI per second)
- SLAB allocator (faster than SLUB in this workload)
- tg3 NIC
- 1.080.000 pps without a single drop at NIC level.
Idea is to add two prefetchw() calls in __alloc_skb(), one to prefetch
first sk_buff cache line, the second to prefetch the shinfo part.
Also using one memset() to initialize all skb_shared_info fields instead
of one by one to reduce number of instructions, using long word moves.
All skb_shared_info fields before 'dataref' are cleared in
__alloc_skb().
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add hlist_for_each_entry_rcu_bh() and
hlist_for_each_entry_continue_rcu_bh() macros, and use them in
ipv6_get_ifaddr(), if6_get_first() and if6_get_next() to fix lockdeps
warnings.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Reviewed-by: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
virtio added mergeable buffers mode where 2 bytes of extra info is put
after vnet header but before actual data (tun does not need this data).
In hindsight, it would have been better to add the new info *before* the
packet: as it is, users need a lot of tricky code to skip the extra 2
bytes in the middle of the iovec, and in fact applications seem to get
it wrong, and only work with specific iovec layout. The fact we might
need to split iovec also means we might in theory overflow iovec max
size.
This patch adds a simpler way for applications to handle this,
and future proofs the interface against further extensions,
by making the size of the virtio net header configurable
from userspace. As a result, tun driver will simply
skip the extra 2 bytes on both input and output.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: David S. Miller <davem@davemloft.net>
Per cpu variable softnet_data.total was shared between IRQ and SoftIRQ context
without any protection. And enqueue_to_backlog should update the netdev_rx_stat
of the target CPU.
This patch renames softnet_data.total to softnet_data.processed: the number of
packets processed in uppper levels(IP stacks).
softnet_stat data is moved into softnet_data.
Signed-off-by: Changli Gao <xiaosuo@gmail.com>
----
include/linux/netdevice.h | 17 +++++++----------
net/core/dev.c | 26 ++++++++++++--------------
net/sched/sch_generic.c | 2 +-
3 files changed, 20 insertions(+), 25 deletions(-)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix this one:
include/net/sock.h: error: two or more data types in declaration specifiers
Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit 6be8ac2f ("[NET]: uninline skb_pull, de-bloats a lot")
we uninlined skb_pull.
But in some critical paths it makes sense to inline this thing
and it helps performance significantly.
Create an skb_pull_inline() so that we can do this in a way that
serves also as annotation.
Based upon a patch by Eric Dumazet.
Signed-off-by: David S. Miller <davem@davemloft.net>
sk_callback_lock rwlock actually protects sk->sk_sleep pointer, so we
need two atomic operations (and associated dirtying) per incoming
packet.
RCU conversion is pretty much needed :
1) Add a new structure, called "struct socket_wq" to hold all fields
that will need rcu_read_lock() protection (currently: a
wait_queue_head_t and a struct fasync_struct pointer).
[Future patch will add a list anchor for wakeup coalescing]
2) Attach one of such structure to each "struct socket" created in
sock_alloc_inode().
3) Respect RCU grace period when freeing a "struct socket_wq"
4) Change sk_sleep pointer in "struct sock" by sk_wq, pointer to "struct
socket_wq"
5) Change sk_sleep() function to use new sk->sk_wq instead of
sk->sk_sleep
6) Change sk_has_sleeper() to wq_has_sleeper() that must be used inside
a rcu_read_lock() section.
7) Change all sk_has_sleeper() callers to :
- Use rcu_read_lock() instead of read_lock(&sk->sk_callback_lock)
- Use wq_has_sleeper() to eventually wakeup tasks.
- Use rcu_read_unlock() instead of read_unlock(&sk->sk_callback_lock)
8) sock_wake_async() is modified to use rcu protection as well.
9) Exceptions :
macvtap, drivers/net/tun.c, af_unix use integrated "struct socket_wq"
instead of dynamically allocated ones. They dont need rcu freeing.
Some cleanups or followups are probably needed, (possible
sk_callback_lock conversion to a spinlock for example...).
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we create the sctp_datamsg and fragment the user data,
we know exactly if we are sending full segments or not and
how they might be bundled. During this time, we can mark
messages a Nagle capable or not. This makes the check at
transmit time much simpler.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Right now, sctp transports are not fully initialized and when
adding any new fields, they have to be explicitely initialized.
This is prone to mistakes. So we switch to calling kzalloc()
which makes things much simpler.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
The 'resent' bit is used to make sure that we don't update
rto estimate based on retransmitted chunks. However, we already
have the 'rto_pending' bit that we test when need to update rto,
so 'resent' bit is just extra. Additionally, we currently have
a bug in that we always set a 'resent' bit and thus rto estimate
is only updated by Heartbeats.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
In current implementation if ABORT chunk is received with T flag is set
and zero verification tag in COOKIE-WAIT state, the ABORT chunk will be
always accepted. This is because in COOKIE-WAIT state, the endpoint does
not know the peer's verification tag, and it's zero in the endpoint.
Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
sock_recv_ts_and_drops() is fat and slow (~ 4% of cpu time on some
profiles)
We can test all socket flags at once to make fast path fast again.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When queueing a skb to socket, we can immediately release its dst if
target socket do not use IP_CMSG_PKTINFO.
tcp_data_queue() can drop dst too.
This to benefit from a hot cache line and avoid the receiver, possibly
on another cpu, to dirty this cache line himself.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit 95766fff ([UDP]: Add memory accounting.),
each received packet needs one extra sock_lock()/sock_release() pair.
This added latency because of possible backlog handling. Then later,
ticket spinlocks added yet another latency source in case of DDOS.
This patch introduces lock_sock_bh() and unlock_sock_bh()
synchronization primitives, avoiding one atomic operation and backlog
processing.
skb_free_datagram_locked() uses them instead of full blown
lock_sock()/release_sock(). skb is orphaned inside locked section for
proper socket memory reclaim, and finally freed outside of it.
UDP receive path now take the socket spinlock only once.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ok, version 4
Change Notes:
1) Minor cleanups, from Vlads notes
Summary:
Hey-
Recently, it was reported to me that the kernel could oops in the
following way:
<5> kernel BUG at net/core/skbuff.c:91!
<5> invalid operand: 0000 [#1]
<5> Modules linked in: sctp netconsole nls_utf8 autofs4 sunrpc iptable_filter
ip_tables cpufreq_powersave parport_pc lp parport vmblock(U) vsock(U) vmci(U)
vmxnet(U) vmmemctl(U) vmhgfs(U) acpiphp dm_mirror dm_mod button battery ac md5
ipv6 uhci_hcd ehci_hcd snd_ens1371 snd_rawmidi snd_seq_device snd_pcm_oss
snd_mixer_oss snd_pcm snd_timer snd_page_alloc snd_ac97_codec snd soundcore
pcnet32 mii floppy ext3 jbd ata_piix libata mptscsih mptsas mptspi mptscsi
mptbase sd_mod scsi_mod
<5> CPU: 0
<5> EIP: 0060:[<c02bff27>] Not tainted VLI
<5> EFLAGS: 00010216 (2.6.9-89.0.25.EL)
<5> EIP is at skb_over_panic+0x1f/0x2d
<5> eax: 0000002c ebx: c033f461 ecx: c0357d96 edx: c040fd44
<5> esi: c033f461 edi: df653280 ebp: 00000000 esp: c040fd40
<5> ds: 007b es: 007b ss: 0068
<5> Process swapper (pid: 0, threadinfo=c040f000 task=c0370be0)
<5> Stack: c0357d96 e0c29478 00000084 00000004 c033f461 df653280 d7883180
e0c2947d
<5> 00000000 00000080 df653490 00000004 de4f1ac0 de4f1ac0 00000004
df653490
<5> 00000001 e0c2877a 08000800 de4f1ac0 df653490 00000000 e0c29d2e
00000004
<5> Call Trace:
<5> [<e0c29478>] sctp_addto_chunk+0xb0/0x128 [sctp]
<5> [<e0c2947d>] sctp_addto_chunk+0xb5/0x128 [sctp]
<5> [<e0c2877a>] sctp_init_cause+0x3f/0x47 [sctp]
<5> [<e0c29d2e>] sctp_process_unk_param+0xac/0xb8 [sctp]
<5> [<e0c29e90>] sctp_verify_init+0xcc/0x134 [sctp]
<5> [<e0c20322>] sctp_sf_do_5_1B_init+0x83/0x28e [sctp]
<5> [<e0c25333>] sctp_do_sm+0x41/0x77 [sctp]
<5> [<c01555a4>] cache_grow+0x140/0x233
<5> [<e0c26ba1>] sctp_endpoint_bh_rcv+0xc5/0x108 [sctp]
<5> [<e0c2b863>] sctp_inq_push+0xe/0x10 [sctp]
<5> [<e0c34600>] sctp_rcv+0x454/0x509 [sctp]
<5> [<e084e017>] ipt_hook+0x17/0x1c [iptable_filter]
<5> [<c02d005e>] nf_iterate+0x40/0x81
<5> [<c02e0bb9>] ip_local_deliver_finish+0x0/0x151
<5> [<c02e0c7f>] ip_local_deliver_finish+0xc6/0x151
<5> [<c02d0362>] nf_hook_slow+0x83/0xb5
<5> [<c02e0bb2>] ip_local_deliver+0x1a2/0x1a9
<5> [<c02e0bb9>] ip_local_deliver_finish+0x0/0x151
<5> [<c02e103e>] ip_rcv+0x334/0x3b4
<5> [<c02c66fd>] netif_receive_skb+0x320/0x35b
<5> [<e0a0928b>] init_stall_timer+0x67/0x6a [uhci_hcd]
<5> [<c02c67a4>] process_backlog+0x6c/0xd9
<5> [<c02c690f>] net_rx_action+0xfe/0x1f8
<5> [<c012a7b1>] __do_softirq+0x35/0x79
<5> [<c0107efb>] handle_IRQ_event+0x0/0x4f
<5> [<c01094de>] do_softirq+0x46/0x4d
Its an skb_over_panic BUG halt that results from processing an init chunk in
which too many of its variable length parameters are in some way malformed.
The problem is in sctp_process_unk_param:
if (NULL == *errp)
*errp = sctp_make_op_error_space(asoc, chunk,
ntohs(chunk->chunk_hdr->length));
if (*errp) {
sctp_init_cause(*errp, SCTP_ERROR_UNKNOWN_PARAM,
WORD_ROUND(ntohs(param.p->length)));
sctp_addto_chunk(*errp,
WORD_ROUND(ntohs(param.p->length)),
param.v);
When we allocate an error chunk, we assume that the worst case scenario requires
that we have chunk_hdr->length data allocated, which would be correct nominally,
given that we call sctp_addto_chunk for the violating parameter. Unfortunately,
we also, in sctp_init_cause insert a sctp_errhdr_t structure into the error
chunk, so the worst case situation in which all parameters are in violation
requires chunk_hdr->length+(sizeof(sctp_errhdr_t)*param_count) bytes of data.
The result of this error is that a deliberately malformed packet sent to a
listening host can cause a remote DOS, described in CVE-2010-1173:
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2010-1173
I've tested the below fix and confirmed that it fixes the issue. We move to a
strategy whereby we allocate a fixed size error chunk and ignore errors we don't
have space to report. Tested by me successfully
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some drivers (e.g. iwlwifi) need to know and try
to figure it out based on other things, but making
it explicit is definitely better.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Changes:
This is a complete re-write of the socket layer. Making the socket
implementation more aligned with the other socket layers and using more
of the support functions available in sock.c. Lots of code is copied
from af_unix (and some from af_irda).
Non-blocking mode should be working as well.
Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Changes:
o Function cfcnfg_disconn_adapt_layer is changed to do asynchronous
disconnect, not waiting for any response from the modem. Due to this
the function cfcnfg_linkdestroy_rsp does nothing anymore.
o Because disconnect may take down a connection before a connect response
is received the function cfcnfg_linkup_rsp is checking if the client is
still waiting for the response, if not a disconnect request is sent to
the modem.
o cfctrl is no longer keeping track of pending disconnect requests.
o Added function cfctrl_cancel_req, which is used for deleting a pending
connect request if disconnect is done before connect response is received.
o Removed unused function cfctrl_insert_req2
o Added better handling of connect reject from modem.
Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Changes:
o Added functions cfsrvl_get and cfsrvl_put.
o Added support release_client to use by socket and net device.
o Increase reference counting for in-flight packets from cfmuxl
Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Changes:
o Renamed cfcnfg_del_adapt_layer to cfcnfg_disconn_adapt_layer
o Fixed typo cfcfg to cfcnfg
o Renamed linkid to channel_id
o Updated documentation in caif_dev.h
o Minor formatting changes
Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sk->sk_data_ready() of sctp socket can be called from both BH and non-BH
contexts, but the default sk->sk_data_ready(), sock_def_readable(), can
not be used in this case. Therefore, we have to make a new function
sctp_data_ready() to grab sk->sk_data_ready() with BH disabling.
=========================================================
[ INFO: possible irq lock inversion dependency detected ]
2.6.33-rc6 #129
---------------------------------------------------------
sctp_darn/1517 just changed the state of lock:
(clock-AF_INET){++.?..}, at: [<c06aab60>] sock_def_readable+0x20/0x80
but this lock took another, SOFTIRQ-unsafe lock in the past:
(slock-AF_INET){+.-...}
and interrupts could create inverse lock ordering between them.
other info that might help us debug this:
1 lock held by sctp_darn/1517:
#0: (sk_lock-AF_INET){+.+.+.}, at: [<cdfe363d>] sctp_sendmsg+0x23d/0xc00 [sctp]
Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now there's no need to use this fuction directly because it's handled by
register_pernet_device. So to make this simple and easy to understand,
make this static to do not tempt potentional users.
Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Current socket backlog limit is not enough to really stop DDOS attacks,
because user thread spend many time to process a full backlog each
round, and user might crazy spin on socket lock.
We should add backlog size and receive_queue size (aka rmem_alloc) to
pace writers, and let user run without being slow down too much.
Introduce a sk_rcvqueues_full() helper, to avoid taking socket lock in
stress situations.
Under huge stress from a multiqueue/RPS enabled NIC, a single flow udp
receiver can now process ~200.000 pps (instead of ~100 pps before the
patch) on a 8 core machine.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
batch skb dequeueing from softnet input_pkt_queue to reduce potential lock
contention when RPS is enabled.
Note: in the worst case, the number of packets in a softnet_data may
be double of netdev_max_backlog.
Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Idea from Eric Dumazet.
As for placement inside of struct sock, I tried to choose a place
that otherwise has a 32-bit hole on 64-bit systems.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
reimplement softnet_data.output_queue as a FIFO queue to keep the
fairness among the qdiscs rescheduled.
Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
----
include/linux/netdevice.h | 1 +
net/core/dev.c | 22 ++++++++++++----------
2 files changed, 13 insertions(+), 10 deletions(-)
Signed-off-by: David S. Miller <davem@davemloft.net>
When scanning, it is somewhat important to scan
on the correct virtual interface. All drivers
that currently implement hw_scan only support a
single virtual interface, but that may change
and then we'd want to be ready.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Now that the mac80211 is choosing dynamic ps timeouts based on the ps-qos
network latency configuration, configure a default value of -1 as the dynamic
ps timeout in cfg80211. This value allows the mac80211 to determine the value
to be used.
Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Determine the dynamic PS timeout based on the configured ps-qos network
latency. For backwards wext compatibility, allow the dynamic PS timeout
configured by the cfg80211 to overrule the automatically determined value.
Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This is used to configure APs to not bridge traffic between connected stations.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This change corrects the fact that we were not reporting Gen2 link speeds
when we were in fact connected at Gen2 rates.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
__sk_dst_set() might be called while no state can be integrated in a
rcu_dereference_check() condition.
So use rcu_dereference_raw() to shutup lockdep warnings (if
CONFIG_PROVE_RCU is set)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
const qualifier on sock argument is misleading, since we can modify rxhash.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
RFC 1122 says the following:
...
Keep-alive packets MUST only be sent when no data or
acknowledgement packets have been received for the
connection within an interval.
...
The acknowledgement packet is reseting the keepalive
timer but the data packet isn't. This patch fixes it by
checking the timestamp of the last received data packet
too when the keepalive timer expires.
Signed-off-by: Flavio Leitner <fleitner@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Calls to twsk_net() are in some cases protected by reference counting
as an alternative to RCU protection. Cases covered by reference counts
include __inet_twsk_kill(), inet_twsk_free(), inet_twdr_do_twkill_work(),
inet_twdr_twcal_tick(), and tcp_timewait_state_process(). RCU is used
by inet_twsk_purge(). Locking is used by established_get_first()
and established_get_next(). Finally, __inet_twsk_hashdance() is an
initialization case.
It appears to be non-trivial to locate the appropriate locks and
reference counts from within twsk_net(), so used rcu_dereference_raw().
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
