TCP timestamping extends SO_TIMESTAMPING to bytestreams.
Bytestreams do not have a 1:1 relationship between send() buffers and
network packets. The feature interprets a send call on a bytestream as
a request for a timestamp for the last byte in that send() buffer.
The choice corresponds to a request for a timestamp when all bytes in
the buffer have been sent. That assumption depends on in-order kernel
transmission. This is the common case. That said, it is possible to
construct a traffic shaping tree that would result in reordering.
The guarantee is strong, then, but not ironclad.
This implementation supports send and sendpages (splice). GSO replaces
one large packet with multiple smaller packets. This patch also copies
the option into the correct smaller packet.
This patch does not yet support timestamping on data in an initial TCP
Fast Open SYN, because that takes a very different data path.
If ID generation in ee_data is enabled, bytestream timestamps return a
byte offset, instead of the packet counter for datagrams.
The implementation supports a single timestamp per packet. It silenty
replaces requests for previous timestamps. To avoid missing tstamps,
flush the tcp queue by disabling Nagle, cork and autocork. Missing
tstamps can be detected by offset when the ee_data ID is enabled.
Implementation details:
- On GSO, the timestamping code can be included in the main loop. I
moved it into its own loop to reduce the impact on the common case
to a single branch.
- To avoid leaking the absolute seqno to userspace, the offset
returned in ee_data must always be relative. It is an offset between
an skb and sk field. The first is always set (also for GSO & ACK).
The second must also never be uninitialized. Only allow the ID
option on sockets in the ESTABLISHED state, for which the seqno
is available. Never reset it to zero (instead, move it to the
current seqno when reenabling the option).
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Kernel transmit latency is often incurred in the packet scheduler.
Introduce a new timestamp on transmission just before entering the
scheduler. When data travels through multiple devices (bonding,
tunneling, ...) each device will export an individual timestamp.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Datagrams timestamped on transmission can coexist in the kernel stack
and be reordered in packet scheduling. When reading looped datagrams
from the socket error queue it is not always possible to unique
correlate looped data with original send() call (for application
level retransmits). Even if possible, it may be expensive and complex,
requiring packet inspection.
Introduce a data-independent ID mechanism to associate timestamps with
send calls. Pass an ID alongside the timestamp in field ee_data of
sock_extended_err.
The ID is a simple 32 bit unsigned int that is associated with the
socket and incremented on each send() call for which software tx
timestamp generation is enabled.
The feature is enabled only if SOF_TIMESTAMPING_OPT_ID is set, to
avoid changing ee_data for existing applications that expect it 0.
The counter is reset each time the flag is reenabled. Reenabling
does not change the ID of already submitted data. It is possible
to receive out of order IDs if the timestamp stream is not quiesced
first.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sk_flags is reaching its limit. New timestamping options will not fit.
Move all of them into a new field sk->sk_tsflags.
Added benefit is that this removes boilerplate code to convert between
SOF_TIMESTAMPING_.. and SOCK_TIMESTAMPING_.. in getsockopt/setsockopt.
SOCK_TIMESTAMPING_RX_SOFTWARE is also used to toggle the receive
timestamp logic (netstamp_needed). That can be simplified and this
last key removed, but will leave that for a separate patch.
Signed-off-by: Willem de Bruijn <willemb@google.com>
----
The u16 in sock can be moved into a 16-bit hole below sk_gso_max_segs,
though that scatters tstamp fields throughout the struct.
Signed-off-by: David S. Miller <davem@davemloft.net>
Applications that request kernel tx timestamps with SO_TIMESTAMPING
read timestamps as recvmsg() ancillary data. The response is defined
implicitly as timespec[3].
1) define struct scm_timestamping explicitly and
2) add support for new tstamp types. On tx, scm_timestamping always
accompanies a sock_extended_err. Define previously unused field
ee_info to signal the type of ts[0]. Introduce SCM_TSTAMP_SND to
define the existing behavior.
The reception path is not modified. On rx, no struct similar to
sock_extended_err is passed along with SCM_TIMESTAMPING.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These belong to the t4 msg header, will ensure there is no accidental code
duplication in the future
Signed-off-by: Anish Bhatt <anish@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit reduces spurious retransmits due to apparent SACK reneging
by only reacting to SACK reneging that persists for a short delay.
When a sequence space hole at snd_una is filled, some TCP receivers
send a series of ACKs as they apparently scan their out-of-order queue
and cumulatively ACK all the packets that have now been consecutiveyly
received. This is essentially misbehavior B in "Misbehaviors in TCP
SACK generation" ACM SIGCOMM Computer Communication Review, April
2011, so we suspect that this is from several common OSes (Windows
2000, Windows Server 2003, Windows XP). However, this issue has also
been seen in other cases, e.g. the netdev thread "TCP being hoodwinked
into spurious retransmissions by lack of timestamps?" from March 2014,
where the receiver was thought to be a BSD box.
Since snd_una would temporarily be adjacent to a previously SACKed
range in these scenarios, this receiver behavior triggered the Linux
SACK reneging code path in the sender. This led the sender to clear
the SACK scoreboard, enter CA_Loss, and spuriously retransmit
(potentially) every packet from the entire write queue at line rate
just a few milliseconds before the ACK for each packet arrives at the
sender.
To avoid such situations, now when a sender sees apparent reneging it
does not yet retransmit, but rather adjusts the RTO timer to give the
receiver a little time (max(RTT/2, 10ms)) to send us some more ACKs
that will restore sanity to the SACK scoreboard. If the reneging
persists until this RTO then, as before, we clear the SACK scoreboard
and enter CA_Loss.
A 10ms delay tolerates a receiver sending such a stream of ACKs at
56Kbit/sec. And to allow for receivers with slower or more congested
paths, we wait for at least RTT/2.
We validated the resulting max(RTT/2, 10ms) delay formula with a mix
of North American and South American Google web server traffic, and
found that for ACKs displaying transient reneging:
(1) 90% of inter-ACK delays were less than 10ms
(2) 99% of inter-ACK delays were less than RTT/2
In tests on Google web servers this commit reduced reneging events by
75%-90% (as measured by the TcpExtTCPSACKReneging counter), without
any measurable impact on latency for user HTTP and SPDY requests.
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Zoltan Kiss says:
====================
xen-netback: Changes around carrier handling
This series starts using carrier off as a way to purge packets when the guest is
not able (or willing) to receive them. It is a much faster way to get rid of
packets waiting for an overwhelmed guest.
The first patch changes current netback code where it relies currently on
netif_carrier_ok.
The second turns off the carrier if the guest times out on a queue, and only
turn it on again if that queue (or queues) resurrects.
====================
Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently when the guest is not able to receive more packets, qdisc layer starts
a timer, and when it goes off, qdisc is started again to deliver a packet again.
This is a very slow way to drain the queues, consumes unnecessary resources and
slows down other guests shutdown.
This patch change the behaviour by turning the carrier off when that timer
fires, so all the packets are freed up which were stucked waiting for that vif.
Instead of the rx_queue_purge bool it uses the VIF_STATUS_RX_PURGE_EVENT bit to
signal the thread that either the timeout happened or an RX interrupt arrived,
so the thread can check what it should do. It also disables NAPI, so the guest
can't transmit, but leaves the interrupts on, so it can resurrect.
Only the queues which brought down the interface can enable it again, the bit
QUEUE_STATUS_RX_STALLED makes sure of that.
Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Cc: netdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: xen-devel@lists.xenproject.org
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces a new state bit VIF_STATUS_CONNECTED to track whether the
vif is in a connected state. Using carrier will not work with the next patch
in this series, which aims to turn the carrier temporarily off if the guest
doesn't seem to be able to receive packets.
Signed-off-by: Zoltan Kiss <zoltan.kiss@citrix.com>
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Cc: netdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: xen-devel@lists.xenproject.org
v2:
- rename the bitshift type to "enum state_bit_shift" here, not in the next patch
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
net/6lowpan/iphc.c
Minor conflicts in iphc.c were changes overlapping with some
style cleanups.
John W. Linville says:
====================
Please pull this last(?) batch of wireless change intended for the
3.17 stream...
For the NFC bits, Samuel says:
"This is a rather quiet one, we have:
- A new driver from ST Microelectronics for their NCI ST21NFCB,
including device tree support.
- p2p support for the ST21NFCA driver
- A few fixes an enhancements for the NFC digital laye"
For the Atheros bits, Kalle says:
"Michal and Janusz did some important RX aggregation fixes, basically we
were missing RX reordering altogether. The 10.1 firmware doesn't support
Ad-Hoc mode and Michal fixed ath10k so that it doesn't advertise Ad-Hoc
support with that firmware. Also he implemented a workaround for a KVM
issue."
For the Bluetooth bits, Gustavo and Johan say:
"To quote Gustavo from his previous request:
'Some last minute fixes for -next. We have a fix for a use after free in
RFCOMM, another fix to an issue with ADV_DIRECT_IND and one for ADV_IND with
auto-connection handling. Last, we added support for reading the codec and
MWS setting for controllers that support these features.'
Additionally there are fixes to LE scanning, an update to conform to the 4.1
core specification as well as fixes for tracking the page scan state. All
of these fixes are important for 3.17."
And,
"We've got:
- 6lowpan fixes/cleanups
- A couple crash fixes, one for the Marvell HCI driver and another in LE SMP.
- Fix for an incorrect connected state check
- Fix for the bondable requirement during pairing (an issue which had
crept in because of using "pairable" when in fact the actual meaning
was "bondable" (these have different meanings in Bluetooth)"
Along with those are some late-breaking hardware support patches in
brcmfmac and b43 as well as a stray ath9k patch.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
With netlink_lookup() conversion to RCU, we need to use appropriate
rcu dereference in netlink_seq_socket_idx() & netlink_seq_next()
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: e341694e3e ("netlink: Convert netlink_lookup() to use RCU protected hash table")
Signed-off-by: David S. Miller <davem@davemloft.net>
tcpm_key is an array inside struct tcp_md5sig, there is no need to check it
against NULL.
Signed-off-by: Dmitry Popov <ixaphire@qrator.net>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Intel did some benchmarking on our network throughput when Linux on Hyper-V
is as used as a gateway. This fix gave us almost a 1 Gbps additional throughput
on about 5Gbps base throughput we hadi, prior to increasing the sendbuf size.
The sendbuf mechanism is a copy based transport that we have which is clearly
more optimal than the copy-free page flipping mechanism (for small packets).
In the forwarding scenario, we deal only with MTU sized packets,
and increasing the size of the senbuf area gave us the additional performance.
For what it is worth, Windows guests on Hyper-V, I am told use similar sendbuf
size as well.
The exact value of sendbuf I think is less important than the fact that it needs
to be larger than what Linux can allocate as physically contiguous memory.
Thus the change over to allocating via vmalloc().
We currently allocate 16MB receive buffer and we use vmalloc there for allocation.
Also the low level channel code has already been modified to deal with physically
dis-contiguous memory in the ringbuffer setup.
Based on experimentation Intel did, they say there was some improvement in throughput
as the sendbuf size was increased up to 16MB and there was no effect on throughput
beyond 16MB. Thus I have chosen 16MB here.
Increasing the sendbuf value makes a material difference in small packet handling
In this version of the patch, based on David's feedback, I have added
additional details in the commit log.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces the use of devm_kzalloc and does away with the
kfrees in the probe and remove functions. Also, a label is removed.
Signed-off-by: Himangi Saraogi <himangi774@gmail.com>
Acked-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
commit 6cbdceeb1c
bridge: Dump vlan information from a bridge port
introduced a comment in an attempt to explain the
code logic. The comment is unfinished so it confuses more
than it explains, remove it.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
As pointed out by the intel guys, there is no need to hold rcu read lock in
cxgbi_inet6addr_handler(), this patch removes it.
Fixes: 759a0cc5a3 ("cxgb4i: Add ipv6 code to driver, call into libcxgbi ipv6 api")
Signed-off-by: Anish Bhatt <anish@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Thomas Graf says:
====================
Lockless netlink_lookup() with new concurrent hash table
Netlink sockets are maintained in a hash table to allow efficient lookup
via the port ID for unicast messages. However, lookups currently require
a read lock to be taken. This series adds a new generic, resizable,
scalable, concurrent hash table based on the paper referenced in the first
patch. It then makes use of the new data type to implement lockless
netlink_lookup().
Patch 3/3 to convert nft_hash is included for reference but should be
merged via the netfilter tree. Inclusion in this series is to provide
context for the suggested API.
Against net-next since the initial user of the new hash table is in net/
Changes:
v4-v5:
- use GFP_KERNEL to alloc Netlink buckets as suggested by Nikolay
Aleksandrov
- free nft hash element on removal as spotted by Nikolay Aleksandrov
and Patrick McHardy
v3-v4:
- fixed wrong shift assignment placement as spotted by Nikolay Aleksandrov
- reverted default size of nft_hash to 4 as requested by Patrick McHardy,
default size for other hash tables remains at 64 if no hint is given
- fixed copyright as requested by Patrick McHardy
v2-v3:
- fixed typo in nft_hash_destroy() when passing rhashtable handle
v1-v2:
- fixed traversal off-by-one as spotted by Tobias Klauser
- removed unlikely() from BUG_ON() as spotted by Josh Triplett
- new 3rd patch to convert nft_hash to rhashtable
- make rhashtable_insert() return void
- nl_sk_hash_lock must be a mutex
- fixed wrong name of rht_shrink_below_30()
- exported symbols rht_grow_above_75() and rht_shrink_below_30()
- allow table freeing with RCU callback
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The sizing of the hash table and the practice of requiring a lookup
to retrieve the pprev to be stored in the element cookie before the
deletion of an entry is left intact.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Patrick McHardy <kaber@trash.net>
Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Heavy Netlink users such as Open vSwitch spend a considerable amount of
time in netlink_lookup() due to the read-lock on nl_table_lock. Use of
RCU relieves the lock contention.
Makes use of the new resizable hash table to avoid locking on the
lookup.
The hash table will grow if entries exceeds 75% of table size up to a
total table size of 64K. It will automatically shrink if usage falls
below 30%.
Also splits nl_table_lock into a separate mutex to protect hash table
mutations and allow synchronize_rcu() to sleep while waiting for readers
during expansion and shrinking.
Before:
9.16% kpktgend_0 [openvswitch] [k] masked_flow_lookup
6.42% kpktgend_0 [pktgen] [k] mod_cur_headers
6.26% kpktgend_0 [pktgen] [k] pktgen_thread_worker
6.23% kpktgend_0 [kernel.kallsyms] [k] memset
4.79% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup
4.37% kpktgend_0 [kernel.kallsyms] [k] memcpy
3.60% kpktgend_0 [openvswitch] [k] ovs_flow_extract
2.69% kpktgend_0 [kernel.kallsyms] [k] jhash2
After:
15.26% kpktgend_0 [openvswitch] [k] masked_flow_lookup
8.12% kpktgend_0 [pktgen] [k] pktgen_thread_worker
7.92% kpktgend_0 [pktgen] [k] mod_cur_headers
5.11% kpktgend_0 [kernel.kallsyms] [k] memset
4.11% kpktgend_0 [openvswitch] [k] ovs_flow_extract
4.06% kpktgend_0 [kernel.kallsyms] [k] _raw_spin_lock
3.90% kpktgend_0 [kernel.kallsyms] [k] jhash2
[...]
0.67% kpktgend_0 [kernel.kallsyms] [k] netlink_lookup
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Willem de Bruijn
Anish Bhatt
Neal Cardwell
David S. Miller
Zoltan Kiss
Ricardo Ribalda
Hans Wennborg
Eric Dumazet
Dmitry Popov
KY Srinivasan
Francois Romieu
Himangi Saraogi
Hariprasad Shenai
Michael S. Tsirkin
Thomas Graf