[ Some fixes went into mainstream before this patch, so I needed
to rebase it upon the current tree, that's why it's different from
the original one posted on the list --pablo ]
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
- introduction of the new Network Coding component. This new mechanism aims to
increase throughput by fusing multiple packets in one transmission.
- minor cleanups
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)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=icrn
-----END PGP SIGNATURE-----
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
Included changes:
- introduction of the new Network Coding component. This new mechanism aims to
increase throughput by fusing multiple packets in one transmission.
- minor cleanups
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch removes chip specific operations from the common hardware
paths, as well as the Makefile change to accomodate the new files.
Signed-off-by: Arvind Bhushan <arvindb@chelsio.com>
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch defines the common operations to support multiple chips. It
includes common header file modifications to support the current chips
(T4 and T5). It also includes the following bug fixes:
- reconfirms the rnode state after an implicit logo.
- corrects the stats array size.
- sets up and checks flags correctly when coming up as master and finding
the card initialized
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Arvind Bhushan <arvindb@chelsio.com>
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch creates a new file for T5 adapter operations.
Signed-off-by: Arvind Bhushan <arvindb@chelsio.com>
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch separates T4 adapter operations into a new file.
Signed-off-by: Arvind Bhushan <arvindb@chelsio.com>
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
T5 adapter does not support onchip queue memory. Present logic fails to
allocate QP for T5 and returns an error. Also, if module parameter ocqp_support
is zero then we are unable to allocate QP which should not be the case. Ideally
if ocqp_support parameter is 0 or onchip queue support is disable then host QP
should be allocated before returning an error.
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Always bump the tcam_full stat. Also, bump wr reply timeout to 30 seconds.
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It enables direct DMA by HW to memory region PBL arrays and fast register PBL
arrays from host memory, vs the T4 way of passing these arrays in the WR itself.
The result is lower latency for memory registration, and larger PBL array
support for fast register operations.
This patch also updates ULP_TX_MEM_WRITE command fields for T5. Ordering bit of
ULP_TX_MEM_WRITE is at bit position 22 in T5 and at 23 in T4.
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Both DB Flow-Control and DB Coalescing are disabled by default on T5
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adds support for Chelsio T5 adapter.
Enables T5's Write Combining feature.
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
All T5 adapters will only support VFs on PF0-3 despite the ability of the
hardware to support them on PF4-7. This keeps our T4 and T5 adapters more
similar which simplifies host driver software.
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements a low latency Write Combining (aka Write Coalescing) work
request path. PCIE maps User Space Doorbell BAR2 region writes to the new
interface to SGE. SGE pulls a new message from PCIE new interface and if its a
coalesced write work request then pushes it for processing. This patch copies
coalesced work request to memory mapped BAR2 space.
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When receiving a network coded packet, the decoding buffer is searched
for a packet to use for decoding. The source, destination, and crc32 from
the coded packet is used to identify the wanted packet. The decoded
packet is passed to the usual unicast receiver function, as had it never
been network coded.
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
To be able to decode a network coded packet, a node must already know
one of the two coded packets. This is done by buffering skbs before
transmission and buffering packets sniffed with promiscuous mode from
other hosts.
Packets are kept in a buffer similar to the one with forward-skbs: A
hash table, where each entry, which corresponds to a src-dst pair, has a
linked list packets.
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Before adding forward-skbs to the coding buffer, the buffer is searched
for a potential coding opportunity. If one is found, the two packets are
network coded and transmitted right away. If not, the forward-skb is
added to the buffer.
Network coded packets are transmitted with information about the two
receivers and the two coded packets. The first receiver is given by the
MAC header, while the second is given in the payload/bat-header. The
second receiver uses promiscuous mode to receive the packet and check
the second destination.
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Two be able to network code two packets, one packet must be buffered
until the next is available. This is done in a "coding buffer", which is
essentially a hash table with lists of packets. Each entry in the hash
table corresponds to a specific src-dst pair, which has a linked list of
packets that are buffered.
This patch adds skbs to the buffer just before forwarding them. The
buffer is traversed every 10 ms, where timed skbs are removed from the
buffer and transmitted. To allow experiments with the network coding
scheme, the timeout is tunable through a file in debugfs.
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
To use network coding efficiently, a relay must know when neighbor nodes
are likely to have enough information to be able to decode a network
coded packet. This is detected by using OGMs from batman-adv to discover
when one neighbor is in range of another neighbor. The relay check the
TLL to detect when an OGM is forwarded from one neighbor by another
neighbor, and thereby knows that the two neighbors are in range and thus
overhear packets sent by each other.
This information is saved in the orig_node struct to be used when
searching for coding opportunities. Two lists are added to the
orig_node struct: One for neighbors that can hear the orig_node
(outgoing nc_nodes) and one for neighbors that the orig_node can hear
(incoming nc_nodes).
Information about nc_nodes is kept for 10 seconds and is available
through debugfs in batman_adv/nc_nodes to use when debugging network
coding.
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Network coding exploits the 802.11 shared medium to allow multiple
packets to be sent in a single transmission. In brief, a relay can XOR
two packets, and send the coded packet to two destinations. The
receivers can decode one of the original packets by XOR'ing the coded
packet with the other original packet. This will lead to increased
throughput in topologies where two packets cross one relay.
In a simple topology with three nodes, it takes four transmissions
without network coding to get one packet from Node A to Node B and one
from Node B to Node A:
1. Node A ---- p1 ---> Node R Node B
2. Node A Node R <--- p2 ---- Node B
3. Node A <--- p2 ---- Node R Node B
4. Node A Node R ---- p1 ---> Node B
With network coding, the relay only needs one transmission, which saves
us one slot of valuable airtime:
1. Node A ---- p1 ---> Node R Node B
2. Node A Node R <--- p2 ---- Node B
3. Node A <- p1 x p2 - Node R - p1 x p2 -> Node B
The same principle holds for a topology including five nodes. Here the
packets from Node A and Node B are overheard by Node C and Node D,
respectively. This allows Node R to send a network coded packet to save
one transmission:
Node A Node B
| \ / |
| p1 p2 |
| \ / |
p1 > Node R < p2
| |
| / \ |
| p1 x p2 p1 x p2 |
v / \ v
/ \
Node C < > Node D
More information is available on the open-mesh.org wiki[1].
This patch adds the initial code to support network coding in
batman-adv. It sets up a worker thread to do house keeping and adds a
sysfs file to enable/disable network coding. The feature is disabled by
default, as it requires a wifi-driver with working promiscuous mode, and
also because it adds a small delay at each hop.
[1] http://www.open-mesh.org/projects/batman-adv/wiki/Catwoman
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
In C standard any expression different from 0 will be converted to
'true' when casting to bool (whatever is the length of the value).
Therefore all the "!!" conversions can be removed.
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
batadv_check_unicast_packet() is changed to return a value based on the
reason to drop the packet, which will be useful information for
future users of batadv_check_unicast_packet().
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
The batadv_priv struct carries a pointer to its own interface
struct. Therefore, it is not necessary to retrieve the soft_iface
via the primary interface.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
The variable dev is initialized but never used
otherwise, so remove the unused variable.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
QLogic applications use these callbacks to perform
o NIC Partitioning (NPAR) configuration and management
o Diagnostic tests
o Flash access and updates
Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mugunthan V N says:
====================
This patch serires implements the following features in CPSW driver
* get/set phy link settings
* interrupt pacing
* get phy id via ioctl cmd SIOCGMIIPHY
Changes from initial version
* Made active-slave common for cpts, ethtool and SIOCGMIIPHY ioctl
* Cleaned CPSW DT binding documentation by seperating slave nodes
under sub-section
* implemented get phy id via ioctl cmd SIOCGMIIPHY
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement get phy_id via ioctl SIOCGMIIPHY. In switch mode active phy_id
is returned and in dual EMAC mode slave's specific phy_id is returned.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements support for interrupt pacing block of CPSW via ethtool
Inetrrupt pacing block is common of both the ethernet interface in
dual emac mode
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements get/set of the phy settings via ethtool apis
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change cpts_active_slave to active_slave so that the same DT property
can be used to ethtool and SIOCGMIIPHY.
CC: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move all the slave note properties to separate section to reduce the
confusion between slave note properties and cpsw node properties
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement to collect ethernet statistical information on ks8851_mll device.
Signed-off-by: David J. Choi <david.choi@micrel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
remove cast for kmalloc/kzalloc return value.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
remove cast for kmalloc return value.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
This has been tested on a qnap TS-119P II. Note that enabling WOL with
"ethtool -s eth0 wol g" is not enough; you also need to tell the PIC
microcontroller inside the qnap that WOL should be enabled by sending
0xF2 with qcontrol(1) and you have to disable EUP ("Energy-using
Products", a European power-saving thing) by sending 0xF4.
Signed-off-by: Michael Stapelberg <michael@stapelberg.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This allows ethernet drivers (such as the mv643xx_eth) to support
Wake on LAN on platforms where PHY registers have to be configured
for Wake on LAN (e.g. the Marvell Kirkwood based qnap TS-119P II).
Signed-off-by: Michael Stapelberg <michael@stapelberg.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is the second of the TLP patch series; it augments the basic TLP
algorithm with a loss detection scheme.
This patch implements a mechanism for loss detection when a Tail
loss probe retransmission plugs a hole thereby masking packet loss
from the sender. The loss detection algorithm relies on counting
TLP dupacks as outlined in Sec. 3 of:
http://tools.ietf.org/html/draft-dukkipati-tcpm-tcp-loss-probe-01
The basic idea is: Sender keeps track of TLP "episode" upon
retransmission of a TLP packet. An episode ends when the sender receives
an ACK above the SND.NXT (tracked by tlp_high_seq) at the time of the
episode. We want to make sure that before the episode ends the sender
receives a "TLP dupack", indicating that the TLP retransmission was
unnecessary, so there was no loss/hole that needed plugging. If the
sender gets no TLP dupack before the end of the episode, then it reduces
ssthresh and the congestion window, because the TLP packet arriving at
the receiver probably plugged a hole.
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch series implement the Tail loss probe (TLP) algorithm described
in http://tools.ietf.org/html/draft-dukkipati-tcpm-tcp-loss-probe-01. The
first patch implements the basic algorithm.
TLP's goal is to reduce tail latency of short transactions. It achieves
this by converting retransmission timeouts (RTOs) occuring due
to tail losses (losses at end of transactions) into fast recovery.
TLP transmits one packet in two round-trips when a connection is in
Open state and isn't receiving any ACKs. The transmitted packet, aka
loss probe, can be either new or a retransmission. When there is tail
loss, the ACK from a loss probe triggers FACK/early-retransmit based
fast recovery, thus avoiding a costly RTO. In the absence of loss,
there is no change in the connection state.
PTO stands for probe timeout. It is a timer event indicating
that an ACK is overdue and triggers a loss probe packet. The PTO value
is set to max(2*SRTT, 10ms) and is adjusted to account for delayed
ACK timer when there is only one oustanding packet.
TLP Algorithm
On transmission of new data in Open state:
-> packets_out > 1: schedule PTO in max(2*SRTT, 10ms).
-> packets_out == 1: schedule PTO in max(2*RTT, 1.5*RTT + 200ms)
-> PTO = min(PTO, RTO)
Conditions for scheduling PTO:
-> Connection is in Open state.
-> Connection is either cwnd limited or no new data to send.
-> Number of probes per tail loss episode is limited to one.
-> Connection is SACK enabled.
When PTO fires:
new_segment_exists:
-> transmit new segment.
-> packets_out++. cwnd remains same.
no_new_packet:
-> retransmit the last segment.
Its ACK triggers FACK or early retransmit based recovery.
ACK path:
-> rearm RTO at start of ACK processing.
-> reschedule PTO if need be.
In addition, the patch includes a small variation to the Early Retransmit
(ER) algorithm, such that ER and TLP together can in principle recover any
N-degree of tail loss through fast recovery. TLP is controlled by the same
sysctl as ER, tcp_early_retrans sysctl.
tcp_early_retrans==0; disables TLP and ER.
==1; enables RFC5827 ER.
==2; delayed ER.
==3; TLP and delayed ER. [DEFAULT]
==4; TLP only.
The TLP patch series have been extensively tested on Google Web servers.
It is most effective for short Web trasactions, where it reduced RTOs by 15%
and improved HTTP response time (average by 6%, 99th percentile by 10%).
The transmitted probes account for <0.5% of the overall transmissions.
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
YOSHIFUJI Hideaki
David S. Miller
Arvind Bhushan
Vipul Pandya
Santosh Rastapur
Martin Hundebøll
Antonio Quartulli
Marek Lindner
Wei Yongjun
Shahed Shaikh
Himanshu Madhani
Mugunthan V N
David J. Choi
Zhang Yanfei
Michael Stapelberg
Nandita Dukkipati