Currently, the TIPC RPS dissector is based only on the incoming packets'
source node address, hence steering all traffic from a node to the same
core. We have seen that this makes the links vulnerable to starvation
and unnecessary resets when we turn down the link tolerance to very low
values.
To reduce the risk of this happening, we exempt probe and probe replies
packets from the convergence to one core per source node. Instead, we do
the opposite, - we try to diverge those packets across as many cores as
possible, by randomizing the flow selector key.
To make such packets identifiable to the dissector, we add a new
'is_keepalive' bit to word 0 of the LINK_PROTOCOL header. This bit is
set both for PROBE and PROBE_REPLY messages, and only for those.
It should be noted that these packets are not part of any flow anyway,
and only constitute a minuscule fraction of all packets sent across a
link. Hence, there is no risk that this will affect overall performance.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit 2f487712b8 ("tipc: guarantee that group broadcast doesn't
bypass group unicast") there was introduced a last-minute rebasing
error that broke non-group communication.
We fix this here.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The following scenario is possible:
- A user joins a group, and immediately sends out a broadcast message
to its members.
- The broadcast message, following a different data path than the
initial JOIN message sent out during the joining procedure, arrives
to a receiver before the latter..
- The receiver drops the message, since it is not ready to accept any
messages until the JOIN has arrived.
We avoid this by treating group protocol JOIN messages like unicast
messages.
- We let them pass through the recipient's multicast input queue, just
like ordinary unicasts.
- We force the first following broadacst to be sent as replicated
unicast and being acknowledged by the recipient before accepting
any more broadcast transmissions.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We need a mechanism guaranteeing that group unicasts sent out from a
socket are not bypassed by later sent broadcasts from the same socket.
We do this as follows:
- Each time a unicast is sent, we set a the broadcast method for the
socket to "replicast" and "mandatory". This forces the first
subsequent broadcast message to follow the same network and data path
as the preceding unicast to a destination, hence preventing it from
overtaking the latter.
- In order to make the 'same data path' statement above true, we let
group unicasts pass through the multicast link input queue, instead
of as previously through the unicast link input queue.
- In the first broadcast following a unicast, we set a new header flag,
requiring all recipients to immediately acknowledge its reception.
- During the period before all the expected acknowledges are received,
the socket refuses to accept any more broadcast attempts, i.e., by
blocking or returning EAGAIN. This period should typically not be
longer than a few microseconds.
- When all acknowledges have been received, the sending socket will
open up for subsequent broadcasts, this time giving the link layer
freedom to itself select the best transmission method.
- The forced and/or abrupt transmission method changes described above
may lead to broadcasts arriving out of order to the recipients. We
remedy this by introducing code that checks and if necessary
re-orders such messages at the receiving end.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As a preparation for introducing flow control for multicast and datagram
messaging we need a more strictly defined framework than we have now. A
socket must be able keep track of exactly how many and which other
sockets it is allowed to communicate with at any moment, and keep the
necessary state for those.
We therefore introduce a new concept we have named Communication Group.
Sockets can join a group via a new setsockopt() call TIPC_GROUP_JOIN.
The call takes four parameters: 'type' serves as group identifier,
'instance' serves as an logical member identifier, and 'scope' indicates
the visibility of the group (node/cluster/zone). Finally, 'flags' makes
it possible to set certain properties for the member. For now, there is
only one flag, indicating if the creator of the socket wants to receive
a copy of broadcast or multicast messages it is sending via the socket,
and if wants to be eligible as destination for its own anycasts.
A group is closed, i.e., sockets which have not joined a group will
not be able to send messages to or receive messages from members of
the group, and vice versa.
Any member of a group can send multicast ('group broadcast') messages
to all group members, optionally including itself, using the primitive
send(). The messages are received via the recvmsg() primitive. A socket
can only be member of one group at a time.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the broadcast send link after 100 attempts has failed to
transfer a packet to all peers, we consider it stale, and reset
it. Thereafter it needs to re-synchronize with the peers, something
currently done by just resetting and re-establishing all links to
all peers. This has turned out to be overkill, with potentially
unwanted consequences for the remaining cluster.
A closer analysis reveals that this can be done much simpler. When
this kind of failure happens, for reasons that may lie outside the
TIPC protocol, it is typically only one peer which is failing to
receive and acknowledge packets. It is hence sufficient to identify
and reset the links only to that peer to resolve the situation, without
having to reset the broadcast link at all. This solution entails a much
lower risk of negative consequences for the own node as well as for
the overall cluster.
We implement this change in this commit.
Reviewed-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pass the new extended ACK reporting struct to all of the generic
netlink parsing functions. For now, pass NULL in almost all callers
(except for some in the core.)
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If the bearer carrying multicast messages supports broadcast, those
messages will be sent to all cluster nodes, irrespective of whether
these nodes host any actual destinations socket or not. This is clearly
wasteful if the cluster is large and there are only a few real
destinations for the message being sent.
In this commit we extend the eligibility of the newly introduced
"replicast" transmit option. We now make it possible for a user to
select which method he wants to be used, either as a mandatory setting
via setsockopt(), or as a relative setting where we let the broadcast
layer decide which method to use based on the ratio between cluster
size and the message's actual number of destination nodes.
In the latter case, a sending socket must stick to a previously
selected method until it enters an idle period of at least 5 seconds.
This eliminates the risk of message reordering caused by method change,
i.e., when changes to cluster size or number of destinations would
otherwise mandate a new method to be used.
Reviewed-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TIPC multicast messages are currently carried over a reliable
'broadcast link', making use of the underlying media's ability to
transport packets as L2 broadcast or IP multicast to all nodes in
the cluster.
When the used bearer is lacking that ability, we can instead emulate
the broadcast service by replicating and sending the packets over as
many unicast links as needed to reach all identified destinations.
We now introduce a new TIPC link-level 'replicast' service that does
this.
Reviewed-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Until now, we allocate memory always with GFP_ATOMIC flag.
When the system is under memory pressure and a user tries to send,
the send fails due to low memory. However, the user application
can wait for free memory if we allocate it using GFP_KERNEL flag.
In this commit, we use allocate memory with GFP_KERNEL for all user
allocation.
Reported-by: Rune Torgersen <runet@innovsys.com>
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The socket code currently handles link congestion by either blocking
and trying to send again when the congestion has abated, or just
returning to the user with -EAGAIN and let him re-try later.
This mechanism is prone to starvation, because the wakeup algorithm is
non-atomic. During the time the link issues a wakeup signal, until the
socket wakes up and re-attempts sending, other senders may have come
in between and occupied the free buffer space in the link. This in turn
may lead to a socket having to make many send attempts before it is
successful. In extremely loaded systems we have observed latency times
of several seconds before a low-priority socket is able to send out a
message.
In this commit, we simplify this mechanism and reduce the risk of the
described scenario happening. When a message is attempted sent via a
congested link, we now let it be added to the link's backlog queue
anyway, thus permitting an oversubscription of one message per source
socket. We still create a wakeup item and return an error code, hence
instructing the sender to block or stop sending. Only when enough space
has been freed up in the link's backlog queue do we issue a wakeup event
that allows the sender to continue with the next message, if any.
The fact that a socket now can consider a message sent even when the
link returns a congestion code means that the sending socket code can
be simplified. Also, since this is a good opportunity to get rid of the
obsolete 'mtu change' condition in the three socket send functions, we
now choose to refactor those functions completely.
Signed-off-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit e4bf4f7696 ("tipc: simplify packet sequence number
handling") we changed the internal representation of the packet
sequence number counters from u32 to u16, reflecting what is really
sent over the wire.
Since then some link statistics counters have been displaying incorrect
values, partially because the counters meant to be used as sequence
number snapshots are now used as direct counters, stored as u32, and
partially because some counter updates are just missing in the code.
In this commit we correct this in two ways. First, we base the
displayed packet sent/received values on direct counters instead
of as previously a calculated difference between current sequence
number and a snapshot. Second, we add the missing updates of the
counters.
This change is compatible with the current netlink API, and requires
no changes to the user space tools.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
commit 817298102b ("tipc: fix link priority propagation") introduced a
compatibility problem between TIPC versions newer than Linux 4.6 and
those older than Linux 4.4. In versions later than 4.4, link STATE
messages only contain a non-zero link priority value when the sender
wants the receiver to change its priority. This has the effect that the
receiver resets itself in order to apply the new priority. This works
well, and is consistent with the said commit.
However, in versions older than 4.4 a valid link priority is present in
all sent link STATE messages, leading to cyclic link establishment and
reset on the 4.6+ node.
We fix this by adding a test that the received value should not only
be valid, but also differ from the current value in order to cause the
receiving link endpoint to reset.
Reported-by: Amar Nv <amar.nv005@gmail.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit 2d18ac4ba7 ("tipc: extend broadcast link initialization
criteria") we tried to fix a problem with the initial synchronization
of broadcast link acknowledge values. Unfortunately that solution is
not sufficient to solve the issue.
We have seen it happen that LINK_PROTOCOL/STATE packets with a valid
non-zero unicast acknowledge number may bypass BCAST_PROTOCOL
initialization, NAME_DISTRIBUTOR and other STATE packets with invalid
broadcast acknowledge numbers, leading to premature opening of the
broadcast link. When the bypassed packets finally arrive, they are
inadvertently accepted, and the already correctly initialized
acknowledge number in the broadcast receive link is overwritten by
the invalid (zero) value of the said packets. After this the broadcast
link goes stale.
We now fix this by marking the packets where we know the acknowledge
value is or may be invalid, and then ignoring the acks from those.
To this purpose, we claim an unused bit in the header to indicate that
the value is invalid. We set the bit to 1 in the initial BCAST_PROTOCOL
synchronization packet and all initial ("bulk") NAME_DISTRIBUTOR
packets, plus those LINK_PROTOCOL packets sent out before the broadcast
links are fully synchronized.
This minor protocol update is fully backwards compatible.
Reported-by: John Thompson <thompa.atl@gmail.com>
Tested-by: John Thompson <thompa.atl@gmail.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because of the risk of an excessive number of NACK messages and
retransissions, receivers have until now abstained from sending
broadcast NACKS directly upon detection of a packet sequence number
gap. We have instead relied on such gaps being detected by link
protocol STATE message exchange, something that by necessity delays
such detection and subsequent retransmissions.
With the introduction of unicast NACK transmission and rate control
of retransmissions we can now remove this limitation. We now allow
receiving nodes to send NACKS immediately, while coordinating the
permission to do so among the nodes in order to avoid NACK storms.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As cluster sizes grow, so does the amount of identical or overlapping
broadcast NACKs generated by the packet receivers. This often leads to
'NACK crunches' resulting in huge numbers of redundant retransmissions
of the same packet ranges.
In this commit, we introduce rate control of broadcast retransmissions,
so that a retransmitted range cannot be retransmitted again until after
at least 10 ms. This reduces the frequency of duplicate, redundant
retransmissions by an order of magnitude, while having a significant
positive impact on overall throughput and scalability.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we send broadcasts in clusters of more 70-80 nodes, we sometimes
see the broadcast link resetting because of an excessive number of
retransmissions. This is caused by a combination of two factors:
1) A 'NACK crunch", where loss of broadcast packets is discovered
and NACK'ed by several nodes simultaneously, leading to multiple
redundant broadcast retransmissions.
2) The fact that the NACKS as such also are sent as broadcast, leading
to excessive load and packet loss on the transmitting switch/bridge.
This commit deals with the latter problem, by moving sending of
broadcast nacks from the dedicated BCAST_PROTOCOL/NACK message type
to regular unicast LINK_PROTOCOL/STATE messages. We allocate 10 unused
bits in word 8 of the said message for this purpose, and introduce a
new capability bit, TIPC_BCAST_STATE_NACK in order to keep the change
backwards compatible.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a link is attempted woken up after congestion, it uses a different,
more generous criteria than when it was originally declared congested.
This has the effect that the link, and the sending process, sometimes
will be woken up unnecessarily, just to immediately return to congestion
when it turns out there is not not enough space in its send queue to
host the pending message. This is a waste of CPU cycles.
We now change the function link_prepare_wakeup() to use exactly the same
criteria as tipc_link_xmit(). However, since we are now excluding the
window limit from the wakeup calculation, and the current backlog limit
for the lowest level is too small to house even a single maximum-size
message, we have to expand this limit. We do this by evaluating an
alternative, minimum value during the setting of the importance limits.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After a new receiver peer has been added to the broadcast transmission
link, we allow immediate transmission of new broadcast packets, trusting
that the new peer will not accept the packets until it has received the
previously sent unicast broadcast initialiation message. In the same
way, the sender must not accept any acknowledges until it has itself
received the broadcast initialization from the peer, as well as
confirmation of the reception of its own initialization message.
Furthermore, when a receiver peer goes down, the sender has to produce
the missing acknowledges from the lost peer locally, in order ensure
correct release of the buffers that were expected to be acknowledged by
the said peer.
In a highly stressed system we have observed that contact with a peer
may come up and be lost before the above mentioned broadcast initial-
ization and confirmation have been received. This leads to the locally
produced acknowledges being rejected, and the non-acknowledged buffers
to linger in the broadcast link transmission queue until it fills up
and the link goes into permanent congestion.
In this commit, we remedy this by temporarily setting the corresponding
broadcast receive link state to ESTABLISHED and the 'bc_peer_is_up'
state to true before we issue the local acknowledges. This ensures that
those acknowledges will always be accepted. The mentioned state values
are restored immediately afterwards when the link is reset.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
At first contact between two nodes, an endpoint might sometimes have
time to send out a LINK_PROTOCOL/STATE packet before it has received
the broadcast initialization packet from the peer, i.e., before it has
received a valid broadcast packet number to add to the 'bc_ack' field
of the protocol message.
This means that the peer endpoint will receive a protocol packet with an
invalid broadcast acknowledge value of 0. Under unlucky circumstances
this may lead to the original, already received acknowledge value being
overwritten, so that the whole broadcast link goes stale after a while.
We fix this by delaying the setting of the link field 'bc_peer_is_up'
until we know that the peer really has received our own broadcast
initialization message. The latter is always sent out as the first
unicast message on a link, and always with seqeunce number 1. Because
of this, we only need to look for a non-zero unicast acknowledge value
in the arriving STATE messages, and once that is confirmed we know we
are safe and can set the mentioned field. Before this moment, we must
ignore all broadcast acknowledges from the peer.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Several cases of overlapping changes, except the packet scheduler
conflicts which deal with the addition of the free list parameter
to qdisc_enqueue().
Signed-off-by: David S. Miller <davem@davemloft.net>
net/tipc/link.c: In function ‘tipc_link_timeout’:
net/tipc/link.c:744:28: warning: ‘mtyp’ may be used uninitialized in this function [-Wuninitialized]
Fixes: 42b18f605f ("tipc: refactor function tipc_link_timeout()")
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TIPC based clusters are by default set up with full-mesh link
connectivity between all nodes. Those links are expected to provide
a short failure detection time, by default set to 1500 ms. Because
of this, the background load for neighbor monitoring in an N-node
cluster increases with a factor N on each node, while the overall
monitoring traffic through the network infrastructure increases at
a ~(N * (N - 1)) rate. Experience has shown that such clusters don't
scale well beyond ~100 nodes unless we significantly increase failure
discovery tolerance.
This commit introduces a framework and an algorithm that drastically
reduces this background load, while basically maintaining the original
failure detection times across the whole cluster. Using this algorithm,
background load will now grow at a rate of ~(2 * sqrt(N)) per node, and
at ~(2 * N * sqrt(N)) in traffic overhead. As an example, each node will
now have to actively monitor 38 neighbors in a 400-node cluster, instead
of as before 399.
This "Overlapping Ring Supervision Algorithm" is completely distributed
and employs no centralized or coordinated state. It goes as follows:
- Each node makes up a linearly ascending, circular list of all its N
known neighbors, based on their TIPC node identity. This algorithm
must be the same on all nodes.
- The node then selects the next M = sqrt(N) - 1 nodes downstream from
itself in the list, and chooses to actively monitor those. This is
called its "local monitoring domain".
- It creates a domain record describing the monitoring domain, and
piggy-backs this in the data area of all neighbor monitoring messages
(LINK_PROTOCOL/STATE) leaving that node. This means that all nodes in
the cluster eventually (default within 400 ms) will learn about
its monitoring domain.
- Whenever a node discovers a change in its local domain, e.g., a node
has been added or has gone down, it creates and sends out a new
version of its node record to inform all neighbors about the change.
- A node receiving a domain record from anybody outside its local domain
matches this against its own list (which may not look the same), and
chooses to not actively monitor those members of the received domain
record that are also present in its own list. Instead, it relies on
indications from the direct monitoring nodes if an indirectly
monitored node has gone up or down. If a node is indicated lost, the
receiving node temporarily activates its own direct monitoring towards
that node in order to confirm, or not, that it is actually gone.
- Since each node is actively monitoring sqrt(N) downstream neighbors,
each node is also actively monitored by the same number of upstream
neighbors. This means that all non-direct monitoring nodes normally
will receive sqrt(N) indications that a node is gone.
- A major drawback with ring monitoring is how it handles failures that
cause massive network partitionings. If both a lost node and all its
direct monitoring neighbors are inside the lost partition, the nodes in
the remaining partition will never receive indications about the loss.
To overcome this, each node also chooses to actively monitor some
nodes outside its local domain. Those nodes are called remote domain
"heads", and are selected in such a way that no node in the cluster
will be more than two direct monitoring hops away. Because of this,
each node, apart from monitoring the member of its local domain, will
also typically monitor sqrt(N) remote head nodes.
- As an optimization, local list status, domain status and domain
records are marked with a generation number. This saves senders from
unnecessarily conveying unaltered domain records, and receivers from
performing unneeded re-adaptations of their node monitoring list, such
as re-assigning domain heads.
- As a measure of caution we have added the possibility to disable the
new algorithm through configuration. We do this by keeping a threshold
value for the cluster size; a cluster that grows beyond this value
will switch from full-mesh to ring monitoring, and vice versa when
it shrinks below the value. This means that if the threshold is set to
a value larger than any anticipated cluster size (default size is 32)
the new algorithm is effectively disabled. A patch set for altering the
threshold value and for listing the table contents will follow shortly.
- This change is fully backwards compatible.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The node keepalive interval is recalculated at each timer expiration
to catch any changes in the link tolerance, and stored in a field in
struct tipc_node. We use jiffies as unit for the stored value.
This is suboptimal, because it makes the calculation unnecessary
complex, including two unit conversions. The conversions also lead to
a rounding error that causes the link "abort limit" to be 3 in the
normal case, instead of 4, as intended. This again leads to unnecessary
link resets when the network is pushed close to its limit, e.g., in an
environment with hundreds of nodes or namesapces.
In this commit, we do instead let the keepalive value be calculated and
stored in milliseconds, so that there is only one conversion and the
rounding error is eliminated.
We also remove a redundant "keepalive" field in struct tipc_link. This
is remnant from the previous implementation.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 42b18f605f ("tipc: refactor function tipc_link_timeout()"),
introduced a bug which prevents sending of probe messages during
link synchronization phase. This leads to hanging links, if the
bearer is disabled/enabled after links are up.
In this commit, we send the probe messages correctly.
Fixes: 42b18f605f ("tipc: refactor function tipc_link_timeout()")
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
According to the link FSM, a received traffic packet can take a link
from state ESTABLISHING to ESTABLISHED, but the link can still not be
fully set up in one atomic operation. This means that even if the the
very first packet on the link is a traffic packet with sequence number
1 (one), it has to be dropped and retransmitted.
This can be avoided if we let the mentioned packet be preceded by a
LINK_PROTOCOL/STATE message, which takes up the endpoint before the
arrival of the traffic.
We add this small feature in this commit.
This is a fully compatible change.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The function tipc_link_timeout() is unnecessary complex, and can
easily be made more readable.
We do that with this commit. The only functional change is that we
remove a redundant test for whether the broadcast link is up or not.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a link is down, it will continuously try to re-establish contact
with the peer by sending out a RESET or an ACTIVATE message at each
timeout interval. The default value for this interval is currently
375 ms. This is wasteful, and may become a problem in very large
clusters with dozens or hundreds of nodes being down simultaneously.
We now introduce a simple backoff algorithm for these cases. The
first five messages are sent at default rate; thereafter a message
is sent only each 16th timer interval.
This will cover the vast majority of link recycling cases, since the
endpoint starting last will transmit at the higher speed, and the link
should normally be established well be before the rate needs to be
reduced.
The only case where we will see a degradation of link re-establishment
times is when the endpoints remain intact, and a glitch in the
transmission media is causing the link reset. We will then experience
a worst-case re-establishing time of 6 seconds, something we deem
acceptable.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a link endpoint is going down locally, e.g., because its interface
is being stopped, it will spontaneously send out a RESET message to
its peer, informing it about this fact. This saves the peer from
detecting the failure via probing, and hence gives both speedier and
less resource consuming failure detection on the peer side.
According to the link FSM, a receiver of a RESET message, ignoring the
reason for it, must now consider the sender ready to come back up, and
starts periodically sending out ACTIVATE messages to the peer in order
to re-establish the link. Also, according to the FSM, the receiver of
an ACTIVATE message can now go directly to state ESTABLISHED and start
sending regular traffic packets. This is a well-proven and robust FSM.
However, in the case of a reboot, there is a small possibilty that link
endpoint on the rebooted node may have been re-created with a new bearer
identity between the moment it sent its (pre-boot) RESET and the moment
it receives the ACTIVATE from the peer. The new bearer identity cannot
be known by the peer according to this scenario, since traffic headers
don't convey such information. This is a problem, because both endpoints
need to know the correct value of the peer's bearer id at any moment in
time in order to be able to produce correct link events for their users.
The only way to guarantee this is to enforce a full setup message
exchange (RESET + ACTIVATE) even after the reboot, since those messages
carry the bearer idientity in their header.
In this commit we do this by introducing and setting a "stopping" bit in
the header of the spontaneously generated RESET messages, informing the
peer that the sender will not be immediately ready to re-establish the
link. A receiver seeing this bit must act as if this were a locally
detected connectivity failure, and hence has to go through a full two-
way setup message exchange before any link can be re-established.
Although never reported, this problem seems to have always been around.
This protocol addition is fully backwards compatible.
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make the c files less cluttered and enable netlink attributes to be
shared between files.
Signed-off-by: Richard Alpe <richard.alpe@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Acked-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Until now, we have kept a pre-allocated protocol message header
aggregated into struct tipc_link. Apart from adding unnecessary
footprint to the link instances, this requires extra code both to
initialize and re-initialize it.
We now remove this sub-optimization. This change also makes it
possible to clean up the function tipc_build_proto_msg() and remove
a couple of small functions that were accessing the mentioned header.
In particular, we can replace all occurrences of the local function
call link_own_addr(link) with the generic tipc_own_addr(net).
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/phy/bcm7xxx.c
drivers/net/phy/marvell.c
drivers/net/vxlan.c
All three conflicts were cases of simple overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Refactor tipc_node_xmit() to fail fast and fail early. Fix several
potential memory leaks in unexpected error paths.
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Richard Alpe <richard.alpe@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently link priority changes isn't handled for active links. In
this patch we resolve this by changing our priority if the peer passes
a valid priority in a state message.
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Richard Alpe <richard.alpe@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Changing certain link attributes (link tolerance and link priority)
from the TIPC management tool is supposed to automatically take
effect at both endpoints of the affected link.
Currently the media address is not instantiated for the link and is
used uninstantiated when crafting protocol messages designated for the
peer endpoint. This means that changing a link property currently
results in the property being changed on the local machine but the
protocol message designated for the peer gets lost. Resulting in
property discrepancy between the endpoints.
In this patch we resolve this by using the media address from the
link entry and using the bearer transmit function to send it. Hence,
we can now eliminate the redundant function tipc_link_prot_xmit() and
the redundant field tipc_link::media_addr.
Fixes: 2af5ae372a (tipc: clean up unused code and structures)
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reported-by: Jason Hu <huzhijiang@gmail.com>
Signed-off-by: Richard Alpe <richard.alpe@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/renesas/ravb_main.c
kernel/bpf/syscall.c
net/ipv4/ipmr.c
All three conflicts were cases of overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit 5266698661 ("tipc: let broadcast packet
reception use new link receive function") the broadcast send
link state was meant to always be set to LINK_ESTABLISHED, since
we don't need this link to follow the regular link FSM rules. It
was also the intention that this state anyway shouldn't impact
the run-time working state of the link, since the latter in
reality is controlled by the number of registered peers.
We have now discovered that this assumption is not quite correct.
If the broadcast link is reset because of too many retransmissions,
its state will inadvertently go to LINK_RESETTING, and never go
back to LINK_ESTABLISHED, because the LINK_FAILURE event was not
anticipated. This will work well once, but if it happens a second
time, the reset on a link in LINK_RESETTING has has no effect, and
neither the broadcast link nor the unicast links will go down as
they should.
Furthermore, it is confusing that the management tool shows that
this link is in UP state when that obviously isn't the case.
We now ensure that this state strictly follows the true working
state of the link. The state is set to LINK_ESTABLISHED when
the number of peers is non-zero, and to LINK_RESET otherwise.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The number of variables with Hungarian notation (l_ptr, n_ptr etc.)
has been significantly reduced over the last couple of years.
We now root out the last traces of this practice.
There are no functional changes in this commit.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We move the definition of struct tipc_link from link.h to link.c in
order to minimize its exposure to the rest of the code.
When needed, we define new functions to make it possible for external
entities to access and set data in the link.
Apart from the above, there are no functional changes.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In our effort to have less code and include dependencies between
entities such as node, link and bearer, we try to narrow down
the exposed interface towards the node as much as possible.
In this commit, we move the definition of struct tipc_node, along
with many of its associated function declarations, from node.h to
node.c. We also move some function definitions from link.c and
name_distr.c to node.c, since they access fields in struct tipc_node
that should not be externally visible. The moved functions are renamed
according to new location, and made static whenever possible.
There are no functional changes in this commit.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
According to the node FSM a node in state SELF_UP_PEER_UP cannot
change state inside a lock context, except when a TUNNEL_PROTOCOL
(SYNCH or FAILOVER) packet arrives. However, the node's individual
links may still change state.
Since each link now is protected by its own spinlock, we finally have
the conditions in place to convert the node spinlock to an rwlock_t.
If the node state and arriving packet type are rigth, we can let the
link directly receive the packet under protection of its own spinlock
and the node lock in read mode. In all other cases we use the node
lock in write mode. This enables full concurrent execution between
parallel links during steady-state traffic situations, i.e., 99+ %
of the time.
This commit implements this change.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As a preparation to allow parallel links to work more independently
from each other we introduce a per-link spinlock, to be stored in the
struct nodes's link entry area. Since the node lock still is a regular
spinlock there is no increase in parallellism at this stage.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit 5cbb28a4bf ("tipc: linearize arriving NAME_DISTR
and LINK_PROTO buffers") we added linearization of NAME_DISTRIBUTOR,
LINK_PROTOCOL/RESET and LINK_PROTOCOL/ACTIVATE to the function
tipc_udp_recv(). The location of the change was selected in order
to make the commit easily appliable to 'net' and 'stable'.
We now move this linearization to where it should be done, in the
functions tipc_named_rcv() and tipc_link_proto_rcv() respectively.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TO: "David S. Miller" <davem@davemloft.net>
CC: netdev@vger.kernel.org
CC: Jon Maloy <jon.maloy@ericsson.com>
CC: Ying Xue <ying.xue@windriver.com>
CC: tipc-discussion@lists.sourceforge.net
CC: linux-kernel@vger.kernel.org
Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After the previous changes in this series, we can now remove some
unused code and structures, both in the broadcast, link aggregation
and link code.
There are no functional changes in this commit.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the recent commit series, we have established a one-way dependency
between the link aggregation (struct tipc_node) instances and their
pertaining tipc_link instances. This has enabled quite significant code
and structure simplifications.
In this commit, we eliminate the field 'owner', which points to an
instance of struct tipc_node, from struct tipc_link, and replace it with
a pointer to struct net, which is the only external reference now needed
by a link instance.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The neighbor discovery function currently uses the function
tipc_bearer_send() for transmitting packets, assuming that the
sent buffers are not consumed by the called function.
We want to change this, in order to avoid unnecessary buffer cloning
elswhere in the code.
This commit introduces a new function tipc_bearer_skb() which consumes
the sent buffers, and let the discoverer functions use this new call
instead. The discoverer does now itself perform the cloning when
that is necessary.
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jon Maloy