A new argument is added to __skb_recv_datagram to provide
an explicit skb destructor, invoked under the receive queue
lock.
The UDP protocol uses such argument to perform memory
reclaiming on dequeue, so that the UDP protocol does not
set anymore skb->desctructor.
Instead explicit memory reclaiming is performed at close() time and
when skbs are removed from the receive queue.
The in kernel UDP protocol users now need to call a
skb_recv_udp() variant instead of skb_recv_datagram() to
properly perform memory accounting on dequeue.
Overall, this allows acquiring only once the receive queue
lock on dequeue.
Tested using pktgen with random src port, 64 bytes packet,
wire-speed on a 10G link as sender and udp_sink as the receiver,
using an l4 tuple rxhash to stress the contention, and one or more
udp_sink instances with reuseport.
nr sinks vanilla patched
1 440 560
3 2150 2300
6 3650 3800
9 4450 4600
12 6250 6450
v1 -> v2:
- do rmem and allocated memory scheduling under the receive lock
- do bulk scheduling in first_packet_length() and in udp_destruct_sock()
- avoid the typdef for the dequeue callback
Suggested-by: Eric Dumazet <edumazet@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ip6_route_output() doesn't return a negative error when it fails, rather
the ->error field of the returned dst_entry struct needs to be checked.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: 75b54cb57c ("rxrpc: Add IPv6 support")
Signed-off-by: David Howells <dhowells@redhat.com>
Fix the following checker warning:
net/rxrpc/call_object.c:279 rxrpc_new_client_call()
warn: passing zero to 'ERR_PTR'
where a value that's always zero is passed to ERR_PTR() so that it can be
passed to a tracepoint in an auxiliary pointer field.
Just pass NULL instead to the tracepoint.
Fixes: a84a46d730 ("rxrpc: Add some additional call tracing")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Don't request an ACK on the last DATA packet of a call's Tx phase as for a
client there will be a reply packet or some sort of ACK to shift phase. If
the ACK is requested, OpenAFS sends a REQUESTED-ACK ACK with soft-ACKs in
it and doesn't follow up with a hard-ACK.
If we don't set the flag, OpenAFS will send a DELAY ACK that hard-ACKs the
reply data, thereby allowing the call to terminate cleanly.
Signed-off-by: David Howells <dhowells@redhat.com>
We need to generate a DELAY ACK from the service end of an operation if we
start doing the actual operation work and it takes longer than expected.
This will hard-ACK the request data and allow the client to release its
resources.
To make this work:
(1) We have to set the ack timer and propose an ACK when the call moves to
the RXRPC_CALL_SERVER_ACK_REQUEST and clear the pending ACK and cancel
the timer when we start transmitting the reply (the first DATA packet
of the reply implicitly ACKs the request phase).
(2) It must be possible to set the timer when the caller is holding
call->state_lock, so split the lock-getting part of the timer function
out.
(3) Add trace notes for the ACK we're requesting and the timer we clear.
Signed-off-by: David Howells <dhowells@redhat.com>
In rxrpc_kernel_recv_data(), when we return the error number incurred by a
failed call, we must negate it before returning it as it's stored as
positive (that's what we have to pass back to userspace).
Signed-off-by: David Howells <dhowells@redhat.com>
The call's background processor work item needs to notify the socket when
it completes a call so that recvmsg() or the AFS fs can deal with it.
Without this, call expiry isn't handled.
Signed-off-by: David Howells <dhowells@redhat.com>
When a call expires, it must be queued for the background processor to deal
with otherwise a service call that is improperly terminated will just sit
there awaiting an ACK and won't expire.
Signed-off-by: David Howells <dhowells@redhat.com>
OpenAFS doesn't always correctly terminate client calls that it makes -
this includes calls the OpenAFS servers make to the cache manager service.
It should end the client call with either:
(1) An ACK that has firstPacket set to one greater than the seq number of
the reply DATA packet with the LAST_PACKET flag set (thereby
hard-ACK'ing all packets). nAcks should be 0 and acks[] should be
empty (ie. no soft-ACKs).
(2) An ACKALL packet.
OpenAFS, though, may send an ACK packet with firstPacket set to the last
seq number or less and soft-ACKs listed for all packets up to and including
the last DATA packet.
The transmitter, however, is obliged to keep the call live and the
soft-ACK'd DATA packets around until they're hard-ACK'd as the receiver is
permitted to drop any merely soft-ACK'd packet and request retransmission
by sending an ACK packet with a NACK in it.
Further, OpenAFS will also terminate a client call by beginning the next
client call on the same connection channel. This implicitly completes the
previous call.
This patch handles implicit ACK of a call on a channel by the reception of
the first packet of the next call on that channel.
If another call doesn't come along to implicitly ACK a call, then we have
to time the call out. There are some bugs there that will be addressed in
subsequent patches.
Signed-off-by: David Howells <dhowells@redhat.com>
Separate the output of PING ACKs from the output of other sorts of ACK so
that if we receive a PING ACK and schedule transmission of a PING RESPONSE
ACK, the response doesn't get cancelled by a PING ACK we happen to be
scheduling transmission of at the same time.
If a PING RESPONSE gets lost, the other side might just sit there waiting
for it and refuse to proceed otherwise.
Signed-off-by: David Howells <dhowells@redhat.com>
Split rxrpc_send_data_packet() to separate ACK generation (which is more
complicated) from ABORT generation. This simplifies the code a bit and
fixes the following warning:
In file included from ../net/rxrpc/output.c:20:0:
net/rxrpc/output.c: In function 'rxrpc_send_call_packet':
net/rxrpc/ar-internal.h:1187:27: error: 'top' may be used uninitialized in this function [-Werror=maybe-uninitialized]
net/rxrpc/output.c:103:24: note: 'top' was declared here
net/rxrpc/output.c:225:25: error: 'hard_ack' may be used uninitialized in this function [-Werror=maybe-uninitialized]
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David Howells <dhowells@redhat.com>
When a reply is deemed lost, we send a ping to find out the other end
received all the request data packets we sent. This should be limited to
client calls and we shouldn't do this on service calls.
Signed-off-by: David Howells <dhowells@redhat.com>
If an call comes in to a local endpoint that isn't listening for any
incoming calls at the moment, an oops will happen. We need to check that
the local endpoint's service pointer isn't NULL before we dereference it.
Signed-off-by: David Howells <dhowells@redhat.com>
struct rxrpc_local->service is marked __rcu - this means that accesses of
it need to be managed using RCU wrappers. There are two such places in
rxrpc_release_sock() where the value is checked and cleared. Fix this by
using the appropriate wrappers.
Signed-off-by: David Howells <dhowells@redhat.com>
The call timer's concept of a call timeout (of which there are three) that
is inactive is that it is the timeout has the same expiration time as the
call expiration timeout (the expiration timer is never inactive). However,
I'm not resetting the timeouts when they expire, leading to repeated
processing of expired timeouts when other timeout events occur.
Fix this by:
(1) Move the timer expiry detection into rxrpc_set_timer() inside the
locked section. This means that if a timeout is set that will expire
immediately, we deal with it immediately.
(2) If a timeout is at or before now then it has expired. When an expiry
is detected, an event is raised, the timeout is automatically
inactivated and the event processor is queued.
(3) If a timeout is at or after the expiry timeout then it is inactive.
Inactive timeouts do not contribute to the timer setting.
(4) The call timer callback can now just call rxrpc_set_timer() to handle
things.
(5) The call processor work function now checks the event flags rather
than checking the timeouts directly.
Signed-off-by: David Howells <dhowells@redhat.com>
Keep that call timeouts as ktimes rather than jiffies so that they can be
expressed as functions of RTT.
Signed-off-by: David Howells <dhowells@redhat.com>
When we receive an ACK from the peer that tells us what the peer's receive
window (rwind) is, we should reduce ssthresh to rwind if rwind is smaller
than ssthresh.
Signed-off-by: David Howells <dhowells@redhat.com>
Switch to Congestion Avoidance mode at cwnd == ssthresh rather than relying
on cwnd getting incremented beyond ssthresh and the window size, the mode
being shifted and then cwnd being corrected.
We need to make sure we switch into CA mode so that we stop marking every
packet for ACK.
Signed-off-by: David Howells <dhowells@redhat.com>
Note the serial number of the packet being ACK'd in the congestion
management trace rather than the serial number of the ACK packet. Whilst
the serial number of the ACK packet is useful for matching ACK packet in
the output of wireshark, the serial number that the ACK is in response to
is of more use in working out how different trace lines relate.
Signed-off-by: David Howells <dhowells@redhat.com>
Set the request-ACK on more DATA packets whilst we're in slow start mode so
that we get sufficient ACKs back to supply information to configure the
window.
Signed-off-by: David Howells <dhowells@redhat.com>
Reduce the rxrpc_local::services list to just a pointer as we don't permit
multiple service endpoints to bind to a single transport endpoints (this is
excluded by rxrpc_lookup_local()).
The reason we don't allow this is that if you send a request to an AFS
filesystem service, it will try to talk back to your cache manager on the
port you sent from (this is how file change notifications are handled). To
prevent someone from stealing your CM callbacks, we don't let AF_RXRPC
sockets share a UDP socket if at least one of them has a service bound.
Signed-off-by: David Howells <dhowells@redhat.com>
In rxrpc_activate_channels(), the connection cache state is checked outside
of the lock, which means it can change whilst we're waking calls up,
thereby changing whether or not we're allowed to wake calls up.
Fix this by moving the check inside the locked region. The check to see if
all the channels are currently busy can stay outside of the locked region.
Whilst we're at it:
(1) Split the locked section out into its own function so that we can call
it from other places in a later patch.
(2) Determine the mask of channels dependent on the state as we're going
to add another state in a later patch that will restrict the number of
simultaneous calls to 1 on a connection.
Signed-off-by: David Howells <dhowells@redhat.com>
In rxrpc_send_data_packet() make the loss-injection path return through the
same code as the transmission path so that the RTT determination is
initiated and any future timer shuffling will be done, despite the packet
having been binned.
Whilst we're at it:
(1) Add to the tx_data tracepoint an indication of whether or not we're
retransmitting a data packet.
(2) When we're deciding whether or not to request an ACK, rather than
checking if we're in fast-retransmit mode check instead if we're
retransmitting.
(3) Don't invoke the lose_skb tracepoint when losing a Tx packet as we're
not altering the sk_buff refcount nor are we just seeing it after
getting it off the Tx list.
(4) The rxrpc_skb_tx_lost note is then no longer used so remove it.
(5) rxrpc_lose_skb() no longer needs to deal with rxrpc_skb_tx_lost.
Signed-off-by: David Howells <dhowells@redhat.com>
Exclusive connections are currently reusable (which they shouldn't be)
because rxrpc_alloc_client_connection() checks the exclusive flag in the
rxrpc_connection struct before it's initialised from the function
parameters. This means that the DONT_REUSE flag doesn't get set.
Fix this by checking the function parameters for the exclusive flag.
Signed-off-by: David Howells <dhowells@redhat.com>
Implement RxRPC slow-start, which is similar to RFC 5681 for TCP. A
tracepoint is added to log the state of the congestion management algorithm
and the decisions it makes.
Notes:
(1) Since we send fixed-size DATA packets (apart from the final packet in
each phase), counters and calculations are in terms of packets rather
than bytes.
(2) The ACK packet carries the equivalent of TCP SACK.
(3) The FLIGHT_SIZE calculation in RFC 5681 doesn't seem particularly
suited to SACK of a small number of packets. It seems that, almost
inevitably, by the time three 'duplicate' ACKs have been seen, we have
narrowed the loss down to one or two missing packets, and the
FLIGHT_SIZE calculation ends up as 2.
(4) In rxrpc_resend(), if there was no data that apparently needed
retransmission, we transmit a PING ACK to ask the peer to tell us what
its Rx window state is.
Signed-off-by: David Howells <dhowells@redhat.com>
If we've sent all the request data in a client call but haven't seen any
sign of the reply data yet, schedule an ACK to be sent to the server to
find out if the reply data got lost.
If the server hasn't yet hard-ACK'd the request data, we send a PING ACK to
demand a response to find out whether we need to retransmit.
If the server says it has received all of the data, we send an IDLE ACK to
tell the server that we haven't received anything in the receive phase as
yet.
To make this work, a non-immediate PING ACK must carry a delay. I've chosen
the same as the IDLE ACK for the moment.
Signed-off-by: David Howells <dhowells@redhat.com>
Generate a summary of the Tx buffer packet state when an ACK is received
for use in a later patch that does congestion management.
Signed-off-by: David Howells <dhowells@redhat.com>
When determining the resend timer value, we have a value in nsec but the
timer is in jiffies which may be a million or more times more coarse.
nsecs_to_jiffies() rounds down - which means that the resend timeout
expressed as jiffies is very likely earlier than the one expressed as
nanoseconds from which it was derived.
The problem is that rxrpc_resend() gets triggered by the timer, but can't
then find anything to resend yet. It sets the timer again - but gets
kicked off immediately again and again until the nanosecond-based expiry
time is reached and we actually retransmit.
Fix this by adding 1 to the jiffies-based resend_at value to counteract the
rounding and make sure that the timer happens after the nanosecond-based
expiry is passed.
Alternatives would be to adjust the timestamp on the packets to align
with the jiffie scale or to switch back to using jiffie-timestamps.
Signed-off-by: David Howells <dhowells@redhat.com>
Clear the ACK reason, ACK timer and resend timer when entering the client
reply phase when the first DATA packet is received. New ACKs will be
proposed once the data is queued.
The resend timer is no longer relevant and we need to cancel ACKs scheduled
to probe for a lost reply.
Signed-off-by: David Howells <dhowells@redhat.com>
Send an immediate ACK if we fill in a hole in the buffer left by an
out-of-sequence packet. This may allow the congestion management in the peer
to avoid a retransmission if packets got reordered on the wire.
Signed-off-by: David Howells <dhowells@redhat.com>
Send an ACK if we haven't sent one for the last two packets we've received.
This keeps the other end apprised of where we've got to - which is
important if they're doing slow-start.
We do this in recvmsg so that we can dispatch a packet directly without the
need to wake up the background thread.
This should possibly be made configurable in future.
Signed-off-by: David Howells <dhowells@redhat.com>
Add a tracepoint to log in rxrpc_resend() which packets will be
retransmitted. Note that if a positive ACK comes in whilst we have dropped
the lock to retransmit another packet, the actual retransmission may not
happen, though some of the effects will (such as altering the congestion
management).
Signed-off-by: David Howells <dhowells@redhat.com>
Add a tracepoint to log proposed ACKs, including whether the proposal is
used to update a pending ACK or is discarded in favour of an easlier,
higher priority ACK.
Whilst we're at it, get rid of the rxrpc_acks() function and access the
name array directly. We do, however, need to validate the ACK reason
number given to trace_rxrpc_rx_ack() to make sure we don't overrun the
array.
Signed-off-by: David Howells <dhowells@redhat.com>
Add a tracepoint to log transmission of DATA packets (including loss
injection).
Adjust the ACK transmission tracepoint to include the packet serial number
and to line this up with the DATA transmission display.
Signed-off-by: David Howells <dhowells@redhat.com>
rxrpc_send_call_packet() is invoking the tx_ack tracepoint before it checks
whether there's an ACK to transmit (another thread may jump in and transmit
it).
Fix this by only invoking the tracepoint if we get a valid ACK to transmit.
Further, only allocate a serial number if we're going to actually transmit
something.
Signed-off-by: David Howells <dhowells@redhat.com>
When the last packet of data to be transmitted on a call is queued, tx_top
is set and then the RXRPC_CALL_TX_LAST flag is set. Unfortunately, this
leaves a race in the ACK processing side of things because the flag affects
the interpretation of tx_top and also allows us to start receiving reply
data before we've finished transmitting.
To fix this, make the following changes:
(1) rxrpc_queue_packet() now sets a marker in the annotation buffer
instead of setting the RXRPC_CALL_TX_LAST flag.
(2) rxrpc_rotate_tx_window() detects the marker and sets the flag in the
same context as the routines that use it.
(3) rxrpc_end_tx_phase() is simplified to just shift the call state.
The Tx window must have been rotated before calling to discard the
last packet.
(4) rxrpc_receiving_reply() is added to handle the arrival of the first
DATA packet of a reply to a client call (which is an implicit ACK of
the Tx phase).
(5) The last part of rxrpc_input_ack() is reordered to perform Tx
rotation, then soft-ACK application and then to end the phase if we've
rotated the last packet. In the event of a terminal ACK, the soft-ACK
application will be skipped as nAcks should be 0.
(6) rxrpc_input_ackall() now has to rotate as well as ending the phase.
In addition:
(7) Alter the transmit tracepoint to log the rotation of the last packet.
(8) Remove the no-longer relevant queue_reqack tracepoint note. The
ACK-REQUESTED packet header flag is now set as needed when we actually
transmit the packet and may vary by retransmission.
Signed-off-by: David Howells <dhowells@redhat.com>
Fix the call timer in the following ways:
(1) If call->resend_at or call->ack_at are before or equal to the current
time, then ignore that timeout.
(2) If call->expire_at is before or equal to the current time, then don't
set the timer at all (possibly we should queue the call).
(3) Don't skip modifying the timer if timer_pending() is true. This
indicates that the timer is working, not that it has expired and is
running/waiting to run its expiry handler.
Also call rxrpc_set_timer() to start the call timer going rather than
calling add_timer().
Signed-off-by: David Howells <dhowells@redhat.com>
When rxrpc_input_soft_acks() is parsing the soft-ACKs from an ACK packet,
it updates the Tx packet annotations in the annotation buffer. If a
soft-ACK is an ACK, then we overwrite unack'd, nak'd or to-be-retransmitted
states and that is fine; but if the soft-ACK is an NACK, we overwrite the
to-be-retransmitted with a nak - which isn't.
Instead, we need to let any scheduled retransmission stand if the packet
was NAK'd.
Note that we don't reissue a resend if the annotation is in the
to-be-retransmitted state because someone else must've scheduled the
resend already.
Signed-off-by: David Howells <dhowells@redhat.com>
When a DATA packet has its initial transmission, we may need to start or
adjust the resend timer. Without this we end up relying on being sent a
NACK to initiate the resend.
Signed-off-by: David Howells <dhowells@redhat.com>
before_eq() and friends should be used to compare serial numbers (when not
checking for (non)equality) rather than casting to int, subtracting and
checking the result.
Signed-off-by: David Howells <dhowells@redhat.com>
Don't send an IDLE ACK at the end of the transmission of the response to a
service call. The service end resends DATA packets until the client sends an
ACK that hard-acks all the send data. At that point, the call is complete.
Signed-off-by: David Howells <dhowells@redhat.com>
Set the timestamp on sk_buffs holding packets to be transmitted before
queueing them because the moment the packet is on the queue it can be seen
by the retransmission algorithm - which may see a completely random
timestamp.
If the retransmission algorithm sees such a timestamp, it may retransmit
the packet and, in future, tell the congestion management algorithm that
the retransmit timer expired.
Signed-off-by: David Howells <dhowells@redhat.com>
