In some circumstances, intermediate devices may lose state in a way
that temporarily prevents the successful delivery of packets from a
TCP peer. For example, a firewall may drop a NAT forwarding table
entry.
Since iPXE spends most of its time downloading files (and hence purely
receiving data, sending only TCP ACKs), this can easily happen in a
situation in which there is no reason for iPXE's TCP stack to generate
any retransmissions. The temporary loss of connectivity can therefore
effectively become permanent.
Work around this problem by sending TCP keepalives after a period of
inactivity on an established connection.
TCP keepalives usually send a single garbage byte in sequence number
space that has already been ACKed by the peer. Since we do not need
to elicit a response from the peer, we instead send pure ACKs (with no
garbage data) in order to keep the transmit code path simple.
Originally-implemented-by: Ladi Prosek <lprosek@redhat.com>
Debugged-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
We currently do not wait for a received FIN before exiting to boot a
loaded OS. In the common case of booting from an HTTP server, this
means that the TCP connection is left consuming resources on the
server side: the server will retransmit the FIN several times before
giving up.
Fix by initiating a graceful close of all TCP connections and waiting
(for up to one second) for all connections to finish closing
gracefully (i.e. for the outgoing FIN to have been sent and ACKed, and
for the incoming FIN to have been received and ACKed at least once).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
We currently shrink the TCP window permanently if we are ever forced
(by a low-memory condition) to discard a previously received TCP
packet. This behaviour was intended to reduce the number of
retransmissions in a lossy network, since lost packets might
potentially result in the entire window contents being retransmitted.
Since commit e0fc8fe ("[tcp] Implement support for TCP Selective
Acknowledgements (SACK)") the cost of lost packets has been reduced by
around one order of magnitude, and the reduction in the window size
(which affects the maximum throughput) is now the more significant
cost.
Remove the code which reduces the TCP maximum window size when a
received packet is discarded.
Reported-by: Wissam Shoukair <wissams@mellanox.com>
Tested-by: Wissam Shoukair <wissams@mellanox.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The TCP Selective Acknowledgement option (specified in RFC2018)
provides a mechanism for the receiver to indicate packets that have
been received out of order (e.g. due to earlier dropped packets).
iPXE often operates in environments in which there is a high
probability of packet loss. For example, the legacy USB keyboard
emulation in some BIOSes involves polling the USB bus from within a
system management interrupt: this introduces an invisible delay of
around 500us which is long enough for around 40 full-length packets to
be dropped. Similarly, almost all 1Gbps USB2 devices will eventually
end up dropping packets because the USB2 bus does not provide enough
bandwidth to sustain a 1Gbps stream, and most devices will not provide
enough internal buffering to hold a full TCP window's worth of
received packets.
Add support for sending TCP Selective Acknowledgements. This provides
the sender with more detailed information about which packets have
been lost, and so allows for a more efficient retransmission strategy.
We include a SACK-permitted option in our SYN packet, since
experimentation shows that at least Linux peers will not include a
SACK-permitted option in the SYN-ACK packet if one was not present in
the initial SYN. (RFC2018 does not seem to mandate this behaviour,
but it is consistent with the approach taken in RFC1323.) We ignore
any received SACK options; this is safe to do since SACK is only ever
advisory and we never have to send non-trivial amounts of data.
Since our TCP receive queue is a candidate for cache discarding under
low memory conditions, we may end up discarding data that has been
reported as received via a SACK option. This is permitted by RFC2018.
We follow the stricture that SACK blocks must not report data which is
no longer held by the receiver: previously-reported blocks are
validated against the current receive queue before being included
within the current SACK block list.
Experiments in a qemu VM using forced packet drops (by setting
NETDEV_DISCARD_RATE to 32) show that implementing SACK improves
throughput by around 400%.
Experiments with a USB2 NIC (an SMSC7500) show that implementing SACK
improves throughput by around 700%, increasing the download rate from
35Mbps up to 250Mbps (which is approximately the usable bandwidth
limit for USB2).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
These files cannot be automatically relicensed by util/relicense.pl
since they either contain unusual but trivial contributions (such as
the addition of __nonnull function attributes), or contain lines
dating back to the initial git revision (and so require manual
knowledge of the code's origin).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When running inside a virtual machine (or when using the UNDI driver),
transmitting packets can be expensive. When we receive several
packets in one poll (e.g. because a slow BIOS timer interrupt routine
has caused us to fall behind in processing), we can safely send just a
single ACK to cover all of the received packets. This reduces the
time spent transmitting and allows us to clear the backlog much
faster.
Various RFCs (starting with RFC1122) state that there should be an ACK
for at least every second segment. We choose not to enforce this
rule. Under normal operation each poll should find at most one
received packet, and we will then not delay any ACKs. We delay
(i.e. omit) ACKs only when under sufficiently heavy load that we are
finding multiple packets per poll; under these conditions it is
important to clear the backlog quickly since any delay may lead to
dropped packets.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently ignores ACKs which do not acknowledge any new data.
(In particular, it does not stop the retransmission timer; this is
done to prevent an immediate retransmission if a duplicate ACK is
received while the transmit queue is non-empty.)
If a peer provides a window size of zero and later sends a duplicate
ACK to update the window size, this update will therefore be ignored
and iPXE will never be able to transmit data.
Fix by updating the window size even for ACKs which do not acknowledge
new data.
Reported-by: Wissam Shoukair <wissams@mellanox.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently advertises a fixed MSS of 1460, which is correct only
for IPv4 over Ethernet. For IPv6 over Ethernet, the value should be
1440 (allowing for the larger IPv6 header). For non-Ethernet link
layers, the value should reflect the MTU of the underlying network
device.
Use tcpip_mtu() to calculate the transport-layer MTU associated with
the peer address, and calculate the MSS to allow for an optionless TCP
header as per RFC 6691.
As a side benefit, we can now fail a connection immediately with a
meaningful error message if we have no route to the destination
address.
Reported-by: Anton D. Kachalov <mouse@yandex-team.ru>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
On large networks with substantial numbers of monitoring agents,
unwanted TCP connection attempts may end up flooding iPXE's ARP cache.
Fix by silently dropping packets received for unrecognised TCP
connections. This should not cause problems, since many firewalls
will also silently drop any such packets.
Reported-by: Jarrod Johnson <jarrod.b.johnson@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Whenever memory pressure causes a queued packet to be discarded (and
so retransmitted), reduce the maximum TCP window to a size that would
have prevented the discard.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Discarding the active ARP cache entry in the middle of a download will
substantially disrupt the TCP stream. Try to minimise any such
disruption by treating ARP cache entries as expensive, and discarding
them only when nothing else is available to discard.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Commit ea61075 ("[tcp] Add support for TCP window scaling") introduced
a potential NULL pointer dereference by referring to the connection's
send window scale before checking whether or not the connection is
known.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The maximum unscaled TCP window (64kB) implies a maximum bandwidth of
around 300kB/s on a WAN link with an RTT of 200ms. Add support for
the TCP window scaling option to remove this upper limit.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
TCP currently neglects to allow sufficient space for its own headers
when allocating I/O buffers. This problem is masked by the fact that
the maximum link-layer header size (802.11) is substantially larger
than the common Ethernet link-layer header.
Fix by allowing sufficient space for any TCP headers, as well as the
network-layer and link-layer headers.
Reported-by: Scott K Logan <logans@cottsay.net>
Debugged-by: Scott K Logan <logans@cottsay.net>
Tested-by: Scott K Logan <logans@cottsay.net>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Commit 3f442d3 ("[tcp] Record ts_recent on first received packet")
failed to achieve its stated intention.
Fix this (and reduce the code size) by moving the ts_recent update to
tcp_rx_seq(). This is the code responsible for advancing the window,
called by both tcp_rx_syn() and tcp_rx_data(), and so the window check
is now redundant.
Reported-by: Frank Weed <zorbustheknight@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Commit 6861304 ("[tcp] Handle out-of-order received packets")
introduced a regression in which ts_recent would not be updated until
the first packet is received in the ESTABLISHED state, i.e. the
timestamp from the SYN+ACK packet would be ignored. This causes the
connection to be dropped by strictly-conforming TCP peers, such as
FreeBSD.
Fix by delaying the timestamp window check until after processing the
received SYN flag.
Reported-by: winders@sonnet.com
Signed-off-by: Michael Brown <mcb30@ipxe.org>
There are several points in the iPXE codebase where
list_for_each_entry() is (ab)used to extract only the first entry from
a list. Add a macro list_first_entry() to make this code easier to
read.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Guarantee that a retry timer cannot go out of scope while the timer is
running, and provide a guarantee to the expiry callback that the timer
will remain in scope during the entire callback (similar to the
guarantee provided to interface methods).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Allow packets in the receive queue to be discarded in order to free up
memory. This avoids a potential deadlock condition in which the
missing packet can never be received because the receive queue is
occupying all of the memory available for further RX buffers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Maintain a queue of received packets, so that lost packets need not
result in retransmission of the entire TCP window.
Increase the TCP window to 8kB, in order that we can potentially
transmit enough duplicate ACKs to trigger Fast Retransmission at the
sender.
Using a 10MB HTTP download in qemu-kvm with an artificial drop rate of
1 in 64 packets, this reduces the download time from around 26s to
around 4s.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently forces sending (i.e. sends a pure ACK even in the
absence of fresh data to send) only in response to packets that
consume sequence space or that lie outside of the receive window.
This ignores the possibility that a previous ACK was not actually sent
(due to, for example, the retransmission timer running).
This does not cause incorrect behaviour, but does cause unnecessary
retransmissions from our peer. For example:
1. Peer sends final data packet (ack 106 seq 521..523)
2. We send FIN (seq 106..107 ack 523)
3. Peer sends FIN (ack 106 seq 523..524)
4. We send nothing since retransmission timer is running for our FIN
5. Peer ACKs our FIN (ack 107 seq 524..524)
6. We send nothing since this packet consumes no sequence space
7. Peer retransmits FIN (ack 107 seq 523..524)
8. We ACK peer's FIN (seq 107..107 ack 524)
What should happen at step (6) is that we should ACK the peer's FIN,
since we can deduce that we have never sent this ACK.
Fix by maintaining an "ACK pending" flag that is set whenever we are
made aware that our peer needs an ACK (whether by consuming sequence
space or by sending a packet that appears out of order), and is
cleared only when the ACK packet has been transmitted.
Reported-by: Piotr Jaroszyński <p.jaroszynski@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently repurposes the retransmission timer to hold the TCP
connection in the TIME_WAIT state (i.e. waiting for up to 2*MSL in
case we are required to re-ACK our peer's FIN due to a lost ACK).
However, the fact that this timer is running will prevent such an ACK
from ever being sent, since the logic in tcp_xmit() assumes that a
running timer indicates that we ourselves are waiting for an ACK and
so blocks the transmission. (We always wait for an ACK before sending
our next packet, to keep our transmit data path as simple as
possible.)
Fix by using an entirely separate timer for the TIME_WAIT state, so
that packets can still be sent.
Reported-by: Piotr Jaroszyński <p.jaroszynski@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Every other scalar integer value in struct tcp_connection is in host
byte order; change the definition of local_port to match.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Remove data-xfer as an interface type, and replace data-xfer
interfaces with generic interfaces supporting the data-xfer methods.
Filter interfaces (as used by the TLS layer) are handled using the
generic pass-through interface capability. A side-effect of this is
that deliver_raw() no longer exists as a data-xfer method. (In
practice this doesn't lose any efficiency, since there are no
instances within the current codebase where xfer_deliver_raw() is used
to pass data to an interface supporting the deliver_raw() method.)
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Standardise on using timer_init() to initialise an embedded retry
timer, to match the coding style used by other embedded objects.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Standardise on using ref_init() to initialise an embedded reference
count, to match the coding style used by other embedded objects.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently updates the TCP sequence number after delivering the
data to the application via xfer_deliver_iob(). If the application
responds to the received data by transmitting more data, this would
result in a stale ACK number appearing in the transmitted packet,
which potentially causes retransmissions and also gives the
undesirable appearance of violating causality (by sending a response
to a message that we claim not to have yet received).
Reported-by: Guo-Fu Tseng <cooldavid@cooldavid.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Access to the gpxe.org and etherboot.org domains and associated
resources has been revoked by the registrant of the domain. Work
around this problem by renaming project from gPXE to iPXE, and
updating URLs to match.
Also update README, LOG and COPYRIGHTS to remove obsolete information.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
In several places, we currently use size_t to represent a difference
between TCP sequence numbers. This can cause compiler warnings
relating to printf format specifiers, since the result of
(uint32_t+size_t) may be an unsigned long on some compilers.
Fix by using uint32_t for all variables that represent a difference
between TCP sequence numbers.
Tested-by: Joshua Oreman <oremanj@xenon.get-linux.org>
Michael Brown