As explained in commit 9f9843a751 ("tcp: properly handle stretch
acks in slow start"), TCP stacks have to consider how many packets
are acknowledged in one single ACK, because of GRO, but also
because of ACK compression or losses.
We plan to add SACK compression in the following patch, we
must therefore not call tcp_enter_quickack_mode()
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
An RTO event indicates the head has not been acked for a long time
after its last (re)transmission. But the other packets are not
necessarily lost if they have been only sent recently (for example
due to application limit). This patch would prohibit marking packets
sent within an RTT to be lost on RTO event, using similar logic in
TCP RACK detection.
Normally the head (SND.UNA) would be marked lost since RTO should
fire strictly after the head was sent. An exception is when the
most recent RACK RTT measurement is larger than the (previous)
RTO. To address this exception the head is always marked lost.
Congestion control interaction: since we may not mark every packet
lost, the congestion window may be more than 1 (inflight plus 1).
But only one packet will be retransmitted after RTO, since
tcp_retransmit_timer() calls tcp_retransmit_skb(...,segs=1). The
connection still performs slow start from one packet (with Cubic
congestion control).
This commit was tested in an A/B test with Google web servers,
and showed a reduction of 2% in (spurious) retransmits post
timeout (SlowStartRetrans), and correspondingly reduced DSACKs
(DSACKIgnoredOld) by 7%.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Create and export a new helper tcp_rack_skb_timeout and move tcp_is_rack
to prepare the final RTO change.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously when TCP times out, it first updates cwnd and ssthresh,
marks packets lost, and then updates congestion state again. This
was fine because everything not yet delivered is marked lost,
so the inflight is always 0 and cwnd can be safely set to 1 to
retransmit one packet on timeout.
But the inflight may not always be 0 on timeout if TCP changes to
mark packets lost based on packet sent time. Therefore we must
first mark the packet lost, then set the cwnd based on the
(updated) inflight.
This is not a pure refactor. Congestion control may potentially
break if it uses (not yet updated) inflight to compute ssthresh.
Fortunately all existing congestion control modules does not do that.
Also it changes the inflight when CA_LOSS_EVENT is called, and only
westwood processes such an event but does not use inflight.
This change has two other minor side benefits:
1) consistent with Fast Recovery s.t. the inflight is updated
first before tcp_enter_recovery flips state to CA_Recovery.
2) avoid intertwining loss marking with state update, making the
code more readable.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Refactor using a new helper, tcp_timeout_mark_loss(), that marks packets
lost upon RTO.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The previous approach for the lost and retransmit bits was to
wipe the slate clean: zero all the lost and retransmit bits,
correspondingly zero the lost_out and retrans_out counters, and
then add back the lost bits (and correspondingly increment lost_out).
The new approach is to treat this very much like marking packets
lost in fast recovery. We don’t wipe the slate clean. We just say
that for all packets that were not yet marked sacked or lost, we now
mark them as lost in exactly the same way we do for fast recovery.
This fixes the lost retransmit accounting at RTO time and greatly
simplifies the RTO code by sharing much of the logic with Fast
Recovery.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a rewrite of NewReno loss recovery implementation that is
simpler and standalone for readability and better performance by
using less states.
Note that NewReno refers to RFC6582 as a modification to the fast
recovery algorithm. It is used only if the connection does not
support SACK in Linux. It should not to be confused with the Reno
(AIMD) congestion control.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch disables RFC6675 loss detection and make sysctl
net.ipv4.tcp_recovery = 1 controls a binary choice between RACK
(1) or RFC6675 (0).
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for the classic DUPACK threshold rule
(#DupThresh) in RACK.
When the number of packets SACKed is greater or equal to the
threshold, RACK sets the reordering window to zero which would
immediately mark all the unsacked packets below the highest SACKed
sequence lost. Since this approach is known to not work well with
reordering, RACK only uses it if no reordering has been observed.
The DUPACK threshold rule is a particularly useful extension to the
fast recoveries triggered by RACK reordering timer. For example
data-center transfers where the RTT is much smaller than a timer
tick, or high RTT path where the default RTT/4 may take too long.
Note that this patch differs slightly from RFC6675. RFC6675
considers a packet lost when at least #DupThresh higher-sequence
packets are SACKed.
With RACK, for connections that have seen reordering, RACK
continues to use a dynamically-adaptive time-based reordering
window to detect losses. But for connections on which we have not
yet seen reordering, this patch considers a packet lost when at
least one higher sequence packet is SACKed and the total number
of SACKed packets is at least DupThresh. For example, suppose a
connection has not seen reordering, and sends 10 packets, and
packets 3, 5, 7 are SACKed. RFC6675 considers packets 1 and 2
lost. RACK considers packets 1, 2, 4, 6 lost.
There is some small risk of spurious retransmits here due to
reordering. However, this is mostly limited to the first flight of
a connection on which the sender receives SACKs from reordering.
And RFC 6675 and FACK loss detection have a similar risk on the
first flight with reordering (it's just that the risk of spurious
retransmits from reordering was slightly narrower for those older
algorithms due to the margin of 3*MSS).
Also the minimum reordering window is reduced from 1 msec to 0
to recover quicker on short RTT transfers. Therefore RACK is more
aggressive in marking packets lost during recovery to reduce the
reordering window timeouts.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When CONFIG_PROC_FS isn't set, variable ipconfig_dir isn't used.
net/ipv4/ipconfig.c:167:31: warning: ‘ipconfig_dir’ defined but not used [-Wunused-variable]
static struct proc_dir_entry *ipconfig_dir;
^~~~~~~~~~~~
Move the declaration of ipconfig_dir inside the CONFIG_PROC_FS ifdef to
fix the warning.
Fixes: c04d2cb200 ("ipconfig: Write NTP server IPs to /proc/net/ipconfig/ntp_servers")
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The bpf syscall and selftests conflicts were trivial
overlapping changes.
The r8169 change involved moving the added mdelay from 'net' into a
different function.
A TLS close bug fix overlapped with the splitting of the TLS state
into separate TX and RX parts. I just expanded the tests in the bug
fix from "ctx->conf == X" into "ctx->tx_conf == X && ctx->rx_conf
== X".
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently the truncated bit is set only when 1) the mirrored packet
is larger than mtu and 2) the ipv4 packet tot_len is larger than
the actual skb->len. This patch adds another case for detecting
whether ipv6 packet is truncated or not, by checking the ipv6 header
payload_len and the skb->len.
Reported-by: Xiaoyan Jin <xiaoyanj@vmware.com>
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
linux-4.16 got support for softirq based hrtimers.
TCP can switch its pacing hrtimer to this variant, since this
avoids going through a tasklet and some atomic operations.
pacing timer logic looks like other (jiffies based) tcp timers.
v2: use hrtimer_try_to_cancel() in tcp_clear_xmit_timers()
to correctly release reference on socket if needed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix more memory leaks in ip_cmsg_send() callers. Part of them were fixed
earlier in 919483096b.
* udp_sendmsg one was there since the beginning when linux sources were
first added to git;
* ping_v4_sendmsg one was copy/pasted in c319b4d76b.
Whenever return happens in udp_sendmsg() or ping_v4_sendmsg() IP options
have to be freed if they were allocated previously.
Add label so that future callers (if any) can use it instead of kfree()
before return that is easy to forget.
Fixes: c319b4d76b (net: ipv4: add IPPROTO_ICMP socket kind)
Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This version has some suggestions by Eric Dumazet:
- Use a local variable for the mark in IPv6 instead of ctl_sk to avoid SMP
races.
- Use the more elegant "IP4_REPLY_MARK(net, skb->mark) ?: sk->sk_mark"
statement.
- Factorize code as sk_fullsock() check is not necessary.
Aidan McGurn from Openwave Mobility systems reported the following bug:
"Marked routing is broken on customer deployment. Its effects are large
increase in Uplink retransmissions caused by the client never receiving
the final ACK to their FINACK - this ACK misses the mark and routes out
of the incorrect route."
Currently marks are added to sk_buffs for replies when the "fwmark_reflect"
sysctl is enabled. But not for TW sockets that had sk->sk_mark set via
setsockopt(SO_MARK..).
Fix this in IPv4/v6 by adding tw->tw_mark for TIME_WAIT sockets. Copy the the
original sk->sk_mark in __inet_twsk_hashdance() to the new tw->tw_mark location.
Then progate this so that the skb gets sent with the correct mark. Do the same
for resets. Give the "fwmark_reflect" sysctl precedence over sk->sk_mark so that
netfilter rules are still honored.
Signed-off-by: Jon Maxwell <jmaxwell37@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
INET_CSK_DEBUG is always set and only is used for 2 pr_debug calls.
EXPORT_SYMBOL(inet_csk_timer_bug_msg) is only used by these 2
pr_debug calls and is also unnecessary as the exported string can
be used directly by these calls.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Damir reported a breakage of SO_BINDTODEVICE for UDP sockets.
In absence of VRF devices, after commit fb74c27735 ("net:
ipv4: add second dif to udp socket lookups") the dif mismatch
isn't fatal anymore for UDP socket lookup with non null
sk_bound_dev_if, breaking SO_BINDTODEVICE semantics.
This changeset addresses the issue making the dif match mandatory
again in the above scenario.
Reported-by: Damir Mansurov <dnman@oktetlabs.ru>
Fixes: fb74c27735 ("net: ipv4: add second dif to udp socket lookups")
Fixes: 1801b570dd ("net: ipv6: add second dif to udp socket lookups")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After route cache is flushed via ipv4_sysctl_rtcache_flush(), we forget
to reset fnhe_mtu_locked in rt_bind_exception(). When pmtu is updated
in __ip_rt_update_pmtu(), it will return directly since the pmtu is
still locked. e.g.
+ ip netns exec client ping 10.10.1.1 -c 1 -s 1400 -M do
PING 10.10.1.1 (10.10.1.1) 1400(1428) bytes of data.
>From 10.10.0.254 icmp_seq=1 Frag needed and DF set (mtu = 0)
Signed-off-by: David S. Miller <davem@davemloft.net>
No changes in refcount semantics -- key init is false; replace
static_key_enable with static_branch_enable
static_key_slow_inc|dec with static_branch_inc|dec
static_key_false with static_branch_unlikely
Added a '_key' suffix to udp and udpv6 encap_needed, for better
self documentation.
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
No changes in refcount semantics -- key init is false; replace
static_key_slow_inc|dec with static_branch_inc|dec
static_key_false with static_branch_unlikely
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch makes it so that if a destructor is not present we avoid trying
to update the skb socket or any reference counting that would be associated
with the NULL socket and/or descriptor. By doing this we can support
traffic coming from another namespace without any issues.
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for a software provided checksum and GSO_PARTIAL
segmentation support. With this we can offload UDP segmentation on devices
that only have partial support for tunnels.
Since we are no longer needing the hardware checksum we can drop the checks
in the segmentation code that were verifying if it was present.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows us to take care of unrolling the first segment and the
last segment of the loop for processing the segmented skb. Part of the
motivation for this is that it makes it easier to process the fact that the
first fame and all of the frames in between should be mostly identical
in terms of header data, and the last frame has differences in the length
and partial checksum.
In addition I am dropping the header length calculation since we don't
really need it for anything but the last frame and it can be easily
obtained by just pulling the data_len and offset of tail from the transport
header.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is meant to allow us to avoid having to recompute the checksum
from scratch and have it passed as a parameter.
Instead of taking that approach we can take advantage of the fact that the
length that was used to compute the existing checksum is included in the
UDP header.
Finally to avoid the need to invert the result we can just call csum16_add
and csum16_sub directly. By doing this we can avoid a number of
instructions in the loop that is handling segmentation.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is no point in passing MSS as a parameter for for the GSO
segmentation call as it is already available via the shared info for the
skb itself.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We need to record the number of segments that will be generated when this
frame is segmented. The expectation is that if gso_size is set then
gso_segs is set as well. Without this some drivers such as ixgbe get
confused if they attempt to offload this as they record 0 segments for the
entire packet instead of the correct value.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter/IPVS updates for net-next
The following patchset contains Netfilter/IPVS updates for your net-next
tree, more relevant updates in this batch are:
1) Add Maglev support to IPVS. Moreover, store lastest server weight in
IPVS since this is needed by maglev, patches from from Inju Song.
2) Preparation works to add iptables flowtable support, patches
from Felix Fietkau.
3) Hand over flows back to conntrack slow path in case of TCP RST/FIN
packet is seen via new teardown state, also from Felix.
4) Add support for extended netlink error reporting for nf_tables.
5) Support for larger timeouts that 23 days in nf_tables, patch from
Florian Westphal.
6) Always set an upper limit to dynamic sets, also from Florian.
7) Allow number generator to make map lookups, from Laura Garcia.
8) Use hash_32() instead of opencode hashing in IPVS, from Vicent Bernat.
9) Extend ip6tables SRH match to support previous, next and last SID,
from Ahmed Abdelsalam.
10) Move Passive OS fingerprint nf_osf.c, from Fernando Fernandez.
11) Expose nf_conntrack_max through ctnetlink, from Florent Fourcot.
12) Several housekeeping patches for xt_NFLOG, x_tables and ebtables,
from Taehee Yoo.
13) Unify meta bridge with core nft_meta, then make nft_meta built-in.
Make rt and exthdr built-in too, again from Florian.
14) Missing initialization of tbl->entries in IPVS, from Cong Wang.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow some non-cached routes to use non-expired fnhe:
1. ip_del_fnhe: moved above and now called by find_exception.
The 4.5+ commit deed49df73 expires fnhe only when caching
routes. Change that to:
1.1. use fnhe for non-cached local output routes, with the help
from (2)
1.2. allow __mkroute_input to detect expired fnhe (outdated
fnhe_gw, for example) when do_cache is false, eg. when itag!=0
for unicast destinations.
2. __mkroute_output: keep fi to allow local routes with orig_oif != 0
to use fnhe info even when the new route will not be cached into fnhe.
After commit 839da4d989 ("net: ipv4: set orig_oif based on fib
result for local traffic") it means all local routes will be affected
because they are not cached. This change is used to solve a PMTU
problem with IPVS (and probably Netfilter DNAT) setups that redirect
local clients from target local IP (local route to Virtual IP)
to new remote IP target, eg. IPVS TUN real server. Loopback has
64K MTU and we need to create fnhe on the local route that will
keep the reduced PMTU for the Virtual IP. Without this change
fnhe_pmtu is updated from ICMP but never exposed to non-cached
local routes. This includes routes with flowi4_oif!=0 for 4.6+ and
with flowi4_oif=any for 4.14+).
3. update_or_create_fnhe: make sure fnhe_expires is not 0 for
new entries
Fixes: 839da4d989 ("net: ipv4: set orig_oif based on fib result for local traffic")
Fixes: d6d5e999e5 ("route: do not cache fib route info on local routes with oif")
Fixes: deed49df73 ("route: check and remove route cache when we get route")
Cc: David Ahern <dsahern@gmail.com>
Cc: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously the bbr->idle_restart tracking was zeroing out the
bbr->idle_restart bit upon ACKs that did not SACK or ACK anything,
e.g. receiving incoming data or receiver window updates. In such
situations BBR would forget that this was a restart-from-idle
situation, and if the min_rtt had expired it would unnecessarily enter
PROBE_RTT (even though we were actually restarting from idle but had
merely forgotten that fact).
The fix is simple: we need to remember we are restarting from idle
until we receive a S/ACK for some data (a S/ACK for the first flight
of data we send as we are restarting).
This commit is a stable candidate for kernels back as far as 4.9.
Fixes: 0f8782ea14 ("tcp_bbr: add BBR congestion control")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: Yousuk Seung <ysseung@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Using the udp_v4_check() function to calculate the pseudo header
for the newly segmented UDP packets results in assigning the complement
of the value to the UDP header checksum field.
Always undo the complement the partial checksum value in order to
match the case where GSO is not used on the UDP transmit path.
Fixes: ee80d1ebe5 ("udp: add udp gso")
Signed-off-by: Sean Tranchetti <stranche@codeaurora.org>
Signed-off-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Applications with many concurrent connections, high variance
in receive queue length and tight memory bounds cannot
allocate worst-case buffer size to drain sockets. Knowing
the size of receive queue length, applications can optimize
how they allocate buffers to read from the socket.
The number of bytes pending on the socket is directly
available through ioctl(FIONREAD/SIOCINQ) and can be
approximated using getsockopt(MEMINFO) (rmem_alloc includes
skb overheads in addition to application data). But, both of
these options add an extra syscall per recvmsg. Moreover,
ioctl(FIONREAD/SIOCINQ) takes the socket lock.
Add the TCP_INQ socket option to TCP. When this socket
option is set, recvmsg() relays the number of bytes available
on the socket for reading to the application via the
TCP_CM_INQ control message.
Calculate the number of bytes after releasing the socket lock
to include the processed backlog, if any. To avoid an extra
branch in the hot path of recvmsg() for this new control
message, move all cmsg processing inside an existing branch for
processing receive timestamps. Since the socket lock is not held
when calculating the size of receive queue, TCP_INQ is a hint.
For example, it can overestimate the queue size by one byte,
if FIN is received.
With this method, applications can start reading from the socket
using a small buffer, and then use larger buffers based on the
remaining data when needed.
V3 change-log:
As suggested by David Miller, added loads with barrier
to check whether we have multiple threads calling recvmsg
in parallel. When that happens we lock the socket to
calculate inq.
V4 change-log:
Removed inline from a static function.
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Suggested-by: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Syzbot managed to send a udp gso packet without checksum offload into
the gso stack by disabling tx checksum (UDP_NO_CHECK6_TX). This
triggered the skb_warn_bad_offload.
RIP: 0010:skb_warn_bad_offload+0x2bc/0x600 net/core/dev.c:2658
skb_gso_segment include/linux/netdevice.h:4038 [inline]
validate_xmit_skb+0x54d/0xd90 net/core/dev.c:3120
__dev_queue_xmit+0xbf8/0x34c0 net/core/dev.c:3577
dev_queue_xmit+0x17/0x20 net/core/dev.c:3618
UDP_NO_CHECK6_TX sets skb->ip_summed to CHECKSUM_NONE just after the
udp gso integrity checks in udp_(v6_)send_skb. Extend those checks to
catch and fail in this case.
After the integrity checks jump directly to the CHECKSUM_PARTIAL case
to avoid reading the no_check_tx flags again (a TOCTTOU race).
Fixes: bec1f6f697 ("udp: generate gso with UDP_SEGMENT")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Called when a TCP segment is acknowledged.
Could be used by application protocols who hold additional
metadata associated with the stream data.
This is required by TLS device offload to release
metadata associated with acknowledged TLS records.
Signed-off-by: Ilya Lesokhin <ilyal@mellanox.com>
Signed-off-by: Boris Pismenny <borisp@mellanox.com>
Signed-off-by: Aviad Yehezkel <aviadye@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently the truncated bit is set only when the mirrored packet
is larger than mtu. For certain cases, the packet might already
been truncated before sending to the erspan tunnel. In this case,
the patch detect whether the IP header's total length is larger
than the actual skb->len. If true, this indicated that the
mirrored packet is truncated and set the erspan truncate bit.
I tested the patch using bpf_skb_change_tail helper function to
shrink the packet size and send to erspan tunnel.
Reported-by: Xiaoyan Jin <xiaoyanj@vmware.com>
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When adding tcp mmap() implementation, I forgot that socket lock
had to be taken before current->mm->mmap_sem. syzbot eventually caught
the bug.
Since we can not lock the socket in tcp mmap() handler we have to
split the operation in two phases.
1) mmap() on a tcp socket simply reserves VMA space, and nothing else.
This operation does not involve any TCP locking.
2) getsockopt(fd, IPPROTO_TCP, TCP_ZEROCOPY_RECEIVE, ...) implements
the transfert of pages from skbs to one VMA.
This operation only uses down_read(¤t->mm->mmap_sem) after
holding TCP lock, thus solving the lockdep issue.
This new implementation was suggested by Andy Lutomirski with great details.
Benefits are :
- Better scalability, in case multiple threads reuse VMAS
(without mmap()/munmap() calls) since mmap_sem wont be write locked.
- Better error recovery.
The previous mmap() model had to provide the expected size of the
mapping. If for some reason one part could not be mapped (partial MSS),
the whole operation had to be aborted.
With the tcp_zerocopy_receive struct, kernel can report how
many bytes were successfuly mapped, and how many bytes should
be read to skip the problematic sequence.
- No more memory allocation to hold an array of page pointers.
16 MB mappings needed 32 KB for this array, potentially using vmalloc() :/
- skbs are freed while mmap_sem has been released
Following patch makes the change in tcp_mmap tool to demonstrate
one possible use of mmap() and setsockopt(... TCP_ZEROCOPY_RECEIVE ...)
Note that memcg might require additional changes.
Fixes: 93ab6cc691 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Suggested-by: Andy Lutomirski <luto@kernel.org>
Cc: linux-mm@kvack.org
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
UDP GSO needs to export __udp_gso_segment to call it from ipv6.
I accidentally exported static ipv4 function __udp4_gso_segment.
Remove that EXPORT_SYMBOL_GPL.
Fixes: ee80d1ebe5 ("udp: add udp gso")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In tcp_select_initial_window(), we only set rcv_wnd to
tcp_default_init_rwnd() if current mss > (1 << wscale). Otherwise,
rcv_wnd is kept at the full receive space of the socket which is a
value way larger than tcp_default_init_rwnd().
With larger initial rcv_wnd value, receive buffer autotuning logic
takes longer to kick in and increase the receive buffer.
In a TCP throughput test where receiver has rmem[2] set to 125MB
(wscale is 11), we see the connection gets recvbuf limited at the
beginning of the connection and gets less throughput overall.
Signed-off-by: Wei Wang <weiwan@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The TCP repair sequence of operation is to first set the socket in
repair mode, then inject the TCP stats into the socket with repair
socket options, then call connect() to re-activate the socket. The
connect syscall simply returns and set state to ESTABLISHED
mode. As a result Fast Open is meaningless for TCP repair.
However allowing sendto() system call with MSG_FASTOPEN flag half-way
during the repair operation could unexpectedly cause data to be
sent, before the operation finishes changing the internal TCP stats
(e.g. MSS). This in turn triggers TCP warnings on inconsistent
packet accounting.
The fix is to simply disallow Fast Open operation once the socket
is in the repair mode.
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow specifying segment size in the send call.
The new control message performs the same function as socket option
UDP_SEGMENT while avoiding the extra system call.
[ Export udp_cmsg_send for ipv6. -DaveM ]
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When sending large datagrams that are later segmented, store data in
page frags to avoid copying from linear in skb_segment.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb_segment by default transfers allocated wmem from the gso skb
to the tail of the segment list. This underreports real truesize
of the list, especially if the tail might be dropped.
Similar to tcp_gso_segment, update wmem_alloc with the aggregate
list truesize and make each segment responsible for its own
share by setting skb->destructor.
Clear gso_skb->destructor prior to calling skb_segment to skip
the default assignment to tail.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Support generic segmentation offload for udp datagrams. Callers can
concatenate and send at once the payload of multiple datagrams with
the same destination.
To set segment size, the caller sets socket option UDP_SEGMENT to the
length of each discrete payload. This value must be smaller than or
equal to the relevant MTU.
A follow-up patch adds cmsg UDP_SEGMENT to specify segment size on a
per send call basis.
Total byte length may then exceed MTU. If not an exact multiple of
segment size, the last segment will be shorter.
The implementation adds a gso_size field to the udp socket, ip(v6)
cmsg cookie and inet_cork structure to be able to set the value at
setsockopt or cmsg time and to work with both lockless and corked
paths.
Initial benchmark numbers show UDP GSO about as expensive as TCP GSO.
tcp tso
3197 MB/s 54232 msg/s 54232 calls/s
6,457,754,262 cycles
tcp gso
1765 MB/s 29939 msg/s 29939 calls/s
11,203,021,806 cycles
tcp without tso/gso *
739 MB/s 12548 msg/s 12548 calls/s
11,205,483,630 cycles
udp
876 MB/s 14873 msg/s 624666 calls/s
11,205,777,429 cycles
udp gso
2139 MB/s 36282 msg/s 36282 calls/s
11,204,374,561 cycles
[*] after reverting commit 0a6b2a1dc2
("tcp: switch to GSO being always on")
Measured total system cycles ('-a') for one core while pinning both
the network receive path and benchmark process to that core:
perf stat -a -C 12 -e cycles \
./udpgso_bench_tx -C 12 -4 -D "$DST" -l 4
Note the reduction in calls/s with GSO. Bytes per syscall drops
increases from 1470 to 61818.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement generic segmentation offload support for udp datagrams. A
follow-up patch adds support to the protocol stack to generate such
packets.
UDP GSO is not UFO. UFO fragments a single large datagram. GSO splits
a large payload into a number of discrete UDP datagrams.
The implementation adds a GSO type SKB_UDP_GSO_L4 to differentiate it
from UFO (SKB_UDP_GSO).
IPPROTO_UDPLITE is excluded, as that protocol has no gso handler
registered.
[ Export __udp_gso_segment for ipv6. -DaveM ]
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
UDP segmentation offload needs access to inet_cork in the udp layer.
Pass the struct to ip(6)_make_skb instead of allocating it on the
stack in that function itself.
This patch is a noop otherwise.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet