linux_dsm_epyc7002/net
Neal Cardwell 299bcb55ec tcp: fix cwnd-limited bug for TSO deferral where we send nothing
When cwnd is not a multiple of the TSO skb size of N*MSS, we can get
into persistent scenarios where we have the following sequence:

(1) ACK for full-sized skb of N*MSS arrives
  -> tcp_write_xmit() transmit full-sized skb with N*MSS
  -> move pacing release time forward
  -> exit tcp_write_xmit() because pacing time is in the future

(2) TSQ callback or TCP internal pacing timer fires
  -> try to transmit next skb, but TSO deferral finds remainder of
     available cwnd is not big enough to trigger an immediate send
     now, so we defer sending until the next ACK.

(3) repeat...

So we can get into a case where we never mark ourselves as
cwnd-limited for many seconds at a time, even with
bulk/infinite-backlog senders, because:

o In case (1) above, every time in tcp_write_xmit() we have enough
cwnd to send a full-sized skb, we are not fully using the cwnd
(because cwnd is not a multiple of the TSO skb size). So every time we
send data, we are not cwnd limited, and so in the cwnd-limited
tracking code in tcp_cwnd_validate() we mark ourselves as not
cwnd-limited.

o In case (2) above, every time in tcp_write_xmit() that we try to
transmit the "remainder" of the cwnd but defer, we set the local
variable is_cwnd_limited to true, but we do not send any packets, so
sent_pkts is zero, so we don't call the cwnd-limited logic to update
tp->is_cwnd_limited.

Fixes: ca8a226343 ("tcp: make cwnd-limited checks measurement-based, and gentler")
Reported-by: Ingemar Johansson <ingemar.s.johansson@ericsson.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20201209035759.1225145-1-ncardwell.kernel@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2020-12-09 16:15:54 -08:00
..
6lowpan
9p
802
8021q
appletalk
atm net: atm: fix update of position index in lec_seq_next 2020-10-31 12:26:30 -07:00
ax25
batman-adv batman-adv: Don't always reallocate the fragmentation skb head 2020-11-27 08:02:55 +01:00
bluetooth Bluetooth: Replace zero-length array with flexible-array member 2020-10-29 17:22:59 -05:00
bpf
bpfilter
bridge bridge: Fix a deadlock when enabling multicast snooping 2020-12-07 17:14:43 -08:00
caif
can can: isotp: isotp_setsockopt(): block setsockopt on bound sockets 2020-12-09 08:44:15 -08:00
ceph
core net: flow_offload: Fix memory leak for indirect flow block 2020-12-09 16:08:33 -08:00
dcb
dccp tcp: fix race condition when creating child sockets from syncookies 2020-11-23 16:32:33 -08:00
decnet
dns_resolver
dsa
ethernet
ethtool ethtool: fix stack overflow in ethnl_parse_bitset() 2020-12-09 15:50:38 -08:00
hsr
ieee802154
ife
ipv4 tcp: fix cwnd-limited bug for TSO deferral where we send nothing 2020-12-09 16:15:54 -08:00
ipv6 tcp: Retain ECT bits for tos reflection 2020-12-09 16:08:23 -08:00
iucv net/af_iucv: set correct sk_protocol for child sockets 2020-11-21 14:43:45 -08:00
kcm
key
l2tp
l3mdev
lapb
llc
mac80211 mac80211: mesh: fix mesh_pathtbl_init() error path 2020-12-04 17:34:25 -08:00
mac802154
mpls mpls: load mpls_gso after mpls_iptunnel 2020-10-20 21:16:45 -07:00
mptcp mptcp: print new line in mptcp_seq_show() if mptcp isn't in use 2020-12-07 17:45:29 -08:00
ncsi net/ncsi: Fix netlink registration 2020-11-12 17:00:13 -08:00
netfilter netfilter: nftables_offload: build mask based from the matching bytes 2020-11-27 12:10:47 +01:00
netlabel netlabel: fix an uninitialized warning in netlbl_unlabel_staticlist() 2020-11-14 12:07:57 -08:00
netlink
netrom
nfc nfc: Ensure presence of NFC_ATTR_FIRMWARE_NAME attribute in nfc_genl_fw_download() 2020-10-20 17:06:22 -07:00
nsh
openvswitch openvswitch: fix error return code in validate_and_copy_dec_ttl() 2020-12-04 15:43:14 -08:00
packet net/packet: fix packet receive on L3 devices without visible hard header 2020-11-23 17:29:36 -08:00
phonet
psample
qrtr
rds RDMA: Add rdma_connect_locked() 2020-10-28 09:14:49 -03:00
rfkill rfkill: Fix use-after-free in rfkill_resume() 2020-11-12 09:18:06 +01:00
rose rose: Fix Null pointer dereference in rose_send_frame() 2020-11-20 10:04:58 -08:00
rxrpc
sched net/sched: fq_pie: initialize timer earlier in fq_pie_init() 2020-12-04 14:15:01 -08:00
sctp sctp: change to hold/put transport for proto_unreach_timer 2020-11-14 11:57:12 -08:00
smc net/smc: fix direct access to ib_gid_addr->ndev in smc_ib_determine_gid() 2020-11-19 10:59:19 -08:00
strparser
sunrpc net/sunrpc: fix useless comparison in proc_do_xprt() 2020-11-08 16:28:25 -05:00
switchdev
tipc net: tipc: prevent possible null deref of link 2020-12-08 15:53:41 -08:00
tls net/tls: Protect from calling tls_dev_del for TLS RX twice 2020-11-25 17:31:06 -08:00
unix
vmw_vsock vsock/virtio: discard packets only when socket is really closed 2020-11-23 16:36:29 -08:00
wimax
wireless cfg80211: initialize rekey_data 2020-12-04 12:35:58 +01:00
x25 net/x25: prevent a couple of overflows 2020-12-02 17:26:36 -08:00
xdp net, xsk: Avoid taking multiple skbuff references 2020-11-24 22:39:56 +01:00
xfrm net: xfrm: fix memory leak in xfrm_user_policy() 2020-11-10 09:14:25 +01:00
compat.c
devres.c
Kconfig
Makefile
socket.c
sysctl_net.c