linux/net/ipv4
Ilpo Järvinen f785a8e28b [TCP]: Fix lost_retrans loop vs fastpath problems
Detection implemented with lost_retrans must work also when
fastpath is taken, yet most of the queue is skipped including
(very likely) those retransmitted skb's we're interested in.
This problem appeared when the hints got added, which removed
a need to always walk over the whole write queue head.
Therefore decicion for the lost_retrans worker loop entry must
be separated from the sacktag processing more than it was
necessary before.

It turns out to be problematic to optimize the worker loop
very heavily because ack_seqs of skb may have a number of
discontinuity points. Maybe similar approach as currently is
implemented could be attempted but that's becoming more and
more complex because the trend is towards less skb walking
in sacktag marker. Trying a simple work until all rexmitted
skbs heve been processed approach.

Maybe after(highest_sack_end_seq, tp->high_seq) checking is not
sufficiently accurate and causes entry too often in no-work-to-do
cases. Since that's not known, I've separated solution to that
from this patch.

Noticed because of report against a related problem from TAKANO
Ryousei <takano@axe-inc.co.jp>. He also provided a patch to
that part of the problem. This patch includes solution to it
(though this patch has to use somewhat different placement).
TAKANO's description and patch is available here:

  http://marc.info/?l=linux-netdev&m=119149311913288&w=2

...In short, TAKANO's problem is that end_seq the loop is using
not necessarily the largest SACK block's end_seq because the
current ACK may still have higher SACK blocks which are later
by the loop.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-11 17:35:41 -07:00
..
ipvs [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
netfilter [NET]: make netlink user -> kernel interface synchronious 2007-10-10 21:15:29 -07:00
Kconfig [NET]: Generic Large Receive Offload for TCP traffic 2007-10-10 16:47:46 -07:00
Makefile [NET]: Generic Large Receive Offload for TCP traffic 2007-10-10 16:47:46 -07:00
af_inet.c [IPV4]: Add ICMPMsgStats MIB (RFC 4293) 2007-10-10 16:51:28 -07:00
ah4.c [IPSEC]: Move IP protocol setting from transforms into xfrm4_input.c 2007-10-10 16:55:56 -07:00
arp.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
cipso_ipv4.c [CIPSO]: remove duplicated code in the cipso_v4_*_getattr() functions 2007-10-10 16:51:17 -07:00
datagram.c [IPV4]: Fix "ipOutNoRoutes" counter error for TCP and UDP 2007-06-03 18:08:50 -07:00
devinet.c [IPV4]: When possible test for IFF_LOOPBACK and not dev == loopback_dev 2007-10-10 16:52:48 -07:00
esp4.c [IPSEC]: Move IP protocol setting from transforms into xfrm4_input.c 2007-10-10 16:55:56 -07:00
fib_frontend.c [NETLINK]: fib_frontend build fixes 2007-10-10 21:32:39 -07:00
fib_hash.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
fib_lookup.h [RTNETLINK]: Fix sending netlink message when replace route. 2007-05-24 16:36:53 -07:00
fib_rules.c [IPV4/IPV6/DECNET]: Small cleanup for fib rules. 2007-10-10 16:51:22 -07:00
fib_semantics.c [NETLINK]: Introduce nested and byteorder flag to netlink attribute 2007-10-10 16:49:16 -07:00
fib_trie.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
icmp.c [IPV4]: Add ICMPMsgStats MIB (RFC 4293) 2007-10-10 16:51:28 -07:00
igmp.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
inet_connection_sock.c [INET]: local port range robustness 2007-10-10 17:30:46 -07:00
inet_diag.c [NET]: make netlink user -> kernel interface synchronious 2007-10-10 21:15:29 -07:00
inet_hashtables.c [INET]: local port range robustness 2007-10-10 17:30:46 -07:00
inet_lro.c [NET]: sparse warning fixes 2007-10-10 16:54:48 -07:00
inet_timewait_sock.c [NET]: DIV_ROUND_UP cleanup (part two) 2007-10-10 16:48:37 -07:00
inetpeer.c [IPV4]: Fix inetpeer gcc-4.2 warnings 2007-07-20 19:39:17 -07:00
ip_forward.c [IPV4] IPSEC: Omit redirect for tunnelled packet. 2007-10-10 16:48:33 -07:00
ip_fragment.c [NET]: Make the device list and device lookups per namespace. 2007-10-10 16:49:10 -07:00
ip_gre.c [NET]: Move hardware header operations out of netdevice. 2007-10-10 16:52:52 -07:00
ip_input.c [NET]: Make packet reception network namespace safe 2007-10-10 16:49:08 -07:00
ip_options.c [IPV4] ip_options.c: kmalloc + memset conversion to kzalloc 2007-07-31 14:06:45 -07:00
ip_output.c [NET]: Move hardware header operations out of netdevice. 2007-10-10 16:52:52 -07:00
ip_sockglue.c [NET]: sparse warning fixes 2007-10-10 16:54:48 -07:00
ipcomp.c [IPSEC]: Move IP protocol setting from transforms into xfrm4_input.c 2007-10-10 16:55:56 -07:00
ipconfig.c [NET]: Wrap netdevice hardware header creation. 2007-10-10 16:52:50 -07:00
ipip.c [NET]: Nuke SET_MODULE_OWNER macro. 2007-10-10 16:51:13 -07:00
ipmr.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
netfilter.c [SK_BUFF]: Introduce ip_hdr(), remove skb->nh.iph 2007-04-25 22:25:10 -07:00
proc.c [NET]: sparse warning fixes 2007-10-10 16:54:48 -07:00
protocol.c [IPV4]: align inet_protos[] on SMP 2007-04-25 22:28:20 -07:00
raw.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
route.c [NET]: Make core networking code use seq_open_private 2007-10-10 16:55:33 -07:00
syncookies.c [SK_BUFF]: Introduce tcp_hdr(), remove skb->h.th 2007-04-25 22:25:26 -07:00
sysctl_net_ipv4.c [INET]: local port range robustness 2007-10-10 17:30:46 -07:00
tcp.c [TCP]: Return useful listenq info in tcp_info and INET_DIAG_INFO. 2007-10-10 16:51:35 -07:00
tcp_bic.c [TCP]: Remove num_acked>0 checks from cong.ctrl mods pkts_acked 2007-10-10 16:47:55 -07:00
tcp_cong.c [TCP]: remove unused argument to cong_avoid op 2007-07-18 01:46:58 -07:00
tcp_cubic.c [TCP]: Remove num_acked>0 checks from cong.ctrl mods pkts_acked 2007-10-10 16:47:55 -07:00
tcp_diag.c [TCP]: Return useful listenq info in tcp_info and INET_DIAG_INFO. 2007-10-10 16:51:35 -07:00
tcp_highspeed.c [TCP]: remove unused argument to cong_avoid op 2007-07-18 01:46:58 -07:00
tcp_htcp.c [TCP]: H-TCP maxRTT estimation at startup 2007-08-07 18:29:05 -07:00
tcp_hybla.c [TCP]: remove unused argument to cong_avoid op 2007-07-18 01:46:58 -07:00
tcp_illinois.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_input.c [TCP]: Fix lost_retrans loop vs fastpath problems 2007-10-11 17:35:41 -07:00
tcp_ipv4.c [INET]: local port range robustness 2007-10-10 17:30:46 -07:00
tcp_lp.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_minisocks.c [TCP]: Move sack_ok access to obviously named funcs & cleanup 2007-10-10 16:48:00 -07:00
tcp_output.c [TCP]: Fix two off-by-one errors in fackets_out adjusting logic 2007-10-10 16:54:47 -07:00
tcp_probe.c [NET]: Make /proc/net per network namespace 2007-10-10 16:49:06 -07:00
tcp_scalable.c [TCP]: remove unused argument to cong_avoid op 2007-07-18 01:46:58 -07:00
tcp_timer.c [TCP]: Move sack_ok access to obviously named funcs & cleanup 2007-10-10 16:48:00 -07:00
tcp_vegas.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_vegas.h [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_veno.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_westwood.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tcp_yeah.c [TCP]: congestion control API pass RTT in microseconds 2007-07-31 02:27:57 -07:00
tunnel4.c [IPSEC]: Changing API of xfrm4_tunnel_register. 2007-02-13 12:54:47 -08:00
udp.c [INET]: local port range robustness 2007-10-10 17:30:46 -07:00
udp_impl.h [UDP]: Randomize port selection. 2007-10-10 16:48:31 -07:00
udplite.c [UDP]: Randomize port selection. 2007-10-10 16:48:31 -07:00
xfrm4_input.c [IPSEC]: Move IP protocol setting from transforms into xfrm4_input.c 2007-10-10 16:55:56 -07:00
xfrm4_mode_beet.c [IPSEC]: Move IP length/checksum setting out of transforms 2007-10-10 16:55:56 -07:00
xfrm4_mode_transport.c [IPSEC]: Use IPv6 calling convention as the convention for x->mode->output 2007-10-10 16:55:54 -07:00
xfrm4_mode_tunnel.c [IPSEC]: Move IP length/checksum setting out of transforms 2007-10-10 16:55:56 -07:00
xfrm4_output.c [IPSEC]: Move IP length/checksum setting out of transforms 2007-10-10 16:55:56 -07:00
xfrm4_policy.c [NET]: Make the loopback device per network namespace. 2007-10-10 16:52:49 -07:00
xfrm4_state.c [IPSEC]: exporting xfrm_state_afinfo 2007-02-08 12:39:00 -08:00
xfrm4_tunnel.c [IPSEC]: Move IP protocol setting from transforms into xfrm4_input.c 2007-10-10 16:55:56 -07:00