linux/net/tipc
Tuong Lien e95584a889 tipc: fix unlimited bundling of small messages
We have identified a problem with the "oversubscription" policy in the
link transmission code.

When small messages are transmitted, and the sending link has reached
the transmit window limit, those messages will be bundled and put into
the link backlog queue. However, bundles of data messages are counted
at the 'CRITICAL' level, so that the counter for that level, instead of
the counter for the real, bundled message's level is the one being
increased.
Subsequent, to-be-bundled data messages at non-CRITICAL levels continue
to be tested against the unchanged counter for their own level, while
contributing to an unrestrained increase at the CRITICAL backlog level.

This leaves a gap in congestion control algorithm for small messages
that can result in starvation for other users or a "real" CRITICAL
user. Even that eventually can lead to buffer exhaustion & link reset.

We fix this by keeping a 'target_bskb' buffer pointer at each levels,
then when bundling, we only bundle messages at the same importance
level only. This way, we know exactly how many slots a certain level
have occupied in the queue, so can manage level congestion accurately.

By bundling messages at the same level, we even have more benefits. Let
consider this:
- One socket sends 64-byte messages at the 'CRITICAL' level;
- Another sends 4096-byte messages at the 'LOW' level;

When a 64-byte message comes and is bundled the first time, we put the
overhead of message bundle to it (+ 40-byte header, data copy, etc.)
for later use, but the next message can be a 4096-byte one that cannot
be bundled to the previous one. This means the last bundle carries only
one payload message which is totally inefficient, as for the receiver
also! Later on, another 64-byte message comes, now we make a new bundle
and the same story repeats...

With the new bundling algorithm, this will not happen, the 64-byte
messages will be bundled together even when the 4096-byte message(s)
comes in between. However, if the 4096-byte messages are sent at the
same level i.e. 'CRITICAL', the bundling algorithm will again cause the
same overhead.

Also, the same will happen even with only one socket sending small
messages at a rate close to the link transmit's one, so that, when one
message is bundled, it's transmitted shortly. Then, another message
comes, a new bundle is created and so on...

We will solve this issue radically by another patch.

Fixes: 365ad353c2 ("tipc: reduce risk of user starvation during link congestion")
Reported-by: Hoang Le <hoang.h.le@dektech.com.au>
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Tuong Lien <tuong.t.lien@dektech.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-10-02 11:02:05 -04:00
..
addr.c tipc: initialise addr_trail_end when setting node addresses 2019-08-11 21:40:04 -07:00
addr.h tipc: add 128-bit node identifier 2018-03-23 13:12:18 -04:00
bcast.c tipc: clean up skb list lock handling on send path 2019-08-18 14:01:07 -07:00
bcast.h tipc: fix a null pointer deref 2019-03-21 09:56:55 -07:00
bearer.c tipc: add loopback device tracking 2019-08-08 22:11:39 -07:00
bearer.h tipc: add loopback device tracking 2019-08-08 22:11:39 -07:00
core.c tipc: add loopback device tracking 2019-08-08 22:11:39 -07:00
core.h tipc: add loopback device tracking 2019-08-08 22:11:39 -07:00
diag.c tipc: switch to rhashtable iterator 2018-08-29 18:04:54 -07:00
discover.c tipc: fix lockdep warning when reinitilaizing sockets 2018-11-17 22:01:31 -08:00
discover.h tipc: some cleanups in the file discover.c 2018-03-23 13:12:17 -04:00
eth_media.c tipc: make media address offset a common define 2015-02-27 18:18:48 -05:00
group.c tipc: clean up skb list lock handling on send path 2019-08-18 14:01:07 -07:00
group.h tipc: extend sock diag for group communication 2018-06-30 21:05:42 +09:00
ib_media.c tipc: rename media/msg related definitions 2015-02-27 18:18:48 -05:00
Kconfig docs: kbuild: convert docs to ReST and rename to *.rst 2019-06-14 14:21:21 -06:00
link.c tipc: fix unlimited bundling of small messages 2019-10-02 11:02:05 -04:00
link.h tipc: fix missing Name entries due to half-failover 2019-05-04 00:59:51 -04:00
Makefile tipc: enable tracepoints in tipc 2018-12-19 11:49:24 -08:00
monitor.c netlink: make nla_nest_start() add NLA_F_NESTED flag 2019-04-27 17:03:44 -04:00
monitor.h tipc: dump monitor attributes 2016-07-26 14:26:42 -07:00
msg.c tipc: fix unlimited bundling of small messages 2019-10-02 11:02:05 -04:00
msg.h Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2019-08-19 11:54:03 -07:00
name_distr.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2019-09-15 14:17:27 +02:00
name_distr.h tipc: permit overlapping service ranges in name table 2018-03-31 22:19:52 -04:00
name_table.c netlink: make nla_nest_start() add NLA_F_NESTED flag 2019-04-27 17:03:44 -04:00
name_table.h tipc: eliminate message disordering during binding table update 2018-10-22 19:29:12 -07:00
net.c netlink: make validation more configurable for future strictness 2019-04-27 17:07:21 -04:00
net.h tipc: fix lockdep warning when reinitilaizing sockets 2018-11-17 22:01:31 -08:00
netlink_compat.c tipc: compat: allow tipc commands without arguments 2019-08-01 18:14:00 -04:00
netlink.c tipc: fix missing indentation in source code 2019-06-24 07:29:45 -07:00
netlink.h tipc: make cluster size threshold for monitoring configurable 2016-07-26 14:26:42 -07:00
node.c tipc: clean up skb list lock handling on send path 2019-08-18 14:01:07 -07:00
node.h tipc: optimize link synching mechanism 2019-07-25 15:55:47 -07:00
socket.c tipc: clean up skb list lock handling on send path 2019-08-18 14:01:07 -07:00
socket.h tipc: add trace_events for tipc socket 2018-12-19 11:49:24 -08:00
subscr.c tipc: fix unbalanced reference counter 2018-04-12 21:46:10 -04:00
subscr.h tipc: fix modprobe tipc failed after switch order of device registration 2019-05-20 10:45:43 -07:00
sysctl.c proc/sysctl: add shared variables for range check 2019-07-18 17:08:07 -07:00
topsrv.c tipc: add loopback device tracking 2019-08-08 22:11:39 -07:00
topsrv.h tipc: rename tipc_server to tipc_topsrv 2018-02-16 15:26:34 -05:00
trace.c tipc: remove unneeded semicolon in trace.c 2019-01-17 22:04:43 -08:00
trace.h tipc: add trace_events for tipc bearer 2018-12-19 11:49:25 -08:00
udp_media.c tipc: use rcu dereference functions properly 2019-07-07 13:18:53 -07:00
udp_media.h tipc: implement configuration of UDP media MTU 2018-04-20 11:04:05 -04:00