linux/net/sctp
Marcelo Ricardo Leitner ba6f5e33bd sctp: avoid refreshing heartbeat timer too often
Currently on high rate SCTP streams the heartbeat timer refresh can
consume quite a lot of resources as timer updates are costly and it
contains a random factor, which a) is also costly and b) invalidates
mod_timer() optimization for not editing a timer to the same value.
It may even cause the timer to be slightly advanced, for no good reason.

As suggested by David Laight this patch now removes this timer update
from hot path by leaving the timer on and re-evaluating upon its
expiration if the heartbeat is still needed or not, similarly to what is
done for TCP. If it's not needed anymore the timer is re-scheduled to
the new timeout, considering the time already elapsed.

For this, we now record the last tx timestamp per transport, updated in
the same spots as hb timer was restarted on tx. Also split up
sctp_transport_reset_timers into sctp_transport_reset_t3_rtx and
sctp_transport_reset_hb_timer, so we can re-arm T3 without re-arming the
heartbeat one.

On loopback with MTU of 65535 and data chunks with 1636, so that we
have a considerable amount of chunks without stressing system calls,
netperf -t SCTP_STREAM -l 30, perf looked like this before:

Samples: 103K of event 'cpu-clock', Event count (approx.): 25833000000
  Overhead  Command  Shared Object      Symbol
+    6,15%  netperf  [kernel.vmlinux]   [k] copy_user_enhanced_fast_string
-    5,43%  netperf  [kernel.vmlinux]   [k] _raw_write_unlock_irqrestore
   - _raw_write_unlock_irqrestore
      - 96,54% _raw_spin_unlock_irqrestore
         - 36,14% mod_timer
            + 97,24% sctp_transport_reset_timers
            + 2,76% sctp_do_sm
         + 33,65% __wake_up_sync_key
         + 28,77% sctp_ulpq_tail_event
         + 1,40% del_timer
      - 1,84% mod_timer
         + 99,03% sctp_transport_reset_timers
         + 0,97% sctp_do_sm
      + 1,50% sctp_ulpq_tail_event

And after this patch, now with netperf -l 60:

Samples: 230K of event 'cpu-clock', Event count (approx.): 57707250000
  Overhead  Command  Shared Object      Symbol
+    5,65%  netperf  [kernel.vmlinux]   [k] memcpy_erms
+    5,59%  netperf  [kernel.vmlinux]   [k] copy_user_enhanced_fast_string
-    5,05%  netperf  [kernel.vmlinux]   [k] _raw_spin_unlock_irqrestore
   - _raw_spin_unlock_irqrestore
      + 49,89% __wake_up_sync_key
      + 45,68% sctp_ulpq_tail_event
      - 2,85% mod_timer
         + 76,51% sctp_transport_reset_t3_rtx
         + 23,49% sctp_do_sm
      + 1,55% del_timer
+    2,50%  netperf  [sctp]             [k] sctp_datamsg_from_user
+    2,26%  netperf  [sctp]             [k] sctp_sendmsg

Throughput-wise, from 6800mbps without the patch to 7050mbps with it,
~3.7%.

Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-04-10 22:22:34 -04:00
..
associola.c sctp: align MTU to a word 2016-03-20 16:31:12 -04:00
auth.c sctp: Use shash 2016-01-27 20:36:15 +08:00
bind_addr.c sctp: fix copying more bytes than expected in sctp_add_bind_addr 2016-03-08 15:04:08 -05:00
chunk.c sctp: allow sctp_transmit_packet and others to use gfp 2016-03-13 22:29:07 -04:00
debug.c sctp: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
endpointola.c sctp: Use shash 2016-01-27 20:36:15 +08:00
input.c sctp: align MTU to a word 2016-03-20 16:31:12 -04:00
inqueue.c net: sctp: fix remote memory pressure from excessive queueing 2014-10-14 12:46:22 -04:00
ipv6.c sctp: lack the check for ports in sctp_v6_cmp_addr 2016-03-01 17:25:12 -05:00
Kconfig net: sctp: get rid of SCTP_DBG_TSNS entirely 2013-07-02 00:08:03 -07:00
Makefile net: sctp: Inline the functions from command.c 2014-07-08 14:38:48 -07:00
objcnt.c sctp: fix checkpatch errors with (foo*)|foo * bar|foo* bar 2013-12-26 13:47:47 -05:00
output.c sctp: flush if we can't fit another DATA chunk 2016-04-05 15:39:44 -04:00
outqueue.c sctp: avoid refreshing heartbeat timer too often 2016-04-10 22:22:34 -04:00
primitive.c sctp: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
probe.c net: sctp: Convert log timestamps to be y2038 safe 2016-03-01 17:18:44 -05:00
proc.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-03-08 12:34:12 -05:00
protocol.c sctp: fix copying more bytes than expected in sctp_add_bind_addr 2016-03-08 15:04:08 -05:00
sm_make_chunk.c sctp: avoid refreshing heartbeat timer too often 2016-04-10 22:22:34 -04:00
sm_sideeffect.c sctp: avoid refreshing heartbeat timer too often 2016-04-10 22:22:34 -04:00
sm_statefuns.c sctp: fix use-after-free in pr_debug statement 2016-01-11 17:13:01 -05:00
sm_statetable.c sctp: fix checkpatch errors with indent 2013-12-26 13:47:48 -05:00
socket.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-03-23 23:25:14 -07:00
ssnmap.c sctp: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
sysctl.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-01-11 23:55:43 -05:00
transport.c sctp: avoid refreshing heartbeat timer too often 2016-04-10 22:22:34 -04:00
tsnmap.c sctp: Fix FSF address in file headers 2013-12-06 12:37:56 -05:00
ulpevent.c sctp: Fixup v4mapped behaviour to comply with Sock API 2014-07-31 21:49:06 -07:00
ulpqueue.c net: introduce SO_INCOMING_CPU 2014-11-11 13:00:06 -05:00