Adds an optional Random Early Detection on each SFQ flow queue.
Traditional SFQ limits count of packets, while RED permits to also
control number of bytes per flow, and adds ECN capability as well.
1) We dont handle the idle time management in this RED implementation,
since each 'new flow' begins with a null qavg. We really want to address
backlogged flows.
2) if headdrop is selected, we try to ecn mark first packet instead of
currently enqueued packet. This gives faster feedback for tcp flows
compared to traditional RED [ marking the last packet in queue ]
Example of use :
tc qdisc add dev $DEV parent 1:1 handle 10: est 1sec 4sec sfq \
limit 3000 headdrop flows 512 divisor 16384 \
redflowlimit 100000 min 8000 max 60000 probability 0.20 ecn
qdisc sfq 10: parent 1:1 limit 3000p quantum 1514b depth 127 headdrop
flows 512/16384 divisor 16384
ewma 6 min 8000b max 60000b probability 0.2 ecn
prob_mark 0 prob_mark_head 4876 prob_drop 6131
forced_mark 0 forced_mark_head 0 forced_drop 0
Sent 1175211782 bytes 777537 pkt (dropped 6131, overlimits 11007
requeues 0)
rate 99483Kbit 8219pps backlog 689392b 456p requeues 0
In this test, with 64 netperf TCP_STREAM sessions, 50% using ECN enabled
flows, we can see number of packets CE marked is smaller than number of
drops (for non ECN flows)
If same test is run, without RED, we can check backlog is much bigger.
qdisc sfq 10: parent 1:1 limit 3000p quantum 1514b depth 127 headdrop
flows 512/16384 divisor 16384
Sent 1148683617 bytes 795006 pkt (dropped 0, overlimits 0 requeues 0)
rate 98429Kbit 8521pps backlog 1221290b 841p requeues 0
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Kernel code and interface.
--------------------------
* Compile time switches
There is only one, but very important, compile time switch.
It is not settable by "make config", but should be selected
manually and after a bit of thinking in <include/net/pkt_sched.h>
PSCHED_CLOCK_SOURCE can take three values:
PSCHED_GETTIMEOFDAY
PSCHED_JIFFIES
PSCHED_CPU
PSCHED_GETTIMEOFDAY
Default setting is the most conservative PSCHED_GETTIMEOFDAY.
It is very slow both because of weird slowness of do_gettimeofday()
and because it forces code to use unnatural "timeval" format,
where microseconds and seconds fields are separate.
Besides that, it will misbehave, when delays exceed 2 seconds
(f.e. very slow links or classes bounded to small slice of bandwidth)
To resume: as only you will get it working, select correct clock
source and forget about PSCHED_GETTIMEOFDAY forever.
PSCHED_JIFFIES
Clock is derived from jiffies. On architectures with HZ=100
granularity of this clock is not enough to make reasonable
bindings to real time. However, taking into account Linux
architecture problems, which force us to use artificial
integrated clock in any case, this switch is not so bad
for schduling even on high speed networks, though policing
is not reliable.
PSCHED_CPU
It is available only for alpha and pentiums with correct
CPU timestamp. It is the fastest way, use it when it is available,
but remember: not all pentiums have this facility, and
a lot of them have clock, broken by APM etc. etc.
6987ecf083