On Tue, Jun 4, 2013 at 1:21 PM, Jesper Dangaard Brouer <jbrouer@redhat.com> wrote:
On Tue, 04 Jun 2013 10:11:48 -0700
Eric Dumazet <eric.dumazet@gmail.com> wrote:

> From: Eric Dumazet <edumazet@google.com>
>
> commit 56b765b79 ("htb: improved accuracy at high rates") added
> another regression for low rates, because it mixes 1ns and 64ns time
> units.
>
> So the maximum delay (mbuffer) was not 60 second, but 937 ms.
>
> Lets convert all time fields to 1ns as 64bit arches are becoming the
> norm.

I'm of-cause happy as I need this fixed for big-HW machines at Red Hat.

But how is this 64-bit usage going to affect performance for smaller
ARM/MIPS based home routers, where shaping at these low rates is more
relevant?  (I'm just asking because I don't know, and just test this
on a 24-CPU machine).

I'm not worried about it but will find out shortly in cerowrt.

I take it this does not fix the DSL/ATM issue however?
 

Tested-by: Jesper Dangaard Brouer <brouer@redhat.com>

I have tested you patch, and it works for me.
Tested shaping at 100Kbit/s:
 - max "rate 100656bit" on dev lo
 - max "rate 102016bit" on a real  1 Gbit/s NIC (dev eth63).
 - max "rate 103256bit" on a real 10 Gbit/s NIC (dev eth31).

The traffic "of-cause" spikes due to the GSO frames, if measuring the
traffic at a higher resolution (I've seen upto 3 sec without traffic).

Thanks for the quick fix! :-)))
--
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Sr. Network Kernel Developer at Red Hat
  Author of http://www.iptv-analyzer.org
  LinkedIn: http://www.linkedin.com/in/brouer

Reproducer commands:
 tc qdisc add dev lo root handle 1: htb default 1
 tc class add dev lo parent 1: classid 1:1 htb rate 100kbit
 netserver
 netperf -t TCP_STREAM -l 60 -H 127.0.0.1 -- -m 1024

Measuring the qdisc "rate" via:
 tc -s -d class show dev lo classid 1:1

Data:
=====
From dev "lo":
--------------
class htb 1:1 root prio 0 quantum 1250 rate 100000bit ceil 100000bit burst 1600b/1 mpu 0b overhead 0b cburst 1600b/1 mpu 0b overhead 0b level 0
 Sent 597719 bytes 64 pkt (dropped 942, overlimits 0 requeues 0)
 rate 100656bit 1pps backlog 0b 1p requeues 0
 lended: 64 borrowed: 0 giants: 0
 tokens: -27387576 ctokens: -27387576

From 1Gbit/s dev eth63:
-----------------------
class htb 1:1 root prio 0 quantum 1250 rate 100000bit ceil 100000bit burst 1600b/1 mpu 0b overhead 0b cburst 1600b/1 mpu 0b overhead 0b level 0
 Sent 1801152 bytes 3224 pkt (dropped 0, overlimits 0 requeues 0)
 rate 102016bit 9pps backlog 0b 12p requeues 0
 lended: 2284 borrowed: 0 giants: 0
 tokens: -18923820 ctokens: -18923820


From 10Gbit/s dev eth31:
------------------------
class htb 1:1 root prio 0 quantum 1250 rate 100000bit ceil 100000bit burst 1600b/1 mpu 0b overhead 0b cburst 1600b/1 mpu 0b overhead 0b level 0
 Sent 855814 bytes 581 pkt (dropped 0, overlimits 0 requeues 0)
 rate 103256bit 9pps backlog 0b 9p requeues 0
 lended: 81 borrowed: 0 giants: 0
 tokens: -45419150 ctokens: -45419150




--
Dave Täht

Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.html