[Cake] BUG_ON vs WARN_ON
Dave Taht
dave.taht at gmail.com
Wed Oct 5 12:38:19 EDT 2016
I cannot repeat that result this morning, with either replace or
change. I *was* running far more extensive tests between changing
things that way than I just did, but a string of quick tests, changing
the bandwidth, changing it to unlimited, etc got the correct behaviors
throughout for both replace and change. But, changing it to 50mbit
shows the capacity estimate here to still be 700Mbit.
d at apu2:~$ tc -s qdisc show dev enp2s0
qdisc cake 800f: root refcnt 9 bandwidth 50Mbit diffserv4 flows rtt 100.0ms raw
Sent 7573781243 bytes 5002790 pkt (dropped 6, overlimits 1270728
requeues 19055)
backlog 0b 0p requeues 19055
memory used: 729504b of 4Mb
capacity estimate: 700Mbit
Bulk Best Effort Video Voice
thresh 3125Kbit 50Mbit 25Mbit 12500Kbit
target 5.8ms 5.0ms 5.0ms 5.0ms
interval 100.8ms 100.0ms 100.0ms 100.0ms
pk_delay 0us 4.6ms 1us 69us
av_delay 0us 3.2ms 0us 4us
sp_delay 0us 5us 0us 3us
pkts 0 2991920 1 89
bytes 0 4529561380 90 26451
way_inds 0 0 0 0
way_miss 0 38 1 2
way_cols 0 0 0 0
drops 0 6 0 0
marks 0 0 0 0
sp_flows 0 0 0 0
bk_flows 0 0 0 1
un_flows 0 0 0 0
max_len 0 19682 90 927
I am still not seeing the number of drops I would expect in the cake
statistics. 4 with 4 flows going at 50mbit on a 30 second test doesn't
strike me as enough. I will do a capture later to see if they add up.
That said, I did have to patch out on this particular kernel, this:
Linux apu2 4.5.0-fqmac3.5-wt-ath # yes, I'm running 3 kernels behind mainline.
diff --git a/sch_cake.c b/sch_cake.c
index f613ab4..92c5078 100644
--- a/sch_cake.c
+++ b/sch_cake.c
@@ -64,7 +64,7 @@
#endif
#if (KERNEL_VERSION(4,4,11) > LINUX_VERSION_CODE) ||
((KERNEL_VERSION(4,5,0) <= LINUX_VERSION_CODE) &&
(KERNEL_VERSION(4,5,5) > LINUX_VERSION_CODE))
-#define qdisc_tree_reduce_backlog(_a,_b,_c) qdisc_tree_decrease_qlen(_a,_b)
+// #define qdisc_tree_reduce_backlog(_a,_b,_c) qdisc_tree_decrease_qlen(_a,_b)
#endif
More information about the Cake
mailing list