Hello! I'm doing some bug triaging on the iproute bugs in Debian.
I've tried to reproduce the problem reported in bug#175462 on my Debian Etch (i386) and Debian Sid (AMD64) computers. The problem seem to persist, even though I got other numbers that triggered the overflows then what was reported in the original description of the problem. On Debian Etch i386: buffer 10240kb -> burst 10Mb buffer 10241kb -> burst 4086kb $ dpkg -l | grep iproute ii iproute 20061002-3 Professional tools to control the networking $ uname -r 2.6.18-4-amd64 $ sudo tc qdisc add dev rtl8139 parent root handle 2: tbf latency 10ms buffer 10240kb mpu 64 rate 40kbit $ sudo tc -s qdisc ls |grep 2: qdisc tbf 2: dev rtl8139 rate 40000bit burst 10Mb lat 9.8ms $ sudo tc qdisc del dev rtl8139 $ sudo tc qdisc add dev rtl8139 parent root handle 2: tbf latency 10ms buffer 10241kb mpu 64 rate 40kbit $ sudo tc -s qdisc ls |grep 2: qdisc tbf 2: dev rtl8139 rate 40000bit burst 4086Mb lat 2197.8s $ sudo tc qdisc del dev rtl8139 root On Debian Sid AMD64: Not reproducible. $ dpkg -l | grep iproute ii iproute 20070313-1 Professional tools to control the networking $ uname -r 2.6.22-1-amd64 $ sudo tc qdisc add dev nvif parent root handle 2: tbf latency 10ms buffer 10738kb mpu 64 rate 40kbit $ /sbin/tc -s qdisc ls | grep 2: qdisc tbf 2: dev nvif rate 40000bit burst 10738Kb lat 10.0ms $ sudo tc qdisc del dev nvif root Using a 32bit/i386 Sid chroot on the same Sid AMD64: buffer 10kb -> burst 10kb buffer 11kb -> burst 4294956559b # dpkg -l | grep iproute ii iproute 20070313-1 Professional tools to control the networking # uname -r 2.6.22-1-amd64 # tc qdisc add dev nvif parent root handle 2: tbf latency 10ms buffer 10kb mpu 64 rate 40kbit # tc -s qdisc ls | grep 2: qdisc tbf 2: dev nvif rate 40000bit burst 10Kb lat 10.0ms # tc qdisc del dev nvif root # tc qdisc add dev nvif parent root handle 2: tbf latency 10ms buffer 11kb mpu 64 rate 40kbit # tc -s qdisc ls | grep 2: qdisc tbf 2: dev nvif rate 40000bit burst 4294956559b lat 115.3ms # tc qdisc del dev nvif root I hope to look closer into this soon and hopefully find out why this happens... -- Regards, Andreas Henriksson
signature.asc
Description: This is a digitally signed message part