netdev - Fw: [Bug 220766] New: Packets dessapear

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <20251107081324.3ab69917@phoenix>
Date: Fri, 7 Nov 2025 08:13:24 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: netdev@...r.kernel.org
Subject: Fw: [Bug 220766] New: Packets dessapear

Begin forwarded message:

Date: Fri, 07 Nov 2025 15:47:37 +0000
From: bugzilla-daemon@...nel.org
To: stephen@...workplumber.org
Subject: [Bug 220766] New: Packets dessapear

https://bugzilla.kernel.org/show_bug.cgi?id=220766

            Bug ID: 220766
           Summary: Packets dessapear
           Product: Networking
           Version: 2.5
          Hardware: All
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: Other
          Assignee: stephen@...workplumber.org
          Reporter: wads31566@...il.com
        Regression: No

Created attachment 308921
  --> https://bugzilla.kernel.org/attachment.cgi?id=308921&action=edit  
Network topology

A network hierarchy (network_topo.png) implementation with namespace and TC get
packets lost unexpectedly.
Linux namespaces(created via "ip netns") and connected via veths to a single
bridge inside of the root-namespace. For setting up a veth settings TC is used
like that:
"$CLI tc qdisc add dev veth-${TARGET} root handle 1: netem delay ${DELAY_MS}ms
${DELAY_DISTRIBUTION_MS}ms distribution normal
$CLI tc qdisc add dev veth-${TARGET} parent 1: handle 2: netem loss 0 rate
100Mbit corrupt 1% duplicate 1%"

At this case, where we have a non-zero delay and banwidth limit. And when both
of those attributes are set it produces some packet-loss, i.e. packets are
counted as sent on the client namespace link statistics, but the received
amount of packets on the client side is always various and real loss percentage
is different from defined one. Usually the loss is about 3-7%.
Loss is counted as a servers-bridge sum of received packets and sum of packets
sent from the namespace link statistics.
But when only delay and no other netem params used - it works fine.

Mine assumption is that is becomes a bug, when several params are used.

Uname: Linux 6.14.0-35-generic #35-Ubuntu SMP PREEMPT_DYNAMIC Sat Oct 11
10:06:31 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
tc -V: tc utility, iproute2-6.14.0, libbpf 1.5.0
Distros: kubuntu, openSUSE Tumbleweed
As a traffic generator iPerf3 was used.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.