netdev - Re: TC: HTB module over limiting when CPU is under load

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

Date: Sat, 13 Jan 2024 15:57:23 +0200
From: Isaac Boukris <iboukris@...il.com>
To: netdev@...r.kernel.org
Subject: Re: TC: HTB module over limiting when CPU is under load

On Sat, Jan 13, 2024 at 2:40 PM Isaac Boukris <iboukris@...il.com> wrote:
>
> The attached script reproduces it for me, when run normally the curl
> command reports ~1800kbps for all modules HTB, HSFC and NETEM, but
> when I run the below command to incur heavy cpu load, I only get
> ~400kbps when using HTB while HSFC and NETEM still work well.

The original tests were made on Rocky 8:
uname -a
Linux Rocky8 4.18.0-513.9.1.el8_9.x86_64 #1 SMP Wed Nov 29 18:55:19
UTC 2023 x86_64 x86_64 x86_64 GNU/Linux

I've now tested with a more recent kernel using Fedora:
uname -a
Linux fedora 6.6.9-200.fc39.x86_64 #1 SMP PREEMPT_DYNAMIC Mon Jan  1
20:05:54 UTC 2024 x86_64 GNU/Linux

On Fedora with the aforementioned openssl command for load, i get
~1400kb with HTB (compare to ~400kb with the older kernel), but when
instead i use stress-ng command for load then HTB only reaches
~1100kb, while HSFC and NETEM still achieve ~1800kb in all cases.

stress-ng --cpu $(nproc)

Thanks!