netdev - RFC: making sch

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: <CAA93jw6OM9hzT2Okk4_tuLg_f=M-TXz+a17ZrRB2CL9uzm_eig@mail.gmail.com>
Date: Mon, 31 Jul 2023 12:10:13 -0700
From: Dave Taht <dave.taht@...il.com>
To: Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: RFC: making sch_cake multicore?

The ongoing, everlasting feature request for CAKE has been to speed it
up on now common multicore embedded hardware. 4 and 8 cores are
increasingly common in my world. I am really out of date on how
feasible this would be, in fact I don't know if older efforts to
remove the locks from pfifo_fast ever succeeded. Similarly, on inbound
shaping, we remain dependent on a tc filter matchall... mirred  into
an ifb, which is quite slow and apparently still single threaded.
Leveraging some shared memory area for some shared parameters in a
mqprio-like design is on my mind, but somehow that would have to share
some data with sub-qdiscs.

Over here is a design document talking about two approaches to this so
far. Please comment there. Thx!

https://docs.google.com/document/d/1tTYBPeaRdCO9AGTGQCpoiuLORQzN_bG3TAkEolJPh28/edit#heading=h.1d4d73u2m8nx

-- 
Podcast: https://www.youtube.com/watch?v=bxmoBr4cBKg
Dave Täht CSO, LibreQos