Merge branch 'improve-the-taprio-qdisc-s-relationship-with-its-children'
authorJakub Kicinski <kuba@kernel.org>
Wed, 9 Aug 2023 22:59:23 +0000 (15:59 -0700)
committerJakub Kicinski <kuba@kernel.org>
Wed, 9 Aug 2023 22:59:24 +0000 (15:59 -0700)
commit29afcd69672a4e3d8604d17206d42004540d6d5c
tree22df015b777476e8faa196541acc50486fed7fd3
parente972a54706e47a034991cee69fae95746540609b
parent29c298d2bc82ccdbd23bf08bbafeb4e874832946
Merge branch 'improve-the-taprio-qdisc-s-relationship-with-its-children'

Vladimir Oltean says:

====================
Improve the taprio qdisc's relationship with its children

v1: https://lore.kernel.org/lkml/20230531173928.1942027-1-vladimir.oltean@nxp.com/

Prompted by Vinicius' request to consolidate some child Qdisc
dereferences in taprio:
https://lore.kernel.org/netdev/87edmxv7x2.fsf@intel.com/

I remembered that I had left some unfinished work in this Qdisc, namely
commit af7b29b1deaa ("Revert "net/sched: taprio: make qdisc_leaf() see
the per-netdev-queue pfifo child qdiscs"").

This patch set represents another stab at, essentially, what's in the
title. Not only does taprio not properly detect when it's grafted as a
non-root qdisc, but it also returns incorrect per-class stats.
Eventually, Vinicius' request is addressed too, although in a different
form than the one he requested (which was purely cosmetic).

Review from people more experienced with Qdiscs than me would be
appreciated. I tried my best to explain what I consider to be problems.
I am deliberately targeting net-next because the changes are too
invasive for net - they were reverted from stable once already.
====================

Link: https://lore.kernel.org/r/20230807193324.4128292-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>