On 22.01.23 05:28, Sasha Levin wrote:
This is a note to let you know that I've just added the patch titled
wifi: mac80211: Drop support for TX push path
to the 6.1-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
wifi-mac80211-drop-support-for-tx-push-path.patch
and it can be found in the queue-6.1 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
We should at least have a discussion about that.
While I think we have sorted out all related regressions it's still way
too early to be sure.
The patch is also changing most mac80211 driver interfaces from queuing
to non-queuing and is thus nothing I would do within a fix release.
All in all it's more likely to cause issues than fix them, at last at
this point in time.
So do we really want to backport that to (all) stable kernels?
I've also just backported the two for stable relevant patches which
depend on the iTXQ transformation:
https://lore.kernel.org/r/20230121223330.389255-2-alexander@xxxxxxxxxxxxxx
https://lore.kernel.org/r/20230121223330.389255-2-alexander@xxxxxxxxxxxxxx
If there are more patches can point them out to me and I'll should be
able port them, too.
All in all I see no pressing need to retire the old push path for stable
kernels at that time.
Question is also where to stop if we back port it now:
The transition to iTXQ is only the first step to get rid of the old push
path in mac80211. Working patch titels are:
1) wifi: mac80211: Always provide the MMPDU TXQ
2) wifi: mac80211: Convert vif->txq to an array
3) wifi: mac80211: add new iTXQs to replace remaining legacy TX
4) wifi: mac80211: Stop using legacy TX path
5) wifi: mac80211: Cleanup legacy TX path - AMPDU
6) WIP: Drop pending
7) wifi: mac80211: integrate PS buffering into iTXQ
8) wifi: mac80211: handle filtered frames within iTXQs
Patch 6) is only a rough skeleton so far and 4-8 still need at least
some moderate work. All in all thinks seem to hash out quite well and
I'm hoping to get them merged for 6.4.
Together they are fundamentally altering the TX path and nothing I would
like to backport to stable.
Alexander