Differential D52301

[rtwn] Migrate to use ieee80211_output_seqno_assign()
ClosedPublic
Actions

Authored by adrian on Aug 31 2025, 3:09 AM.

Details

Reviewers

Group Reviewers

wireless

Commits

rGdff11c4f8007: rtwn: enable seqno offload; migrate to use ieee80211_output_seqno_assign()

Summary

This is currently a straight substitution.

The sequence number code can likely be cleaned up to just always assign
it in the transmit path in the new world order, including some extra
flags to determine whether the frame belongs in an A-MPDU or not.

Locally tested:

RTL8812AU, STA mode

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

adrian created this revision.Aug 31 2025, 3:09 AM

Herald added a subscriber: imp. · View Herald TranscriptAug 31 2025, 3:09 AM

adrian requested review of this revision.Aug 31 2025, 3:09 AM

Harbormaster completed remote builds in B66705: Diff 161281.Aug 31 2025, 3:09 AM

adrian added a parent revision: D52300: [iwx] tell net80211 not to originate NULL data frames.Aug 31 2025, 3:09 AM

adrian added a child revision: D52302: [net80211] clean up M_SEQNO_SET and M_SEQNO_GET() to always limit to the sequence number range.

adrian mentioned this in D50693: rtwn: enable seqno offload.Aug 31 2025, 3:16 AM

adrian added a reviewer: wireless.

adrian added a project: wireless.

Change order of commits. And squash the updated this one into D50693?

sys/dev/rtwn/rtl8192c/r92c_tx.c
458	So what's the point of adding the % here if in the next change the macro does it? You'll have to undo the % again then?
sys/dev/rtwn/rtl8812a/r12a_tx.c
438	dito

This revision now requires changes to proceed.Sep 1 2025, 1:11 AM

remove the % operator in the driver now that the net80211 macros do it for us

Harbormaster completed remote builds in B66773: Diff 161403.Sep 3 2025, 3:32 AM

bz accepted this revision as: bz.Sep 3 2025, 11:44 AM

Yeah I was /just/ thinking about squishing them together into one commit.

Do I do that and just have two Differential Revision lines in the commit message?

In D52301#1196749, @adrian wrote:

Yeah I was /just/ thinking about squishing them together into one commit.

Do I do that and just have two Differential Revision lines in the commit message?

I'd be fine with that. If the magic works of auto-closing I have no idea. phabric-admins might know.

Or you get this in and then rebase the other one to just the one "enable ...". line setting FEXT?

This revision was not accepted when it landed; it landed in state Needs Review.Sep 6 2025, 6:25 AM

Closed by commit rGdff11c4f8007: rtwn: enable seqno offload; migrate to use ieee80211_output_seqno_assign() (authored by adrian). · Explain Why

This revision was automatically updated to reflect the committed changes.

adrian added a commit: rGdff11c4f8007: rtwn: enable seqno offload; migrate to use ieee80211_output_seqno_assign().