Details

Reviewers

Group Reviewers

Summary

Split the amrr_update() routine into HT and legacy paths.
Currently they're the same, minus HT / non-HT specific comments
and some logging changes.

HT rates don't monotonically go up (ie, going MCS 7 -> MCS 8 is wrong,
as MCS8 is "just" two stream MCS0) and failing a rate by a little
shouldn't necessarily preclude testing the next rate up.

This should be a no-op, besides the logging changes.

Locally tested:

RTL8192CU, STA mode

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Build Status

Buildable 61989
Build 58873: arc lint + arc unit

Event Timeline

adrian created this revision.Sun, Dec 29, 7:41 PM

Herald added subscribers: glebius, melifaro, imp. · View Herald TranscriptSun, Dec 29, 7:41 PM

adrian requested review of this revision.Sun, Dec 29, 7:41 PM

Harbormaster completed remote builds in B61396: Diff 148535.Sun, Dec 29, 7:41 PM

adrian added a parent revision: D48240: ath_hal_ar9300: implement the TX/RX chainmask override for AR9300 HAL.Sun, Dec 29, 7:42 PM

adrian added a reviewer: wireless.Sun, Dec 29, 11:39 PM

adrian added a project: wireless.

Note: the goal here is to create a third path for VHT rates, even if it's making inefficient choices.

I have a local diff here that I'm not putting up for review (yet!) that turns it from an rix +/- change into a table based change, so we can configure which MCS rates it steps up / down through. It's a lot easier to do that with amrr refactored and cleaned up.

Assuming you did the c&p replication in full this seems fine as a start modulo style.

sys/net80211/ieee80211_amrr.c
235	No blank after the cast.
300	No need to initialize here.
356	What's this initialization still good for? The if/else below will do the job.
361	No need for {} {} here.
370	Should we apply return () the new code?

This revision now requires changes to proceed.Tue, Jan 7, 6:45 PM

adrian added a parent revision: D48210: iwm: add sku_cap_mimo_disable; populate sku_cap_11n_enable.Thu, Jan 9, 5:12 PM

adrian removed a parent revision: D48240: ath_hal_ar9300: implement the TX/RX chainmask override for AR9300 HAL.

adrian added a child revision: D48428: rtwn: fix RTL8192EU power-down / power-up hanging.Sat, Jan 11, 3:04 PM

adrian added a child revision: D48429: amrr: refactor amrr_node_init() into HT and legacy paths.Sat, Jan 11, 3:07 PM

adrian removed a child revision: D48428: rtwn: fix RTL8192EU power-down / power-up hanging.

feedback from bz

Harbormaster completed remote builds in B61749: Diff 149319.Thu, Jan 16, 2:29 AM

updated, please re-review!

I just looked through the delta changes, so the original comment about the duplication assumption stays. The changes seem fine. Thank you for doing them!

This revision is now accepted and ready to land.Thu, Jan 16, 11:36 AM

adrian edited parent revisions, added: D48610: net80211: add ieee80211_vht_node_check_valid_mcs(); removed: D48210: iwm: add sku_cap_mimo_disable; populate sku_cap_11n_enable.Thu, Jan 23, 2:40 AM