Page MenuHomeFreeBSD

aq(4): arm PHY thermal shutdown from the admin poll
AcceptedPublic

Authored by nick_spun.io on Fri, Sep 11, 3:16 PM.
Referenced Files
F171577029: D59593.id186470.diff
Fri, Sep 11, 9:03 PM
F171576370: D59593.diff
Fri, Sep 11, 8:57 PM
F171538888: D59593.diff
Fri, Sep 11, 3:57 PM
F171538734: D59593.id186470.diff
Fri, Sep 11, 3:56 PM
Subscribers

Details

Reviewers
adrian
emaste
Summary

move thermal shutdown arming to the admin poll

this gives a more reasonable delay prior to the first attempt, and also
allows us to retry and make the option runtime-tuneable via a new
disable_thermal_arm sysctl

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 76779
Build 73662: arc lint + arc unit

Event Timeline

As a general rule I prefer positive-sense controls - so e.g. "enable_thermal_limit" or "thermal_limit.enable" (if we're also going to have a "thermal_limit.count or such), and it can default to true.

share/man/man4/aq.4
177

At first I read this as the driver applies the setting, which restores the firmware's own default -- it's the reset of the adapter which restored the firmware's own default, and we're overriding that, right? For the purpose of the user-facing man page I think we could just leave the note about firmware defaults out.

As a general rule I prefer positive-sense controls - so e.g. "enable_thermal_limit" or "thermal_limit.enable" (if we're also going to have a "thermal_limit.count or such), and it can default to true.

Can do - the A1 cards disable it by default and the A2 cards enable it by default so we're overriding one of them either way

invert knob and add parent thermal node, clean up verbiage in man page

This revision is now accepted and ready to land.Fri, Sep 11, 4:26 PM