Page MenuHomeFreeBSD

mac_veriexec: remove opt_global.h workarounds
Needs ReviewPublic

Authored by kevans on Jan 30 2020, 3:08 AM.
Tags
None
Referenced Files
Unknown Object (File)
Mon, Mar 9, 6:33 AM
Unknown Object (File)
Mon, Mar 9, 5:49 AM
Unknown Object (File)
Fri, Feb 27, 12:54 PM
Unknown Object (File)
Fri, Feb 27, 6:26 AM
Unknown Object (File)
Feb 9 2026, 7:18 AM
Unknown Object (File)
Dec 1 2025, 10:30 AM
Unknown Object (File)
Nov 21 2025, 12:32 PM
Unknown Object (File)
Nov 20 2025, 4:41 PM
Subscribers

Details

Reviewers
stevek
sjg
Summary

Upon first glance, I can't actually see that any parts of this (or any headers included) actually need MAC defined. However,
earlier commits started producing an opt_global for independent modules builds and rS357274 started defining MAC in it since it's almost universally enabled with exception to mips, which doesn't really have a solid kernel config to represent the majority. Ideally, all of these opt_global bits should be removed from modules/ in favor of defining proper defaults in sys/conf/config.mk.

[reviewer's note]
It does occur to me that Juniper may have other need for MAC to be defined beyond what I can identify by inspecting our in-tree headers and source -- in that case, this would ideally be reflected as an unconditional default down in config.mk in their tree. This file sees little churn and shouldn't be ripe for merge conflicts, other than the recent churn as I've identified defaults that modules currently care about and moved them into common bits.

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 29028

Event Timeline

sys/modules/mac_veriexec/Makefile
23

Shouldn't this be surrounded by the following since opt_mac.h from the kernel config should be used when KERNELBUILDDIR is set?
.if !defined(KERNELBUILDDIR)
.endif