Page MenuHomeFreeBSD

i386: Fix the nodevice apic build
ClosedPublic

Authored by markj on Apr 7 2022, 7:13 PM.
Tags
None
Referenced Files
Unknown Object (File)
Fri, Apr 26, 6:13 PM
Unknown Object (File)
Feb 11 2024, 10:54 PM
Unknown Object (File)
Feb 4 2024, 4:27 PM
Unknown Object (File)
Dec 20 2023, 4:39 AM
Unknown Object (File)
Dec 12 2023, 12:49 PM
Unknown Object (File)
Nov 28 2023, 6:13 PM
Unknown Object (File)
Nov 24 2023, 4:38 PM
Unknown Object (File)
Nov 24 2023, 9:52 AM
Subscribers

Details

Summary

PR: 263124
Fixes: 62d09b46ad75 ("x86: Defer LAPIC calibration until after timecounters are available")

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

markj requested review of this revision.Apr 7 2022, 7:13 PM
This revision is now accepted and ready to land.Apr 7 2022, 11:46 PM

I wonder if we should just make 'device apic' non-optional on i386 at this point. Any non-ancient machine that can actually run FreeBSD/i386 at this point should have an APIC and you are better off using an I/O APIC to avoid interrupt sharing.

In D34830#789462, @jhb wrote:

I wonder if we should just make 'device apic' non-optional on i386 at this point. Any non-ancient machine that can actually run FreeBSD/i386 at this point should have an APIC and you are better off using an I/O APIC to avoid interrupt sharing.

Weren't there Pentiums (aka 586) and Pentium II/III (686) which had no IOAPIC on m/b? I am not sure completely, but I suspect that LAPIC cannot work in this physical configuration.

emaste added a subscriber: emaste.

I wonder if we should just make 'device apic' non-optional on i386

That sounds reasonable to me; we could follow up in main with that change (and not MFC).

This revision was automatically updated to reflect the committed changes.