Re: [PATCH v2] x86/smpboot: Remove 486-isms from the modern AP boot path

[Brian Gerst]

On Wed, Apr 1, 2020 at 8:14 AM Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
>
> On 01/04/2020 12:39, Brian Gerst wrote:
> > On Wed, Apr 1, 2020 at 5:22 AM Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
> >> On 31/03/2020 23:53, Brian Gerst wrote:
> >>> On Tue, Mar 31, 2020 at 6:44 PM Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
> >>>> On 31/03/2020 23:23, Brian Gerst wrote:
> >>>>> On Tue, Mar 31, 2020 at 1:59 PM Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
> >>>>>> Linux has an implementation of the Universal Start-up Algorithm (MP spec,
> >>>>>> Appendix B.4, Application Processor Startup), which includes unconditionally
> >>>>>> writing to the Bios Data Area and CMOS registers.
> >>>>>>
> >>>>>> The warm reset vector is only necessary in the non-integrated Local APIC case.
> >>>>>> UV and Jailhouse already have an opt-out for this behaviour, but blindly using
> >>>>>> the BDA and CMOS on a UEFI or other reduced hardware system isn't clever.
> >>>>>>
> >>>>>> We could make this conditional on the integrated-ness of the Local APIC, but
> >>>>>> 486-era SMP isn't supported.  Drop the logic completely, tidying up the includ
> >>>>>> list and header files as appropriate.
> >>>>>>

> >>>>> You removed x86_platform.legacy.warm_reset in the original patch, but
> >>>>> that is missing in V2.
> >>>> Second hunk?  Or are you referring to something different?
> >>> Removing the warm_reset field from struct x86_legacy_features.
> >> Ok, but that is still present as the 2nd hunk of the patch.
> > My apologies, Gmail was hiding that section of the patch because it
> > was a reply to the original patch.  For future reference, add the
> > version number to the title when resubmitting a patch (ie. [PATCH
> > v2]).
>
> Erm... is Gmail hiding that too?
>
> Lore thinks it is there:
> https://lore.kernel.org/lkml/CAMzpN2g0LS5anGc7CXco4pgBHhGzc8hw+shMOg8WEWGsx+BHpg@xxxxxxxxxxxxxx/

Ugh, yes.  I thought it was the title that Gmail threaded on, but it
must be the In-Reply-To: header.  Sorry for the confusion.

That said, I think the v1 patch is probably the better way to go (but
adjusting the comments to include early Pentium-era systems without
integrated APICs).  Then the decision to drop support for external
APICs could be a separate patch.

--
Brian Gerst