Re: [PATCH v3 08/10] x86/mtrr: let cache_aps_delayed_init replace mtrr_aps_delayed_init

[Borislav Petkov]

On Fri, Sep 30, 2022 at 03:11:07PM +0200, Juergen Gross wrote:
> Yes, this can be done. It would practically have to be the first one just
> after CPUHP_BRINGUP_CPU.

Right.

> The question is whether we really want to call the MTRR/PAT initialization
> on hotplugged cpus only after enabling interrupts. Note that the callbacks
> are activated only at the end of start_secondary(), while today MTRR/PAT
> initialization is called some time earlier by:
> 
>   start_secondary()
>     smp_callin()
>       smp_store_cpu_info()
>         identify_secondary_cpu()
>           mtrr_ap_init()
> 
> I don't think this is a real problem, but I wanted to mention it.

Yep, I saw that too but I don't think there will be a problem either.
I mean, it should be early enough as you point out not to need proper
MTRR/PAT settings yet.

But we'll make sure we test this real good too.

> The next question would be, why MTRR/PAT init should be special
> (meaning: why are all the other functions called that early not
> realized via callbacks)?

Well, our init code is crazy. Frankly, I don't see why not more of the
"init stuff on the freshly hotplugged CPU" work is done there...

> Is it just because of the special handling during boot/resume?

... unless this is the case, ofc. Right.

> It might be worth a discussion whether there shouldn't be a special group
> of callbacks activated BEFORE interrupts are being enabled.

That's a good question. /me writes it down to ask tglx when he gets back.

I mean, that early I don't think it matters whether IRQs are enabled
or not. But this'll need to be audited on a case by case basis. As I
said, our boot code is nuts with stuff bolted on everywhere for whatever
reasons.

> Thanks. I'll write a patch for that.

Thanks too.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette