Re: [PATCH] x86/mce: Schedule mce_setup() on correct CPU for CPER decoding

[Yazen Ghannam]

On 6/15/2023 1:02 PM, Yazen Ghannam wrote:
> On 6/15/2023 12:20 PM, Borislav Petkov wrote:
> > On Thu, Jun 15, 2023 at 11:34:21AM -0400, Yazen Ghannam wrote:
> > > We can solve the NMI issue if it ever comes up in the future. Unless 
> > > there's
> > > an obvious change to avoid this now. Any suggestions?
> >
> > Yes, solve it right from the get-go. "It cannot happen now" is not good
> > enough. It should not be even technically possible.
>
> Okay, understood.
>
> > Just report what's logged into BERT - nothing more. Whoever needs the
> > remaining info, can dump it from the machine.
>
> Will do.

How about these changes? I can split this into separate preemption and 
PPIN patches.

The PPIN isn't coming directly from BERT. But it seems we can use the 
per_cpu value which is set up during CPU init.

Thanks,
Yazen

diff --git a/arch/x86/kernel/cpu/mce/apei.c b/arch/x86/kernel/cpu/mce/apei.c
index 8ed341714686..db16dc3c7b03 100644
--- a/arch/x86/kernel/cpu/mce/apei.c
+++ b/arch/x86/kernel/cpu/mce/apei.c
@@ -97,15 +97,19 @@ int apei_smca_report_x86_error(struct 
cper_ia_proc_ctx *ctx_info, u64 lapic_id)
        if (ctx_info->reg_arr_size < 48)
                return -EINVAL;

+       get_cpu();
        mce_setup(&m);
+       put_cpu();

        m.extcpu = -1;
        m.socketid = -1;
+       m.ppin = 0;

        for_each_possible_cpu(cpu) {
                if (cpu_data(cpu).initial_apicid == lapic_id) {
                        m.extcpu = cpu;
                        m.socketid = cpu_data(m.extcpu).phys_proc_id;
+                       m.ppin = cpu_data(m.extcpu).ppin;
                        break;
                }
        }