Re: [PATCH 3/3] EDAC/mce_amd: Add support for FRU Text in MCA

[Yazen Ghannam]

On Mon, Jul 04, 2022 at 11:13:56AM +0200, Borislav Petkov wrote:
> On Mon, Apr 18, 2022 at 05:44:40PM +0000, Yazen Ghannam wrote:
> > @@ -1254,11 +1255,10 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
> >  		((m->status & MCI_STATUS_PCC)	? "PCC"	  : "-"));
> >  
> >  	if (boot_cpu_has(X86_FEATURE_SMCA)) {
> > -		u32 low, high;
> >  		u32 addr = MSR_AMD64_SMCA_MCx_CONFIG(m->bank);
> >  
> > -		if (!rdmsr_safe(addr, &low, &high) &&
> > -		    (low & MCI_CONFIG_MCAX))
> > +		if (!rdmsrl_safe_on_cpu(m->extcpu, addr, &mca_config) &&
> 
> This change needs to be mentioned in the commit message.
>

Okay.

> > +		    (mca_config & MCI_CONFIG_MCAX))
> >  			pr_cont("|%s", ((m->status & MCI_STATUS_TCC) ? "TCC" : "-"));
> >  
> >  		pr_cont("|%s", ((m->status & MCI_STATUS_SYNDV) ? "SyndV" : "-"));
> > @@ -1300,6 +1300,17 @@ amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data)
> >  		pr_cont("\n");
> 
> So here above the code prints SYND1 and SYND2.
> 
> If they contain FRU strings, then this printing should be an if-else by
> checking bit 9.
> 
> 	if (BIT(9))
> 		print fru text
> 	else
> 		print naked syndromes
> 
>

Okay, will change.

> >  		decode_smca_error(m);
> > +
> > +		if (mca_config & BIT(9)) {
> > +			char frutext[32];
> 
> Why 32?
>

I picked the next power of 2 after 16 in order to have some space for
terminating NULL in the string. I can't think of a good reason it needs to be
a power of 2, so I can change this to 17.

> > +			memset(frutext, 0, sizeof(frutext));
> > +			memcpy(&frutext[0], &m->synd1, 8);
> > +			memcpy(&frutext[8], &m->synd2, 8);
> > +
> > +			pr_emerg(HW_ERR "FRU Text: %s\n", frutext);
> > +		}
> > +
> >  		goto err_code;
> >  	}
>

Thanks,
Yazen