linux-kernel - Re: [PATCH 06/20] x86/mce/amd: Use helper for GPU UMC bank type checks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <79e63555-a46b-43aa-ab2c-161e1cb80904@amd.com>
Date:   Mon, 27 Nov 2023 10:12:16 -0500
From:   Yazen Ghannam <yazen.ghannam@....com>
To:     Borislav Petkov <bp@...en8.de>
Cc:     yazen.ghannam@....com, linux-edac@...r.kernel.org,
        linux-kernel@...r.kernel.org, tony.luck@...el.com, x86@...nel.org,
        Avadhut.Naik@....com, Smita.KoralahalliChannabasappa@....com,
        amd-gfx@...ts.freedesktop.org, linux-trace-kernel@...r.kernel.org
Subject: Re: [PATCH 06/20] x86/mce/amd: Use helper for GPU UMC bank type
 checks

On 11/27/2023 6:46 AM, Borislav Petkov wrote:
> On Sat, Nov 18, 2023 at 01:32:34PM -0600, Yazen Ghannam wrote:
>> +/* GPU UMCs have MCATYPE=0x1.*/
>> +bool smca_gpu_umc_bank_type(u64 ipid)
>> +{
>> +	if (!smca_umc_bank_type(ipid))
>> +		return false;
>> +
>> +	return FIELD_GET(MCI_IPID_MCATYPE, ipid) == 0x1;
>> +}
> 
> And now this tells you that you want to use
> 
> 	u32 hwid_mcatype;       /* (hwid,mcatype) tuple */
> 
> everywhere and do your macros around that thing.
> 
> No need for yet another helper as this all is static information which
> doesn't change.
> 
>> +EXPORT_SYMBOL_GPL(smca_gpu_umc_bank_type);
> 
> Definitely not another export.
> 

Right, we already use the tuple thing. Patch 8 in this set uses the 
tuple to look up (search for) the bank type at runtime. This is in order 
to get rid of all the per-CPU stuff.

Now I figured it'd be good to have special helpers to do a quick check 
for the UMC bank types. Because memory errors will be the most commonly 
reported errors.

But it's likely not important to save a few cycles here. Especially 
since the helpers will be used in the notifier chain and not in the #MC 
handler, etc.

I'll think on it a bit more, but this patch and the previous one can 
likely be dropped.

Thanks,
Yazen