linux-kernel - Re: [PATCH 2/2] x86/mce: Add support for Extended Physical Address MCA changes

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

Message-ID: <20210713162112.GB16744@aus-x-yghannam.amd.com>
Date:   Tue, 13 Jul 2021 12:21:12 -0400
From:   Yazen Ghannam <yazen.ghannam@....com>
To:     Smita Koralahalli <Smita.KoralahalliChannabasappa@....com>
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org,
        linux-edac@...r.kernel.org, Tony Luck <tony.luck@...el.com>,
        "H . Peter Anvin" <hpa@...or.com>,
        Muralidhara M K <muralimk@....com>,
        Akshay Gupta <Akshay.Gupta@....com>,
        Youquan Song <youquan.song@...el.com>,
        Zhen Lei <thunder.leizhen@...wei.com>
Subject: Re: [PATCH 2/2] x86/mce: Add support for Extended Physical Address
 MCA changes

On Thu, Jun 24, 2021 at 08:33:41PM -0500, Smita Koralahalli wrote:
> Newer AMD processors such as AMD 'Milan' support more physical address
> bits.
> 
> That is the MCA_ADDR registers on Scalable MCA systems contain the
> ErrorAddr in bits [56:0] instead of [55:0]. Hence the existing LSB field
> from bits [61:56] in MCA_ADDR must be moved around to accommodate the
> larger ErrorAddr size.
> 
> MCA_CONFIG[McaLsbInStatusSupported] indicates this change. If set, the
> LSB field will be found in MCA_STATUS rather than MCA_ADDR.
> 
> Each logical CPU has unique MCA bank in hardware and is not shared with
> other logical CPUs. Additionally on SMCA systems, each feature bit may be
> different for each bank within same logical CPU.
> 
> Check for MCA_CONFIG[McaLsbInStatusSupported] for each MCA bank and for
> each CPU.
> 
> Signed-off-by: Smita Koralahalli <Smita.KoralahalliChannabasappa@....com>
> ---

Reviewed-by: Yazen Ghannam <yazen.ghannam@....com>
Tested-by: Yazen Ghannam <yazen.ghannam@....com>

Thanks,
Yazen

P.S. A good test case for this feature is data poison consumption on
Milan. This case will generate two MCA errors. One will be a deferred
error in the UMC bank. This will be handled through the deferred error
interrupt handler. Also, the UMC bank does not have the "LSB in Status"
feature. The other error will be a #MC in the Load Store bank. This
bank does have the "LSB in Status" feature.