linux-kernel - Re: [PATCH v3] dcdbas: Add support for WSMT ACPI table

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 for Android: free password hash cracker in your pocket

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

Message-ID: <4e9a34c5-e0e6-b060-40fa-30d738d1eb69@gmail.com>
Date:   Tue, 3 Jul 2018 08:52:22 -0500
From:   Stuart Hayes <stuart.w.hayes@...il.com>
To:     Mario.Limonciello@...l.com, andy.shevchenko@...il.com
Cc:     dvhart@...radead.org, linux-kernel@...r.kernel.org,
        platform-driver-x86@...r.kernel.org
Subject: Re: [PATCH v3] dcdbas: Add support for WSMT ACPI table



On 7/2/2018 11:15 AM, Mario.Limonciello@...l.com wrote:
>>
>>> I don't believe SMM communication ACPI table has ever been implemented by
>> Dell
>>> on server or client BIOS.  I do agree this table describes the behavior that DCDBAS
>> driver
>>> has used since before even UEFI BIOS pretty accurately.
>>
>> So, EPS table has been for ages in Dell machines?
>> Can we consider it as a predecessor of that SMM communication ACPI table?
> 
> No, EPS is new this year, specifically for server BIOS to be able to support SMM communication
> when WSMT is enabled.  The code tests in Stuart's patch will detect if WSMT is enabled
> and if it's enabled test if EPS was defined.  On server BIOS when EPS is defined dcdbas
> will be able to communicate using addresses defined in EPS.
> 
> Server BIOS will support EPS for applications using dcdbas interface and may at a later time
> introduce same WMI interface as client too (but applications will need time to update so
> they need to support both).
> 
> Actually Stuart's patch will cause client BIOS that has WSMT enabled make dcdbas fail
> initialization (as it should because dcdbas doesn't have a region that it can successfully
> communicate).  
> 
> In client machines we moved this communication to ACPI buffer allocated by WMI, which
> is why we have dell-smbios-wmi now in kernel.
> 
> I think once some variation of Stuart's patch is merged, I'll send a follow
> up patch to drop this test because it's no longer necessary:
> https://github.com/torvalds/linux/blob/master/drivers/platform/x86/dell-smbios-smm.c#L106
> 
> 
>>
>>> Stuart and I did discuss with server BIOS (who uses this EPS mechanism) to see if
>> its possible
>>> to move EPS to SMM communication ACPI table however since it's been
>> deprecated by
>>> UEFI 2.7 they weren't willing to adopt it.
>>
>> It's pity, but the motivation to deprecate is "lack of use" which is
>> not true. That's why I would suggest to escalate this to UEFI
>> committee.
>>
>>> Stuart, anything else you want to add here?
>>
>> Darren, what's your opinion about this?
>>
>> P.S. I'm not against this approach (just some technical comments I
>> already shared), but on the other hand it would be nice to have undo
>> that deprecation and follow the standard in new firmwares.
>> Would you agree?
> 
> Sure.  Due to the timing of how long this will take, even if SMM communication
> ACPI table is undone from deprecation we may have to still support both EPS
> and SMM communication ACPI table though (maybe it would be order of preference).
> 
> 

I have confirmation that the EPS table will be 16-byte aligned, so I can make that
change.  I'll send a v5 with that and the updated comment.