Re: [dmidecode] [PATCH] update dmidecode to parse Modern Management Controller blocks

[Neil Horman]

On Thu, Aug 09, 2018 at 09:50:54AM +0200, Jean Delvare wrote:
> Hi Neil,
> 
> On Wed, 8 Aug 2018 08:08:48 -0400, Neil Horman wrote:
> > (...) Regardless, do you
> > think we need to add some sanity check for SMBIOS version 3 where that 
> > length
> > isn't specified, and refuse to decode it for the ambiguity it provides?
> 
> Unspecified length is not "SMBIOS version 3". It spans from SMBIOS
> specification 2.7.0 to 3.1.1. That's a huge period of time (almost 8
> years), but in practice I only know of 2 systems implementing a type 42
> record during that time, and both records are broken. So the effort to
> deal with this case should be minimum. I would just silently skip
> decoding them (as is done today already.) This can always be revisited
> later if and only if at least one real world system would benefit from
> additional work.
> 
Understood, then I think the best course of action is to modify the version
check in the case statement to only call
decode_management_controller_structure() if the version is greater than 0x311,
rather than 0x300 as it is today.  If its less than that, we can just do the
traditional decode of Interface type and Vendor ID.  I'll update the patch and
test it on my new system, which should hopefully be available to me today, and
post v4.

Neil

> Thanks,
> -- 
> Jean Delvare
> SUSE L3 Support
>