A little confusing inconsistency
Jukka A. Ukkonen
jau789 at gmail.com
Mon Mar 21 10:35:33 UTC 2016
On 03/20/16 18:17, Ian Lepore wrote:
> On Sun, 2016-03-20 at 15:30 +0200, Jukka A. Ukkonen wrote:
>> Hello all,
>>
>> Why does sysctl report hw.platform on arm and hw.model on amd64?
>> The content is apparently intended to be analogous.
>>
>> E.g. on RPI2 ...
>> hw.platform: bcm2836
>>
>> and then on amd64 ...
>> hw.model: AMD Opteron(tm) Processor 4162 EE
>>
>> Is this just a little lapsus or intentional for some reason?
>> I noticed this when I tried bsdstats on RPI2. It complained
>> about missing OID hw.model.
>
> In the armv6 world, driven by FDT data, hw.model should be the value
> from the device tree model property. But that isn't going to lead to
> "consistency" either, because then it will be "Wandboard Quad" or
> something similar.
Somehow my mind works such that I would expect hw.model to contain
additional information about a specific CPU model because there can
be a lot of variants of the same CPU architecture. E.g. bcm2836
certainly matches that purpose. Similarly I would expect hw.platform
to tell something about the mother board or that sort of info, like
"Wandboard Quad", "Raspberry Pi 2 Model B", "Super Micro H8DCL-6F",
etc.
In my mind it would make sense to have sysctl report something like...
hw.machine: arm
hw.machine_arch: armv6hf
hw.model: bcm2836
hw.platform: Raspberry Pi 2 Model B
Of course armv6hf could be only armv6, and bcm2836 might equally
well be something like "bcm2836,cortex-a7". Anyhow some consistency
about which OIDs can be expected to be available and what is their
content would be nice.
--jau
More information about the freebsd-arm
mailing list