Re: [dmidecode] OEM type 236 for HPe Gen10(+) servers

[Erwan Velu]

[...]

> I would like to blame the vendors for doing weird things, however I
> can't find anything in the specification that would prevent them from
> putting the entries in the DMI table in any order they like. Therefore
> I'm afraid we have to support all cases.
>

Yeah, was on the same page but HPe exactly told me that.
There is no enforcement of the specification to do in this logic.


>
> The only way I can think of to actually support that is to go for a
> 2-pass decoding. First pass will only look for entry type 1, record the
> vendor and stop. Second pass will be identical to what we have today,
> except that it won't set the vendor when seeing type 1, because that
> was already done during pass 1. This will make decoding a bit slower
> obviously, but thankfully most vendors *do* sort the entries by type
> number, so type 1 should be found quickly and the first pass should be
> pretty fast in most cases.
>
> The following fix should do the trick:
>

Oh thanks for this patch, that's very reactive.

I rebased my patch on top of yours and it works perfectly.
Please find attached the v2 of my patch which is now into the hpe specific
case.

If you agree, I'd love to sort this hpe specific part into a numerical
order ;)

Erwan,

v2-0001-oem-Adding-HPE-ProLiant-HDD-Backplane-type-236.patch
Description: Text Data