[PATCH libdrm v3 1/1] amdgpu: move asic id table to a separate file

Alex Deucher alexdeucher at gmail.com
Wed May 31 13:19:39 UTC 2017


On Wed, May 31, 2017 at 7:19 AM, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> On 30 May 2017 at 22:59, Li, Samuel <Samuel.Li at amd.com> wrote:
>>>  - Marketing can make mistakes or have IT glitches The inconsistent use of "(TM)" and using a 67C2:00 is something one wants to double-check with them.
>> Marketing names are there for a lot of reasons. The code here is to pass the names only.
>> If you are interested in a vendor's marketing names, please reach out to the vendor, e.g. through your contacts in the vendor who also shares your interest.
>>
> True, yet orthogonal to what I'm saying. As people point potential
> mistakes you (perhaps not personally) want to check if those are
> genuine or not.
> If things are correct, say 67C2:00 is valid, simply mention "yes A/B
> is bit unusual, yet it's the correct name".
>
>>>- Having a separate file so that clients can update/edit it does not help much.
>> Please say it to pci.ids/usb.ids :)
>>
> Those files have many more users than the amdgpu.ids, have existed for
> ~20 years. Since you refer to pci.ids - why not reuse it but grow a
> local copy instead?

The problem with pci.ids is that it doesn't take into account pci
revisions so it would be comparable work either way.

Alex

>
>>> Adding ~200 loc for ~170 devices entries sounds like a step in the wrong direction.
>> Check the vendor's entries in pci.ids, and you might have some better idea.
>>
> The file lists ~2.8k entries for 1002 ATI/AMD, yet those include north
> bridges and others which are not applicable here.
> I'm afraid your argument is too subtle.
>
> As they say "Not my circus not my moneys" (don't take this the wrong way).
>
> Regards,
> Emil
> P.S. Please convince your email client to quote properly?


More information about the amd-gfx mailing list