[systemd-devel] acpi_index value reported as 0 - not allowed?

[Andrei Borzenkov]

14.09.2018 17:08, Joe Hershberger пишет:
> 14 07:39:03 UTC 2018, Andrei Borzenkov wrote:
>> 13.09.2018 22:05, Joe Hershberger пишет:
>>> Hi,
>>>
>>> I have a system which reports acpi_index as 0, which matches the
>>> labeling on the device. Looking at the source in udev, it seems to be
>>> expecting an index greater than 0.
>>>
>>
>> You should have at least mentioned what system it is.
> 
> Apologies, I should have elaborated. The system is one that we are in
> the middle of developing. It is an x64 target. We are using the DSDT
> table to return the function 7 in _DSM.
> 
>>
>>> Looking at https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_2.7.1.pdf
>>
>> This is SMBIOS specification while acpi_index is obtained by ACPI _DSM
>> method. Unfortunately PCI specifications that define it are not publicly
>> accessible.
> 
> Is there a statement in the PCI specs that rule out using 0 or are you
> simply stating that you can't reference it here?
> 

I do not have access to official final specs. The change request is
available and does not restrict value range. Actually the only
requirement is that instance numbers be unique.

https://pcisig.com/sites/default/files/specification_documents/ECR_Slot_Naming-10.pdf


> Thanks,
> -Joe
> 
>>
>>> it provides an example that uses 1 and 2, but I don't see anything
>>> that claims that 0 is allowed or disallowed.
>>>
>>> Would it be reasonable to send a patch that makes udev allow an index
>>> of 0 or is there a reason not to allow 0?
>>>
>>> Thanks,
>>> -Joe