[pve-devel] applied: [PATCH v2 qemu-server 1/1] Add support for up to 16 PCI(e) devices

[Thomas Lamprecht]

On 06.09.19 14:50, Aaron Lauterer wrote:
> 
> 
> On 9/6/19 2:45 PM, Dominik Csapak wrote:
>> looks mostly ok, one (important) comment inline
>>
>>>       # win7 is picky about pcie assignments
>>>       hostpci0bus0 => { bus => "pcie.0", addr => 16 },
>>>       hostpci1bus0 => { bus => "pcie.0", addr => 17 },
>>>       hostpci2bus0 => { bus => "pcie.0", addr => 18 },
>>>       hostpci3bus0 => { bus => "pcie.0", addr => 19 },
>>>       ivshmem => { bus => 'pcie.0', addr => 20 },
>>> +    hostpci4bus0 => { bus => "pcie.0", addr => 9 },
>>> +    hostpci5bus0 => { bus => "pcie.0", addr => 10 },
>>> +    hostpci6bus0 => { bus => "pcie.0", addr => 11 },
>>> +    hostpci7bus0 => { bus => "pcie.0", addr => 12 },
>>> +    hostpci8bus0 => { bus => "pcie.0", addr => 13 },
>>> +    hostpci9bus0 => { bus => "pcie.0", addr => 14 },
>>> +    hostpci10bus0 => { bus => "pcie.0", addr => 15 },
>>> +    hostpci11bus0 => { bus => "pcie.0", addr => 20 },
>>
>> addr 20 is already used by ivshmem
>>
>> i would prefer to have the list in order of the addresses, so that this
>> will be more obvious and does not happen. also thomas mentioned offlist that it would be nice to have a test that automatically checks this, and i agree, but no one had time to do this (for now)
>>
> 
> Thanks for catching that one!
> 

Applied, with a fixup for the address conflicts by incrementing the addr
noted in hostpci10bus0 to hostpci15bus0 by one, holler if I did something
stupid.

Oh, and I just added the test for PCI/PCIe "bus:addr" conflicts, I do not
like that we had no safety net whatsoever here and just hoped someone along
the chain catches a duplicated, may have worked 'til now but IMO that would
gone wrong sooner or later, and probably me who'd overlook it..

I throw in a slight cleanup of the config_to_command hostpci loop, and two
additional config2command tests for hostpci with q35 and pc machine types,
pci and pcie pass-through and windows/linux os types..

While the pcie hostpci pass-through may change under certain circumstances,
as it hinders live migration anyway (but snapshots?), it does not hurt to
actively track this..