Re: [Qemu-ppc] [PATCH] ppc/pnv: Add model for Power8 PHB3 PCIe Host bridge

[Cédric Le Goater]

On 06/27/2018 10:41 AM, Benjamin Herrenschmidt wrote:
> On Wed, 2018-06-27 at 09:46 +0200, Cédric Le Goater wrote:
>> So the "IBM PHB3 PCIE Root Port" is already user createable.
>>
>> I can take a look at user createable PHB3s. I think this is OK from a model
>> perspective. The object is rather standalone, it needs the machine for 
>> the XICS fabric and a couple of ids, phb id and chip id. These can come
>> from the command line.
>>
>> We want at least one PHB3 per socket/chip though. 
> 
> We don't want the user to specify the SCOM addresses though

We don't need to.

>  (for the MMIO windows we should get skiboot to assign them).

yes. That's a small fix yet to be done. It can come later.

> If the user gets to specify a thing it would be which of the 3 or 4 HW
> PHBs of the chip it is, the SCOM addresses gets deduced.


So I modified a bit the model to support user creatable PHB3 devices and
could start a QEMU powernv machine with these options :
  
Raid controller on PHB0 :

  -device megasas,id=scsi0,bus=pci.0,addr=0x1
  -drive file=$file,if=none,id=drive-scsi0-0-0-0,format=qcow2,cache=none
  -device 
scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=1

Ethernet controller on PHB0 :

  -device e1000,netdev=net0,mac=C0:FF:EE:00:00:02,bus=pci.0,addr=0x2
  -netdev 
bridge,id=net0,helper=/usr/libexec/qemu-bridge-helper,br=virbr0,id=net0

USB controller on PHB0 :

  -device nec-usb-xhci,bus=pci.0,addr=0x7

Extra PHB1 on chip 0 :

  -device pnv-phb3,chip-id=0,phb-id=1,id=phb0.1

USB controller on PHB1 :

  -device nec-usb-xhci,bus=pci.1,addr=0x7 

Most of the code changes are fine and even fix a few problems.  

But, user created PHB3 devices are not parented to the PnvChip and 
this means that we need to : 

 1 - scan the full object hierarchy to populate the device tree 
 2 - look for the owning chip in the PHP3 realize routine 
 3 - grab the pnv machine to get the XICS fabric

2 & 3 are somewhat violation of the QOM models but we can live with 
that. 1 is annoying. I think.

So may be we should forget about the -device possibility and use fixed
values for the number of PHBs. The value would depend on the chip flavor.
 

Cheers,

C.