[Skiboot] [PATCH 3/3] platforms/qemu: Add slot table entries for a PCIe switch

[Cédric Le Goater]

On 06/11/2019 12:54, Oliver O'Halloran wrote:
> On Wed, Nov 6, 2019 at 10:48 PM Cédric Le Goater <clg at kaod.org> wrote:
>>
>> On 05/11/2019 08:43, Oliver O'Halloran wrote:
>>> Add a more complex slot table example to the Qemu platform. A more complex
>>> PCIe topology in qemu is useful for testing and it's simple enough to
>>> serve as an good example of how to use the slot table construction macros.
>>>
>>> Test in Qemu with:
>>>
>>> ./qemu-system-ppc64 -machine powernv -nographic -m 2G  \
>>> -bios ./skiboot.lid -kernel ./vmlinux -initrd ./petitfs \
>>> -device x3130-upstream,id=sw0_up,bus=pcie.2 \
>>>       -device xio3130-downstream,id=sw0_down0,chassis=1,slot=0,bus=sw0_up,addr=0 \
>>>               -drive file=./disk1.img,format=raw,if=none,id=nvme1 \
>>>               -device nvme,drive=nvme1,bus=sw0_down0,serial=1 \
>>>       -device xio3130-downstream,id=sw0_down1,chassis=1,slot=1,bus=sw0_up,addr=1 \
>>>       -device xio3130-downstream,id=sw0_down2,chassis=1,slot=2,bus=sw0_up,addr=2 \
>>>               -device e1000e,bus=sw0_down2
>>>
>>> Results in:
>>>
>>> PHB#0002:00:00.0 [ROOT] 1014 04c1 R:00 C:060400 B:01..05 SLOT=pcie.2
>>> PHB#0002:01:00.0 [SWUP] 104c 8232 R:02 C:060400 B:02..05 SLOT=sw0_up
>>> PHB#0002:02:00.0 [SWDN] 104c 8233 R:01 C:060400 B:03..03 SLOT=sw0_down0
>>> PHB#0002:03:00.0 [EP  ] 8086 5845 R:02 C:010802 (  mass-storage) LOC_CODE=sw0_down0
>>> PHB#0002:02:01.0 [SWDN] 104c 8233 R:01 C:060400 B:04..04 SLOT=sw0_down1
>>> PHB#0002:02:02.0 [SWDN] 104c 8233 R:01 C:060400 B:05..05 SLOT=sw0_down2
>>> PHB#0002:05:00.0 [EP  ] 8086 10d3 R:00 C:020000 (      ethernet) LOC_CODE=sw0_down2
>>>
>>> Cc: Cédric Le Goater <clg at kaod.org>
>>> Signed-off-by: Oliver O'Halloran <oohall at gmail.com>
>>
>>
>> Reviewed-by: Cédric Le Goater <clg at kaod.org>
>>
>>> ---
>>> Cedric, should we look at adding something like this to the default
>>> qemu machine?
>>
>>
>> Are you talking about defining a default set of PCIe devices for
>> the QEMU machine to represent better the HW ?
>>
>> We had that in the past but it was very incorrect and the QEMU
>> maintainers were asking us to be able to remove all devices with
>> the -nodefaults option. I just removed them.
>>
>> I agree that adding a new default set representing correctly HW
>> would be useful. Can we agree on a topology ? witherspoon has :
>>
>>  - one NIC (pec1)
>>  - one USB (pec1)
>>  - one bridge with a SATA device  (pec2)
>>
>> Thanks,
>>
>> C.
> 
> The actual devices I'm not that concerned about. It's more about
> having a a PCIe topology that's more than just a single device under
> every PHB. The P8 model was supposed to reflect a palmetto which had a
> switch under one of the PHBs so I figured we should have something
> similar in the default machine model.

ok. Yes. We should do the same for P9 (witherspoon).

C.