[Qemu-devel] Re: May I use -device in qemu-system command to attach to PCIe bus?

[Wei Xu]

Isaku, 

We are on same page now; I also consider the qdev_find way. Both ways can
work and let's wait for others' opinions...

Wei


On 10/28/10 4:25 AM, "Isaku Yamahata" <address@hidden> wrote:

> On Thu, Oct 28, 2010 at 03:37:27AM -0700, Wei Xu wrote:
>> Isaku,
>> 
>> To make things clear, let me rephrase the problem. With q35/vPCIe, in VMM
>> monitor, we can do hotplug like:
>>  pci_add auto|<bus:dev> nic|storage
>> to hotplug a device to PCIe/PCI bus.
>> 
>> But we have two problems  here:
>> (1) command line for example, "-net nic,addr=<bus:dev>" always failed
>> because it cannot find the bus.
>> (2) If "pci_add auto" in monitor or no "addr=<bus:dev" in command line, the
>> device will be attached to PCI bus, in stead of PCIe ports
>> 
>> I solved the first problem. The root cause is pcie_root_write_config (which
>> init SECONDARY_BUS config) happened after pci_nic_init. The latter use
>> pci_find_bus to find the parent bus but failed because config space for
>> secondary bus still not initialized.
> 
> Okay, now the issue is clear to me.
> When I tried to clean up pci bridge code, I included the similar lines
> in bridge initialization function.
> But Michael complained about it, so I dropped the lines for merge.
> 
> I think there are two ways to address it.
> - initialize secondary/subordinate bus register by qemu
>   i.e. something like the patch below.
> 
> - use qdev device path
>   i.e. qbus_find() instead of pci_find_bus().
>   I don't know if the corresponding command line option already exists or not.
> 
> My preference is the former, but discussion is necessary to get consensus.
> My plan was to raise the issue when I address pv pci bus numbering issue.
> I've expected that other developers wouldn't think the issue important
> because multiple pci bus can't be used easily right now.
> 
> thanks,
> 
>> 
>> The fix is like:
>> 
>> address@hidden contents]$ git diff hw/pci_bridge.c
>> diff --git a/hw/pci_bridge.c b/hw/pci_bridge.c
>> index c048305..042c1f9 100644
>> --- a/hw/pci_bridge.c
>> +++ b/hw/pci_bridge.c
>> @@ -203,6 +203,7 @@ void pci_bridge_set_bus_number(PCIBridge *br,
>>      uint16_t domain;
>>      uint8_t primary_bus;
>>      uint8_t devfn;
>> +    uint8_t *conf;
>>  
>>      if (!pci_bus_number) {
>>          pci_bus_number = qemu_mallocz(sizeof(*pci_bus_number));
>> @@ -215,10 +216,15 @@ void pci_bridge_set_bus_number(PCIBridge *br,
>>  
>>      d = &br->dev;
>>      devfn = d->devfn;
>> +       conf = d->config;
>>  
>>      br->sec_bus = secondary;
>>      br->sub_bus = subordinate;
>>  
>> +       //TODO wexu2 workaround of "-net nic,addr=<bus:dev>"
>> +    pci_set_byte(conf + PCI_SECONDARY_BUS, secondary);
>> +    pci_set_byte(conf + PCI_SUBORDINATE_BUS, subordinate);
>> +
>>      bus = d->bus;
>>      primary_bus = 0;
>>      if ((d = pci_bridge_get_device(bus)) != NULL){
>> 
>> The fix is a little bit hacky -- welcome any advice.
>> 
>> For 2nd issue, current code assume PCI bus is flat and not straight forward
>> to change the behavior.
>> 
>> Wei Xu
>> 
>> 
>> On 10/27/10 10:16 PM, "Isaku Yamahata" <address@hidden> wrote:
>> 
>>> Oh, now I'm seeing your point. Some codes of qemu assume that
>>> the pci bus is flat. They need fixes.
>>> 
>>> Right now I've found
>>> - parse_pci_devfn in qdev-properties.c
>>>   This corresponds to (1)
>>> - pci_get_bus_devfn() in pci.c
>>>   This corresponds to (2)
>>> 
>>> thanks,
>>> 
>>> On Wed, Oct 27, 2010 at 08:51:29PM -0700, Wei Xu wrote:
>>>> Isaku,
>>>> 
>>>> I found two issues:
>>>> (1) command line option is not working: even with addr=<bus>, it complained
>>>> bus cannot be found;
>>>> (2) monitor: pci_add auto ... cannot work.
>>>> 
>>>> Wei
>>>> 
>>>> 
>>>> On 10/27/10 8:03 PM, "Isaku Yamahata" <address@hidden> wrote:
>>>> 
>>>>> On Wed, Oct 27, 2010 at 09:51:47AM -0700, Wei Xu wrote:
>>>>>> Isaku,
>>>>>> 
>>>>>> I have one question regarding EP and PCIe bus: now only way is to use
>>>>>> "pci_add <bus:dev> .." to hot-plug the device to a PCIe port. Looks like
>>>>>> command line with -device parameter automatically use PCI bus? Is it
>>>>>> possible to support it? Have to provide the <bus:dev> in -device
>>>>>> parameter?
>>>>> 
>>>>> I'm not sure I got your question right.
>>>>> We can populate the device under an express port with -device command line
>>>>> option. At least it should be possible.
>>>>> 
>>>>> For implementation detail, qemu doesn't distinguish pci express
>>>>> from conventional pci so strictly.
>>>>> The only difference is whether the bridge(pci host bridge or
>>>>> pci-to-pci bridge) is just express port because I implemented express
>>>>> ports as pci-to-pci bridge.
>>>>> 
>>>>> Or do you want to add express port with command line option?
>>>>> If so, not possible at the moment.
>>>> 
>>