[libvirt] [PATCH v3] pci address conflict when virtio disk with drive type

[Xu He Jie]

于 2011年10月28日 06:28, Eric Blake 写道:
> On 10/23/2011 05:31 AM, Xu He Jie wrote:
>> When using the xml as below:
>> ------------------------------------------------------
>> <devices>
>> <emulator>/home/soulxu/data/work-code/qemu-kvm/x86_64-softmmu/qemu-system-x86_64</emulator> 
>>
>> <disk type='file' device='disk'>
>> <driver name='qemu' type='qcow2'/>
>> <source file='/home/soulxu/data/VM/images/linux.img'/>
>> <target dev='vda' bus='virtio'/>
>> <address type='drive' controller='0' bus='0' unit='0'/>
>> </disk>
>> <input type='mouse' bus='ps2'/>
>> <graphics type='vnc' port='-1' autoport='yes'/>
>> <video>
>> <model type='cirrus' vram='9216' heads='1'/>
>> <address type='pci' domain='0x0000' bus='0x00' slot='0x02' 
>> function='0x0'/>
>> </video>
>> <memballoon model='virtio'>
>> <address type='pci' domain='0x0000' bus='0x00' slot='0x03' 
>> function='0x0'/>
>> </memballoon>
>> </devices>
>
> Lack of indentation made this harder to read.
>
>> ------------------------------------------------------
>>
>> Then can't startup qemu, the error message as below:
>> virsh # start test-vm
>> error: Failed to start domain test-vm
>> error: internal error process exited while connecting to monitor: 
>> qemu-system-x86_64: -device 
>> virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x3: PCI: slot 3 
>> function 0 not available for virtio-balloon-pci, in use by 
>> virtio-blk-pci
>> qemu-system-x86_64: -device 
>> virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x3: Device 
>> 'virtio-balloon-pci' could not be initialized
>>
>> So adding check for bus type and address type. Only the address of 
>> pci type support by virtio bus.
>>
>> Signed-off-by: Xu He Jie<xuhj at linux.vnet.ibm.com>
>> ---
>> src/qemu/qemu_command.c | 10 +++++++---
>> 1 files changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
>> index 30c0be6..7c4bc0a 100644
>> --- a/src/qemu/qemu_command.c
>> +++ b/src/qemu/qemu_command.c
>> @@ -1321,13 +1321,17 @@ qemuAssignDevicePCISlots(virDomainDefPtr def, 
>> qemuDomainPCIAddressSetPtr addrs)
>>
>> /* Disks (VirtIO only for now */
>
> While we're here, add the closing ).
>
>> for (i = 0; i< def->ndisks ; i++) {
>> - if (def->disks[i]->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE)
>> - continue;
>> -
>> /* Only VirtIO disks use PCI addrs */
>> if (def->disks[i]->bus != VIR_DOMAIN_DISK_BUS_VIRTIO)
>> continue;
>>
>> + if ((def->disks[i]->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)&&
>> + (def->disks[i]->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE)) {
>> + qemuReportError(VIR_ERR_INTERNAL_ERROR,
>
> This is user-visible, so VIR_ERR_CONFIG_UNSUPPORTED is probably better.
>
>> + _("virtio only supports device address type 'PCI' "));
>
> no trailing space in the message
>
>> + goto error;
>> + }
>
> Your patch fails 'make check', so it needs another respin that 
> resolves this ('make check -C tests TESTS=qemuxml2argvtest 
> VIR_TEST_DEBUG=1'):
>
> 58) QEMU XML-2-ARGV disk-ioeventfd ...
> Offset 387
> Expect [4,drive=drive-virtio-disk0,id=virtio-disk0 -drive 
> file=/var/lib/libvirt/Fedora-14-x86_64-Live-KDE.iso,if=none,media=cdrom,id=drive-ide0-1-0 
> -device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 
> -device 
> virtio-net-pci,tx=bh,ioeventfd=off,vlan=0,id=net0,mac=52:54:00:e5:48:58,bus=pci.0,addr=0x3 
> -net user,vlan=0,name=hostnet0 -serial pty -usb -vnc 127.0.0.1:-809 
> -std-vga -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5]
> Actual [5,drive=drive-virtio-disk0,id=virtio-disk0 -drive 
> file=/var/lib/libvirt/Fedora-14-x86_64-Live-KDE.iso,if=none,media=cdrom,id=drive-ide0-1-0 
> -device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 
> -device 
> virtio-net-pci,tx=bh,ioeventfd=off,vlan=0,id=net0,mac=52:54:00:e5:48:58,bus=pci.0,addr=0x3 
> -net user,vlan=0,name=hostnet0 -serial pty -usb -vnc 127.0.0.1:-809 
> -std-vga -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x7]
>
> ... FAILED
>
> In other words, your patch silently caused pci devices to be 
> renumbered. If it was a bug in our tests for using the wrong addresses 
> in the first place, then let's fix our tests; but more likely, this 
> means your patch was incorrect and would break guest ABI stability if 
> it were applied.
>

I checked the code again. My patch was incorrect. if address's type is 
'VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI', it should not execute 
'qemuDomainPCIAddressSetNextAddr' again.

I will send patch v4 later.

Thanks for your review.