[libvirt] [PATCH] qemu: Honor the original properties of PCI device when detaching

[Eric Blake]

On 10/17/2011 05:07 AM, Osier Yang wrote:
> This patch fixes two problems:
>      1) The device will be reattached to host even if it's not
>         managed, as there is a "pciDeviceSetManaged".
>      2) The device won't be reattached to host with original
>         driver properly. As it doesn't honor the device original
>         properties which are maintained by driver->activePciHostdevs.
> ---
>   src/qemu/qemu_hotplug.c |   16 ++++++++--------
>   1 files changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
> index bfa524b..f3f0060 100644
> --- a/src/qemu/qemu_hotplug.c
> +++ b/src/qemu/qemu_hotplug.c
> @@ -1956,6 +1956,7 @@ int qemuDomainDetachHostPciDevice(struct qemud_driver *driver,
>       qemuDomainObjPrivatePtr priv = vm->privateData;
>       int i, ret;
>       pciDevice *pci;
> +    pciDevice *activePci;
>
>       for (i = 0 ; i<  vm->def->nhostdevs ; i++) {
>           if (vm->def->hostdevs[i]->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS ||
> @@ -2015,16 +2016,15 @@ int qemuDomainDetachHostPciDevice(struct qemud_driver *driver,
>                          detach->source.subsys.u.pci.bus,
>                          detach->source.subsys.u.pci.slot,
>                          detach->source.subsys.u.pci.function);
> -    if (!pci)
> -        ret = -1;
> -    else {
> -        pciDeviceSetManaged(pci, detach->managed);
> -        pciDeviceListDel(driver->activePciHostdevs, pci);
> -        if (pciResetDevice(pci, driver->activePciHostdevs, NULL)<  0)
> +    if (pci) {
> +        activePci = pciDeviceListSteal(driver->activePciHostdevs, pci);
> +        if (pciResetDevice(activePci, driver->activePciHostdevs, NULL)<  0)
>               ret = -1;
> -        pciDeviceReAttachInit(pci);
> -        qemuReattachPciDevice(pci, driver);
> +        qemuReattachPciDevice(activePci, driver);

This calls qemuReattachPciDevice even if pciResetDevice failed.  I don't 
think that is right.

However, I agree with you that hot-unplug cannot make a device managed 
if it was not already managed, and also that it should be using the 
device attributes from driver->activePciHostdevs rather than from the 
just-created pci in order to properly decide how much work to do on the 
unplug event.

-- 
Eric Blake   eblake at redhat.com    +1-801-349-2682
Libvirt virtualization library http://libvirt.org