| Message ID | 1447341076-11573-2-git-send-email-marcel@redhat.com |
|---|---|
| State | New |
| Headers | show |
On Do, 2015-11-12 at 17:11 +0200, Marcel Apfelbaum wrote: > As part of porting the pxb device to Q35 remove the internal pci-2-pci > bridge. The only way to hot-pug devices on the extra PCI root buses > is by adding a pci-2-pci to the pxb before the firmware assign the > IO/mem resources. I think this breaks live migration and is guest-visible too. So not going to fly without backward-compatibility mode for old machine types ... Possibly it is easier to just have have pci and pcie pxb devices look differenty rather than adding all the compatibility fluff. cheers, Gerd
On 11/13/2015 11:13 AM, Gerd Hoffmann wrote: > On Do, 2015-11-12 at 17:11 +0200, Marcel Apfelbaum wrote: >> As part of porting the pxb device to Q35 remove the internal pci-2-pci >> bridge. The only way to hot-pug devices on the extra PCI root buses >> is by adding a pci-2-pci to the pxb before the firmware assign the >> IO/mem resources. > > I think this breaks live migration and is guest-visible too. So not > going to fly without backward-compatibility mode for old machine > types ... > > Possibly it is easier to just have have pci and pcie pxb devices look > differenty rather than adding all the compatibility fluff. > Hi Gerd, Thank you for the review. Yes, you are right about migration, but I prefer to add a compat-prop rather than different devices because I do want to remove the internal pci-bridge even for pxb PCI. The reason is the feedback I got from libvirt that "composed" device are not recommended. In conclusion, since I am going to get rid of the pci-bridge, the devices will be exactly the same (except for the internal bus). I want to use the same concept as virtio-pci devices that get to be "express" on PCIe, but remain PCI on PCI buses. Thanks, Marcel > cheers, > Gerd > >
diff --git a/hw/pci-bridge/pci_expander_bridge.c b/hw/pci-bridge/pci_expander_bridge.c index 57f8a37..0464dec 100644 --- a/hw/pci-bridge/pci_expander_bridge.c +++ b/hw/pci-bridge/pci_expander_bridge.c @@ -196,7 +196,7 @@ static gint pxb_compare(gconstpointer a, gconstpointer b) static int pxb_dev_initfn(PCIDevice *dev) { PXBDev *pxb = PXB_DEV(dev); - DeviceState *ds, *bds; + DeviceState *ds; PCIBus *bus; const char *dev_name = NULL; @@ -211,18 +211,13 @@ static int pxb_dev_initfn(PCIDevice *dev) } ds = qdev_create(NULL, TYPE_PXB_HOST); - bus = pci_bus_new(ds, "pxb-internal", NULL, NULL, 0, TYPE_PXB_BUS); + bus = pci_bus_new(ds, dev_name, NULL, NULL, 0, TYPE_PXB_BUS); bus->parent_dev = dev; bus->address_space_mem = dev->bus->address_space_mem; bus->address_space_io = dev->bus->address_space_io; bus->map_irq = pxb_map_irq_fn; - bds = qdev_create(BUS(bus), "pci-bridge"); - bds->id = dev_name; - qdev_prop_set_uint8(bds, PCI_BRIDGE_DEV_PROP_CHASSIS_NR, pxb->bus_nr); - qdev_prop_set_bit(bds, PCI_BRIDGE_DEV_PROP_SHPC, false); - PCI_HOST_BRIDGE(ds)->bus = bus; if (pxb_register_bus(dev, bus)) { @@ -230,7 +225,6 @@ static int pxb_dev_initfn(PCIDevice *dev) } qdev_init_nofail(ds); - qdev_init_nofail(bds); pci_word_test_and_set_mask(dev->config + PCI_STATUS, PCI_STATUS_66MHZ | PCI_STATUS_FAST_BACK);
As part of porting the pxb device to Q35 remove the internal pci-2-pci bridge. The only way to hot-pug devices on the extra PCI root buses is by adding a pci-2-pci to the pxb before the firmware assign the IO/mem resources. Signed-off-by: Marcel Apfelbaum <marcel@redhat.com> --- hw/pci-bridge/pci_expander_bridge.c | 10 ++-------- 1 file changed, 2 insertions(+), 8 deletions(-)