On Mon, Jan 10, 2011 at 11:25:48PM +0200, Michael S. Tsirkin wrote: > On Mon, Jan 10, 2011 at 04:55:03PM -0200, Marcelo Tosatti wrote: > > Expose no_hotplug attribute via I/O port, so ACPI BIOS can indicate > > removability status to guest OS. > > > > An updated seabios is required to make use of this feature (seabios.git > > commit ID 3c241edf3d7ef29c21). > > > > Signed-off-by: Marcelo Tosatti <mtosa...@redhat.com> > > Tested-by: Gleb Natapov <g...@redhat.com> > > > > Index: qemu/hw/acpi_piix4.c > > =================================================================== > > --- qemu.orig/hw/acpi_piix4.c > > +++ qemu/hw/acpi_piix4.c > > @@ -37,6 +37,7 @@ > > #define GPE_BASE 0xafe0 > > #define PCI_BASE 0xae00 > > #define PCI_EJ_BASE 0xae08 > > +#define PCI_RMV_BASE 0xae0c > > > > #define PIIX4_PCI_HOTPLUG_STATUS 2 > > > > @@ -73,6 +74,7 @@ typedef struct PIIX4PMState { > > /* for pci hotplug */ > > struct gpe_regs gpe; > > struct pci_status pci0_status; > > + uint32_t pci0_hotplug_enable; > > } PIIX4PMState; > > > > static void piix4_acpi_system_hot_add_init(PCIBus *bus, PIIX4PMState *s); > > @@ -322,6 +324,25 @@ static const VMStateDescription vmstate_ > > } > > }; > > > > +static void piix4_update_hotplug(PIIX4PMState *s) > > +{ > > + PCIDevice *dev = &s->dev; > > + BusState *bus = qdev_get_parent_bus(&dev->qdev); > > + DeviceState *qdev, *next; > > + > > + s->pci0_hotplug_enable = 0; > > + > > + QLIST_FOREACH_SAFE(qdev, &bus->children, sibling, next) { > > + PCIDeviceInfo *info = container_of(qdev->info, PCIDeviceInfo, > > qdev); > > + PCIDevice *pdev = DO_UPCAST(PCIDevice, qdev, qdev); > > + int slot = PCI_SLOT(pdev->devfn); > > + > > + if (!info->no_hotplug) { > > + s->pci0_hotplug_enable |= (1 << slot); > > + } > > + } > > +} > > + > > So a slot starts as non-removable and becomes removable when you add a > hotpluggable device? Isn't this a bit strange? > I would say all slots should be removable unless there's a > non hotpluggable device there ...
Ugh, thats right, thanks.