On Tue, 2015-06-09 at 11:37 +0800, Chen Fan wrote:
> Signed-off-by: Chen Fan <chen.fan.f...@cn.fujitsu.com>
> ---
> hw/vfio/pci.c | 81
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 81 insertions(+)
>
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index 97af143..a4e8658 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -156,6 +156,7 @@ typedef struct VFIOPCIDevice {
> PCIHostDeviceAddress host;
> EventNotifier err_notifier;
> EventNotifier req_notifier;
> + Notifier sec_bus_reset_notifier;
> int (*resetfn)(struct VFIOPCIDevice *);
> uint32_t features;
> #define VFIO_FEATURE_ENABLE_VGA_BIT 0
> @@ -171,6 +172,7 @@ typedef struct VFIOPCIDevice {
> bool req_enabled;
> bool has_flr;
> bool has_pm_reset;
> + bool needs_bus_reset;
> bool rom_read_failed;
> } VFIOPCIDevice;
>
> @@ -3085,6 +3087,81 @@ static int vfio_check_devices_host_bus_reset(void)
> return 0;
> }
>
> +/* Update all the reset state of the affected devices in VM */
> +static void vfio_pci_host_needs_bus_reset(Notifier *n, void *opaque)
> +{
> + VFIOPCIDevice *vdev = container_of(n, VFIOPCIDevice,
> sec_bus_reset_notifier);
> + VFIODevice *vbasedev = &vdev->vbasedev;
> + PCIDevice *parent = opaque;
> + VFIOGroup *group;
> + struct vfio_pci_hot_reset_info *info = NULL;
> + struct vfio_pci_dependent_device *devices;
> + int i, ret;
> +
> + ret = vfio_get_hot_reset_info(vdev, &info);
> + if (ret) {
> + goto out;
> + }
> +
> + vdev->needs_bus_reset = true;
> + /*
> + * qbus_reset_all always reset the devices from the depth level,
> + * here only need to do the reset for the device encounter the aer,
> + * in order to avoid reset the affected device first below the bus,
> + * we only set the top level devices as needs_reset.
> + */
> + if (parent == pci_bridge_get_device(vdev->pdev.bus)) {
> + vbasedev->needs_reset = true;
> + }
> +
> + /* List all affected devices by bus reset */
> + devices = &info->devices[0];
> +
> + /* Verify that we have all the groups required */
> + for (i = 0; i < info->count; i++) {
> + PCIHostDeviceAddress host;
> + VFIOPCIDevice *tmp;
> + VFIODevice *vbasedev_iter;
> +
> + host.domain = devices[i].segment;
> + host.bus = devices[i].bus;
> + host.slot = PCI_SLOT(devices[i].devfn);
> + host.function = PCI_FUNC(devices[i].devfn);
> +
> + /* Skip the current device */
> + if (vfio_pci_host_match(&host, &vdev->host)) {
> + continue;
> + }
> +
> + /* Ensure we own the group of the affected device */
> + QLIST_FOREACH(group, &vfio_group_list, next) {
> + if (group->groupid == devices[i].group_id) {
> + break;
> + }
This is insufficient now that groups can be opened without a container.
> + }
> +
> + if (!group) {
> + goto out;
> + }
> +
> + /* Ensure affected devices for reset under the same bus */
> + QLIST_FOREACH(vbasedev_iter, &group->device_list, next) {
> + if (vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
> + continue;
> + }
> + tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
> + if (vfio_pci_host_match(&host, &tmp->host)) {
> + tmp->needs_bus_reset = true;
> + tmp->vbasedev.needs_reset = false;
> + break;
> + }
> + }
> + }
> +
> +out:
> + g_free(info);
> +}
> +
> static int vfio_setup_aer(VFIOPCIDevice *vdev, uint8_t cap_ver,
> int pos, uint16_t size)
> {
> @@ -3156,6 +3233,9 @@ static int vfio_setup_aer(VFIOPCIDevice *vdev, uint8_t
> cap_ver,
> pdev->exp.aer_cap + PCI_ERR_UNCOR_SEVER, 4);
> pci_long_test_and_clear_mask(exp_cap + PCI_ERR_UNCOR_SEVER, ~severity);
>
> + vdev->sec_bus_reset_notifier.notify = vfio_pci_host_needs_bus_reset;
> + pci_bus_add_reset_notifier(pdev->bus, &vdev->sec_bus_reset_notifier);
> +
> return 0;
>
> error:
> @@ -4129,6 +4209,7 @@ static void vfio_exitfn(PCIDevice *pdev)
>
> vfio_unregister_req_notifier(vdev);
> vfio_unregister_err_notifier(vdev);
> + pci_bus_remove_reset_notifier(&vdev->sec_bus_reset_notifier);
> pci_device_set_intx_routing_notifier(&vdev->pdev, NULL);
> vfio_disable_interrupts(vdev);
> if (vdev->intx.mmap_timer) {
