Re: [Qemu-devel] [PATCH RFC v2 09/12] vfio/ccw: get irqs info and set the eventfd fd

[Alex Williamson]

On Thu, 12 Jan 2017 08:25:10 +0100
Dong Jia Shi <address@hidden> wrote:

> From: Xiao Feng Ren <address@hidden>
> 
> vfio-ccw resorts to the eventfd mechanism to communicate with userspace.
> We fetch the irqs info via the ioctl VFIO_DEVICE_GET_IRQ_INFO,
> register a event notifier to get the eventfd fd which is sent
> to kernel via the ioctl VFIO_DEVICE_SET_IRQS, then we can implement
> read operation once kernel sends the signal.
> 
> Signed-off-by: Xiao Feng Ren <address@hidden>
> ---
>  hw/vfio/ccw.c | 102 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 102 insertions(+)
> 
> diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
> index 93394c2..c6bfce7 100644
> --- a/hw/vfio/ccw.c
> +++ b/hw/vfio/ccw.c
> @@ -21,6 +21,7 @@
>  #include "hw/vfio/vfio-common.h"
>  #include "hw/s390x/s390-ccw.h"
>  #include "hw/s390x/ccw-device.h"
> +#include "qemu/error-report.h"
>  #include "standard-headers/asm-s390/vfio_ccw.h"
>  
>  #define TYPE_VFIO_CCW "vfio-ccw"
> @@ -30,6 +31,7 @@ typedef struct VFIOCCWDevice {
>      uint64_t io_region_size;
>      uint64_t io_region_offset;
>      struct ccw_io_region *io_region;
> +    EventNotifier io_notifier;
>  } VFIOCCWDevice;
>  
>  static void vfio_ccw_compute_needs_reset(VFIODevice *vdev)
> @@ -54,6 +56,98 @@ static void vfio_ccw_reset(DeviceState *dev)
>      ioctl(vcdev->vdev.fd, VFIO_DEVICE_RESET);
>  }
>  
> +static void vfio_ccw_io_notifier_handler(void *opaque)
> +{
> +    VFIOCCWDevice *vcdev = opaque;
> +
> +    if (!event_notifier_test_and_clear(&vcdev->io_notifier)) {
> +        return;
> +    }
> +}
> +
> +static void vfio_ccw_register_io_notifier(VFIOCCWDevice *vcdev, Error **errp)
> +{
> +    VFIODevice *vdev = &vcdev->vdev;
> +    struct vfio_irq_info *irq_info;
> +    struct vfio_irq_set *irq_set;
> +    size_t argsz;
> +    int32_t *pfd;
> +
> +    if (vdev->num_irqs != VFIO_CCW_NUM_IRQS) {
> +        error_setg(errp, "vfio: unexpected number of io irqs %u",
> +                   vdev->num_irqs);
> +        return;
> +    }

Same issue here as with region info, we may find reasons to add new
interrupts in the kernel, it's happened for PCI support with things
like device error recovery and requests to release the device.  Don't
lock us into never expanding these, just make sure that the index that
this implementation depends on is here.

> +
> +    argsz = sizeof(*irq_set);
> +    irq_info = g_malloc0(argsz);
> +    irq_info->index = VFIO_CCW_IO_IRQ_INDEX;
> +    irq_info->argsz = argsz;
> +    if (ioctl(vdev->fd, VFIO_DEVICE_GET_IRQ_INFO,
> +              irq_info) < 0 || irq_info->count < 1) {
> +        error_setg(errp, "vfio: Error getting irq info");
> +        goto get_error;
> +    }
> +
> +    if (event_notifier_init(&vcdev->io_notifier, 0)) {
> +        error_setg(errp, "vfio: Unable to init event notifier for IO");
> +        goto get_error;
> +    }
> +
> +    argsz = sizeof(*irq_set) + sizeof(*pfd);
> +    irq_set = g_malloc0(argsz);
> +    irq_set->argsz = argsz;
> +    irq_set->flags = VFIO_IRQ_SET_DATA_EVENTFD |
> +                     VFIO_IRQ_SET_ACTION_TRIGGER;
> +    irq_set->index = VFIO_CCW_IO_IRQ_INDEX;
> +    irq_set->start = 0;
> +    irq_set->count = 1;
> +    pfd = (int32_t *) &irq_set->data;
> +
> +    *pfd = event_notifier_get_fd(&vcdev->io_notifier);
> +    qemu_set_fd_handler(*pfd, vfio_ccw_io_notifier_handler, NULL, vcdev);
> +    if (ioctl(vdev->fd, VFIO_DEVICE_SET_IRQS, irq_set)) {
> +        error_setg(errp, "vfio: Failed to set up io notification");
> +        qemu_set_fd_handler(*pfd, NULL, NULL, vcdev);
> +        event_notifier_cleanup(&vcdev->io_notifier);
> +        goto set_error;
> +    }
> +
> +set_error:
> +    g_free(irq_set);
> +
> +get_error:
> +    g_free(irq_info);
> +}
> +
> +static void vfio_ccw_unregister_io_notifier(VFIOCCWDevice *vcdev)
> +{
> +    struct vfio_irq_set *irq_set;
> +    size_t argsz;
> +    int32_t *pfd;
> +
> +    argsz = sizeof(*irq_set) + sizeof(*pfd);
> +    irq_set = g_malloc0(argsz);
> +    irq_set->argsz = argsz;
> +    irq_set->flags = VFIO_IRQ_SET_DATA_EVENTFD |
> +                     VFIO_IRQ_SET_ACTION_TRIGGER;
> +    irq_set->index = VFIO_CCW_IO_IRQ_INDEX;
> +    irq_set->start = 0;
> +    irq_set->count = 1;
> +    pfd = (int32_t *) &irq_set->data;
> +    *pfd = -1;
> +
> +    if (ioctl(vcdev->vdev.fd, VFIO_DEVICE_SET_IRQS, irq_set)) {
> +        error_report("vfio: Failed to de-assign device io fd");
> +    }
> +
> +    qemu_set_fd_handler(event_notifier_get_fd(&vcdev->io_notifier),
> +                        NULL, NULL, vcdev);
> +    event_notifier_cleanup(&vcdev->io_notifier);
> +
> +    g_free(irq_set);
> +}
> +
>  static void vfio_ccw_get_region(VFIOCCWDevice *vcdev, Error **errp)
>  {
>      VFIODevice *vdev = &vcdev->vdev;
> @@ -193,8 +287,15 @@ static void vfio_ccw_realize(DeviceState *dev, Error 
> **errp)
>          goto out_region_err;
>      }
>  
> +    vfio_ccw_register_io_notifier(vcdev, errp);
> +    if (*errp) {
> +        goto out_notifier_err;
> +    }
> +
>      return;
>  
> +out_notifier_err:
> +    vfio_ccw_put_region(vcdev);
>  out_region_err:
>      vfio_put_device(vcdev);
>  out_device_err:
> @@ -217,6 +318,7 @@ static void vfio_ccw_unrealize(DeviceState *dev, Error 
> **errp)
>          cdc->unrealize(cdev, errp);
>      }
>  
> +    vfio_ccw_unregister_io_notifier(vcdev);
>      vfio_ccw_put_region(vcdev);
>      vfio_put_device(vcdev);
>      vfio_put_group(group);