Re: [PATCH v5 05/14] vfio/mdev: idxd: add basic mdev registration and helper functions

From: Dave Jiang
Date: Tue Mar 02 2021 - 02:46:06 EST



On 2/10/2021 4:59 PM, Jason Gunthorpe wrote:
On Fri, Feb 05, 2021 at 01:53:24PM -0700, Dave Jiang wrote:

<-- cut for brevity -->


+static int vdcm_idxd_set_msix_trigger(struct vdcm_idxd *vidxd,
+ unsigned int index, unsigned int start,
+ unsigned int count, uint32_t flags,
+ void *data)
+{
+ int i, rc = 0;
+
+ if (count > VIDXD_MAX_MSIX_ENTRIES - 1)
+ count = VIDXD_MAX_MSIX_ENTRIES - 1;
+
+ if (count == 0 && (flags & VFIO_IRQ_SET_DATA_NONE)) {
+ /* Disable all MSIX entries */
+ for (i = 0; i < VIDXD_MAX_MSIX_ENTRIES; i++) {
+ rc = msix_trigger_unregister(vidxd, i);
+ if (rc < 0)
+ return rc;
+ }
+ return 0;
+ }
+
+ for (i = 0; i < count; i++) {
+ if (flags & VFIO_IRQ_SET_DATA_EVENTFD) {
+ u32 fd = *(u32 *)(data + i * sizeof(u32));
+
+ rc = msix_trigger_register(vidxd, fd, i);
+ if (rc < 0)
+ return rc;
+ } else if (flags & VFIO_IRQ_SET_DATA_NONE) {
+ rc = msix_trigger_unregister(vidxd, i);
+ if (rc < 0)
+ return rc;
+ }
+ }
+ return rc;
+}
+
+static int idxd_vdcm_set_irqs(struct vdcm_idxd *vidxd, uint32_t flags,
+ unsigned int index, unsigned int start,
+ unsigned int count, void *data)
+{
+ int (*func)(struct vdcm_idxd *vidxd, unsigned int index,
+ unsigned int start, unsigned int count, uint32_t flags,
+ void *data) = NULL;
+ struct mdev_device *mdev = vidxd->vdev.mdev;
+ struct device *dev = mdev_dev(mdev);
+
+ switch (index) {
+ case VFIO_PCI_INTX_IRQ_INDEX:
+ dev_warn(dev, "intx interrupts not supported.\n");
+ break;
+ case VFIO_PCI_MSI_IRQ_INDEX:
+ dev_dbg(dev, "msi interrupt.\n");
+ switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
+ case VFIO_IRQ_SET_ACTION_MASK:
+ case VFIO_IRQ_SET_ACTION_UNMASK:
+ break;
+ case VFIO_IRQ_SET_ACTION_TRIGGER:
+ func = vdcm_idxd_set_msix_trigger;
This would be a good place to insert a common VFIO helper library to
take care of the MSI-X emulation for IMS.

Hi Jason,

So after looking at the code in vfio_pci_intrs.c, I agree that the set_irqs code between VFIO_PCI and this driver can be made in common. Given that Alex doesn't want a vfio_pci device embedded in the driver, I think we'll need some sort of generic VFIO device that can be used from the vfio_pci side and vfio_mdev side to pass down in order to have common support library functions. Do you have any thoughts on how to do this cleanly architecturally? Also, with vfio_pci common split [1] still being worked on, do you think we can defer the work on making the interrupt setup code common until the vfio_pci split work settles? Thanks!

[1]: https://lore.kernel.org/kvm/20210201162828.5938-1-mgurtovoy@xxxxxxxxxx/