[RFC PATCH v4 24/24] vfio/iommufd: Make vfio cdev pre-openable by passing a file handle

[Zhenzhong Duan <zhenzhong.duan@intel.com>]

This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu.

This way qemu never needs to have privilege to open a VFIO or iommu
cdev node.

There is no easy way to check if a device is mdev with FD passing,
so fail the x-balloon-allowed check unconditionally in this case.

There is also no easy way to get BDF as name with FD passing, so
we fake a name by VFIO_FD[fd].

Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
 hw/vfio/iommufd.c | 12 +++++++----
 hw/vfio/pci.c     | 53 +++++++++++++++++++++++++++++++++++++++++------
 2 files changed, 55 insertions(+), 10 deletions(-)

diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index e532eed2ac..87f1960eed 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -343,11 +343,15 @@ static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
     uint32_t ioas_id;
     Error *err = NULL;
 
-    devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
-    if (devfd < 0) {
-        return devfd;
+    if (vbasedev->fd < 0) {
+        devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
+        if (devfd < 0) {
+            return devfd;
+        }
+        vbasedev->fd = devfd;
+    } else {
+        devfd = vbasedev->fd;
     }
-    vbasedev->fd = devfd;
 
     space = vfio_get_address_space(as);
 
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 9aa6c21200..6429172f30 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -44,6 +44,7 @@
 #include "migration/qemu-file.h"
 #include "linux/iommufd.h"
 #include "sysemu/iommufd.h"
+#include "monitor/monitor.h"
 
 #define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
 
@@ -3185,12 +3186,18 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
     }
 #endif
 
-    if (!vbasedev->sysfsdev) {
+    if (vbasedev->fd < 0 && !vbasedev->sysfsdev) {
         if (!(~vdev->host.domain || ~vdev->host.bus ||
               ~vdev->host.slot || ~vdev->host.function)) {
             error_setg(errp, "No provided host device");
+#ifdef CONFIG_IOMMUFD
+            error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F, "
+                              "-device vfio-pci,sysfsdev=PATH_TO_DEVICE "
+                              "or -device vfio-pci,fd=DEVICE_FD\n");
+#else
             error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F "
                               "or -device vfio-pci,sysfsdev=PATH_TO_DEVICE\n");
+#endif
             return;
         }
         vbasedev->sysfsdev =
@@ -3199,13 +3206,28 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
                             vdev->host.slot, vdev->host.function);
     }
 
-    if (stat(vbasedev->sysfsdev, &st) < 0) {
-        error_setg_errno(errp, errno, "no such host device");
-        error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
-        return;
+    if (vbasedev->fd < 0) {
+        if (stat(vbasedev->sysfsdev, &st) < 0) {
+            error_setg_errno(errp, errno, "no such host device");
+            error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+            return;
+        }
+        vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+    }
+#ifdef CONFIG_IOMMUFD
+    else {
+        if (!vbasedev->iommufd) {
+            error_setg(errp, "Use FD passing only with iommufd backend");
+            return;
+        }
+        /*
+         * Give a name with fd so any function printing out vbasedev->name
+         * will not break.
+         */
+        vbasedev->name = g_strdup_printf("VFIO_FD%d", vbasedev->fd);
     }
+#endif
 
-    vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
     vbasedev->ops = &vfio_pci_ops;
     vbasedev->type = VFIO_DEVICE_TYPE_PCI;
     vbasedev->dev = DEVICE(vdev);
@@ -3574,6 +3596,7 @@ static void vfio_instance_init(Object *obj)
     vdev->host.bus = ~0U;
     vdev->host.slot = ~0U;
     vdev->host.function = ~0U;
+    vdev->vbasedev.fd = -1;
 
     vdev->nv_gpudirect_clique = 0xFF;
 
@@ -3634,6 +3657,21 @@ static Property vfio_pci_dev_properties[] = {
     DEFINE_PROP_END_OF_LIST(),
 };
 
+#ifdef CONFIG_IOMMUFD
+static void vfio_pci_set_fd(Object *obj, const char *str, Error **errp)
+{
+    VFIOPCIDevice *vdev = VFIO_PCI(obj);
+    int fd = -1;
+
+    fd = monitor_fd_param(monitor_cur(), str, errp);
+    if (fd == -1) {
+        error_prepend(errp, "Could not parse remote object fd %s:", str);
+        return;
+    }
+    vdev->vbasedev.fd = fd;
+}
+#endif
+
 static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
@@ -3641,6 +3679,9 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
 
     dc->reset = vfio_pci_reset;
     device_class_set_props(dc, vfio_pci_dev_properties);
+#ifdef CONFIG_IOMMUFD
+    object_class_property_add_str(klass, "fd", NULL, vfio_pci_set_fd);
+#endif
     dc->desc = "VFIO-based PCI device assignment";
     set_bit(DEVICE_CATEGORY_MISC, dc->categories);
     pdc->realize = vfio_realize;
-- 
2.34.1