vfio-user: run vfio-user context

Setup a handler to run vfio-user context. The context is driven by
messages to the file descriptor associated with it - get the fd for
the context and hook up the handler with it

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: e934b0090529d448b6a7972b21dfc3d7421ce494.1655151679.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

hw/remote/vfio-user-obj.c

@@ -27,6 +27,9 @@
  *
  * device - id of a device on the server, a required option. PCI devices
  *          alone are supported presently.
+ *
+ * notes - x-vfio-user-server could block IO and monitor during the
+ *         initialization phase.
  */
 
 #include "qemu/osdep.h"
@@ -40,11 +43,14 @@
 #include "hw/remote/machine.h"
 #include "qapi/error.h"
 #include "qapi/qapi-visit-sockets.h"
+#include "qapi/qapi-events-misc.h"
 #include "qemu/notify.h"
+#include "qemu/thread.h"
 #include "sysemu/sysemu.h"
 #include "libvfio-user.h"
 #include "hw/qdev-core.h"
 #include "hw/pci/pci.h"
+#include "qemu/timer.h"
 
 #define TYPE_VFU_OBJECT "x-vfio-user-server"
 OBJECT_DECLARE_TYPE(VfuObject, VfuObjectClass, VFU_OBJECT)
@@ -86,6 +92,8 @@ struct VfuObject {
     PCIDevice *pci_dev;
 
     Error *unplug_blocker;
+
+    int vfu_poll_fd;
 };
 
 static void vfu_object_init_ctx(VfuObject *o, Error **errp);
@@ -164,6 +172,78 @@ static void vfu_object_set_device(Object *obj, const char *str, Error **errp)
     vfu_object_init_ctx(o, errp);
 }
 
+static void vfu_object_ctx_run(void *opaque)
+{
+    VfuObject *o = opaque;
+    const char *vfu_id;
+    char *vfu_path, *pci_dev_path;
+    int ret = -1;
+
+    while (ret != 0) {
+        ret = vfu_run_ctx(o->vfu_ctx);
+        if (ret < 0) {
+            if (errno == EINTR) {
+                continue;
+            } else if (errno == ENOTCONN) {
+                vfu_id = object_get_canonical_path_component(OBJECT(o));
+                vfu_path = object_get_canonical_path(OBJECT(o));
+                g_assert(o->pci_dev);
+                pci_dev_path = object_get_canonical_path(OBJECT(o->pci_dev));
+                 /* o->device is a required property and is non-NULL here */
+                g_assert(o->device);
+                qapi_event_send_vfu_client_hangup(vfu_id, vfu_path,
+                                                  o->device, pci_dev_path);
+                qemu_set_fd_handler(o->vfu_poll_fd, NULL, NULL, NULL);
+                o->vfu_poll_fd = -1;
+                object_unparent(OBJECT(o));
+                g_free(vfu_path);
+                g_free(pci_dev_path);
+                break;
+            } else {
+                VFU_OBJECT_ERROR(o, "vfu: Failed to run device %s - %s",
+                                 o->device, strerror(errno));
+                break;
+            }
+        }
+    }
+}
+
+static void vfu_object_attach_ctx(void *opaque)
+{
+    VfuObject *o = opaque;
+    GPollFD pfds[1];
+    int ret;
+
+    qemu_set_fd_handler(o->vfu_poll_fd, NULL, NULL, NULL);
+
+    pfds[0].fd = o->vfu_poll_fd;
+    pfds[0].events = G_IO_IN | G_IO_HUP | G_IO_ERR;
+
+retry_attach:
+    ret = vfu_attach_ctx(o->vfu_ctx);
+    if (ret < 0 && (errno == EAGAIN || errno == EWOULDBLOCK)) {
+        /**
+         * vfu_object_attach_ctx can block QEMU's main loop
+         * during attach - the monitor and other IO
+         * could be unresponsive during this time.
+         */
+        (void)qemu_poll_ns(pfds, 1, 500 * (int64_t)SCALE_MS);
+        goto retry_attach;
+    } else if (ret < 0) {
+        VFU_OBJECT_ERROR(o, "vfu: Failed to attach device %s to context - %s",
+                         o->device, strerror(errno));
+        return;
+    }
+
+    o->vfu_poll_fd = vfu_get_poll_fd(o->vfu_ctx);
+    if (o->vfu_poll_fd < 0) {
+        VFU_OBJECT_ERROR(o, "vfu: Failed to get poll fd %s", o->device);
+        return;
+    }
+
+    qemu_set_fd_handler(o->vfu_poll_fd, vfu_object_ctx_run, NULL, o);
+}
+
 /*
  * TYPE_VFU_OBJECT depends on the availability of the 'socket' and 'device'
  * properties. It also depends on devices instantiated in QEMU. These
@@ -184,6 +264,20 @@ static void vfu_object_machine_done(Notifier *notifier, void *data)
     }
 }
 
+/**
+ * vfu_object_init_ctx: Create and initialize libvfio-user context. Add
+ *     an unplug blocker for the associated PCI device. Setup a FD handler
+ *     to process incoming messages in the context's socket.
+ *
+ *     The socket and device properties are mandatory, and this function
+ *     will not create the context without them - the setters for these
+ *     properties should call this function when the property is set. The
+ *     machine should also be ready when this function is invoked - it is
+ *     because QEMU objects are initialized before devices, and the
+ *     associated PCI device wouldn't be available at the object
+ *     initialization time. Until these conditions are satisfied, this
+ *     function would return early without performing any task.
+ */
 static void vfu_object_init_ctx(VfuObject *o, Error **errp)
 {
     ERRP_GUARD();
@@ -202,7 +296,8 @@ static void vfu_object_init_ctx(VfuObject *o, Error **errp)
         return;
     }
 
-    o->vfu_ctx = vfu_create_ctx(VFU_TRANS_SOCK, o->socket->u.q_unix.path, 0,
+    o->vfu_ctx = vfu_create_ctx(VFU_TRANS_SOCK, o->socket->u.q_unix.path,
+                                LIBVFIO_USER_FLAG_ATTACH_NB,
                                 o, VFU_DEV_TYPE_PCI);
     if (o->vfu_ctx == NULL) {
         error_setg(errp, "vfu: Failed to create context - %s", strerror(errno));
@@ -241,6 +336,21 @@ static void vfu_object_init_ctx(VfuObject *o, Error **errp)
                TYPE_VFU_OBJECT, o->device);
     qdev_add_unplug_blocker(DEVICE(o->pci_dev), o->unplug_blocker);
 
+    ret = vfu_realize_ctx(o->vfu_ctx);
+    if (ret < 0) {
+        error_setg(errp, "vfu: Failed to realize device %s- %s",
+                   o->device, strerror(errno));
+        goto fail;
+    }
+
+    o->vfu_poll_fd = vfu_get_poll_fd(o->vfu_ctx);
+    if (o->vfu_poll_fd < 0) {
+        error_setg(errp, "vfu: Failed to get poll fd %s", o->device);
+        goto fail;
+    }
+
+    qemu_set_fd_handler(o->vfu_poll_fd, vfu_object_attach_ctx, NULL, o);
+
     return;
 
 fail:
@@ -275,6 +385,7 @@ static void vfu_object_init(Object *obj)
         qemu_add_machine_init_done_notifier(&o->machine_done);
     }
 
+    o->vfu_poll_fd = -1;
 }
 
 static void vfu_object_finalize(Object *obj)
@@ -288,6 +399,11 @@ static void vfu_object_finalize(Object *obj)
 
     o->socket = NULL;
 
+    if (o->vfu_poll_fd != -1) {
+        qemu_set_fd_handler(o->vfu_poll_fd, NULL, NULL, NULL);
+        o->vfu_poll_fd = -1;
+    }
+
     if (o->vfu_ctx) {
         vfu_destroy_ctx(o->vfu_ctx);
         o->vfu_ctx = NULL;

qapi/misc.json

@@ -553,3 +553,34 @@
 ##
 { 'event': 'RTC_CHANGE',
   'data': { 'offset': 'int', 'qom-path': 'str' } }
+
+##
+# @VFU_CLIENT_HANGUP:
+#
+# Emitted when the client of a TYPE_VFIO_USER_SERVER closes the
+# communication channel
+#
+# @vfu-id: ID of the TYPE_VFIO_USER_SERVER object. It is the last component
+#          of @vfu-qom-path referenced below
+#
+# @vfu-qom-path: path to the TYPE_VFIO_USER_SERVER object in the QOM tree
+#
+# @dev-id: ID of attached PCI device
+#
+# @dev-qom-path: path to attached PCI device in the QOM tree
+#
+# Since: 7.1
+#
+# Example:
+#
+# <- { "event": "VFU_CLIENT_HANGUP",
+#      "data": { "vfu-id": "vfu1",
+#                "vfu-qom-path": "/objects/vfu1",
+#                "dev-id": "sas1",
+#                "dev-qom-path": "/machine/peripheral/sas1" },
+#      "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+#
+##
+{ 'event': 'VFU_CLIENT_HANGUP',
+  'data': { 'vfu-id': 'str', 'vfu-qom-path': 'str',
+            'dev-id': 'str', 'dev-qom-path': 'str' } }