vhost-scsi: Reduce vhost_scsi_mutex use

We on longer need to hold the vhost_scsi_mutex the entire time we
set/clear the endpoint. The tv_tpg_mutex handles tpg accesses not related
to the tpg list, the port link/unlink functions use the tv_tpg_mutex while
accessing the tpg->vhost_scsi pointer, vhost_scsi_do_plug will no longer
queue events after the virtqueue's backend has been cleared and flushed,
and we don't drop our refcount to the tpg until after we have stopped
cmds and wait for outstanding cmds to complete.

This moves the vhost_scsi_mutex use to it's documented use of being used
to access the tpg list. We then don't need to hold it while a flush is
being performed causing other device's vhost_scsi_set_endpoint
and vhost_scsi_make_tpg/vhost_scsi_drop_tpg calls to have to wait on a
flakey device.
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Message-Id: <20230321020624.13323-8-michael.christie@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

drivers/vhost/scsi.c

@@ -229,7 +229,10 @@ struct vhost_scsi_ctx {
 	struct iov_iter out_iter;
 };
 
-/* Global spinlock to protect vhost_scsi TPG list for vhost IOCTL access */
+/*
+ * Global mutex to protect vhost_scsi TPG list for vhost IOCTLs and LIO
+ * configfs management operations.
+ */
 static DEFINE_MUTEX(vhost_scsi_mutex);
 static LIST_HEAD(vhost_scsi_list);
 
@@ -1526,7 +1529,7 @@ static int vhost_scsi_setup_vq_cmds(struct vhost_virtqueue *vq, int max_cmds)
  * vhost_scsi_tpg with an active struct vhost_scsi_nexus
  *
  *  The lock nesting rule is:
- *    vhost_scsi_mutex -> vs->dev.mutex -> tpg->tv_tpg_mutex -> vq->mutex
+ *    vs->dev.mutex -> vhost_scsi_mutex -> tpg->tv_tpg_mutex -> vq->mutex
  */
 static int
 vhost_scsi_set_endpoint(struct vhost_scsi *vs,
@@ -1540,7 +1543,6 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 	int index, ret, i, len;
 	bool match = false;
 
-	mutex_lock(&vhost_scsi_mutex);
 	mutex_lock(&vs->dev.mutex);
 
 	/* Verify that ring has been setup correctly. */
@@ -1561,6 +1563,7 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 	if (vs->vs_tpg)
 		memcpy(vs_tpg, vs->vs_tpg, len);
 
+	mutex_lock(&vhost_scsi_mutex);
 	list_for_each_entry(tpg, &vhost_scsi_list, tv_tpg_list) {
 		mutex_lock(&tpg->tv_tpg_mutex);
 		if (!tpg->tpg_nexus) {
@@ -1576,6 +1579,7 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 		if (!strcmp(tv_tport->tport_name, t->vhost_wwpn)) {
 			if (vs->vs_tpg && vs->vs_tpg[tpg->tport_tpgt]) {
 				mutex_unlock(&tpg->tv_tpg_mutex);
+				mutex_unlock(&vhost_scsi_mutex);
 				ret = -EEXIST;
 				goto undepend;
 			}
@@ -1590,6 +1594,7 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 			if (ret) {
 				pr_warn("target_depend_item() failed: %d\n", ret);
 				mutex_unlock(&tpg->tv_tpg_mutex);
+				mutex_unlock(&vhost_scsi_mutex);
 				goto undepend;
 			}
 			tpg->tv_tpg_vhost_count++;
@@ -1599,6 +1604,7 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 		}
 		mutex_unlock(&tpg->tv_tpg_mutex);
 	}
+	mutex_unlock(&vhost_scsi_mutex);
 
 	if (match) {
 		memcpy(vs->vs_vhost_wwpn, t->vhost_wwpn,
@@ -1654,7 +1660,6 @@ vhost_scsi_set_endpoint(struct vhost_scsi *vs,
 	kfree(vs_tpg);
 out:
 	mutex_unlock(&vs->dev.mutex);
-	mutex_unlock(&vhost_scsi_mutex);
 	return ret;
 }
 
@@ -1670,7 +1675,6 @@ vhost_scsi_clear_endpoint(struct vhost_scsi *vs,
 	int index, ret, i;
 	u8 target;
 
-	mutex_lock(&vhost_scsi_mutex);
 	mutex_lock(&vs->dev.mutex);
 	/* Verify that ring has been setup correctly. */
 	for (index = 0; index < vs->dev.nvqs; ++index) {
@@ -1757,12 +1761,10 @@ vhost_scsi_clear_endpoint(struct vhost_scsi *vs,
 	vs->vs_tpg = NULL;
 	WARN_ON(vs->vs_events_nr);
 	mutex_unlock(&vs->dev.mutex);
-	mutex_unlock(&vhost_scsi_mutex);
 	return 0;
 
 err_dev:
 	mutex_unlock(&vs->dev.mutex);
-	mutex_unlock(&vhost_scsi_mutex);
 	return ret;
 }