scst_sysfs: Add support for cluster_name

When dlm_new_lockspace is called with a NULL cluster_name the
kernel will emit an error message "dlm cluster name '%s' is being
used without an application provided cluster name".

Therefore, provide a mechanism to set a cluster_name and use it
when calling dlm_new_lockspace.

scst/src/scst_dlm.c

@@ -1510,4 +1510,6 @@ const struct scst_cl_ops scst_dlm_cl_ops = {
 	.reserve		= scst_dlm_reserve,
 };
 
+char *scst_dlm_cluster_name;
+
 #endif

scst/src/scst_priv.h

@@ -176,6 +176,7 @@ extern wait_queue_head_t scst_dev_cmd_waitQ;
 
 extern const struct scst_cl_ops scst_no_dlm_cl_ops;
 extern const struct scst_cl_ops scst_dlm_cl_ops;
+extern char *scst_dlm_cluster_name;
 
 extern unsigned int scst_setup_id;
 
@@ -417,8 +418,8 @@ static inline int scst_dlm_new_lockspace(const char *name, int namelen,
 					 uint32_t flags,
 					 int lvblen)
 {
-	return dlm_new_lockspace(name, NULL, flags, lvblen, NULL, NULL, NULL,
-				 lockspace);
+	return dlm_new_lockspace(name, scst_dlm_cluster_name, flags, lvblen,
+				 NULL, NULL, NULL, lockspace);
 }
 
 int scst_alloc_space(struct scst_cmd *cmd);

scst/src/scst_sysfs.c

@@ -7594,6 +7594,70 @@ static struct kobj_attribute scst_last_sysfs_mgmt_res_attr =
 	__ATTR(last_sysfs_mgmt_res, S_IRUGO,
 		scst_last_sysfs_mgmt_res_show, NULL);
 
+static ssize_t scst_cluster_name_show(struct kobject *kobj,
+	struct kobj_attribute *attr, char *buf)
+{
+	int res = 0;
+
+	TRACE_ENTRY();
+
+	if (scst_dlm_cluster_name != NULL)
+		res = sprintf(buf, "%s\n%s", scst_dlm_cluster_name,
+			SCST_SYSFS_KEY_MARK "\n");
+
+	TRACE_EXIT_RES(res);
+	return res;
+}
+
+static ssize_t scst_cluster_name_store(struct kobject *kobj,
+	struct kobj_attribute *attr, const char *buf, size_t count)
+{
+	int res = 0;
+	int len;
+
+	TRACE_ENTRY();
+
+	if ((buf == NULL) || (count == 0)) {
+		goto out;
+	}
+
+	len = strnlen(buf, count);
+	if (buf[count-1] == '\n')
+		len--;
+
+	if (len == 0) {
+		kfree(scst_dlm_cluster_name);
+		scst_dlm_cluster_name = NULL;
+		goto out_done;
+	}
+
+	if (len >= DLM_LOCKSPACE_LEN) {
+		PRINT_ERROR("cluster_name string too long (len %d)", len);
+		res = -EINVAL;
+		goto out;
+	}
+
+	kfree(scst_dlm_cluster_name);
+	scst_dlm_cluster_name = kstrndup(buf, len, GFP_KERNEL);
+	if (!scst_dlm_cluster_name) {
+		PRINT_ERROR("Unable to alloc cluster_name string (len %d)",
+			len+1);
+		res = -ENOMEM;
+		goto out;
+	}
+
+out_done:
+	res = count;
+
+out:
+	TRACE_EXIT_RES(res);
+	return res;
+}
+
+static struct kobj_attribute scst_cluster_name_attr =
+	__ATTR(cluster_name, S_IRUGO | S_IWUSR, scst_cluster_name_show,
+	       scst_cluster_name_store);
+
 static struct attribute *scst_sysfs_root_def_attrs[] = {
 	&scst_measure_latency_attr.attr,
 	&scst_threads_attr.attr,
@@ -7609,6 +7673,7 @@ static struct attribute *scst_sysfs_root_def_attrs[] = {
 	&scst_trace_mcmds_attr.attr,
 	&scst_version_attr.attr,
 	&scst_last_sysfs_mgmt_res_attr.attr,
+	&scst_cluster_name_attr.attr,
 	NULL,
 };
 #if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 2, 0)
@@ -7932,6 +7997,7 @@ void scst_sysfs_cleanup(void)
 	TRACE_ENTRY();
 
 	PRINT_INFO("%s", "Exiting SCST sysfs hierarchy...");
+	kfree(scst_dlm_cluster_name);
 
 	scst_del_put_sgv_kobj();