SUNRPC: Transport fault injection

It has been exceptionally useful to exercise the logic that handles
local immediate errors and RDMA connection loss.  To enable
developers to test this regularly and repeatably, add logic to
simulate connection loss every so often.

Fault injection is disabled by default. It is enabled with

  $ sudo echo xxx > /sys/kernel/debug/sunrpc/inject_fault/disconnect

where "xxx" is a large positive number of transport method calls
before a disconnect. A value of several thousand is usually a good
number that allows reasonable forward progress while still causing a
lot of connection drops.

These hooks are disabled when SUNRPC_DEBUG is turned off.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>

net/sunrpc/debugfs.c

@@ -10,9 +10,12 @@
 #include "netns.h"
 
 static struct dentry *topdir;
+static struct dentry *rpc_fault_dir;
 static struct dentry *rpc_clnt_dir;
 static struct dentry *rpc_xprt_dir;
 
+unsigned int rpc_inject_disconnect;
+
 struct rpc_clnt_iter {
 	struct rpc_clnt	*clnt;
 	loff_t		pos;
@@ -257,6 +260,8 @@ rpc_xprt_debugfs_register(struct rpc_xprt *xprt)
 		debugfs_remove_recursive(xprt->debugfs);
 		xprt->debugfs = NULL;
 	}
+
+	atomic_set(&xprt->inject_disconnect, rpc_inject_disconnect);
 }
 
 void
@@ -266,11 +271,78 @@ rpc_xprt_debugfs_unregister(struct rpc_xprt *xprt)
 	xprt->debugfs = NULL;
 }
 
+static int
+fault_open(struct inode *inode, struct file *filp)
+{
+	filp->private_data = kmalloc(128, GFP_KERNEL);
+	if (!filp->private_data)
+		return -ENOMEM;
+	return 0;
+}
+
+static int
+fault_release(struct inode *inode, struct file *filp)
+{
+	kfree(filp->private_data);
+	return 0;
+}
+
+static ssize_t
+fault_disconnect_read(struct file *filp, char __user *user_buf,
+		      size_t len, loff_t *offset)
+{
+	char *buffer = (char *)filp->private_data;
+	size_t size;
+
+	size = sprintf(buffer, "%u\n", rpc_inject_disconnect);
+	return simple_read_from_buffer(user_buf, len, offset, buffer, size);
+}
+
+static ssize_t
+fault_disconnect_write(struct file *filp, const char __user *user_buf,
+		       size_t len, loff_t *offset)
+{
+	char buffer[16];
+
+	len = min(len, sizeof(buffer) - 1);
+	if (copy_from_user(buffer, user_buf, len))
+		return -EFAULT;
+	buffer[len] = '\0';
+	if (kstrtouint(buffer, 10, &rpc_inject_disconnect))
+		return -EINVAL;
+	return len;
+}
+
+static const struct file_operations fault_disconnect_fops = {
+	.owner		= THIS_MODULE,
+	.open		= fault_open,
+	.read		= fault_disconnect_read,
+	.write		= fault_disconnect_write,
+	.release	= fault_release,
+};
+
+static struct dentry *
+inject_fault_dir(struct dentry *topdir)
+{
+	struct dentry *faultdir;
+
+	faultdir = debugfs_create_dir("inject_fault", topdir);
+	if (!faultdir)
+		return NULL;
+
+	if (!debugfs_create_file("disconnect", S_IFREG | S_IRUSR, faultdir,
+				 NULL, &fault_disconnect_fops))
+		return NULL;
+
+	return faultdir;
+}
+
 void __exit
 sunrpc_debugfs_exit(void)
 {
 	debugfs_remove_recursive(topdir);
 	topdir = NULL;
+	rpc_fault_dir = NULL;
 	rpc_clnt_dir = NULL;
 	rpc_xprt_dir = NULL;
 }
@@ -282,6 +354,10 @@ sunrpc_debugfs_init(void)
 	if (!topdir)
 		return;
 
+	rpc_fault_dir = inject_fault_dir(topdir);
+	if (!rpc_fault_dir)
+		goto out_remove;
+
 	rpc_clnt_dir = debugfs_create_dir("rpc_clnt", topdir);
 	if (!rpc_clnt_dir)
 		goto out_remove;
@@ -294,5 +370,6 @@ sunrpc_debugfs_init(void)
 out_remove:
 	debugfs_remove_recursive(topdir);
 	topdir = NULL;
+	rpc_fault_dir = NULL;
 	rpc_clnt_dir = NULL;
 }