[PATCH v2 5/5] nfsd: add the infrastructure to handle the cld upcall

[Jeff Layton <jlayton@redhat.com>]

...and add a mechanism for switching between the "legacy" tracker and
the new one. The decision is made by looking to see whether the
v4recoverydir exists. If it does, then the legacy client tracker is
used.

If it's not, then the kernel will create a "cld" pipe in rpc_pipefs.
That pipe is used to talk to a daemon for handling the upcall.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
---
 fs/nfsd/nfs4recover.c |  318 ++++++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 317 insertions(+), 1 deletions(-)

diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 62fd534..ae3d936 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -1,5 +1,6 @@
 /*
 *  Copyright (c) 2004 The Regents of the University of Michigan.
+*  Copyright (c) 2011 Jeff Layton <jlayton@redhat.com>
 *  All rights reserved.
 *
 *  Andy Adamson <andros@citi.umich.edu>
@@ -36,6 +37,10 @@
 #include <linux/namei.h>
 #include <linux/crypto.h>
 #include <linux/sched.h>
+#include <linux/fs.h>
+#include <linux/sunrpc/rpc_pipe_fs.h>
+#include <linux/sunrpc/clnt.h>
+#include <linux/nfsd/cld.h>
 
 #include "nfsd.h"
 #include "state.h"
@@ -467,12 +472,323 @@ static const struct nfsd4_client_tracking_ops nfsd4_legacy_tracking_ops = {
 	.grace_done	= nfsd4_recdir_purge_old,
 };
 
+/* Globals */
+#define NFSD_PIPE_DIR		"/nfsd"
+
+static struct dentry *cld_pipe;
+
+/* list of cld_msg's that are currently in use */
+static DEFINE_SPINLOCK(cld_lock);
+static LIST_HEAD(cld_list);
+static unsigned int cld_xid;
+
+struct cld_upcall {
+	struct list_head	cu_list;
+	struct task_struct *	cu_task;
+	struct cld_msg	cu_msg;
+};
+
+static int
+nfsd4_cld_upcall(struct cld_msg *cmsg)
+{
+	int ret;
+	struct rpc_pipe_msg msg;
+
+	memset(&msg, 0, sizeof(msg));
+	msg.data = cmsg;
+	msg.len = sizeof(*cmsg);
+
+	ret = rpc_queue_upcall(cld_pipe->d_inode, &msg);
+	if (ret < 0)
+		goto out;
+
+	set_current_state(TASK_UNINTERRUPTIBLE);
+	schedule();
+	__set_current_state(TASK_RUNNING);
+
+out:
+	return ret;
+}
+
+ssize_t cld_pipe_downcall(struct file *filp, const char __user *src,
+			     size_t mlen)
+{
+	struct cld_upcall *tmp, *cup;
+	struct cld_msg *cmsg = (struct cld_msg *)src;
+	uint32_t xid;
+
+	if (mlen != sizeof(*cmsg)) {
+		dprintk("%s: got %lu bytes, expected %lu\n", __func__, mlen,
+			sizeof(*cmsg));
+		return -EINVAL;
+	}
+
+	/* copy just the xid so we can try to find that */
+	if (copy_from_user(&xid, &cmsg->cm_xid, sizeof(xid)) != 0) {
+		dprintk("%s: error when copying xid from userspace", __func__);
+		return -EFAULT;
+	}
+
+	/* walk the list and find corresponding xid */
+	cup = NULL;
+	spin_lock(&cld_lock);
+	list_for_each_entry(tmp, &cld_list, cu_list) {
+		if (get_unaligned(&tmp->cu_msg.cm_xid) == xid) {
+			cup = tmp;
+			list_del_init(&cup->cu_list);
+			break;
+		}
+	}
+	spin_unlock(&cld_lock);
+
+	/* couldn't find upcall? */
+	if (!cup) {
+		dprintk("%s: couldn't find upcall -- xid=%u\n", __func__,
+			cup->cu_msg.cm_xid);
+		return -EINVAL;
+	}
+
+	if (copy_from_user(&cup->cu_msg, src, mlen) != 0)
+		return -EFAULT;
+
+	wake_up_process(cup->cu_task);
+	return mlen;
+}
+
+void
+cld_pipe_destroy_msg(struct rpc_pipe_msg *msg)
+{
+	struct cld_msg *cmsg = msg->data;
+	struct cld_upcall *cup = container_of(cmsg, struct cld_upcall,
+						 cu_msg);
+
+	if (msg->errno >= 0)
+		return;
+	wake_up_process(cup->cu_task);
+}
+
+static const struct rpc_pipe_ops cld_upcall_ops = {
+	.upcall		= rpc_pipe_generic_upcall,
+	.downcall	= cld_pipe_downcall,
+	.destroy_msg	= cld_pipe_destroy_msg,
+};
+
+int
+nfsd4_init_cld_pipe(void)
+{
+	int ret;
+	struct path path;
+	struct vfsmount *mnt;
+
+	if (cld_pipe)
+		return 0;
+
+	mnt = rpc_get_mount();
+	if (IS_ERR(mnt))
+		return PTR_ERR(mnt);
+
+	ret = vfs_path_lookup(mnt->mnt_root, mnt, NFSD_PIPE_DIR, 0, &path);
+	if (ret)
+		goto err;
+
+	cld_pipe = rpc_mkpipe(path.dentry, "cld", NULL,
+				 &cld_upcall_ops, RPC_PIPE_WAIT_FOR_OPEN);
+	path_put(&path);
+	if (!IS_ERR(cld_pipe))
+		return 0;
+
+	ret = PTR_ERR(cld_pipe);
+err:
+	rpc_put_mount();
+	return ret;
+}
+
+void
+nfsd4_remove_cld_pipe(void)
+{
+	int ret;
+
+	ret = rpc_unlink(cld_pipe);
+	if (ret)
+		printk(KERN_ERR "NFSD: error removing cld pipe: %d\n", ret);
+	cld_pipe = NULL;
+	rpc_put_mount();
+}
+
+static struct cld_upcall *
+alloc_cld_upcall(void)
+{
+	struct cld_upcall *new, *tmp;
+
+	new = kzalloc(sizeof(*new), GFP_KERNEL);
+	if (!new)
+		return new;
+
+	/* FIXME: hard cap on number in flight? */
+restart_search:
+	spin_lock(&cld_lock);
+	list_for_each_entry(tmp, &cld_list, cu_list) {
+		if (tmp->cu_msg.cm_xid == cld_xid) {
+			cld_xid++;
+			spin_unlock(&cld_lock);
+			goto restart_search;
+		}
+	}
+	new->cu_task = current;
+	new->cu_msg.cm_vers = CLD_MAX_UPCALL_VERSION;
+	put_unaligned(cld_xid++, &new->cu_msg.cm_xid);
+	list_add(&new->cu_list, &cld_list);
+	spin_unlock(&cld_lock);
+
+	dprintk("%s: allocated xid %u\n", __func__, new->cu_msg.cm_xid);
+
+	return new;
+}
+
+static void
+free_cld_upcall(struct cld_upcall *victim)
+{
+	spin_lock(&cld_lock);
+	list_del(&victim->cu_list);
+	spin_unlock(&cld_lock);
+	kfree(victim);
+}
+
+/* Ask daemon to create a new record */
+static int
+nfsd4_cld_create(struct nfs4_client *clp)
+{
+	int ret;
+	struct cld_upcall *cup;
+
+	/* Don't upcall if it's already stored */
+	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
+		return -EEXIST;
+
+	cup = alloc_cld_upcall();
+	if (!cup)
+		return -ENOMEM;
+
+	cup->cu_msg.cm_cmd = Cld_Create;
+	cup->cu_msg.cm_u.cm_name.cn_len = clp->cl_name.len;
+	memcpy(cup->cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
+			clp->cl_name.len);
+
+	ret = nfsd4_cld_upcall(&cup->cu_msg);
+	if (!ret) {
+		ret = cup->cu_msg.cm_status;
+		set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
+	}
+
+	free_cld_upcall(cup);
+	return ret;
+}
+
+/* Ask daemon to remove a record */
+static int
+nfsd4_cld_remove(struct nfs4_client *clp)
+{
+	int ret;
+	struct cld_upcall *cup;
+
+	/* Don't upcall if it's already stored */
+	if (!test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
+		return -ENOENT;
+
+	cup = alloc_cld_upcall();
+	if (!cup)
+		return -ENOMEM;
+
+	cup->cu_msg.cm_cmd = Cld_Expire;
+	cup->cu_msg.cm_u.cm_name.cn_len = clp->cl_name.len;
+	memcpy(cup->cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
+			clp->cl_name.len);
+
+	ret = nfsd4_cld_upcall(&cup->cu_msg);
+	if (!ret) {
+		ret = cup->cu_msg.cm_status;
+		clear_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
+	}
+
+	free_cld_upcall(cup);
+	return ret;
+}
+
+/* Check for presence of a record, and update its timestamp */
+static int
+nfsd4_cld_check(struct nfs4_client *clp)
+{
+	int ret;
+	struct cld_upcall *cup;
+
+	/* Don't upcall if one was already stored during this grace pd */
+	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
+		return 0;
+
+	cup = alloc_cld_upcall();
+	if (!cup)
+		return -ENOMEM;
+
+	cup->cu_msg.cm_cmd = Cld_Allow;
+	cup->cu_msg.cm_u.cm_name.cn_len = clp->cl_name.len;
+	memcpy(cup->cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
+			clp->cl_name.len);
+
+	ret = nfsd4_cld_upcall(&cup->cu_msg);
+	if (!ret) {
+		ret = cup->cu_msg.cm_status;
+		set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
+	}
+
+	free_cld_upcall(cup);
+	return ret;
+}
+
+static int
+nfsd4_cld_grace_done(time_t boot_time)
+{
+	int ret;
+	struct cld_upcall *cup;
+
+	cup = alloc_cld_upcall();
+	if (!cup)
+		return -ENOMEM;
+
+	cup->cu_msg.cm_cmd = Cld_GraceDone;
+	cup->cu_msg.cm_u.cm_gracetime = (int64_t)boot_time;
+	ret = nfsd4_cld_upcall(&cup->cu_msg);
+	if (!ret)
+		ret = cup->cu_msg.cm_status;
+
+	free_cld_upcall(cup);
+	return ret;
+}
+
+static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops = {
+	.init		= nfsd4_init_cld_pipe,
+	.exit		= nfsd4_remove_cld_pipe,
+	.create		= nfsd4_cld_create,
+	.remove		= nfsd4_cld_remove,
+	.check		= nfsd4_cld_check,
+	.grace_done	= nfsd4_cld_grace_done,
+};
+
 int
 nfsd4_client_tracking_init(void)
 {
 	int status;
+	struct path path;
 
-	client_tracking_ops = &nfsd4_legacy_tracking_ops;
+	if (!client_tracking_ops) {
+		client_tracking_ops = &nfsd4_cld_tracking_ops;
+		status = kern_path(nfs4_recoverydir(), LOOKUP_FOLLOW, &path);
+		if (!status) {
+			if (S_ISDIR(path.dentry->d_inode->i_mode))
+				client_tracking_ops =
+						&nfsd4_legacy_tracking_ops;
+			path_put(&path);
+		}
+	}
 	status = client_tracking_ops->init();
 	if (status) {
 		printk(KERN_WARNING "NFSD: Unable to initialize client "
-- 
1.7.1