sunrpc: add debugfs file for displaying client rpc_task queue

It's possible to get a dump of the RPC task queue by writing a value to
/proc/sys/sunrpc/rpc_debug. If you write any value to that file, you get
a dump of the RPC client task list into the log buffer. This is a rather
inconvenient interface however, and makes it hard to get immediate info
about the task queue.

Add a new directory hierarchy under debugfs:

    sunrpc/
        rpc_clnt/
            <clientid>/

Within each clientid directory we create a new "tasks" file that will
dump info similar to what shows up in the log buffer, but with a few
small differences -- we avoid printing raw kernel addresses in favor of
symbolic names and the XID is also displayed.

Signed-off-by: Jeff Layton <jlayton@primarydata.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>

5 files changed, 210 insertions, 1 deletions

diff --git a/net/sunrpc/Kconfig b/net/sunrpc/Kconfig
index 0754d0f466d2..fb78117b896c 100644
--- a/net/sunrpc/Kconfig
+++ b/net/sunrpc/Kconfig
@@ -35,6 +35,7 @@ config RPCSEC_GSS_KRB5
 config SUNRPC_DEBUG
         bool "RPC: Enable dprintk debugging"
         depends on SUNRPC && SYSCTL
+        select DEBUG_FS
         help
           This option enables a sysctl-based debugging interface
           that is be used by the 'rpcdebug' utility to turn on or off
diff --git a/net/sunrpc/Makefile b/net/sunrpc/Makefile
index e5a7a1cac8f3..15e6f6c23c5d 100644
--- a/net/sunrpc/Makefile
+++ b/net/sunrpc/Makefile
@@ -14,6 +14,7 @@ sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \
             addr.o rpcb_clnt.o timer.o xdr.o \
             sunrpc_syms.o cache.o rpc_pipe.o \
             svc_xprt.o
+sunrpc-$(CONFIG_SUNRPC_DEBUG) += debugfs.o
 sunrpc-$(CONFIG_SUNRPC_BACKCHANNEL) += backchannel_rqst.o bc_svc.o
 sunrpc-$(CONFIG_PROC_FS) += stats.o
 sunrpc-$(CONFIG_SYSCTL) += sysctl.o
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 36c64ef460cf..05da12a33945 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -305,6 +305,10 @@ static int rpc_client_register(struct rpc_clnt *clnt,
         struct super_block *pipefs_sb;
         int err;
 
+        err = rpc_clnt_debugfs_register(clnt);
+        if (err)
+                return err;
+
         pipefs_sb = rpc_get_sb_net(net);
         if (pipefs_sb) {
                 err = rpc_setup_pipedir(pipefs_sb, clnt);
@@ -331,6 +335,7 @@ err_auth:
 out:
         if (pipefs_sb)
                 rpc_put_sb_net(net);
+        rpc_clnt_debugfs_unregister(clnt);
         return err;
 }
 
@@ -670,6 +675,7 @@ int rpc_switch_client_transport(struct rpc_clnt *clnt,
 
         rpc_unregister_client(clnt);
         __rpc_clnt_remove_pipedir(clnt);
+        rpc_clnt_debugfs_unregister(clnt);
 
         /*
          * A new transport was created.  "clnt" therefore
@@ -771,6 +777,7 @@ rpc_free_client(struct rpc_clnt *clnt)
                         rcu_dereference(clnt->cl_xprt)->servername);
         if (clnt->cl_parent != clnt)
                 parent = clnt->cl_parent;
+        rpc_clnt_debugfs_unregister(clnt);
         rpc_clnt_remove_pipedir(clnt);
         rpc_unregister_client(clnt);
         rpc_free_iostats(clnt->cl_metrics);
@@ -1397,7 +1404,8 @@ rpc_restart_call(struct rpc_task *task)
 EXPORT_SYMBOL_GPL(rpc_restart_call);
 
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
-static const char *rpc_proc_name(const struct rpc_task *task)
+const char
+*rpc_proc_name(const struct rpc_task *task)
 {
         const struct rpc_procinfo *proc = task->tk_msg.rpc_proc;
 
diff --git a/net/sunrpc/debugfs.c b/net/sunrpc/debugfs.c
new file mode 100644
index 000000000000..3d7745683ca3
--- /dev/null
+++ b/net/sunrpc/debugfs.c
@@ -0,0 +1,191 @@
+/**
+ * debugfs interface for sunrpc
+ *
+ * (c) 2014 Jeff Layton <jlayton@primarydata.com>
+ */
+
+#include <linux/debugfs.h>
+#include <linux/sunrpc/sched.h>
+#include <linux/sunrpc/clnt.h>
+#include "netns.h"
+
+static struct dentry *topdir;
+static struct dentry *rpc_clnt_dir;
+
+struct rpc_clnt_iter {
+        struct rpc_clnt *clnt;
+        loff_t          pos;
+};
+
+static int
+tasks_show(struct seq_file *f, void *v)
+{
+        u32 xid = 0;
+        struct rpc_task *task = v;
+        struct rpc_clnt *clnt = task->tk_client;
+        const char *rpc_waitq = "none";
+
+        if (RPC_IS_QUEUED(task))
+                rpc_waitq = rpc_qname(task->tk_waitqueue);
+
+        if (task->tk_rqstp)
+                xid = be32_to_cpu(task->tk_rqstp->rq_xid);
+
+        seq_printf(f, "%5u %04x %6d 0x%x 0x%x %8ld %ps %sv%u %s a:%ps q:%s\n",
+                task->tk_pid, task->tk_flags, task->tk_status,
+                clnt->cl_clid, xid, task->tk_timeout, task->tk_ops,
+                clnt->cl_program->name, clnt->cl_vers, rpc_proc_name(task),
+                task->tk_action, rpc_waitq);
+        return 0;
+}
+
+static void *
+tasks_start(struct seq_file *f, loff_t *ppos)
+        __acquires(&clnt->cl_lock)
+{
+        struct rpc_clnt_iter *iter = f->private;
+        loff_t pos = *ppos;
+        struct rpc_clnt *clnt = iter->clnt;
+        struct rpc_task *task;
+
+        iter->pos = pos + 1;
+        spin_lock(&clnt->cl_lock);
+        list_for_each_entry(task, &clnt->cl_tasks, tk_task)
+                if (pos-- == 0)
+                        return task;
+        return NULL;
+}
+
+static void *
+tasks_next(struct seq_file *f, void *v, loff_t *pos)
+{
+        struct rpc_clnt_iter *iter = f->private;
+        struct rpc_clnt *clnt = iter->clnt;
+        struct rpc_task *task = v;
+        struct list_head *next = task->tk_task.next;
+
+        ++iter->pos;
+        ++*pos;
+
+        /* If there's another task on list, return it */
+        if (next == &clnt->cl_tasks)
+                return NULL;
+        return list_entry(next, struct rpc_task, tk_task);
+}
+
+static void
+tasks_stop(struct seq_file *f, void *v)
+        __releases(&clnt->cl_lock)
+{
+        struct rpc_clnt_iter *iter = f->private;
+        struct rpc_clnt *clnt = iter->clnt;
+
+        spin_unlock(&clnt->cl_lock);
+}
+
+static const struct seq_operations tasks_seq_operations = {
+        .start  = tasks_start,
+        .next   = tasks_next,
+        .stop   = tasks_stop,
+        .show   = tasks_show,
+};
+
+static int tasks_open(struct inode *inode, struct file *filp)
+{
+        int ret = seq_open_private(filp, &tasks_seq_operations,
+                                        sizeof(struct rpc_clnt_iter));
+
+        if (!ret) {
+                struct seq_file *seq = filp->private_data;
+                struct rpc_clnt_iter *iter = seq->private;
+
+                iter->clnt = inode->i_private;
+
+                if (!atomic_inc_not_zero(&iter->clnt->cl_count)) {
+                        seq_release_private(inode, filp);
+                        ret = -EINVAL;
+                }
+        }
+
+        return ret;
+}
+
+static int
+tasks_release(struct inode *inode, struct file *filp)
+{
+        struct seq_file *seq = filp->private_data;
+        struct rpc_clnt_iter *iter = seq->private;
+
+        rpc_release_client(iter->clnt);
+        return seq_release_private(inode, filp);
+}
+
+static const struct file_operations tasks_fops = {
+        .owner          = THIS_MODULE,
+        .open           = tasks_open,
+        .read           = seq_read,
+        .llseek         = seq_lseek,
+        .release        = tasks_release,
+};
+
+int
+rpc_clnt_debugfs_register(struct rpc_clnt *clnt)
+{
+        int len;
+        char name[9]; /* 8 for hex digits + NULL terminator */
+
+        /* Already registered? */
+        if (clnt->cl_debugfs)
+                return 0;
+
+        len = snprintf(name, sizeof(name), "%x", clnt->cl_clid);
+        if (len >= sizeof(name))
+                return -EINVAL;
+
+        /* make the per-client dir */
+        clnt->cl_debugfs = debugfs_create_dir(name, rpc_clnt_dir);
+        if (!clnt->cl_debugfs)
+                return -ENOMEM;
+
+        /* make tasks file */
+        if (!debugfs_create_file("tasks", S_IFREG | S_IRUSR, clnt->cl_debugfs,
+                                 clnt, &tasks_fops)) {
+                debugfs_remove_recursive(clnt->cl_debugfs);
+                clnt->cl_debugfs = NULL;
+                return -ENOMEM;
+        }
+
+        return 0;
+}
+
+void
+rpc_clnt_debugfs_unregister(struct rpc_clnt *clnt)
+{
+        debugfs_remove_recursive(clnt->cl_debugfs);
+        clnt->cl_debugfs = NULL;
+}
+
+void __exit
+sunrpc_debugfs_exit(void)
+{
+        debugfs_remove_recursive(topdir);
+}
+
+int __init
+sunrpc_debugfs_init(void)
+{
+        topdir = debugfs_create_dir("sunrpc", NULL);
+        if (!topdir)
+                goto out;
+
+        rpc_clnt_dir = debugfs_create_dir("rpc_clnt", topdir);
+        if (!rpc_clnt_dir)
+                goto out_remove;
+
+        return 0;
+out_remove:
+        debugfs_remove_recursive(topdir);
+        topdir = NULL;
+out:
+        return -ENOMEM;
+}
diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c
index f632e476ab6c..e37fbed87956 100644
--- a/net/sunrpc/sunrpc_syms.c
+++ b/net/sunrpc/sunrpc_syms.c
@@ -97,6 +97,11 @@ init_sunrpc(void)
         err = register_rpc_pipefs();
         if (err)
                 goto out4;
+
+        err = sunrpc_debugfs_init();
+        if (err)
+                goto out5;
+
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
         rpc_register_sysctl();
 #endif
@@ -104,6 +109,8 @@ init_sunrpc(void)
         init_socket_xprt();     /* clnt sock transport */
         return 0;
 
+out5:
+        unregister_rpc_pipefs();
 out4:
         unregister_pernet_subsys(&sunrpc_net_ops);
 out3:
@@ -120,6 +127,7 @@ cleanup_sunrpc(void)
         rpcauth_remove_module();
         cleanup_socket_xprt();
         svc_cleanup_xprt_sock();
+        sunrpc_debugfs_exit();
         unregister_rpc_pipefs();
         rpc_destroy_mempool();
         unregister_pernet_subsys(&sunrpc_net_ops);