[PATCH] v9fs: VFS superblock operations and glue
authorEric Van Hensbergen <ericvh@gmail.com>
Fri, 9 Sep 2005 20:04:20 +0000 (13:04 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Fri, 9 Sep 2005 20:57:56 +0000 (13:57 -0700)
This part of the patch contains VFS superblock and mapping code.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
fs/9p/v9fs.c [new file with mode: 0644]
fs/9p/v9fs.h [new file with mode: 0644]
fs/9p/v9fs_vfs.h [new file with mode: 0644]
fs/9p/vfs_super.c [new file with mode: 0644]

diff --git a/fs/9p/v9fs.c b/fs/9p/v9fs.c
new file mode 100644 (file)
index 0000000..14d663e
--- /dev/null
@@ -0,0 +1,448 @@
+/*
+ *  linux/fs/9p/v9fs.c
+ *
+ *  This file contains functions assisting in mapping VFS to 9P2000
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/fs.h>
+#include <linux/parser.h>
+#include <linux/idr.h>
+
+#include "debug.h"
+#include "v9fs.h"
+#include "9p.h"
+#include "v9fs_vfs.h"
+#include "transport.h"
+#include "mux.h"
+#include "conv.h"
+
+/* TODO: sysfs or debugfs interface */
+int v9fs_debug_level = 0;      /* feature-rific global debug level  */
+
+/*
+  * Option Parsing (code inspired by NFS code)
+  *
+  */
+
+enum {
+       /* Options that take integer arguments */
+       Opt_port, Opt_msize, Opt_uid, Opt_gid, Opt_afid, Opt_debug,
+       Opt_rfdno, Opt_wfdno,
+       /* String options */
+       Opt_name, Opt_remotename,
+       /* Options that take no arguments */
+       Opt_legacy, Opt_nodevmap, Opt_unix, Opt_tcp, Opt_fd,
+       /* Error token */
+       Opt_err
+};
+
+static match_table_t tokens = {
+       {Opt_port, "port=%u"},
+       {Opt_msize, "msize=%u"},
+       {Opt_uid, "uid=%u"},
+       {Opt_gid, "gid=%u"},
+       {Opt_afid, "afid=%u"},
+       {Opt_rfdno, "rfdno=%u"},
+       {Opt_wfdno, "wfdno=%u"},
+       {Opt_debug, "debug=%u"},
+       {Opt_name, "name=%s"},
+       {Opt_remotename, "aname=%s"},
+       {Opt_unix, "proto=unix"},
+       {Opt_tcp, "proto=tcp"},
+       {Opt_fd, "proto=fd"},
+       {Opt_tcp, "tcp"},
+       {Opt_unix, "unix"},
+       {Opt_fd, "fd"},
+       {Opt_legacy, "noextend"},
+       {Opt_nodevmap, "nodevmap"},
+       {Opt_err, NULL}
+};
+
+/*
+ *  Parse option string.
+ */
+
+/**
+ * v9fs_parse_options - parse mount options into session structure
+ * @options: options string passed from mount
+ * @v9ses: existing v9fs session information
+ *
+ */
+
+static void v9fs_parse_options(char *options, struct v9fs_session_info *v9ses)
+{
+       char *p;
+       substring_t args[MAX_OPT_ARGS];
+       int option;
+       int ret;
+
+       /* setup defaults */
+       v9ses->port = V9FS_PORT;
+       v9ses->maxdata = 9000;
+       v9ses->proto = PROTO_TCP;
+       v9ses->extended = 1;
+       v9ses->afid = ~0;
+       v9ses->debug = 0;
+       v9ses->rfdno = ~0;
+       v9ses->wfdno = ~0;
+
+       if (!options)
+               return;
+
+       while ((p = strsep(&options, ",")) != NULL) {
+               int token;
+               if (!*p)
+                       continue;
+               token = match_token(p, tokens, args);
+               if (token < Opt_name) {
+                       if ((ret = match_int(&args[0], &option)) < 0) {
+                               dprintk(DEBUG_ERROR,
+                                       "integer field, but no integer?\n");
+                               continue;
+                       }
+
+               }
+               switch (token) {
+               case Opt_port:
+                       v9ses->port = option;
+                       break;
+               case Opt_msize:
+                       v9ses->maxdata = option;
+                       break;
+               case Opt_uid:
+                       v9ses->uid = option;
+                       break;
+               case Opt_gid:
+                       v9ses->gid = option;
+                       break;
+               case Opt_afid:
+                       v9ses->afid = option;
+                       break;
+               case Opt_rfdno:
+                       v9ses->rfdno = option;
+                       break;
+               case Opt_wfdno:
+                       v9ses->wfdno = option;
+                       break;
+               case Opt_debug:
+                       v9ses->debug = option;
+                       break;
+               case Opt_tcp:
+                       v9ses->proto = PROTO_TCP;
+                       break;
+               case Opt_unix:
+                       v9ses->proto = PROTO_UNIX;
+                       break;
+               case Opt_fd:
+                       v9ses->proto = PROTO_FD;
+                       break;
+               case Opt_name:
+                       match_strcpy(v9ses->name, &args[0]);
+                       break;
+               case Opt_remotename:
+                       match_strcpy(v9ses->remotename, &args[0]);
+                       break;
+               case Opt_legacy:
+                       v9ses->extended = 0;
+                       break;
+               case Opt_nodevmap:
+                       v9ses->nodev = 1;
+                       break;
+               default:
+                       continue;
+               }
+       }
+}
+
+/**
+ * v9fs_inode2v9ses - safely extract v9fs session info from super block
+ * @inode: inode to extract information from
+ *
+ * Paranoid function to extract v9ses information from superblock,
+ * if anything is missing it will report an error.
+ *
+ */
+
+struct v9fs_session_info *v9fs_inode2v9ses(struct inode *inode)
+{
+       return (inode->i_sb->s_fs_info);
+}
+
+/**
+ * v9fs_get_idpool - allocate numeric id from pool
+ * @p - pool to allocate from
+ *
+ * XXX - This seems to be an awful generic function, should it be in idr.c with
+ *            the lock included in struct idr?
+ */
+
+int v9fs_get_idpool(struct v9fs_idpool *p)
+{
+       int i = 0;
+       int error;
+
+retry:
+       if (idr_pre_get(&p->pool, GFP_KERNEL) == 0)
+               return 0;
+
+       if (down_interruptible(&p->lock) == -EINTR) {
+               eprintk(KERN_WARNING, "Interrupted while locking\n");
+               return -1;
+       }
+
+       error = idr_get_new(&p->pool, NULL, &i);
+       up(&p->lock);
+
+       if (error == -EAGAIN)
+               goto retry;
+       else if (error)
+               return -1;
+
+       return i;
+}
+
+/**
+ * v9fs_put_idpool - release numeric id from pool
+ * @p - pool to allocate from
+ *
+ * XXX - This seems to be an awful generic function, should it be in idr.c with
+ *            the lock included in struct idr?
+ */
+
+void v9fs_put_idpool(int id, struct v9fs_idpool *p)
+{
+       if (down_interruptible(&p->lock) == -EINTR) {
+               eprintk(KERN_WARNING, "Interrupted while locking\n");
+               return;
+       }
+       idr_remove(&p->pool, id);
+       up(&p->lock);
+}
+
+/**
+ * v9fs_session_init - initialize session
+ * @v9ses: session information structure
+ * @dev_name: device being mounted
+ * @data: options
+ *
+ */
+
+int
+v9fs_session_init(struct v9fs_session_info *v9ses,
+                 const char *dev_name, char *data)
+{
+       struct v9fs_fcall *fcall = NULL;
+       struct v9fs_transport *trans_proto;
+       int n = 0;
+       int newfid = -1;
+       int retval = -EINVAL;
+
+       v9ses->name = __getname();
+       if (!v9ses->name)
+               return -ENOMEM;
+
+       v9ses->remotename = __getname();
+       if (!v9ses->remotename) {
+               putname(v9ses->name);
+               return -ENOMEM;
+       }
+
+       strcpy(v9ses->name, V9FS_DEFUSER);
+       strcpy(v9ses->remotename, V9FS_DEFANAME);
+
+       v9fs_parse_options(data, v9ses);
+
+       /* set global debug level */
+       v9fs_debug_level = v9ses->debug;
+
+       /* id pools that are session-dependent: FIDs and TIDs */
+       idr_init(&v9ses->fidpool.pool);
+       init_MUTEX(&v9ses->fidpool.lock);
+       idr_init(&v9ses->tidpool.pool);
+       init_MUTEX(&v9ses->tidpool.lock);
+
+
+       switch (v9ses->proto) {
+       case PROTO_TCP:
+               trans_proto = &v9fs_trans_tcp;
+               break;
+       case PROTO_UNIX:
+               trans_proto = &v9fs_trans_unix;
+               *v9ses->remotename = 0;
+               break;
+       case PROTO_FD:
+               trans_proto = &v9fs_trans_fd;
+               *v9ses->remotename = 0;
+               if((v9ses->wfdno == ~0) || (v9ses->rfdno == ~0)) {
+                       printk(KERN_ERR "v9fs: Insufficient options for proto=fd\n");
+                       retval = -ENOPROTOOPT;
+                       goto SessCleanUp;
+               }
+               break;
+       default:
+               printk(KERN_ERR "v9fs: Bad mount protocol %d\n", v9ses->proto);
+               retval = -ENOPROTOOPT;
+               goto SessCleanUp;
+       };
+
+       v9ses->transport = trans_proto;
+
+       if ((retval = v9ses->transport->init(v9ses, dev_name, data)) < 0) {
+               eprintk(KERN_ERR, "problem initializing transport\n");
+               goto SessCleanUp;
+       }
+
+       v9ses->inprogress = 0;
+       v9ses->shutdown = 0;
+       v9ses->session_hung = 0;
+
+       if ((retval = v9fs_mux_init(v9ses, dev_name)) < 0) {
+               dprintk(DEBUG_ERROR, "problem initializing mux\n");
+               goto SessCleanUp;
+       }
+
+       if (v9ses->afid == ~0) {
+               if (v9ses->extended)
+                       retval =
+                           v9fs_t_version(v9ses, v9ses->maxdata, "9P2000.u",
+                                          &fcall);
+               else
+                       retval = v9fs_t_version(v9ses, v9ses->maxdata, "9P2000",
+                                               &fcall);
+
+               if (retval < 0) {
+                       dprintk(DEBUG_ERROR, "v9fs_t_version failed\n");
+                       goto FreeFcall;
+               }
+
+               /* Really should check for 9P1 and report error */
+               if (!strcmp(fcall->params.rversion.version, "9P2000.u")) {
+                       dprintk(DEBUG_9P, "9P2000 UNIX extensions enabled\n");
+                       v9ses->extended = 1;
+               } else {
+                       dprintk(DEBUG_9P, "9P2000 legacy mode enabled\n");
+                       v9ses->extended = 0;
+               }
+
+               n = fcall->params.rversion.msize;
+               kfree(fcall);
+
+               if (n < v9ses->maxdata)
+                       v9ses->maxdata = n;
+       }
+
+       newfid = v9fs_get_idpool(&v9ses->fidpool);
+       if (newfid < 0) {
+               eprintk(KERN_WARNING, "couldn't allocate FID\n");
+               retval = -ENOMEM;
+               goto SessCleanUp;
+       }
+       /* it is a little bit ugly, but we have to prevent newfid */
+       /* being the same as afid, so if it is, get a new fid     */
+       if (v9ses->afid != ~0 && newfid == v9ses->afid) {
+               newfid = v9fs_get_idpool(&v9ses->fidpool);
+               if (newfid < 0) {
+                       eprintk(KERN_WARNING, "couldn't allocate FID\n");
+                       retval = -ENOMEM;
+                       goto SessCleanUp;
+               }
+       }
+
+       if ((retval =
+            v9fs_t_attach(v9ses, v9ses->name, v9ses->remotename, newfid,
+                          v9ses->afid, NULL))
+           < 0) {
+               dprintk(DEBUG_ERROR, "cannot attach\n");
+               goto SessCleanUp;
+       }
+
+       if (v9ses->afid != ~0) {
+               if (v9fs_t_clunk(v9ses, v9ses->afid, NULL))
+                       dprintk(DEBUG_ERROR, "clunk failed\n");
+       }
+
+       return newfid;
+
+      FreeFcall:
+       kfree(fcall);
+
+      SessCleanUp:
+       v9fs_session_close(v9ses);
+       return retval;
+}
+
+/**
+ * v9fs_session_close - shutdown a session
+ * @v9ses: session information structure
+ *
+ */
+
+void v9fs_session_close(struct v9fs_session_info *v9ses)
+{
+       if (v9ses->recvproc) {
+               send_sig(SIGKILL, v9ses->recvproc, 1);
+               wait_for_completion(&v9ses->proccmpl);
+       }
+
+       if (v9ses->transport)
+               v9ses->transport->close(v9ses->transport);
+
+       putname(v9ses->name);
+       putname(v9ses->remotename);
+}
+
+extern int v9fs_error_init(void);
+
+/**
+ * v9fs_init - Initialize module
+ *
+ */
+
+static int __init init_v9fs(void)
+{
+       v9fs_error_init();
+
+       printk(KERN_INFO "Installing v9fs 9P2000 file system support\n");
+
+       return register_filesystem(&v9fs_fs_type);
+}
+
+/**
+ * v9fs_init - shutdown module
+ *
+ */
+
+static void __exit exit_v9fs(void)
+{
+       unregister_filesystem(&v9fs_fs_type);
+}
+
+module_init(init_v9fs)
+module_exit(exit_v9fs)
+
+MODULE_AUTHOR("Eric Van Hensbergen <ericvh@gmail.com>");
+MODULE_AUTHOR("Ron Minnich <rminnich@lanl.gov>");
+MODULE_LICENSE("GPL");
diff --git a/fs/9p/v9fs.h b/fs/9p/v9fs.h
new file mode 100644 (file)
index 0000000..5220302
--- /dev/null
@@ -0,0 +1,105 @@
+/*
+ * V9FS definitions.
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+/*
+  * Idpool structure provides lock and id management
+  *
+  */
+
+struct v9fs_idpool {
+       struct semaphore lock;
+       struct idr pool;
+};
+
+/*
+  * Session structure provides information for an opened session
+  *
+  */
+
+struct v9fs_session_info {
+       /* options */
+       unsigned int maxdata;
+       unsigned char extended; /* set to 1 if we are using UNIX extensions */
+       unsigned char nodev;    /* set to 1 if no disable device mapping */
+       unsigned short port;    /* port to connect to */
+       unsigned short debug;   /* debug level */
+       unsigned short proto;   /* protocol to use */
+       unsigned int afid;      /* authentication fid */
+       unsigned int rfdno;     /* read file descriptor number */
+       unsigned int wfdno;     /* write file descriptor number */
+
+
+       char *name;             /* user name to mount as */
+       char *remotename;       /* name of remote hierarchy being mounted */
+       unsigned int uid;       /* default uid/muid for legacy support */
+       unsigned int gid;       /* default gid for legacy support */
+
+       /* book keeping */
+       struct v9fs_idpool fidpool;     /* The FID pool for file descriptors */
+       struct v9fs_idpool tidpool;     /* The TID pool for transactions ids */
+
+       /* transport information */
+       struct v9fs_transport *transport;
+
+       int inprogress;         /* session in progress => true */
+       int shutdown;           /* session shutting down. no more attaches. */
+       unsigned char session_hung;
+
+       /* mux private data */
+       struct v9fs_fcall *curfcall;
+       wait_queue_head_t read_wait;
+       struct completion fcread;
+       struct completion proccmpl;
+       struct task_struct *recvproc;
+
+       spinlock_t muxlock;
+       struct list_head mux_fcalls;
+};
+
+/* possible values of ->proto */
+enum {
+       PROTO_TCP,
+       PROTO_UNIX,
+       PROTO_FD,
+};
+
+int v9fs_session_init(struct v9fs_session_info *, const char *, char *);
+struct v9fs_session_info *v9fs_inode2v9ses(struct inode *);
+void v9fs_session_close(struct v9fs_session_info *v9ses);
+int v9fs_get_idpool(struct v9fs_idpool *p);
+void v9fs_put_idpool(int id, struct v9fs_idpool *p);
+int v9fs_get_option(char *opts, char *name, char *buf, int buflen);
+long long v9fs_get_int_option(char *opts, char *name, long long dflt);
+int v9fs_parse_tcp_devname(const char *devname, char **addr, char **remotename);
+
+#define V9FS_MAGIC 0x01021997
+
+/* other default globals */
+#define V9FS_PORT              564
+#define V9FS_DEFUSER   "nobody"
+#define V9FS_DEFANAME  ""
+
+/* inital pool sizes for fids and tags */
+#define V9FS_START_FIDS 8192
+#define V9FS_START_TIDS 256
diff --git a/fs/9p/v9fs_vfs.h b/fs/9p/v9fs_vfs.h
new file mode 100644 (file)
index 0000000..2f2cea7
--- /dev/null
@@ -0,0 +1,53 @@
+/*
+ * V9FS VFS extensions.
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+/* plan9 semantics are that created files are implicitly opened.
+ * But linux semantics are that you call create, then open.
+ * the plan9 approach is superior as it provides an atomic
+ * open.
+ * we track the create fid here. When the file is opened, if fidopen is
+ * non-zero, we use the fid and can skip some steps.
+ * there may be a better way to do this, but I don't know it.
+ * one BAD way is to clunk the fid on create, then open it again:
+ * you lose the atomicity of file open
+ */
+
+/* special case:
+ * unlink calls remove, which is an implicit clunk. So we have to track
+ * that kind of thing so that we don't try to clunk a dead fid.
+ */
+
+extern struct file_system_type v9fs_fs_type;
+extern struct file_operations v9fs_file_operations;
+extern struct file_operations v9fs_dir_operations;
+extern struct dentry_operations v9fs_dentry_operations;
+
+struct inode *v9fs_get_inode(struct super_block *sb, int mode);
+ino_t v9fs_qid2ino(struct v9fs_qid *qid);
+void v9fs_mistat2inode(struct v9fs_stat *, struct inode *,
+                      struct super_block *);
+int v9fs_dir_release(struct inode *inode, struct file *filp);
+int v9fs_file_open(struct inode *inode, struct file *file);
+void v9fs_inode2mistat(struct inode *inode, struct v9fs_stat *mistat);
+void v9fs_dentry_release(struct dentry *);
diff --git a/fs/9p/vfs_super.c b/fs/9p/vfs_super.c
new file mode 100644 (file)
index 0000000..ce0778a
--- /dev/null
@@ -0,0 +1,271 @@
+/*
+ *  linux/fs/9p/vfs_super.c
+ *
+ * This file contians superblock ops for 9P2000. It is intended that
+ * you mount this file system on directories.
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/fs.h>
+#include <linux/file.h>
+#include <linux/stat.h>
+#include <linux/string.h>
+#include <linux/smp_lock.h>
+#include <linux/inet.h>
+#include <linux/pagemap.h>
+#include <linux/seq_file.h>
+#include <linux/mount.h>
+#include <linux/idr.h>
+
+#include "debug.h"
+#include "v9fs.h"
+#include "9p.h"
+#include "v9fs_vfs.h"
+#include "conv.h"
+#include "fid.h"
+
+static void v9fs_clear_inode(struct inode *);
+static struct super_operations v9fs_super_ops;
+
+/**
+ * v9fs_clear_inode - release an inode
+ * @inode: inode to release
+ *
+ */
+
+static void v9fs_clear_inode(struct inode *inode)
+{
+       filemap_fdatawrite(inode->i_mapping);
+}
+
+/**
+ * v9fs_set_super - set the superblock
+ * @s: super block
+ * @data: file system specific data
+ *
+ */
+
+static int v9fs_set_super(struct super_block *s, void *data)
+{
+       s->s_fs_info = data;
+       return set_anon_super(s, data);
+}
+
+/**
+ * v9fs_fill_super - populate superblock with info
+ * @sb: superblock
+ * @v9ses: session information
+ *
+ */
+
+static void
+v9fs_fill_super(struct super_block *sb, struct v9fs_session_info *v9ses,
+               int flags)
+{
+       sb->s_maxbytes = MAX_LFS_FILESIZE;
+       sb->s_blocksize_bits = fls(v9ses->maxdata - 1);
+       sb->s_blocksize = 1 << sb->s_blocksize_bits;
+       sb->s_magic = V9FS_MAGIC;
+       sb->s_op = &v9fs_super_ops;
+
+       sb->s_flags = flags | MS_ACTIVE | MS_SYNCHRONOUS | MS_DIRSYNC |
+           MS_NODIRATIME | MS_NOATIME;
+}
+
+/**
+ * v9fs_get_sb - mount a superblock
+ * @fs_type: file system type
+ * @flags: mount flags
+ * @dev_name: device name that was mounted
+ * @data: mount options
+ *
+ */
+
+static struct super_block *v9fs_get_sb(struct file_system_type
+                                      *fs_type, int flags,
+                                      const char *dev_name, void *data)
+{
+       struct super_block *sb = NULL;
+       struct v9fs_fcall *fcall = NULL;
+       struct inode *inode = NULL;
+       struct dentry *root = NULL;
+       struct v9fs_session_info *v9ses = NULL;
+       struct v9fs_fid *root_fid = NULL;
+       int mode = S_IRWXUGO | S_ISVTX;
+       uid_t uid = current->fsuid;
+       gid_t gid = current->fsgid;
+       int stat_result = 0;
+       int newfid = 0;
+       int retval = 0;
+
+       dprintk(DEBUG_VFS, " \n");
+
+       v9ses = kcalloc(1, sizeof(struct v9fs_session_info), GFP_KERNEL);
+       if (!v9ses)
+               return ERR_PTR(-ENOMEM);
+
+       if ((newfid = v9fs_session_init(v9ses, dev_name, data)) < 0) {
+               dprintk(DEBUG_ERROR, "problem initiating session\n");
+               retval = newfid;
+               goto free_session;
+       }
+
+       sb = sget(fs_type, NULL, v9fs_set_super, v9ses);
+
+       v9fs_fill_super(sb, v9ses, flags);
+
+       inode = v9fs_get_inode(sb, S_IFDIR | mode);
+       if (IS_ERR(inode)) {
+               retval = PTR_ERR(inode);
+               goto put_back_sb;
+       }
+
+       inode->i_uid = uid;
+       inode->i_gid = gid;
+
+       root = d_alloc_root(inode);
+
+       if (!root) {
+               retval = -ENOMEM;
+               goto release_inode;
+       }
+
+       sb->s_root = root;
+
+       /* Setup the Root Inode */
+       root_fid = v9fs_fid_create(root);
+       if (root_fid == NULL) {
+               retval = -ENOMEM;
+               goto release_dentry;
+       }
+
+       root_fid->fidopen = 0;
+       root_fid->v9ses = v9ses;
+
+       stat_result = v9fs_t_stat(v9ses, newfid, &fcall);
+       if (stat_result < 0) {
+               dprintk(DEBUG_ERROR, "stat error\n");
+               v9fs_t_clunk(v9ses, newfid, NULL);
+               v9fs_put_idpool(newfid, &v9ses->fidpool);
+       } else {
+               root_fid->fid = newfid;
+               root_fid->qid = fcall->params.rstat.stat->qid;
+               root->d_inode->i_ino =
+                   v9fs_qid2ino(&fcall->params.rstat.stat->qid);
+               v9fs_mistat2inode(fcall->params.rstat.stat, root->d_inode, sb);
+       }
+
+       kfree(fcall);
+
+       if (stat_result < 0) {
+               retval = stat_result;
+               goto release_dentry;
+       }
+
+       return sb;
+
+      release_dentry:
+       dput(sb->s_root);
+
+      release_inode:
+       iput(inode);
+
+      put_back_sb:
+       up_write(&sb->s_umount);
+       deactivate_super(sb);
+       v9fs_session_close(v9ses);
+
+      free_session:
+       kfree(v9ses);
+
+       return ERR_PTR(retval);
+}
+
+/**
+ * v9fs_kill_super - Kill Superblock
+ * @s: superblock
+ *
+ */
+
+static void v9fs_kill_super(struct super_block *s)
+{
+       struct v9fs_session_info *v9ses = s->s_fs_info;
+
+       dprintk(DEBUG_VFS, " %p\n", s);
+
+       v9fs_dentry_release(s->s_root); /* clunk root */
+
+       kill_anon_super(s);
+
+       v9fs_session_close(v9ses);
+       kfree(v9ses);
+       dprintk(DEBUG_VFS, "exiting kill_super\n");
+}
+
+/**
+ * v9fs_show_options - Show mount options in /proc/mounts
+ * @m: seq_file to write to
+ * @mnt: mount descriptor
+ *
+ */
+
+static int v9fs_show_options(struct seq_file *m, struct vfsmount *mnt)
+{
+       struct v9fs_session_info *v9ses = mnt->mnt_sb->s_fs_info;
+
+       if (v9ses->debug != 0)
+               seq_printf(m, ",debug=%u", v9ses->debug);
+       if (v9ses->port != V9FS_PORT)
+               seq_printf(m, ",port=%u", v9ses->port);
+       if (v9ses->maxdata != 9000)
+               seq_printf(m, ",msize=%u", v9ses->maxdata);
+       if (v9ses->afid != ~0)
+               seq_printf(m, ",afid=%u", v9ses->afid);
+       if (v9ses->proto == PROTO_UNIX)
+               seq_puts(m, ",proto=unix");
+       if (v9ses->extended == 0)
+               seq_puts(m, ",noextend");
+       if (v9ses->nodev == 1)
+               seq_puts(m, ",nodevmap");
+       seq_printf(m, ",name=%s", v9ses->name);
+       seq_printf(m, ",aname=%s", v9ses->remotename);
+       seq_printf(m, ",uid=%u", v9ses->uid);
+       seq_printf(m, ",gid=%u", v9ses->gid);
+       return 0;
+}
+
+static struct super_operations v9fs_super_ops = {
+       .statfs = simple_statfs,
+       .clear_inode = v9fs_clear_inode,
+       .show_options = v9fs_show_options,
+};
+
+struct file_system_type v9fs_fs_type = {
+       .name = "9P",
+       .get_sb = v9fs_get_sb,
+       .kill_sb = v9fs_kill_super,
+       .owner = THIS_MODULE,
+};