incubator-nuttx/Documentation/NfsHowto.html

389 lines
13 KiB
HTML

<html>
<head>
<title>NFS Client How-To</title>
</head>
<body background="backgd.gif">
<hr><hr>
<table width ="100%">
<tr align="center" bgcolor="#e4e4e4">
<td>
<h1><big><font color="#3c34ec"><i>NFS Client How-To</i></font></big></h1>
<p>Last Updated: June 18, 2012</p>
</td>
</tr>
</table>
<hr><hr>
<table width ="100%">
<tr bgcolor="#e4e4e4">
<td>
<h1>Table of Contents</h1>
</td>
</tr>
</table>
<center><table width ="80%">
<tr>
<td>
<table>
<tr>
<td valign="top" width="22"><img height="20" width="20" src="favicon.ico"></td>
<td>
<a href="#nfsconfiguration">Adding NFS to the NuttX Configuration</a>
</td>
</tr>
</table>
<table>
<tr>
<td valign="top" width="22"><img height="20" width="20" src="favicon.ico"></td>
<td>
<a href="#mountinterface">Mount Interface</a>
</td>
</tr>
</table>
<table>
<tr>
<td valign="top" width="22"><img height="20" width="20" src="favicon.ico"></td>
<td>
<a href="#nfsmount">NFS Mount Command</a>
</td>
</tr>
</table>
<table>
<tr>
<td valign="top" width="22"><img height="20" width="20" src="favicon.ico"></td>
<td>
<a href="#serverconfig">Configuring the NFS server (Ubuntu)</a>
</td>
</tr>
</table>
</td>
</tr>
</table></center>
<table width ="100%">
<tr bgcolor="#e4e4e4">
<td>
<a name="nfsconfiguration"><h1>Adding NFS to the NuttX Configuration</h1></a>
</td>
</tr>
</table>
<p>
The NFS client is easily added to your configuration:
You simply need to add <code>CONFIG_NFS</code> to your <code>nuttx/.config</code> file.
There are, however, a few dependencies on other system settings:
</p>
<ol>
<li>
First, there are number of things that you must configure in order to be able to use any file system:
</li>
<ul>
<li>
<code>CONFIG_NFILE_DESCRIPTORS > 0</code>. You must include support for file descriptors.
</li>
<li>
<code>CONFIG_DISABLE_MOUNTPOINT=n</code>. You must include support for mount points in the pseudo-file system.
</li>
</ul>
<li>
And there are several dependencies on the networking configuration.
At a minimum, you need to have the following selections:
</li>
<ul>
<li>
<code>CONFIG_NET=y</code>. General networking support.
</li>
<li>
<code>CONFIG_NET_UDP=y</code>. Support for UDP.
</li>
</ul>
</ol>
<table width ="100%">
<tr bgcolor="#e4e4e4">
<td>
<a name="mountinterface"><h1>Mount Interface</h1></a>
</td>
</tr>
</table>
<p>
A low-level, C-callable interface is provided to mount a file system.
That interface is called <code>mount()</code> and is mentioned in the <a href="NuttxPortingGuide.html#NxFileSystem"><code>porting guide</code></a> and is prototyped in the header file <code>include/sys/mount.h</code>:
</p>
<ul><pre>
int mount(const char *source, const char *target, const char *filesystemtype, unsigned long mountflags, const void *data);
</pre></ul>
<p>
<b>Synopsis</b>:
<code>mount()</code> attaches the filesystem specified by the <code>source</code> block device name into the root file system at the path specified by <code>target</code>.
</p>
<p>
<b>Input Paramters</b>:
<ul>
<li><code>source</code>. A null-terminated string providing the fill path to a block driver in the NuttX pseudo-file system.
<li><code>target</code>. The location in the NuttX pseudo-file system where the volume will be mounted.
<li><code>filesystemtype</code>. A string identifying the type of file system to use.
<li><code>mountflags</code>. Various flags that can be used to qualify how the file system is mounted.
<li><code>data</code>. Opaque data that is passed to the file system with the mount occurs.
</ul>
</p>
<p>
<b>Returned Values</b>
Zero is returned on success; -1 is returned on an error and <code>errno</code> is set appropriately:
<ul>
<li><code>EACCES</code>.
A component of a path was not searchable or mounting a read-onlyfilesystem was attempted without giving the <code>MS_RDONLY</code> flag.
</li>
<li><code>EBUSY</code>.
<code>source</code> is already mounted.
</li>
<li><code>EFAULT</code>.
One of the pointer arguments points outside the user address space.
</li>
<li><code>EINVAL</code>.
<code>source</code> had an invalid superblock.
</li>
<li><code>ENODEV</code>.
<code>filesystemtype</code> not configured
</li>
<li><code>ENOENT</code>.
A pathname was empty or had a nonexistent component.
</li>
<li><code>ENOMEM</code>.
Could not allocate a memory to copy filenames or data into.
</li>
<li><code>ENOTBLK</code>.
<code>source</code> is not a block device
</li>
</ul>
</p>
<p>
This same interface can be used to mount a remote, NFS file system using some special parameters.
The NFS mount differs from the <i>normal</i> file system mount in that: (1) there is no block driver for the NFS file system, and (2) special parameters must be passed as <code>data</code> to describe the remote NFS server.
Thus the following code snippet might represent how an NFS file system is mounted:
</p>
<ul><pre>
#include &lt;sys/mount.h&gt;
#include &lt;nuttx/fs/nfs.h&gt;
struct nfs_args data;
char *mountpoint;
ret = mount(NULL, mountpoint, string &quot;nfs&quot;, 0, (FAR void *)&data);
</pre></ul>
<p>
NOTE that: (1) the block driver paramter is <code>NULL</code>.
The <code>mount()</code> is smart enough to know that no block driver is needed with the NFS file system.
(2) The NFS file system is identified with the simple string &quot;nfs&quot;
(3) A reference to <code>struct nfs_args</code> is passed as an NFS-specific argument.
</p>
<p>
The NFS-specific interface is described in the file <code>include/nuttx/fs/nfs.h</code>.
There you can see that <code>struct nfs_args</code> is defined as:
</p>
<ul><pre>
struct nfs_args
{
uint8_t addrlen; /* Length of address */
uint8_t sotype; /* Socket type */
uint8_t flags; /* Flags, determines if following are valid: */
uint8_t timeo; /* Time value in deciseconds (with NFSMNT_TIMEO) */
uint8_t retrans; /* Times to retry send (with NFSMNT_RETRANS) */
uint16_t wsize; /* Write size in bytes (with NFSMNT_WSIZE) */
uint16_t rsize; /* Read size in bytes (with NFSMNT_RSIZE) */
uint16_t readdirsize; /* readdir size in bytes (with NFSMNT_READDIRSIZE) */
char *path; /* Server's path of the directory being mount */
struct sockaddr_storage addr; /* File server address (requires 32-bit alignment) */
};
</pre></ul>
<table width ="100%">
<tr bgcolor="#e4e4e4">
<td>
<a name="nfsmount"><h1>NFS Mount Command</h1></a>
</td>
</tr>
</table>
<p>
The <a href="NuttShell.html">NuttShell (NSH)</a> also supports a command called <code>nfsmount</code>
that can be used to mount a remote file system via the NSH command line.
</p>
<p>
<b>Command Syntax:</b>
</p>
<ul><pre>
nfsmount &lt;server-address&gt; &lt;mount-point&gt; &lt;remote-path&gt;
</pre></ul>
<p>
<b>Synopsis</b>.
The <code>nfsmount</code> command mounts a network file system in the NuttX pseudo filesystem.
The <code>nfsmount</code> will use NFSv3 UDP protocol to mount the remote file system.
</p>
<p>
<b>Command Line Arguments</b>.
The <code>nfsmount</code> takes three arguments:
</p>
<ol>
<li>
The <code>&lt;server-address&gt;</code> is the IP address of the server exporting the file system you wish to mount.
This implementation of NFS for the NuttX RTOS is only for a local area network, so the server and client must be in the same network.
</li>
<li>
The <code>&lt;mount-point &gt;</code> is the location in the NuttX pseudo filesystem where the mounted volume will appear.
This mount point can only reside in the NuttX pseudo filesystem.
By convention, this mount point is a subdirectory under <code>/mnt</code>.
The mount command will create whatever pseudo directories that may be needed to complete the full path (but the full path must not already exist).
</li>
<li>
The <code>&lt;remote-path&gt;</code> is the file system <code>/</code> directory being exported from server.
This <code>/</code> directory must have been configured for exportation on the server before when the NFS server was set up.
</li>
</ol>
<p>
After the volume has been mounted in the NuttX pseudo filesystem, it may be access in the same way as other objects in the file system.
</p>
<p>
<b>Example</b>.
Suppose the the NFS server has been configured to export the directory <code>/export/shared</code>.
The the following command would mount that file system (assuming that the target also has privileges to mount the file system).
</p>
<ul><pre>
NuttShell (NSH)
nsh&gt; ls /mnt
/mnt:
nsh: ls: no such directory: /mnt
nsh&gt; nfsmount 10.0.0.1 /mnt/nfs /export/shared
nsh&gt; ls -l /mnt/nfs
/mnt/nfs:
drwxrwxrwx 4096 ..
drwxrwxrwx 4096 testdir/
-rw-rw-rw- 6 ctest.txt
-rw-r--r-- 15 btest.txt
drwxrwxrwx 4096 .
nsh&gt; echo &quot;This is a test&quot; &gt;/mnt/nfs/testdir/testfile.txt
nsh&gt; ls -l /mnt/nfs/testdir
/mnt/nfs/testdir:
-rw-rw-rw- 21 another.txt
drwxrwxrwx 4096 ..
drwxrwxrwx 4096 .
-rw-rw-rw- 16 testfile.txt
nsh&gt; cat /mnt/nfs/testdir/testfile.txt
This is a test
</pre></ul>
<table width ="100%">
<tr bgcolor="#e4e4e4">
<td>
<a name="serverconfig"><h1>Configuring the NFS server (Ubuntu)</h1></a>
</td>
</tr>
</table>
<p>
Setting up the server will be done in two steps:
First, setting up the configuration file for NFS, and then starting the NFS services.
But first, you need to install the nfs server on Ubuntu with the these two commands:
</p>
<ul><pre>
# sudo apt-get install nfs-common</FONT>
# sudo apt-get install nfs-kernel-server</FONT>
</pre></ul>
<p>
After that, we need to make or choose the directory we want to export from the NFS server.
In our case, we are going to make a new directory called <code>/export</code>.
</p>
<ul><pre>
# sudo mkdir /export
</pre></ul>
<p>
It is important that <code>/export</code> directory allow access to everyone (777 permissions) as we will be accessing the NFS share from the client with no authentication.
</p>
<ul><pre>
# sudo chmod 777 /export
</pre></ul>
<p>
When all this is done, we will need to edit the configuration file to set up an NFS server: <code>/etc/exports</code>.
This file contains a list of entries;
each entry indicates a volume that is shared and how it is shared.
For more information for a complete description of all the setup options for this file you can check in the man pages (<code>man export</code>).</p>
An entry in <code>/etc/exports</code> will typically look like this:
</p>
<ul><pre>
directory machine1(option11,option12)
</pre></ul>
<p>
So for our example we export <coce>/export</code> to the client 10.0.0.2 add the entry:
</p>
<ul><pre>
/export 10.0.0.2(rw)
</pre></ul>
<p>
In our case we are using all the default options except for the <code>ro</code> that we replaced with <code>rw</code> so that our client will have read and write access to the directory that we are exporting.
</p>
</p>
After we do all the require configurations, we are ready to start the server with the next command:
</p>
<ul><pre>
# sudo /etc/init.d/nfs-kernel-server start
</pre></ul>
</p>
Note: If you later decide to add more NFS exports to the /etc/exports file, you will need to either restart NFS daemon
or run command exportfs.
</p>
<ul><pre>
# sudo /etc/init.d/nfs-kernel-server start
</pre></ul>
<p>Or</p>
<ul><pre>
# exportfs -ra
</pre></ul>
<p>
Now we can check if the export directory and our mount point is properly set up.
</p>
<ul><pre>
# sudo showmount -e
# sudo showmount -a
</pre></ul>
<p>
And also we can verify if NFS is running in the system with:
</p>
<P STYLE="margin-left: 0.49in; margin-bottom: 0in; line-height: 100%">
<ul><pre>
# rpcinfo &ndash;p</FONT>
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100011 1 udp 749 rquotad
100011 2 udp 749 rquotad
100005 1 udp 759 mountd
100005 1 tcp 761 mountd
100005 2 udp 764 mountd
100005 2 tcp 766 mountd
100005 3 udp 769 mountd
100005 3 tcp 771 mountd
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
300019 1 tcp 830 amd
300019 1 udp 831 amd
100024 1 udp 944 status
100024 1 tcp 946 status
100021 1 udp 1042 nlockmgr
100021 3 udp 1042 nlockmgr
100021 4 udp 1042 nlockmgr
100021 1 tcp 1629 nlockmgr
100021 3 tcp 1629 nlockmgr
100021 4 tcp 1629 nlockmgr
</pre></ul>
<p>
Now your NFS sever is sharing <code>/export</code> directory to be accessed.
</p>
</body>
</html>