path: root/hurd/documentation/translators.html
diff options
Diffstat (limited to 'hurd/documentation/translators.html')
1 files changed, 236 insertions, 0 deletions
diff --git a/hurd/documentation/translators.html b/hurd/documentation/translators.html
new file mode 100644
index 00000000..4e47a9c0
--- /dev/null
+++ b/hurd/documentation/translators.html
@@ -0,0 +1,236 @@
+[[meta copyright="Copyright © 1998, 1999, 2007, 2008 Free Software Foundation,
+[[meta license="Verbatim copying and distribution of this entire article is
+permitted in any medium, provided this notice is preserved."]]
+[[meta title="Translators"]]
+By Marcus Brinkmann.
+<li><a href="#concept" name="TOC_concept">Concept</a></li>
+<li><a href="#examples" name="TOC_examples">Examples</a></li>
+<li><a href="#actpas" name="TOC_actpas">Passive Translators, Active Translators</a></li>
+<li><a href="#manage" name="TOC_manage">Managing Translators</a></li>
+<h3><a href="#TOC_concept" name="concept">Concept</a></h3>
+Before we take a closer look at translators, let us consider regular
+filesystems. A filesystem is store for a hierarchical tree of directories
+and files. You access directories and files by a special character string,
+the path. Furthermore, there are symbolic links to refer to one file at
+several places in the tree, there are hard links to give one and the same
+file several names. There are also special device files for communication
+with the hardware device drivers of the kernel, and there are mount points
+to include other stores in the directory tree. Then there are obscure
+objects like fifos and hard links.</p>
+Although these objects are very different, they share some common
+properties, for example, they have all an owner and a group associated with
+them as well as access rights (permissions). This information is written in
+inodes. This is a actually a further commonality: Every object has exactly
+one inode associated with it (hard links are somewhat special as they share
+one and the same inode). Sometimes, the inode has further information
+stored in it. For example, the inode can contain the target of a symbolic
+However, these commonalities are usually not exploited in the
+implementations, despite the common programming interface to them. All
+inodes can be accessed through the standard POSIX calls, for example
+<code>read()</code> and <code>write()</code>. For example, to add a new
+object type (for example a new link type) to a common monolithic unix
+kernel, you would need to modify the code for each filesystem
+In the Hurd, things work differently. Although in the Hurd a special
+filesystem server can exploit special properties of standard object types
+like links (in the ext2 filesystem with fast links, for example), it has a
+general interface to add such features without modifying existing code.</p>
+The trick is to allow a program to be inserted between the actual content of
+a file and the user accessing this file. Such a program is called a
+translator, because it is able to process the incoming requests in many
+different ways. In other words, a translator is a Hurd server which provides
+the basic filesystem interface.</p>
+Translators have very interesting properties. From the kernels point of
+view, they are just another user process. This means, translators can be run
+by any user. You don't need root priviligies to install or modify a
+translator, you only need the access rights for the underlying inode the
+translator is attached to. Many translators don't require an actual file to
+operate, they can provide information by their own means. This is why
+the information about translators is stored in the inode.</p>
+Translators are responsible to serve all file system operations that involve
+the inode they are attached to. Because they are not restricted to the usual
+set of objects (device file, link etc), they are free to return anything
+that makes sense to the programmer. One could imagine a translator that
+behaves like a directory when accessed by <code>cd</code> or
+<code>ls</code> and at the same time behaves like a file when accessed by
+<h3><a href="#TOC_examples" name="examples">Examples</a></h3>
+<h4>Mount Points</h4>
+A mount point can be seen as an inode that has a special translator attached
+to it. Its purpose would be to translate filesystem operations on the mount
+point in filesystem operations on another store, let's say, another
+Indeed, this is how filesystems are implemented under the Hurd. A
+filesystem is a translator. This translator takes a store as its argument,
+and is able to serve all filesystem operations transparently.</p>
+<h4>Device Files</h4>
+There are many different device files, and in systems with a monolithical
+kernel, they are all provided by the kernel itself. In the Hurd, all device
+files are provided by translators. One translator can provide support for
+many similar device files, for example all hard disk partitions. This way,
+the number of actual translators needed is quite small. However, note that
+for each device file accessed, a seperate translator task is started.
+Because the Hurd is heavily multi threaded, this is very cheap.</p>
+When hardware is involved, a translator usually starts to communicate with
+the kernel to get the data from the hardware. However, if no hardware access
+is necessary, the kernel does not need to be involved. For example,
+<code>/dev/zero</code> does not require hardware access, and can therefore
+be implemented completely in user space.</p>
+<h4>Symbolic Links</h4>
+A symbolic link can be seen as a translator. Accesing the symbolic link
+would start up the translator, which would forward the request to the
+filesystem that contains the file the link points to.</p>
+However, for better performance, filesystems that have native support
+for symbolic links can take advantage of this feature and implement
+symbolic links differently. Internally, accessing a symbolic link would not
+start a new translator process. However, to the user, it would still look
+as if a passive translator is involved (see below for an explanation what a
+passsive translator is).</p>
+Because the Hurd ships with a symlink translator, any filesystem server that
+provides support for translators automatically has support for symlinks (and
+firmlinks, and device files etc)! This means, you can get a working
+filesystem very fast, and add native support for symlinks and other features
+<h3><a href="#TOC_actpas" name="actpas">Passive Translators, Active Translators</a></h3>
+There are two types of translators, passive and active. They are really
+completely different things, so don't mix them up, but they have a close
+relation to each other.</p>
+<h4>Active Translators</h4>
+An active translator is a running translator process, as introduced above.
+You can set and remove active translators using the
+<code>settrans -a</code></a>
+command. The <code>-a</code> option is necessary to tell
+<code>settrans</code> that you want to modify the active translator.</p>
+The <code>settrans</code> command takes three kind of arguments. First, you
+can set options for the <code>settrans</code> command itself, like
+<code>-a</code> to modify the active translator. Then you set the inode you
+want to modify. Remember that a translator is always associated with an
+inode in the directory hierarchy. You can only modify one inode at a time.
+If you do not specify any more arguments, <code>settrans</code> will try to
+remove an existing translator. How hard it tries depends on the force
+options you specify (if the translator is in use by any process, you will
+get "device or resource busy" error message unless you force it to go away).</p>
+But if you specify further arguments, it will be interpreted as a command
+line to run the translator. This means, the next argument is the filename of
+the translator executable. Further arguments are options to the translator,
+and not to the <code>settrans</code> command.</p>
+For example, to mount an ext2fs partition, you can run
+<code>settrans -a -c /mnt /hurd/ext2fs /dev/hd2s5</code>. The
+<code>-c</code> option will create the mount point for you if it doesn't
+exist already. This does not need to be a directory, by the way. To unmount,
+you would try <code>settrans -a /mnt</code>.</p>
+<h4>Passive Translators</h4>
+A passive translator is set and modified with the same syntax as the active
+translator (just leave away the <code>-a</code>, so everything said above is
+true for passive translators, too. However, there is a difference: passive
+translators are not yet started.</p>
+This makes sense, because this is what you usually want. You don't want the
+partition mounted unless you really access files on this partition. You
+don't want to bring up the network unless there is some traffic and so
+Instead, the first time the passive translator is accessed, it is
+automatically read out of the inode and an active translator is started on
+top of it using the command line that was stored in the inode. This is
+similar to the Linux automounter functionality. However, it does not come as
+an additional bonus that you have to set up manually, but an integral part of
+the system. So, setting passive translators defers starting the translator
+task until you really need it. By the way, if the active translator dies for
+some reason, the next time the inode is accessed the translator is
+There is a further difference: active translators can die or get lost. As
+soon as the active translator process is killed (for example, because you
+reboot the machine) it is lost forever. Passive translators are not transient
+and stay in the inode during reboots until you modify them with the
+<code>settrans</code> program or delete the inodes they are attached to.
+This means, you don't need to maintain a configuration file with your mount
+One last point: Even if you have set a passive translator, you can still
+set a different active translator. Only if the translator is automatically
+started because there was no active translator the time the inode was
+accessed the passive translator is considered.</p>
+<h3><a href="#TOC_manage" name="manage">Managing Translators</a></h3>
+As mentioned above, you can use
+to set and alter passive and active translators. There are a lot of options
+to change the behaviour of <code>settrans</code> in case something goes
+wrong, and to conditionalize its action. Here are some common usages:</p>
+<ul><li><code>settrans -c /mnt /hurd/ext2fs /dev/hd2s5</code> mounts a
+partition, the translator will stay across reboots.</li>
+<li><code>settrans -a /mnt /hurd/ext2fs ~/dummy.fs</code> mounts a
+filesystem inside a data file, the translator will go away if it dies.</li>
+<li><code>settrans -fg /nfs-data</code> forces a translator to go away.</li>
+You can use the <code>showtrans</code></a>
+command to see if a translator is attached to an inode. This will only show
+you the passive translator though.</p>
+You can change the options of an active (filesystem) translator with
+<code>fsysopts</code> without actually restarting it. This is very
+convenient. For example, you can do what is called "remounting a
+partition read-only" under Linux simply by running <code>fsysopts
+/mntpoint --readonly</code>. The running active translator
+will change its behaviour according to your request if possible.
+<code>fsysopts /mntpoint</code> without a parameter shows you the current
+I recommend that you start by reading the <code>/bin/mount</code> command,
+it is only a small script. Because setting filesystem translators is
+similar to mounting partitions, you can easily grasp the concept this way.
+Make a file system image with <code>dd if=/dev/zero of=dummy.fs bs=1024k
+count=8; mke2fs dummy.fs</code> and "mount" it with <code>settrans -c dummy
+/hurd/ext2fs `pwd`/dummy.fs</code>. Note that the translator is not started
+yet, no new <code>ext2fs</code> process is running (verify with <code>ps
+Aux</code>). Check that everything is correct using <code>showtrans</code></p>
+Now type <code>ls dummy</code> and you will notice the short delay that
+occurs while the translator is started. After that, there will be no more
+delays accessing dummy. Under Linux, one would say that you automounted a
+loop file system. Check with <code>ps Aux</code> that there is an <code>ext2fs
+dummy</code> process up and running now. Now put some files into the new
+directory. Try to make the filesystem read-only with <code>fsysopts</code>.
+Note how further write attempts fail now. Try to kill the active translator
+with <code>settrans -g</code>.</p>
+You should have some understanding of what is going on now. Now remember
+that this was only <em>one</em> special server, the Hurd ext2fs server.
+There are many more server in the <code>hurd</code> directory. Some of them
+are for filesystems. Some are needed for file system features like links.
+Some are needed for device files. Some are useful for networking. Imagine
+"mounting" an FTP Server with <code>settrans</code> and downloading files
+simply with the standard <code>cp</code> command. Or editing your web sites
+with <code>emacs /ftp/</code>!</p>