<html>
<head>
<title>Implementing a new API in Libvirt</title>
</head>
<body>
<h1>Implementing a new API in Libvirt</h1>
<ul id="toc"></ul>
<p>
This document walks you through the process of implementing a new
API in libvirt. It uses as an example the addition of the node device
create and destroy APIs.
</p>
<p>
Before you begin coding, it is critical that you propose your
changes on the libvirt mailing list and get feedback on your ideas to
make sure what you're proposing fits with the general direction of the
project. Even before doing a proof of concept implementation, send an
email giving an overview of the functionality you think should be
added to libvirt. Someone may already be working on the feature you
want. Also, recognize that everything you write is likely to undergo
significant rework as you discuss it with the other developers, so
don't wait too long before getting feedback.
</p>
<p>
Adding a new API to libvirt is not difficult, but there are quite a
few steps. This document assumes that you are familiar with C
programming and have checked out the libvirt code from the source code
repository and successfully built the existing tree. Instructions on
how to check out and build the code can be found at:
</p>
<p>
<a href="http://libvirt.org/downloads.html">http://libvirt.org/downloads.html</a>
</p>
<p>
Once you have a working development environment, the steps to create a
new API are:
</p>
<ol>
<li>define the public API</li>
<li>define the internal driver API</li>
<li>implement the public API</li>
<li>define the wire protocol format</li>
<li>implement the RPC client</li>
<li>implement the server side dispatcher</li>
<li>implement the driver methods</li>
<li>add virsh support</li>
</ol>
<p>
It is, of course, possible to implement the pieces in any order, but
if the development tasks are completed in the order listed, the code
will compile after each step. Given the number of changes required,
verification after each step is highly recommended.
</p>
<p>
Submit new code in the form shown in the example code: one patch
per step. That's not to say submit patches before you have working
functionality--get the whole thing working and make sure you're happy
with it. Then use git or some other version control system that lets
you rewrite your commit history and break patches into pieces so you
don't drop a big blob of code on the mailing list at one go. For
example, I didn't follow my own advice when I originally submitted the
example code to the libvirt list but rather submitted it in several
large chunks. I've used git's ability to rewrite my commit history to
break the code apart into the example patches shown.
</p>
<p>
Don't mix anything else into the patches you submit. The patches
should be the minimal changes required to implement the functionality
you're adding. If you notice a bug in unrelated code (i.e., code you
don't have to touch to implement your API change) during development,
create a patch that just addresses that bug and submit it
separately.
</p>
<p>With that said, let's begin.</p>
<h2><a name='publicapi'>Defining the public API</a></h2>
<p>The first task is to define the public API and add it to:</p>
<p><code>include/libvirt/libvirt.h.in</code></p>
<p>
This task is in many ways the most important to get right, since once
the API has been committed to the repository, it's libvirt's policy
never to change it. Mistakes in the implementation are bugs that you
can fix. Make a mistake in the API definition and you're stuck with
it, so think carefully about the interface and don't be afraid to
rework it as you go through the process of implementing it.
</p>
<p>Once you have defined the API, you have to add the symbol names to:</p>
<p><code>src/libvirt_public.syms</code></p>
<p class="example">See <a href="api_extension/0001-Step-1-of-8-Define-the-public-API.patch">0001-Step-1-of-8-Define-the-public-API.patch</a> for example code.</p>
<h2><a name='internalapi'>Defining the internal API</a></h2>
<p>
Each public API call is associated with a driver, such as a host
virtualization driver, a network virtualization driver, a storage
virtualization driver, a state driver, or a device monitor. Adding
the internal API is ordinarily a matter of adding a new member to the
struct representing one of these drivers.
</p>
<p>
Of course, it's possible that the new API will involve the creation of
an entire new driver type, in which case the changes will include the
creation of a new struct type to represent the new driver type.
</p>
<p>The driver structs are defined in:</p>
<p><code>src/driver.h</code></p>
<p>
To define the internal API, first typedef the driver function
prototype and then add a new field for it to the relevant driver
struct.
</p>
<p class="example">See <a href="api_extension/0002-Step-2-of-8-Define-the-internal-driver-API.patch">0002-Step-2-of-8-Define-the-internal-driver-API.patch</a></p>
<h2><a name='implpublic'>Implementing the public API</a></h2>
<p>
Implementing the public API is largely a formality in which we wire up
public API to the internal driver API. The public API implementation
takes care of some basic validity checks before passing control to the
driver implementation. In RFC 2119 vocabulary, this function:
</p>
<ol class="ordinarylist">
<li>SHOULD log a message with VIR_DEBUG() indicating that it is
being called and its parameters;</li>
<li>MUST call virResetLastError();</li>
<li>SHOULD confirm that the connection is valid with
VIR_IS_CONNECT(conn);</li>
<li><strong>SECURITY: If the API requires a connection with write
privileges, MUST confirm that the connection flags do not
indicate that the connection is read-only;</strong></li>
<li>SHOULD do basic validation of the parameters that are being
passed in;</li>
<li>MUST confirm that the driver for this connection exists and that
it implements this function;</li>
<li>MUST call the internal API;</li>
<li>SHOULD log a message with VIR_DEBUG() indicating that it is
returning, its return value, and status.</li>
<li>MUST return status to the caller.</li>
</ol>
<p>The public API calls are implemented in:</p>
<p><code>src/libvirt.c</code></p>
<p class="example">See <a href="api_extension/0003-Step-3-of-8-Implement-the-public-API.patch">0003-Step-3-of-8-Implement-the-public-API.patch</a></p>
<h2><a name='wireproto'>Defining the wire protocol format</a></h2>
<p>
Defining the wire protocol is essentially a straightforward exercise
which is probably most easily understood by referring to the existing
remote protocol wire format definitions and the example patch. It
involves making two additions to:
</p>
<p><code>qemud/remote_protocol.x</code></p>
<p>
First, create two new structs for each new function that you're adding
to the API. One struct describes the parameters to be passed to the
remote function, and a second struct describes the value returned by
the remote function. The one exception to this rule is that functions
that return only integer status do not require a struct for returned
data.
</p>
<p>
Second, add values to the remote_procedure enum for each new function
added to the API.
</p>
<p class="example">See <a href="api_extension/0004-Step-4-of-8-Define-the-wire-protocol-format.patch">0004-Step-4-of-8-Define-the-wire-protocol-format.patch</a></p>
<p>
Once these changes are in place, it's necessary to run 'make rpcgen'
in the qemud directory to create the .c and .h files required by the
remote protocol code. This must be done on a Linux host using the
GLibC rpcgen program. Other rpcgen versions may generate code which
results in bogus compile time warnings
</p>
<h2><a name='rpcclient'>Implement the RPC client</a></h2>
<p>
Implementing the RPC client is also relatively mechanical, so refer to
the exising code and example patch for guidance. The RPC client uses
the rpcgen generated .h files. The remote method calls go in:
</p>
<p><code>src/remote_internal.c</code></p>
<p>Each remote method invocation does the following:</p>
<ol class="ordinarylist">
<li>locks the remote driver;</li>
<li>sets up the method arguments;</li>
<li>invokes the remote function;</li>
<li>checks the return value, if necessary;</li>
<li>extracts any returned data;</li>
<li>frees any returned data;</li>
<li>unlocks the remote driver.</li>
</ol>
<p>
Once you have created the remote method calls, you have to add fields
for them to the driver structs for the appropriate remote driver.
</p>
<p class="example">See <a href="api_extension/0005-Step-5-of-8-Implement-the-RPC-client.patch">0005-Step-5-of-8-Implement-the-RPC-client.patch</a></p>
<h2><a name="serverdispatch">Implement the server side dispatcher</a></h2>
<p>
Implementing the server side of the remote function calls is simply a
matter of deserializing the parameters passed in from the remote
caller and passing them to the corresponding internal API function.
The server side dispatchers are implemented in:
</p>
<p><code>qemud/remote.c</code></p>
<p>Again, this step uses the .h files generated by make rpcgen.</p>
<p class="example">See <a href="api_extension/0006-Step-6-of-8-Implement-the-server-side-dispatcher.patch">0006-Step-6-of-8-Implement-the-server-side-dispatcher.patch</a></p>
<h2><a name="driverimpl">Implement the driver methods</a></h2>
<p>
So, after all that, we get to the fun part. All functionality in
libvirt is implemented inside a driver. Thus, here is where you
implement whatever functionality you're adding to libvirt. You'll
either need to add additional files to the src directory or extend
files that are already there, depending on what functionality you're
adding.
</p>
<p>
In the example code, the extension is only an additional two function
calls in the node device API, so most of the new code is additions to
existing files. The only new files are there for multi-platform
implementation convenience, as some of the new code is Linux specific.
</p>
<p>
The example code is probably uninteresting unless you're concerned
with libvirt storage, but I've included it here to show how new files
are added to the build environment.
</p>
<p class="example">See <a href="api_extension/0007-Step-7-of-8-Implement-the-driver-methods.patch">0007-Step-7-of-8-Implement-the-driver-methods.patch</a></p>
<h2><a name="virsh">Implement virsh commands</a></h2>
<p>
Once you have the new functionality in place, the easiest way to test
it and also to provide it to end users is to implement support for it
in virsh.
</p>
<p>
A virsh command is composed of a few pieces of code. You need to
define an array of vshCmdInfo structs for each new command that
contain the help text and the command description text. You also need
an array of vshCmdOptDef structs to describe the command options.
Once you have those pieces of data in place you can write the function
implementing the virsh command. Finally, you need to add the new
command to the commands[] array.
</p>
<p class="example">See <a href="api_extension/0008-Step-8-of-8-Add-virsh-support.patch">0008-Step-8-of-8-Add-virsh-support.patch</a></p>
<p>Once you have working functionality, run make check and make
syntax-check before generating patches.</p>
</body>
</html>
