<!DOCTYPE html>
<html lang="en" data-content_root="../">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<title>1. How To Write Linux PCI Drivers — The Linux Kernel documentation</title>
<link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=fa44fd50" />
<link rel="stylesheet" type="text/css" href="../_static/alabaster.css?v=3918102e" />
<script src="../_static/documentation_options.js?v=5929fcd5"></script>
<script src="../_static/doctools.js?v=9bcbadda"></script>
<script src="../_static/sphinx_highlight.js?v=dc90522c"></script>
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="2. The PCI Express Port Bus Driver Guide HOWTO" href="pciebus-howto.html" />
<link rel="prev" title="PCI Bus Subsystem" href="index.html" />
<link rel="stylesheet" href="../_static/custom.css" type="text/css" />
</head><body>
<div class="document">
<div class="sphinxsidebar" role="navigation" aria-label="Main">
<div class="sphinxsidebarwrapper">
<p class="logo"><a href="../index.html">
<img class="logo" src="../_static/logo.svg" alt="Logo of The Linux Kernel"/>
</a></p>
<h1 class="logo"><a href="../index.html">The Linux Kernel</a></h1>
<p class="blurb">6.18.50</p>
<search id="searchbox" style="display: none" role="search">
<h3 id="searchlabel">Quick search</h3>
<div class="searchformwrapper">
<form class="search" action="../search.html" method="get">
<input type="text" name="q" aria-labelledby="searchlabel" autocomplete="off" autocorrect="off" autocapitalize="off" spellcheck="false"/>
<input type="submit" value="Go" />
</form>
</div>
</search>
<script>document.getElementById('searchbox').style.display = "block"</script>
<p>
<h3 class="kernel-toc-contents">Contents</h3>
<input type="checkbox" class="kernel-toc-toggle" id = "kernel-toc-toggle" checked>
<label class="kernel-toc-title" for="kernel-toc-toggle"></label>
<div class="kerneltoc" id="kerneltoc">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../process/development-process.html">Development process</a></li>
<li class="toctree-l1"><a class="reference internal" href="../process/submitting-patches.html">Submitting patches</a></li>
<li class="toctree-l1"><a class="reference internal" href="../process/code-of-conduct.html">Code of conduct</a></li>
<li class="toctree-l1"><a class="reference internal" href="../maintainer/index.html">Maintainer handbook</a></li>
<li class="toctree-l1"><a class="reference internal" href="../process/index.html">All development-process docs</a></li>
</ul>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="../core-api/index.html">Core API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../driver-api/index.html">Driver APIs</a></li>
<li class="toctree-l1 current"><a class="reference internal" href="../subsystem-apis.html">Subsystems</a><ul class="current">
<li class="toctree-l2"><a class="reference internal" href="../subsystem-apis.html#core-subsystems">Core subsystems</a></li>
<li class="toctree-l2"><a class="reference internal" href="../subsystem-apis.html#human-interfaces">Human interfaces</a></li>
<li class="toctree-l2"><a class="reference internal" href="../subsystem-apis.html#networking-interfaces">Networking interfaces</a></li>
<li class="toctree-l2"><a class="reference internal" href="../subsystem-apis.html#storage-interfaces">Storage interfaces</a></li>
<li class="toctree-l2 current"><a class="reference internal" href="../subsystem-apis.html#other-subsystems">Other subsystems</a><ul class="current">
<li class="toctree-l3"><a class="reference internal" href="../accounting/index.html">Accounting</a></li>
<li class="toctree-l3"><a class="reference internal" href="../cpu-freq/index.html">CPUFreq - CPU frequency and voltage scaling code in the Linux(TM) kernel</a></li>
<li class="toctree-l3"><a class="reference internal" href="../edac/index.html">EDAC Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../fpga/index.html">FPGA</a></li>
<li class="toctree-l3"><a class="reference internal" href="../i2c/index.html">I2C/SMBus Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../iio/index.html">Industrial I/O</a></li>
<li class="toctree-l3"><a class="reference internal" href="../pcmcia/index.html">PCMCIA</a></li>
<li class="toctree-l3"><a class="reference internal" href="../spi/index.html">Serial Peripheral Interface (SPI)</a></li>
<li class="toctree-l3"><a class="reference internal" href="../w1/index.html">1-Wire Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../watchdog/index.html">Watchdog Support</a></li>
<li class="toctree-l3"><a class="reference internal" href="../virt/index.html">Virtualization Support</a></li>
<li class="toctree-l3"><a class="reference internal" href="../hwmon/index.html">Hardware Monitoring</a></li>
<li class="toctree-l3"><a class="reference internal" href="../accel/index.html">Compute Accelerators</a></li>
<li class="toctree-l3"><a class="reference internal" href="../security/index.html">Security Documentation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../crypto/index.html">Crypto API</a></li>
<li class="toctree-l3"><a class="reference internal" href="../bpf/index.html">BPF Documentation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../usb/index.html">USB support</a></li>
<li class="toctree-l3 current"><a class="reference internal" href="index.html">PCI Bus Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../misc-devices/index.html">Assorted Miscellaneous Devices Documentation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../peci/index.html">PECI Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../wmi/index.html">WMI Subsystem</a></li>
<li class="toctree-l3"><a class="reference internal" href="../tee/index.html">TEE Subsystem</a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../locking/index.html">Locking</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../process/license-rules.html">Licensing rules</a></li>
<li class="toctree-l1"><a class="reference internal" href="../doc-guide/index.html">Writing documentation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../dev-tools/index.html">Development tools</a></li>
<li class="toctree-l1"><a class="reference internal" href="../dev-tools/testing-overview.html">Testing guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="../kernel-hacking/index.html">Hacking guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="../trace/index.html">Tracing</a></li>
<li class="toctree-l1"><a class="reference internal" href="../fault-injection/index.html">Fault injection</a></li>
<li class="toctree-l1"><a class="reference internal" href="../livepatch/index.html">Livepatching</a></li>
<li class="toctree-l1"><a class="reference internal" href="../rust/index.html">Rust</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../admin-guide/index.html">Administration</a></li>
<li class="toctree-l1"><a class="reference internal" href="../kbuild/index.html">Build system</a></li>
<li class="toctree-l1"><a class="reference internal" href="../admin-guide/reporting-issues.html">Reporting issues</a></li>
<li class="toctree-l1"><a class="reference internal" href="../tools/index.html">Userspace tools</a></li>
<li class="toctree-l1"><a class="reference internal" href="../userspace-api/index.html">Userspace API</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../firmware-guide/index.html">Firmware</a></li>
<li class="toctree-l1"><a class="reference internal" href="../devicetree/index.html">Firmware and Devicetree</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../arch/index.html">CPU architectures</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../staging/index.html">Unsorted documentation</a></li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../translations/index.html">Translations</a></li>
</ul>
</div>
<script type="text/javascript"> <!--
var sbar = document.getElementsByClassName("sphinxsidebar")[0];
let currents = document.getElementsByClassName("current")
if (currents.length) {
sbar.scrollTop = currents[currents.length - 1].offsetTop;
}
--> </script>
<div role="note" aria-label="source link">
<h3>This Page</h3>
<ul class="this-page-menu">
<li><a href="../_sources/PCI/pci.rst.txt"
rel="nofollow">Show Source</a></li>
</ul>
</div>
</div>
</div>
<div class="documentwrapper">
<div class="bodywrapper">
<div class="body" role="main">
<div class="language-selection">
English
<ul>
<li><a href="../translations/zh_CN/PCI/pci.html">Chinese (Simplified)</a></li>
</ul>
</div>
<section id="how-to-write-linux-pci-drivers">
<h1><span class="section-number">1. </span>How To Write Linux PCI Drivers<a class="headerlink" href="#how-to-write-linux-pci-drivers" title="Link to this heading">¶</a></h1>
<dl class="field-list simple">
<dt class="field-odd">Authors<span class="colon">:</span></dt>
<dd class="field-odd"><ul class="simple">
<li><p>Martin Mares <<a class="reference external" href="mailto:mj%40ucw.cz">mj<span>@</span>ucw<span>.</span>cz</a>></p></li>
<li><p>Grant Grundler <<a class="reference external" href="mailto:grundler%40parisc-linux.org">grundler<span>@</span>parisc-linux<span>.</span>org</a>></p></li>
</ul>
</dd>
</dl>
<p>The world of PCI is vast and full of (mostly unpleasant) surprises.
Since each CPU architecture implements different chip-sets and PCI devices
have different requirements (erm, “features”), the result is the PCI support
in the Linux kernel is not as trivial as one would wish. This short paper
tries to introduce all potential driver authors to Linux APIs for
PCI device drivers.</p>
<p>A more complete resource is the third edition of “Linux Device Drivers”
by Jonathan Corbet, Alessandro Rubini, and Greg Kroah-Hartman.
LDD3 is available for free (under Creative Commons License) from:
<a class="reference external" href="https://lwn.net/Kernel/LDD3/">https://lwn.net/Kernel/LDD3/</a>.</p>
<p>However, keep in mind that all documents are subject to “bit rot”.
Refer to the source code if things are not working as described here.</p>
<p>Please send questions/comments/patches about Linux PCI API to the
“Linux PCI” <<a class="reference external" href="mailto:linux-pci%40atrey.karlin.mff.cuni.cz">linux-pci<span>@</span>atrey<span>.</span>karlin<span>.</span>mff<span>.</span>cuni<span>.</span>cz</a>> mailing list.</p>
<section id="structure-of-pci-drivers">
<h2><span class="section-number">1.1. </span>Structure of PCI drivers<a class="headerlink" href="#structure-of-pci-drivers" title="Link to this heading">¶</a></h2>
<p>PCI drivers “discover” PCI devices in a system via <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_register_driver()</span></code>.
Actually, it’s the other way around. When the PCI generic code discovers
a new device, the driver with a matching “description” will be notified.
Details on this below.</p>
<p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_register_driver()</span></code> leaves most of the probing for devices to
the PCI layer and supports online insertion/removal of devices [thus
supporting hot-pluggable PCI, CardBus, and Express-Card in a single driver].
<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_register_driver()</span></code> call requires passing in a table of function
pointers and thus dictates the high level structure of a driver.</p>
<p>Once the driver knows about a PCI device and takes ownership, the
driver generally needs to perform the following initialization:</p>
<blockquote>
<div><ul class="simple">
<li><p>Enable the device</p></li>
<li><p>Request MMIO/IOP resources</p></li>
<li><p>Set the DMA mask size (for both coherent and streaming DMA)</p></li>
<li><p>Allocate and initialize shared control data (<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_allocate_coherent()</span></code>)</p></li>
<li><p>Access device configuration space (if needed)</p></li>
<li><p>Register IRQ handler (<a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a>)</p></li>
<li><p>Initialize non-PCI (i.e. LAN/SCSI/etc parts of the chip)</p></li>
<li><p>Enable DMA/processing engines</p></li>
</ul>
</div></blockquote>
<p>When done using the device, and perhaps the module needs to be unloaded,
the driver needs to take the following steps:</p>
<blockquote>
<div><ul class="simple">
<li><p>Disable the device from generating IRQs</p></li>
<li><p>Release the IRQ (<a class="reference internal" href="../core-api/genericirq.html#c.free_irq" title="free_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">free_irq()</span></code></a>)</p></li>
<li><p>Stop all DMA activity</p></li>
<li><p>Release DMA buffers (both streaming and coherent)</p></li>
<li><p>Unregister from other subsystems (e.g. scsi or netdev)</p></li>
<li><p>Release MMIO/IOP resources</p></li>
<li><p>Disable the device</p></li>
</ul>
</div></blockquote>
<p>Most of these topics are covered in the following sections.
For the rest look at LDD3 or <linux/pci.h> .</p>
<p>If the PCI subsystem is not configured (CONFIG_PCI is not set), most of
the PCI functions described below are defined as inline functions either
completely empty or just returning an appropriate error codes to avoid
lots of ifdefs in the drivers.</p>
</section>
<section id="pci-register-driver-call">
<h2><span class="section-number">1.2. </span>pci_register_driver() call<a class="headerlink" href="#pci-register-driver-call" title="Link to this heading">¶</a></h2>
<p>PCI device drivers call <code class="docutils literal notranslate"><span class="pre">pci_register_driver()</span></code> during their
initialization with a pointer to a structure describing the driver
(<code class="docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_driver</span></code>):</p>
<dl class="c struct">
<dt class="sig sig-object c" id="c.pci_driver">
<span class="k"><span class="pre">struct</span></span><span class="w"> </span><span class="sig-name descname"><span class="n"><span class="pre">pci_driver</span></span></span><a class="headerlink" href="#c.pci_driver" title="Link to this definition">¶</a><br /></dt>
<dd><p>PCI driver structure</p>
</dd></dl>
<div class="kernelindent docutils container">
<p><strong>Definition</strong>:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>struct pci_driver {
const char *name;
const struct pci_device_id *id_table;
int (*probe)(struct pci_dev *dev, const struct pci_device_id *id);
void (*remove)(struct pci_dev *dev);
int (*suspend)(struct pci_dev *dev, pm_message_t state);
int (*resume)(struct pci_dev *dev);
void (*shutdown)(struct pci_dev *dev);
int (*sriov_configure)(struct pci_dev *dev, int num_vfs);
int (*sriov_set_msix_vec_count)(struct pci_dev *vf, int msix_vec_count);
u32 (*sriov_get_vf_total_msix)(struct pci_dev *pf);
const struct pci_error_handlers *err_handler;
const struct attribute_group **groups;
const struct attribute_group **dev_groups;
struct device_driver driver;
struct pci_dynids dynids;
bool driver_managed_dma;
};
</pre></div>
</div>
<p><strong>Members</strong></p>
<dl class="simple">
<dt><code class="docutils literal notranslate"><span class="pre">name</span></code></dt><dd><p>Driver name.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">id_table</span></code></dt><dd><p>Pointer to table of device IDs the driver is
interested in. Most drivers should export this
table using MODULE_DEVICE_TABLE(pci,...).</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">probe</span></code></dt><dd><p>This probing function gets called (during execution
of <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_register_driver()</span></code> for already existing
devices or later if a new device gets inserted) for
all PCI devices which match the ID table and are not
“owned” by the other drivers yet. This function gets
passed a “<code class="xref c c-struct broken_xref docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_dev</span></code> *” for each device whose
entry in the ID table matches the device. The probe
function returns zero when the driver chooses to
take “ownership” of the device or an error code
(negative number) otherwise.
The probe function always gets called from process
context, so it can sleep.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">remove</span></code></dt><dd><p>The <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">remove()</span></code> function gets called whenever a device
being handled by this driver is removed (either during
deregistration of the driver or when it’s manually
pulled out of a hot-pluggable slot).
The remove function always gets called from process
context, so it can sleep.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">suspend</span></code></dt><dd><p>Put device into low power state.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">resume</span></code></dt><dd><p>Wake device from low power state.
(Please see <a class="reference internal" href="../power/pci.html"><span class="doc">PCI Power Management</span></a> for descriptions
of PCI Power Management and the related functions.)</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">shutdown</span></code></dt><dd><p>Hook into reboot_notifier_list (kernel/sys.c).
Intended to stop any idling DMA operations.
Useful for enabling wake-on-lan (NIC) or changing
the power state of a device before reboot.
e.g. drivers/net/e100.c.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">sriov_configure</span></code></dt><dd><p>Optional driver callback to allow configuration of
number of VFs to enable via sysfs “sriov_numvfs” file.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">sriov_set_msix_vec_count</span></code></dt><dd><p>PF Driver callback to change number of MSI-X
vectors on a VF. Triggered via sysfs “sriov_vf_msix_count”.
This will change MSI-X Table Size in the VF Message Control
registers.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">sriov_get_vf_total_msix</span></code></dt><dd><p>PF driver callback to get the total number of
MSI-X vectors available for distribution to the VFs.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">err_handler</span></code></dt><dd><p>See <a class="reference internal" href="pci-error-recovery.html"><span class="doc">PCI Error Recovery</span></a></p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">groups</span></code></dt><dd><p>Sysfs attribute groups.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">dev_groups</span></code></dt><dd><p>Attributes attached to the device that will be
created once it is bound to the driver.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">driver</span></code></dt><dd><p>Driver model structure.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">dynids</span></code></dt><dd><p>List of dynamically added device IDs.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">driver_managed_dma</span></code></dt><dd><p>Device driver doesn’t use kernel DMA API for DMA.
For most device drivers, no need to care about this flag
as long as all DMAs are handled through the kernel DMA API.
For some special ones, for example VFIO drivers, they know
how to manage the DMA themselves and set this flag so that
the IOMMU layer will allow them to setup and manage their
own I/O address space.</p>
</dd>
</dl>
</div>
<p>The ID table is an array of <code class="docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_device_id</span></code> entries ending with an
all-zero entry. Definitions with static const are generally preferred.</p>
<dl class="c struct">
<dt class="sig sig-object c" id="c.pci_device_id">
<span class="k"><span class="pre">struct</span></span><span class="w"> </span><span class="sig-name descname"><span class="n"><span class="pre">pci_device_id</span></span></span><a class="headerlink" href="#c.pci_device_id" title="Link to this definition">¶</a><br /></dt>
<dd><p>PCI device ID structure</p>
</dd></dl>
<div class="kernelindent docutils container">
<p><strong>Definition</strong>:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>struct pci_device_id {
__u32 vendor, device;
__u32 subvendor, subdevice;
__u32 class, class_mask;
kernel_ulong_t driver_data;
__u32 override_only;
};
</pre></div>
</div>
<p><strong>Members</strong></p>
<dl class="simple">
<dt><code class="docutils literal notranslate"><span class="pre">vendor</span></code></dt><dd><p>Vendor ID to match (or PCI_ANY_ID)</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">device</span></code></dt><dd><p>Device ID to match (or PCI_ANY_ID)</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">subvendor</span></code></dt><dd><p>Subsystem vendor ID to match (or PCI_ANY_ID)</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">subdevice</span></code></dt><dd><p>Subsystem device ID to match (or PCI_ANY_ID)</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">class</span></code></dt><dd><p>Device class, subclass, and “interface” to match.
See Appendix D of the PCI Local Bus Spec or
include/linux/pci_ids.h for a full list of classes.
Most drivers do not need to specify class/class_mask
as vendor/device is normally sufficient.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">class_mask</span></code></dt><dd><p>Limit which sub-fields of the class field are compared.
See drivers/scsi/sym53c8xx_2/ for example of usage.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">driver_data</span></code></dt><dd><p>Data private to the driver.
Most drivers don’t need to use driver_data field.
Best practice is to use driver_data as an index
into a static list of equivalent device types,
instead of using it as a pointer.</p>
</dd>
<dt><code class="docutils literal notranslate"><span class="pre">override_only</span></code></dt><dd><p>Match only when dev->driver_override is this driver.</p>
</dd>
</dl>
</div>
<p>Most drivers only need <code class="docutils literal notranslate"><span class="pre">PCI_DEVICE()</span></code> or <code class="docutils literal notranslate"><span class="pre">PCI_DEVICE_CLASS()</span></code> to set up
a pci_device_id table.</p>
<p>New PCI IDs may be added to a device driver pci_ids table at runtime
as shown below:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>echo "vendor device subvendor subdevice class class_mask driver_data" > \
/sys/bus/pci/drivers/{driver}/new_id
</pre></div>
</div>
<p>All fields are passed in as hexadecimal values (no leading 0x).
The vendor and device fields are mandatory, the others are optional. Users
need pass only as many optional fields as necessary:</p>
<blockquote>
<div><ul class="simple">
<li><p>subvendor and subdevice fields default to PCI_ANY_ID (FFFFFFFF)</p></li>
<li><p>class and classmask fields default to 0</p></li>
<li><p>driver_data defaults to 0UL.</p></li>
<li><p>override_only field defaults to 0.</p></li>
</ul>
</div></blockquote>
<p>Note that driver_data must match the value used by any of the pci_device_id
entries defined in the driver. This makes the driver_data field mandatory
if all the pci_device_id entries have a non-zero driver_data value.</p>
<p>Once added, the driver probe routine will be invoked for any unclaimed
PCI devices listed in its (newly updated) pci_ids list.</p>
<p>When the driver exits, it just calls <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_unregister_driver" title="pci_unregister_driver"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_unregister_driver()</span></code></a> and the PCI layer
automatically calls the remove hook for all devices handled by the driver.</p>
<section id="attributes-for-driver-functions-data">
<h3><span class="section-number">1.2.1. </span>“Attributes” for driver functions/data<a class="headerlink" href="#attributes-for-driver-functions-data" title="Link to this heading">¶</a></h3>
<p>Please mark the initialization and cleanup functions where appropriate
(the corresponding macros are defined in <linux/init.h>):</p>
<blockquote>
<div><table class="docutils align-default">
<tbody>
<tr class="row-odd"><td><p>__init</p></td>
<td><p>Initialization code. Thrown away after the driver
initializes.</p></td>
</tr>
<tr class="row-even"><td><p>__exit</p></td>
<td><p>Exit code. Ignored for non-modular drivers.</p></td>
</tr>
</tbody>
</table>
</div></blockquote>
<dl class="simple">
<dt>Tips on when/where to use the above attributes:</dt><dd><ul class="simple">
<li><p>The <a class="reference internal" href="../driver-api/basics.html#c.module_init" title="module_init"><code class="xref c c-func docutils literal notranslate"><span class="pre">module_init()</span></code></a>/<a class="reference internal" href="../driver-api/basics.html#c.module_exit" title="module_exit"><code class="xref c c-func docutils literal notranslate"><span class="pre">module_exit()</span></code></a> functions (and all
initialization functions called _only_ from these)
should be marked __init/__exit.</p></li>
<li><p>Do not mark the <a class="reference internal" href="#c.pci_driver" title="pci_driver"><code class="xref c c-struct docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_driver</span></code></a>.</p></li>
<li><p>Do NOT mark a function if you are not sure which mark to use.
Better to not mark the function than mark the function wrong.</p></li>
</ul>
</dd>
</dl>
</section>
</section>
<section id="how-to-find-pci-devices-manually">
<h2><span class="section-number">1.3. </span>How to find PCI devices manually<a class="headerlink" href="#how-to-find-pci-devices-manually" title="Link to this heading">¶</a></h2>
<p>PCI drivers should have a really good reason for not using the
<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_register_driver()</span></code> interface to search for PCI devices.
The main reason PCI devices are controlled by multiple drivers
is because one PCI device implements several different HW services.
E.g. combined serial/parallel port/floppy controller.</p>
<p>A manual search may be performed using the following constructs:</p>
<p>Searching by vendor and device ID:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>struct pci_dev *dev = NULL;
while (dev = pci_get_device(VENDOR_ID, DEVICE_ID, dev))
configure_device(dev);
</pre></div>
</div>
<p>Searching by class ID (iterate in a similar way):</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>pci_get_class(CLASS_ID, dev)
</pre></div>
</div>
<p>Searching by both vendor/device and subsystem vendor/device ID:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>pci_get_subsys(VENDOR_ID,DEVICE_ID, SUBSYS_VENDOR_ID, SUBSYS_DEVICE_ID, dev).
</pre></div>
</div>
<p>You can use the constant PCI_ANY_ID as a wildcard replacement for
VENDOR_ID or DEVICE_ID. This allows searching for any device from a
specific vendor, for example.</p>
<p>These functions are hotplug-safe. They increment the reference count on
the pci_dev that they return. You must eventually (possibly at module unload)
decrement the reference count on these devices by calling <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_dev_put" title="pci_dev_put"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_dev_put()</span></code></a>.</p>
</section>
<section id="device-initialization-steps">
<h2><span class="section-number">1.4. </span>Device Initialization Steps<a class="headerlink" href="#device-initialization-steps" title="Link to this heading">¶</a></h2>
<p>As noted in the introduction, most PCI drivers need the following steps
for device initialization:</p>
<blockquote>
<div><ul class="simple">
<li><p>Enable the device</p></li>
<li><p>Request MMIO/IOP resources</p></li>
<li><p>Set the DMA mask size (for both coherent and streaming DMA)</p></li>
<li><p>Allocate and initialize shared control data (<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_allocate_coherent()</span></code>)</p></li>
<li><p>Access device configuration space (if needed)</p></li>
<li><p>Register IRQ handler (<a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a>)</p></li>
<li><p>Initialize non-PCI (i.e. LAN/SCSI/etc parts of the chip)</p></li>
<li><p>Enable DMA/processing engines.</p></li>
</ul>
</div></blockquote>
<p>The driver can access PCI config space registers at any time.
(Well, almost. When running BIST, config space can go away...but
that will just result in a PCI Bus Master Abort and config reads
will return garbage).</p>
<section id="enable-the-pci-device">
<h3><span class="section-number">1.4.1. </span>Enable the PCI device<a class="headerlink" href="#enable-the-pci-device" title="Link to this heading">¶</a></h3>
<p>Before touching any device registers, the driver needs to enable
the PCI device by calling <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_enable_device" title="pci_enable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_enable_device()</span></code></a>. This will:</p>
<blockquote>
<div><ul class="simple">
<li><p>wake up the device if it was in suspended state,</p></li>
<li><p>allocate I/O and memory regions of the device (if BIOS did not),</p></li>
<li><p>allocate an IRQ (if BIOS did not).</p></li>
</ul>
</div></blockquote>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_enable_device" title="pci_enable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_enable_device()</span></code></a> can fail! Check the return value.</p>
</div>
<div class="admonition warning">
<p class="admonition-title">Warning</p>
<p>OS BUG: we don’t check resource allocations before enabling those
resources. The sequence would make more sense if we called
<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_request_resources()</span></code> before calling <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_enable_device" title="pci_enable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_enable_device()</span></code></a>.
Currently, the device drivers can’t detect the bug when two
devices have been allocated the same range. This is not a common
problem and unlikely to get fixed soon.</p>
<p>This has been discussed before but not changed as of 2.6.19:
<a class="reference external" href="https://lore.kernel.org/r/20060302180025.GC28895@flint.arm.linux.org.uk/">https://lore.kernel.org/r/20060302180025.GC28895@flint.arm.linux.org.uk/</a></p>
</div>
<p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_set_master" title="pci_set_master"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_set_master()</span></code></a> will enable DMA by setting the bus master bit
in the PCI_COMMAND register. It also fixes the latency timer value if
it’s set to something bogus by the BIOS. <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_clear_master" title="pci_clear_master"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_clear_master()</span></code></a> will
disable DMA by clearing the bus master bit.</p>
<p>If the PCI device can use the PCI Memory-Write-Invalidate transaction,
call <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_set_mwi" title="pci_set_mwi"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_set_mwi()</span></code></a>. This enables the PCI_COMMAND bit for Mem-Wr-Inval
and also ensures that the cache line size register is set correctly.
Check the return value of <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_set_mwi" title="pci_set_mwi"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_set_mwi()</span></code></a> as not all architectures
or chip-sets may support Memory-Write-Invalidate. Alternatively,
if Mem-Wr-Inval would be nice to have but is not required, call
<a class="reference internal" href="../driver-api/pci/pci.html#c.pci_try_set_mwi" title="pci_try_set_mwi"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_try_set_mwi()</span></code></a> to have the system do its best effort at enabling
Mem-Wr-Inval.</p>
</section>
<section id="request-mmio-iop-resources">
<h3><span class="section-number">1.4.2. </span>Request MMIO/IOP resources<a class="headerlink" href="#request-mmio-iop-resources" title="Link to this heading">¶</a></h3>
<p>Memory (MMIO), and I/O port addresses should NOT be read directly
from the PCI device config space. Use the values in the pci_dev structure
as the PCI “bus address” might have been remapped to a “host physical”
address by the arch/chip-set specific kernel support.</p>
<p>See <a class="reference internal" href="../driver-api/io-mapping.html"><span class="doc">The io_mapping functions</span></a> for how to access device registers
or device memory.</p>
<p>The device driver needs to call <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_request_region" title="pci_request_region"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_request_region()</span></code></a> to verify
no other device is already using the same address resource.
Conversely, drivers should call <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_release_region" title="pci_release_region"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_release_region()</span></code></a> AFTER
calling <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_disable_device" title="pci_disable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_disable_device()</span></code></a>.
The idea is to prevent two devices colliding on the same address range.</p>
<div class="admonition tip">
<p class="admonition-title">Tip</p>
<p>See OS BUG comment above. Currently (2.6.19), The driver can only
determine MMIO and IO Port resource availability _after_ calling
<a class="reference internal" href="../driver-api/pci/pci.html#c.pci_enable_device" title="pci_enable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_enable_device()</span></code></a>.</p>
</div>
<p>Generic flavors of <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_request_region" title="pci_request_region"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_request_region()</span></code></a> are <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">request_mem_region()</span></code>
(for MMIO ranges) and <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">request_region()</span></code> (for IO Port ranges).
Use these for address resources that are not described by “normal” PCI
BARs.</p>
<p>Also see <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_request_selected_regions" title="pci_request_selected_regions"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_request_selected_regions()</span></code></a> below.</p>
</section>
<section id="set-the-dma-mask-size">
<h3><span class="section-number">1.4.3. </span>Set the DMA mask size<a class="headerlink" href="#set-the-dma-mask-size" title="Link to this heading">¶</a></h3>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>If anything below doesn’t make sense, please refer to
<a class="reference internal" href="../core-api/dma-api.html"><span class="doc">Dynamic DMA mapping using the generic device</span></a>. This section is just a reminder that
drivers need to indicate DMA capabilities of the device and is not
an authoritative source for DMA interfaces.</p>
</div>
<p>While all drivers should explicitly indicate the DMA capability
(e.g. 32 or 64 bit) of the PCI bus master, devices with more than
32-bit bus master capability for streaming data need the driver
to “register” this capability by calling <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">dma_set_mask()</span></code> with
appropriate parameters. In general this allows more efficient DMA
on systems where System RAM exists above 4G _physical_ address.</p>
<p>Drivers for all PCI-X and PCIe compliant devices must call
<code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">dma_set_mask()</span></code> as they are 64-bit DMA devices.</p>
<p>Similarly, drivers must also “register” this capability if the device
can directly address “coherent memory” in System RAM above 4G physical
address by calling <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">dma_set_coherent_mask()</span></code>.
Again, this includes drivers for all PCI-X and PCIe compliant devices.
Many 64-bit “PCI” devices (before PCI-X) and some PCI-X devices are
64-bit DMA capable for payload (“streaming”) data but not control
(“coherent”) data.</p>
</section>
<section id="setup-shared-control-data">
<h3><span class="section-number">1.4.4. </span>Setup shared control data<a class="headerlink" href="#setup-shared-control-data" title="Link to this heading">¶</a></h3>
<p>Once the DMA masks are set, the driver can allocate “coherent” (a.k.a. shared)
memory. See <a class="reference internal" href="../core-api/dma-api.html"><span class="doc">Dynamic DMA mapping using the generic device</span></a> for a full description of
the DMA APIs. This section is just a reminder that it needs to be done
before enabling DMA on the device.</p>
</section>
<section id="initialize-device-registers">
<h3><span class="section-number">1.4.5. </span>Initialize device registers<a class="headerlink" href="#initialize-device-registers" title="Link to this heading">¶</a></h3>
<p>Some drivers will need specific “capability” fields programmed
or other “vendor specific” register initialized or reset.
E.g. clearing pending interrupts.</p>
</section>
<section id="register-irq-handler">
<h3><span class="section-number">1.4.6. </span>Register IRQ handler<a class="headerlink" href="#register-irq-handler" title="Link to this heading">¶</a></h3>
<p>While calling <a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a> is the last step described here,
this is often just another intermediate step to initialize a device.
This step can often be deferred until the device is opened for use.</p>
<p>All interrupt handlers for IRQ lines should be registered with IRQF_SHARED
and use the devid to map IRQs to devices (remember that all PCI IRQ lines
can be shared).</p>
<p><a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a> will associate an interrupt handler and device handle
with an interrupt number. Historically interrupt numbers represent
IRQ lines which run from the PCI device to the Interrupt controller.
With MSI and MSI-X (more below) the interrupt number is a CPU “vector”.</p>
<p><a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a> also enables the interrupt. Make sure the device is
quiesced and does not have any interrupts pending before registering
the interrupt handler.</p>
<p>MSI and MSI-X are PCI capabilities. Both are “Message Signaled Interrupts”
which deliver interrupts to the CPU via a DMA write to a Local APIC.
The fundamental difference between MSI and MSI-X is how multiple
“vectors” get allocated. MSI requires contiguous blocks of vectors
while MSI-X can allocate several individual ones.</p>
<p>MSI capability can be enabled by calling <a class="reference internal" href="msi-howto.html#c.pci_alloc_irq_vectors" title="pci_alloc_irq_vectors"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_alloc_irq_vectors()</span></code></a> with the
PCI_IRQ_MSI and/or PCI_IRQ_MSIX flags before calling <a class="reference internal" href="../core-api/genericirq.html#c.request_irq" title="request_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">request_irq()</span></code></a>. This
causes the PCI support to program CPU vector data into the PCI device
capability registers. Many architectures, chip-sets, or BIOSes do NOT
support MSI or MSI-X and a call to pci_alloc_irq_vectors with just
the PCI_IRQ_MSI and PCI_IRQ_MSIX flags will fail, so try to always
specify PCI_IRQ_INTX as well.</p>
<p>Drivers that have different interrupt handlers for MSI/MSI-X and
legacy INTx should chose the right one based on the msi_enabled
and msix_enabled flags in the pci_dev structure after calling
pci_alloc_irq_vectors.</p>
<p>There are (at least) two really good reasons for using MSI:</p>
<ol class="arabic simple">
<li><p>MSI is an exclusive interrupt vector by definition.
This means the interrupt handler doesn’t have to verify
its device caused the interrupt.</p></li>
<li><p>MSI avoids DMA/IRQ race conditions. DMA to host memory is guaranteed
to be visible to the host CPU(s) when the MSI is delivered. This
is important for both data coherency and avoiding stale control data.
This guarantee allows the driver to omit MMIO reads to flush
the DMA stream.</p></li>
</ol>
<p>See drivers/infiniband/hw/mthca/ or drivers/net/tg3.c for examples
of MSI/MSI-X usage.</p>
</section>
</section>
<section id="pci-device-shutdown">
<h2><span class="section-number">1.5. </span>PCI device shutdown<a class="headerlink" href="#pci-device-shutdown" title="Link to this heading">¶</a></h2>
<p>When a PCI device driver is being unloaded, most of the following
steps need to be performed:</p>
<blockquote>
<div><ul class="simple">
<li><p>Disable the device from generating IRQs</p></li>
<li><p>Release the IRQ (<a class="reference internal" href="../core-api/genericirq.html#c.free_irq" title="free_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">free_irq()</span></code></a>)</p></li>
<li><p>Stop all DMA activity</p></li>
<li><p>Release DMA buffers (both streaming and coherent)</p></li>
<li><p>Unregister from other subsystems (e.g. scsi or netdev)</p></li>
<li><p>Disable device from responding to MMIO/IO Port addresses</p></li>
<li><p>Release MMIO/IO Port resource(s)</p></li>
</ul>
</div></blockquote>
<section id="stop-irqs-on-the-device">
<h3><span class="section-number">1.5.1. </span>Stop IRQs on the device<a class="headerlink" href="#stop-irqs-on-the-device" title="Link to this heading">¶</a></h3>
<p>How to do this is chip/device specific. If it’s not done, it opens
the possibility of a “screaming interrupt” if (and only if)
the IRQ is shared with another device.</p>
<p>When the shared IRQ handler is “unhooked”, the remaining devices
using the same IRQ line will still need the IRQ enabled. Thus if the
“unhooked” device asserts IRQ line, the system will respond assuming
it was one of the remaining devices asserted the IRQ line. Since none
of the other devices will handle the IRQ, the system will “hang” until
it decides the IRQ isn’t going to get handled and masks the IRQ (100,000
iterations later). Once the shared IRQ is masked, the remaining devices
will stop functioning properly. Not a nice situation.</p>
<p>This is another reason to use MSI or MSI-X if it’s available.
MSI and MSI-X are defined to be exclusive interrupts and thus
are not susceptible to the “screaming interrupt” problem.</p>
</section>
<section id="release-the-irq">
<h3><span class="section-number">1.5.2. </span>Release the IRQ<a class="headerlink" href="#release-the-irq" title="Link to this heading">¶</a></h3>
<p>Once the device is quiesced (no more IRQs), one can call <a class="reference internal" href="../core-api/genericirq.html#c.free_irq" title="free_irq"><code class="xref c c-func docutils literal notranslate"><span class="pre">free_irq()</span></code></a>.
This function will return control once any pending IRQs are handled,
“unhook” the drivers IRQ handler from that IRQ, and finally release
the IRQ if no one else is using it.</p>
</section>
<section id="stop-all-dma-activity">
<h3><span class="section-number">1.5.3. </span>Stop all DMA activity<a class="headerlink" href="#stop-all-dma-activity" title="Link to this heading">¶</a></h3>
<p>It’s extremely important to stop all DMA operations BEFORE attempting
to deallocate DMA control data. Failure to do so can result in memory
corruption, hangs, and on some chip-sets a hard crash.</p>
<p>Stopping DMA after stopping the IRQs can avoid races where the
IRQ handler might restart DMA engines.</p>
<p>While this step sounds obvious and trivial, several “mature” drivers
didn’t get this step right in the past.</p>
</section>
<section id="release-dma-buffers">
<h3><span class="section-number">1.5.4. </span>Release DMA buffers<a class="headerlink" href="#release-dma-buffers" title="Link to this heading">¶</a></h3>
<p>Once DMA is stopped, clean up streaming DMA first.
I.e. unmap data buffers and return buffers to “upstream”
owners if there is one.</p>
<p>Then clean up “coherent” buffers which contain the control data.</p>
<p>See <a class="reference internal" href="../core-api/dma-api.html"><span class="doc">Dynamic DMA mapping using the generic device</span></a> for details on unmapping interfaces.</p>
</section>
<section id="unregister-from-other-subsystems">
<h3><span class="section-number">1.5.5. </span>Unregister from other subsystems<a class="headerlink" href="#unregister-from-other-subsystems" title="Link to this heading">¶</a></h3>
<p>Most low level PCI device drivers support some other subsystem
like USB, ALSA, SCSI, NetDev, Infiniband, etc. Make sure your
driver isn’t losing resources from that other subsystem.
If this happens, typically the symptom is an Oops (panic) when
the subsystem attempts to call into a driver that has been unloaded.</p>
</section>
<section id="disable-device-from-responding-to-mmio-io-port-addresses">
<h3><span class="section-number">1.5.6. </span>Disable Device from responding to MMIO/IO Port addresses<a class="headerlink" href="#disable-device-from-responding-to-mmio-io-port-addresses" title="Link to this heading">¶</a></h3>
<p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">io_unmap()</span></code> MMIO or IO Port resources and then call <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_disable_device" title="pci_disable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_disable_device()</span></code></a>.
This is the symmetric opposite of <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_enable_device" title="pci_enable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_enable_device()</span></code></a>.
Do not access device registers after calling <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_disable_device" title="pci_disable_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_disable_device()</span></code></a>.</p>
</section>
<section id="release-mmio-io-port-resource-s">
<h3><span class="section-number">1.5.7. </span>Release MMIO/IO Port Resource(s)<a class="headerlink" href="#release-mmio-io-port-resource-s" title="Link to this heading">¶</a></h3>
<p>Call <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_release_region" title="pci_release_region"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_release_region()</span></code></a> to mark the MMIO or IO Port range as available.
Failure to do so usually results in the inability to reload the driver.</p>
</section>
</section>
<section id="how-to-access-pci-config-space">
<h2><span class="section-number">1.6. </span>How to access PCI config space<a class="headerlink" href="#how-to-access-pci-config-space" title="Link to this heading">¶</a></h2>
<p>You can use <cite>pci_(read|write)_config_(byte|word|dword)</cite> to access the config
space of a device represented by <cite><code class="xref c c-struct broken_xref docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_dev</span></code> *</cite>. All these functions return
0 when successful or an error code (<cite>PCIBIOS_...</cite>) which can be translated to a
text string by pcibios_strerror. Most drivers expect that accesses to valid PCI
devices don’t fail.</p>
<p>If you don’t have a <code class="xref c c-struct broken_xref docutils literal notranslate"><span class="pre">struct</span> <span class="pre">pci_dev</span></code> available, you can call
<cite>pci_bus_(read|write)_config_(byte|word|dword)</cite> to access a given device
and function on that bus.</p>
<p>If you access fields in the standard portion of the config header, please
use symbolic names of locations and bits declared in <linux/pci.h>.</p>
<p>If you need to access Extended PCI Capability registers, just call
<a class="reference internal" href="../driver-api/pci/pci.html#c.pci_find_capability" title="pci_find_capability"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_find_capability()</span></code></a> for the particular capability and it will find the
corresponding register block for you.</p>
</section>
<section id="other-interesting-functions">
<h2><span class="section-number">1.7. </span>Other interesting functions<a class="headerlink" href="#other-interesting-functions" title="Link to this heading">¶</a></h2>
<table class="docutils align-default">
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_domain_bus_and_slot" title="pci_get_domain_bus_and_slot"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_domain_bus_and_slot()</span></code></a></p></td>
<td><p>Find pci_dev corresponding to given domain,
bus and slot and number. If the device is
found, its reference count is increased.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_set_power_state" title="pci_set_power_state"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_set_power_state()</span></code></a></p></td>
<td><p>Set PCI Power Management state (0=D0 ... 3=D3)</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_find_capability" title="pci_find_capability"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_find_capability()</span></code></a></p></td>
<td><p>Find specified capability in device’s capability
list.</p></td>
</tr>
<tr class="row-even"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_resource_start()</span></code></p></td>
<td><p>Returns bus start address for a given PCI region</p></td>
</tr>
<tr class="row-odd"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_resource_end()</span></code></p></td>
<td><p>Returns bus end address for a given PCI region</p></td>
</tr>
<tr class="row-even"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_resource_len()</span></code></p></td>
<td><p>Returns the byte length of a PCI region</p></td>
</tr>
<tr class="row-odd"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_set_drvdata()</span></code></p></td>
<td><p>Set private driver data pointer for a pci_dev</p></td>
</tr>
<tr class="row-even"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_get_drvdata()</span></code></p></td>
<td><p>Return private driver data pointer for a pci_dev</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_set_mwi" title="pci_set_mwi"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_set_mwi()</span></code></a></p></td>
<td><p>Enable Memory-Write-Invalidate transactions.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_clear_mwi" title="pci_clear_mwi"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_clear_mwi()</span></code></a></p></td>
<td><p>Disable Memory-Write-Invalidate transactions.</p></td>
</tr>
</tbody>
</table>
</section>
<section id="miscellaneous-hints">
<h2><span class="section-number">1.8. </span>Miscellaneous hints<a class="headerlink" href="#miscellaneous-hints" title="Link to this heading">¶</a></h2>
<p>When displaying PCI device names to the user (for example when a driver wants
to tell the user what card has it found), please use pci_name(pci_dev).</p>
<p>Always refer to the PCI devices by a pointer to the pci_dev structure.
All PCI layer functions use this identification and it’s the only
reasonable one. Don’t use bus/slot/function numbers except for very
special purposes -- on systems with multiple primary buses their semantics
can be pretty complex.</p>
<p>Don’t try to turn on Fast Back to Back writes in your driver. All devices
on the bus need to be capable of doing it, so this is something which needs
to be handled by platform and generic code, not individual drivers.</p>
</section>
<section id="vendor-and-device-identifications">
<h2><span class="section-number">1.9. </span>Vendor and device identifications<a class="headerlink" href="#vendor-and-device-identifications" title="Link to this heading">¶</a></h2>
<p>Do not add new device or vendor IDs to include/linux/pci_ids.h unless they
are shared across multiple drivers. You can add private definitions in
your driver if they’re helpful, or just use plain hex constants.</p>
<p>The device IDs are arbitrary hex numbers (vendor controlled) and normally used
only in a single location, the pci_device_id table.</p>
<p>Please DO submit new vendor/device IDs to <a class="reference external" href="https://pci-ids.ucw.cz/">https://pci-ids.ucw.cz/</a>.
There’s a mirror of the pci.ids file at <a class="reference external" href="https://github.com/pciutils/pciids">https://github.com/pciutils/pciids</a>.</p>
</section>
<section id="obsolete-functions">
<h2><span class="section-number">1.10. </span>Obsolete functions<a class="headerlink" href="#obsolete-functions" title="Link to this heading">¶</a></h2>
<p>There are several functions which you might come across when trying to
port an old driver to the new PCI interface. They are no longer present
in the kernel as they aren’t compatible with hotplug or PCI domains or
having sane locking.</p>
<table class="docutils align-default">
<tbody>
<tr class="row-odd"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_find_device()</span></code></p></td>
<td><p>Superseded by <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_device" title="pci_get_device"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_device()</span></code></a></p></td>
</tr>
<tr class="row-even"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_find_subsys()</span></code></p></td>
<td><p>Superseded by <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_subsys" title="pci_get_subsys"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_subsys()</span></code></a></p></td>
</tr>
<tr class="row-odd"><td><p><code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">pci_find_slot()</span></code></p></td>
<td><p>Superseded by <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_domain_bus_and_slot" title="pci_get_domain_bus_and_slot"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_domain_bus_and_slot()</span></code></a></p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_slot" title="pci_get_slot"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_slot()</span></code></a></p></td>
<td><p>Superseded by <a class="reference internal" href="../driver-api/pci/pci.html#c.pci_get_domain_bus_and_slot" title="pci_get_domain_bus_and_slot"><code class="xref c c-func docutils literal notranslate"><span class="pre">pci_get_domain_bus_and_slot()</span></code></a></p></td>
</tr>
</tbody>
</table>
<p>The alternative is the traditional PCI device driver that walks PCI
device lists. This is still possible but discouraged.</p>
</section>
<section id="mmio-space-and-write-posting">
<h2><span class="section-number">1.11. </span>MMIO Space and “Write Posting”<a class="headerlink" href="#mmio-space-and-write-posting" title="Link to this heading">¶</a></h2>
<p>Converting a driver from using I/O Port space to using MMIO space
often requires some additional changes. Specifically, “write posting”
needs to be handled. Many drivers (e.g. tg3, acenic, sym53c8xx_2)
already do this. I/O Port space guarantees write transactions reach the PCI
device before the CPU can continue. Writes to MMIO space allow the CPU
to continue before the transaction reaches the PCI device. HW weenies
call this “Write Posting” because the write completion is “posted” to
the CPU before the transaction has reached its destination.</p>
<p>Thus, timing sensitive code should add <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">readl()</span></code> where the CPU is
expected to wait before doing other work. The classic “bit banging”
sequence works fine for I/O Port space:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>for (i = 8; --i; val >>= 1) {
outb(val & 1, ioport_reg); /* write bit */
udelay(10);
}
</pre></div>
</div>
<p>The same sequence for MMIO space should be:</p>
<div class="highlight-none notranslate"><div class="highlight"><pre><span></span>for (i = 8; --i; val >>= 1) {
writeb(val & 1, mmio_reg); /* write bit */
readb(safe_mmio_reg); /* flush posted write */
udelay(10);
}
</pre></div>
</div>
<p>It is important that “safe_mmio_reg” not have any side effects that
interferes with the correct operation of the device.</p>
<p>Another case to watch out for is when resetting a PCI device. Use PCI
Configuration space reads to flush the <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">writel()</span></code>. This will gracefully
handle the PCI master abort on all platforms if the PCI device is
expected to not respond to a <code class="xref c c-func broken_xref docutils literal notranslate"><span class="pre">readl()</span></code>. Most x86 platforms will allow
MMIO reads to master abort (a.k.a. “Soft Fail”) and return garbage
(e.g. ~0). But many RISC platforms will crash (a.k.a.”Hard Fail”).</p>
</section>
</section>
</div>
</div>
</div>
<div class="clearer"></div>
</div>
<div class="footer">
©The kernel development community.
|
Powered by <a href="https://www.sphinx-doc.org/">Sphinx 8.1.3</a>
& <a href="https://alabaster.readthedocs.io">Alabaster 0.7.16</a>
|
<a href="../_sources/PCI/pci.rst.txt"
rel="nofollow">Page source</a>
</div>
</body>
</html>