About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Documentation / DocBook / drm.tmpl

Custom Search

Based on kernel version 3.15.4. Page generated on 2014-07-07 09:02 EST.

1	<?xml version="1.0" encoding="UTF-8"?>
2	<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
3		"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd" []>
5	<book id="drmDevelopersGuide">
6	  <bookinfo>
7	    <title>Linux DRM Developer's Guide</title>
9	    <authorgroup>
10	      <author>
11		<firstname>Jesse</firstname>
12		<surname>Barnes</surname>
13		<contrib>Initial version</contrib>
14		<affiliation>
15		  <orgname>Intel Corporation</orgname>
16		  <address>
17		    <email>jesse.barnes@intel.com</email>
18		  </address>
19		</affiliation>
20	      </author>
21	      <author>
22		<firstname>Laurent</firstname>
23		<surname>Pinchart</surname>
24		<contrib>Driver internals</contrib>
25		<affiliation>
26		  <orgname>Ideas on board SPRL</orgname>
27		  <address>
28		    <email>laurent.pinchart@ideasonboard.com</email>
29		  </address>
30		</affiliation>
31	      </author>
32	      <author>
33		<firstname>Daniel</firstname>
34		<surname>Vetter</surname>
35		<contrib>Contributions all over the place</contrib>
36		<affiliation>
37		  <orgname>Intel Corporation</orgname>
38		  <address>
39		    <email>daniel.vetter@ffwll.ch</email>
40		  </address>
41		</affiliation>
42	      </author>
43	    </authorgroup>
45	    <copyright>
46	      <year>2008-2009</year>
47	      <year>2013-2014</year>
48	      <holder>Intel Corporation</holder>
49	    </copyright>
50	    <copyright>
51	      <year>2012</year>
52	      <holder>Laurent Pinchart</holder>
53	    </copyright>
55	    <legalnotice>
56	      <para>
57		The contents of this file may be used under the terms of the GNU
58		General Public License version 2 (the "GPL") as distributed in
59		the kernel source COPYING file.
60	      </para>
61	    </legalnotice>
63	    <revhistory>
64	      <!-- Put document revisions here, newest first. -->
65	      <revision>
66		<revnumber>1.0</revnumber>
67		<date>2012-07-13</date>
68		<authorinitials>LP</authorinitials>
69		<revremark>Added extensive documentation about driver internals.
70		</revremark>
71	      </revision>
72	    </revhistory>
73	  </bookinfo>
75	<toc></toc>
77	<part id="drmCore">
78	  <title>DRM Core</title>
79	  <partintro>
80	    <para>
81	      This first part of the DRM Developer's Guide documents core DRM code,
82	      helper libraries for writing drivers and generic userspace interfaces
83	      exposed by DRM drivers.
84	    </para>
85	  </partintro>
87	  <chapter id="drmIntroduction">
88	    <title>Introduction</title>
89	    <para>
90	      The Linux DRM layer contains code intended to support the needs
91	      of complex graphics devices, usually containing programmable
92	      pipelines well suited to 3D graphics acceleration.  Graphics
93	      drivers in the kernel may make use of DRM functions to make
94	      tasks like memory management, interrupt handling and DMA easier,
95	      and provide a uniform interface to applications.
96	    </para>
97	    <para>
98	      A note on versions: this guide covers features found in the DRM
99	      tree, including the TTM memory manager, output configuration and
100	      mode setting, and the new vblank internals, in addition to all
101	      the regular features found in current kernels.
102	    </para>
103	    <para>
104	      [Insert diagram of typical DRM stack here]
105	    </para>
106	  </chapter>
108	  <!-- Internals -->
110	  <chapter id="drmInternals">
111	    <title>DRM Internals</title>
112	    <para>
113	      This chapter documents DRM internals relevant to driver authors
114	      and developers working to add support for the latest features to
115	      existing drivers.
116	    </para>
117	    <para>
118	      First, we go over some typical driver initialization
119	      requirements, like setting up command buffers, creating an
120	      initial output configuration, and initializing core services.
121	      Subsequent sections cover core internals in more detail,
122	      providing implementation notes and examples.
123	    </para>
124	    <para>
125	      The DRM layer provides several services to graphics drivers,
126	      many of them driven by the application interfaces it provides
127	      through libdrm, the library that wraps most of the DRM ioctls.
128	      These include vblank event handling, memory
129	      management, output management, framebuffer management, command
130	      submission &amp; fencing, suspend/resume support, and DMA
131	      services.
132	    </para>
134	  <!-- Internals: driver init -->
136	  <sect1>
137	    <title>Driver Initialization</title>
138	    <para>
139	      At the core of every DRM driver is a <structname>drm_driver</structname>
140	      structure. Drivers typically statically initialize a drm_driver structure,
141	      and then pass it to one of the <function>drm_*_init()</function> functions
142	      to register it with the DRM subsystem.
143	    </para>
144	    <para>
145	      The <structname>drm_driver</structname> structure contains static
146	      information that describes the driver and features it supports, and
147	      pointers to methods that the DRM core will call to implement the DRM API.
148	      We will first go through the <structname>drm_driver</structname> static
149	      information fields, and will then describe individual operations in
150	      details as they get used in later sections.
151	    </para>
152	    <sect2>
153	      <title>Driver Information</title>
154	      <sect3>
155	        <title>Driver Features</title>
156	        <para>
157	          Drivers inform the DRM core about their requirements and supported
158	          features by setting appropriate flags in the
159	          <structfield>driver_features</structfield> field. Since those flags
160	          influence the DRM core behaviour since registration time, most of them
161	          must be set to registering the <structname>drm_driver</structname>
162	          instance.
163	        </para>
164	        <synopsis>u32 driver_features;</synopsis>
165	        <variablelist>
166	          <title>Driver Feature Flags</title>
167	          <varlistentry>
168	            <term>DRIVER_USE_AGP</term>
169	            <listitem><para>
170	              Driver uses AGP interface, the DRM core will manage AGP resources.
171	            </para></listitem>
172	          </varlistentry>
173	          <varlistentry>
174	            <term>DRIVER_REQUIRE_AGP</term>
175	            <listitem><para>
176	              Driver needs AGP interface to function. AGP initialization failure
177	              will become a fatal error.
178	            </para></listitem>
179	          </varlistentry>
180	          <varlistentry>
181	            <term>DRIVER_PCI_DMA</term>
182	            <listitem><para>
183	              Driver is capable of PCI DMA, mapping of PCI DMA buffers to
184	              userspace will be enabled. Deprecated.
185	            </para></listitem>
186	          </varlistentry>
187	          <varlistentry>
188	            <term>DRIVER_SG</term>
189	            <listitem><para>
190	              Driver can perform scatter/gather DMA, allocation and mapping of
191	              scatter/gather buffers will be enabled. Deprecated.
192	            </para></listitem>
193	          </varlistentry>
194	          <varlistentry>
195	            <term>DRIVER_HAVE_DMA</term>
196	            <listitem><para>
197	              Driver supports DMA, the userspace DMA API will be supported.
198	              Deprecated.
199	            </para></listitem>
200	          </varlistentry>
201	          <varlistentry>
202	            <term>DRIVER_HAVE_IRQ</term><term>DRIVER_IRQ_SHARED</term>
203	            <listitem><para>
204	              DRIVER_HAVE_IRQ indicates whether the driver has an IRQ handler
205	              managed by the DRM Core. The core will support simple IRQ handler
206	              installation when the flag is set. The installation process is
207	              described in <xref linkend="drm-irq-registration"/>.</para>
208	              <para>DRIVER_IRQ_SHARED indicates whether the device &amp; handler
209	              support shared IRQs (note that this is required of PCI  drivers).
210	            </para></listitem>
211	          </varlistentry>
212	          <varlistentry>
213	            <term>DRIVER_GEM</term>
214	            <listitem><para>
215	              Driver use the GEM memory manager.
216	            </para></listitem>
217	          </varlistentry>
218	          <varlistentry>
219	            <term>DRIVER_MODESET</term>
220	            <listitem><para>
221	              Driver supports mode setting interfaces (KMS).
222	            </para></listitem>
223	          </varlistentry>
224	          <varlistentry>
225	            <term>DRIVER_PRIME</term>
226	            <listitem><para>
227	              Driver implements DRM PRIME buffer sharing.
228	            </para></listitem>
229	          </varlistentry>
230	          <varlistentry>
231	            <term>DRIVER_RENDER</term>
232	            <listitem><para>
233	              Driver supports dedicated render nodes.
234	            </para></listitem>
235	          </varlistentry>
236	        </variablelist>
237	      </sect3>
238	      <sect3>
239	        <title>Major, Minor and Patchlevel</title>
240	        <synopsis>int major;
241	int minor;
242	int patchlevel;</synopsis>
243	        <para>
244	          The DRM core identifies driver versions by a major, minor and patch
245	          level triplet. The information is printed to the kernel log at
246	          initialization time and passed to userspace through the
247	          DRM_IOCTL_VERSION ioctl.
248	        </para>
249	        <para>
250	          The major and minor numbers are also used to verify the requested driver
251	          API version passed to DRM_IOCTL_SET_VERSION. When the driver API changes
252	          between minor versions, applications can call DRM_IOCTL_SET_VERSION to
253	          select a specific version of the API. If the requested major isn't equal
254	          to the driver major, or the requested minor is larger than the driver
255	          minor, the DRM_IOCTL_SET_VERSION call will return an error. Otherwise
256	          the driver's set_version() method will be called with the requested
257	          version.
258	        </para>
259	      </sect3>
260	      <sect3>
261	        <title>Name, Description and Date</title>
262	        <synopsis>char *name;
263	char *desc;
264	char *date;</synopsis>
265	        <para>
266	          The driver name is printed to the kernel log at initialization time,
267	          used for IRQ registration and passed to userspace through
268	          DRM_IOCTL_VERSION.
269	        </para>
270	        <para>
271	          The driver description is a purely informative string passed to
272	          userspace through the DRM_IOCTL_VERSION ioctl and otherwise unused by
273	          the kernel.
274	        </para>
275	        <para>
276	          The driver date, formatted as YYYYMMDD, is meant to identify the date of
277	          the latest modification to the driver. However, as most drivers fail to
278	          update it, its value is mostly useless. The DRM core prints it to the
279	          kernel log at initialization time and passes it to userspace through the
280	          DRM_IOCTL_VERSION ioctl.
281	        </para>
282	      </sect3>
283	    </sect2>
284	    <sect2>
285	      <title>Driver Load</title>
286	      <para>
287	        The <methodname>load</methodname> method is the driver and device
288	        initialization entry point. The method is responsible for allocating and
289		initializing driver private data, performing resource allocation and
290		mapping (e.g. acquiring
291	        clocks, mapping registers or allocating command buffers), initializing
292	        the memory manager (<xref linkend="drm-memory-management"/>), installing
293	        the IRQ handler (<xref linkend="drm-irq-registration"/>), setting up
294	        vertical blanking handling (<xref linkend="drm-vertical-blank"/>), mode
295		setting (<xref linkend="drm-mode-setting"/>) and initial output
296		configuration (<xref linkend="drm-kms-init"/>).
297	      </para>
298	      <note><para>
299	        If compatibility is a concern (e.g. with drivers converted over from
300	        User Mode Setting to Kernel Mode Setting), care must be taken to prevent
301	        device initialization and control that is incompatible with currently
302	        active userspace drivers. For instance, if user level mode setting
303	        drivers are in use, it would be problematic to perform output discovery
304	        &amp; configuration at load time. Likewise, if user-level drivers
305	        unaware of memory management are in use, memory management and command
306	        buffer setup may need to be omitted. These requirements are
307	        driver-specific, and care needs to be taken to keep both old and new
308	        applications and libraries working.
309	      </para></note>
310	      <synopsis>int (*load) (struct drm_device *, unsigned long flags);</synopsis>
311	      <para>
312	        The method takes two arguments, a pointer to the newly created
313		<structname>drm_device</structname> and flags. The flags are used to
314		pass the <structfield>driver_data</structfield> field of the device id
315		corresponding to the device passed to <function>drm_*_init()</function>.
316		Only PCI devices currently use this, USB and platform DRM drivers have
317		their <methodname>load</methodname> method called with flags to 0.
318	      </para>
319	      <sect3>
320	        <title>Driver Private Data</title>
321	        <para>
322	          The driver private hangs off the main
323	          <structname>drm_device</structname> structure and can be used for
324	          tracking various device-specific bits of information, like register
325	          offsets, command buffer status, register state for suspend/resume, etc.
326	          At load time, a driver may simply allocate one and set
327	          <structname>drm_device</structname>.<structfield>dev_priv</structfield>
328	          appropriately; it should be freed and
329	          <structname>drm_device</structname>.<structfield>dev_priv</structfield>
330	          set to NULL when the driver is unloaded.
331	        </para>
332	      </sect3>
333	      <sect3 id="drm-irq-registration">
334	        <title>IRQ Registration</title>
335	        <para>
336	          The DRM core tries to facilitate IRQ handler registration and
337	          unregistration by providing <function>drm_irq_install</function> and
338	          <function>drm_irq_uninstall</function> functions. Those functions only
339	          support a single interrupt per device, devices that use more than one
340	          IRQs need to be handled manually.
341	        </para>
342	        <sect4>
343	          <title>Managed IRQ Registration</title>
344	          <para>
345	            Both the <function>drm_irq_install</function> and
346		    <function>drm_irq_uninstall</function> functions get the device IRQ by
347		    calling <function>drm_dev_to_irq</function>. This inline function will
348		    call a bus-specific operation to retrieve the IRQ number. For platform
349		    devices, <function>platform_get_irq</function>(..., 0) is used to
350		    retrieve the IRQ number.
351	          </para>
352	          <para>
353	            <function>drm_irq_install</function> starts by calling the
354	            <methodname>irq_preinstall</methodname> driver operation. The operation
355	            is optional and must make sure that the interrupt will not get fired by
356	            clearing all pending interrupt flags or disabling the interrupt.
357	          </para>
358	          <para>
359	            The IRQ will then be requested by a call to
360	            <function>request_irq</function>. If the DRIVER_IRQ_SHARED driver
361	            feature flag is set, a shared (IRQF_SHARED) IRQ handler will be
362	            requested.
363	          </para>
364	          <para>
365	            The IRQ handler function must be provided as the mandatory irq_handler
366	            driver operation. It will get passed directly to
367	            <function>request_irq</function> and thus has the same prototype as all
368	            IRQ handlers. It will get called with a pointer to the DRM device as the
369	            second argument.
370	          </para>
371	          <para>
372	            Finally the function calls the optional
373	            <methodname>irq_postinstall</methodname> driver operation. The operation
374	            usually enables interrupts (excluding the vblank interrupt, which is
375	            enabled separately), but drivers may choose to enable/disable interrupts
376	            at a different time.
377	          </para>
378	          <para>
379	            <function>drm_irq_uninstall</function> is similarly used to uninstall an
380	            IRQ handler. It starts by waking up all processes waiting on a vblank
381	            interrupt to make sure they don't hang, and then calls the optional
382	            <methodname>irq_uninstall</methodname> driver operation. The operation
383	            must disable all hardware interrupts. Finally the function frees the IRQ
384	            by calling <function>free_irq</function>.
385	          </para>
386	        </sect4>
387	        <sect4>
388	          <title>Manual IRQ Registration</title>
389	          <para>
390	            Drivers that require multiple interrupt handlers can't use the managed
391	            IRQ registration functions. In that case IRQs must be registered and
392	            unregistered manually (usually with the <function>request_irq</function>
393	            and <function>free_irq</function> functions, or their devm_* equivalent).
394	          </para>
395	          <para>
396	            When manually registering IRQs, drivers must not set the DRIVER_HAVE_IRQ
397	            driver feature flag, and must not provide the
398		    <methodname>irq_handler</methodname> driver operation. They must set the
399		    <structname>drm_device</structname> <structfield>irq_enabled</structfield>
400		    field to 1 upon registration of the IRQs, and clear it to 0 after
401		    unregistering the IRQs.
402	          </para>
403	        </sect4>
404	      </sect3>
405	      <sect3>
406	        <title>Memory Manager Initialization</title>
407	        <para>
408	          Every DRM driver requires a memory manager which must be initialized at
409	          load time. DRM currently contains two memory managers, the Translation
410	          Table Manager (TTM) and the Graphics Execution Manager (GEM).
411	          This document describes the use of the GEM memory manager only. See
412	          <xref linkend="drm-memory-management"/> for details.
413	        </para>
414	      </sect3>
415	      <sect3>
416	        <title>Miscellaneous Device Configuration</title>
417	        <para>
418	          Another task that may be necessary for PCI devices during configuration
419	          is mapping the video BIOS. On many devices, the VBIOS describes device
420	          configuration, LCD panel timings (if any), and contains flags indicating
421	          device state. Mapping the BIOS can be done using the pci_map_rom() call,
422	          a convenience function that takes care of mapping the actual ROM,
423	          whether it has been shadowed into memory (typically at address 0xc0000)
424	          or exists on the PCI device in the ROM BAR. Note that after the ROM has
425	          been mapped and any necessary information has been extracted, it should
426	          be unmapped; on many devices, the ROM address decoder is shared with
427	          other BARs, so leaving it mapped could cause undesired behaviour like
428	          hangs or memory corruption.
429	  <!--!Fdrivers/pci/rom.c pci_map_rom-->
430	        </para>
431	      </sect3>
432	    </sect2>
433	  </sect1>
435	  <!-- Internals: memory management -->
437	  <sect1 id="drm-memory-management">
438	    <title>Memory management</title>
439	    <para>
440	      Modern Linux systems require large amount of graphics memory to store
441	      frame buffers, textures, vertices and other graphics-related data. Given
442	      the very dynamic nature of many of that data, managing graphics memory
443	      efficiently is thus crucial for the graphics stack and plays a central
444	      role in the DRM infrastructure.
445	    </para>
446	    <para>
447	      The DRM core includes two memory managers, namely Translation Table Maps
448	      (TTM) and Graphics Execution Manager (GEM). TTM was the first DRM memory
449	      manager to be developed and tried to be a one-size-fits-them all
450	      solution. It provides a single userspace API to accommodate the need of
451	      all hardware, supporting both Unified Memory Architecture (UMA) devices
452	      and devices with dedicated video RAM (i.e. most discrete video cards).
453	      This resulted in a large, complex piece of code that turned out to be
454	      hard to use for driver development.
455	    </para>
456	    <para>
457	      GEM started as an Intel-sponsored project in reaction to TTM's
458	      complexity. Its design philosophy is completely different: instead of
459	      providing a solution to every graphics memory-related problems, GEM
460	      identified common code between drivers and created a support library to
461	      share it. GEM has simpler initialization and execution requirements than
462	      TTM, but has no video RAM management capabilities and is thus limited to
463	      UMA devices.
464	    </para>
465	    <sect2>
466	      <title>The Translation Table Manager (TTM)</title>
467	      <para>
468		TTM design background and information belongs here.
469	      </para>
470	      <sect3>
471		<title>TTM initialization</title>
472	        <warning><para>This section is outdated.</para></warning>
473	        <para>
474	          Drivers wishing to support TTM must fill out a drm_bo_driver
475	          structure. The structure contains several fields with function
476	          pointers for initializing the TTM, allocating and freeing memory,
477	          waiting for command completion and fence synchronization, and memory
478	          migration. See the radeon_ttm.c file for an example of usage.
479		</para>
480		<para>
481		  The ttm_global_reference structure is made up of several fields:
482		</para>
483		<programlisting>
484		  struct ttm_global_reference {
485		  	enum ttm_global_types global_type;
486		  	size_t size;
487		  	void *object;
488		  	int (*init) (struct ttm_global_reference *);
489		  	void (*release) (struct ttm_global_reference *);
490		  };
491		</programlisting>
492		<para>
493		  There should be one global reference structure for your memory
494		  manager as a whole, and there will be others for each object
495		  created by the memory manager at runtime.  Your global TTM should
496		  have a type of TTM_GLOBAL_TTM_MEM.  The size field for the global
497		  object should be sizeof(struct ttm_mem_global), and the init and
498		  release hooks should point at your driver-specific init and
499		  release routines, which probably eventually call
500		  ttm_mem_global_init and ttm_mem_global_release, respectively.
501		</para>
502		<para>
503		  Once your global TTM accounting structure is set up and initialized
504		  by calling ttm_global_item_ref() on it,
505		  you need to create a buffer object TTM to
506		  provide a pool for buffer object allocation by clients and the
507		  kernel itself.  The type of this object should be TTM_GLOBAL_TTM_BO,
508		  and its size should be sizeof(struct ttm_bo_global).  Again,
509		  driver-specific init and release functions may be provided,
510		  likely eventually calling ttm_bo_global_init() and
511		  ttm_bo_global_release(), respectively.  Also, like the previous
512		  object, ttm_global_item_ref() is used to create an initial reference
513		  count for the TTM, which will call your initialization function.
514		</para>
515	      </sect3>
516	    </sect2>
517	    <sect2 id="drm-gem">
518	      <title>The Graphics Execution Manager (GEM)</title>
519	      <para>
520	        The GEM design approach has resulted in a memory manager that doesn't
521	        provide full coverage of all (or even all common) use cases in its
522	        userspace or kernel API. GEM exposes a set of standard memory-related
523	        operations to userspace and a set of helper functions to drivers, and let
524	        drivers implement hardware-specific operations with their own private API.
525	      </para>
526	      <para>
527	        The GEM userspace API is described in the
528	        <ulink url="http://lwn.net/Articles/283798/"><citetitle>GEM - the Graphics
529	        Execution Manager</citetitle></ulink> article on LWN. While slightly
530	        outdated, the document provides a good overview of the GEM API principles.
531	        Buffer allocation and read and write operations, described as part of the
532	        common GEM API, are currently implemented using driver-specific ioctls.
533	      </para>
534	      <para>
535	        GEM is data-agnostic. It manages abstract buffer objects without knowing
536	        what individual buffers contain. APIs that require knowledge of buffer
537	        contents or purpose, such as buffer allocation or synchronization
538	        primitives, are thus outside of the scope of GEM and must be implemented
539	        using driver-specific ioctls.
540	      </para>
541	      <para>
542		On a fundamental level, GEM involves several operations:
543		<itemizedlist>
544		  <listitem>Memory allocation and freeing</listitem>
545		  <listitem>Command execution</listitem>
546		  <listitem>Aperture management at command execution time</listitem>
547		</itemizedlist>
548		Buffer object allocation is relatively straightforward and largely
549	        provided by Linux's shmem layer, which provides memory to back each
550	        object.
551	      </para>
552	      <para>
553	        Device-specific operations, such as command execution, pinning, buffer
554		read &amp; write, mapping, and domain ownership transfers are left to
555	        driver-specific ioctls.
556	      </para>
557	      <sect3>
558	        <title>GEM Initialization</title>
559	        <para>
560	          Drivers that use GEM must set the DRIVER_GEM bit in the struct
561	          <structname>drm_driver</structname>
562	          <structfield>driver_features</structfield> field. The DRM core will
563	          then automatically initialize the GEM core before calling the
564	          <methodname>load</methodname> operation. Behind the scene, this will
565	          create a DRM Memory Manager object which provides an address space
566	          pool for object allocation.
567	        </para>
568	        <para>
569	          In a KMS configuration, drivers need to allocate and initialize a
570	          command ring buffer following core GEM initialization if required by
571	          the hardware. UMA devices usually have what is called a "stolen"
572	          memory region, which provides space for the initial framebuffer and
573	          large, contiguous memory regions required by the device. This space is
574	          typically not managed by GEM, and must be initialized separately into
575	          its own DRM MM object.
576	        </para>
577	      </sect3>
578	      <sect3>
579	        <title>GEM Objects Creation</title>
580	        <para>
581	          GEM splits creation of GEM objects and allocation of the memory that
582	          backs them in two distinct operations.
583	        </para>
584	        <para>
585	          GEM objects are represented by an instance of struct
586	          <structname>drm_gem_object</structname>. Drivers usually need to extend
587	          GEM objects with private information and thus create a driver-specific
588	          GEM object structure type that embeds an instance of struct
589	          <structname>drm_gem_object</structname>.
590	        </para>
591	        <para>
592	          To create a GEM object, a driver allocates memory for an instance of its
593	          specific GEM object type and initializes the embedded struct
594	          <structname>drm_gem_object</structname> with a call to
595	          <function>drm_gem_object_init</function>. The function takes a pointer to
596	          the DRM device, a pointer to the GEM object and the buffer object size
597	          in bytes.
598	        </para>
599	        <para>
600	          GEM uses shmem to allocate anonymous pageable memory.
601	          <function>drm_gem_object_init</function> will create an shmfs file of
602	          the requested size and store it into the struct
603	          <structname>drm_gem_object</structname> <structfield>filp</structfield>
604	          field. The memory is used as either main storage for the object when the
605	          graphics hardware uses system memory directly or as a backing store
606	          otherwise.
607	        </para>
608	        <para>
609	          Drivers are responsible for the actual physical pages allocation by
610	          calling <function>shmem_read_mapping_page_gfp</function> for each page.
611	          Note that they can decide to allocate pages when initializing the GEM
612	          object, or to delay allocation until the memory is needed (for instance
613	          when a page fault occurs as a result of a userspace memory access or
614	          when the driver needs to start a DMA transfer involving the memory).
615	        </para>
616	        <para>
617	          Anonymous pageable memory allocation is not always desired, for instance
618	          when the hardware requires physically contiguous system memory as is
619	          often the case in embedded devices. Drivers can create GEM objects with
620	          no shmfs backing (called private GEM objects) by initializing them with
621	          a call to <function>drm_gem_private_object_init</function> instead of
622	          <function>drm_gem_object_init</function>. Storage for private GEM
623	          objects must be managed by drivers.
624	        </para>
625	        <para>
626	          Drivers that do not need to extend GEM objects with private information
627	          can call the <function>drm_gem_object_alloc</function> function to
628	          allocate and initialize a struct <structname>drm_gem_object</structname>
629	          instance. The GEM core will call the optional driver
630	          <methodname>gem_init_object</methodname> operation after initializing
631	          the GEM object with <function>drm_gem_object_init</function>.
632	          <synopsis>int (*gem_init_object) (struct drm_gem_object *obj);</synopsis>
633	        </para>
634	        <para>
635	          No alloc-and-init function exists for private GEM objects.
636	        </para>
637	      </sect3>
638	      <sect3>
639	        <title>GEM Objects Lifetime</title>
640	        <para>
641	          All GEM objects are reference-counted by the GEM core. References can be
642	          acquired and release by <function>calling drm_gem_object_reference</function>
643	          and <function>drm_gem_object_unreference</function> respectively. The
644	          caller must hold the <structname>drm_device</structname>
645	          <structfield>struct_mutex</structfield> lock. As a convenience, GEM
646	          provides the <function>drm_gem_object_reference_unlocked</function> and
647	          <function>drm_gem_object_unreference_unlocked</function> functions that
648	          can be called without holding the lock.
649	        </para>
650	        <para>
651	          When the last reference to a GEM object is released the GEM core calls
652	          the <structname>drm_driver</structname>
653	          <methodname>gem_free_object</methodname> operation. That operation is
654	          mandatory for GEM-enabled drivers and must free the GEM object and all
655	          associated resources.
656	        </para>
657	        <para>
658	          <synopsis>void (*gem_free_object) (struct drm_gem_object *obj);</synopsis>
659	          Drivers are responsible for freeing all GEM object resources, including
660	          the resources created by the GEM core. If an mmap offset has been
661	          created for the object (in which case
662	          <structname>drm_gem_object</structname>::<structfield>map_list</structfield>::<structfield>map</structfield>
663	          is not NULL) it must be freed by a call to
664	          <function>drm_gem_free_mmap_offset</function>. The shmfs backing store
665	          must be released by calling <function>drm_gem_object_release</function>
666	          (that function can safely be called if no shmfs backing store has been
667	          created).
668	        </para>
669	      </sect3>
670	      <sect3>
671	        <title>GEM Objects Naming</title>
672	        <para>
673	          Communication between userspace and the kernel refers to GEM objects
674	          using local handles, global names or, more recently, file descriptors.
675	          All of those are 32-bit integer values; the usual Linux kernel limits
676	          apply to the file descriptors.
677	        </para>
678	        <para>
679	          GEM handles are local to a DRM file. Applications get a handle to a GEM
680	          object through a driver-specific ioctl, and can use that handle to refer
681	          to the GEM object in other standard or driver-specific ioctls. Closing a
682	          DRM file handle frees all its GEM handles and dereferences the
683	          associated GEM objects.
684	        </para>
685	        <para>
686	          To create a handle for a GEM object drivers call
687	          <function>drm_gem_handle_create</function>. The function takes a pointer
688	          to the DRM file and the GEM object and returns a locally unique handle.
689	          When the handle is no longer needed drivers delete it with a call to
690	          <function>drm_gem_handle_delete</function>. Finally the GEM object
691	          associated with a handle can be retrieved by a call to
692	          <function>drm_gem_object_lookup</function>.
693	        </para>
694	        <para>
695	          Handles don't take ownership of GEM objects, they only take a reference
696	          to the object that will be dropped when the handle is destroyed. To
697	          avoid leaking GEM objects, drivers must make sure they drop the
698	          reference(s) they own (such as the initial reference taken at object
699	          creation time) as appropriate, without any special consideration for the
700	          handle. For example, in the particular case of combined GEM object and
701	          handle creation in the implementation of the
702	          <methodname>dumb_create</methodname> operation, drivers must drop the
703	          initial reference to the GEM object before returning the handle.
704	        </para>
705	        <para>
706	          GEM names are similar in purpose to handles but are not local to DRM
707	          files. They can be passed between processes to reference a GEM object
708	          globally. Names can't be used directly to refer to objects in the DRM
709	          API, applications must convert handles to names and names to handles
710	          using the DRM_IOCTL_GEM_FLINK and DRM_IOCTL_GEM_OPEN ioctls
711	          respectively. The conversion is handled by the DRM core without any
712	          driver-specific support.
713	        </para>
714		<para>
715		  GEM also supports buffer sharing with dma-buf file descriptors through
716		  PRIME. GEM-based drivers must use the provided helpers functions to
717		  implement the exporting and importing correctly. See <xref linkend="drm-prime-support" />.
718		  Since sharing file descriptors is inherently more secure than the
719		  easily guessable and global GEM names it is the preferred buffer
720		  sharing mechanism. Sharing buffers through GEM names is only supported
721		  for legacy userspace. Furthermore PRIME also allows cross-device
722		  buffer sharing since it is based on dma-bufs.
723		</para>
724	      </sect3>
725	      <sect3 id="drm-gem-objects-mapping">
726	        <title>GEM Objects Mapping</title>
727	        <para>
728	          Because mapping operations are fairly heavyweight GEM favours
729	          read/write-like access to buffers, implemented through driver-specific
730	          ioctls, over mapping buffers to userspace. However, when random access
731	          to the buffer is needed (to perform software rendering for instance),
732	          direct access to the object can be more efficient.
733	        </para>
734	        <para>
735	          The mmap system call can't be used directly to map GEM objects, as they
736	          don't have their own file handle. Two alternative methods currently
737	          co-exist to map GEM objects to userspace. The first method uses a
738	          driver-specific ioctl to perform the mapping operation, calling
739	          <function>do_mmap</function> under the hood. This is often considered
740	          dubious, seems to be discouraged for new GEM-enabled drivers, and will
741	          thus not be described here.
742	        </para>
743	        <para>
744	          The second method uses the mmap system call on the DRM file handle.
745	          <synopsis>void *mmap(void *addr, size_t length, int prot, int flags, int fd,
746	             off_t offset);</synopsis>
747	          DRM identifies the GEM object to be mapped by a fake offset passed
748	          through the mmap offset argument. Prior to being mapped, a GEM object
749	          must thus be associated with a fake offset. To do so, drivers must call
750	          <function>drm_gem_create_mmap_offset</function> on the object. The
751	          function allocates a fake offset range from a pool and stores the
752	          offset divided by PAGE_SIZE in
753	          <literal>obj-&gt;map_list.hash.key</literal>. Care must be taken not to
754	          call <function>drm_gem_create_mmap_offset</function> if a fake offset
755	          has already been allocated for the object. This can be tested by
756	          <literal>obj-&gt;map_list.map</literal> being non-NULL.
757	        </para>
758	        <para>
759	          Once allocated, the fake offset value
760	          (<literal>obj-&gt;map_list.hash.key &lt;&lt; PAGE_SHIFT</literal>)
761	          must be passed to the application in a driver-specific way and can then
762	          be used as the mmap offset argument.
763	        </para>
764	        <para>
765	          The GEM core provides a helper method <function>drm_gem_mmap</function>
766	          to handle object mapping. The method can be set directly as the mmap
767	          file operation handler. It will look up the GEM object based on the
768	          offset value and set the VMA operations to the
769	          <structname>drm_driver</structname> <structfield>gem_vm_ops</structfield>
770	          field. Note that <function>drm_gem_mmap</function> doesn't map memory to
771	          userspace, but relies on the driver-provided fault handler to map pages
772	          individually.
773	        </para>
774	        <para>
775	          To use <function>drm_gem_mmap</function>, drivers must fill the struct
776	          <structname>drm_driver</structname> <structfield>gem_vm_ops</structfield>
777	          field with a pointer to VM operations.
778	        </para>
779	        <para>
780	          <synopsis>struct vm_operations_struct *gem_vm_ops
782	  struct vm_operations_struct {
783	          void (*open)(struct vm_area_struct * area);
784	          void (*close)(struct vm_area_struct * area);
785	          int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
786	  };</synopsis>
787	        </para>
788	        <para>
789	          The <methodname>open</methodname> and <methodname>close</methodname>
790	          operations must update the GEM object reference count. Drivers can use
791	          the <function>drm_gem_vm_open</function> and
792	          <function>drm_gem_vm_close</function> helper functions directly as open
793	          and close handlers.
794	        </para>
795	        <para>
796	          The fault operation handler is responsible for mapping individual pages
797	          to userspace when a page fault occurs. Depending on the memory
798	          allocation scheme, drivers can allocate pages at fault time, or can
799	          decide to allocate memory for the GEM object at the time the object is
800	          created.
801	        </para>
802	        <para>
803	          Drivers that want to map the GEM object upfront instead of handling page
804	          faults can implement their own mmap file operation handler.
805	        </para>
806	      </sect3>
807	      <sect3>
808	        <title>Memory Coherency</title>
809	        <para>
810	          When mapped to the device or used in a command buffer, backing pages
811	          for an object are flushed to memory and marked write combined so as to
812	          be coherent with the GPU. Likewise, if the CPU accesses an object
813	          after the GPU has finished rendering to the object, then the object
814	          must be made coherent with the CPU's view of memory, usually involving
815	          GPU cache flushing of various kinds. This core CPU&lt;-&gt;GPU
816	          coherency management is provided by a device-specific ioctl, which
817	          evaluates an object's current domain and performs any necessary
818	          flushing or synchronization to put the object into the desired
819	          coherency domain (note that the object may be busy, i.e. an active
820	          render target; in that case, setting the domain blocks the client and
821	          waits for rendering to complete before performing any necessary
822	          flushing operations).
823	        </para>
824	      </sect3>
825	      <sect3>
826	        <title>Command Execution</title>
827	        <para>
828		  Perhaps the most important GEM function for GPU devices is providing a
829	          command execution interface to clients. Client programs construct
830	          command buffers containing references to previously allocated memory
831	          objects, and then submit them to GEM. At that point, GEM takes care to
832	          bind all the objects into the GTT, execute the buffer, and provide
833	          necessary synchronization between clients accessing the same buffers.
834	          This often involves evicting some objects from the GTT and re-binding
835	          others (a fairly expensive operation), and providing relocation
836	          support which hides fixed GTT offsets from clients. Clients must take
837	          care not to submit command buffers that reference more objects than
838	          can fit in the GTT; otherwise, GEM will reject them and no rendering
839	          will occur. Similarly, if several objects in the buffer require fence
840	          registers to be allocated for correct rendering (e.g. 2D blits on
841	          pre-965 chips), care must be taken not to require more fence registers
842	          than are available to the client. Such resource management should be
843	          abstracted from the client in libdrm.
844	        </para>
845	      </sect3>
846	      <sect3>
847	        <title>GEM Function Reference</title>
848	!Edrivers/gpu/drm/drm_gem.c
849	      </sect3>
850	      </sect2>
851	      <sect2>
852		<title>VMA Offset Manager</title>
853	!Pdrivers/gpu/drm/drm_vma_manager.c vma offset manager
854	!Edrivers/gpu/drm/drm_vma_manager.c
855	!Iinclude/drm/drm_vma_manager.h
856	      </sect2>
857	      <sect2 id="drm-prime-support">
858		<title>PRIME Buffer Sharing</title>
859		<para>
860		  PRIME is the cross device buffer sharing framework in drm, originally
861		  created for the OPTIMUS range of multi-gpu platforms. To userspace
862		  PRIME buffers are dma-buf based file descriptors.
863		</para>
864		<sect3>
865		  <title>Overview and Driver Interface</title>
866		  <para>
867		    Similar to GEM global names, PRIME file descriptors are
868		    also used to share buffer objects across processes. They offer
869		    additional security: as file descriptors must be explicitly sent over
870		    UNIX domain sockets to be shared between applications, they can't be
871		    guessed like the globally unique GEM names.
872		  </para>
873		  <para>
874		    Drivers that support the PRIME
875		    API must set the DRIVER_PRIME bit in the struct
876		    <structname>drm_driver</structname>
877		    <structfield>driver_features</structfield> field, and implement the
878		    <methodname>prime_handle_to_fd</methodname> and
879		    <methodname>prime_fd_to_handle</methodname> operations.
880		  </para>
881		  <para>
882		    <synopsis>int (*prime_handle_to_fd)(struct drm_device *dev,
883				  struct drm_file *file_priv, uint32_t handle,
884				  uint32_t flags, int *prime_fd);
885	int (*prime_fd_to_handle)(struct drm_device *dev,
886				  struct drm_file *file_priv, int prime_fd,
887				  uint32_t *handle);</synopsis>
888		    Those two operations convert a handle to a PRIME file descriptor and
889		    vice versa. Drivers must use the kernel dma-buf buffer sharing framework
890		    to manage the PRIME file descriptors. Similar to the mode setting
891		    API PRIME is agnostic to the underlying buffer object manager, as
892		    long as handles are 32bit unsigned integers.
893		  </para>
894		  <para>
895		    While non-GEM drivers must implement the operations themselves, GEM
896		    drivers must use the <function>drm_gem_prime_handle_to_fd</function>
897		    and <function>drm_gem_prime_fd_to_handle</function> helper functions.
898		    Those helpers rely on the driver
899		    <methodname>gem_prime_export</methodname> and
900		    <methodname>gem_prime_import</methodname> operations to create a dma-buf
901		    instance from a GEM object (dma-buf exporter role) and to create a GEM
902		    object from a dma-buf instance (dma-buf importer role).
903		  </para>
904		  <para>
905		    <synopsis>struct dma_buf * (*gem_prime_export)(struct drm_device *dev,
906					     struct drm_gem_object *obj,
907					     int flags);
908	struct drm_gem_object * (*gem_prime_import)(struct drm_device *dev,
909						    struct dma_buf *dma_buf);</synopsis>
910		    These two operations are mandatory for GEM drivers that support
911		    PRIME.
912		  </para>
913		</sect3>
914	        <sect3>
915	          <title>PRIME Helper Functions</title>
916	!Pdrivers/gpu/drm/drm_prime.c PRIME Helpers
917	        </sect3>
918	      </sect2>
919	      <sect2>
920		<title>PRIME Function References</title>
921	!Edrivers/gpu/drm/drm_prime.c
922	      </sect2>
923	      <sect2>
924		<title>DRM MM Range Allocator</title>
925		<sect3>
926		  <title>Overview</title>
927	!Pdrivers/gpu/drm/drm_mm.c Overview
928		</sect3>
929		<sect3>
930		  <title>LRU Scan/Eviction Support</title>
931	!Pdrivers/gpu/drm/drm_mm.c lru scan roaster
932		</sect3>
933	      </sect2>
934	      <sect2>
935		<title>DRM MM Range Allocator Function References</title>
936	!Edrivers/gpu/drm/drm_mm.c
937	!Iinclude/drm/drm_mm.h
938	      </sect2>
939	  </sect1>
941	  <!-- Internals: mode setting -->
943	  <sect1 id="drm-mode-setting">
944	    <title>Mode Setting</title>
945	    <para>
946	      Drivers must initialize the mode setting core by calling
947	      <function>drm_mode_config_init</function> on the DRM device. The function
948	      initializes the <structname>drm_device</structname>
949	      <structfield>mode_config</structfield> field and never fails. Once done,
950	      mode configuration must be setup by initializing the following fields.
951	    </para>
952	    <itemizedlist>
953	      <listitem>
954	        <synopsis>int min_width, min_height;
955	int max_width, max_height;</synopsis>
956	        <para>
957		  Minimum and maximum width and height of the frame buffers in pixel
958		  units.
959		</para>
960	      </listitem>
961	      <listitem>
962	        <synopsis>struct drm_mode_config_funcs *funcs;</synopsis>
963		<para>Mode setting functions.</para>
964	      </listitem>
965	    </itemizedlist>
966	    <sect2>
967	      <title>Display Modes Function Reference</title>
968	!Iinclude/drm/drm_modes.h
969	!Edrivers/gpu/drm/drm_modes.c
970	    </sect2>
971	    <sect2>
972	      <title>Frame Buffer Creation</title>
973	      <synopsis>struct drm_framebuffer *(*fb_create)(struct drm_device *dev,
974					     struct drm_file *file_priv,
975					     struct drm_mode_fb_cmd2 *mode_cmd);</synopsis>
976	      <para>
977	        Frame buffers are abstract memory objects that provide a source of
978	        pixels to scanout to a CRTC. Applications explicitly request the
979	        creation of frame buffers through the DRM_IOCTL_MODE_ADDFB(2) ioctls and
980	        receive an opaque handle that can be passed to the KMS CRTC control,
981	        plane configuration and page flip functions.
982	      </para>
983	      <para>
984	        Frame buffers rely on the underneath memory manager for low-level memory
985	        operations. When creating a frame buffer applications pass a memory
986	        handle (or a list of memory handles for multi-planar formats) through
987		the <parameter>drm_mode_fb_cmd2</parameter> argument. For drivers using
988		GEM as their userspace buffer management interface this would be a GEM
989		handle.  Drivers are however free to use their own backing storage object
990		handles, e.g. vmwgfx directly exposes special TTM handles to userspace
991		and so expects TTM handles in the create ioctl and not GEM handles.
992	      </para>
993	      <para>
994	        Drivers must first validate the requested frame buffer parameters passed
995	        through the mode_cmd argument. In particular this is where invalid
996	        sizes, pixel formats or pitches can be caught.
997	      </para>
998	      <para>
999	        If the parameters are deemed valid, drivers then create, initialize and
1000	        return an instance of struct <structname>drm_framebuffer</structname>.
1001	        If desired the instance can be embedded in a larger driver-specific
1002		structure. Drivers must fill its <structfield>width</structfield>,
1003		<structfield>height</structfield>, <structfield>pitches</structfield>,
1004	        <structfield>offsets</structfield>, <structfield>depth</structfield>,
1005	        <structfield>bits_per_pixel</structfield> and
1006	        <structfield>pixel_format</structfield> fields from the values passed
1007	        through the <parameter>drm_mode_fb_cmd2</parameter> argument. They
1008	        should call the <function>drm_helper_mode_fill_fb_struct</function>
1009	        helper function to do so.
1010	      </para>
1012	      <para>
1013		The initialization of the new framebuffer instance is finalized with a
1014		call to <function>drm_framebuffer_init</function> which takes a pointer
1015		to DRM frame buffer operations (struct
1016		<structname>drm_framebuffer_funcs</structname>). Note that this function
1017		publishes the framebuffer and so from this point on it can be accessed
1018		concurrently from other threads. Hence it must be the last step in the
1019		driver's framebuffer initialization sequence. Frame buffer operations
1020		are
1021	        <itemizedlist>
1022	          <listitem>
1023	            <synopsis>int (*create_handle)(struct drm_framebuffer *fb,
1024			     struct drm_file *file_priv, unsigned int *handle);</synopsis>
1025	            <para>
1026	              Create a handle to the frame buffer underlying memory object. If
1027	              the frame buffer uses a multi-plane format, the handle will
1028	              reference the memory object associated with the first plane.
1029	            </para>
1030	            <para>
1031	              Drivers call <function>drm_gem_handle_create</function> to create
1032	              the handle.
1033	            </para>
1034	          </listitem>
1035	          <listitem>
1036	            <synopsis>void (*destroy)(struct drm_framebuffer *framebuffer);</synopsis>
1037	            <para>
1038	              Destroy the frame buffer object and frees all associated
1039	              resources. Drivers must call
1040	              <function>drm_framebuffer_cleanup</function> to free resources
1041	              allocated by the DRM core for the frame buffer object, and must
1042	              make sure to unreference all memory objects associated with the
1043	              frame buffer. Handles created by the
1044	              <methodname>create_handle</methodname> operation are released by
1045	              the DRM core.
1046	            </para>
1047	          </listitem>
1048	          <listitem>
1049	            <synopsis>int (*dirty)(struct drm_framebuffer *framebuffer,
1050		     struct drm_file *file_priv, unsigned flags, unsigned color,
1051		     struct drm_clip_rect *clips, unsigned num_clips);</synopsis>
1052	            <para>
1053	              This optional operation notifies the driver that a region of the
1054	              frame buffer has changed in response to a DRM_IOCTL_MODE_DIRTYFB
1055	              ioctl call.
1056	            </para>
1057	          </listitem>
1058	        </itemizedlist>
1059	      </para>
1060	      <para>
1061		The lifetime of a drm framebuffer is controlled with a reference count,
1062		drivers can grab additional references with
1063		<function>drm_framebuffer_reference</function>and drop them
1064		again with <function>drm_framebuffer_unreference</function>. For
1065		driver-private framebuffers for which the last reference is never
1066		dropped (e.g. for the fbdev framebuffer when the struct
1067		<structname>drm_framebuffer</structname> is embedded into the fbdev
1068		helper struct) drivers can manually clean up a framebuffer at module
1069		unload time with
1070		<function>drm_framebuffer_unregister_private</function>.
1071	      </para>
1072	    </sect2>
1073	    <sect2>
1074	      <title>Dumb Buffer Objects</title>
1075	      <para>
1076		The KMS API doesn't standardize backing storage object creation and
1077		leaves it to driver-specific ioctls. Furthermore actually creating a
1078		buffer object even for GEM-based drivers is done through a
1079		driver-specific ioctl - GEM only has a common userspace interface for
1080		sharing and destroying objects. While not an issue for full-fledged
1081		graphics stacks that include device-specific userspace components (in
1082		libdrm for instance), this limit makes DRM-based early boot graphics
1083		unnecessarily complex.
1084	      </para>
1085	      <para>
1086	        Dumb objects partly alleviate the problem by providing a standard
1087	        API to create dumb buffers suitable for scanout, which can then be used
1088	        to create KMS frame buffers.
1089	      </para>
1090	      <para>
1091	        To support dumb objects drivers must implement the
1092	        <methodname>dumb_create</methodname>,
1093	        <methodname>dumb_destroy</methodname> and
1094	        <methodname>dumb_map_offset</methodname> operations.
1095	      </para>
1096	      <itemizedlist>
1097	        <listitem>
1098	          <synopsis>int (*dumb_create)(struct drm_file *file_priv, struct drm_device *dev,
1099	                   struct drm_mode_create_dumb *args);</synopsis>
1100	          <para>
1101	            The <methodname>dumb_create</methodname> operation creates a driver
1102		    object (GEM or TTM handle) suitable for scanout based on the
1103		    width, height and depth from the struct
1104		    <structname>drm_mode_create_dumb</structname> argument. It fills the
1105		    argument's <structfield>handle</structfield>,
1106		    <structfield>pitch</structfield> and <structfield>size</structfield>
1107		    fields with a handle for the newly created object and its line
1108	            pitch and size in bytes.
1109	          </para>
1110	        </listitem>
1111	        <listitem>
1112	          <synopsis>int (*dumb_destroy)(struct drm_file *file_priv, struct drm_device *dev,
1113	                    uint32_t handle);</synopsis>
1114	          <para>
1115	            The <methodname>dumb_destroy</methodname> operation destroys a dumb
1116	            object created by <methodname>dumb_create</methodname>.
1117	          </para>
1118	        </listitem>
1119	        <listitem>
1120	          <synopsis>int (*dumb_map_offset)(struct drm_file *file_priv, struct drm_device *dev,
1121	                       uint32_t handle, uint64_t *offset);</synopsis>
1122	          <para>
1123	            The <methodname>dumb_map_offset</methodname> operation associates an
1124	            mmap fake offset with the object given by the handle and returns
1125	            it. Drivers must use the
1126	            <function>drm_gem_create_mmap_offset</function> function to
1127	            associate the fake offset as described in
1128	            <xref linkend="drm-gem-objects-mapping"/>.
1129	          </para>
1130	        </listitem>
1131	      </itemizedlist>
1132	      <para>
1133	        Note that dumb objects may not be used for gpu acceleration, as has been
1134		attempted on some ARM embedded platforms. Such drivers really must have
1135		a hardware-specific ioctl to allocate suitable buffer objects.
1136	      </para>
1137	    </sect2>
1138	    <sect2>
1139	      <title>Output Polling</title>
1140	      <synopsis>void (*output_poll_changed)(struct drm_device *dev);</synopsis>
1141	      <para>
1142	        This operation notifies the driver that the status of one or more
1143	        connectors has changed. Drivers that use the fb helper can just call the
1144	        <function>drm_fb_helper_hotplug_event</function> function to handle this
1145	        operation.
1146	      </para>
1147	    </sect2>
1148	    <sect2>
1149	      <title>Locking</title>
1150	      <para>
1151	        Beside some lookup structures with their own locking (which is hidden
1152		behind the interface functions) most of the modeset state is protected
1153		by the <code>dev-&lt;mode_config.lock</code> mutex and additionally
1154		per-crtc locks to allow cursor updates, pageflips and similar operations
1155		to occur concurrently with background tasks like output detection.
1156		Operations which cross domains like a full modeset always grab all
1157		locks. Drivers there need to protect resources shared between crtcs with
1158		additional locking. They also need to be careful to always grab the
1159		relevant crtc locks if a modset functions touches crtc state, e.g. for
1160		load detection (which does only grab the <code>mode_config.lock</code>
1161		to allow concurrent screen updates on live crtcs).
1162	      </para>
1163	    </sect2>
1164	  </sect1>
1166	  <!-- Internals: kms initialization and cleanup -->
1168	  <sect1 id="drm-kms-init">
1169	    <title>KMS Initialization and Cleanup</title>
1170	    <para>
1171	      A KMS device is abstracted and exposed as a set of planes, CRTCs, encoders
1172	      and connectors. KMS drivers must thus create and initialize all those
1173	      objects at load time after initializing mode setting.
1174	    </para>
1175	    <sect2>
1176	      <title>CRTCs (struct <structname>drm_crtc</structname>)</title>
1177	      <para>
1178	        A CRTC is an abstraction representing a part of the chip that contains a
1179		pointer to a scanout buffer. Therefore, the number of CRTCs available
1180		determines how many independent scanout buffers can be active at any
1181		given time. The CRTC structure contains several fields to support this:
1182		a pointer to some video memory (abstracted as a frame buffer object), a
1183		display mode, and an (x, y) offset into the video memory to support
1184		panning or configurations where one piece of video memory spans multiple
1185		CRTCs.
1186	      </para>
1187	      <sect3>
1188	        <title>CRTC Initialization</title>
1189	        <para>
1190	          A KMS device must create and register at least one struct
1191	          <structname>drm_crtc</structname> instance. The instance is allocated
1192	          and zeroed by the driver, possibly as part of a larger structure, and
1193	          registered with a call to <function>drm_crtc_init</function> with a
1194	          pointer to CRTC functions.
1195	        </para>
1196	      </sect3>
1197	      <sect3 id="drm-kms-crtcops">
1198	        <title>CRTC Operations</title>
1199	        <sect4>
1200	          <title>Set Configuration</title>
1201	          <synopsis>int (*set_config)(struct drm_mode_set *set);</synopsis>
1202	          <para>
1203	            Apply a new CRTC configuration to the device. The configuration
1204	            specifies a CRTC, a frame buffer to scan out from, a (x,y) position in
1205	            the frame buffer, a display mode and an array of connectors to drive
1206	            with the CRTC if possible.
1207	          </para>
1208	          <para>
1209	            If the frame buffer specified in the configuration is NULL, the driver
1210	            must detach all encoders connected to the CRTC and all connectors
1211	            attached to those encoders and disable them.
1212	          </para>
1213	          <para>
1214	            This operation is called with the mode config lock held.
1215	          </para>
1216	          <note><para>
1217		    Note that the drm core has no notion of restoring the mode setting
1218		    state after resume, since all resume handling is in the full
1219		    responsibility of the driver. The common mode setting helper library
1220		    though provides a helper which can be used for this:
1221		    <function>drm_helper_resume_force_mode</function>.
1222	          </para></note>
1223	        </sect4>
1224	        <sect4>
1225	          <title>Page Flipping</title>
1226	          <synopsis>int (*page_flip)(struct drm_crtc *crtc, struct drm_framebuffer *fb,
1227	                   struct drm_pending_vblank_event *event);</synopsis>
1228	          <para>
1229	            Schedule a page flip to the given frame buffer for the CRTC. This
1230	            operation is called with the mode config mutex held.
1231	          </para>
1232	          <para>
1233	            Page flipping is a synchronization mechanism that replaces the frame
1234	            buffer being scanned out by the CRTC with a new frame buffer during
1235	            vertical blanking, avoiding tearing. When an application requests a page
1236	            flip the DRM core verifies that the new frame buffer is large enough to
1237	            be scanned out by  the CRTC in the currently configured mode and then
1238	            calls the CRTC <methodname>page_flip</methodname> operation with a
1239	            pointer to the new frame buffer.
1240	          </para>
1241	          <para>
1242	            The <methodname>page_flip</methodname> operation schedules a page flip.
1243	            Once any pending rendering targeting the new frame buffer has
1244	            completed, the CRTC will be reprogrammed to display that frame buffer
1245	            after the next vertical refresh. The operation must return immediately
1246	            without waiting for rendering or page flip to complete and must block
1247	            any new rendering to the frame buffer until the page flip completes.
1248	          </para>
1249	          <para>
1250	            If a page flip can be successfully scheduled the driver must set the
1251	            <code>drm_crtc-&lt;fb</code> field to the new framebuffer pointed to
1252	            by <code>fb</code>. This is important so that the reference counting
1253	            on framebuffers stays balanced.
1254	          </para>
1255	          <para>
1256	            If a page flip is already pending, the
1257	            <methodname>page_flip</methodname> operation must return
1258	            -<errorname>EBUSY</errorname>.
1259	          </para>
1260	          <para>
1261	            To synchronize page flip to vertical blanking the driver will likely
1262	            need to enable vertical blanking interrupts. It should call
1263	            <function>drm_vblank_get</function> for that purpose, and call
1264	            <function>drm_vblank_put</function> after the page flip completes.
1265	          </para>
1266	          <para>
1267	            If the application has requested to be notified when page flip completes
1268	            the <methodname>page_flip</methodname> operation will be called with a
1269	            non-NULL <parameter>event</parameter> argument pointing to a
1270	            <structname>drm_pending_vblank_event</structname> instance. Upon page
1271	            flip completion the driver must call <methodname>drm_send_vblank_event</methodname>
1272	            to fill in the event and send to wake up any waiting processes.
1273	            This can be performed with
1274	            <programlisting><![CDATA[
1275	            spin_lock_irqsave(&dev->event_lock, flags);
1276	            ...
1277	            drm_send_vblank_event(dev, pipe, event);
1278	            spin_unlock_irqrestore(&dev->event_lock, flags);
1279	            ]]></programlisting>
1280	          </para>
1281	          <note><para>
1282	            FIXME: Could drivers that don't need to wait for rendering to complete
1283	            just add the event to <literal>dev-&gt;vblank_event_list</literal> and
1284	            let the DRM core handle everything, as for "normal" vertical blanking
1285	            events?
1286	          </para></note>
1287	          <para>
1288	            While waiting for the page flip to complete, the
1289	            <literal>event-&gt;base.link</literal> list head can be used freely by
1290	            the driver to store the pending event in a driver-specific list.
1291	          </para>
1292	          <para>
1293	            If the file handle is closed before the event is signaled, drivers must
1294	            take care to destroy the event in their
1295	            <methodname>preclose</methodname> operation (and, if needed, call
1296	            <function>drm_vblank_put</function>).
1297	          </para>
1298	        </sect4>
1299	        <sect4>
1300	          <title>Miscellaneous</title>
1301	          <itemizedlist>
1302	            <listitem>
1303	              <synopsis>void (*set_property)(struct drm_crtc *crtc,
1304	                     struct drm_property *property, uint64_t value);</synopsis>
1305	              <para>
1306	                Set the value of the given CRTC property to
1307	                <parameter>value</parameter>. See <xref linkend="drm-kms-properties"/>
1308	                for more information about properties.
1309	              </para>
1310	            </listitem>
1311	            <listitem>
1312	              <synopsis>void (*gamma_set)(struct drm_crtc *crtc, u16 *r, u16 *g, u16 *b,
1313	                        uint32_t start, uint32_t size);</synopsis>
1314	              <para>
1315	                Apply a gamma table to the device. The operation is optional.
1316	              </para>
1317	            </listitem>
1318	            <listitem>
1319	              <synopsis>void (*destroy)(struct drm_crtc *crtc);</synopsis>
1320	              <para>
1321	                Destroy the CRTC when not needed anymore. See
1322	                <xref linkend="drm-kms-init"/>.
1323	              </para>
1324	            </listitem>
1325	          </itemizedlist>
1326	        </sect4>
1327	      </sect3>
1328	    </sect2>
1329	    <sect2>
1330	      <title>Planes (struct <structname>drm_plane</structname>)</title>
1331	      <para>
1332	        A plane represents an image source that can be blended with or overlayed
1333		on top of a CRTC during the scanout process. Planes are associated with
1334		a frame buffer to crop a portion of the image memory (source) and
1335		optionally scale it to a destination size. The result is then blended
1336		with or overlayed on top of a CRTC.
1337	      </para>
1338	      <para>
1339	      The DRM core recognizes three types of planes:
1340	      <itemizedlist>
1341	        <listitem>
1342	        DRM_PLANE_TYPE_PRIMARY represents a "main" plane for a CRTC.  Primary
1343	        planes are the planes operated upon by by CRTC modesetting and flipping
1344	        operations described in <xref linkend="drm-kms-crtcops"/>.
1345	        </listitem>
1346	        <listitem>
1347	        DRM_PLANE_TYPE_CURSOR represents a "cursor" plane for a CRTC.  Cursor
1348	        planes are the planes operated upon by the DRM_IOCTL_MODE_CURSOR and
1349	        DRM_IOCTL_MODE_CURSOR2 ioctls.
1350	        </listitem>
1351	        <listitem>
1352	        DRM_PLANE_TYPE_OVERLAY represents all non-primary, non-cursor planes.
1353	        Some drivers refer to these types of planes as "sprites" internally.
1354	        </listitem>
1355	      </itemizedlist>
1356	      For compatibility with legacy userspace, only overlay planes are made
1357	      available to userspace by default.  Userspace clients may set the
1358	      DRM_CLIENT_CAP_UNIVERSAL_PLANES client capability bit to indicate that
1359	      they wish to receive a universal plane list containing all plane types.
1360	      </para>
1361	      <sect3>
1362	        <title>Plane Initialization</title>
1363	        <para>
1364	          To create a plane, a KMS drivers allocates and
1365	          zeroes an instances of struct <structname>drm_plane</structname>
1366	          (possibly as part of a larger structure) and registers it with a call
1367	          to <function>drm_universal_plane_init</function>. The function takes a bitmask
1368	          of the CRTCs that can be associated with the plane, a pointer to the
1369	          plane functions, a list of format supported formats, and the type of
1370	          plane (primary, cursor, or overlay) being initialized.
1371	        </para>
1372	        <para>
1373	          Cursor and overlay planes are optional.  All drivers should provide
1374	          one primary plane per CRTC (although this requirement may change in
1375	          the future); drivers that do not wish to provide special handling for
1376	          primary planes may make use of the helper functions described in
1377	          <xref linkend="drm-kms-planehelpers"/> to create and register a
1378	          primary plane with standard capabilities.
1379	        </para>
1380	      </sect3>
1381	      <sect3>
1382	        <title>Plane Operations</title>
1383	        <itemizedlist>
1384	          <listitem>
1385	            <synopsis>int (*update_plane)(struct drm_plane *plane, struct drm_crtc *crtc,
1386	                        struct drm_framebuffer *fb, int crtc_x, int crtc_y,
1387	                        unsigned int crtc_w, unsigned int crtc_h,
1388	                        uint32_t src_x, uint32_t src_y,
1389	                        uint32_t src_w, uint32_t src_h);</synopsis>
1390	            <para>
1391	              Enable and configure the plane to use the given CRTC and frame buffer.
1392	            </para>
1393	            <para>
1394	              The source rectangle in frame buffer memory coordinates is given by
1395	              the <parameter>src_x</parameter>, <parameter>src_y</parameter>,
1396	              <parameter>src_w</parameter> and <parameter>src_h</parameter>
1397	              parameters (as 16.16 fixed point values). Devices that don't support
1398	              subpixel plane coordinates can ignore the fractional part.
1399	            </para>
1400	            <para>
1401	              The destination rectangle in CRTC coordinates is given by the
1402	              <parameter>crtc_x</parameter>, <parameter>crtc_y</parameter>,
1403	              <parameter>crtc_w</parameter> and <parameter>crtc_h</parameter>
1404	              parameters (as integer values). Devices scale the source rectangle to
1405	              the destination rectangle. If scaling is not supported, and the source
1406	              rectangle size doesn't match the destination rectangle size, the
1407	              driver must return a -<errorname>EINVAL</errorname> error.
1408	            </para>
1409	          </listitem>
1410	          <listitem>
1411	            <synopsis>int (*disable_plane)(struct drm_plane *plane);</synopsis>
1412	            <para>
1413	              Disable the plane. The DRM core calls this method in response to a
1414	              DRM_IOCTL_MODE_SETPLANE ioctl call with the frame buffer ID set to 0.
1415	              Disabled planes must not be processed by the CRTC.
1416	            </para>
1417	          </listitem>
1418	          <listitem>
1419	            <synopsis>void (*destroy)(struct drm_plane *plane);</synopsis>
1420	            <para>
1421	              Destroy the plane when not needed anymore. See
1422	              <xref linkend="drm-kms-init"/>.
1423	            </para>
1424	          </listitem>
1425	        </itemizedlist>
1426	      </sect3>
1427	    </sect2>
1428	    <sect2>
1429	      <title>Encoders (struct <structname>drm_encoder</structname>)</title>
1430	      <para>
1431	        An encoder takes pixel data from a CRTC and converts it to a format
1432		suitable for any attached connectors. On some devices, it may be
1433		possible to have a CRTC send data to more than one encoder. In that
1434		case, both encoders would receive data from the same scanout buffer,
1435		resulting in a "cloned" display configuration across the connectors
1436		attached to each encoder.
1437	      </para>
1438	      <sect3>
1439	        <title>Encoder Initialization</title>
1440	        <para>
1441	          As for CRTCs, a KMS driver must create, initialize and register at
1442	          least one struct <structname>drm_encoder</structname> instance. The
1443	          instance is allocated and zeroed by the driver, possibly as part of a
1444	          larger structure.
1445	        </para>
1446	        <para>
1447	          Drivers must initialize the struct <structname>drm_encoder</structname>
1448	          <structfield>possible_crtcs</structfield> and
1449	          <structfield>possible_clones</structfield> fields before registering the
1450	          encoder. Both fields are bitmasks of respectively the CRTCs that the
1451	          encoder can be connected to, and sibling encoders candidate for cloning.
1452	        </para>
1453	        <para>
1454	          After being initialized, the encoder must be registered with a call to
1455	          <function>drm_encoder_init</function>. The function takes a pointer to
1456	          the encoder functions and an encoder type. Supported types are
1457	          <itemizedlist>
1458	            <listitem>
1459	              DRM_MODE_ENCODER_DAC for VGA and analog on DVI-I/DVI-A
1460	              </listitem>
1461	            <listitem>
1462	              DRM_MODE_ENCODER_TMDS for DVI, HDMI and (embedded) DisplayPort
1463	            </listitem>
1464	            <listitem>
1465	              DRM_MODE_ENCODER_LVDS for display panels
1466	            </listitem>
1467	            <listitem>
1468	              DRM_MODE_ENCODER_TVDAC for TV output (Composite, S-Video, Component,
1469	              SCART)
1470	            </listitem>
1471	            <listitem>
1472	              DRM_MODE_ENCODER_VIRTUAL for virtual machine displays
1473	            </listitem>
1474	          </itemizedlist>
1475	        </para>
1476	        <para>
1477	          Encoders must be attached to a CRTC to be used. DRM drivers leave
1478	          encoders unattached at initialization time. Applications (or the fbdev
1479	          compatibility layer when implemented) are responsible for attaching the
1480	          encoders they want to use to a CRTC.
1481	        </para>
1482	      </sect3>
1483	      <sect3>
1484	        <title>Encoder Operations</title>
1485	        <itemizedlist>
1486	          <listitem>
1487	            <synopsis>void (*destroy)(struct drm_encoder *encoder);</synopsis>
1488	            <para>
1489	              Called to destroy the encoder when not needed anymore. See
1490	              <xref linkend="drm-kms-init"/>.
1491	            </para>
1492	          </listitem>
1493	          <listitem>
1494	            <synopsis>void (*set_property)(struct drm_plane *plane,
1495	                     struct drm_property *property, uint64_t value);</synopsis>
1496	            <para>
1497	              Set the value of the given plane property to
1498	              <parameter>value</parameter>. See <xref linkend="drm-kms-properties"/>
1499	              for more information about properties.
1500	            </para>
1501	          </listitem>
1502	        </itemizedlist>
1503	      </sect3>
1504	    </sect2>
1505	    <sect2>
1506	      <title>Connectors (struct <structname>drm_connector</structname>)</title>
1507	      <para>
1508	        A connector is the final destination for pixel data on a device, and
1509		usually connects directly to an external display device like a monitor
1510		or laptop panel. A connector can only be attached to one encoder at a
1511		time. The connector is also the structure where information about the
1512		attached display is kept, so it contains fields for display data, EDID
1513		data, DPMS &amp; connection status, and information about modes
1514		supported on the attached displays.
1515	      </para>
1516	      <sect3>
1517	        <title>Connector Initialization</title>
1518	        <para>
1519	          Finally a KMS driver must create, initialize, register and attach at
1520	          least one struct <structname>drm_connector</structname> instance. The
1521	          instance is created as other KMS objects and initialized by setting the
1522	          following fields.
1523	        </para>
1524	        <variablelist>
1525	          <varlistentry>
1526	            <term><structfield>interlace_allowed</structfield></term>
1527	            <listitem><para>
1528	              Whether the connector can handle interlaced modes.
1529	            </para></listitem>
1530	          </varlistentry>
1531	          <varlistentry>
1532	            <term><structfield>doublescan_allowed</structfield></term>
1533	            <listitem><para>
1534	              Whether the connector can handle doublescan.
1535	            </para></listitem>
1536	          </varlistentry>
1537	          <varlistentry>
1538	            <term><structfield>display_info
1539	            </structfield></term>
1540	            <listitem><para>
1541	              Display information is filled from EDID information when a display
1542	              is detected. For non hot-pluggable displays such as flat panels in
1543	              embedded systems, the driver should initialize the
1544	              <structfield>display_info</structfield>.<structfield>width_mm</structfield>
1545	              and
1546	              <structfield>display_info</structfield>.<structfield>height_mm</structfield>
1547	              fields with the physical size of the display.
1548	            </para></listitem>
1549	          </varlistentry>
1550	          <varlistentry>
1551	            <term id="drm-kms-connector-polled"><structfield>polled</structfield></term>
1552	            <listitem><para>
1553	              Connector polling mode, a combination of
1554	              <variablelist>
1555	                <varlistentry>
1556	                  <term>DRM_CONNECTOR_POLL_HPD</term>
1557	                  <listitem><para>
1558	                    The connector generates hotplug events and doesn't need to be
1559	                    periodically polled. The CONNECT and DISCONNECT flags must not
1560	                    be set together with the HPD flag.
1561	                  </para></listitem>
1562	                </varlistentry>
1563	                <varlistentry>
1564	                  <term>DRM_CONNECTOR_POLL_CONNECT</term>
1565	                  <listitem><para>
1566	                    Periodically poll the connector for connection.
1567	                  </para></listitem>
1568	                </varlistentry>
1569	                <varlistentry>
1570	                  <term>DRM_CONNECTOR_POLL_DISCONNECT</term>
1571	                  <listitem><para>
1572	                    Periodically poll the connector for disconnection.
1573	                  </para></listitem>
1574	                </varlistentry>
1575	              </variablelist>
1576	              Set to 0 for connectors that don't support connection status
1577	              discovery.
1578	            </para></listitem>
1579	          </varlistentry>
1580	        </variablelist>
1581	        <para>
1582	          The connector is then registered with a call to
1583	          <function>drm_connector_init</function> with a pointer to the connector
1584	          functions and a connector type, and exposed through sysfs with a call to
1585	          <function>drm_sysfs_connector_add</function>.
1586	        </para>
1587	        <para>
1588	          Supported connector types are
1589	          <itemizedlist>
1590	            <listitem>DRM_MODE_CONNECTOR_VGA</listitem>
1591	            <listitem>DRM_MODE_CONNECTOR_DVII</listitem>
1592	            <listitem>DRM_MODE_CONNECTOR_DVID</listitem>
1593	            <listitem>DRM_MODE_CONNECTOR_DVIA</listitem>
1594	            <listitem>DRM_MODE_CONNECTOR_Composite</listitem>
1595	            <listitem>DRM_MODE_CONNECTOR_SVIDEO</listitem>
1596	            <listitem>DRM_MODE_CONNECTOR_LVDS</listitem>
1597	            <listitem>DRM_MODE_CONNECTOR_Component</listitem>
1598	            <listitem>DRM_MODE_CONNECTOR_9PinDIN</listitem>
1599	            <listitem>DRM_MODE_CONNECTOR_DisplayPort</listitem>
1600	            <listitem>DRM_MODE_CONNECTOR_HDMIA</listitem>
1601	            <listitem>DRM_MODE_CONNECTOR_HDMIB</listitem>
1602	            <listitem>DRM_MODE_CONNECTOR_TV</listitem>
1603	            <listitem>DRM_MODE_CONNECTOR_eDP</listitem>
1604	            <listitem>DRM_MODE_CONNECTOR_VIRTUAL</listitem>
1605	          </itemizedlist>
1606	        </para>
1607	        <para>
1608	          Connectors must be attached to an encoder to be used. For devices that
1609	          map connectors to encoders 1:1, the connector should be attached at
1610	          initialization time with a call to
1611	          <function>drm_mode_connector_attach_encoder</function>. The driver must
1612	          also set the <structname>drm_connector</structname>
1613	          <structfield>encoder</structfield> field to point to the attached
1614	          encoder.
1615	        </para>
1616	        <para>
1617	          Finally, drivers must initialize the connectors state change detection
1618	          with a call to <function>drm_kms_helper_poll_init</function>. If at
1619	          least one connector is pollable but can't generate hotplug interrupts
1620	          (indicated by the DRM_CONNECTOR_POLL_CONNECT and
1621	          DRM_CONNECTOR_POLL_DISCONNECT connector flags), a delayed work will
1622	          automatically be queued to periodically poll for changes. Connectors
1623	          that can generate hotplug interrupts must be marked with the
1624	          DRM_CONNECTOR_POLL_HPD flag instead, and their interrupt handler must
1625	          call <function>drm_helper_hpd_irq_event</function>. The function will
1626	          queue a delayed work to check the state of all connectors, but no
1627	          periodic polling will be done.
1628	        </para>
1629	      </sect3>
1630	      <sect3>
1631	        <title>Connector Operations</title>
1632	        <note><para>
1633	          Unless otherwise state, all operations are mandatory.
1634	        </para></note>
1635	        <sect4>
1636	          <title>DPMS</title>
1637	          <synopsis>void (*dpms)(struct drm_connector *connector, int mode);</synopsis>
1638	          <para>
1639	            The DPMS operation sets the power state of a connector. The mode
1640	            argument is one of
1641	            <itemizedlist>
1642	              <listitem><para>DRM_MODE_DPMS_ON</para></listitem>
1643	              <listitem><para>DRM_MODE_DPMS_STANDBY</para></listitem>
1644	              <listitem><para>DRM_MODE_DPMS_SUSPEND</para></listitem>
1645	              <listitem><para>DRM_MODE_DPMS_OFF</para></listitem>
1646	            </itemizedlist>
1647	          </para>
1648	          <para>
1649	            In all but DPMS_ON mode the encoder to which the connector is attached
1650	            should put the display in low-power mode by driving its signals
1651	            appropriately. If more than one connector is attached to the encoder
1652	            care should be taken not to change the power state of other displays as
1653	            a side effect. Low-power mode should be propagated to the encoders and
1654	            CRTCs when all related connectors are put in low-power mode.
1655	          </para>
1656	        </sect4>
1657	        <sect4>
1658	          <title>Modes</title>
1659	          <synopsis>int (*fill_modes)(struct drm_connector *connector, uint32_t max_width,
1660	                      uint32_t max_height);</synopsis>
1661	          <para>
1662	            Fill the mode list with all supported modes for the connector. If the
1663	            <parameter>max_width</parameter> and <parameter>max_height</parameter>
1664	            arguments are non-zero, the implementation must ignore all modes wider
1665	            than <parameter>max_width</parameter> or higher than
1666	            <parameter>max_height</parameter>.
1667	          </para>
1668	          <para>
1669	            The connector must also fill in this operation its
1670	            <structfield>display_info</structfield>
1671	            <structfield>width_mm</structfield> and
1672	            <structfield>height_mm</structfield> fields with the connected display
1673	            physical size in millimeters. The fields should be set to 0 if the value
1674	            isn't known or is not applicable (for instance for projector devices).
1675	          </para>
1676	        </sect4>
1677	        <sect4>
1678	          <title>Connection Status</title>
1679	          <para>
1680	            The connection status is updated through polling or hotplug events when
1681	            supported (see <xref linkend="drm-kms-connector-polled"/>). The status
1682	            value is reported to userspace through ioctls and must not be used
1683	            inside the driver, as it only gets initialized by a call to
1684	            <function>drm_mode_getconnector</function> from userspace.
1685	          </para>
1686	          <synopsis>enum drm_connector_status (*detect)(struct drm_connector *connector,
1687	                                        bool force);</synopsis>
1688	          <para>
1689	            Check to see if anything is attached to the connector. The
1690	            <parameter>force</parameter> parameter is set to false whilst polling or
1691	            to true when checking the connector due to user request.
1692	            <parameter>force</parameter> can be used by the driver to avoid
1693	            expensive, destructive operations during automated probing.
1694	          </para>
1695	          <para>
1696	            Return connector_status_connected if something is connected to the
1697	            connector, connector_status_disconnected if nothing is connected and
1698	            connector_status_unknown if the connection state isn't known.
1699	          </para>
1700	          <para>
1701	            Drivers should only return connector_status_connected if the connection
1702	            status has really been probed as connected. Connectors that can't detect
1703	            the connection status, or failed connection status probes, should return
1704	            connector_status_unknown.
1705	          </para>
1706	        </sect4>
1707	        <sect4>
1708	          <title>Miscellaneous</title>
1709	          <itemizedlist>
1710	            <listitem>
1711	              <synopsis>void (*set_property)(struct drm_connector *connector,
1712	                     struct drm_property *property, uint64_t value);</synopsis>
1713	              <para>
1714	                Set the value of the given connector property to
1715	                <parameter>value</parameter>. See <xref linkend="drm-kms-properties"/>
1716	                for more information about properties.
1717	              </para>
1718	            </listitem>
1719	            <listitem>
1720	              <synopsis>void (*destroy)(struct drm_connector *connector);</synopsis>
1721	              <para>
1722	                Destroy the connector when not needed anymore. See
1723	                <xref linkend="drm-kms-init"/>.
1724	              </para>
1725	            </listitem>
1726	          </itemizedlist>
1727	        </sect4>
1728	      </sect3>
1729	    </sect2>
1730	    <sect2>
1731	      <title>Cleanup</title>
1732	      <para>
1733	        The DRM core manages its objects' lifetime. When an object is not needed
1734		anymore the core calls its destroy function, which must clean up and
1735		free every resource allocated for the object. Every
1736		<function>drm_*_init</function> call must be matched with a
1737		corresponding <function>drm_*_cleanup</function> call to cleanup CRTCs
1738		(<function>drm_crtc_cleanup</function>), planes
1739		(<function>drm_plane_cleanup</function>), encoders
1740		(<function>drm_encoder_cleanup</function>) and connectors
1741		(<function>drm_connector_cleanup</function>). Furthermore, connectors
1742		that have been added to sysfs must be removed by a call to
1743		<function>drm_sysfs_connector_remove</function> before calling
1744		<function>drm_connector_cleanup</function>.
1745	      </para>
1746	      <para>
1747	        Connectors state change detection must be cleanup up with a call to
1748		<function>drm_kms_helper_poll_fini</function>.
1749	      </para>
1750	    </sect2>
1751	    <sect2>
1752	      <title>Output discovery and initialization example</title>
1753	      <programlisting><![CDATA[
1754	void intel_crt_init(struct drm_device *dev)
1755	{
1756		struct drm_connector *connector;
1757		struct intel_output *intel_output;
1759		intel_output = kzalloc(sizeof(struct intel_output), GFP_KERNEL);
1760		if (!intel_output)
1761			return;
1763		connector = &intel_output->base;
1764		drm_connector_init(dev, &intel_output->base,
1765				   &intel_crt_connector_funcs, DRM_MODE_CONNECTOR_VGA);
1767		drm_encoder_init(dev, &intel_output->enc, &intel_crt_enc_funcs,
1770		drm_mode_connector_attach_encoder(&intel_output->base,
1771						  &intel_output->enc);
1773		/* Set up the DDC bus. */
1774		intel_output->ddc_bus = intel_i2c_create(dev, GPIOA, "CRTDDC_A");
1775		if (!intel_output->ddc_bus) {
1776			dev_printk(KERN_ERR, &dev->pdev->dev, "DDC bus registration "
1777				   "failed.\n");
1778			return;
1779		}
1781		intel_output->type = INTEL_OUTPUT_ANALOG;
1782		connector->interlace_allowed = 0;
1783		connector->doublescan_allowed = 0;
1785		drm_encoder_helper_add(&intel_output->enc, &intel_crt_helper_funcs);
1786		drm_connector_helper_add(connector, &intel_crt_connector_helper_funcs);
1788		drm_sysfs_connector_add(connector);
1789	}]]></programlisting>
1790	      <para>
1791	        In the example above (taken from the i915 driver), a CRTC, connector and
1792	        encoder combination is created. A device-specific i2c bus is also
1793	        created for fetching EDID data and performing monitor detection. Once
1794	        the process is complete, the new connector is registered with sysfs to
1795	        make its properties available to applications.
1796	      </para>
1797	    </sect2>
1798	    <sect2>
1799	      <title>KMS API Functions</title>
1800	!Edrivers/gpu/drm/drm_crtc.c
1801	    </sect2>
1802	  </sect1>
1804	  <!-- Internals: kms helper functions -->
1806	  <sect1>
1807	    <title>Mode Setting Helper Functions</title>
1808	    <para>
1809	      The plane, CRTC, encoder and connector functions provided by the drivers
1810	      implement the DRM API. They're called by the DRM core and ioctl handlers
1811	      to handle device state changes and configuration request. As implementing
1812	      those functions often requires logic not specific to drivers, mid-layer
1813	      helper functions are available to avoid duplicating boilerplate code.
1814	    </para>
1815	    <para>
1816	      The DRM core contains one mid-layer implementation. The mid-layer provides
1817	      implementations of several plane, CRTC, encoder and connector functions
1818	      (called from the top of the mid-layer) that pre-process requests and call
1819	      lower-level functions provided by the driver (at the bottom of the
1820	      mid-layer). For instance, the
1821	      <function>drm_crtc_helper_set_config</function> function can be used to
1822	      fill the struct <structname>drm_crtc_funcs</structname>
1823	      <structfield>set_config</structfield> field. When called, it will split
1824	      the <methodname>set_config</methodname> operation in smaller, simpler
1825	      operations and call the driver to handle them.
1826	    </para>
1827	    <para>
1828	      To use the mid-layer, drivers call <function>drm_crtc_helper_add</function>,
1829	      <function>drm_encoder_helper_add</function> and
1830	      <function>drm_connector_helper_add</function> functions to install their
1831	      mid-layer bottom operations handlers, and fill the
1832	      <structname>drm_crtc_funcs</structname>,
1833	      <structname>drm_encoder_funcs</structname> and
1834	      <structname>drm_connector_funcs</structname> structures with pointers to
1835	      the mid-layer top API functions. Installing the mid-layer bottom operation
1836	      handlers is best done right after registering the corresponding KMS object.
1837	    </para>
1838	    <para>
1839	      The mid-layer is not split between CRTC, encoder and connector operations.
1840	      To use it, a driver must provide bottom functions for all of the three KMS
1841	      entities.
1842	    </para>
1843	    <sect2>
1844	      <title>Helper Functions</title>
1845	      <itemizedlist>
1846	        <listitem>
1847	          <synopsis>int drm_crtc_helper_set_config(struct drm_mode_set *set);</synopsis>
1848	          <para>
1849	            The <function>drm_crtc_helper_set_config</function> helper function
1850	            is a CRTC <methodname>set_config</methodname> implementation. It
1851	            first tries to locate the best encoder for each connector by calling
1852	            the connector <methodname>best_encoder</methodname> helper
1853	            operation.
1854	          </para>
1855	          <para>
1856	            After locating the appropriate encoders, the helper function will
1857	            call the <methodname>mode_fixup</methodname> encoder and CRTC helper
1858	            operations to adjust the requested mode, or reject it completely in
1859	            which case an error will be returned to the application. If the new
1860	            configuration after mode adjustment is identical to the current
1861	            configuration the helper function will return without performing any
1862	            other operation.
1863	          </para>
1864	          <para>
1865	            If the adjusted mode is identical to the current mode but changes to
1866	            the frame buffer need to be applied, the
1867	            <function>drm_crtc_helper_set_config</function> function will call
1868	            the CRTC <methodname>mode_set_base</methodname> helper operation. If
1869	            the adjusted mode differs from the current mode, or if the
1870	            <methodname>mode_set_base</methodname> helper operation is not
1871	            provided, the helper function performs a full mode set sequence by
1872	            calling the <methodname>prepare</methodname>,
1873	            <methodname>mode_set</methodname> and
1874	            <methodname>commit</methodname> CRTC and encoder helper operations,
1875	            in that order.
1876	          </para>
1877	        </listitem>
1878	        <listitem>
1879	          <synopsis>void drm_helper_connector_dpms(struct drm_connector *connector, int mode);</synopsis>
1880	          <para>
1881	            The <function>drm_helper_connector_dpms</function> helper function
1882	            is a connector <methodname>dpms</methodname> implementation that
1883	            tracks power state of connectors. To use the function, drivers must
1884	            provide <methodname>dpms</methodname> helper operations for CRTCs
1885	            and encoders to apply the DPMS state to the device.
1886	          </para>
1887	          <para>
1888	            The mid-layer doesn't track the power state of CRTCs and encoders.
1889	            The <methodname>dpms</methodname> helper operations can thus be
1890	            called with a mode identical to the currently active mode.
1891	          </para>
1892	        </listitem>
1893	        <listitem>
1894	          <synopsis>int drm_helper_probe_single_connector_modes(struct drm_connector *connector,
1895	                                            uint32_t maxX, uint32_t maxY);</synopsis>
1896	          <para>
1897	            The <function>drm_helper_probe_single_connector_modes</function> helper
1898	            function is a connector <methodname>fill_modes</methodname>
1899	            implementation that updates the connection status for the connector
1900	            and then retrieves a list of modes by calling the connector
1901	            <methodname>get_modes</methodname> helper operation.
1902	          </para>
1903	          <para>
1904	            The function filters out modes larger than
1905	            <parameter>max_width</parameter> and <parameter>max_height</parameter>
1906	            if specified. It then calls the connector
1907	            <methodname>mode_valid</methodname> helper operation for  each mode in
1908	            the probed list to check whether the mode is valid for the connector.
1909	          </para>
1910	        </listitem>
1911	      </itemizedlist>
1912	    </sect2>
1913	    <sect2>
1914	      <title>CRTC Helper Operations</title>
1915	      <itemizedlist>
1916	        <listitem id="drm-helper-crtc-mode-fixup">
1917	          <synopsis>bool (*mode_fixup)(struct drm_crtc *crtc,
1918	                       const struct drm_display_mode *mode,
1919	                       struct drm_display_mode *adjusted_mode);</synopsis>
1920	          <para>
1921	            Let CRTCs adjust the requested mode or reject it completely. This
1922	            operation returns true if the mode is accepted (possibly after being
1923	            adjusted) or false if it is rejected.
1924	          </para>
1925	          <para>
1926	            The <methodname>mode_fixup</methodname> operation should reject the
1927	            mode if it can't reasonably use it. The definition of "reasonable"
1928	            is currently fuzzy in this context. One possible behaviour would be
1929	            to set the adjusted mode to the panel timings when a fixed-mode
1930	            panel is used with hardware capable of scaling. Another behaviour
1931	            would be to accept any input mode and adjust it to the closest mode
1932	            supported by the hardware (FIXME: This needs to be clarified).
1933	          </para>
1934	        </listitem>
1935	        <listitem>
1936	          <synopsis>int (*mode_set_base)(struct drm_crtc *crtc, int x, int y,
1937	                     struct drm_framebuffer *old_fb)</synopsis>
1938	          <para>
1939	            Move the CRTC on the current frame buffer (stored in
1940	            <literal>crtc-&gt;fb</literal>) to position (x,y). Any of the frame
1941	            buffer, x position or y position may have been modified.
1942	          </para>
1943	          <para>
1944	            This helper operation is optional. If not provided, the
1945	            <function>drm_crtc_helper_set_config</function> function will fall
1946	            back to the <methodname>mode_set</methodname> helper operation.
1947	          </para>
1948	          <note><para>
1949	            FIXME: Why are x and y passed as arguments, as they can be accessed
1950	            through <literal>crtc-&gt;x</literal> and
1951	            <literal>crtc-&gt;y</literal>?
1952	          </para></note>
1953	        </listitem>
1954	        <listitem>
1955	          <synopsis>void (*prepare)(struct drm_crtc *crtc);</synopsis>
1956	          <para>
1957	            Prepare the CRTC for mode setting. This operation is called after
1958	            validating the requested mode. Drivers use it to perform
1959	            device-specific operations required before setting the new mode.
1960	          </para>
1961	        </listitem>
1962	        <listitem>
1963	          <synopsis>int (*mode_set)(struct drm_crtc *crtc, struct drm_display_mode *mode,
1964	                struct drm_display_mode *adjusted_mode, int x, int y,
1965	                struct drm_framebuffer *old_fb);</synopsis>
1966	          <para>
1967	            Set a new mode, position and frame buffer. Depending on the device
1968	            requirements, the mode can be stored internally by the driver and
1969	            applied in the <methodname>commit</methodname> operation, or
1970	            programmed to the hardware immediately.
1971	          </para>
1972	          <para>
1973	            The <methodname>mode_set</methodname> operation returns 0 on success
1974		    or a negative error code if an error occurs.
1975	          </para>
1976	        </listitem>
1977	        <listitem>
1978	          <synopsis>void (*commit)(struct drm_crtc *crtc);</synopsis>
1979	          <para>
1980	            Commit a mode. This operation is called after setting the new mode.
1981	            Upon return the device must use the new mode and be fully
1982	            operational.
1983	          </para>
1984	        </listitem>
1985	      </itemizedlist>
1986	    </sect2>
1987	    <sect2>
1988	      <title>Encoder Helper Operations</title>
1989	      <itemizedlist>
1990	        <listitem>
1991	          <synopsis>bool (*mode_fixup)(struct drm_encoder *encoder,
1992	                       const struct drm_display_mode *mode,
1993	                       struct drm_display_mode *adjusted_mode);</synopsis>
1994	          <para>
1995	            Let encoders adjust the requested mode or reject it completely. This
1996	            operation returns true if the mode is accepted (possibly after being
1997	            adjusted) or false if it is rejected. See the
1998	            <link linkend="drm-helper-crtc-mode-fixup">mode_fixup CRTC helper
1999	            operation</link> for an explanation of the allowed adjustments.
2000	          </para>
2001	        </listitem>
2002	        <listitem>
2003	          <synopsis>void (*prepare)(struct drm_encoder *encoder);</synopsis>
2004	          <para>
2005	            Prepare the encoder for mode setting. This operation is called after
2006	            validating the requested mode. Drivers use it to perform
2007	            device-specific operations required before setting the new mode.
2008	          </para>
2009	        </listitem>
2010	        <listitem>
2011	          <synopsis>void (*mode_set)(struct drm_encoder *encoder,
2012	                 struct drm_display_mode *mode,
2013	                 struct drm_display_mode *adjusted_mode);</synopsis>
2014	          <para>
2015	            Set a new mode. Depending on the device requirements, the mode can
2016	            be stored internally by the driver and applied in the
2017	            <methodname>commit</methodname> operation, or programmed to the
2018	            hardware immediately.
2019	          </para>
2020	        </listitem>
2021	        <listitem>
2022	          <synopsis>void (*commit)(struct drm_encoder *encoder);</synopsis>
2023	          <para>
2024	            Commit a mode. This operation is called after setting the new mode.
2025	            Upon return the device must use the new mode and be fully
2026	            operational.
2027	          </para>
2028	        </listitem>
2029	      </itemizedlist>
2030	    </sect2>
2031	    <sect2>
2032	      <title>Connector Helper Operations</title>
2033	      <itemizedlist>
2034	        <listitem>
2035	          <synopsis>struct drm_encoder *(*best_encoder)(struct drm_connector *connector);</synopsis>
2036	          <para>
2037	            Return a pointer to the best encoder for the connecter. Device that
2038	            map connectors to encoders 1:1 simply return the pointer to the
2039	            associated encoder. This operation is mandatory.
2040	          </para>
2041	        </listitem>
2042	        <listitem>
2043	          <synopsis>int (*get_modes)(struct drm_connector *connector);</synopsis>
2044	          <para>
2045	            Fill the connector's <structfield>probed_modes</structfield> list
2046	            by parsing EDID data with <function>drm_add_edid_modes</function> or
2047	            calling <function>drm_mode_probed_add</function> directly for every
2048	            supported mode and return the number of modes it has detected. This
2049	            operation is mandatory.
2050	          </para>
2051	          <para>
2052	            When adding modes manually the driver creates each mode with a call to
2053	            <function>drm_mode_create</function> and must fill the following fields.
2054	            <itemizedlist>
2055	              <listitem>
2056	                <synopsis>__u32 type;</synopsis>
2057	                <para>
2058	                  Mode type bitmask, a combination of
2059	                  <variablelist>
2060	                    <varlistentry>
2061	                      <term>DRM_MODE_TYPE_BUILTIN</term>
2062	                      <listitem><para>not used?</para></listitem>
2063	                    </varlistentry>
2064	                    <varlistentry>
2065	                      <term>DRM_MODE_TYPE_CLOCK_C</term>
2066	                      <listitem><para>not used?</para></listitem>
2067	                    </varlistentry>
2068	                    <varlistentry>
2069	                      <term>DRM_MODE_TYPE_CRTC_C</term>
2070	                      <listitem><para>not used?</para></listitem>
2071	                    </varlistentry>
2072	                    <varlistentry>
2073	                      <term>
2074	        DRM_MODE_TYPE_PREFERRED - The preferred mode for the connector
2075	                      </term>
2076	                      <listitem>
2077	                        <para>not used?</para>
2078	                      </listitem>
2079	                    </varlistentry>
2080	                    <varlistentry>
2081	                      <term>DRM_MODE_TYPE_DEFAULT</term>
2082	                      <listitem><para>not used?</para></listitem>
2083	                    </varlistentry>
2084	                    <varlistentry>
2085	                      <term>DRM_MODE_TYPE_USERDEF</term>
2086	                      <listitem><para>not used?</para></listitem>
2087	                    </varlistentry>
2088	                    <varlistentry>
2089	                      <term>DRM_MODE_TYPE_DRIVER</term>
2090	                      <listitem>
2091	                        <para>
2092	                          The mode has been created by the driver (as opposed to
2093	                          to user-created modes).
2094	                        </para>
2095	                      </listitem>
2096	                    </varlistentry>
2097	                  </variablelist>
2098	                  Drivers must set the DRM_MODE_TYPE_DRIVER bit for all modes they
2099	                  create, and set the DRM_MODE_TYPE_PREFERRED bit for the preferred
2100	                  mode.
2101	                </para>
2102	              </listitem>
2103	              <listitem>
2104	                <synopsis>__u32 clock;</synopsis>
2105	                <para>Pixel clock frequency in kHz unit</para>
2106	              </listitem>
2107	              <listitem>
2108	                <synopsis>__u16 hdisplay, hsync_start, hsync_end, htotal;
2109	    __u16 vdisplay, vsync_start, vsync_end, vtotal;</synopsis>
2110	                <para>Horizontal and vertical timing information</para>
2111	                <screen><![CDATA[
2112	             Active                 Front           Sync           Back
2113	             Region                 Porch                          Porch
2114	    <-----------------------><----------------><-------------><-------------->
2116	      //////////////////////|
2117	     ////////////////////// |
2118	    //////////////////////  |..................               ................
2119	                                               _______________
2121	    <----- [hv]display ----->
2122	    <------------- [hv]sync_start ------------>
2123	    <--------------------- [hv]sync_end --------------------->
2124	    <-------------------------------- [hv]total ----------------------------->
2125	]]></screen>
2126	              </listitem>
2127	              <listitem>
2128	                <synopsis>__u16 hskew;
2129	    __u16 vscan;</synopsis>
2130	                <para>Unknown</para>
2131	              </listitem>
2132	              <listitem>
2133	                <synopsis>__u32 flags;</synopsis>
2134	                <para>
2135	                  Mode flags, a combination of
2136	                  <variablelist>
2137	                    <varlistentry>
2138	                      <term>DRM_MODE_FLAG_PHSYNC</term>
2139	                      <listitem><para>
2140	                        Horizontal sync is active high
2141	                      </para></listitem>
2142	                    </varlistentry>
2143	                    <varlistentry>
2144	                      <term>DRM_MODE_FLAG_NHSYNC</term>
2145	                      <listitem><para>
2146	                        Horizontal sync is active low
2147	                      </para></listitem>
2148	                    </varlistentry>
2149	                    <varlistentry>
2150	                      <term>DRM_MODE_FLAG_PVSYNC</term>
2151	                      <listitem><para>
2152	                        Vertical sync is active high
2153	                      </para></listitem>
2154	                    </varlistentry>
2155	                    <varlistentry>
2156	                      <term>DRM_MODE_FLAG_NVSYNC</term>
2157	                      <listitem><para>
2158	                        Vertical sync is active low
2159	                      </para></listitem>
2160	                    </varlistentry>
2161	                    <varlistentry>
2162	                      <term>DRM_MODE_FLAG_INTERLACE</term>
2163	                      <listitem><para>
2164	                        Mode is interlaced
2165	                      </para></listitem>
2166	                    </varlistentry>
2167	                    <varlistentry>
2168	                      <term>DRM_MODE_FLAG_DBLSCAN</term>
2169	                      <listitem><para>
2170	                        Mode uses doublescan
2171	                      </para></listitem>
2172	                    </varlistentry>
2173	                    <varlistentry>
2174	                      <term>DRM_MODE_FLAG_CSYNC</term>
2175	                      <listitem><para>
2176	                        Mode uses composite sync
2177	                      </para></listitem>
2178	                    </varlistentry>
2179	                    <varlistentry>
2180	                      <term>DRM_MODE_FLAG_PCSYNC</term>
2181	                      <listitem><para>
2182	                        Composite sync is active high
2183	                      </para></listitem>
2184	                    </varlistentry>
2185	                    <varlistentry>
2186	                      <term>DRM_MODE_FLAG_NCSYNC</term>
2187	                      <listitem><para>
2188	                        Composite sync is active low
2189	                      </para></listitem>
2190	                    </varlistentry>
2191	                    <varlistentry>
2192	                      <term>DRM_MODE_FLAG_HSKEW</term>
2193	                      <listitem><para>
2194	                        hskew provided (not used?)
2195	                      </para></listitem>
2196	                    </varlistentry>
2197	                    <varlistentry>
2198	                      <term>DRM_MODE_FLAG_BCAST</term>
2199	                      <listitem><para>
2200	                        not used?
2201	                      </para></listitem>
2202	                    </varlistentry>
2203	                    <varlistentry>
2204	                      <term>DRM_MODE_FLAG_PIXMUX</term>
2205	                      <listitem><para>
2206	                        not used?
2207	                      </para></listitem>
2208	                    </varlistentry>
2209	                    <varlistentry>
2210	                      <term>DRM_MODE_FLAG_DBLCLK</term>
2211	                      <listitem><para>
2212	                        not used?
2213	                      </para></listitem>
2214	                    </varlistentry>
2215	                    <varlistentry>
2216	                      <term>DRM_MODE_FLAG_CLKDIV2</term>
2217	                      <listitem><para>
2218	                        ?
2219	                      </para></listitem>
2220	                    </varlistentry>
2221	                  </variablelist>
2222	                </para>
2223	                <para>
2224	                  Note that modes marked with the INTERLACE or DBLSCAN flags will be
2225	                  filtered out by
2226	                  <function>drm_helper_probe_single_connector_modes</function> if
2227	                  the connector's <structfield>interlace_allowed</structfield> or
2228	                  <structfield>doublescan_allowed</structfield> field is set to 0.
2229	                </para>
2230	              </listitem>
2231	              <listitem>
2232	                <synopsis>char name[DRM_DISPLAY_MODE_LEN];</synopsis>
2233	                <para>
2234	                  Mode name. The driver must call
2235	                  <function>drm_mode_set_name</function> to fill the mode name from
2236	                  <structfield>hdisplay</structfield>,
2237	                  <structfield>vdisplay</structfield> and interlace flag after
2238	                  filling the corresponding fields.
2239	                </para>
2240	              </listitem>
2241	            </itemizedlist>
2242	          </para>
2243	          <para>
2244	            The <structfield>vrefresh</structfield> value is computed by
2245	            <function>drm_helper_probe_single_connector_modes</function>.
2246	          </para>
2247	          <para>
2248	            When parsing EDID data, <function>drm_add_edid_modes</function> fill the
2249	            connector <structfield>display_info</structfield>
2250	            <structfield>width_mm</structfield> and
2251	            <structfield>height_mm</structfield> fields. When creating modes
2252	            manually the <methodname>get_modes</methodname> helper operation must
2253	            set the <structfield>display_info</structfield>
2254	            <structfield>width_mm</structfield> and
2255	            <structfield>height_mm</structfield> fields if they haven't been set
2256	            already (for instance at initialization time when a fixed-size panel is
2257	            attached to the connector). The mode <structfield>width_mm</structfield>
2258	            and <structfield>height_mm</structfield> fields are only used internally
2259	            during EDID parsing and should not be set when creating modes manually.
2260	          </para>
2261	        </listitem>
2262	        <listitem>
2263	          <synopsis>int (*mode_valid)(struct drm_connector *connector,
2264			  struct drm_display_mode *mode);</synopsis>
2265	          <para>
2266	            Verify whether a mode is valid for the connector. Return MODE_OK for
2267	            supported modes and one of the enum drm_mode_status values (MODE_*)
2268	            for unsupported modes. This operation is mandatory.
2269	          </para>
2270	          <para>
2271	            As the mode rejection reason is currently not used beside for
2272	            immediately removing the unsupported mode, an implementation can
2273	            return MODE_BAD regardless of the exact reason why the mode is not
2274	            valid.
2275	          </para>
2276	          <note><para>
2277	            Note that the <methodname>mode_valid</methodname> helper operation is
2278	            only called for modes detected by the device, and
2279	            <emphasis>not</emphasis> for modes set by the user through the CRTC
2280	            <methodname>set_config</methodname> operation.
2281	          </para></note>
2282	        </listitem>
2283	      </itemizedlist>
2284	    </sect2>
2285	    <sect2>
2286	      <title>Modeset Helper Functions Reference</title>
2287	!Edrivers/gpu/drm/drm_crtc_helper.c
2288	    </sect2>
2289	    <sect2>
2290	      <title>Output Probing Helper Functions Reference</title>
2291	!Pdrivers/gpu/drm/drm_probe_helper.c output probing helper overview
2292	!Edrivers/gpu/drm/drm_probe_helper.c
2293	    </sect2>
2294	    <sect2>
2295	      <title>fbdev Helper Functions Reference</title>
2296	!Pdrivers/gpu/drm/drm_fb_helper.c fbdev helpers
2297	!Edrivers/gpu/drm/drm_fb_helper.c
2298	!Iinclude/drm/drm_fb_helper.h
2299	    </sect2>
2300	    <sect2>
2301	      <title>Display Port Helper Functions Reference</title>
2302	!Pdrivers/gpu/drm/drm_dp_helper.c dp helpers
2303	!Iinclude/drm/drm_dp_helper.h
2304	!Edrivers/gpu/drm/drm_dp_helper.c
2305	    </sect2>
2306	    <sect2>
2307	      <title>EDID Helper Functions Reference</title>
2308	!Edrivers/gpu/drm/drm_edid.c
2309	    </sect2>
2310	    <sect2>
2311	      <title>Rectangle Utilities Reference</title>
2312	!Pinclude/drm/drm_rect.h rect utils
2313	!Iinclude/drm/drm_rect.h
2314	!Edrivers/gpu/drm/drm_rect.c
2315	    </sect2>
2316	    <sect2>
2317	      <title>Flip-work Helper Reference</title>
2318	!Pinclude/drm/drm_flip_work.h flip utils
2319	!Iinclude/drm/drm_flip_work.h
2320	!Edrivers/gpu/drm/drm_flip_work.c
2321	    </sect2>
2322	    <sect2>
2323	      <title>HDMI Infoframes Helper Reference</title>
2324	      <para>
2325		Strictly speaking this is not a DRM helper library but generally useable
2326		by any driver interfacing with HDMI outputs like v4l or alsa drivers.
2327		But it nicely fits into the overall topic of mode setting helper
2328		libraries and hence is also included here.
2329	      </para>
2330	!Iinclude/linux/hdmi.h
2331	!Edrivers/video/hdmi.c
2332	    </sect2>
2333	    <sect2>
2334	      <title id="drm-kms-planehelpers">Plane Helper Reference</title>
2335	!Edrivers/gpu/drm/drm_plane_helper.c Plane Helpers
2336	    </sect2>
2337	  </sect1>
2339	  <!-- Internals: kms properties -->
2341	  <sect1 id="drm-kms-properties">
2342	    <title>KMS Properties</title>
2343	    <para>
2344	      Drivers may need to expose additional parameters to applications than
2345	      those described in the previous sections. KMS supports attaching
2346	      properties to CRTCs, connectors and planes and offers a userspace API to
2347	      list, get and set the property values.
2348	    </para>
2349	    <para>
2350	      Properties are identified by a name that uniquely defines the property
2351	      purpose, and store an associated value. For all property types except blob
2352	      properties the value is a 64-bit unsigned integer.
2353	    </para>
2354	    <para>
2355	      KMS differentiates between properties and property instances. Drivers
2356	      first create properties and then create and associate individual instances
2357	      of those properties to objects. A property can be instantiated multiple
2358	      times and associated with different objects. Values are stored in property
2359	      instances, and all other property information are stored in the property
2360	      and shared between all instances of the property.
2361	    </para>
2362	    <para>
2363	      Every property is created with a type that influences how the KMS core
2364	      handles the property. Supported property types are
2365	      <variablelist>
2366	        <varlistentry>
2367	          <term>DRM_MODE_PROP_RANGE</term>
2368	          <listitem><para>Range properties report their minimum and maximum
2369	            admissible values. The KMS core verifies that values set by
2370	            application fit in that range.</para></listitem>
2371	        </varlistentry>
2372	        <varlistentry>
2373	          <term>DRM_MODE_PROP_ENUM</term>
2374	          <listitem><para>Enumerated properties take a numerical value that
2375	            ranges from 0 to the number of enumerated values defined by the
2376	            property minus one, and associate a free-formed string name to each
2377	            value. Applications can retrieve the list of defined value-name pairs
2378	            and use the numerical value to get and set property instance values.
2379	            </para></listitem>
2380	        </varlistentry>
2381	        <varlistentry>
2382	          <term>DRM_MODE_PROP_BITMASK</term>
2383	          <listitem><para>Bitmask properties are enumeration properties that
2384	            additionally restrict all enumerated values to the 0..63 range.
2385	            Bitmask property instance values combine one or more of the
2386	            enumerated bits defined by the property.</para></listitem>
2387	        </varlistentry>
2388	        <varlistentry>
2389	          <term>DRM_MODE_PROP_BLOB</term>
2390	          <listitem><para>Blob properties store a binary blob without any format
2391	            restriction. The binary blobs are created as KMS standalone objects,
2392	            and blob property instance values store the ID of their associated
2393	            blob object.</para>
2394		    <para>Blob properties are only used for the connector EDID property
2395		    and cannot be created by drivers.</para></listitem>
2396	        </varlistentry>
2397	      </variablelist>
2398	    </para>
2399	    <para>
2400	      To create a property drivers call one of the following functions depending
2401	      on the property type. All property creation functions take property flags
2402	      and name, as well as type-specific arguments.
2403	      <itemizedlist>
2404	        <listitem>
2405	          <synopsis>struct drm_property *drm_property_create_range(struct drm_device *dev, int flags,
2406	                                               const char *name,
2407	                                               uint64_t min, uint64_t max);</synopsis>
2408	          <para>Create a range property with the given minimum and maximum
2409	            values.</para>
2410	        </listitem>
2411	        <listitem>
2412	          <synopsis>struct drm_property *drm_property_create_enum(struct drm_device *dev, int flags,
2413	                                              const char *name,
2414	                                              const struct drm_prop_enum_list *props,
2415	                                              int num_values);</synopsis>
2416	          <para>Create an enumerated property. The <parameter>props</parameter>
2417	            argument points to an array of <parameter>num_values</parameter>
2418	            value-name pairs.</para>
2419	        </listitem>
2420	        <listitem>
2421	          <synopsis>struct drm_property *drm_property_create_bitmask(struct drm_device *dev,
2422	                                                 int flags, const char *name,
2423	                                                 const struct drm_prop_enum_list *props,
2424	                                                 int num_values);</synopsis>
2425	          <para>Create a bitmask property. The <parameter>props</parameter>
2426	            argument points to an array of <parameter>num_values</parameter>
2427	            value-name pairs.</para>
2428	        </listitem>
2429	      </itemizedlist>
2430	    </para>
2431	    <para>
2432	      Properties can additionally be created as immutable, in which case they
2433	      will be read-only for applications but can be modified by the driver. To
2434	      create an immutable property drivers must set the DRM_MODE_PROP_IMMUTABLE
2435	      flag at property creation time.
2436	    </para>
2437	    <para>
2438	      When no array of value-name pairs is readily available at property
2439	      creation time for enumerated or range properties, drivers can create
2440	      the property using the <function>drm_property_create</function> function
2441	      and manually add enumeration value-name pairs by calling the
2442	      <function>drm_property_add_enum</function> function. Care must be taken to
2443	      properly specify the property type through the <parameter>flags</parameter>
2444	      argument.
2445	    </para>
2446	    <para>
2447	      After creating properties drivers can attach property instances to CRTC,
2448	      connector and plane objects by calling the
2449	      <function>drm_object_attach_property</function>. The function takes a
2450	      pointer to the target object, a pointer to the previously created property
2451	      and an initial instance value.
2452	    </para>
2453	  </sect1>
2455	  <!-- Internals: vertical blanking -->
2457	  <sect1 id="drm-vertical-blank">
2458	    <title>Vertical Blanking</title>
2459	    <para>
2460	      Vertical blanking plays a major role in graphics rendering. To achieve
2461	      tear-free display, users must synchronize page flips and/or rendering to
2462	      vertical blanking. The DRM API offers ioctls to perform page flips
2463	      synchronized to vertical blanking and wait for vertical blanking.
2464	    </para>
2465	    <para>
2466	      The DRM core handles most of the vertical blanking management logic, which
2467	      involves filtering out spurious interrupts, keeping race-free blanking
2468	      counters, coping with counter wrap-around and resets and keeping use
2469	      counts. It relies on the driver to generate vertical blanking interrupts
2470	      and optionally provide a hardware vertical blanking counter. Drivers must
2471	      implement the following operations.
2472	    </para>
2473	    <itemizedlist>
2474	      <listitem>
2475	        <synopsis>int (*enable_vblank) (struct drm_device *dev, int crtc);
2476	void (*disable_vblank) (struct drm_device *dev, int crtc);</synopsis>
2477	        <para>
2478		  Enable or disable vertical blanking interrupts for the given CRTC.
2479		</para>
2480	      </listitem>
2481	      <listitem>
2482	        <synopsis>u32 (*get_vblank_counter) (struct drm_device *dev, int crtc);</synopsis>
2483	        <para>
2484		  Retrieve the value of the vertical blanking counter for the given
2485		  CRTC. If the hardware maintains a vertical blanking counter its value
2486		  should be returned. Otherwise drivers can use the
2487		  <function>drm_vblank_count</function> helper function to handle this
2488		  operation.
2489		</para>
2490	      </listitem>
2491	    </itemizedlist>
2492	    <para>
2493	      Drivers must initialize the vertical blanking handling core with a call to
2494	      <function>drm_vblank_init</function> in their
2495	      <methodname>load</methodname> operation. The function will set the struct
2496	      <structname>drm_device</structname>
2497	      <structfield>vblank_disable_allowed</structfield> field to 0. This will
2498	      keep vertical blanking interrupts enabled permanently until the first mode
2499	      set operation, where <structfield>vblank_disable_allowed</structfield> is
2500	      set to 1. The reason behind this is not clear. Drivers can set the field
2501	      to 1 after <function>calling drm_vblank_init</function> to make vertical
2502	      blanking interrupts dynamically managed from the beginning.
2503	    </para>
2504	    <para>
2505	      Vertical blanking interrupts can be enabled by the DRM core or by drivers
2506	      themselves (for instance to handle page flipping operations). The DRM core
2507	      maintains a vertical blanking use count to ensure that the interrupts are
2508	      not disabled while a user still needs them. To increment the use count,
2509	      drivers call <function>drm_vblank_get</function>. Upon return vertical
2510	      blanking interrupts are guaranteed to be enabled.
2511	    </para>
2512	    <para>
2513	      To decrement the use count drivers call
2514	      <function>drm_vblank_put</function>. Only when the use count drops to zero
2515	      will the DRM core disable the vertical blanking interrupts after a delay
2516	      by scheduling a timer. The delay is accessible through the vblankoffdelay
2517	      module parameter or the <varname>drm_vblank_offdelay</varname> global
2518	      variable and expressed in milliseconds. Its default value is 5000 ms.
2519	    </para>
2520	    <para>
2521	      When a vertical blanking interrupt occurs drivers only need to call the
2522	      <function>drm_handle_vblank</function> function to account for the
2523	      interrupt.
2524	    </para>
2525	    <para>
2526	      Resources allocated by <function>drm_vblank_init</function> must be freed
2527	      with a call to <function>drm_vblank_cleanup</function> in the driver
2528	      <methodname>unload</methodname> operation handler.
2529	    </para>
2530	  </sect1>
2532	  <!-- Internals: open/close, file operations and ioctls -->
2534	  <sect1>
2535	    <title>Open/Close, File Operations and IOCTLs</title>
2536	    <sect2>
2537	      <title>Open and Close</title>
2538	      <synopsis>int (*firstopen) (struct drm_device *);
2539	void (*lastclose) (struct drm_device *);
2540	int (*open) (struct drm_device *, struct drm_file *);
2541	void (*preclose) (struct drm_device *, struct drm_file *);
2542	void (*postclose) (struct drm_device *, struct drm_file *);</synopsis>
2543	      <abstract>Open and close handlers. None of those methods are mandatory.
2544	      </abstract>
2545	      <para>
2546	        The <methodname>firstopen</methodname> method is called by the DRM core
2547		for legacy UMS (User Mode Setting) drivers only when an application
2548		opens a device that has no other opened file handle. UMS drivers can
2549		implement it to acquire device resources. KMS drivers can't use the
2550		method and must acquire resources in the <methodname>load</methodname>
2551		method instead.
2552	      </para>
2553	      <para>
2554		Similarly the <methodname>lastclose</methodname> method is called when
2555		the last application holding a file handle opened on the device closes
2556		it, for both UMS and KMS drivers. Additionally, the method is also
2557		called at module unload time or, for hot-pluggable devices, when the
2558		device is unplugged. The <methodname>firstopen</methodname> and
2559		<methodname>lastclose</methodname> calls can thus be unbalanced.
2560	      </para>
2561	      <para>
2562	        The <methodname>open</methodname> method is called every time the device
2563		is opened by an application. Drivers can allocate per-file private data
2564		in this method and store them in the struct
2565		<structname>drm_file</structname> <structfield>driver_priv</structfield>
2566		field. Note that the <methodname>open</methodname> method is called
2567		before <methodname>firstopen</methodname>.
2568	      </para>
2569	      <para>
2570	        The close operation is split into <methodname>preclose</methodname> and
2571		<methodname>postclose</methodname> methods. Drivers must stop and
2572		cleanup all per-file operations in the <methodname>preclose</methodname>
2573		method. For instance pending vertical blanking and page flip events must
2574		be cancelled. No per-file operation is allowed on the file handle after
2575		returning from the <methodname>preclose</methodname> method.
2576	      </para>
2577	      <para>
2578	        Finally the <methodname>postclose</methodname> method is called as the
2579		last step of the close operation, right before calling the
2580		<methodname>lastclose</methodname> method if no other open file handle
2581		exists for the device. Drivers that have allocated per-file private data
2582		in the <methodname>open</methodname> method should free it here.
2583	      </para>
2584	      <para>
2585	        The <methodname>lastclose</methodname> method should restore CRTC and
2586		plane properties to default value, so that a subsequent open of the
2587		device will not inherit state from the previous user. It can also be
2588		used to execute delayed power switching state changes, e.g. in
2589		conjunction with the vga-switcheroo infrastructure. Beyond that KMS
2590		drivers should not do any further cleanup. Only legacy UMS drivers might
2591		need to clean up device state so that the vga console or an independent
2592		fbdev driver could take over.
2593	      </para>
2594	    </sect2>
2595	    <sect2>
2596	      <title>File Operations</title>
2597	      <synopsis>const struct file_operations *fops</synopsis>
2598	      <abstract>File operations for the DRM device node.</abstract>
2599	      <para>
2600	        Drivers must define the file operations structure that forms the DRM
2601		userspace API entry point, even though most of those operations are
2602		implemented in the DRM core. The <methodname>open</methodname>,
2603		<methodname>release</methodname> and <methodname>ioctl</methodname>
2604		operations are handled by
2605		<programlisting>
2606		.owner = THIS_MODULE,
2607		.open = drm_open,
2608		.release = drm_release,
2609		.unlocked_ioctl = drm_ioctl,
2610	  #ifdef CONFIG_COMPAT
2611		.compat_ioctl = drm_compat_ioctl,
2612	  #endif
2613	        </programlisting>
2614	      </para>
2615	      <para>
2616	        Drivers that implement private ioctls that requires 32/64bit
2617		compatibility support must provide their own
2618		<methodname>compat_ioctl</methodname> handler that processes private
2619		ioctls and calls <function>drm_compat_ioctl</function> for core ioctls.
2620	      </para>
2621	      <para>
2622	        The <methodname>read</methodname> and <methodname>poll</methodname>
2623		operations provide support for reading DRM events and polling them. They
2624		are implemented by
2625		<programlisting>
2626		.poll = drm_poll,
2627		.read = drm_read,
2628		.llseek = no_llseek,
2629		</programlisting>
2630	      </para>
2631	      <para>
2632	        The memory mapping implementation varies depending on how the driver
2633		manages memory. Pre-GEM drivers will use <function>drm_mmap</function>,
2634		while GEM-aware drivers will use <function>drm_gem_mmap</function>. See
2635		<xref linkend="drm-gem"/>.
2636		<programlisting>
2637		.mmap = drm_gem_mmap,
2638		</programlisting>
2639	      </para>
2640	      <para>
2641	        No other file operation is supported by the DRM API.
2642	      </para>
2643	    </sect2>
2644	    <sect2>
2645	      <title>IOCTLs</title>
2646	      <synopsis>struct drm_ioctl_desc *ioctls;
2647	int num_ioctls;</synopsis>
2648	      <abstract>Driver-specific ioctls descriptors table.</abstract>
2649	      <para>
2650	        Driver-specific ioctls numbers start at DRM_COMMAND_BASE. The ioctls
2651		descriptors table is indexed by the ioctl number offset from the base
2652		value. Drivers can use the DRM_IOCTL_DEF_DRV() macro to initialize the
2653		table entries.
2654	      </para>
2655	      <para>
2656	        <programlisting>DRM_IOCTL_DEF_DRV(ioctl, func, flags)</programlisting>
2657		<para>
2658		  <parameter>ioctl</parameter> is the ioctl name. Drivers must define
2659		  the DRM_##ioctl and DRM_IOCTL_##ioctl macros to the ioctl number
2660		  offset from DRM_COMMAND_BASE and the ioctl number respectively. The
2661		  first macro is private to the device while the second must be exposed
2662		  to userspace in a public header.
2663		</para>
2664		<para>
2665		  <parameter>func</parameter> is a pointer to the ioctl handler function
2666		  compatible with the <type>drm_ioctl_t</type> type.
2667		  <programlisting>typedef int drm_ioctl_t(struct drm_device *dev, void *data,
2668			struct drm_file *file_priv);</programlisting>
2669		</para>
2670		<para>
2671		  <parameter>flags</parameter> is a bitmask combination of the following
2672		  values. It restricts how the ioctl is allowed to be called.
2673		  <itemizedlist>
2674		    <listitem><para>
2675		      DRM_AUTH - Only authenticated callers allowed
2676		    </para></listitem>
2677		    <listitem><para>
2678		      DRM_MASTER - The ioctl can only be called on the master file
2679		      handle
2680		    </para></listitem>
2681	            <listitem><para>
2682		      DRM_ROOT_ONLY - Only callers with the SYSADMIN capability allowed
2683		    </para></listitem>
2684	            <listitem><para>
2685		      DRM_CONTROL_ALLOW - The ioctl can only be called on a control
2686		      device
2687		    </para></listitem>
2688	            <listitem><para>
2689		      DRM_UNLOCKED - The ioctl handler will be called without locking
2690		      the DRM global mutex
2691		    </para></listitem>
2692		  </itemizedlist>
2693		</para>
2694	      </para>
2695	    </sect2>
2696	  </sect1>
2697	  <sect1>
2698	    <title>Legacy Support Code</title>
2699	    <para>
2700	      The section very briefly covers some of the old legacy support code which
2701	      is only used by old DRM drivers which have done a so-called shadow-attach
2702	      to the underlying device instead of registering as a real driver. This
2703	      also includes some of the old generic buffer management and command
2704	      submission code. Do not use any of this in new and modern drivers.
2705	    </para>
2707	    <sect2>
2708	      <title>Legacy Suspend/Resume</title>
2709	      <para>
2710		The DRM core provides some suspend/resume code, but drivers wanting full
2711		suspend/resume support should provide save() and restore() functions.
2712		These are called at suspend, hibernate, or resume time, and should perform
2713		any state save or restore required by your device across suspend or
2714		hibernate states.
2715	      </para>
2716	      <synopsis>int (*suspend) (struct drm_device *, pm_message_t state);
2717	  int (*resume) (struct drm_device *);</synopsis>
2718	      <para>
2719		Those are legacy suspend and resume methods which
2720		<emphasis>only</emphasis> work with the legacy shadow-attach driver
2721		registration functions. New driver should use the power management
2722		interface provided by their bus type (usually through
2723		the struct <structname>device_driver</structname> dev_pm_ops) and set
2724		these methods to NULL.
2725	      </para>
2726	    </sect2>
2728	    <sect2>
2729	      <title>Legacy DMA Services</title>
2730	      <para>
2731		This should cover how DMA mapping etc. is supported by the core.
2732		These functions are deprecated and should not be used.
2733	      </para>
2734	    </sect2>
2735	  </sect1>
2736	  </chapter>
2738	<!-- TODO
2740	- Add a glossary
2741	- Document the struct_mutex catch-all lock
2742	- Document connector properties
2744	- Why is the load method optional?
2745	- What are drivers supposed to set the initial display state to, and how?
2746	  Connector's DPMS states are not initialized and are thus equal to
2747	  DRM_MODE_DPMS_ON. The fbcon compatibility layer calls
2748	  drm_helper_disable_unused_functions(), which disables unused encoders and
2749	  CRTCs, but doesn't touch the connectors' DPMS state, and
2750	  drm_helper_connector_dpms() in reaction to fbdev blanking events. Do drivers
2751	  that don't implement (or just don't use) fbcon compatibility need to call
2752	  those functions themselves?
2753	- KMS drivers must call drm_vblank_pre_modeset() and drm_vblank_post_modeset()
2754	  around mode setting. Should this be done in the DRM core?
2755	- vblank_disable_allowed is set to 1 in the first drm_vblank_post_modeset()
2756	  call and never set back to 0. It seems to be safe to permanently set it to 1
2757	  in drm_vblank_init() for KMS driver, and it might be safe for UMS drivers as
2758	  well. This should be investigated.
2759	- crtc and connector .save and .restore operations are only used internally in
2760	  drivers, should they be removed from the core?
2761	- encoder mid-layer .save and .restore operations are only used internally in
2762	  drivers, should they be removed from the core?
2763	- encoder mid-layer .detect operation is only used internally in drivers,
2764	  should it be removed from the core?
2765	-->
2767	  <!-- External interfaces -->
2769	  <chapter id="drmExternals">
2770	    <title>Userland interfaces</title>
2771	    <para>
2772	      The DRM core exports several interfaces to applications,
2773	      generally intended to be used through corresponding libdrm
2774	      wrapper functions.  In addition, drivers export device-specific
2775	      interfaces for use by userspace drivers &amp; device-aware
2776	      applications through ioctls and sysfs files.
2777	    </para>
2778	    <para>
2779	      External interfaces include: memory mapping, context management,
2780	      DMA operations, AGP management, vblank control, fence
2781	      management, memory management, and output management.
2782	    </para>
2783	    <para>
2784	      Cover generic ioctls and sysfs layout here.  We only need high-level
2785	      info, since man pages should cover the rest.
2786	    </para>
2788	  <!-- External: render nodes -->
2790	    <sect1>
2791	      <title>Render nodes</title>
2792	      <para>
2793	        DRM core provides multiple character-devices for user-space to use.
2794	        Depending on which device is opened, user-space can perform a different
2795	        set of operations (mainly ioctls). The primary node is always created
2796	        and called card&lt;num&gt;. Additionally, a currently
2797	        unused control node, called controlD&lt;num&gt; is also
2798	        created. The primary node provides all legacy operations and
2799	        historically was the only interface used by userspace. With KMS, the
2800	        control node was introduced. However, the planned KMS control interface
2801	        has never been written and so the control node stays unused to date.
2802	      </para>
2803	      <para>
2804	        With the increased use of offscreen renderers and GPGPU applications,
2805	        clients no longer require running compositors or graphics servers to
2806	        make use of a GPU. But the DRM API required unprivileged clients to
2807	        authenticate to a DRM-Master prior to getting GPU access. To avoid this
2808	        step and to grant clients GPU access without authenticating, render
2809	        nodes were introduced. Render nodes solely serve render clients, that
2810	        is, no modesetting or privileged ioctls can be issued on render nodes.
2811	        Only non-global rendering commands are allowed. If a driver supports
2812	        render nodes, it must advertise it via the DRIVER_RENDER
2813	        DRM driver capability. If not supported, the primary node must be used
2814	        for render clients together with the legacy drmAuth authentication
2815	        procedure.
2816	      </para>
2817	      <para>
2818	        If a driver advertises render node support, DRM core will create a
2819	        separate render node called renderD&lt;num&gt;. There will
2820	        be one render node per device. No ioctls except  PRIME-related ioctls
2821	        will be allowed on this node. Especially GEM_OPEN will be
2822	        explicitly prohibited. Render nodes are designed to avoid the
2823	        buffer-leaks, which occur if clients guess the flink names or mmap
2824	        offsets on the legacy interface. Additionally to this basic interface,
2825	        drivers must mark their driver-dependent render-only ioctls as
2826	        DRM_RENDER_ALLOW so render clients can use them. Driver
2827	        authors must be careful not to allow any privileged ioctls on render
2828	        nodes.
2829	      </para>
2830	      <para>
2831	        With render nodes, user-space can now control access to the render node
2832	        via basic file-system access-modes. A running graphics server which
2833	        authenticates clients on the privileged primary/legacy node is no longer
2834	        required. Instead, a client can open the render node and is immediately
2835	        granted GPU access. Communication between clients (or servers) is done
2836	        via PRIME. FLINK from render node to legacy node is not supported. New
2837	        clients must not use the insecure FLINK interface.
2838	      </para>
2839	      <para>
2840	        Besides dropping all modeset/global ioctls, render nodes also drop the
2841	        DRM-Master concept. There is no reason to associate render clients with
2842	        a DRM-Master as they are independent of any graphics server. Besides,
2843	        they must work without any running master, anyway.
2844	        Drivers must be able to run without a master object if they support
2845	        render nodes. If, on the other hand, a driver requires shared state
2846	        between clients which is visible to user-space and accessible beyond
2847	        open-file boundaries, they cannot support render nodes.
2848	      </para>
2849	    </sect1>
2851	  <!-- External: vblank handling -->
2853	    <sect1>
2854	      <title>VBlank event handling</title>
2855	      <para>
2856	        The DRM core exposes two vertical blank related ioctls:
2857	        <variablelist>
2858	          <varlistentry>
2859	            <term>DRM_IOCTL_WAIT_VBLANK</term>
2860	            <listitem>
2861	              <para>
2862	                This takes a struct drm_wait_vblank structure as its argument,
2863	                and it is used to block or request a signal when a specified
2864	                vblank event occurs.
2865	              </para>
2866	            </listitem>
2867	          </varlistentry>
2868	          <varlistentry>
2869	            <term>DRM_IOCTL_MODESET_CTL</term>
2870	            <listitem>
2871	              <para>
2872	                This should be called by application level drivers before and
2873	                after mode setting, since on many devices the vertical blank
2874	                counter is reset at that time.  Internally, the DRM snapshots
2875	                the last vblank count when the ioctl is called with the
2876	                _DRM_PRE_MODESET command, so that the counter won't go backwards
2877	                (which is dealt with when _DRM_POST_MODESET is used).
2878	              </para>
2879	            </listitem>
2880	          </varlistentry>
2881	        </variablelist>
2882	<!--!Edrivers/char/drm/drm_irq.c-->
2883	      </para>
2884	    </sect1>
2886	  </chapter>
2887	</part>
2888	<part id="drmDrivers">
2889	  <title>DRM Drivers</title>
2891	  <partintro>
2892	    <para>
2893	      This second part of the DRM Developer's Guide documents driver code,
2894	      implementation details and also all the driver-specific userspace
2895	      interfaces. Especially since all hardware-acceleration interfaces to
2896	      userspace are driver specific for efficiency and other reasons these
2897	      interfaces can be rather substantial. Hence every driver has its own
2898	      chapter.
2899	    </para>
2900	  </partintro>
2902	  <chapter id="drmI915">
2903	    <title>drm/i915 Intel GFX Driver</title>
2904	    <para>
2905	      The drm/i915 driver supports all (with the exception of some very early
2906	      models) integrated GFX chipsets with both Intel display and rendering
2907	      blocks. This excludes a set of SoC platforms with an SGX rendering unit,
2908	      those have basic support through the gma500 drm driver.
2909	    </para>
2910	    <sect1>
2911	      <title>Display Hardware Handling</title>
2912	      <para>
2913	        This section covers everything related to the display hardware including
2914	        the mode setting infrastructure, plane, sprite and cursor handling and
2915	        display, output probing and related topics.
2916	      </para>
2917	      <sect2>
2918	        <title>Mode Setting Infrastructure</title>
2919	        <para>
2920	          The i915 driver is thus far the only DRM driver which doesn't use the
2921	          common DRM helper code to implement mode setting sequences. Thus it
2922	          has its own tailor-made infrastructure for executing a display
2923	          configuration change.
2924	        </para>
2925	      </sect2>
2926	      <sect2>
2927	        <title>Plane Configuration</title>
2928	        <para>
2929		  This section covers plane configuration and composition with the
2930		  primary plane, sprites, cursors and overlays. This includes the
2931		  infrastructure to do atomic vsync'ed updates of all this state and
2932		  also tightly coupled topics like watermark setup and computation,
2933		  framebuffer compression and panel self refresh.
2934	        </para>
2935	      </sect2>
2936	      <sect2>
2937	        <title>Output Probing</title>
2938	        <para>
2939		  This section covers output probing and related infrastructure like the
2940		  hotplug interrupt storm detection and mitigation code. Note that the
2941		  i915 driver still uses most of the common DRM helper code for output
2942		  probing, so those sections fully apply.
2943	        </para>
2944	      </sect2>
2945	    </sect1>
2947	    <sect1>
2948	      <title>Memory Management and Command Submission</title>
2949	      <para>
2950		This sections covers all things related to the GEM implementation in the
2951		i915 driver.
2952	      </para>
2953	    </sect1>
2954	  </chapter>
2955	</part>
2956	</book>
Hide Line Numbers
About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Information is copyright its respective author. All material is available from the Linux Kernel Source distributed under a GPL License. This page is provided as a free service by mjmwired.net.