7 This section contains a list of smaller janitorial tasks in the kernel DRM
8 graphics subsystem useful as newbie projects. Or for slow rainy days.
13 To make it easier task are categorized into different levels:
15 Starter: Good tasks to get started with the DRM subsystem.
17 Intermediate: Tasks which need some experience with working in the DRM
18 subsystem, or some specific GPU/display graphics knowledge. For debugging issue
19 it's good to have the relevant hardware (or a virtual driver set up) available
22 Advanced: Tricky tasks that need fairly good understanding of the DRM subsystem
23 and graphics topics. Generally need the relevant hardware for development and
26 Expert: Only attempt these if you've successfully completed some tricky
27 refactorings already and are an expert in the specific area
29 Subsystem-wide refactorings
30 ===========================
32 Remove custom dumb_map_offset implementations
33 ---------------------------------------------
35 All GEM based drivers should be using drm_gem_create_mmap_offset() instead.
36 Audit each individual driver, make sure it'll work with the generic
37 implementation (there's lots of outdated locking leftovers in various
38 implementations), and then remove it.
40 Contact: Daniel Vetter, respective driver maintainers
44 Convert existing KMS drivers to atomic modesetting
45 --------------------------------------------------
47 3.19 has the atomic modeset interfaces and helpers, so drivers can now be
48 converted over. Modern compositors like Wayland or Surfaceflinger on Android
49 really want an atomic modeset interface, so this is all about the bright
52 There is a conversion guide for atomic [1]_ and all you need is a GPU for a
53 non-converted driver. The "Atomic mode setting design overview" series [2]_
54 [3]_ at LWN.net can also be helpful.
56 As part of this drivers also need to convert to universal plane (which means
57 exposing primary & cursor as proper plane objects). But that's much easier to
58 do by directly using the new atomic helper driver callbacks.
60 .. [1] https://blog.ffwll.ch/2014/11/atomic-modeset-support-for-kms-drivers.html
61 .. [2] https://lwn.net/Articles/653071/
62 .. [3] https://lwn.net/Articles/653466/
64 Contact: Daniel Vetter, respective driver maintainers
68 Clean up the clipped coordination confusion around planes
69 ---------------------------------------------------------
71 We have a helper to get this right with drm_plane_helper_check_update(), but
72 it's not consistently used. This should be fixed, preferably in the atomic
73 helpers (and drivers then moved over to clipped coordinates). Probably the
74 helper should also be moved from drm_plane_helper.c to the atomic helpers, to
75 avoid confusion - the other helpers in that file are all deprecated legacy
78 Contact: Ville Syrjälä, Daniel Vetter, driver maintainers
82 Improve plane atomic_check helpers
83 ----------------------------------
85 Aside from the clipped coordinates right above there's a few suboptimal things
86 with the current helpers:
88 - drm_plane_helper_funcs->atomic_check gets called for enabled or disabled
89 planes. At best this seems to confuse drivers, worst it means they blow up
90 when the plane is disabled without the CRTC. The only special handling is
91 resetting values in the plane state structures, which instead should be moved
92 into the drm_plane_funcs->atomic_duplicate_state functions.
94 - Once that's done, helpers could stop calling ->atomic_check for disabled
97 - Then we could go through all the drivers and remove the more-or-less confused
98 checks for plane_state->fb and plane_state->crtc.
100 Contact: Daniel Vetter
104 Convert early atomic drivers to async commit helpers
105 ----------------------------------------------------
107 For the first year the atomic modeset helpers didn't support asynchronous /
108 nonblocking commits, and every driver had to hand-roll them. This is fixed
109 now, but there's still a pile of existing drivers that easily could be
110 converted over to the new infrastructure.
112 One issue with the helpers is that they require that drivers handle completion
113 events for atomic commits correctly. But fixing these bugs is good anyway.
115 Somewhat related is the legacy_cursor_update hack, which should be replaced with
116 the new atomic_async_check/commit functionality in the helpers in drivers that
117 still look at that flag.
119 Contact: Daniel Vetter, respective driver maintainers
123 Fallout from atomic KMS
124 -----------------------
126 ``drm_atomic_helper.c`` provides a batch of functions which implement legacy
127 IOCTLs on top of the new atomic driver interface. Which is really nice for
128 gradual conversion of drivers, but unfortunately the semantic mismatches are
129 a bit too severe. So there's some follow-up work to adjust the function
130 interfaces to fix these issues:
132 * atomic needs the lock acquire context. At the moment that's passed around
133 implicitly with some horrible hacks, and it's also allocate with
134 ``GFP_NOFAIL`` behind the scenes. All legacy paths need to start allocating
135 the acquire context explicitly on stack and then also pass it down into
136 drivers explicitly so that the legacy-on-atomic functions can use them.
138 Except for some driver code this is done. This task should be finished by
139 adding WARN_ON(!drm_drv_uses_atomic_modeset) in drm_modeset_lock_all().
141 * A bunch of the vtable hooks are now in the wrong place: DRM has a split
142 between core vfunc tables (named ``drm_foo_funcs``), which are used to
143 implement the userspace ABI. And then there's the optional hooks for the
144 helper libraries (name ``drm_foo_helper_funcs``), which are purely for
145 internal use. Some of these hooks should be move from ``_funcs`` to
146 ``_helper_funcs`` since they are not part of the core ABI. There's a
147 ``FIXME`` comment in the kerneldoc for each such case in ``drm_crtc.h``.
149 Contact: Daniel Vetter
153 Get rid of dev->struct_mutex from GEM drivers
154 ---------------------------------------------
156 ``dev->struct_mutex`` is the Big DRM Lock from legacy days and infested
157 everything. Nowadays in modern drivers the only bit where it's mandatory is
158 serializing GEM buffer object destruction. Which unfortunately means drivers
159 have to keep track of that lock and either call ``unreference`` or
160 ``unreference_locked`` depending upon context.
162 Core GEM doesn't have a need for ``struct_mutex`` any more since kernel 4.8,
163 and there's a GEM object ``free`` callback for any drivers which are
164 entirely ``struct_mutex`` free.
166 For drivers that need ``struct_mutex`` it should be replaced with a driver-
167 private lock. The tricky part is the BO free functions, since those can't
168 reliably take that lock any more. Instead state needs to be protected with
169 suitable subordinate locks or some cleanup work pushed to a worker thread. For
170 performance-critical drivers it might also be better to go with a more
171 fine-grained per-buffer object and per-context lockings scheme. Currently only
172 the ``msm`` and `i915` drivers use ``struct_mutex``.
174 Contact: Daniel Vetter, respective driver maintainers
178 Move Buffer Object Locking to dma_resv_lock()
179 ---------------------------------------------
181 Many drivers have their own per-object locking scheme, usually using
182 mutex_lock(). This causes all kinds of trouble for buffer sharing, since
183 depending which driver is the exporter and importer, the locking hierarchy is
186 To solve this we need one standard per-object locking mechanism, which is
187 dma_resv_lock(). This lock needs to be called as the outermost lock, with all
188 other driver specific per-object locks removed. The problem is that rolling out
189 the actual change to the locking contract is a flag day, due to struct dma_buf
194 Convert logging to drm_* functions with drm_device parameter
195 ------------------------------------------------------------
197 For drivers which could have multiple instances, it is necessary to
198 differentiate between which is which in the logs. Since DRM_INFO/WARN/ERROR
199 don't do this, drivers used dev_info/warn/err to make this differentiation. We
200 now have drm_* variants of the drm print functions, so we can start to convert
201 those drivers back to using drm-formatted specific log messages.
203 Before you start this conversion please contact the relevant maintainers to make
204 sure your work will be merged - not everyone agrees that the DRM dmesg macros
207 Contact: Sean Paul, Maintainer of the driver you plan to convert
211 Convert drivers to use simple modeset suspend/resume
212 ----------------------------------------------------
214 Most drivers (except i915 and nouveau) that use
215 drm_atomic_helper_suspend/resume() can probably be converted to use
216 drm_mode_config_helper_suspend/resume(). Also there's still open-coded version
217 of the atomic suspend/resume code in older atomic modeset drivers.
219 Contact: Maintainer of the driver you plan to convert
223 Convert drivers to use drm_fbdev_generic_setup()
224 ------------------------------------------------
226 Most drivers can use drm_fbdev_generic_setup(). Driver have to implement
227 atomic modesetting and GEM vmap support. Historically, generic fbdev emulation
228 expected the framebuffer in system memory or system-like memory. By employing
229 struct iosys_map, drivers with frambuffers in I/O memory can be supported
232 Contact: Maintainer of the driver you plan to convert
236 Reimplement functions in drm_fbdev_fb_ops without fbdev
237 -------------------------------------------------------
239 A number of callback functions in drm_fbdev_fb_ops could benefit from
240 being rewritten without dependencies on the fbdev module. Some of the
241 helpers could further benefit from using struct iosys_map instead of
244 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Daniel Vetter
248 Benchmark and optimize blitting and format-conversion function
249 --------------------------------------------------------------
251 Drawing to display memory quickly is crucial for many applications'
254 On at least x86-64, sys_imageblit() is significantly slower than
255 cfb_imageblit(), even though both use the same blitting algorithm and
256 the latter is written for I/O memory. It turns out that cfb_imageblit()
257 uses movl instructions, while sys_imageblit apparently does not. This
258 seems to be a problem with gcc's optimizer. DRM's format-conversion
259 helpers might be subject to similar issues.
261 Benchmark and optimize fbdev's sys_() helpers and DRM's format-conversion
262 helpers. In cases that can be further optimized, maybe implement a different
263 algorithm. For micro-optimizations, use movl/movq instructions explicitly.
264 That might possibly require architecture-specific helpers (e.g., storel()
267 Contact: Thomas Zimmermann <tzimmermann@suse.de>
271 drm_framebuffer_funcs and drm_mode_config_funcs.fb_create cleanup
272 -----------------------------------------------------------------
274 A lot more drivers could be switched over to the drm_gem_framebuffer helpers.
277 - Need to switch over to the generic dirty tracking code using
278 drm_atomic_helper_dirtyfb first (e.g. qxl).
280 - Need to switch to drm_fbdev_generic_setup(), otherwise a lot of the custom fb
281 setup code can't be deleted.
283 - Need to switch to drm_gem_fb_create(), as now drm_gem_fb_create() checks for
284 valid formats for atomic drivers.
286 - Many drivers subclass drm_framebuffer, we'd need a embedding compatible
287 version of the varios drm_gem_fb_create functions. Maybe called
288 drm_gem_fb_create/_with_dirty/_with_funcs as needed.
290 Contact: Daniel Vetter
294 Generic fbdev defio support
295 ---------------------------
297 The defio support code in the fbdev core has some very specific requirements,
298 which means drivers need to have a special framebuffer for fbdev. The main
299 issue is that it uses some fields in struct page itself, which breaks shmem
300 gem objects (and other things). To support defio, affected drivers require
301 the use of a shadow buffer, which may add CPU and memory overhead.
303 Possible solution would be to write our own defio mmap code in the drm fbdev
304 emulation. It would need to fully wrap the existing mmap ops, forwarding
305 everything after it has done the write-protect/mkwrite trickery:
307 - In the drm_fbdev_fb_mmap helper, if we need defio, change the
308 default page prots to write-protected with something like this::
310 vma->vm_page_prot = pgprot_wrprotect(vma->vm_page_prot);
312 - Set the mkwrite and fsync callbacks with similar implementions to the core
313 fbdev defio stuff. These should all work on plain ptes, they don't actually
314 require a struct page. uff. These should all work on plain ptes, they don't
315 actually require a struct page.
317 - Track the dirty pages in a separate structure (bitfield with one bit per page
318 should work) to avoid clobbering struct page.
320 Might be good to also have some igt testcases for this.
322 Contact: Daniel Vetter, Noralf Tronnes
326 connector register/unregister fixes
327 -----------------------------------
329 - For most connectors it's a no-op to call drm_connector_register/unregister
330 directly from driver code, drm_dev_register/unregister take care of this
331 already. We can remove all of them.
333 - For dp drivers it's a bit more a mess, since we need the connector to be
334 registered when calling drm_dp_aux_register. Fix this by instead calling
335 drm_dp_aux_init, and moving the actual registering into a late_register
336 callback as recommended in the kerneldoc.
340 Remove load/unload callbacks from all non-DRIVER_LEGACY drivers
341 ---------------------------------------------------------------
343 The load/unload callbacks in struct &drm_driver are very much midlayers, plus
344 for historical reasons they get the ordering wrong (and we can't fix that)
345 between setting up the &drm_driver structure and calling drm_dev_register().
347 - Rework drivers to no longer use the load/unload callbacks, directly coding the
348 load/unload sequence into the driver's probe function.
350 - Once all non-DRIVER_LEGACY drivers are converted, disallow the load/unload
351 callbacks for all modern drivers.
353 Contact: Daniel Vetter
357 Replace drm_detect_hdmi_monitor() with drm_display_info.is_hdmi
358 ---------------------------------------------------------------
360 Once EDID is parsed, the monitor HDMI support information is available through
361 drm_display_info.is_hdmi. Many drivers still call drm_detect_hdmi_monitor() to
362 retrieve the same information, which is less efficient.
364 Audit each individual driver calling drm_detect_hdmi_monitor() and switch to
365 drm_display_info.is_hdmi if applicable.
367 Contact: Laurent Pinchart, respective driver maintainers
371 Consolidate custom driver modeset properties
372 --------------------------------------------
374 Before atomic modeset took place, many drivers where creating their own
375 properties. Among other things, atomic brought the requirement that custom,
376 driver specific properties should not be used.
378 For this task, we aim to introduce core helpers or reuse the existing ones
381 A quick, unconfirmed, examples list.
383 Introduce core helpers:
384 - audio (amdgpu, intel, gma500, radeon)
385 - brightness, contrast, etc (armada, nouveau) - overlay only (?)
386 - broadcast rgb (gma500, intel)
387 - colorkey (armada, nouveau, rcar) - overlay only (?)
388 - dither (amdgpu, nouveau, radeon) - varies across drivers
389 - underscan family (amdgpu, radeon, nouveau)
393 - tv format names, enhancements (gma500, intel)
394 - tv overscan, margins, etc. (gma500, intel)
395 - zorder (omapdrm) - same as zpos (?)
398 Contact: Emil Velikov, respective driver maintainers
402 Use struct iosys_map throughout codebase
403 ----------------------------------------
405 Pointers to shared device memory are stored in struct iosys_map. Each
406 instance knows whether it refers to system or I/O memory. Most of the DRM-wide
407 interface have been converted to use struct iosys_map, but implementations
408 often still use raw pointers.
410 The task is to use struct iosys_map where it makes sense.
412 * Memory managers should use struct iosys_map for dma-buf-imported buffers.
413 * TTM might benefit from using struct iosys_map internally.
414 * Framebuffer copying and blitting helpers should operate on struct iosys_map.
416 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Christian König, Daniel Vetter
420 Review all drivers for setting struct drm_mode_config.{max_width,max_height} correctly
421 --------------------------------------------------------------------------------------
423 The values in struct drm_mode_config.{max_width,max_height} describe the
424 maximum supported framebuffer size. It's the virtual screen size, but many
425 drivers treat it like limitations of the physical resolution.
427 The maximum width depends on the hardware's maximum scanline pitch. The
428 maximum height depends on the amount of addressable video memory. Review all
429 drivers to initialize the fields to the correct values.
431 Contact: Thomas Zimmermann <tzimmermann@suse.de>
435 Request memory regions in all drivers
436 -------------------------------------
438 Go through all drivers and add code to request the memory regions that the
439 driver uses. This requires adding calls to request_mem_region(),
440 pci_request_region() or similar functions. Use helpers for managed cleanup
443 Drivers are pretty bad at doing this and there used to be conflicts among
444 DRM and fbdev drivers. Still, it's the correct thing to do.
446 Contact: Thomas Zimmermann <tzimmermann@suse.de>
450 Remove driver dependencies on FB_DEVICE
451 ---------------------------------------
453 A number of fbdev drivers provide attributes via sysfs and therefore depend
454 on CONFIG_FB_DEVICE to be selected. Review each driver and attempt to make
455 any dependencies on CONFIG_FB_DEVICE optional. At the minimum, the respective
456 code in the driver could be conditionalized via ifdef CONFIG_FB_DEVICE. Not
457 all drivers might be able to drop CONFIG_FB_DEVICE.
459 Contact: Thomas Zimmermann <tzimmermann@suse.de>
463 Clean up checks for already prepared/enabled in panels
464 ------------------------------------------------------
466 In a whole pile of panel drivers, we have code to make the
467 prepare/unprepare/enable/disable callbacks behave as no-ops if they've already
468 been called. To get some idea of the duplicated code, try::
470 git grep 'if.*>prepared' -- drivers/gpu/drm/panel
471 git grep 'if.*>enabled' -- drivers/gpu/drm/panel
473 In the patch ("drm/panel: Check for already prepared/enabled in drm_panel")
474 we've moved this check to the core. Now we can most definitely remove the
475 check from the individual panels and save a pile of code.
477 In adition to removing the check from the individual panels, it is believed
478 that even the core shouldn't need this check and that should be considered
479 an error if other code ever relies on this check. The check in the core
480 currently prints a warning whenever something is relying on this check with
481 dev_warn(). After a little while, we likely want to promote this to a
482 WARN(1) to help encourage folks not to rely on this behavior.
484 Contact: Douglas Anderson <dianders@chromium.org>
486 Level: Starter/Intermediate
492 Make panic handling work
493 ------------------------
495 This is a really varied tasks with lots of little bits and pieces:
497 * The panic path can't be tested currently, leading to constant breaking. The
498 main issue here is that panics can be triggered from hardirq contexts and
499 hence all panic related callback can run in hardirq context. It would be
500 awesome if we could test at least the fbdev helper code and driver code by
501 e.g. trigger calls through drm debugfs files. hardirq context could be
502 achieved by using an IPI to the local processor.
504 * There's a massive confusion of different panic handlers. DRM fbdev emulation
505 helpers had their own (long removed), but on top of that the fbcon code itself
506 also has one. We need to make sure that they stop fighting over each other.
507 This is worked around by checking ``oops_in_progress`` at various entry points
508 into the DRM fbdev emulation helpers. A much cleaner approach here would be to
509 switch fbcon to the `threaded printk support
510 <https://lwn.net/Articles/800946/>`_.
512 * ``drm_can_sleep()`` is a mess. It hides real bugs in normal operations and
513 isn't a full solution for panic paths. We need to make sure that it only
514 returns true if there's a panic going on for real, and fix up all the
517 * The panic handler must never sleep, which also means it can't ever
518 ``mutex_lock()``. Also it can't grab any other lock unconditionally, not
519 even spinlocks (because NMI and hardirq can panic too). We need to either
520 make sure to not call such paths, or trylock everything. Really tricky.
522 * A clean solution would be an entirely separate panic output support in KMS,
523 bypassing the current fbcon support. See `[PATCH v2 0/3] drm: Add panic handling
524 <https://lore.kernel.org/dri-devel/20190311174218.51899-1-noralf@tronnes.org/>`_.
526 * Encoding the actual oops and preceding dmesg in a QR might help with the
527 dread "important stuff scrolled away" problem. See `[RFC][PATCH] Oops messages
528 transfer using QR codes
529 <https://lore.kernel.org/lkml/1446217392-11981-1-git-send-email-alexandru.murtaza@intel.com/>`_
530 for some example code that could be reused.
532 Contact: Daniel Vetter
536 Clean up the debugfs support
537 ----------------------------
539 There's a bunch of issues with it:
541 - Convert drivers to support the drm_debugfs_add_files() function instead of
542 the drm_debugfs_create_files() function.
544 - Improve late-register debugfs by rolling out the same debugfs pre-register
545 infrastructure for connector and crtc too. That way, the drivers won't need to
546 split their setup code into init and register anymore.
548 - We probably want to have some support for debugfs files on crtc/connectors and
549 maybe other kms objects directly in core. There's even drm_print support in
550 the funcs for these objects to dump kms state, so it's all there. And then the
551 ->show() functions should obviously give you a pointer to the right object.
553 - The drm_driver->debugfs_init hooks we have is just an artifact of the old
554 midlayered load sequence. DRM debugfs should work more like sysfs, where you
555 can create properties/files for an object anytime you want, and the core
556 takes care of publishing/unpuplishing all the files at register/unregister
557 time. Drivers shouldn't need to worry about these technicalities, and fixing
558 this (together with the drm_minor->drm_device move) would allow us to remove
561 Contact: Daniel Vetter
565 Object lifetime fixes
566 ---------------------
568 There's two related issues here
570 - Cleanup up the various ->destroy callbacks, which often are all the same
573 - Lots of drivers erroneously allocate DRM modeset objects using devm_kzalloc,
574 which results in use-after free issues on driver unload. This can be serious
575 trouble even for drivers for hardware integrated on the SoC due to
576 EPROBE_DEFERRED backoff.
578 Both these problems can be solved by switching over to drmm_kzalloc(), and the
579 various convenience wrappers provided, e.g. drmm_crtc_alloc_with_planes(),
580 drmm_universal_plane_alloc(), ... and so on.
582 Contact: Daniel Vetter
586 Remove automatic page mapping from dma-buf importing
587 ----------------------------------------------------
589 When importing dma-bufs, the dma-buf and PRIME frameworks automatically map
590 imported pages into the importer's DMA area. drm_gem_prime_fd_to_handle() and
591 drm_gem_prime_handle_to_fd() require that importers call dma_buf_attach()
592 even if they never do actual device DMA, but only CPU access through
593 dma_buf_vmap(). This is a problem for USB devices, which do not support DMA
596 To fix the issue, automatic page mappings should be removed from the
597 buffer-sharing code. Fixing this is a bit more involved, since the import/export
598 cache is also tied to &drm_gem_object.import_attach. Meanwhile we paper over
599 this problem for USB devices by fishing out the USB host controller device, as
600 long as that supports DMA. Otherwise importing can still needlessly fail.
602 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Daniel Vetter
610 Add unit tests using the Kernel Unit Testing (KUnit) framework
611 --------------------------------------------------------------
613 The `KUnit <https://www.kernel.org/doc/html/latest/dev-tools/kunit/index.html>`_
614 provides a common framework for unit tests within the Linux kernel. Having a
615 test suite would allow to identify regressions earlier.
617 A good candidate for the first unit tests are the format-conversion helpers in
618 ``drm_format_helper.c``.
620 Contact: Javier Martinez Canillas <javierm@redhat.com>
624 Enable trinity for DRM
625 ----------------------
627 And fix up the fallout. Should be really interesting ...
631 Make KMS tests in i-g-t generic
632 -------------------------------
634 The i915 driver team maintains an extensive testsuite for the i915 DRM driver,
635 including tons of testcases for corner-cases in the modesetting API. It would
636 be awesome if those tests (at least the ones not relying on Intel-specific GEM
637 features) could be made to run on any KMS driver.
639 Basic work to run i-g-t tests on non-i915 is done, what's now missing is mass-
640 converting things over. For modeset tests we also first need a bit of
641 infrastructure to use dumb buffers for untiled buffers, to be able to run all
642 the non-i915 specific modeset tests.
646 Extend virtual test driver (VKMS)
647 ---------------------------------
649 See the documentation of :ref:`VKMS <vkms>` for more details. This is an ideal
650 internship task, since it only requires a virtual machine and can be sized to
651 fit the available time.
655 Backlight Refactoring
656 ---------------------
658 Backlight drivers have a triple enable/disable state, which is a bit overkill.
661 1. Roll out backlight_enable() and backlight_disable() helpers everywhere. This
663 2. In all, only look at one of the three status bits set by the above helpers.
664 3. Remove the other two status bits.
666 Contact: Daniel Vetter
673 AMD DC Display Driver
674 ---------------------
676 AMD DC is the display driver for AMD devices starting with Vega. There has been
677 a bunch of progress cleaning it up but there's still plenty of work to be done.
679 See drivers/gpu/drm/amd/display/TODO for tasks.
681 Contact: Harry Wentland, Alex Deucher
686 There is support in place now for writing internal DRM clients making it
687 possible to pick up the bootsplash work that was rejected because it was written
690 - [v6,8/8] drm/client: Hack: Add bootsplash example
691 https://patchwork.freedesktop.org/patch/306579/
693 - [RFC PATCH v2 00/13] Kernel based bootsplash
694 https://lore.kernel.org/r/20171213194755.3409-1-mstaudt@suse.de
696 Contact: Sam Ravnborg
700 Brightness handling on devices with multiple internal panels
701 ============================================================
703 On x86/ACPI devices there can be multiple backlight firmware interfaces:
704 (ACPI) video, vendor specific and others. As well as direct/native (PWM)
705 register programming by the KMS driver.
707 To deal with this backlight drivers used on x86/ACPI call
708 acpi_video_get_backlight_type() which has heuristics (+quirks) to select
709 which backlight interface to use; and backlight drivers which do not match
710 the returned type will not register themselves, so that only one backlight
711 device gets registered (in a single GPU setup, see below).
713 At the moment this more or less assumes that there will only
714 be 1 (internal) panel on a system.
716 On systems with 2 panels this may be a problem, depending on
717 what interface acpi_video_get_backlight_type() selects:
719 1. native: in this case the KMS driver is expected to know which backlight
720 device belongs to which output so everything should just work.
721 2. video: this does support controlling multiple backlights, but some work
722 will need to be done to get the output <-> backlight device mapping
724 The above assumes both panels will require the same backlight interface type.
725 Things will break on systems with multiple panels where the 2 panels need
726 a different type of control. E.g. one panel needs ACPI video backlight control,
727 where as the other is using native backlight control. Currently in this case
728 only one of the 2 required backlight devices will get registered, based on
729 the acpi_video_get_backlight_type() return value.
731 If this (theoretical) case ever shows up, then supporting this will need some
732 work. A possible solution here would be to pass a device and connector-name
733 to acpi_video_get_backlight_type() so that it can deal with this.
735 Note in a way we already have a case where userspace sees 2 panels,
736 in dual GPU laptop setups with a mux. On those systems we may see
737 either 2 native backlight devices; or 2 native backlight devices.
739 Userspace already has code to deal with this by detecting if the related
740 panel is active (iow which way the mux between the GPU and the panels
741 points) and then uses that backlight device. Userspace here very much
742 assumes a single panel though. It picks only 1 of the 2 backlight devices
743 and then only uses that one.
745 Note that all userspace code (that I know off) is currently hardcoded
746 to assume a single panel.
748 Before the recent changes to not register multiple (e.g. video + native)
749 /sys/class/backlight devices for a single panel (on a single GPU laptop),
750 userspace would see multiple backlight devices all controlling the same
753 To deal with this userspace had to always picks one preferred device under
754 /sys/class/backlight and will ignore the others. So to support brightness
755 control on multiple panels userspace will need to be updated too.
757 There are plans to allow brightness control through the KMS API by adding
758 a "display brightness" property to drm_connector objects for panels. This
759 solves a number of issues with the /sys/class/backlight API, including not
760 being able to map a sysfs backlight device to a specific connector. Any
761 userspace changes to add support for brightness control on devices with
762 multiple panels really should build on top of this new KMS property.
764 Contact: Hans de Goede
771 Convert fbdev drivers to DRM
772 ----------------------------
774 There are plenty of fbdev drivers for older hardware. Some hardware has
775 become obsolete, but some still provides good(-enough) framebuffers. The
776 drivers that are still useful should be converted to DRM and afterwards
779 Very simple fbdev drivers can best be converted by starting with a new
780 DRM driver. Simple KMS helpers and SHMEM should be able to handle any
781 existing hardware. The new driver's call-back functions are filled from
784 More complex fbdev drivers can be refactored step-by-step into a DRM
785 driver with the help of the DRM fbconv helpers [4]_. These helpers provide
786 the transition layer between the DRM core infrastructure and the fbdev
787 driver interface. Create a new DRM driver on top of the fbconv helpers,
788 copy over the fbdev driver, and hook it up to the DRM code. Examples for
789 several fbdev drivers are available in Thomas Zimmermann's fbconv tree
790 [4]_, as well as a tutorial of this process [5]_. The result is a primitive
791 DRM driver that can run X11 and Weston.
793 .. [4] https://gitlab.freedesktop.org/tzimmermann/linux/tree/fbconv
794 .. [5] https://gitlab.freedesktop.org/tzimmermann/linux/blob/fbconv/drivers/gpu/drm/drm_fbconv_helper.c
796 Contact: Thomas Zimmermann <tzimmermann@suse.de>