1 .. _todo:
2
3 =========
4 TODO list
5 =========
6
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.
9
10 Difficulty
11 ----------
12
13 To make it easier task are categorized into different levels:
14
15 Starter: Good tasks to get started with the DRM subsystem.
16
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
20 for testing.
21
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
24 testing.
25
26 Expert: Only attempt these if you've successfully completed some tricky
27 refactorings already and are an expert in the specific area
28
29 Subsystem-wide refactorings
30 ===========================
31
32 Remove custom dumb_map_offset implementations
33 ---------------------------------------------
34
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37 implementation (there's lots of outdated locking leftovers in various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.
39
40 Contact: Simona Vetter, respective driver maintainers
41
42 Level: Intermediate
43
44 Convert existing KMS drivers to atomic modesetting
45 --------------------------------------------------
46
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
50 future.
51
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.
55
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.
59
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/
63
64 Contact: Simona Vetter, respective driver maintainers
65
66 Level: Advanced
67
68 Clean up the clipped coordination confusion around planes
69 ---------------------------------------------------------
70
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
76 helpers.
77
78 Contact: Ville Syrjälä, Simona Vetter, driver maintainers
79
80 Level: Advanced
81
82 Improve plane atomic_check helpers
83 ----------------------------------
84
85 Aside from the clipped coordinates right above there's a few suboptimal things
86 with the current helpers:
87
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.
93
94 - Once that's done, helpers could stop calling ->atomic_check for disabled
95 planes.
96
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.
99
100 Contact: Simona Vetter
101
102 Level: Advanced
103
104 Convert early atomic drivers to async commit helpers
105 ----------------------------------------------------
106
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.
111
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.
114
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.
118
119 Contact: Simona Vetter, respective driver maintainers
120
121 Level: Advanced
122
123 Rename drm_atomic_state
124 -----------------------
125
126 The KMS framework uses two slightly different definitions for the ``state``
127 concept. For a given object (plane, CRTC, encoder, etc., so
128 ``drm_$OBJECT_state``), the state is the entire state of that object. However,
129 at the device level, ``drm_atomic_state`` refers to a state update for a
130 limited number of objects.
131
132 The state isn't the entire device state, but only the full state of some
133 objects in that device. This is confusing to newcomers, and
134 ``drm_atomic_state`` should be renamed to something clearer like
135 ``drm_atomic_commit``.
136
137 In addition to renaming the structure itself, it would also imply renaming some
138 related functions (``drm_atomic_state_alloc``, ``drm_atomic_state_get``,
139 ``drm_atomic_state_put``, ``drm_atomic_state_init``,
140 ``__drm_atomic_state_free``, etc.).
141
142 Contact: Maxime Ripard <mripard@kernel.org>
143
144 Level: Advanced
145
146 Fallout from atomic KMS
147 -----------------------
148
149 ``drm_atomic_helper.c`` provides a batch of functions which implement legacy
150 IOCTLs on top of the new atomic driver interface. Which is really nice for
151 gradual conversion of drivers, but unfortunately the semantic mismatches are
152 a bit too severe. So there's some follow-up work to adjust the function
153 interfaces to fix these issues:
154
155 * atomic needs the lock acquire context. At the moment that's passed around
156 implicitly with some horrible hacks, and it's also allocate with
157 ``GFP_NOFAIL`` behind the scenes. All legacy paths need to start allocating
158 the acquire context explicitly on stack and then also pass it down into
159 drivers explicitly so that the legacy-on-atomic functions can use them.
160
161 Except for some driver code this is done. This task should be finished by
162 adding WARN_ON(!drm_drv_uses_atomic_modeset) in drm_modeset_lock_all().
163
164 * A bunch of the vtable hooks are now in the wrong place: DRM has a split
165 between core vfunc tables (named ``drm_foo_funcs``), which are used to
166 implement the userspace ABI. And then there's the optional hooks for the
167 helper libraries (name ``drm_foo_helper_funcs``), which are purely for
168 internal use. Some of these hooks should be move from ``_funcs`` to
169 ``_helper_funcs`` since they are not part of the core ABI. There's a
170 ``FIXME`` comment in the kerneldoc for each such case in ``drm_crtc.h``.
171
172 Contact: Simona Vetter
173
174 Level: Intermediate
175
176 Get rid of dev->struct_mutex from GEM drivers
177 ---------------------------------------------
178
179 ``dev->struct_mutex`` is the Big DRM Lock from legacy days and infested
180 everything. Nowadays in modern drivers the only bit where it's mandatory is
181 serializing GEM buffer object destruction. Which unfortunately means drivers
182 have to keep track of that lock and either call ``unreference`` or
183 ``unreference_locked`` depending upon context.
184
185 Core GEM doesn't have a need for ``struct_mutex`` any more since kernel 4.8,
186 and there's a GEM object ``free`` callback for any drivers which are
187 entirely ``struct_mutex`` free.
188
189 For drivers that need ``struct_mutex`` it should be replaced with a driver-
190 private lock. The tricky part is the BO free functions, since those can't
191 reliably take that lock any more. Instead state needs to be protected with
192 suitable subordinate locks or some cleanup work pushed to a worker thread. For
193 performance-critical drivers it might also be better to go with a more
194 fine-grained per-buffer object and per-context lockings scheme. Currently only
195 the ``msm`` and `i915` drivers use ``struct_mutex``.
196
197 Contact: Simona Vetter, respective driver maintainers
198
199 Level: Advanced
200
201 Move Buffer Object Locking to dma_resv_lock()
202 ---------------------------------------------
203
204 Many drivers have their own per-object locking scheme, usually using
205 mutex_lock(). This causes all kinds of trouble for buffer sharing, since
206 depending which driver is the exporter and importer, the locking hierarchy is
207 reversed.
208
209 To solve this we need one standard per-object locking mechanism, which is
210 dma_resv_lock(). This lock needs to be called as the outermost lock, with all
211 other driver specific per-object locks removed. The problem is that rolling out
212 the actual change to the locking contract is a flag day, due to struct dma_buf
213 buffer sharing.
214
215 Level: Expert
216
217 Convert logging to drm_* functions with drm_device parameter
218 ------------------------------------------------------------
219
220 For drivers which could have multiple instances, it is necessary to
221 differentiate between which is which in the logs. Since DRM_INFO/WARN/ERROR
222 don't do this, drivers used dev_info/warn/err to make this differentiation. We
223 now have drm_* variants of the drm print functions, so we can start to convert
224 those drivers back to using drm-formatted specific log messages.
225
226 Before you start this conversion please contact the relevant maintainers to make
227 sure your work will be merged - not everyone agrees that the DRM dmesg macros
228 are better.
229
230 Contact: Sean Paul, Maintainer of the driver you plan to convert
231
232 Level: Starter
233
234 Convert drivers to use simple modeset suspend/resume
235 ----------------------------------------------------
236
237 Most drivers (except i915 and nouveau) that use
238 drm_atomic_helper_suspend/resume() can probably be converted to use
239 drm_mode_config_helper_suspend/resume(). Also there's still open-coded version
240 of the atomic suspend/resume code in older atomic modeset drivers.
241
242 Contact: Maintainer of the driver you plan to convert
243
244 Level: Intermediate
245
246 Reimplement functions in drm_fbdev_fb_ops without fbdev
247 -------------------------------------------------------
248
249 A number of callback functions in drm_fbdev_fb_ops could benefit from
250 being rewritten without dependencies on the fbdev module. Some of the
251 helpers could further benefit from using struct iosys_map instead of
252 raw pointers.
253
254 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Simona Vetter
255
256 Level: Advanced
257
258 Benchmark and optimize blitting and format-conversion function
259 --------------------------------------------------------------
260
261 Drawing to display memory quickly is crucial for many applications'
262 performance.
263
264 On at least x86-64, sys_imageblit() is significantly slower than
265 cfb_imageblit(), even though both use the same blitting algorithm and
266 the latter is written for I/O memory. It turns out that cfb_imageblit()
267 uses movl instructions, while sys_imageblit apparently does not. This
268 seems to be a problem with gcc's optimizer. DRM's format-conversion
269 helpers might be subject to similar issues.
270
271 Benchmark and optimize fbdev's sys_() helpers and DRM's format-conversion
272 helpers. In cases that can be further optimized, maybe implement a different
273 algorithm. For micro-optimizations, use movl/movq instructions explicitly.
274 That might possibly require architecture-specific helpers (e.g., storel()
275 storeq()).
276
277 Contact: Thomas Zimmermann <tzimmermann@suse.de>
278
279 Level: Intermediate
280
281 drm_framebuffer_funcs and drm_mode_config_funcs.fb_create cleanup
282 -----------------------------------------------------------------
283
284 A lot more drivers could be switched over to the drm_gem_framebuffer helpers.
285 Various hold-ups:
286
287 - Need to switch over to the generic dirty tracking code using
288 drm_atomic_helper_dirtyfb first (e.g. qxl).
289
290 - Need to switch to drm_fbdev_generic_setup(), otherwise a lot of the custom fb
291 setup code can't be deleted.
292
293 - Need to switch to drm_gem_fb_create(), as now drm_gem_fb_create() checks for
294 valid formats for atomic drivers.
295
296 - Many drivers subclass drm_framebuffer, we'd need a embedding compatible
297 version of the varios drm_gem_fb_create functions. Maybe called
298 drm_gem_fb_create/_with_dirty/_with_funcs as needed.
299
300 Contact: Simona Vetter
301
302 Level: Intermediate
303
304 Generic fbdev defio support
305 ---------------------------
306
307 The defio support code in the fbdev core has some very specific requirements,
308 which means drivers need to have a special framebuffer for fbdev. The main
309 issue is that it uses some fields in struct page itself, which breaks shmem
310 gem objects (and other things). To support defio, affected drivers require
311 the use of a shadow buffer, which may add CPU and memory overhead.
312
313 Possible solution would be to write our own defio mmap code in the drm fbdev
314 emulation. It would need to fully wrap the existing mmap ops, forwarding
315 everything after it has done the write-protect/mkwrite trickery:
316
317 - In the drm_fbdev_fb_mmap helper, if we need defio, change the
318 default page prots to write-protected with something like this::
319
320 vma->vm_page_prot = pgprot_wrprotect(vma->vm_page_prot);
321
322 - Set the mkwrite and fsync callbacks with similar implementions to the core
323 fbdev defio stuff. These should all work on plain ptes, they don't actually
324 require a struct page. uff. These should all work on plain ptes, they don't
325 actually require a struct page.
326
327 - Track the dirty pages in a separate structure (bitfield with one bit per page
328 should work) to avoid clobbering struct page.
329
330 Might be good to also have some igt testcases for this.
331
332 Contact: Simona Vetter, Noralf Tronnes
333
334 Level: Advanced
335
336 connector register/unregister fixes
337 -----------------------------------
338
339 - For most connectors it's a no-op to call drm_connector_register/unregister
340 directly from driver code, drm_dev_register/unregister take care of this
341 already. We can remove all of them.
342
343 - For dp drivers it's a bit more a mess, since we need the connector to be
344 registered when calling drm_dp_aux_register. Fix this by instead calling
345 drm_dp_aux_init, and moving the actual registering into a late_register
346 callback as recommended in the kerneldoc.
347
348 Level: Intermediate
349
350 Remove load/unload callbacks
351 ----------------------------
352
353 The load/unload callbacks in struct &drm_driver are very much midlayers, plus
354 for historical reasons they get the ordering wrong (and we can't fix that)
355 between setting up the &drm_driver structure and calling drm_dev_register().
356
357 - Rework drivers to no longer use the load/unload callbacks, directly coding the
358 load/unload sequence into the driver's probe function.
359
360 - Once all drivers are converted, remove the load/unload callbacks.
361
362 Contact: Simona Vetter
363
364 Level: Intermediate
365
366 Replace drm_detect_hdmi_monitor() with drm_display_info.is_hdmi
367 ---------------------------------------------------------------
368
369 Once EDID is parsed, the monitor HDMI support information is available through
370 drm_display_info.is_hdmi. Many drivers still call drm_detect_hdmi_monitor() to
371 retrieve the same information, which is less efficient.
372
373 Audit each individual driver calling drm_detect_hdmi_monitor() and switch to
374 drm_display_info.is_hdmi if applicable.
375
376 Contact: Laurent Pinchart, respective driver maintainers
377
378 Level: Intermediate
379
380 Consolidate custom driver modeset properties
381 --------------------------------------------
382
383 Before atomic modeset took place, many drivers where creating their own
384 properties. Among other things, atomic brought the requirement that custom,
385 driver specific properties should not be used.
386
387 For this task, we aim to introduce core helpers or reuse the existing ones
388 if available:
389
390 A quick, unconfirmed, examples list.
391
392 Introduce core helpers:
393 - audio (amdgpu, intel, gma500, radeon)
394 - brightness, contrast, etc (armada, nouveau) - overlay only (?)
395 - broadcast rgb (gma500, intel)
396 - colorkey (armada, nouveau, rcar) - overlay only (?)
397 - dither (amdgpu, nouveau, radeon) - varies across drivers
398 - underscan family (amdgpu, radeon, nouveau)
399
400 Already in core:
401 - colorspace (sti)
402 - tv format names, enhancements (gma500, intel)
403 - tv overscan, margins, etc. (gma500, intel)
404 - zorder (omapdrm) - same as zpos (?)
405
406
407 Contact: Emil Velikov, respective driver maintainers
408
409 Level: Intermediate
410
411 Use struct iosys_map throughout codebase
412 ----------------------------------------
413
414 Pointers to shared device memory are stored in struct iosys_map. Each
415 instance knows whether it refers to system or I/O memory. Most of the DRM-wide
416 interface have been converted to use struct iosys_map, but implementations
417 often still use raw pointers.
418
419 The task is to use struct iosys_map where it makes sense.
420
421 * Memory managers should use struct iosys_map for dma-buf-imported buffers.
422 * TTM might benefit from using struct iosys_map internally.
423 * Framebuffer copying and blitting helpers should operate on struct iosys_map.
424
425 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Christian König, Simona Vetter
426
427 Level: Intermediate
428
429 Review all drivers for setting struct drm_mode_config.{max_width,max_height} correctly
430 --------------------------------------------------------------------------------------
431
432 The values in struct drm_mode_config.{max_width,max_height} describe the
433 maximum supported framebuffer size. It's the virtual screen size, but many
434 drivers treat it like limitations of the physical resolution.
435
436 The maximum width depends on the hardware's maximum scanline pitch. The
437 maximum height depends on the amount of addressable video memory. Review all
438 drivers to initialize the fields to the correct values.
439
440 Contact: Thomas Zimmermann <tzimmermann@suse.de>
441
442 Level: Intermediate
443
444 Request memory regions in all drivers
445 -------------------------------------
446
447 Go through all drivers and add code to request the memory regions that the
448 driver uses. This requires adding calls to request_mem_region(),
449 pci_request_region() or similar functions. Use helpers for managed cleanup
450 where possible.
451
452 Drivers are pretty bad at doing this and there used to be conflicts among
453 DRM and fbdev drivers. Still, it's the correct thing to do.
454
455 Contact: Thomas Zimmermann <tzimmermann@suse.de>
456
457 Level: Starter
458
459 Remove driver dependencies on FB_DEVICE
460 ---------------------------------------
461
462 A number of fbdev drivers provide attributes via sysfs and therefore depend
463 on CONFIG_FB_DEVICE to be selected. Review each driver and attempt to make
464 any dependencies on CONFIG_FB_DEVICE optional. At the minimum, the respective
465 code in the driver could be conditionalized via ifdef CONFIG_FB_DEVICE. Not
466 all drivers might be able to drop CONFIG_FB_DEVICE.
467
468 Contact: Thomas Zimmermann <tzimmermann@suse.de>
469
470 Level: Starter
471
472 Remove disable/unprepare in remove/shutdown in panel-simple and panel-edp
473 -------------------------------------------------------------------------
474
475 As of commit d2aacaf07395 ("drm/panel: Check for already prepared/enabled in
476 drm_panel"), we have a check in the drm_panel core to make sure nobody
477 double-calls prepare/enable/disable/unprepare. Eventually that should probably
478 be turned into a WARN_ON() or somehow made louder.
479
480 At the moment, we expect that we may still encounter the warnings in the
481 drm_panel core when using panel-simple and panel-edp. Since those panel
482 drivers are used with a lot of different DRM modeset drivers they still
483 make an extra effort to disable/unprepare the panel themsevles at shutdown
484 time. Specifically we could still encounter those warnings if the panel
485 driver gets shutdown() _before_ the DRM modeset driver and the DRM modeset
486 driver properly calls drm_atomic_helper_shutdown() in its own shutdown()
487 callback. Warnings could be avoided in such a case by using something like
488 device links to ensure that the panel gets shutdown() after the DRM modeset
489 driver.
490
491 Once all DRM modeset drivers are known to shutdown properly, the extra
492 calls to disable/unprepare in remove/shutdown in panel-simple and panel-edp
493 should be removed and this TODO item marked complete.
494
495 Contact: Douglas Anderson <dianders@chromium.org>
496
497 Level: Intermediate
498
499 Transition away from using mipi_dsi_*_write_seq()
500 -------------------------------------------------
501
502 The macros mipi_dsi_generic_write_seq() and mipi_dsi_dcs_write_seq() are
503 non-intuitive because, if there are errors, they return out of the *caller's*
504 function. We should move all callers to use mipi_dsi_generic_write_seq_multi()
505 and mipi_dsi_dcs_write_seq_multi() macros instead.
506
507 Once all callers are transitioned, the macros and the functions that they call,
508 mipi_dsi_generic_write_chatty() and mipi_dsi_dcs_write_buffer_chatty(), can
509 probably be removed. Alternatively, if people feel like the _multi() variants
510 are overkill for some use cases, we could keep the mipi_dsi_*_write_seq()
511 variants but change them not to return out of the caller.
512
513 Contact: Douglas Anderson <dianders@chromium.org>
514
515 Level: Starter
516
517
518 Core refactorings
519 =================
520
521 Make panic handling work
522 ------------------------
523
524 This is a really varied tasks with lots of little bits and pieces:
525
526 * The panic path can't be tested currently, leading to constant breaking. The
527 main issue here is that panics can be triggered from hardirq contexts and
528 hence all panic related callback can run in hardirq context. It would be
529 awesome if we could test at least the fbdev helper code and driver code by
530 e.g. trigger calls through drm debugfs files. hardirq context could be
531 achieved by using an IPI to the local processor.
532
533 * There's a massive confusion of different panic handlers. DRM fbdev emulation
534 helpers had their own (long removed), but on top of that the fbcon code itself
535 also has one. We need to make sure that they stop fighting over each other.
536 This is worked around by checking ``oops_in_progress`` at various entry points
537 into the DRM fbdev emulation helpers. A much cleaner approach here would be to
538 switch fbcon to the `threaded printk support
539 <https://lwn.net/Articles/800946/>`_.
540
541 * ``drm_can_sleep()`` is a mess. It hides real bugs in normal operations and
542 isn't a full solution for panic paths. We need to make sure that it only
543 returns true if there's a panic going on for real, and fix up all the
544 fallout.
545
546 * The panic handler must never sleep, which also means it can't ever
547 ``mutex_lock()``. Also it can't grab any other lock unconditionally, not
548 even spinlocks (because NMI and hardirq can panic too). We need to either
549 make sure to not call such paths, or trylock everything. Really tricky.
550
551 * A clean solution would be an entirely separate panic output support in KMS,
552 bypassing the current fbcon support. See `[PATCH v2 0/3] drm: Add panic handling
553 <https://lore.kernel.org/dri-devel/20190311174218.51899-1-noralf@tronnes.org/">https://lore.kernel.org/dri-devel/20190311174218.51899-1-noralf@tronnes.org/>`_.
554
555 * Encoding the actual oops and preceding dmesg in a QR might help with the
556 dread "important stuff scrolled away" problem. See `[RFC][PATCH] Oops messages
557 transfer using QR codes
558 <https://lore.kernel.org/lkml/1446217392-11981-1-git-send-email-alexandru.murtaza@intel.com/">https://lore.kernel.org/lkml/1446217392-11981-1-git-send-email-alexandru.murtaza@intel.com/>`_
559 for some example code that could be reused.
560
561 Contact: Simona Vetter
562
563 Level: Advanced
564
565 Clean up the debugfs support
566 ----------------------------
567
568 There's a bunch of issues with it:
569
570 - Convert drivers to support the drm_debugfs_add_files() function instead of
571 the drm_debugfs_create_files() function.
572
573 - Improve late-register debugfs by rolling out the same debugfs pre-register
574 infrastructure for connector and crtc too. That way, the drivers won't need to
575 split their setup code into init and register anymore.
576
577 - We probably want to have some support for debugfs files on crtc/connectors and
578 maybe other kms objects directly in core. There's even drm_print support in
579 the funcs for these objects to dump kms state, so it's all there. And then the
580 ->show() functions should obviously give you a pointer to the right object.
581
582 - The drm_driver->debugfs_init hooks we have is just an artifact of the old
583 midlayered load sequence. DRM debugfs should work more like sysfs, where you
584 can create properties/files for an object anytime you want, and the core
585 takes care of publishing/unpuplishing all the files at register/unregister
586 time. Drivers shouldn't need to worry about these technicalities, and fixing
587 this (together with the drm_minor->drm_device move) would allow us to remove
588 debugfs_init.
589
590 Contact: Simona Vetter
591
592 Level: Intermediate
593
594 Object lifetime fixes
595 ---------------------
596
597 There's two related issues here
598
599 - Cleanup up the various ->destroy callbacks, which often are all the same
600 simple code.
601
602 - Lots of drivers erroneously allocate DRM modeset objects using devm_kzalloc,
603 which results in use-after free issues on driver unload. This can be serious
604 trouble even for drivers for hardware integrated on the SoC due to
605 EPROBE_DEFERRED backoff.
606
607 Both these problems can be solved by switching over to drmm_kzalloc(), and the
608 various convenience wrappers provided, e.g. drmm_crtc_alloc_with_planes(),
609 drmm_universal_plane_alloc(), ... and so on.
610
611 Contact: Simona Vetter
612
613 Level: Intermediate
614
615 Remove automatic page mapping from dma-buf importing
616 ----------------------------------------------------
617
618 When importing dma-bufs, the dma-buf and PRIME frameworks automatically map
619 imported pages into the importer's DMA area. drm_gem_prime_fd_to_handle() and
620 drm_gem_prime_handle_to_fd() require that importers call dma_buf_attach()
621 even if they never do actual device DMA, but only CPU access through
622 dma_buf_vmap(). This is a problem for USB devices, which do not support DMA
623 operations.
624
625 To fix the issue, automatic page mappings should be removed from the
626 buffer-sharing code. Fixing this is a bit more involved, since the import/export
627 cache is also tied to &drm_gem_object.import_attach. Meanwhile we paper over
628 this problem for USB devices by fishing out the USB host controller device, as
629 long as that supports DMA. Otherwise importing can still needlessly fail.
630
631 Contact: Thomas Zimmermann <tzimmermann@suse.de>, Simona Vetter
632
633 Level: Advanced
634
635
636 Better Testing
637 ==============
638
639 Add unit tests using the Kernel Unit Testing (KUnit) framework
640 --------------------------------------------------------------
641
642 The `KUnit <https://www.kernel.org/doc/html/latest/dev-tools/kunit/index.html>`_
643 provides a common framework for unit tests within the Linux kernel. Having a
644 test suite would allow to identify regressions earlier.
645
646 A good candidate for the first unit tests are the format-conversion helpers in
647 ``drm_format_helper.c``.
648
649 Contact: Javier Martinez Canillas <javierm@redhat.com>
650
651 Level: Intermediate
652
653 Clean up and document former selftests suites
654 ---------------------------------------------
655
656 Some KUnit test suites (drm_buddy, drm_cmdline_parser, drm_damage_helper,
657 drm_format, drm_framebuffer, drm_dp_mst_helper, drm_mm, drm_plane_helper and
658 drm_rect) are former selftests suites that have been converted over when KUnit
659 was first introduced.
660
661 These suites were fairly undocumented, and with different goals than what unit
662 tests can be. Trying to identify what each test in these suites actually test
663 for, whether that makes sense for a unit test, and either remove it if it
664 doesn't or document it if it does would be of great help.
665
666 Contact: Maxime Ripard <mripard@kernel.org>
667
668 Level: Intermediate
669
670 Enable trinity for DRM
671 ----------------------
672
673 And fix up the fallout. Should be really interesting ...
674
675 Level: Advanced
676
677 Make KMS tests in i-g-t generic
678 -------------------------------
679
680 The i915 driver team maintains an extensive testsuite for the i915 DRM driver,
681 including tons of testcases for corner-cases in the modesetting API. It would
682 be awesome if those tests (at least the ones not relying on Intel-specific GEM
683 features) could be made to run on any KMS driver.
684
685 Basic work to run i-g-t tests on non-i915 is done, what's now missing is mass-
686 converting things over. For modeset tests we also first need a bit of
687 infrastructure to use dumb buffers for untiled buffers, to be able to run all
688 the non-i915 specific modeset tests.
689
690 Level: Advanced
691
692 Extend virtual test driver (VKMS)
693 ---------------------------------
694
695 See the documentation of :ref:`VKMS <vkms>` for more details. This is an ideal
696 internship task, since it only requires a virtual machine and can be sized to
697 fit the available time.
698
699 Level: See details
700
701 Backlight Refactoring
702 ---------------------
703
704 Backlight drivers have a triple enable/disable state, which is a bit overkill.
705 Plan to fix this:
706
707 1. Roll out backlight_enable() and backlight_disable() helpers everywhere. This
708 has started already.
709 2. In all, only look at one of the three status bits set by the above helpers.
710 3. Remove the other two status bits.
711
712 Contact: Simona Vetter
713
714 Level: Intermediate
715
716 Driver Specific
717 ===============
718
719 AMD DC Display Driver
720 ---------------------
721
722 AMD DC is the display driver for AMD devices starting with Vega. There has been
723 a bunch of progress cleaning it up but there's still plenty of work to be done.
724
725 See drivers/gpu/drm/amd/display/TODO for tasks.
726
727 Contact: Harry Wentland, Alex Deucher
728
729 Bootsplash
730 ==========
731
732 There is support in place now for writing internal DRM clients making it
733 possible to pick up the bootsplash work that was rejected because it was written
734 for fbdev.
735
736 - [v6,8/8] drm/client: Hack: Add bootsplash example
737 https://patchwork.freedesktop.org/patch/306579/
738
739 - [RFC PATCH v2 00/13] Kernel based bootsplash
740 https://lore.kernel.org/r/20171213194755.3409-1-mstaudt@suse.de
741
742 Contact: Sam Ravnborg
743
744 Level: Advanced
745
746 Brightness handling on devices with multiple internal panels
747 ============================================================
748
749 On x86/ACPI devices there can be multiple backlight firmware interfaces:
750 (ACPI) video, vendor specific and others. As well as direct/native (PWM)
751 register programming by the KMS driver.
752
753 To deal with this backlight drivers used on x86/ACPI call
754 acpi_video_get_backlight_type() which has heuristics (+quirks) to select
755 which backlight interface to use; and backlight drivers which do not match
756 the returned type will not register themselves, so that only one backlight
757 device gets registered (in a single GPU setup, see below).
758
759 At the moment this more or less assumes that there will only
760 be 1 (internal) panel on a system.
761
762 On systems with 2 panels this may be a problem, depending on
763 what interface acpi_video_get_backlight_type() selects:
764
765 1. native: in this case the KMS driver is expected to know which backlight
766 device belongs to which output so everything should just work.
767 2. video: this does support controlling multiple backlights, but some work
768 will need to be done to get the output <-> backlight device mapping
769
770 The above assumes both panels will require the same backlight interface type.
771 Things will break on systems with multiple panels where the 2 panels need
772 a different type of control. E.g. one panel needs ACPI video backlight control,
773 where as the other is using native backlight control. Currently in this case
774 only one of the 2 required backlight devices will get registered, based on
775 the acpi_video_get_backlight_type() return value.
776
777 If this (theoretical) case ever shows up, then supporting this will need some
778 work. A possible solution here would be to pass a device and connector-name
779 to acpi_video_get_backlight_type() so that it can deal with this.
780
781 Note in a way we already have a case where userspace sees 2 panels,
782 in dual GPU laptop setups with a mux. On those systems we may see
783 either 2 native backlight devices; or 2 native backlight devices.
784
785 Userspace already has code to deal with this by detecting if the related
786 panel is active (iow which way the mux between the GPU and the panels
787 points) and then uses that backlight device. Userspace here very much
788 assumes a single panel though. It picks only 1 of the 2 backlight devices
789 and then only uses that one.
790
791 Note that all userspace code (that I know off) is currently hardcoded
792 to assume a single panel.
793
794 Before the recent changes to not register multiple (e.g. video + native)
795 /sys/class/backlight devices for a single panel (on a single GPU laptop),
796 userspace would see multiple backlight devices all controlling the same
797 backlight.
798
799 To deal with this userspace had to always picks one preferred device under
800 /sys/class/backlight and will ignore the others. So to support brightness
801 control on multiple panels userspace will need to be updated too.
802
803 There are plans to allow brightness control through the KMS API by adding
804 a "display brightness" property to drm_connector objects for panels. This
805 solves a number of issues with the /sys/class/backlight API, including not
806 being able to map a sysfs backlight device to a specific connector. Any
807 userspace changes to add support for brightness control on devices with
808 multiple panels really should build on top of this new KMS property.
809
810 Contact: Hans de Goede
811
812 Level: Advanced
813
814 Buffer age or other damage accumulation algorithm for buffer damage
815 ===================================================================
816
817 Drivers that do per-buffer uploads, need a buffer damage handling (rather than
818 frame damage like drivers that do per-plane or per-CRTC uploads), but there is
819 no support to get the buffer age or any other damage accumulation algorithm.
820
821 For this reason, the damage helpers just fallback to a full plane update if the
822 framebuffer attached to a plane has changed since the last page-flip. Drivers
823 set &drm_plane_state.ignore_damage_clips to true as indication to
824 drm_atomic_helper_damage_iter_init() and drm_atomic_helper_damage_iter_next()
825 helpers that the damage clips should be ignored.
826
827 This should be improved to get damage tracking properly working on drivers that
828 do per-buffer uploads.
829
830 More information about damage tracking and references to learning materials can
831 be found in :ref:`damage_tracking_properties`.
832
833 Contact: Javier Martinez Canillas <javierm@redhat.com>
834
835 Level: Advanced
836
837 Outside DRM
838 ===========
839
840 Convert fbdev drivers to DRM
841 ----------------------------
842
843 There are plenty of fbdev drivers for older hardware. Some hardware has
844 become obsolete, but some still provides good(-enough) framebuffers. The
845 drivers that are still useful should be converted to DRM and afterwards
846 removed from fbdev.
847
848 Very simple fbdev drivers can best be converted by starting with a new
849 DRM driver. Simple KMS helpers and SHMEM should be able to handle any
850 existing hardware. The new driver's call-back functions are filled from
851 existing fbdev code.
852
853 More complex fbdev drivers can be refactored step-by-step into a DRM
854 driver with the help of the DRM fbconv helpers [4]_. These helpers provide
855 the transition layer between the DRM core infrastructure and the fbdev
856 driver interface. Create a new DRM driver on top of the fbconv helpers,
857 copy over the fbdev driver, and hook it up to the DRM code. Examples for
858 several fbdev drivers are available in Thomas Zimmermann's fbconv tree
859 [4]_, as well as a tutorial of this process [5]_. The result is a primitive
860 DRM driver that can run X11 and Weston.
861
862 .. [4] https://gitlab.freedesktop.org/tzimmermann/linux/tree/fbconv
863 .. [5] https://gitlab.freedesktop.org/tzimmermann/linux/blob/fbconv/drivers/gpu/drm/drm_fbconv_helper.c
864
865 Contact: Thomas Zimmermann <tzimmermann@suse.de>
866
867 Level: Advanced
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.