868 lines · plain
1.. _todo:2 3=========4TODO list5=========6 7This section contains a list of smaller janitorial tasks in the kernel DRM8graphics subsystem useful as newbie projects. Or for slow rainy days.9 10Difficulty11----------12 13To make it easier task are categorized into different levels:14 15Starter: Good tasks to get started with the DRM subsystem.16 17Intermediate: Tasks which need some experience with working in the DRM18subsystem, or some specific GPU/display graphics knowledge. For debugging issue19it's good to have the relevant hardware (or a virtual driver set up) available20for testing.21 22Advanced: Tricky tasks that need fairly good understanding of the DRM subsystem23and graphics topics. Generally need the relevant hardware for development and24testing.25 26Expert: Only attempt these if you've successfully completed some tricky27refactorings already and are an expert in the specific area28 29Subsystem-wide refactorings30===========================31 32Remove custom dumb_map_offset implementations33---------------------------------------------34 35All GEM based drivers should be using drm_gem_create_mmap_offset() instead.36Audit each individual driver, make sure it'll work with the generic37implementation (there's lots of outdated locking leftovers in various38implementations), and then remove it.39 40Contact: Simona Vetter, respective driver maintainers41 42Level: Intermediate43 44Convert existing KMS drivers to atomic modesetting45--------------------------------------------------46 473.19 has the atomic modeset interfaces and helpers, so drivers can now be48converted over. Modern compositors like Wayland or Surfaceflinger on Android49really want an atomic modeset interface, so this is all about the bright50future.51 52There is a conversion guide for atomic [1]_ and all you need is a GPU for a53non-converted driver. The "Atomic mode setting design overview" series [2]_54[3]_ at LWN.net can also be helpful.55 56As part of this drivers also need to convert to universal plane (which means57exposing primary & cursor as proper plane objects). But that's much easier to58do by directly using the new atomic helper driver callbacks.59 60 .. [1] https://blog.ffwll.ch/2014/11/atomic-modeset-support-for-kms-drivers.html61 .. [2] https://lwn.net/Articles/653071/62 .. [3] https://lwn.net/Articles/653466/63 64Contact: Simona Vetter, respective driver maintainers65 66Level: Advanced67 68Clean up the clipped coordination confusion around planes69---------------------------------------------------------70 71We have a helper to get this right with drm_plane_helper_check_update(), but72it's not consistently used. This should be fixed, preferably in the atomic73helpers (and drivers then moved over to clipped coordinates). Probably the74helper should also be moved from drm_plane_helper.c to the atomic helpers, to75avoid confusion - the other helpers in that file are all deprecated legacy76helpers.77 78Contact: Ville Syrjälä, Simona Vetter, driver maintainers79 80Level: Advanced81 82Improve plane atomic_check helpers83----------------------------------84 85Aside from the clipped coordinates right above there's a few suboptimal things86with the current helpers:87 88- drm_plane_helper_funcs->atomic_check gets called for enabled or disabled89 planes. At best this seems to confuse drivers, worst it means they blow up90 when the plane is disabled without the CRTC. The only special handling is91 resetting values in the plane state structures, which instead should be moved92 into the drm_plane_funcs->atomic_duplicate_state functions.93 94- Once that's done, helpers could stop calling ->atomic_check for disabled95 planes.96 97- Then we could go through all the drivers and remove the more-or-less confused98 checks for plane_state->fb and plane_state->crtc.99 100Contact: Simona Vetter101 102Level: Advanced103 104Convert early atomic drivers to async commit helpers105----------------------------------------------------106 107For the first year the atomic modeset helpers didn't support asynchronous /108nonblocking commits, and every driver had to hand-roll them. This is fixed109now, but there's still a pile of existing drivers that easily could be110converted over to the new infrastructure.111 112One issue with the helpers is that they require that drivers handle completion113events for atomic commits correctly. But fixing these bugs is good anyway.114 115Somewhat related is the legacy_cursor_update hack, which should be replaced with116the new atomic_async_check/commit functionality in the helpers in drivers that117still look at that flag.118 119Contact: Simona Vetter, respective driver maintainers120 121Level: Advanced122 123Rename drm_atomic_state124-----------------------125 126The KMS framework uses two slightly different definitions for the ``state``127concept. For a given object (plane, CRTC, encoder, etc., so128``drm_$OBJECT_state``), the state is the entire state of that object. However,129at the device level, ``drm_atomic_state`` refers to a state update for a130limited number of objects.131 132The state isn't the entire device state, but only the full state of some133objects in that device. This is confusing to newcomers, and134``drm_atomic_state`` should be renamed to something clearer like135``drm_atomic_commit``.136 137In addition to renaming the structure itself, it would also imply renaming some138related 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 142Contact: Maxime Ripard <mripard@kernel.org>143 144Level: Advanced145 146Fallout from atomic KMS147-----------------------148 149``drm_atomic_helper.c`` provides a batch of functions which implement legacy150IOCTLs on top of the new atomic driver interface. Which is really nice for151gradual conversion of drivers, but unfortunately the semantic mismatches are152a bit too severe. So there's some follow-up work to adjust the function153interfaces to fix these issues:154 155* atomic needs the lock acquire context. At the moment that's passed around156 implicitly with some horrible hacks, and it's also allocate with157 ``GFP_NOFAIL`` behind the scenes. All legacy paths need to start allocating158 the acquire context explicitly on stack and then also pass it down into159 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 by162 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 split165 between core vfunc tables (named ``drm_foo_funcs``), which are used to166 implement the userspace ABI. And then there's the optional hooks for the167 helper libraries (name ``drm_foo_helper_funcs``), which are purely for168 internal use. Some of these hooks should be move from ``_funcs`` to169 ``_helper_funcs`` since they are not part of the core ABI. There's a170 ``FIXME`` comment in the kerneldoc for each such case in ``drm_crtc.h``.171 172Contact: Simona Vetter173 174Level: Intermediate175 176Get rid of dev->struct_mutex from GEM drivers177---------------------------------------------178 179``dev->struct_mutex`` is the Big DRM Lock from legacy days and infested180everything. Nowadays in modern drivers the only bit where it's mandatory is181serializing GEM buffer object destruction. Which unfortunately means drivers182have to keep track of that lock and either call ``unreference`` or183``unreference_locked`` depending upon context.184 185Core GEM doesn't have a need for ``struct_mutex`` any more since kernel 4.8,186and there's a GEM object ``free`` callback for any drivers which are187entirely ``struct_mutex`` free.188 189For drivers that need ``struct_mutex`` it should be replaced with a driver-190private lock. The tricky part is the BO free functions, since those can't191reliably take that lock any more. Instead state needs to be protected with192suitable subordinate locks or some cleanup work pushed to a worker thread. For193performance-critical drivers it might also be better to go with a more194fine-grained per-buffer object and per-context lockings scheme. Currently only195the ``msm`` and `i915` drivers use ``struct_mutex``.196 197Contact: Simona Vetter, respective driver maintainers198 199Level: Advanced200 201Move Buffer Object Locking to dma_resv_lock()202---------------------------------------------203 204Many drivers have their own per-object locking scheme, usually using205mutex_lock(). This causes all kinds of trouble for buffer sharing, since206depending which driver is the exporter and importer, the locking hierarchy is207reversed.208 209To solve this we need one standard per-object locking mechanism, which is210dma_resv_lock(). This lock needs to be called as the outermost lock, with all211other driver specific per-object locks removed. The problem is that rolling out212the actual change to the locking contract is a flag day, due to struct dma_buf213buffer sharing.214 215Level: Expert216 217Convert logging to drm_* functions with drm_device parameter218------------------------------------------------------------219 220For drivers which could have multiple instances, it is necessary to221differentiate between which is which in the logs. Since DRM_INFO/WARN/ERROR222don't do this, drivers used dev_info/warn/err to make this differentiation. We223now have drm_* variants of the drm print functions, so we can start to convert224those drivers back to using drm-formatted specific log messages.225 226Before you start this conversion please contact the relevant maintainers to make227sure your work will be merged - not everyone agrees that the DRM dmesg macros228are better.229 230Contact: Sean Paul, Maintainer of the driver you plan to convert231 232Level: Starter233 234Convert drivers to use simple modeset suspend/resume235----------------------------------------------------236 237Most drivers (except i915 and nouveau) that use238drm_atomic_helper_suspend/resume() can probably be converted to use239drm_mode_config_helper_suspend/resume(). Also there's still open-coded version240of the atomic suspend/resume code in older atomic modeset drivers.241 242Contact: Maintainer of the driver you plan to convert243 244Level: Intermediate245 246Reimplement functions in drm_fbdev_fb_ops without fbdev247-------------------------------------------------------248 249A number of callback functions in drm_fbdev_fb_ops could benefit from250being rewritten without dependencies on the fbdev module. Some of the251helpers could further benefit from using struct iosys_map instead of252raw pointers.253 254Contact: Thomas Zimmermann <tzimmermann@suse.de>, Simona Vetter255 256Level: Advanced257 258Benchmark and optimize blitting and format-conversion function259--------------------------------------------------------------260 261Drawing to display memory quickly is crucial for many applications'262performance.263 264On at least x86-64, sys_imageblit() is significantly slower than265cfb_imageblit(), even though both use the same blitting algorithm and266the latter is written for I/O memory. It turns out that cfb_imageblit()267uses movl instructions, while sys_imageblit apparently does not. This268seems to be a problem with gcc's optimizer. DRM's format-conversion269helpers might be subject to similar issues.270 271Benchmark and optimize fbdev's sys_() helpers and DRM's format-conversion272helpers. In cases that can be further optimized, maybe implement a different273algorithm. For micro-optimizations, use movl/movq instructions explicitly.274That might possibly require architecture-specific helpers (e.g., storel()275storeq()).276 277Contact: Thomas Zimmermann <tzimmermann@suse.de>278 279Level: Intermediate280 281drm_framebuffer_funcs and drm_mode_config_funcs.fb_create cleanup282-----------------------------------------------------------------283 284A lot more drivers could be switched over to the drm_gem_framebuffer helpers.285Various hold-ups:286 287- Need to switch over to the generic dirty tracking code using288 drm_atomic_helper_dirtyfb first (e.g. qxl).289 290- Need to switch to drm_fbdev_generic_setup(), otherwise a lot of the custom fb291 setup code can't be deleted.292 293- Need to switch to drm_gem_fb_create(), as now drm_gem_fb_create() checks for294 valid formats for atomic drivers.295 296- Many drivers subclass drm_framebuffer, we'd need a embedding compatible297 version of the varios drm_gem_fb_create functions. Maybe called298 drm_gem_fb_create/_with_dirty/_with_funcs as needed.299 300Contact: Simona Vetter301 302Level: Intermediate303 304Generic fbdev defio support305---------------------------306 307The defio support code in the fbdev core has some very specific requirements,308which means drivers need to have a special framebuffer for fbdev. The main309issue is that it uses some fields in struct page itself, which breaks shmem310gem objects (and other things). To support defio, affected drivers require311the use of a shadow buffer, which may add CPU and memory overhead.312 313Possible solution would be to write our own defio mmap code in the drm fbdev314emulation. It would need to fully wrap the existing mmap ops, forwarding315everything after it has done the write-protect/mkwrite trickery:316 317- In the drm_fbdev_fb_mmap helper, if we need defio, change the318 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 core323 fbdev defio stuff. These should all work on plain ptes, they don't actually324 require a struct page. uff. These should all work on plain ptes, they don't325 actually require a struct page.326 327- Track the dirty pages in a separate structure (bitfield with one bit per page328 should work) to avoid clobbering struct page.329 330Might be good to also have some igt testcases for this.331 332Contact: Simona Vetter, Noralf Tronnes333 334Level: Advanced335 336connector register/unregister fixes337-----------------------------------338 339- For most connectors it's a no-op to call drm_connector_register/unregister340 directly from driver code, drm_dev_register/unregister take care of this341 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 be344 registered when calling drm_dp_aux_register. Fix this by instead calling345 drm_dp_aux_init, and moving the actual registering into a late_register346 callback as recommended in the kerneldoc.347 348Level: Intermediate349 350Remove load/unload callbacks351----------------------------352 353The load/unload callbacks in struct &drm_driver are very much midlayers, plus354for historical reasons they get the ordering wrong (and we can't fix that)355between 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 the358 load/unload sequence into the driver's probe function.359 360- Once all drivers are converted, remove the load/unload callbacks.361 362Contact: Simona Vetter363 364Level: Intermediate365 366Replace drm_detect_hdmi_monitor() with drm_display_info.is_hdmi367---------------------------------------------------------------368 369Once EDID is parsed, the monitor HDMI support information is available through370drm_display_info.is_hdmi. Many drivers still call drm_detect_hdmi_monitor() to371retrieve the same information, which is less efficient.372 373Audit each individual driver calling drm_detect_hdmi_monitor() and switch to374drm_display_info.is_hdmi if applicable.375 376Contact: Laurent Pinchart, respective driver maintainers377 378Level: Intermediate379 380Consolidate custom driver modeset properties381--------------------------------------------382 383Before atomic modeset took place, many drivers where creating their own384properties. Among other things, atomic brought the requirement that custom,385driver specific properties should not be used.386 387For this task, we aim to introduce core helpers or reuse the existing ones388if available:389 390A quick, unconfirmed, examples list.391 392Introduce 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 drivers398- underscan family (amdgpu, radeon, nouveau)399 400Already 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 407Contact: Emil Velikov, respective driver maintainers408 409Level: Intermediate410 411Use struct iosys_map throughout codebase412----------------------------------------413 414Pointers to shared device memory are stored in struct iosys_map. Each415instance knows whether it refers to system or I/O memory. Most of the DRM-wide416interface have been converted to use struct iosys_map, but implementations417often still use raw pointers.418 419The 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 425Contact: Thomas Zimmermann <tzimmermann@suse.de>, Christian König, Simona Vetter426 427Level: Intermediate428 429Review all drivers for setting struct drm_mode_config.{max_width,max_height} correctly430--------------------------------------------------------------------------------------431 432The values in struct drm_mode_config.{max_width,max_height} describe the433maximum supported framebuffer size. It's the virtual screen size, but many434drivers treat it like limitations of the physical resolution.435 436The maximum width depends on the hardware's maximum scanline pitch. The437maximum height depends on the amount of addressable video memory. Review all438drivers to initialize the fields to the correct values.439 440Contact: Thomas Zimmermann <tzimmermann@suse.de>441 442Level: Intermediate443 444Request memory regions in all drivers445-------------------------------------446 447Go through all drivers and add code to request the memory regions that the448driver uses. This requires adding calls to request_mem_region(),449pci_request_region() or similar functions. Use helpers for managed cleanup450where possible.451 452Drivers are pretty bad at doing this and there used to be conflicts among453DRM and fbdev drivers. Still, it's the correct thing to do.454 455Contact: Thomas Zimmermann <tzimmermann@suse.de>456 457Level: Starter458 459Remove driver dependencies on FB_DEVICE460---------------------------------------461 462A number of fbdev drivers provide attributes via sysfs and therefore depend463on CONFIG_FB_DEVICE to be selected. Review each driver and attempt to make464any dependencies on CONFIG_FB_DEVICE optional. At the minimum, the respective465code in the driver could be conditionalized via ifdef CONFIG_FB_DEVICE. Not466all drivers might be able to drop CONFIG_FB_DEVICE.467 468Contact: Thomas Zimmermann <tzimmermann@suse.de>469 470Level: Starter471 472Remove disable/unprepare in remove/shutdown in panel-simple and panel-edp473-------------------------------------------------------------------------474 475As of commit d2aacaf07395 ("drm/panel: Check for already prepared/enabled in476drm_panel"), we have a check in the drm_panel core to make sure nobody477double-calls prepare/enable/disable/unprepare. Eventually that should probably478be turned into a WARN_ON() or somehow made louder.479 480At the moment, we expect that we may still encounter the warnings in the481drm_panel core when using panel-simple and panel-edp. Since those panel482drivers are used with a lot of different DRM modeset drivers they still483make an extra effort to disable/unprepare the panel themsevles at shutdown484time. Specifically we could still encounter those warnings if the panel485driver gets shutdown() _before_ the DRM modeset driver and the DRM modeset486driver properly calls drm_atomic_helper_shutdown() in its own shutdown()487callback. Warnings could be avoided in such a case by using something like488device links to ensure that the panel gets shutdown() after the DRM modeset489driver.490 491Once all DRM modeset drivers are known to shutdown properly, the extra492calls to disable/unprepare in remove/shutdown in panel-simple and panel-edp493should be removed and this TODO item marked complete.494 495Contact: Douglas Anderson <dianders@chromium.org>496 497Level: Intermediate498 499Transition away from using mipi_dsi_*_write_seq()500-------------------------------------------------501 502The macros mipi_dsi_generic_write_seq() and mipi_dsi_dcs_write_seq() are503non-intuitive because, if there are errors, they return out of the *caller's*504function. We should move all callers to use mipi_dsi_generic_write_seq_multi()505and mipi_dsi_dcs_write_seq_multi() macros instead.506 507Once all callers are transitioned, the macros and the functions that they call,508mipi_dsi_generic_write_chatty() and mipi_dsi_dcs_write_buffer_chatty(), can509probably be removed. Alternatively, if people feel like the _multi() variants510are overkill for some use cases, we could keep the mipi_dsi_*_write_seq()511variants but change them not to return out of the caller.512 513Contact: Douglas Anderson <dianders@chromium.org>514 515Level: Starter516 517 518Core refactorings519=================520 521Make panic handling work522------------------------523 524This 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. The527 main issue here is that panics can be triggered from hardirq contexts and528 hence all panic related callback can run in hardirq context. It would be529 awesome if we could test at least the fbdev helper code and driver code by530 e.g. trigger calls through drm debugfs files. hardirq context could be531 achieved by using an IPI to the local processor.532 533* There's a massive confusion of different panic handlers. DRM fbdev emulation534 helpers had their own (long removed), but on top of that the fbcon code itself535 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 points537 into the DRM fbdev emulation helpers. A much cleaner approach here would be to538 switch fbcon to the `threaded printk support539 <https://lwn.net/Articles/800946/>`_.540 541* ``drm_can_sleep()`` is a mess. It hides real bugs in normal operations and542 isn't a full solution for panic paths. We need to make sure that it only543 returns true if there's a panic going on for real, and fix up all the544 fallout.545 546* The panic handler must never sleep, which also means it can't ever547 ``mutex_lock()``. Also it can't grab any other lock unconditionally, not548 even spinlocks (because NMI and hardirq can panic too). We need to either549 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 handling553 <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 the556 dread "important stuff scrolled away" problem. See `[RFC][PATCH] Oops messages557 transfer using QR codes558 <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 561Contact: Simona Vetter562 563Level: Advanced564 565Clean up the debugfs support566----------------------------567 568There's a bunch of issues with it:569 570- Convert drivers to support the drm_debugfs_add_files() function instead of571 the drm_debugfs_create_files() function.572 573- Improve late-register debugfs by rolling out the same debugfs pre-register574 infrastructure for connector and crtc too. That way, the drivers won't need to575 split their setup code into init and register anymore.576 577- We probably want to have some support for debugfs files on crtc/connectors and578 maybe other kms objects directly in core. There's even drm_print support in579 the funcs for these objects to dump kms state, so it's all there. And then the580 ->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 old583 midlayered load sequence. DRM debugfs should work more like sysfs, where you584 can create properties/files for an object anytime you want, and the core585 takes care of publishing/unpuplishing all the files at register/unregister586 time. Drivers shouldn't need to worry about these technicalities, and fixing587 this (together with the drm_minor->drm_device move) would allow us to remove588 debugfs_init.589 590Contact: Simona Vetter591 592Level: Intermediate593 594Object lifetime fixes595---------------------596 597There's two related issues here598 599- Cleanup up the various ->destroy callbacks, which often are all the same600 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 serious604 trouble even for drivers for hardware integrated on the SoC due to605 EPROBE_DEFERRED backoff.606 607Both these problems can be solved by switching over to drmm_kzalloc(), and the608various convenience wrappers provided, e.g. drmm_crtc_alloc_with_planes(),609drmm_universal_plane_alloc(), ... and so on.610 611Contact: Simona Vetter612 613Level: Intermediate614 615Remove automatic page mapping from dma-buf importing616----------------------------------------------------617 618When importing dma-bufs, the dma-buf and PRIME frameworks automatically map619imported pages into the importer's DMA area. drm_gem_prime_fd_to_handle() and620drm_gem_prime_handle_to_fd() require that importers call dma_buf_attach()621even if they never do actual device DMA, but only CPU access through622dma_buf_vmap(). This is a problem for USB devices, which do not support DMA623operations.624 625To fix the issue, automatic page mappings should be removed from the626buffer-sharing code. Fixing this is a bit more involved, since the import/export627cache is also tied to &drm_gem_object.import_attach. Meanwhile we paper over628this problem for USB devices by fishing out the USB host controller device, as629long as that supports DMA. Otherwise importing can still needlessly fail.630 631Contact: Thomas Zimmermann <tzimmermann@suse.de>, Simona Vetter632 633Level: Advanced634 635 636Better Testing637==============638 639Add unit tests using the Kernel Unit Testing (KUnit) framework640--------------------------------------------------------------641 642The `KUnit <https://www.kernel.org/doc/html/latest/dev-tools/kunit/index.html>`_643provides a common framework for unit tests within the Linux kernel. Having a644test suite would allow to identify regressions earlier.645 646A good candidate for the first unit tests are the format-conversion helpers in647``drm_format_helper.c``.648 649Contact: Javier Martinez Canillas <javierm@redhat.com>650 651Level: Intermediate652 653Clean up and document former selftests suites654---------------------------------------------655 656Some KUnit test suites (drm_buddy, drm_cmdline_parser, drm_damage_helper,657drm_format, drm_framebuffer, drm_dp_mst_helper, drm_mm, drm_plane_helper and658drm_rect) are former selftests suites that have been converted over when KUnit659was first introduced.660 661These suites were fairly undocumented, and with different goals than what unit662tests can be. Trying to identify what each test in these suites actually test663for, whether that makes sense for a unit test, and either remove it if it664doesn't or document it if it does would be of great help.665 666Contact: Maxime Ripard <mripard@kernel.org>667 668Level: Intermediate669 670Enable trinity for DRM671----------------------672 673And fix up the fallout. Should be really interesting ...674 675Level: Advanced676 677Make KMS tests in i-g-t generic678-------------------------------679 680The i915 driver team maintains an extensive testsuite for the i915 DRM driver,681including tons of testcases for corner-cases in the modesetting API. It would682be awesome if those tests (at least the ones not relying on Intel-specific GEM683features) could be made to run on any KMS driver.684 685Basic work to run i-g-t tests on non-i915 is done, what's now missing is mass-686converting things over. For modeset tests we also first need a bit of687infrastructure to use dumb buffers for untiled buffers, to be able to run all688the non-i915 specific modeset tests.689 690Level: Advanced691 692Extend virtual test driver (VKMS)693---------------------------------694 695See the documentation of :ref:`VKMS <vkms>` for more details. This is an ideal696internship task, since it only requires a virtual machine and can be sized to697fit the available time.698 699Level: See details700 701Backlight Refactoring702---------------------703 704Backlight drivers have a triple enable/disable state, which is a bit overkill.705Plan to fix this:706 7071. Roll out backlight_enable() and backlight_disable() helpers everywhere. This708 has started already.7092. In all, only look at one of the three status bits set by the above helpers.7103. Remove the other two status bits.711 712Contact: Simona Vetter713 714Level: Intermediate715 716Driver Specific717===============718 719AMD DC Display Driver720---------------------721 722AMD DC is the display driver for AMD devices starting with Vega. There has been723a bunch of progress cleaning it up but there's still plenty of work to be done.724 725See drivers/gpu/drm/amd/display/TODO for tasks.726 727Contact: Harry Wentland, Alex Deucher728 729Bootsplash730==========731 732There is support in place now for writing internal DRM clients making it733possible to pick up the bootsplash work that was rejected because it was written734for fbdev.735 736- [v6,8/8] drm/client: Hack: Add bootsplash example737 https://patchwork.freedesktop.org/patch/306579/738 739- [RFC PATCH v2 00/13] Kernel based bootsplash740 https://lore.kernel.org/r/20171213194755.3409-1-mstaudt@suse.de741 742Contact: Sam Ravnborg743 744Level: Advanced745 746Brightness handling on devices with multiple internal panels747============================================================748 749On x86/ACPI devices there can be multiple backlight firmware interfaces:750(ACPI) video, vendor specific and others. As well as direct/native (PWM)751register programming by the KMS driver.752 753To deal with this backlight drivers used on x86/ACPI call754acpi_video_get_backlight_type() which has heuristics (+quirks) to select755which backlight interface to use; and backlight drivers which do not match756the returned type will not register themselves, so that only one backlight757device gets registered (in a single GPU setup, see below).758 759At the moment this more or less assumes that there will only760be 1 (internal) panel on a system.761 762On systems with 2 panels this may be a problem, depending on763what interface acpi_video_get_backlight_type() selects:764 7651. native: in this case the KMS driver is expected to know which backlight766 device belongs to which output so everything should just work.7672. video: this does support controlling multiple backlights, but some work768 will need to be done to get the output <-> backlight device mapping769 770The above assumes both panels will require the same backlight interface type.771Things will break on systems with multiple panels where the 2 panels need772a different type of control. E.g. one panel needs ACPI video backlight control,773where as the other is using native backlight control. Currently in this case774only one of the 2 required backlight devices will get registered, based on775the acpi_video_get_backlight_type() return value.776 777If this (theoretical) case ever shows up, then supporting this will need some778work. A possible solution here would be to pass a device and connector-name779to acpi_video_get_backlight_type() so that it can deal with this.780 781Note in a way we already have a case where userspace sees 2 panels,782in dual GPU laptop setups with a mux. On those systems we may see783either 2 native backlight devices; or 2 native backlight devices.784 785Userspace already has code to deal with this by detecting if the related786panel is active (iow which way the mux between the GPU and the panels787points) and then uses that backlight device. Userspace here very much788assumes a single panel though. It picks only 1 of the 2 backlight devices789and then only uses that one.790 791Note that all userspace code (that I know off) is currently hardcoded792to assume a single panel.793 794Before 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),796userspace would see multiple backlight devices all controlling the same797backlight.798 799To deal with this userspace had to always picks one preferred device under800/sys/class/backlight and will ignore the others. So to support brightness801control on multiple panels userspace will need to be updated too.802 803There are plans to allow brightness control through the KMS API by adding804a "display brightness" property to drm_connector objects for panels. This805solves a number of issues with the /sys/class/backlight API, including not806being able to map a sysfs backlight device to a specific connector. Any807userspace changes to add support for brightness control on devices with808multiple panels really should build on top of this new KMS property.809 810Contact: Hans de Goede811 812Level: Advanced813 814Buffer age or other damage accumulation algorithm for buffer damage815===================================================================816 817Drivers that do per-buffer uploads, need a buffer damage handling (rather than818frame damage like drivers that do per-plane or per-CRTC uploads), but there is819no support to get the buffer age or any other damage accumulation algorithm.820 821For this reason, the damage helpers just fallback to a full plane update if the822framebuffer attached to a plane has changed since the last page-flip. Drivers823set &drm_plane_state.ignore_damage_clips to true as indication to824drm_atomic_helper_damage_iter_init() and drm_atomic_helper_damage_iter_next()825helpers that the damage clips should be ignored.826 827This should be improved to get damage tracking properly working on drivers that828do per-buffer uploads.829 830More information about damage tracking and references to learning materials can831be found in :ref:`damage_tracking_properties`.832 833Contact: Javier Martinez Canillas <javierm@redhat.com>834 835Level: Advanced836 837Outside DRM838===========839 840Convert fbdev drivers to DRM841----------------------------842 843There are plenty of fbdev drivers for older hardware. Some hardware has844become obsolete, but some still provides good(-enough) framebuffers. The845drivers that are still useful should be converted to DRM and afterwards846removed from fbdev.847 848Very simple fbdev drivers can best be converted by starting with a new849DRM driver. Simple KMS helpers and SHMEM should be able to handle any850existing hardware. The new driver's call-back functions are filled from851existing fbdev code.852 853More complex fbdev drivers can be refactored step-by-step into a DRM854driver with the help of the DRM fbconv helpers [4]_. These helpers provide855the transition layer between the DRM core infrastructure and the fbdev856driver interface. Create a new DRM driver on top of the fbconv helpers,857copy over the fbdev driver, and hook it up to the DRM code. Examples for858several fbdev drivers are available in Thomas Zimmermann's fbconv tree859[4]_, as well as a tutorial of this process [5]_. The result is a primitive860DRM driver that can run X11 and Weston.861 862 .. [4] https://gitlab.freedesktop.org/tzimmermann/linux/tree/fbconv863 .. [5] https://gitlab.freedesktop.org/tzimmermann/linux/blob/fbconv/drivers/gpu/drm/drm_fbconv_helper.c864 865Contact: Thomas Zimmermann <tzimmermann@suse.de>866 867Level: Advanced868