]>
Commit | Line | Data |
---|---|---|
0e70dad0 TR |
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 | Subsystem-wide refactorings | |
11 | =========================== | |
12 | ||
13 | De-midlayer drivers | |
14 | ------------------- | |
15 | ||
16 | With the recent ``drm_bus`` cleanup patches for 3.17 it is no longer required | |
17 | to have a ``drm_bus`` structure set up. Drivers can directly set up the | |
18 | ``drm_device`` structure instead of relying on bus methods in ``drm_usb.c`` | |
085c6c09 | 19 | and ``drm_pci.c``. The goal is to get rid of the driver's ``->load`` / |
0e70dad0 TR |
20 | ``->unload`` callbacks and open-code the load/unload sequence properly, using |
21 | the new two-stage ``drm_device`` setup/teardown. | |
22 | ||
23 | Once all existing drivers are converted we can also remove those bus support | |
24 | files for USB and platform devices. | |
25 | ||
26 | All you need is a GPU for a non-converted driver (currently almost all of | |
27 | them, but also all the virtual ones used by KVM, so everyone qualifies). | |
28 | ||
29 | Contact: Daniel Vetter, Thierry Reding, respective driver maintainers | |
30 | ||
39dea70d DV |
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37 | implementation (there's lots of outdated locking leftovers in various | |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: Daniel Vetter, respective driver maintainers | |
41 | ||
0e70dad0 TR |
42 | Convert existing KMS drivers to atomic modesetting |
43 | -------------------------------------------------- | |
44 | ||
45 | 3.19 has the atomic modeset interfaces and helpers, so drivers can now be | |
46 | converted over. Modern compositors like Wayland or Surfaceflinger on Android | |
47 | really want an atomic modeset interface, so this is all about the bright | |
48 | future. | |
49 | ||
50 | There is a conversion guide for atomic and all you need is a GPU for a | |
51 | non-converted driver (again virtual HW drivers for KVM are still all | |
52 | suitable). | |
53 | ||
54 | As part of this drivers also need to convert to universal plane (which means | |
55 | exposing primary & cursor as proper plane objects). But that's much easier to | |
56 | do by directly using the new atomic helper driver callbacks. | |
57 | ||
58 | Contact: Daniel Vetter, respective driver maintainers | |
1a80cc1c DV |
59 | |
60 | Clean up the clipped coordination confusion around planes | |
61 | --------------------------------------------------------- | |
62 | ||
63 | We have a helper to get this right with drm_plane_helper_check_update(), but | |
64 | it's not consistently used. This should be fixed, preferrably in the atomic | |
65 | helpers (and drivers then moved over to clipped coordinates). Probably the | |
66 | helper should also be moved from drm_plane_helper.c to the atomic helpers, to | |
67 | avoid confusion - the other helpers in that file are all deprecated legacy | |
68 | helpers. | |
69 | ||
70 | Contact: Ville Syrjälä, Daniel Vetter, driver maintainers | |
4e8be453 | 71 | |
0e70dad0 TR |
72 | Convert early atomic drivers to async commit helpers |
73 | ---------------------------------------------------- | |
74 | ||
75 | For the first year the atomic modeset helpers didn't support asynchronous / | |
76 | nonblocking commits, and every driver had to hand-roll them. This is fixed | |
77 | now, but there's still a pile of existing drivers that easily could be | |
78 | converted over to the new infrastructure. | |
79 | ||
80 | One issue with the helpers is that they require that drivers handle completion | |
81 | events for atomic commits correctly. But fixing these bugs is good anyway. | |
82 | ||
83 | Contact: Daniel Vetter, respective driver maintainers | |
84 | ||
f217f554 DV |
85 | Better manual-upload support for atomic |
86 | --------------------------------------- | |
87 | ||
88 | This would be especially useful for tinydrm: | |
89 | ||
90 | - Add a struct drm_rect dirty_clip to drm_crtc_state. When duplicating the | |
91 | crtc state, clear that to the max values, x/y = 0 and w/h = MAX_INT, in | |
92 | __drm_atomic_helper_crtc_duplicate_state(). | |
93 | ||
0cac6ac1 TR |
94 | - Move tinydrm_merge_clips into drm_framebuffer.c, dropping the tinydrm\_ |
95 | prefix ofc and using drm_fb\_. drm_framebuffer.c makes sense since this | |
f217f554 DV |
96 | is a function useful to implement the fb->dirty function. |
97 | ||
98 | - Create a new drm_fb_dirty function which does essentially what e.g. | |
99 | mipi_dbi_fb_dirty does. You can use e.g. drm_atomic_helper_update_plane as the | |
100 | template. But instead of doing a simple full-screen plane update, this new | |
101 | helper also sets crtc_state->dirty_clip to the right coordinates. And of | |
102 | course it needs to check whether the fb is actually active (and maybe where), | |
103 | so there's some book-keeping involved. There's also some good fun involved in | |
104 | scaling things appropriately. For that case we might simply give up and | |
105 | declare the entire area covered by the plane as dirty. | |
106 | ||
107 | Contact: Noralf Trønnes, Daniel Vetter | |
108 | ||
0e70dad0 TR |
109 | Fallout from atomic KMS |
110 | ----------------------- | |
111 | ||
112 | ``drm_atomic_helper.c`` provides a batch of functions which implement legacy | |
113 | IOCTLs on top of the new atomic driver interface. Which is really nice for | |
114 | gradual conversion of drivers, but unfortunately the semantic mismatches are | |
115 | a bit too severe. So there's some follow-up work to adjust the function | |
116 | interfaces to fix these issues: | |
117 | ||
118 | * atomic needs the lock acquire context. At the moment that's passed around | |
119 | implicitly with some horrible hacks, and it's also allocate with | |
120 | ``GFP_NOFAIL`` behind the scenes. All legacy paths need to start allocating | |
121 | the acquire context explicitly on stack and then also pass it down into | |
122 | drivers explicitly so that the legacy-on-atomic functions can use them. | |
123 | ||
2ec04b33 DV |
124 | Except for some driver code this is done. This task should be finished by |
125 | adding WARN_ON(!drm_drv_uses_atomic_modeset) in drm_modeset_lock_all(). | |
be05fe13 | 126 | |
0e70dad0 TR |
127 | * A bunch of the vtable hooks are now in the wrong place: DRM has a split |
128 | between core vfunc tables (named ``drm_foo_funcs``), which are used to | |
129 | implement the userspace ABI. And then there's the optional hooks for the | |
130 | helper libraries (name ``drm_foo_helper_funcs``), which are purely for | |
131 | internal use. Some of these hooks should be move from ``_funcs`` to | |
132 | ``_helper_funcs`` since they are not part of the core ABI. There's a | |
133 | ``FIXME`` comment in the kerneldoc for each such case in ``drm_crtc.h``. | |
134 | ||
0e70dad0 TR |
135 | Contact: Daniel Vetter |
136 | ||
137 | Get rid of dev->struct_mutex from GEM drivers | |
138 | --------------------------------------------- | |
139 | ||
140 | ``dev->struct_mutex`` is the Big DRM Lock from legacy days and infested | |
141 | everything. Nowadays in modern drivers the only bit where it's mandatory is | |
142 | serializing GEM buffer object destruction. Which unfortunately means drivers | |
143 | have to keep track of that lock and either call ``unreference`` or | |
144 | ``unreference_locked`` depending upon context. | |
145 | ||
146 | Core GEM doesn't have a need for ``struct_mutex`` any more since kernel 4.8, | |
147 | and there's a ``gem_free_object_unlocked`` callback for any drivers which are | |
148 | entirely ``struct_mutex`` free. | |
149 | ||
150 | For drivers that need ``struct_mutex`` it should be replaced with a driver- | |
151 | private lock. The tricky part is the BO free functions, since those can't | |
152 | reliably take that lock any more. Instead state needs to be protected with | |
153 | suitable subordinate locks or some cleanup work pushed to a worker thread. For | |
154 | performance-critical drivers it might also be better to go with a more | |
2ec04b33 DV |
155 | fine-grained per-buffer object and per-context lockings scheme. Currently only the |
156 | ``msm`` driver still use ``struct_mutex``. | |
0e70dad0 | 157 | |
085c6c09 | 158 | Contact: Daniel Vetter, respective driver maintainers |
0e70dad0 | 159 | |
45ae2787 SP |
160 | Convert instances of dev_info/dev_err/dev_warn to their DRM_DEV_* equivalent |
161 | ---------------------------------------------------------------------------- | |
162 | ||
163 | For drivers which could have multiple instances, it is necessary to | |
164 | differentiate between which is which in the logs. Since DRM_INFO/WARN/ERROR | |
165 | don't do this, drivers used dev_info/warn/err to make this differentiation. We | |
166 | now have DRM_DEV_* variants of the drm print macros, so we can start to convert | |
167 | those drivers back to using drm-formwatted specific log messages. | |
168 | ||
9f446781 DV |
169 | Before you start this conversion please contact the relevant maintainers to make |
170 | sure your work will be merged - not everyone agrees that the DRM dmesg macros | |
171 | are better. | |
172 | ||
45ae2787 SP |
173 | Contact: Sean Paul, Maintainer of the driver you plan to convert |
174 | ||
3233fc0a NT |
175 | Convert drivers to use simple modeset suspend/resume |
176 | ---------------------------------------------------- | |
177 | ||
178 | Most drivers (except i915 and nouveau) that use | |
179 | drm_atomic_helper_suspend/resume() can probably be converted to use | |
2ec04b33 DV |
180 | drm_mode_config_helper_suspend/resume(). Also there's still open-coded version |
181 | of the atomic suspend/resume code in older atomic modeset drivers. | |
3233fc0a NT |
182 | |
183 | Contact: Maintainer of the driver you plan to convert | |
184 | ||
ee05baa0 NT |
185 | Convert drivers to use drm_fb_helper_fbdev_setup/teardown() |
186 | ----------------------------------------------------------- | |
187 | ||
188 | Most drivers can use drm_fb_helper_fbdev_setup() except maybe: | |
189 | ||
190 | - amdgpu which has special logic to decide whether to call | |
191 | drm_helper_disable_unused_functions() | |
192 | ||
193 | - armada which isn't atomic and doesn't call | |
194 | drm_helper_disable_unused_functions() | |
195 | ||
196 | - i915 which calls drm_fb_helper_initial_config() in a worker | |
197 | ||
198 | Drivers that use drm_framebuffer_remove() to clean up the fbdev framebuffer can | |
199 | probably use drm_fb_helper_fbdev_teardown(). | |
200 | ||
201 | Contact: Maintainer of the driver you plan to convert | |
202 | ||
66499101 DV |
203 | Clean up mmap forwarding |
204 | ------------------------ | |
205 | ||
206 | A lot of drivers forward gem mmap calls to dma-buf mmap for imported buffers. | |
207 | And also a lot of them forward dma-buf mmap to the gem mmap implementations. | |
208 | Would be great to refactor this all into a set of small common helpers. | |
209 | ||
210 | Contact: Daniel Vetter | |
211 | ||
212 | Put a reservation_object into drm_gem_object | |
213 | -------------------------------------------- | |
214 | ||
215 | This would remove the need for the ->gem_prime_res_obj callback. It would also | |
216 | allow us to implement generic helpers for waiting for a bo, allowing for quite a | |
217 | bit of refactoring in the various wait ioctl implementations. | |
218 | ||
219 | Contact: Daniel Vetter | |
220 | ||
1aecabb5 DV |
221 | idr_init_base() |
222 | --------------- | |
223 | ||
224 | DRM core&drivers uses a lot of idr (integer lookup directories) for mapping | |
225 | userspace IDs to internal objects, and in most places ID=0 means NULL and hence | |
226 | is never used. Switching to idr_init_base() for these would make the idr more | |
227 | efficient. | |
228 | ||
229 | Contact: Daniel Vetter | |
230 | ||
f0014881 NT |
231 | Defaults for .gem_prime_import and export |
232 | ----------------------------------------- | |
233 | ||
234 | Most drivers don't need to set drm_driver->gem_prime_import and | |
235 | ->gem_prime_export now that drm_gem_prime_import() and drm_gem_prime_export() | |
236 | are the default. | |
237 | ||
b39b5394 NT |
238 | struct drm_gem_object_funcs |
239 | --------------------------- | |
240 | ||
241 | GEM objects can now have a function table instead of having the callbacks on the | |
242 | DRM driver struct. This is now the preferred way and drivers can be moved over. | |
243 | ||
0e70dad0 TR |
244 | Core refactorings |
245 | ================= | |
246 | ||
0e70dad0 TR |
247 | Clean up the DRM header mess |
248 | ---------------------------- | |
249 | ||
2ec04b33 DV |
250 | The DRM subsystem originally had only one huge global header, ``drmP.h``. This |
251 | is now split up, but many source files still include it. The remaining part of | |
252 | the cleanup work here is to replace any ``#include <drm/drmP.h>`` by only the | |
253 | headers needed (and fixing up any missing pre-declarations in the headers). | |
0e70dad0 | 254 | |
085c6c09 DV |
255 | In the end no .c file should need to include ``drmP.h`` anymore. |
256 | ||
0e70dad0 TR |
257 | Contact: Daniel Vetter |
258 | ||
259 | Add missing kerneldoc for exported functions | |
260 | -------------------------------------------- | |
261 | ||
262 | The DRM reference documentation is still lacking kerneldoc in a few areas. The | |
263 | task would be to clean up interfaces like moving functions around between | |
264 | files to better group them and improving the interfaces like dropping return | |
265 | values for functions that never fail. Then write kerneldoc for all exported | |
ff41c419 | 266 | functions and an overview section and integrate it all into the drm book. |
0e70dad0 TR |
267 | |
268 | See https://dri.freedesktop.org/docs/drm/ for what's there already. | |
269 | ||
270 | Contact: Daniel Vetter | |
271 | ||
0e70dad0 TR |
272 | Make panic handling work |
273 | ------------------------ | |
274 | ||
275 | This is a really varied tasks with lots of little bits and pieces: | |
276 | ||
277 | * The panic path can't be tested currently, leading to constant breaking. The | |
278 | main issue here is that panics can be triggered from hardirq contexts and | |
279 | hence all panic related callback can run in hardirq context. It would be | |
280 | awesome if we could test at least the fbdev helper code and driver code by | |
281 | e.g. trigger calls through drm debugfs files. hardirq context could be | |
282 | achieved by using an IPI to the local processor. | |
283 | ||
284 | * There's a massive confusion of different panic handlers. DRM fbdev emulation | |
285 | helpers have one, but on top of that the fbcon code itself also has one. We | |
286 | need to make sure that they stop fighting over each another. | |
287 | ||
288 | * ``drm_can_sleep()`` is a mess. It hides real bugs in normal operations and | |
289 | isn't a full solution for panic paths. We need to make sure that it only | |
290 | returns true if there's a panic going on for real, and fix up all the | |
291 | fallout. | |
292 | ||
293 | * The panic handler must never sleep, which also means it can't ever | |
294 | ``mutex_lock()``. Also it can't grab any other lock unconditionally, not | |
295 | even spinlocks (because NMI and hardirq can panic too). We need to either | |
296 | make sure to not call such paths, or trylock everything. Really tricky. | |
297 | ||
298 | * For the above locking troubles reasons it's pretty much impossible to | |
299 | attempt a synchronous modeset from panic handlers. The only thing we could | |
300 | try to achive is an atomic ``set_base`` of the primary plane, and hope that | |
301 | it shows up. Everything else probably needs to be delayed to some worker or | |
302 | something else which happens later on. Otherwise it just kills the box | |
303 | harder, prevent the panic from going out on e.g. netconsole. | |
304 | ||
305 | * There's also proposal for a simplied DRM console instead of the full-blown | |
306 | fbcon and DRM fbdev emulation. Any kind of panic handling tricks should | |
307 | obviously work for both console, in case we ever get kmslog merged. | |
308 | ||
309 | Contact: Daniel Vetter | |
310 | ||
0cad7f71 DV |
311 | Clean up the debugfs support |
312 | ---------------------------- | |
313 | ||
314 | There's a bunch of issues with it: | |
315 | ||
316 | - The drm_info_list ->show() function doesn't even bother to cast to the drm | |
317 | structure for you. This is lazy. | |
318 | ||
319 | - We probably want to have some support for debugfs files on crtc/connectors and | |
320 | maybe other kms objects directly in core. There's even drm_print support in | |
321 | the funcs for these objects to dump kms state, so it's all there. And then the | |
322 | ->show() functions should obviously give you a pointer to the right object. | |
323 | ||
324 | - The drm_info_list stuff is centered on drm_minor instead of drm_device. For | |
325 | anything we want to print drm_device (or maybe drm_file) is the right thing. | |
326 | ||
327 | - The drm_driver->debugfs_init hooks we have is just an artifact of the old | |
328 | midlayered load sequence. DRM debugfs should work more like sysfs, where you | |
329 | can create properties/files for an object anytime you want, and the core | |
330 | takes care of publishing/unpuplishing all the files at register/unregister | |
331 | time. Drivers shouldn't need to worry about these technicalities, and fixing | |
332 | this (together with the drm_minor->drm_device move) would allow us to remove | |
333 | debugfs_init. | |
334 | ||
335 | Contact: Daniel Vetter | |
336 | ||
81a7bd4a DV |
337 | KMS cleanups |
338 | ------------ | |
339 | ||
340 | Some of these date from the very introduction of KMS in 2008 ... | |
341 | ||
342 | - drm_mode_config.crtc_idr is misnamed, since it contains all KMS object. Should | |
343 | be renamed to drm_mode_config.object_idr. | |
344 | ||
345 | - drm_display_mode doesn't need to be derived from drm_mode_object. That's | |
346 | leftovers from older (never merged into upstream) KMS designs where modes | |
347 | where set using their ID, including support to add/remove modes. | |
348 | ||
e6a3e405 DV |
349 | - Make ->funcs and ->helper_private vtables optional. There's a bunch of empty |
350 | function tables in drivers, but before we can remove them we need to make sure | |
351 | that all the users in helpers and drivers do correctly check for a NULL | |
352 | vtable. | |
353 | ||
354 | - Cleanup up the various ->destroy callbacks. A lot of them just wrapt the | |
355 | drm_*_cleanup implementations and can be removed. Some tack a kfree() at the | |
356 | end, for which we could add drm_*_cleanup_kfree(). And then there's the (for | |
357 | historical reasons) misnamed drm_primary_helper_destroy() function. | |
358 | ||
0e70dad0 TR |
359 | Better Testing |
360 | ============== | |
361 | ||
362 | Enable trinity for DRM | |
363 | ---------------------- | |
364 | ||
365 | And fix up the fallout. Should be really interesting ... | |
366 | ||
367 | Make KMS tests in i-g-t generic | |
368 | ------------------------------- | |
369 | ||
370 | The i915 driver team maintains an extensive testsuite for the i915 DRM driver, | |
371 | including tons of testcases for corner-cases in the modesetting API. It would | |
372 | be awesome if those tests (at least the ones not relying on Intel-specific GEM | |
373 | features) could be made to run on any KMS driver. | |
374 | ||
375 | Basic work to run i-g-t tests on non-i915 is done, what's now missing is mass- | |
376 | converting things over. For modeset tests we also first need a bit of | |
377 | infrastructure to use dumb buffers for untiled buffers, to be able to run all | |
378 | the non-i915 specific modeset tests. | |
379 | ||
ad9ff96f HM |
380 | Extend virtual test driver (VKMS) |
381 | --------------------------------- | |
382 | ||
383 | See the documentation of :ref:`VKMS <vkms>` for more details. This is an ideal | |
384 | internship task, since it only requires a virtual machine and can be sized to | |
385 | fit the available time. | |
386 | ||
0e70dad0 TR |
387 | Contact: Daniel Vetter |
388 | ||
0e70dad0 TR |
389 | Driver Specific |
390 | =============== | |
391 | ||
f217f554 DV |
392 | tinydrm |
393 | ------- | |
394 | ||
395 | Tinydrm is the helper driver for really simple fb drivers. The goal is to make | |
396 | those drivers as simple as possible, so lots of room for refactoring: | |
397 | ||
398 | - backlight helpers, probably best to put them into a new drm_backlight.c. | |
399 | This is because drivers/video is de-facto unmaintained. We could also | |
400 | move drivers/video/backlight to drivers/gpu/backlight and take it all | |
9949b355 MM |
401 | over within drm-misc, but that's more work. Backlight helpers require a fair |
402 | bit of reworking and refactoring. A simple example is the enabling of a backlight. | |
403 | Tinydrm has helpers for this. It would be good if other drivers can also use the | |
404 | helper. However, there are various cases we need to consider i.e different | |
405 | drivers seem to have different ways of enabling/disabling a backlight. | |
406 | We also need to consider the backlight drivers (like gpio_backlight). The situation | |
407 | is further complicated by the fact that the backlight is tied to fbdev | |
408 | via fb_notifier_callback() which has complicated logic. For further details, refer | |
409 | to the following discussion thread: | |
410 | https://groups.google.com/forum/#!topic/outreachy-kernel/8rBe30lwtdA | |
f217f554 DV |
411 | |
412 | - spi helpers, probably best put into spi core/helper code. Thierry said | |
413 | the spi maintainer is fast&reactive, so shouldn't be a big issue. | |
414 | ||
415 | - extract the mipi-dbi helper (well, the non-tinydrm specific parts at | |
416 | least) into a separate helper, like we have for mipi-dsi already. Or follow | |
417 | one of the ideas for having a shared dsi/dbi helper, abstracting away the | |
418 | transport details more. | |
419 | ||
f217f554 DV |
420 | - tinydrm_gem_cma_prime_import_sg_table should probably go into the cma |
421 | helpers, as a _vmapped variant (since not every driver needs the vmap). | |
422 | And tinydrm_gem_cma_free_object could the be merged into | |
423 | drm_gem_cma_free_object(). | |
424 | ||
425 | - tinydrm_fb_create we could move into drm_simple_pipe, only need to add | |
426 | the fb_create hook to drm_simple_pipe_funcs, which would again simplify a | |
427 | bunch of things (since it gives you a one-stop vfunc for simple drivers). | |
428 | ||
429 | - Quick aside: The unregister devm stuff is kinda getting the lifetimes of | |
430 | a drm_device wrong. Doesn't matter, since everyone else gets it wrong | |
431 | too :-) | |
432 | ||
f217f554 DV |
433 | - also rework the drm_framebuffer_funcs->dirty hook wire-up, see above. |
434 | ||
435 | Contact: Noralf Trønnes, Daniel Vetter | |
436 | ||
0a26a45d HW |
437 | AMD DC Display Driver |
438 | --------------------- | |
439 | ||
440 | AMD DC is the display driver for AMD devices starting with Vega. There has been | |
441 | a bunch of progress cleaning it up but there's still plenty of work to be done. | |
442 | ||
443 | See drivers/gpu/drm/amd/display/TODO for tasks. | |
444 | ||
445 | Contact: Harry Wentland, Alex Deucher | |
446 | ||
7b3b61b6 DV |
447 | i915 |
448 | ---- | |
449 | ||
450 | - Our early/late pm callbacks could be removed in favour of using | |
451 | device_link_add to model the dependency between i915 and snd_had. See | |
452 | https://dri.freedesktop.org/docs/drm/driver-api/device_link.html | |
453 | ||
0e70dad0 TR |
454 | Outside DRM |
455 | =========== |