1What are these headers ? 2------------------------ 3This is the canonical source of drm headers that user space should use for 4communicating with the kernel DRM subsystem. 5 6They flow from the kernel, thus any changes must be merged there first. 7Do _not_ attempt to "fix" these by deviating from the kernel ones ! 8 9 10Non-linux platforms - changes/patches 11------------------------------------- 12If your platform has local changes, please send them upstream for inclusion. 13Even if your patches don't get accepted in their current form, devs will 14give you feedback on how to address things properly. 15 16git send-email --subject-prefix="PATCH libdrm" your patches to dri-devel 17mailing list. 18 19Before doing so, please consider the following: 20 - Have the [libdrm vs kernel] headers on your platform deviated ? 21Consider unifying them first. 22 23 - Have you introduced additional ABI that's not available in Linux ? 24Propose it for [Linux kernel] upstream inclusion. 25If that doesn't work out (hopefully it never does), move it to another header 26and/or keep the change(s) local ? 27 28 - Are your changes DRI1/UMS specific ? 29There is virtually no interest/power in keeping those legacy interfaces. They 30are around due to the kernel "thou shalt not break existing user space" rule. 31 32Consider porting the driver to DRI2/KMS - all (almost?) sensible hardware is 33capable of supporting those. 34 35 36Which headers go where ? 37------------------------ 38A snipped from the, now removed, Makefile.am used to state: 39 40 XXX airlied says, nothing besides *_drm.h and drm*.h should be necessary. 41 however, r300 and via need their reg headers installed in order to build. 42 better solutions are welcome. 43 44Obviously the r300 and via headers are no longer around ;-) 45 46Reason behind is that the drm headers can be used as a basic communications 47channel with the respective kernel modules. If more advanced functionality is 48required one can pull the specific libdrm_$driver which is free to pull 49additional files from the kernel. 50 51For example: nouveau has nouveau/nvif/*.h while vc4 has vc4/*.h 52 53If your driver is still in prototyping/staging state, consider moving the 54$driver_drm.h into $driver and _not_ installing it. An header providing opaque 55definitions and access [via $driver_drmif.h or similar] would be better fit. 56 57 58When and which headers to update 59-------------------------------- 60Ideally all files will be synced (updated) with the latest released kernel on 61each libdrm release. Sadly that's not yet possible since quite a few headers 62differ significantly - see Outdated or Broken Headers section below. 63 64That said, it's up-to the individual developers to sync with newer version 65(from drm-next) as they see fit. 66 67 68When and how to update these files 69---------------------------------- 70In order to update the files do the following: 71 - Switch to a Linux kernel tree/branch which is not rebased. 72For example: airlied/drm-next 73 - Install the headers via `make headers_install' to a separate location. 74 - Copy the drm header[s] + git add + git commit. 75 - Note: Your commit message must include: 76 a) Brief summary on the delta. If there's any change that looks like an 77API/ABI break one _must_ explicitly state why it's safe to do so. 78 b) "Generated using make headers_install." 79 c) "Generated from $tree/branch commit $sha" 80 81 82Outdated or Broken Headers 83-------------------------- 84This section contains a list of headers and the respective "issues" they might 85have relative to their kernel equivalent. 86 87Nearly all headers: 88 - Missing extern C notation. 89Status: Trivial. 90 91Most UMS headers: 92 - Not using fixed size integers - compat ioctls are broken. 93Status: ? 94Promote to fixed size ints, which match the current (32bit) ones. 95 96 97amdgpu_drm.h 98 - Using the stdint.h uint*_t over the respective __u* ones 99Status: Trivial. 100 101drm_mode.h 102 - Missing DPI encode/connector pair. 103Status: Trivial. 104 105i915_drm.h 106 - Missing PARAMS - HAS_POOLED_EU, MIN_EU_IN_POOL CONTEXT_PARAM_NO_ERROR_CAPTURE 107Status: Trivial. 108 109mga_drm.h 110 - Typo fix, use struct over typedef. 111Status: Trivial. 112 113nouveau_drm.h 114 - Missing macros NOUVEAU_GETPARAM*, NOUVEAU_DRM_HEADER_PATCHLEVEL, structs, 115enums, using stdint.h over the __u* types. 116Status: ? 117 118qxl_drm.h 119 - Using the stdint.h uint*_t over the respective __u* ones 120Status: Trivial. 121 122r128_drm.h 123 - Broken compat ioctls. 124 125radeon_drm.h 126 - Missing RADEON_TILING_R600_NO_SCANOUT, CIK_TILE_MODE_*, broken UMS ioctls, 127using stdint types. 128 - Both kernel and libdrm: missing padding - 129drm_radeon_gem_{create,{g,s}et_tiling,set_domain} others ? 130Status: ? 131 132savage_drm.h 133 - Renamed ioctls - DRM_IOCTL_SAVAGE_{,BCI}_EVENT_EMIT, compat ioctls are broken. 134Status: ? 135 136sis_drm.h 137 - Borken ioctls + libdrm uses int vs kernel long 138Status: ? 139 140via_drm.h 141 - Borken ioctls - libdrm int vs kernel long 142Status: ? 143 144 145omap_drm.h (living in $TOP/omap) 146 - License mismatch, missing DRM_IOCTL_OMAP_GEM_NEW and related struct 147Status: ? 148 149msm_drm.h (located in $TOP/freedreno/msm/) 150 - License mismatch, missing MSM_PIPE_*, MSM_SUBMIT_*. Renamed 151drm_msm_gem_submit::flags, missing drm_msm_gem_submit::fence_fd. 152Status: ? 153 154exynos_drm.h (living in $TOP/exynos) 155 - License mismatch, now using fixed size ints (but not everywhere). Lots of 156new stuff. 157Status: ? 158