aboutsummaryrefslogtreecommitdiff
path: root/backend/drm
AgeCommit message (Collapse)Author
2017-09-23Simplify retry_pageflipScott Anderson
2017-09-23Add error recovery for failed page flipsScott Anderson
2017-09-23Add timeout to DRM cleanupScott Anderson
This prevents a potential infinite loop preventing us from closing properly.
2017-09-23Print connector name on atomic commit failureScott Anderson
2017-09-23Allow forcing legacy DRM interfaceScott Anderson
2017-09-10Remove non-blocking flag on modesetting commitsScott Anderson
This seems to make hotplugging DisplayPort MST connectors more reliable.
2017-09-09Retry atomic commit on failureScott Anderson
So we don't lose any previously "good" properties that were set.
2017-09-09Check for disappearing DRM connectorsScott Anderson
DisplayPort MST connectors will disappear when they're disconnected, so we need to check for that.
2017-09-06Add set_gamma and get_gamma_size to wlr_output_implemersion
2017-08-29Merge pull request #120 from acrisci/feature/wlr-cursorDrew DeVault
wlr_cursor
2017-08-26add wlr_cursor basic implementationTony Crisci
2017-08-26Merge wlr_udev into wlr_sessionScott Anderson
2017-08-19wlr renderer/texture: rename init to create when it does allocDominique Martinet
2017-08-18drm realloc_crtcs: check we found an outputDominique Martinet
This follows-up on #105 that just initialized to 0. Instead, make sure we actually found an output that matches
2017-08-18fix uninitialized warningTony Crisci
`index` might be uninitialized. Warning can be produced with gcc and `-01` flag.
2017-08-16Fix DRM problem introduced in 901c14cDrew DeVault
2017-08-16Prevent alloc errors from crashing in `list_t`Calvin Lee
This commit changes the `list_t` api so that alloc errors can be detected and worked around. Also fixes errors not found in 5cc7342
2017-08-14Refactor out wlr_output_mode_stateDrew DeVault
2017-08-13Add backend detection functionsCalvin Lee
Resolves #80
2017-08-13Remove wlr_output_state, update backendsDrew DeVault
2017-08-13wlr_backend_init -> wlr_backend_startDrew DeVault
Also renames create to init. We'll use create for anything that allocates and init for anything that takes a pointer and initializes it.
2017-08-12Refactor wlr_backend_state out of DRM backendDrew DeVault
2017-08-12Use libinput_event_destroy/xkb_*_unrefDominique Martinet
Also sneak in a missing drmModeFreePlaneResources
2017-08-11Complete a few destroy functionsDominique Martinet
This adds missing free calls for: - drm outputs - libinput backend in general - final udev free - output mode state
2017-08-10Merge pull request #53 from nyorain/drm_bufferDrew DeVault
Implement drm (egl) buffer attaching
2017-08-10Refactor EGL handlingDrew DeVault
2017-08-11Fix possible use of CRTC after monitor is unpluggedScott Anderson
2017-08-10Implement drm (egl) buffer attachingnyorain
2017-08-10Added null check on crtcScott Anderson
2017-08-09Atomic modesettingScott Anderson
2017-08-08Rename wlr_surface -> wlr_texture; attach -> uploadnyorain
2017-08-07Use gbm_bo_map for cursorScott Anderson
2017-08-07Style changesScott Anderson
2017-08-06Moved headersScott Anderson
2017-08-06Renamed some variables to be less genericScott Anderson
2017-08-06Updated DRM cursor renderingScott Anderson
2017-08-06Clean up resources when plane is reassignedScott Anderson
2017-08-06More cleanupsScott Anderson
2017-08-06Various cleanupsScott Anderson
2017-08-06Split off some functions into drm-utilScott Anderson
2017-08-06Use plane for renderingScott Anderson
2017-08-06Add planes.Scott Anderson
2017-08-06Upgraded CRTC and Encoder matchingScott Anderson
2017-08-06Various changes/cleanupsScott Anderson
2017-08-06DRM resource initalisationScott Anderson
2017-08-06Added DRM property infrastructureScott Anderson
2017-08-06Changed header paths.Scott Anderson
2017-07-27Use ARGB8888 pixel format for cursor boTony Crisci
The previous pixel format (RGBA8888) is not supported on all devices for cursors. Weston uses ARGB8888 exclusively for its cursor buffer object which leads me to believe this pixel format is better supported.
2017-07-24bugfix: add null check on output gbm on pageflipTony Crisci
The gbm for the output might be null for the pageflip in the case that the output has been disconnected. The gbm might be set to null by wlr_drm_output_cleanup() in this case. If the output is cleaned up before the pageflip, then a double free will crash the compositor on the call to gbm_surface_release_buffer() in the pageflip handler. The outputs buffer object bo[1] will point to invalid memory.
2017-06-26Log max cursor size on errorDrew DeVault