summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge remote-tracking branch 'androidx86/marshmallow-x86' into replicant-6.0HEADreplicant-6.0-0001replicant-6.0Wolfgang Wiedmeyer2017-02-121-7/+2
|\
| * remove PIPE_BIND_TRANSFER_READ/WRITE for compatibility with mesa 12.1Mauro Rossi2016-09-101-5/+0
| | | | | | | | | | | | | | | | | | | | | | This changes are the necessary complement to corresponding mesa commit 5981ab5 "gallium: remove PIPE_BIND_TRANSFER_READ/WRITE" Fixes the following building errors: external/drm_gralloc/gralloc_drm_pipe.c:102:11: error: use of undeclared identifier 'PIPE_BIND_TRANSFER_READ' ... external/drm_gralloc/gralloc_drm_pipe.c:104:11: error: use of undeclared identifier 'PIPE_BIND_TRANSFER_WRITE'
| * android: fix gralloc_drm_pipe building error with mesa 12.1Mauro Rossi2016-08-271-2/+2
| | | | | | | | | | | | | | | | Fixes a building error introduced by commit 9daaa6f "gallium: add a pipe_context parameter to resource_get_handle" The new parameter was introduced for radeonsi future uses and can be optionally set to NULL for now
* | enable drm_gralloc module for swrastWolfgang Wiedmeyer2016-08-271-5/+5
|/ | | | | | | | | Even if only swrast/llvmpipe is selected, there is still a dependency on drm_gralloc. Removing the dependency seems to be too much work. The path for the Mesa repo needs to be adapted. Signed-off-by: Wolfgang Wiedmeyer <wolfgit@wiedmeyer.de>
* gralloc_drm: show the unsupported error if drv is nullChih-Wei Huang2016-07-111-2/+2
| | | | | | | | | The unsupported driver message was suppressed by a subtle bug of commit cbf96c0d which treats the latest checking as an else case. Remove the unnecessary else. Besides, logically on entering ENABLE_PIPE snippet, drv is always null. No need to check it.
* drm_gralloc: fix random crash with wildpointerMark Yao2016-07-061-1/+10
| | | | | | | | | | | two drm handle may use same bo, but there is no reference protect. if one of the drm handle release the bo, another handle's bo become a wildpointer, any read/write on the wildpointer will cause system unstable, crash. Change-Id: Ieaca522e3372dba82c48961499b9b657ca33cd15 Signed-off-by: Mark Yao <mark.yao@rock-chips.com>
* Support multiple exteranl dispaly in mirror modeWuZhen2016-06-272-124/+107
| | | | | | and extend support beyong hdmi. Change-Id: Ib9550eac0330807ccd4c3919609cbf02ad98299d
* Maintain a list of connector outputsWuZhen2016-06-277-85/+130
| | | | | | Refactor code to support multiple mirrored outputs following code. Change-Id: I60781d707ae68db2dbd2c4c4122b9183633a74be
* Rename hdmi_output_modeWuZhen2016-06-272-8/+10
| | | | | | | Rename it to drm_output_mode and move it into gralloc_drm_output in preparation for folowing changes. Change-Id: I488f395feede9c9a20eadb9145ebf46295ba4a71
* Best fit the external displayMa Jian2016-06-271-5/+54
| | | | | | | | | | | | | | | NO_REF_TASK tested: 1) set persist.remixos.disp_best_fit to true 2) plugin the HDMI, the resolution should be the one fit the primary screen (i.e. the resolution should be larger than or equal to primery resolution, and it is the closest one) 3) set persist.remixos.disp_best_fit to false 4) plugin HDMI again, the external display will use the default resolution. Change-Id: I80b0fe6e7cc676ce1915f07f267204734bfc645f
* nouveau: add support for latest Kepler and Maxwell chipsetsMauro Rossi2016-06-131-0/+2
| | | | 0x100 and 0x120 chipsets were added in gralloc_drm_nouveau.c
* android: fix gralloc_drm_pipe building error with mesa 11.3Mauro Rossi2016-06-131-3/+3
| | | | | | | | Fixes a building error introduced by commit 82db518 "gallium: add external usage flags to resource_from(get)_handle (v2)" parameter usage is required as a combination of PIPE_HANDLE_USAGE_* flags, PIPE_HANDLE_USAGE_READ_WRITE was used as the most general case
* avoid issue with devices that don't support blit.WuZhen2016-05-121-0/+5
| | | | Change-Id: I398258147f9adcfbbcf2d92c7ba19887ea917bd0
* Revert "TEMP HACK: make non-standalone surfaceflinger work"WuZhen2016-05-121-1/+0
| | | | | | This reverts commit b19bbdbf98742f0428f6f4265b5e80d569608abd. Change-Id: I351b1c3f149924f38635058004eb91d84351646a
* support hotplug with vanilla drm ueventWuZhen2016-05-121-60/+134
| | | | Change-Id: Ifc83db7bfe6f330ca68e4454516f7a84b2433823
* support hdmi clone more for intel GEN8+ gpusWuZhen2016-05-122-8/+53
| | | | Change-Id: I4e83eb55f2813c92873b8481b8ae9328add4c4f7
* gralloc_drm: use pipe as a fallback instead of defaultMauro Rossi2016-05-021-7/+6
| | | | | | | | Current implementation prefers gralloc_drm_pipe over gralloc_drm_{driver} while it may be the target solution, for nouveau it causes problems because it is not anymore possible to use gralloc_drm_nouveau whenever there is a driver ( e.g. swrast) in BoardConfig.mk BOARD_GPU_DRIVERS not matching the list of valid drivers in drm_gralloc Android.mk
* drm_gralloc: pipe: add error handling for NULL load_pipe_screenRob Herring2016-04-011-9/+13
| | | | | | | Check load_pipe_screen for NULL and return in case dlsym returns a NULL pointer. This can happen if the correct gallium driver is not enabled. Signed-off-by: Rob Herring <robh@kernel.org>
* Fix gralloc_drm_handle_t size mismatch on 32 and 64 bit builds, take 2Rob Herring2016-04-011-1/+1
| | | | | | | | | On x86 there was still a size mismatch due to x86-32 structs not following natural alignment for 64-bit values. Force the alignment of the union. Also, there appeared to be some padding on the end, so move the union to the end of the struct. Signed-off-by: Rob Herring <robh@kernel.org>
* gralloc_drm_pipe: ignore mode_quirk_vmwgfxChih-Wei Huang2016-03-181-8/+2
| | | | Probably we don't need this quirk anymore?
* Revert "Changed format constants to match what vmwgfx driver expects"Chih-Wei Huang2016-03-171-6/+4
| | | | This partially reverts commit dc21193e3252ba2d6526546ba1d0c02116baf3ee.
* Android.mk: remove unneeded static librariesRob Herring2016-03-171-3/+0
| | | | | | With the load_pipe_screen support, these libraries are no longer needed. Signed-off-by: Rob Herring <robh@kernel.org>
* Build pipe support by defaultRob Herring2016-03-171-9/+4
| | | | | | | Build pipe support by default when any unlisted driver is enabled. This way gralloc doesn't have to be updated for every new driver. Signed-off-by: Rob Herring <robh@kernel.org>
* WIP: synchronize gralloc entry pointsRob Clark2016-03-171-15/+55
| | | | | | Note entirely sure what synchronization guarantees android provides before calling in to module.. but stuff inside gralloc module is definately not thread-safe so slap a big lock around the outside.
* Add gralloc_drm_get_prime_fd functionRob Herring2016-03-173-0/+11
| | | | | | Mesa EGL needs to retrieve prime fds from gralloc handles. Signed-off-by: Rob Herring <robh@kernel.org>
* pipe: use gallium loader functionRob Herring2016-03-173-280/+23
| | | | | | | Based on Rob Clark's work to avoid duplicating gallium driver loading code and statically linking each driver's pipe and winsys libraries. Signed-off-by: Rob Herring <robh@kernel.org>
* fix pipe building with upstreamRob Herring2016-03-171-2/+2
| | | | Signed-off-by: Rob Herring <robh@kernel.org>
* add padding to gralloc_drm_handle_tWuZhen2016-01-061-1/+4
| | | | | | | gralloc_drm_handle_t size differs for 32/64 bit process. thus on x86_64, this will crash any 32bit app, add propper padding. Change-Id: I03663b36dd841bf69c84973fd2f5e99741317c15
* Merge remote-tracking branch 'x86/lollipop-x86' into marshmallow-x86Chih-Wei Huang2015-11-141-3/+6
|\
| * fix GRALLOC_DRM_HANDLE_NUM_INTS definition for 64-bitMauro Rossi2015-11-141-3/+6
| | | | | | | | | | | | | | | | | | As highlighted by pstglia current definition of GRALLOC_DRM_HANDLE_NUM_INTS causes issues with 64-bit, because a pointer is accounted as having size of an integer, which is not applicable to 64-bit pointer The same definition adopted in AOSP hardware/drm_gralloc master branch has been used, without changing the structure of gralloc_drm_handle_t.
* | Revert "gralloc: Remove localized hw specific pixel format"Chih-Wei Huang2015-10-102-2/+2
| | | | | | | | This reverts commit 3e00d3255ba6de08761a6b8f40b838680d502d0b.
* | drm_gralloc: Add drm_gralloc actions to gralloc_drm.hSean Paul2015-10-101-0/+8
|/ | | | | Change-Id: If640f6ad7c6b904df59b5c5e164f184f1fc97170 Signed-off-by: Sean Paul <seanpaul@chromium.org>
* WIP: always use gralloc_drm_get_pid()Rob Clark2015-10-061-1/+1
| | | | Rather than accessing gralloc_drm_pid directly.
* android: export drm_gralloc headersMauro Rossi2015-09-181-0/+3
|
* Merge remote-tracking branch 'x86/kitkat-x86' into lollipop-x86Chih-Wei Huang2015-07-271-2/+0
|\
| * gralloc_drm_pipe: remove unnecessary headersChih-Wei Huang2015-06-191-2/+0
| |
* | Correct the casting type for radeon drm callsPaulo Sergio Travaglia2015-07-261-2/+2
| | | | | | | | | | | | | | | | It could be wrong to cast a pointer to a long type (32-bit signed integer) and extend it to a uint64_t (64-bit unsigned integer). Cast it to uintptr_t as the xorg-video-ati does.
* | Merge remote-tracking branch 'x86/kitkat-x86' into lollipop-x86Chih-Wei Huang2015-05-112-1/+59
|\ \ | |/
| * Include newer/missing chipset familiesPaulo Sergio Travaglia2015-05-112-1/+59
| | | | | | | | | | | | | | | | - HAWAII and MULLINS added - changed 0x9649 from CHIP_FAMILY_SUMO to CHIP_FAMILY_SUMO2 - added missing chipInfo for CHIP_FAMILY_VERDE, CHIP_FAMILY_OLAND, CHIP_FAMILY_BONAIRE and CHIP_FAMILY_KAVERI
* | gralloc_drm: make GRALLOC_USAGE_SW_READ_OFTEN be a valid usageChih-Wei Huang2015-04-301-2/+3
| | | | | | | | | | | | | | It allows the camera to get a lock on the buffer. This is required when a picture is taken. Credited to Edgardo Gho <edgardogho@gmail.com>.
* | Merge remote-tracking branch 'x86/kitkat-x86' into lollipop-x86Chih-Wei Huang2015-04-301-20/+1
|\ \ | |/
| * do not use pipe driver for r600gChih-Wei Huang2015-04-291-8/+0
| | | | | | | | Seems it is buggy.
| * remove unnecessary paths in LOCAL_C_INCLUDESChih-Wei Huang2015-04-291-12/+1
| | | | | | | | Now the include paths are imported automatically from libdrm*.
* | Enable radeonsi on Android-x86Paulo Sergio Travaglia2015-04-301-1/+1
| |
* | Merge remote-tracking branch 'x86/kitkat-x86' into lollipop-x86Chih-Wei Huang2015-04-199-128/+454
|\ \ | |/
| * Changed format constants to match what vmwgfx driver expectsPaulo Sergio Travaglia2015-04-192-4/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | When allocating a new buffer using pipe driver, pipe constants are used to define a format based on corresponding HAL_PIXEL_FORMAT However, vmwgfx driver expects other constants (SVGA3D_A8R8G8B8, SVGA3D_X8R8G8B8, SVGA3D_R5G6B5, etc) Changed the returning constants to match what driver expects. v2: [cwhuang] - Use the constants defined in svga3d_types.h. - Use C99 standard to avoid building errors with mesa 10.6.
| * nouveau: undefine SW_INDICATOR_FULLY_DISABLES_TILINGChih-Wei Huang2015-01-291-1/+1
| | | | | | | | Based on the work by pstglia.
| * gralloc_drm_pipe: fix building errors with mesa 10.4Chih-Wei Huang2015-01-222-19/+11
| |
| * add radeonsi to pci_id_driver_map.hChih-Wei Huang2015-01-221-0/+7
| |
| * nouveau: bring some sw_indicator loveEmil Velikov2015-01-221-4/+34
| | | | | | | | | | | | | | | | | | | | Based on some work by pstglia, one needs to switch off (partially or fully?) the tiling when creating the BO. Comment/uncomment the SW_INDICATOR_FULLY_DISABLES_TILING define to toggle between the two. Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>