summaryrefslogtreecommitdiffstats
path: root/sdm
Commit message (Collapse)AuthorAgeFilesLines
* hwc2: Add saturation matrixlinpeter2019-01-256-158/+315
| | | | | | | | | Using transfer matrix to boost adaptive color mode Bug: 121231574 Test: Color pattern check and switch mode test Change-Id: I1761800e3f0dec5a5f2ba961ea2da602c2215210 (cherry picked from commit a654775321b1c56626cc08a37cad9e0609cc1c5d)
* hwc2: Ignore Notify to non-existance file nodeTommy Chiu2018-12-251-0/+4
| | | | | | | | | | | | Secure display will try to signal both wled and labibb. However the display function did not use WLED to drive backlight, nor use LAB/IBB of the PMIC to provide the LCD bias on B4S4. So just bypass the file node writing if it's invalid. Bug: 111504659 Test: Manually tested with keystore_cli_v2 Change-Id: Ia7ea7c13a7635a7f81ae6d76dd1627791c7e49bf
* hwc2: White point compensation for Adaptive color modelinpeter2018-12-206-6/+217
| | | | | | | | | | | | | | | | Adaptive color doesn't have white point calibration. Using transfer matrix to compensate white point. Compensted values are saved in persist/display/calibrated_rgb. Bug: 116282483 Test: switch color mode and check transfer matrix adb shell vndservice call display.qservice 399 i32 0 i32 0 Check dump rgb value adb shell vndservice call display.qservice 21 i32 1 i32 1 i32 2 Change-Id: Iaa7a5f8ccc58be8ecdcfebe814cf7660ef34cc93 Signed-off-by: linpeter <linpeter@google.com>
* Merge remote-tracking branch ↵Adam Shih2018-12-045-4/+16
|\ | | | | | | | | | | | | 'goog/qcom/release/LA.UM.7.8.9.08.00.00.478.137' into pi-dev Bug: 120390132 Change-Id: I5e4edd6e3a484313dd220c48ed5ef2123bc317d4
| * hwc2: Fix compile errors in switch statement.Anjaneya Prasad Musunuri2018-11-261-0/+4
| | | | | | | | | | | | | | | | Fix compile errors because of fall through switch statements. CRs-Fixed: 2352498 Change-Id: I5ec641a2eeee1cfc4bf6750903c6287c5a137d9e
| * sdm: Clear solid fill stages upon flushTharaga Balachandran2018-11-142-0/+7
| | | | | | | | | | | | | | | | | | | | Solid fill stages persist after empty commit because the property DIM_STAGES_V1 is set during solid fill creation but not cleared by HWDeviceDRM::Flush(). This commit clears solid fill stages and updates the property. CRs-Fixed: 2348615 Change-Id: I017b44094286f89c315d16e0de6717f6ffae7b76
| * Merge "hwc2: Check and return valid error from ControlIdlePowerCollapse"Linux Build Service Account2018-11-131-4/+4
| |\
| | * hwc2: Check and return valid error from ControlIdlePowerCollapseRamakant Singh2018-11-051-4/+4
| | | | | | | | | | | | | | | | | | | | | Idle Power collapse is not supported by video mode panel. Change-Id: I5be0f2efcb2f3b22951968fabb74b48d76e6fb22 CRs-Fixed: 2320046
| * | sdm: Update HW layer buffer id from the SDM layerRajavenu Kyatham2018-11-021-0/+1
| |/ | | | | | | | | | | | | - updated buffer id is used for rotator caching decision Change-Id: I6db58a32517436986357dc9e0f6ca068969a1796 CRs-Fixed: 2344358
* | Merge remote-tracking branch ↵ChengYou Ho2018-11-193-7/+10
|\| | | | | | | | | | | | | 'goog/qcom/release/LA.UM.7.8.9.08.00.00.478.122' into pi-dev Bug: 119743470 Change-Id: Ib13d11696dac9c61fbf7c69f2152719e546b277a
| * [REVERT] sdm: Allow poweron before first commit.Pullakavi Srinivas2018-10-251-3/+1
| | | | | | | | | | | | This change reverts I5f726c3b9c106fdc047d67591918d82939e45311 CRs-Fixed: 2305135 Change-Id: Iab7b8691ba97821e2e60f2af6e28ba865b61325a
| * Merge "hwc2: Reset validate on surface damage region update."Linux Build Service Account2018-10-171-1/+1
| |\
| | * hwc2: Reset validate on surface damage region update.Pullakavi Srinivas2018-10-101-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | Video mode panels make use of dirty regions to recompute aggregated fb roi that needs to be cropped from cached buffer. Reset validate to update appropriate rect. CRs-fixed: 2326636 Change-Id: I9e6519ca833f84c36d9fa6fb8270ccf3d1fcf83d
| * | Merge "hwc2: Fix PU disable for multiple Concurrent Writeback frames"Linux Build Service Account2018-10-171-1/+1
| |\ \
| | * | hwc2: Fix PU disable for multiple Concurrent Writeback framesSushil Chauhan2018-10-111-1/+1
| | |/ | | | | | | | | | | | | | | | | | | | | | Fix the disabling of Partial Update feature, in case of Concurrent Writeback for multiple frames. CRs-Fixed: 2318849 Change-Id: Id34eb53b101cf445eacb7d7e9716ae5b8d3a7ac7
| * / sdm: Fix Concurrent Writeback fence handlingPrabhanjan Kandula2018-10-151-5/+8
| |/ | | | | | | | | | | | | Fence get is now part of atomic commit. Update fence get property for Concurrent Writeback before pushing commit. Change-Id: Ib089633408dd801cff8a0b779401521e1c3bbfec
* | Revert "sdm: Allow poweron before first commit."Samuel.lh Huang2018-11-091-4/+1
| | | | | | | | | | | | | | | | | | | | | | | | This reverts commit 2b08f933f8a44e2af39d59cf06da7203ab9eeda0. The poweron function do the kick off the frame buffer. But the DRM plane are empty because boot animation doesn't start yet. Therefore the device will show the black screen. Bug: 118594229 Test: no black screen after revert. Change-Id: If88c04d608007bdc04e9bdf31b3e6f1d94817c07
* | Merge remote-tracking branch ↵ChengYou Ho2018-10-161-1/+4
|\| | | | | | | | | | | | | 'goog/qcom/release/LA.UM.7.8.9.08.00.00.478.098' into pi-dev Bug: 117810284 Change-Id: I01c6d91c6a39ba09d5f4edaf42c16bc5f83cf2ab
| * sdm: Allow poweron before first commit.Pullakavi Srinivas2018-09-251-1/+4
| | | | | | | | | | | | | | | | Set CRTC mode while poweron, so that poweron and poweroff can be performed without commit in between. CRs-Fixed: 2305135 Change-Id: I5f726c3b9c106fdc047d67591918d82939e45311
* | Merge remote-tracking branch ↵Cyan_Hsieh2018-09-2522-35/+403
|\| | | | | | | | | | | | | 'goog/qcom/release/LA.UM.7.8.9.08.00.00.478.078' into pi-dev Bug: 116473492 Change-Id: I57b01175e9300fad89683201bfb7dd271c58fd05
| * Merge "hwc2: Fix GetDozeSupport api to handle failure case"Linux Build Service Account2018-09-092-7/+14
| |\
| | * hwc2: Fix GetDozeSupport api to handle failure caseGurpreet Singh Dhami2018-08-232-7/+14
| | | | | | | | | | | | | | | | | | | | | Fix for GetDozeSupport api to return BAD_DISPLAY error, if that display doesn't exist. Change-Id: I53c61d0d0adc3850ef9e82da189c4d8141466c93
| * | hwc2: Fix error handling for invalid ColorMode and RenderIntentGurpreet Singh Dhami2018-09-041-0/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | Fix to return BAD_PARAMETER for invalid RenderIntent value in GetRenderIntents api and invalid ColorMode in SetColorModeWithRenderIntent api. Currently they return Unsupported error. Change-Id: Ic56322f56170d2e421161c69ebf4900f86ec554c
| * | Merge "hwc2: Wait for retire fence after populating it from layer stack"Linux Build Service Account2018-08-301-1/+6
| |\ \
| | * | hwc2: Wait for retire fence after populating it from layer stackRamkumar Radhakrishnan2018-08-171-1/+6
| | |/ | | | | | | | | | | | | | | | | | | | | | Wait for retire fence and notify pmic about secure display end after populating the retire fence from layer stack. Change-Id: I61661d2b4939bbd184aeb8dac310ff156026800c CRs-Fixed: 2298981
| * / sdm: Define client interface to enable/disable idle pcRamkumar Radhakrishnan2018-08-2122-27/+263
| |/ | | | | | | | | | | | | | | | | | | | | | | | | 1. Define client interface to enable/disable idle power collapse 2. Maintain refcount to handle concurrent enable/disable 3. Trigger the screen refresh and wait for the next frame commit done event before returning the control to the client on disable idle pc 4. Enable idle pc on suspend and reset the refcount 5. Add binder support to enable/disable idle pc CRs-Fixed: 2255316 Change-Id:Ibcaf9d4edca502cc91e9b201be822bd48313a635
| * sdm: Notify secure display start/end to PMIC driverRamkumar Radhakrishnan2018-08-092-1/+96
| | | | | | | | | | | | | | | | | | | | Notify secure display start/end to PMIC driver to disable PMIC irqs before the start of secure display session and enable PMIC irqs after the end of secure display session to avoid register access by PMIC drivers during secure display session. Change-Id: I6c71b7f0cb8c78be4144bd785b752c4e1ad0c86b CRs-Fixed: 2285937
* | hwc2: Ignore non integral source crops for solid color layersMidas Chien2018-09-051-1/+2
|/ | | | | | | | | | | The solid color source rectangle isn't honored by SDM anyway, ignore it to workaround. Bug:112620861 Test: ColorLayer can be composed by Device even with non integral source crops Change-Id: Ic74e217baaf73e6ebf9ce0d48f93dcb8bfa2b29d
* hwc2: Fix GetReadbackBufferAttributes apiSushil Chauhan2018-07-312-3/+29
| | | | | | | | Fix GetReadbackBufferAttributes api to get the Dataspace from current ColorMode of HWC Display. CRs-Fixed: 2288505 Change-Id: I75325d21b01e517236eb3c8124c706952103ae9b
* Merge "hwc: Use HWC for videos with non-integral source crop"Linux Build Service Account2018-07-281-1/+3
|\
| * hwc: Use HWC for videos with non-integral source cropNaseer Ahmed2018-07-181-1/+3
| | | | | | | | | | Change-Id: I2b2f525278219f24f48894d04db891871215a017 CRs-Fixed: 2281150
* | Merge "sdm: Allow Skip Validate when Partial Update is disabled"Linux Build Service Account2018-07-186-21/+46
|\ \
| * | sdm: Allow Skip Validate when Partial Update is disabledSushil Chauhan2018-07-176-21/+46
| |/ | | | | | | | | | | | | | | Allow Skip Validate, if the Partial Update feature is not enabled on Display and surface is getting updated on the updating layer(s). CRs-Fixed: 2280394 Change-Id: I0524818fb4cbb11dcc8fc993fa7ae536bd555709
* / hwc2: Update FBT frame rate.Pullakavi Srinivas2018-07-161-0/+3
|/ | | | | | | | | -- FBT refresh rate updates on SetClientTarget call. -- Set it upfront(prepare) so that SDM doesn't misinterpret its frame rate. CRs-Fixed: 2274297 Change-Id: Ia45fb173522b2fb392d02b74c535cc216dba5e8c
* sdm: Update debug property to treble compliance.Tharaga Balachandran2018-07-051-1/+1
| | | | | | | | Update vsync skew debug property to treble compliance. Vendor prefix should be present in all vendor properties. Change-Id: I322a2ef5e9d392f59e73c80f2adafa8346e0be7e CRs-Fixed: 2265938
* hwc2: Skip SDM prepare for consecutive GPU composed framesSushil Chauhan2018-07-032-0/+40
| | | | | | | | | | | Skip SDM prepare, if all the layers in the current draw cycle are marked as Skip and previous draw cycle had GPU Composition, since the resources for GPU Target layer, have already been validated and configured to driver. It avoids unnecessary CPU cycles in Strategy, Resource Manager, DAL, etc. in SDM and the Validate IOCTL in driver. CRs-Fixed: 2269870 Change-Id: I019a13d0eaf08c01d710555ff99f104d2a3d24ce
* sdm: drm: Remove special handling for first_cycleNaseer Ahmed2018-06-261-6/+0
| | | | Change-Id: Id9dab7861e41b9b69270235d0cff4db469029684
* hwc: Fix HWC Writeback wrong resolution shown with dumped framesGurpreet Singh Dhami2018-06-131-1/+2
| | | | Change-Id: Ia4c680381801f14305515bb2e6b75cd49794743d
* hwc: Fix HWC Writeback one less frame dumpGurpreet Singh Dhami2018-06-121-6/+10
| | | | | | Fix HWC Writeback failure to capture first frame if directory does not exist. Change-Id: I051238462cd3f12c1bda5345201f8cf0c7ed4701
* hwc2: Add support for optional user preferencesSushil Chauhan2018-06-077-14/+42
| | | | | | | | Add support for optional user preferences of output buffer format and post-processed flag, in the HWC binder command to dump frame output. CRs-Fixed: 2254012 Change-Id: I33298fab5122c1fa8e687b72dfa429e287261afb
* Merge "sdm: Reduce the fb_id cache limit for UI layers" into display.lnx.4.9Linux Build Service Account2018-06-062-15/+30
|\
| * sdm: Reduce the fb_id cache limit for UI layersSushil Chauhan2018-06-042-15/+30
| | | | | | | | | | | | | | | | | | Reduce the fb_id cache limit for UI layers to 3 and for Rotator use cases to 2. This is required to reduce orphaned allocations in pre-rotation use cases. CRs-Fixed: 2253973 Change-Id: I3989c0179d5a97b596605dd5405b89abbba5c509
* | hwc: Fix typoNaseer Ahmed2018-06-041-1/+1
|/ | | | Change-Id: Ia5166410bb7a08a6cf0d8e8c07973ef34342b980
* Merge "sdm: Clear fb_id map if it exceeds the size limit" into display.lnx.4.9Linux Build Service Account2018-05-242-4/+6
|\
| * sdm: Clear fb_id map if it exceeds the size limitSushil Chauhan2018-05-242-4/+6
| | | | | | | | | | | | | | | | | | | | Clear the fb_id map of SDM HW layer, if it exceeds the size limit. It is required for cases like pre-rotation where new buffer handles are allocated and new unique handle ids are used on same HWC layer for consecutive draw cycles involving device rotations. CRs-Fixed: 2247747 Change-Id: Iec71c42572f589717ad8f5677f84ff16e0aaaf01
* | hwc: Do not reapply mode if already setNaseer Ahmed2018-05-241-0/+4
|/ | | | | Change-Id: Ica6ad6941dc04a190b0088581a2d6293c2024202 CRs-Fixed: 2248553
* hwc: Use default blend space as sRGBNaseer Ahmed2018-05-231-15/+9
| | | | | | | | | When in saturated mode, default mode should be assumed as sRGB and the remaining non sRGB layers should fall back to GPU. CRs-Fixed: 2247695 Bug: 77939611 Change-Id: I58d42709eaffbdf79ba9103ec61bc690afb0b96a
* hwc: Mark only the scRGB layer as skipNaseer Ahmed2018-05-231-4/+2
| | | | | | | | | | Since the color mode is now being used as the working color space, scRGB layers can be marked as skip and the entire composition no longer needs to fall back to GPU. CRs-Fixed: 2247597 Bug: 80082357 Change-Id: Ic32ab3495037a0dcb2f2f3b427bd55e37e17d0cd
* Merge "hwc: tonemapper: Update handle ID for inverse tonemapper"Linux Build Service Account2018-05-171-0/+1
|\
| * hwc: tonemapper: Update handle ID for inverse tonemapperNaseer Ahmed2018-05-171-0/+1
| | | | | | | | | | | | | | | | The recent fb_id optimization requires that the handle ID be updated from the gralloc buffer ID Change-Id: I456dc58b87dcfeed61baf082930a16785cd9545a CRs-Fixed: 2244112
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-23 09:15:47 +0000