This repository has been archived by the owner on Jan 7, 2023. It is now read-only.
Debugging #95
Open
Debugging #95
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Set the ro.hardware.gralloc property to 'minigbm' to have the new module name be loaded preferentially. Bug: 77276633 Change-Id: Ib247ee08486887bd76b86ed9c8b7d62c58a31af7 Signed-off-by: Alistair Strachan <[email protected]>
am: 3fbaca6 Change-Id: I884b6bd6e0658b9b931b53d42bcee1ecbc16fc95
am: ff234ae Change-Id: I3a6a0db16f4bd7048aa9b985fdd7444a8fda5cc7
Dri backend is not used for the all use flags. When same fd is
shared between dri backend and the minigbm drv, they end up using the
same kernel DRI context.
BUG=b:111081134
TEST=cheets_SurfaceComposition
graphics_Gbm
graphics_Gralloc
Change-Id: I47ae12391acfc9c947e8281de11472fcdf5dec5b
Signed-off-by: Satyajit Sahu <[email protected]>
Reviewed-on: https://chromium-review.googlesource.com/1164972
Tested-by: Deepak Sharma <[email protected]>
Reviewed-by: Tomasz Figa <[email protected]>
Reviewed-by: Gurchetan Singh <[email protected]>
This patch simply changes some methods' names in order to align them with the upstream ones. It's needed because Chromium/Wayland implementation is going to have a gbm implementation used by a GPU split effort. In order to avoid to have a per compile gn flags, we would like to standartize the gbm APIs and be able to use the same binary without recompilations. In follow-up CLs, I will fix callers and finally remove the old APIs namings in favor of new ones. Change-Id: I850824e7194077c47baa761cddb2947959dff9c3 Reviewed-on: https://chromium-review.googlesource.com/1162170 Commit-Ready: Michael Spang <[email protected]> Tested-by: Michael Spang <[email protected]> Reviewed-by: Kristian H. Kristensen <[email protected]>
Adding missing plane handle API renaming. Change-Id: I33c4b4bcc5097ca568c95eb8c3718c3f790fd996 Reviewed-on: https://chromium-review.googlesource.com/1170773 Commit-Ready: Maksim Sisov <[email protected]> Tested-by: Maksim Sisov <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
When the format is NV12, additional buffer is required for the UV plane. Though size adjustment for this requirement is currently implemented only in the msm driver, it is not specific to this device. This commit moves the adjustment to helper function 'drv_dumb_bo_create', which is called from each driver. BUG=chromium:852302 TEST=video playback on msm Change-Id: I690f79ccc6a2c99e680b0d293ffa6002f74ae6bf Signed-off-by: Keiichi Watanabe <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1163582 Reviewed-by: Tomasz Figa <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
To use vivid on VM as an external camera, virtio_gpu has to support NV12. BUG=chromium:852302 TEST=Use camera app on betty by using vivid as an external camera Change-Id: I84ddcbae473a06337918328afe1db8d32003ad84 Signed-off-by: Keiichi Watanabe <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1160434 Reviewed-by: Tomasz Figa <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Though presubmit.sh requires formatting by clang-format, some files are not formatted. BUG=None TEST=compile Change-Id: I828dd85536cb94dc98cd11e83ec86af53a425070 Signed-off-by: Keiichi Watanabe <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1172241 Reviewed-by: Gurchetan Singh <[email protected]>
Because HAL_PIXEL_FORMAT_YV12 requires that height is not aligned, adjustments are needed when the format is DRM_FORMAT_YVU420_ANDROID. This adjustment was done in each driver, but it can be moved to the helper function. BUG=None TEST=VDA test in import mode on Cheza Change-Id: I23a8a6e674664e80c30ea66b75d79e10edc28d5d Signed-off-by: Keiichi Watanabe <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1172262 Reviewed-by: Tomasz Figa <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
…USE_PROTECTED Originally, Android's HAL_PIXEL_FORMAT_BLOB always regards tilable. It is not true if it is allocated with BO_USE_PROTECTED. This CL enables to allocated protected non-tiled Android's HAL_PIXEL_FORMAT_BLOB. BUG=b:112565837 TEST=Play protected videos with ARC++ at eve Change-Id: I0e45f658f170cd380a74b9207ba57b08b6edab8b Reviewed-on: https://chromium-review.googlesource.com/1175641 Commit-Ready: Hirokazu Honda <[email protected]> Tested-by: Hirokazu Honda <[email protected]> Reviewed-by: Tomasz Figa <[email protected]>
Since we should be using llvmpipe with MSM for now, add formats we know work well with vgem (which also uses llvmpipe). BUG=none TEST=none Change-Id: I88bf183a23a593029ed303c16ef960eb36d59f9f Reviewed-on: https://chromium-review.googlesource.com/1194189 Commit-Ready: Douglas Anderson <[email protected]> Tested-by: Douglas Anderson <[email protected]> Tested-by: Gurchetan Singh <[email protected]> Reviewed-by: Douglas Anderson <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Without doing this it's possible for the CPU to access the memory when it's in use by another device. TEST=adb shell "am instrument -w --abi armeabi-v7a -e class\ android.media.cts.EncodeDecodeTest#testEncodeDecodeVideoFromPersistentSurfaceToSurfaceQCIF\ android.media.cts/android.support.test.runner.AndroidJUnitRunner" BUG=b:71835379 Change-Id: I3b1508a0eab3b020b7c42978cb1e1099ebc029fd Signed-off-by: Luigi Santivetti <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1193302 Commit-Ready: Gurchetan Singh <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Before this change, a NULL pointer could have been passed to munmap(), making impossible for it to unmap previoulsy mapped memory. After this change, munmap() always receives a valid pointer. BUG=b:71835379 b:114699642 TEST=adb shell "am instrument -w --abi armeabi-v7a -e class\ android.uirendering.cts.testclasses.InfrastructureTests#testScreenshot\ android.uirendering.cts/android.support.test.runner.AndroidJUnitRunner" Change-Id: Ic1b15807623209ab28d0d4ed63fe0d6ef2dcc6f8 Signed-off-by: Luigi Santivetti <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1221667 Commit-Ready: Kazuhiro Inaba <[email protected]> Tested-by: Pin-chih Lin <[email protected]> Reviewed-by: Stéphane Marchesin <[email protected]> Reviewed-by: Daniel Kurtz <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Change-Id: I9b5c3db33bd0c7914a214e2181d655aec84632c5 Reviewed-on: https://chromium-review.googlesource.com/1229436 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Tested-by: Gurchetan Singh <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Change-Id: Ie5f8b148e3f051edde08b0b46ab22a80f034ff80 Reviewed-on: https://chromium-review.googlesource.com/1229437 Commit-Ready: Gurchetan Singh <[email protected]> Tested-by: Gurchetan Singh <[email protected]> Reviewed-by: Stéphane Marchesin <[email protected]>
We defined the function, but it's since gone unused. Change-Id: I23e7e57413c4a7c82d9da302b79e6b9c5068654d Reviewed-on: https://chromium-review.googlesource.com/1229438 Commit-Ready: Gurchetan Singh <[email protected]> Tested-by: Gurchetan Singh <[email protected]> Reviewed-by: Stéphane Marchesin <[email protected]>
Mesa drivers can't use it, in general.
I'm leaving in Rockchip since I haven't gotten any reports
about this test failing on kevin-arcnext.
Fixes: abe44f ("minigbm: add support for BG24")
BUG=b:77876551, b:115564746
TEST=The following tests should pass on Eve/Grunt arc-next:
android.hardware.nativehardware.cts.AHardwareBufferNativeTests#SingleLayer_ColorTest_GpuColorOutputAndSampledImage_R8G8B8_UNORM
android.hardware.nativehardware.cts.AHardwareBufferNativeTests#SingleLayer_ColorTest_GpuColorOutputCpuRead_R8G8B8_UNORM
android.hardware.nativehardware.cts.AHardwareBufferNativeTests#SingleLayer_ColorTest_GpuColorOutputIsRenderable_R8G8B8_UNORM
android.hardware.nativehardware.cts.AHardwareBufferNativeTests#SingleLayer_ColorTest_GpuSampledImageCanBeSampled_R8G8B8_UNORM
Change-Id: Ic7aec07c89fdc21e0c8392238e833f7980062049
Reviewed-on: https://chromium-review.googlesource.com/1229439
Commit-Ready: Gurchetan Singh <[email protected]>
Tested-by: Gurchetan Singh <[email protected]>
Reviewed-by: Stéphane Marchesin <[email protected]>
DRM_IOCTL_MSM_GEM_NEW is used to create buffer and DRM_IOCTL_MSM_GEM_INFO is used to map buffer. BUG=none TEST=mmap_test, plane_test Change-Id: I7db0cf36b03625f02828b63946550c768c422419 Signed-off-by: Tanmay Shah <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1134481 Reviewed-by: Kristian H. Kristensen <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
map stride is bigger than the allocation stride. Currently gralloc_lock does not consider map_stride. Align the width so that stride get alinged to 256. BUG=b:115946221 TEST= CtsNativeHardwareTestCases GpuColorOutputCpuRead* tests Change-Id: I0e1ccfba4ec981702498a76fa5a0b2c662b6e728 Signed-off-by: Satyajit Sahu <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1242767 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Tested-by: Deepak Sharma <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
This patch does not bring any functionality changes, but just adds a MINIGBM define in order to be able to distiguish if Chromium, for example, uses a system libgbm (needed by Ozone/Wayland builds for Linux due to some diffirences, which we cannot fix yet) or minigbm. Change-Id: I82366ce9a1c302124b75a6c21028a67127497065 Reviewed-on: https://chromium-review.googlesource.com/1281887 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Tested-by: Maksim Sisov <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
Some camera use cases, such as camera output buffers shared between video encoder and hwcomposer, require CPU access to the buffers (the ARC++ encoder stack is not zero-copy yet) and having uncached mapping slows them down significantly, to the point that respective CTS tests start failing. Fix it by excluding buffers with BO_CAMERA_READ and BO_CAMERA_WRITE from the WC mapping condition. BUG=b:117978452 TEST=android.hardware.camera2.cts.RecordingTest#testVideoPreviewSurfaceSharing on Nocturne Change-Id: Icc3c7104029403fbf64a7fcc8476d52a3f9ea3f0 Reviewed-on: https://chromium-review.googlesource.com/1309914 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Tested-by: Tomasz Figa <[email protected]> Reviewed-by: Tomasz Figa <[email protected]>
We only need this workaround for DRI Android buffers, not all buffers. Also, we don't need this for buffers with non-SW uses. BUG=b:117942643 TEST=compile Change-Id: I7ea24b0dc3b040b726f6c1223c96998f078a20fc Signed-off-by: Satyajit Sahu <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1300595 Reviewed-by: Robert Tarasov <[email protected]>
some of the android.media.cts.EncodeDecodeTest in media test are failing when enabled. Bug=b:115585732 TEST=run android.media.cts.EncodeDecodeTest tests Change-Id: Id54236a4441e6441bd8500a81f83c00e2fa405d8 Signed-off-by: Deepak Sharma <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1329993 Commit-Ready: Deepak Sharma <[email protected]> Tested-by: Deepak Sharma <[email protected]> Reviewed-by: Bas Nieuwenhuizen <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
For linear path we don't wait for GPU to finish before CPU access which may case issues in gralloc lock() case. Bug=b:115585732 TEST=android.video.cts.VideoEncoderDecoderTest#testAvcGoog0Qual1920x1080 Change-Id: If827fe4a18726797082046b4f24fc8c75ff1bcb7 Signed-off-by: Deepak Sharma <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1340484 Commit-Ready: Deepak Sharma <[email protected]> Tested-by: Deepak Sharma <[email protected]> Reviewed-by: Deepak Sharma <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
This CL aligns the minigbm more with the upstream GBM by defining GBM_BO_IMPORT_FD_MODIFIER and using gbm_import_fd_modifier_data instead. That is, the main difference between the old gbm_import_fd_planar_data one and the new one is the format_modifiers variable. In the upstream GBM, it's a single variable. In the minigbm, it is an array. As we know there are no cases when modifiers would be different for each plane. Thus, it's safe to eliminate that and adapt more to the upstream. Change-Id: Iaae062ef1fe9fc9ab0ead09c5f4bfa91d2db67c3 Reviewed-on: https://chromium-review.googlesource.com/1360771 Commit-Ready: Maksim Sisov <[email protected]> Tested-by: Maksim Sisov <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
This reverts commit bc667c3. Reason for revert: This change might cause crbug.com/913329. Original change's description: > minigbm: define GBM_BO_IMPORT_FD_MODIFIER > > This CL aligns the minigbm more with the upstream GBM > by defining GBM_BO_IMPORT_FD_MODIFIER and using > gbm_import_fd_modifier_data instead. > > That is, the main difference between the old > gbm_import_fd_planar_data one and the new one is > the format_modifiers variable. In the upstream > GBM, it's a single variable. In the minigbm, it is > an array. > > As we know there are no cases when modifiers would > be different for each plane. Thus, it's safe to eliminate > that and adapt more to the upstream. > > Change-Id: Iaae062ef1fe9fc9ab0ead09c5f4bfa91d2db67c3 > Reviewed-on: https://chromium-review.googlesource.com/1360771 > Commit-Ready: Maksim Sisov <[email protected]> > Tested-by: Maksim Sisov <[email protected]> > Reviewed-by: Gurchetan Singh <[email protected]> Change-Id: I7dec62abd4243554847930cce33f0d5db069c3f4 Reviewed-on: https://chromium-review.googlesource.com/1369469 Commit-Ready: Naoki Fukino <[email protected]> Tested-by: Naoki Fukino <[email protected]> Reviewed-by: Naoki Fukino <[email protected]>
Venus driver requires extra padding for NV12 buffers. BUG=b:120118851 TEST=plane_test -f NV12 TEST=plane_test -f NV12 -z 640x480 Change-Id: If86c548350278cc8d87159b58e56802b2598d448 Signed-off-by: Tanmay Shah <[email protected]> Reviewed-on: https://chromium-review.googlesource.com/1362202 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
No functional changs. BUG=none TEST=compile Change-Id: I6b302e9bbf9d46489bfaa03d3b5cf2b070cb1d24 Reviewed-on: https://chromium-review.googlesource.com/1356697 Commit-Ready: ChromeOS CL Exonerator Bot <[email protected]> Tested-by: Stéphane Marchesin <[email protected]> Reviewed-by: Stéphane Marchesin <[email protected]> Reviewed-by: Gurchetan Singh <[email protected]>
This format is used in VTS tests VTS use it as default color format
This is to help fix the building issues in IA HWC
SetInterface is used by MSDK for the interalce info of the buffer SetProtectionInfo is used by MSDK for protection info for the buffer union type added for protection info
Otherwise it will cause sramble sceen issues in Android
lockFlex fills outFlexLayout->planes with pointer to global variable ycbcrplanes, doesn't work from multiple threads. Workaround: change the variable to thread_local.
We allocate buffer by using allocate4, but lockFlex is called by gralloc1. Bufore we use buffer, we need retain/release.
Add NV12_Y_TILED_INTEL video format support
We use one wrong align heigh for NV12_Y_TILED_INTEL video format. Fix this issue with this changes.
The DRM_FORMAT_ABGR2101010 format will be used by Android Media framework for software codec of AV1Hdr, and the usage flag request as: BO_USE_SW_READ_RARELY BO_USE_SW_READ_OFTEN BO_USE_SW_WRITE_RARELY BO_USE_SW_WRITE_OFTEN Tracked-On: OAM-95578 Signed-off-by: Yang, Dong <[email protected]>
Add YUYV format for 264 video which is used by HWC. Align 263 video with 64, if we use 32, we may encounter VA create surface failures in HWC. Tracked-On: OAM-95589 Signed-off-by: Ren Chenglei <[email protected]>
Fix PlaneLayout width/height metadata by using the horizontal/vertical subsampling of each plane. TEST=vts -m VtsHalGraphicsMapperV4_0TargetTest Tracked-On: OAM-95590 Signed-off-by: ason Macnak <[email protected]>
Camera preview use YCbCr, and will be converted to resolved format NV12. NV12 is one private defined format. So the descriptor.format is not same with crosHandle->droid_format. Suppress the check here. Camera works as one buffer producer for media. In previous, we may allocate & lock buffer with one same "driver". But when we switch to allocator & mapper 4.0, some components still use allocate & mapper 2.0, and which will use two different drivers. So, we need retain before we use the buffer allocated by other driver. In future, those components should switch to allocator & mapper 4.0 which is asked by API 30. Tracked-On: OAM-95595 Signed-off-by: Ren Chenglei <[email protected]>
With allocator & mapper 4.0, there is some regression in CTS CtsNativeHardwareTestCases. Compared with allocator & mapper 2.0, we still not support two mipmap usage, and also not support to allocate more than 1 layer. Tracked-On: OAM-95612 Signed-off-by: Ren Chenglei [email protected]
For RGBA_FP16 format, we still not support RENDERING and TEXTURE usage. Tracked-On: OAM-95612 Signed-off-by: Ren Chenglei [email protected]
For DRM_FORMAT_R8, we don't need align the stride. This change could help fix allocator issues in CtsNNAPITestCases. Tracked-On: OAM-95724 Signed-off-by: Ren Chenglei [email protected]
Add BO_USE_SW_MASK usage to DRM_FORMAT_RGB888 format, this is required by Android R CTS. Test: android.hardware.cts.HardwareBufferTest#testCreate Tracked-On: OAM-95726 Signed-off-by: Ren Chenglei [email protected]
After we rebased our allocator & mapper from 2.0 to 4.0, we use allocator 4.0 to allocate buffer, which can't be accessed by mapper 2.0 directly. Here, we add freeBuffer function, if we want to use mapper 2.0 to lock & unlock buffer(allocated by allocator 4.0), we need importBuffer & freeBuffer before/after lock and unlock. Tracked-On: OAM-95916 Signed-off-by: Ren Chenglei <[email protected]>
Issue has been fixed from media driver, revert back to 32, which can help pass VTS failure Tracked-On: OAM-96724 Signed-off-by: Ren Chenglei <[email protected]>
On virgl backend, we use vhal camera. During camera preview, we use RGBA format, so add camera usage for it. Tracked-On: OAM-96725 Signed-off-by: Ren Chenglei [email protected]
This is needed for codec2.0 Tracked-On: OAM-96726 Signed-off-by: Yang, Dong <[email protected]> Signed-off-by: Ren Chenglei <[email protected]>
1. native_handle_* need native_handle_t instead of buffer_handle_t, this conflicts with mapper 2.0 release function 2. importBuffer call native_handle_clone, which will have "malloc", then we need use "free" by native_handle_delete in case of memory leak, which is called in private defined API freeBuffer 3. We should use Android official define API release instead of private defined API freeBuffer; so we need remove native_handle_clone in importBuffer and use release directly Tracked-On: OAM-96807 Signed-off-by: Chenglei Ren <[email protected]>
Inside the release function (CrosGralloc1::release()) native_handle_close() and native_handle_delete() functions are getting called after driver release function call (driver->release()). In driver release function these are already cleaned up. Inside driver release function native_handle_close() is called inside destructor of cros_gralloc_buffer. Tracked-On: OAM-97019 Signed-off-by: Kaushlendra Kumar <[email protected]> Signed-off-by: Chenglei Ren <[email protected]>
getBackingStore() should return a globally unique value for a dedicated graphics buffer, the graphics buffers will share between different modules like display and codec, the unique value will be used to identify specifi memory. Tracked-On: OAM-97062 Signed-off-by: Yang, Dong <[email protected]>
Signed-off-by: Shaofeng Tang <[email protected]>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.