hardware-intel-common-vaapi

Fork

test: fix AVCEContextTest/RateControl regression

Update AVCEContextTest/RateControl expectations that changed
since PR #197

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

test: add status expectation in I965TestFixture::createConfig

Add VAStatus parameter to I965TestFixture::createConfig to allow
users to pass in the expected status for testing.

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

avoid NULL pointer deref in i965_CreateConfig

The attrib_found pointer variable might be NULL in many cases.
Thus, check it before attempting to dereference it.

Fixes #246

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

add brc mode check for AVC when creating the config

v0:this add the RC check for AVC encoder

Fixes #35

Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>

VAEncROI on GEN7 chipset fix

Copying line intel_h264_enc_roi_config() from gen75_vme.c to gen7_vme.c

Add the configure attributes of EncMaxRefs for GEN9 HEVC encoder

Fixes #200

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Do not apply AUD hack to non-RawData headers

Headers which are not VAEncPackedHeaderRawData should not be inserted
at the front of the access unit by the AUD hack - if they are, they will
end up duplicated if they match the test because they are also placed
later in the access unit.

Signed-off-by: Mark Thompson <sw@jkqxz.net>

Update new media kernels for HEVC encoder

Fixes #240

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Correct the data type

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>

Don't align the bitrate to 1000 for VDEnc AVC encoding

The HuC accepts the real bitrate setting

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>

Initialize the HME mv cost table for VDEnc AVC encoding

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>

add VC-1 intensity compensation for SNB

implement intensity compensation for VC-1 decoding

Intensity compensation was not present for B-frames, but only for P-frames. When a P-frame flags intensity compensation for its forward reference frame, all subsequent B-frames that use this reference frame as well, need to do intensity compensation as well.

test: fix GCC 7.1.1 warnings/errors

Fixes #236: add explicit braces to avoid 'ambiguous else' warning.

Signed-off-by: Kelly Ledford <kelly.ledford@intel.com>

Encoder: Avoid memory leak

An internal surface is created when the input surface doesn't meet the
HW requirement for encoding, however this surface is not destroyed when
the corresponding context is destroyed which causes the memory leak
issue. This patch fixes this issue.

This fixes https://github.com/01org/libyami/issues/751

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>

Remove useless frame dimension check for VP9

Signed-off-by: Mark Thompson <sw@jkqxz.net>

Don't check the stride in the y direction for a single plane surface

obj_surface->height is used to calculate the offset for U/V plane. for a
surface with single plane, the check is unnecessary

This fixes https://github.com/01org/intel-vaapi-driver/issues/222

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Daniel Charles <daniel.charles@intel.com>
Tested-by: Daniel Charles <daniel.charles@intel.com>

Add HEVC ROI attributes for VAConfigAttribEncROI

Fixes #179

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Add the support of ROI for HEVC encoder on GEN9 SKL+

Fixes #179

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Parse ROI Misc parameters for CQP mode

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Update media kernels for hevc encoder

Signed-off-by: peng.chen <peng.c.chen@intel.com>

Fix VDENC_PIPE_BUF_ADDR_STATE for AVC VDEnc encoding

DW22-DW27 are the first 2 reference surfaces in list0, not the down
scaling referece surfaces.

This fixes https://github.com/01org/intel-vaapi-driver/issues/224

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>

Enc: add max ref number query for AVC encoder on SKL+

Fixes #199

Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>

Remove generated file from tree and on clean

Signed-off-by: Philipp Kerling <pkerling@casix.org>

wayland: Also support wl_drm version 1

Just using version 2 without checking would lead to a protocol error
bringing down the entire application, and wl_drm version 1 is still
supported since v2 only adds PRIME capabilities.

Fix the ROI bottom setting issue in vdenc

Signed-off-by: peng.chen <peng.c.chen@intel.com>

wayland: Use private event queue for compositor communication

(Ab)using the default queue that the application might itself use
already and work on in parallel to initializing libva is not
thread-safe. Make it thread-safe by setting a private queue on a
wrapped wl_display. Also print some more error messages in case things
go wrong.

Signed-off-by: Philipp Kerling <pkerling@casix.org>
Signed-off-by: Olivier Crete <olivier.crete@collabora.com>

wayland: Bump wayland-client dependency to 1.11.0

Signed-off-by: Philipp Kerling <pkerling@casix.org>

wayland: Cleanup wl_registry on terminate

Signed-off-by: Philipp Kerling <pkerling@casix.org>
Signed-off-by: Olivier Crete <olivier.crete@collabora.com>

Bump intel-vaapi-driver to 1.8.4.pre1 for development

Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>