OSDN Git Service
peng.chen [Thu, 13 Apr 2017 05:23:17 +0000 (13:23 +0800)]
Improve the HEVC encoding quality on SKL/APL/KBL
V1:
1, Update the commit date
2, Add extern to a function prototype
3, Reallocate resource for different resolution pictures
4, Move static constant array ftq_25i[]
5, Free res_pak_slice_batch_buffer before flush the batch buffer
6, Add cache attribute setting
7, Fix write domain issue of output surface
8, Remove the wrong bit setting in pipe_select
9, Correct the licensing declaration
10, Move hevc enc apis declaration to the enc common api file
11, Add two field bits in PAK pic_state
V2:
1, Move QM/FM default tables out of the function
2, Set the private_data for obj_surface before allocating the surface
3, Move hevc_qpc_table[] into gen9_hevc_encoder.c
4, Alocate independent ME gpe context for each HMEs
5, Reconfigure the VFE state
6, Remove the i965_zero_gpe_resource in ALLOC_GPE_RESOURCE
7, Unify the coding style
V3:
1, Fix memory clear size mistake in intra_distortion
2, Disable multi-slices support for HEVC encoding
Signed-off-by: peng.chen <peng.c.chen@intel.com>
peng.chen [Thu, 13 Apr 2017 05:23:17 +0000 (13:23 +0800)]
Add the new kernel binary for the improved HEVC encoder on SKL/APL/KBL
V1:
Move kernel binary into new file
V2:
Unify the coding style
Signed-off-by: peng.chen <peng.c.chen@intel.com>
Wang, Tiantian [Mon, 10 Apr 2017 06:12:24 +0000 (02:12 -0400)]
Check and allocate (if not exist) bo for dst_obj_surface in vpp
This fixes #20
https://github.com/01org/intel-vaapi-driver/issues/20
Signed-off-by: Wang Tiantian <tiantian.wang@intel.com>
Pengfei Qu [Wed, 5 Apr 2017 02:59:48 +0000 (10:59 +0800)]
ENC: add AUD nal unit at the beginning of picture
v0:parse nal type with generic method
Fixes #110
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Víctor Manuel Jáquez Leal [Tue, 11 Apr 2017 17:43:38 +0000 (19:43 +0200)]
dri: error for more unimplemented surface formats
Such as commit
396bf01a, RGB formats are not display correctly
in X11. Hence, return unimplemented error.
Fixes #131
Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
Víctor Manuel Jáquez Leal [Tue, 11 Apr 2017 12:03:34 +0000 (14:03 +0200)]
Fix: check height alignment against hpitch
Commit
43a75ea introduced a regression when importing dmabuf from v4l2,
returning VA_STATUS_ERROR_INVALID_PARAMETER.
This is because, when checking the check height alignment, the wpitch
was used, when hpitch is the one to compare.
Fixes #129
Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
Xiang, Haihao [Fri, 24 Mar 2017 01:13:24 +0000 (09:13 +0800)]
Add pre-commit hook
Check that the code follows a consistant coding style before committing.
The current command will be interrupted if the commit doesn't follow the
code style
This fixes https://github.com/01org/intel-vaapi-driver/issues/99
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 23 Mar 2017 07:31:47 +0000 (15:31 +0800)]
Unify the coding style in the driver
Linux coding style is used in the driver source code. Use the command
below to format/indent .c/.h files
$> astyle --style=linux -cnpUH -s4 -M120 <file>
A script of style_unify is added in the top-level directory to
handle all .c/.h files in the driver. There is no change to any
functionality
This fixes https://github.com/01org/intel-vaapi-driver/issues/99
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Sreerenj Balachandran [Thu, 6 Apr 2017 19:05:38 +0000 (12:05 -0700)]
gen9_avc_encoder: macro optimization
Avoid duplicated multiplication, do the frame size
calculation once and reuse the result.
Signed-off-by: Sreerenj Balachandran <sreerenj.balachandran@intel.com>
Wang, Tiantian [Thu, 30 Mar 2017 01:22:03 +0000 (21:22 -0400)]
Add imported surface alignment check
This fixes issues #62
https://github.com/01org/intel-vaapi-driver/issues/62
Signed-off-by: Wang Tiantian <tiantian.wang@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Wang, Tiantian [Thu, 30 Mar 2017 01:35:55 +0000 (21:35 -0400)]
Fix wrong y_cb/cr_offset when the imported surfaces's fourcc is VA_FOURCC_411P
Signed-off-by: Wang Tiantian <tiantian.wang@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
jkyu [Thu, 30 Mar 2017 23:28:02 +0000 (19:28 -0400)]
Add the video encoder support for GLK
HEVC encoder will be enabled later.
Fix https://github.com/01org/intel-vaapi-driver/issues/111
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Wang, TianTian<tiantian.wang@intel.com>
Signed-off-by: jkyu <jiankang.yu@intel.com>
Xiang, Haihao [Thu, 6 Apr 2017 13:03:23 +0000 (21:03 +0800)]
Bump intel-vaapi-driver to 1.8.2.pre1 for development
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 6 Apr 2017 14:26:09 +0000 (22:26 +0800)]
test: check whether MVC encoding is support
This fixes https://github.com/01org/intel-vaapi-driver/issues/123
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Pengfei Qu [Wed, 5 Apr 2017 05:52:13 +0000 (13:52 +0800)]
ENC: Disable the ROI feature of AVC encoder on SKL/APL/KBL
this is WA to disable the feature as the new improved AVC encoder does not support it by now.
Fixes #108
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Wed, 5 Apr 2017 06:21:29 +0000 (14:21 +0800)]
ENC:disable SVC feature of AVC encoder on SKL/APL/KBL
this is WA to disable SVC featrue as new improved AVC does not support it
Fixes #116
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Wed, 5 Apr 2017 09:28:23 +0000 (17:28 +0800)]
ENC: Add the attribute of VAConfigAttribEncMaxSlices for AVC/HEVC encoder
this patch enable the VAConfigAttribEncMaxSlices attribute and
this also add WA to disable multi slcie feature on APL/KBL/SKL as the improved AVC encoder does not support it.
Fixes #117
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Wed, 5 Apr 2017 06:00:41 +0000 (14:00 +0800)]
ENC: disable MVC feature of AVC encoder on SKL/APL/KBL
this is WA to disable the MVC feature as the new improved AVC encoder does not support it by now.
Fixes #115
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Sebastian Ramacher [Sat, 1 Apr 2017 18:50:22 +0000 (20:50 +0200)]
Fix spelling of 'extended'
Signed-off-by: Sebastian Ramacher <sramacher@debian.org>
Víctor Manuel Jáquez Leal [Mon, 27 Mar 2017 16:05:40 +0000 (09:05 -0700)]
gen8: accept P010 as valid format
The functions gen8_gpe_set_surface2_state() and
gen8_gpe_media_chroma_surface_setup() asserts for the surface color
format, admiting only NV12.
Nonetheless, gen8 also can encoder surfaces with P010 color format.
This patch adds P010 color format in the assertion condition.
Fixes #106
Pengfei Qu [Mon, 27 Mar 2017 01:43:30 +0000 (09:43 +0800)]
Fix: WA the AVC B frame issue with high profile
this is WA to disable the transform_8x8_mode_flag in the driver.
Fixes #97
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Fri, 24 Mar 2017 08:09:24 +0000 (16:09 +0800)]
Fix: wrong type define and cause the memory access error
Fixes #101
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Xiang, Haihao [Thu, 23 Mar 2017 03:43:11 +0000 (11:43 +0800)]
Update shaders for VP8 encoding
This fixes https://github.com/01org/intel-vaapi-driver/issues/59
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:31 +0000 (14:02 +0800)]
Encoder: AVC encoder improved on KBL
v1:
add gtest support for quality range on KBL
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:28 +0000 (14:02 +0800)]
Fix:code refine and replace 52 with Marco defines
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:24 +0000 (14:02 +0800)]
Encoder: add AVC encoder pipeline support on KBL
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:20 +0000 (14:02 +0800)]
ENC: use common gpe structure in AVC encoder
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:17 +0000 (14:02 +0800)]
Encoder: add mbenc structure for KBL
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:02:13 +0000 (14:02 +0800)]
Encoder: add AVC const data for KBL
add member variable for KBL.
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Mon, 20 Mar 2017 06:01:09 +0000 (14:01 +0800)]
Encoder: add AVC kernel binary on KBL
Fixes #73
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Xiang, Haihao [Thu, 16 Mar 2017 05:32:13 +0000 (13:32 +0800)]
Avoid NULL pointer in avc_get_first_mb_bit_offset_with_epb()
This fixes https://github.com/01org/intel-vaapi-driver/issues/92
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 14 Mar 2017 07:18:37 +0000 (15:18 +0800)]
Check pointers against NULL before using
This fixes https://github.com/01org/intel-vaapi-driver/issues/90
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
peng.chen [Fri, 10 Mar 2017 07:19:25 +0000 (02:19 -0500)]
Add the support for GLK in the intel-driver
Decoder and VPP are enabled with this patch.
Encoder will be enabled later.
Fix https://github.com/01org/intel-vaapi-driver/issues/80
Signed-off-by: Peng.Chen <peng.c.chen@intel.com>
Signed-off-by: Yu, JianKang <jiankang.yu@intel.com>
Signed-off-by: Wang, TianTian<tiantian.wang@intel.com>
Pengfei Qu [Wed, 15 Mar 2017 03:16:47 +0000 (11:16 +0800)]
Fix:ROI number should be 4
Fixes #88
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Daniel Charles [Mon, 13 Mar 2017 17:41:13 +0000 (10:41 -0700)]
configure: change package tarball name to use hyphens
Compilation scripts on different distributions (i.e. arch, gentoo) rely
on the tarball name to use hyphens between words.
This patch will create tarballs with hyphenated names and keep the
legacy package naming
TEST="make dist, should produce intel-vaapi-driver.tar.*"
Fixes #85
Signed-off-by: Daniel Charles <daniel.charles@intel.com>
Leonardo Taccari [Mon, 6 Mar 2017 21:40:25 +0000 (22:40 +0100)]
i965_decoder_utils: replace alloca(3) with malloc(3) and free(3)
alloca(3) usage is less portable than malloc(3)/free(3) (this fixes
issue #76 pointed out by building intel-vaapi-driver on NetBSD).
Suggested and thanks to @xhaihao and @yakuizhao via pull request #77.
Signed-off-by: Leonardo Taccari <iamleot@gmail.com>
Lim Siew Hoon [Wed, 8 Mar 2017 11:51:22 +0000 (19:51 +0800)]
Fix src_obj_surf->orig_height assign to src_rect.height.
Fixes https://github.com/01org/intel-vaapi-driver/issues/79
Signed-off-by: Lim Siew Hoon <siew.hoon.lim@intel.com>
Xiang, Haihao [Tue, 7 Mar 2017 15:53:17 +0000 (23:53 +0800)]
Bump intel-vaapi-driver to 1.8.1.pre1 for development
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Pengfei Qu [Fri, 3 Mar 2017 09:46:19 +0000 (17:46 +0800)]
Fix the quality range regression issue
v1:
use bool to define the member
define the one map data for quality range according to the platform type
Fixed #72
For low quali# Please enter the commit message for your changes. Lines starting
Zhao Yakui [Mon, 6 Mar 2017 05:38:17 +0000 (13:38 +0800)]
VDENC: Avoid using the first index reference frame for the invalid reference
Maybe the ref_list_idx[0][X] is also initialized to zero even for the invalid
reference frame, which is also regarded as valid. In such case the
VDENC_PIPE_BUF will program the valid GPU address for the invalid reference.
>For example: FWD Ref1 is invalid
Fix #70
https://github.com/01org/intel-vaapi-driver/issues/70
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Pengfei Qu [Tue, 14 Feb 2017 09:51:25 +0000 (17:51 +0800)]
ENC:support more quality level and switch to new AVC encoder solution on SKL/APL
v1:
add AVC encoder support on APL
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Pengfei Qu [Tue, 14 Feb 2017 09:46:12 +0000 (17:46 +0800)]
ENC: add MFX pipeline for AVC encoder
MFX pipeline:
add MFX command for AVC encoder
add MFX Picture slice level command init for AVC
add MFX pipeline init prepare run for AVC encode
add VME/MFX context init for AVC encoder
v1:
set memory MOCS property
v2:
add Macro definition of the AVC quality level range
v3:
add check for the reference frame num.
Fixes #43
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Fri, 13 Jan 2017 05:56:04 +0000 (13:56 +0800)]
ENC: add VME pipeline for AVC encoder
VME pipeline:
add resource and surface allocation and free function
add init table for frame mbbrc update
add scaling kernel for AVC encoder
add BRC init reset kernel for AVC RC logic
add BRC frame update-kernel for AVC RC logic
add BRC MB level update kernel for AVC RC logic
add REF frame QA caculation and MB level const data
add MBENC kernel for AVC encoder
add ME kernel for AVC encoder
add WP/SFD kernel for AVC encoder
add kernel init/destroy function for AVC encoder
add kernel related parameter check function for AVC
add VME pipeline init prepare/run function for AVC encoder
v1:
allocate the 32x surface accordlingly.
Fixes #43
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Tue, 7 Feb 2017 08:55:40 +0000 (16:55 +0800)]
ENC: add AVC common structure and functions
v1:add kernel pointer for different platform
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Pengfei Qu [Wed, 28 Dec 2016 02:17:02 +0000 (10:17 +0800)]
ENC: add kernel related structure and define for AVC
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Pengfei Qu [Wed, 28 Dec 2016 02:13:35 +0000 (10:13 +0800)]
ENC: add AVC kernel binary on SKL
v1:move kernal array into the c file
v2:add file into Makefile.am
v3:use the Apache 2 license header for media kernel binary
Fixes #43
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Pengfei Qu [Wed, 28 Dec 2016 02:10:06 +0000 (10:10 +0800)]
ENC: add const data/table for AVC encoder
v1:
add fies in the Makefile.am
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Pengfei Qu [Wed, 28 Dec 2016 02:07:26 +0000 (10:07 +0800)]
ENC: add common structure for AVC encoder
v1:
add context init function for AVC encoder
v2:
add file in the Makefile.am
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Pengfei Qu [Wed, 28 Dec 2016 01:48:15 +0000 (09:48 +0800)]
ENC: move gpe related function into src/i965_gpe_utils.h/c
v1:
add align version for obj surface conversion to gpe surface
remove comments and enum value
v2:
use intel->media_mocs to configure the memory property
v3:
add maro definition for AVC status/ctl register
Fixes #43
Signed-off-by: Pengfei Qu <Pengfei.Qu@intel.com>
Reviewed-by: Sean V Kelley<seanvk@posteo.de>
Xiang, Haihao [Mon, 27 Feb 2017 04:58:55 +0000 (12:58 +0800)]
Update license header for VP8/VP9 encoding kernels
The VP8/VP9 encoding kernels are licensed under the Apache License, Version 2.0
This fixes https://github.com/01org/intel-vaapi-driver/issues/65
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Hyunjun Ko [Wed, 22 Feb 2017 10:33:34 +0000 (19:33 +0900)]
JPEG Encoder: return supported pixel formats for vaQuerySurfaceAttributes
Since commit
d540e43 landed, we can query with VAEntrypointEncPicture
entrypoint for jpeg encoder. But it doesn't still return some formats,
which jpeg encoder is capable of handling.
Fixes #60
Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Zhao Yakui [Thu, 23 Feb 2017 08:27:43 +0000 (16:27 +0800)]
VDENC: Fix the incorrect shift in REF_IDX_STATE command
This is to fix the issue #63
https://github.com/01org/intel-vaapi-driver/issues/63
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Xiang, Haihao [Mon, 27 Feb 2017 02:41:34 +0000 (10:41 +0800)]
Update the README
Kaby Lake supports VP9 8bit encode and HEVC 10bit encode
This fixes https://github.com/01org/intel-vaapi-driver/issues/66
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Daniel Charles [Tue, 21 Feb 2017 23:31:46 +0000 (15:31 -0800)]
configure: check for x11 on path
enable-x11 is set to yes by default and during configuration it has to
check for X11 header files on the path then the build system is prepared
to avoid compiling X11 related code if headers are not present.
Fixes #48
TEST="autogen.sh --prefix=<some prefix>, make sure X11 headers are not installed"
TEST="x11 should not print out as supported on the configure summary"
Signed-off-by: Daniel Charles <daniel.charles@intel.com>
Xiang, Haihao [Tue, 10 Jan 2017 22:02:02 +0000 (14:02 -0800)]
Set the pipeline to use the new VP8 encoding shaders on SKL/BXT/KBL
v2: correct the license header in gen9_encoder_vp8.c
This is a portion of patches for VP8 rework on SKL/BXT/KBL
Fixes github.com/01org/intel-vaapi-driver/issues/40 now
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 10 Jan 2017 22:02:01 +0000 (14:02 -0800)]
Add new VP8 encoding shaders for SKL/BXT/KBL
v2: add license header to each shader
This is a portion of patches for VP8 rework on SKL/BXT/KBL
github.com/01org/intel-vaapi-driver/issues/40
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Xiang, Haihao [Tue, 10 Jan 2017 22:02:00 +0000 (14:02 -0800)]
Set the pipeline to use the new VP8 encoding shaders on BSW
Currently only one temporal layer is supported
v2:
Use OUT_RELOC64 instead of OUT_RELOC for 48 bits address
Fix the calculation of min_bit_rate in i965_encoder_vp8_get_misc_parameters()
Fix the calculation of average_bitrate and max_bitrate in i965_encoder_vp8_vme_brc_init_reset_set_curbe()
Move the constant tables copied & pasted from libvpx to separate files
Use the common MOCS state
v3:
Move all VP8 probability tables from a .h to a .c file
Rename vp8_quant.h to vpx_quant.h which can be used for both VP8 and VP9
Remove the unused search path in vp8_search_path and update the comment
Correct the license header in some files
v4:
Move all VP8 quantization tables from a .h to a .c file
Add 'static' or 'static const' to variables in local file scope
Free the private surface data if required
This is a portion of patches for VP8 rework on BSW
Fixes github.com/01org/intel-vaapi-driver/issues/39 now
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Xiang, Haihao [Tue, 10 Jan 2017 22:01:59 +0000 (14:01 -0800)]
Add new VP8 encoding shaders for BSW
v2: Add the license header to each shader
This is a portion of patches for VP8 rework on BSW
github.com/01org/intel-vaapi-driver/issues/39
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Zhao Yakui [Fri, 17 Feb 2017 07:55:12 +0000 (15:55 +0800)]
Fix the incorrect usage of buffer domain for VDENC
At the same time the cache is fixed for VDENC reference.
Fix the issue #56
https://github.com/01org/intel-vaapi-driver/issues/56
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Hyunjun Ko [Fri, 17 Feb 2017 05:27:58 +0000 (14:27 +0900)]
wayland: return error rather than assertion in case of invalid format
return error instead of assertion, so that appilcation could handle
these cases.
Fixes #54
Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
peng.chen [Wed, 15 Feb 2017 05:17:15 +0000 (13:17 +0800)]
Move the down-sampled NV12 surface handlers into the reconstructed surface's private data from the input surface's for hevc encoder
Fixes #42.
Signed-off-by: peng.chen <peng.c.chen@intel.com>
peng.chen [Tue, 14 Feb 2017 01:07:44 +0000 (09:07 +0800)]
Fix the linear P010 aborts bug for hevc main10 encoding.
Fixes #41.
Signed-off-by: peng.chen <peng.c.chen@intel.com>
Hans de Goede [Wed, 15 Feb 2017 14:21:12 +0000 (15:21 +0100)]
Wayland-drm: Fix not finding wl_drm_interface on systems with libglvnd
We do not want just any libEGL.so.1 we want mesa's libEGL.so.1 as that
is the only way which defines the wl_drm_interface symbol we need,
one systems with libglvnd libEGL.so.1 is a dispatcher library provided
by libglvnd and the actual mesa libEGL we want is named libEGL_mesa.so.0
so try that first.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Xiang, Haihao [Tue, 14 Feb 2017 06:25:51 +0000 (14:25 +0800)]
Do not return VA_STATUS_SUCCESS when checking (VAProfileVP9Profile0, VAEntrypointEncSlice) on SKL
SKL doesn't support VP9 hardware accelerated encoding, the hybrid driver also doesn't
support VP9 encoding.
This fixes https://github.com/01org/intel-vaapi-driver/issues/46
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Tue, 14 Feb 2017 06:52:40 +0000 (14:52 +0800)]
Query the Huc on APL/KBL to check whether VDENC BRC is supported
The VDENC BRC has the dependency on Huc. When the HUC is loaded,
it will return the corresponding BRC attribute.
Fix the issue #45
https://github.com/01org/intel-vaapi-driver/issues/44
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Tue, 14 Feb 2017 06:52:40 +0000 (14:52 +0800)]
Follow the HW requirement to configure VDENC on KBL
V2->V3: Remove the write of MI_FLUSH between multi-slices
V1->V2: Save/Restore the macroblock_address for multi-slices
Remove the unnecessary code of insert_one_zero_byte
as the WA is always needed.
Fix the issue #44
https://github.com/01org/intel-vaapi-driver/issues/44
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Tue, 14 Feb 2017 06:52:40 +0000 (14:52 +0800)]
Fix the incorrect ver-coord for VDENC encoding
Currently the ver-coord is wrong for VDENC command especially when multi-slices
are used.
Fix the issue #44
https://github.com/01org/intel-vaapi-driver/issues/44
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Thu, 9 Feb 2017 09:20:46 +0000 (17:20 +0800)]
VDENC: add the const prefix to avoid updating the const table
Fix the issue #36
https://github.com/01org/intel-vaapi-driver/issues/36
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Víctor Manuel Jáquez Leal [Wed, 8 Feb 2017 18:54:00 +0000 (19:54 +0100)]
JPEG Encoder: support vaQuerySurfaceAttributes() pixel format
When querying the surface attributes, using the JPEG encoder
(profile VAProfileJPEGBaseline, entrypoint VAEntrypointEncPicture)
the function doesn't return the available VASurfaceAttribPixelFormat
because it is not considered in the driver's code.
This patch returns the default pixel formats for encoders entry
points.
This error was reported in
https://bugzilla.gnome.org/show_bug.cgi?id=775490
Fixes #38
Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
Sean V Kelley [Sat, 4 Feb 2017 08:04:44 +0000 (00:04 -0800)]
config: Explicitly handle wayland-client pkgconfig
Ensure deps are more robustly covered.
fixes #29
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V. Kelley [Sat, 4 Feb 2017 06:29:33 +0000 (22:29 -0800)]
Merge pull request #28 from xhaihao/master
1037U doesn't support encoding
Xiang, Haihao [Fri, 3 Feb 2017 08:08:26 +0000 (16:08 +0800)]
1037U doesn't support encoding
1037U is recognized as Ivy Bridge, but actually 1037U doesn't
support encoding, so we have to add it into the cpu hook list.
Refer to https://github.com/01org/intel-vaapi-driver/issues/3 for more info.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Sean V. Kelley [Thu, 2 Feb 2017 20:05:22 +0000 (12:05 -0800)]
Merge pull request #25 from tmm1/add-missing-ifdef
i965: remove unnecessary code from i965_Terminate when compiling without HAVE_HYBRID_CODEC
Aman Gupta [Thu, 2 Feb 2017 17:42:09 +0000 (09:42 -0800)]
i965_drv: fix building with --enable-static
Avoids compiling some unnecessary code in i965_Terminate when
building without HAVE_HYBRID_CODEC.
In particular, this removes the invocation to dlclose(), which
means the driver can be compiled without a dependency on libdl.
Previously, the generated libi965_drv_video.a could not be used
statically because it referenced dlclose(). The corresponding
dlopen() is already guarded behind the HAVE_HYBRID_CODEC flag.
Fixes #27
Signed-off-by: Aman Gupta <aman@tmm1.net>
Sean V. Kelley [Wed, 1 Feb 2017 23:56:02 +0000 (15:56 -0800)]
Merge pull request #23 from seanvk/remove_debiancontrols
debian: Remove legacy packaging
Sean V Kelley [Wed, 1 Feb 2017 19:44:10 +0000 (11:44 -0800)]
debian: Remove legacy packaging
debian control files are maintained by packagers
Yes there is a "convenience" factor to roll your
own debian package, but this is not maintained by
us and represents a fork from debian packagers.
As with the bit rot of libva.spec, remove this.
fixes #22
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Wed, 1 Feb 2017 04:52:48 +0000 (20:52 -0800)]
config: Update for github
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Wed, 1 Feb 2017 04:12:21 +0000 (20:12 -0800)]
config: Update project name
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Tue, 31 Jan 2017 05:40:07 +0000 (21:40 -0800)]
travis: Added new security list for scans
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Tue, 31 Jan 2017 05:08:05 +0000 (21:08 -0800)]
github: Removed coveralls integration
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Tue, 31 Jan 2017 04:51:32 +0000 (20:51 -0800)]
github: Added slack team info
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Sun, 29 Jan 2017 23:05:58 +0000 (15:05 -0800)]
travis: Add support for coverity scan to travis
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Sun, 29 Jan 2017 02:18:30 +0000 (18:18 -0800)]
github: Updated name for kanban
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Thu, 26 Jan 2017 23:10:19 +0000 (15:10 -0800)]
github: Updated mailing list
Also corrected link to filing new issues.
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Thu, 26 Jan 2017 22:43:53 +0000 (14:43 -0800)]
github: Update contact info
Update with new mailing list and irc channel.
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 04:21:31 +0000 (20:21 -0800)]
travis: Add hook to slack
Add notifications for travis to slack
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 02:22:16 +0000 (18:22 -0800)]
github : Corrected project names
Switched from interface to driver
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 02:17:42 +0000 (18:17 -0800)]
github: Add badges for github services
Add badges for waffle.io, travis, and coveralls
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 02:03:07 +0000 (18:03 -0800)]
travis: Removed libva-dev from deps
Building it locally, also had a typo.
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 01:56:48 +0000 (17:56 -0800)]
travis: Added preliminary travis yml file
This is a simple travis file.
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Mon, 23 Jan 2017 01:22:05 +0000 (17:22 -0800)]
intel-vaapi-driver: Add github docs
Adding documentation for the github site.
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Kuang-che Wu [Wed, 18 Jan 2017 18:51:29 +0000 (10:51 -0800)]
libva-intel-driver: fix i965 encoder wrong bits shift operation
shift uint32_t by 32 bits is undefined behavior.
For this particular case: when invoke avc_bitstream_put_ui() with 32
bits value at byte position of multiple of 4, existing 32 bits garbage
data in the buffer may be retained instead of cleared. The result is,
the position of NALU start code (0x00000001) looks like overwritten by
garbage value.
Patch has been tested and used upstream:
https://chromium-review.googlesource.com/#/c/410541/
Signed-off-by: Kuang-che Wu <kcwu@chromium.org>
Signed-off-by: Sean V Kelley <seanvk@posteo.de>
Reviewed-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Tue, 17 Jan 2017 00:40:20 +0000 (08:40 +0800)]
Fix the incorrect reference count of reconstructed_object for VP9 enc
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Zhao Yakui [Tue, 17 Jan 2017 00:40:19 +0000 (08:40 +0800)]
Fix the incorrect 48-bit address reallocation
Currently it works well for 32-bit address. But it will cause that the
upper 32-bit address is incorrect if the 48-bit address buffer is allocated.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Zhao Yakui [Tue, 17 Jan 2017 00:40:18 +0000 (08:40 +0800)]
Follow the HW spec to configure the buffer cache on Gen9+
The MOCS field is used to define the cache type for the given buffer. From the
SKL+, the MOCS field is interpreted as the index that is used to find the corresponding
cache type in kernel driver. The current MOCS setting causes that buffer uses
the wrong cache type.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Sean V Kelley <sean.v.kelley@intel.com>
Zhao Yakui [Tue, 17 Jan 2017 00:40:17 +0000 (08:40 +0800)]
Fix the incorrect configuration of media_pipeline power domain
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Zhao Yakui [Tue, 17 Jan 2017 00:40:16 +0000 (08:40 +0800)]
H264 Encoding:Free aux_batchbuffer to configure access domain correctly for PAK_OBJ command buffer
The access domain is not configured correctly for PAK_OBJ command buffer.
And it causes that the buffer content is not synchronized correctly.
At the same time the 64-byte is aligned for the boundary between
CPU and GPU access instead of 16-byte.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Sean V Kelley <seanvk@posteo.de>
Sean V Kelley [Tue, 17 Jan 2017 23:01:36 +0000 (15:01 -0800)]
Revert "ENC: move gpe related function into src/i965_gpe_utils.h/c"
This reverts commit
e1783804bcb35385ffabaeeeb8b4cc3c03d513e0.
Sean V Kelley [Tue, 17 Jan 2017 23:01:35 +0000 (15:01 -0800)]
Revert "ENC: add common structure for AVC/HEVC encoder"
This reverts commit
d710462a41cac9b6d12aefab89734674084ab15f.
Sean V Kelley [Tue, 17 Jan 2017 23:01:34 +0000 (15:01 -0800)]
Revert "ENC:add context init function for AVC/HEVC encoder"
This reverts commit
4679bb2de3d9f00287535c2abc5d5f16b389b69b.