remove test
Moving test to omapdrmtest tree, to be able to re-use display and other
utility code for displaying video directly with drm plane (overlay) or
with dri2 video.
The v4l2/camera util code and fill pattern generation should also make
it easier to add an encoder test in the future.
Moving test to omapdrmtest tree, to be able to re-use display and other
utility code for displaying video directly with drm plane (overlay) or
with dri2 video.
The v4l2/camera util code and fill pattern generation should also make
it easier to add an encoder test in the future.
ducati: update for only core0 (sysm3) builds
Currently upstream remoteproc/rpmsg only supports first core. Since
we don't really need both cores, just move everything over to first
core. Requires 'adjust memory maps to move everything over to core0'
commit in sysbios-rpmsg tree.
Currently upstream remoteproc/rpmsg only supports first core. Since
we don't really need both cores, just move everything over to first
core. Requires 'adjust memory maps to move everything over to core0'
commit in sysbios-rpmsg tree.
begin porting linux/client side
No longer are there syslink or memmgr userspace libs. Instead libdrm
and libdrm_omap are utilized.
No longer are there syslink or memmgr userspace libs. Instead libdrm
and libdrm_omap are utilized.
ducati: add compile time flag to disable requesting slpm_APPM3 if DCE_REQUEST_APPM3==0
Helps integrating DCE in firmware images where slpm_APPM3 is already acquired.
Helps integrating DCE in firmware images where slpm_APPM3 is already acquired.
Fix proxy init/deinit refcount
Fixes crashes when using more than one engine
Fixes crashes when using more than one engine
Rework the resource cleanup code
Fixes m3 crashes when cleaning up for processes that have both encoders and
decoders instantiated
Fixes m3 crashes when cleaning up for processes that have both encoders and
decoders instantiated
Comment out some HDVICP_Reset code that causes hangs switching codec instances.
Fixes hangs when transcoding or using different decoders/encoders in the same
pipeline.
Fixes hangs when transcoding or using different decoders/encoders in the same
pipeline.
base_image: load mpeg4enc codec
..and install the mpeg4enc header
Add mpeg4enc header
dce_app_m3.cfg: uncomment use of ti.omap.deh.Deh again
Workaround for encoders that need getBufferFxn != NULL
dce-test: link to libtimemmgr
Add h264enc header
Add VIDENC2 headers
Initial VIDENC2 implementation
initial setup of ducati build for dce-rpmsg
Please note that the build instructions change (see README). Now
both sysm3 and appm3 images are built.
Please note that the build instructions change (see README). Now
both sysm3 and appm3 images are built.
add videnc2 header for encoders
update README
re-sync with wiki page
re-sync with wiki page
test: add "-1" arg to allocate 1D output buffers
If first arg is "-1" then 1D "paged mode" buffers will be used for
decoder output.
If first arg is "-1" then 1D "paged mode" buffers will be used for
decoder output.
Fix IVC1VDEC_Status definition
Update codec headers
dce_app_m3.cfg: update heap sizes to match the new memory layout
ducatiplatform.xs: update to match the glp1.4 memory layout
dce_app_m3.cfg: update linker section mappings
ducati: fix incorrect allocation of non-tiled requests
sizeDim1 is only used for 2d requests.
sizeDim1 is only used for 2d requests.
test: use base class params
Don't use codec specific derived structs, since formats may change
over time and we don't need to set any h264 specific settings.
Don't use codec specific derived structs, since formats may change
over time and we don't need to set any h264 specific settings.
fix pkg-config dependencies
update firmware
These pre-built images work with ubuntu 11.04 (or probably any kernel
with syslink-2.0 and memmgr-2.0)
These pre-built images work with ubuntu 11.04 (or probably any kernel
with syslink-2.0 and memmgr-2.0)
add files to build ducati (m3) side firmware
remove ducati submodule
Will be moving this in tree now that codecs are released. See:
http://software-dl.ti.com/dsps/dsps_public_sw/codecs/HDVICP2/index_FDS.html
Will be moving this in tree now that codecs are released. See:
http://software-dl.ti.com/dsps/dsps_public_sw/codecs/HDVICP2/index_FDS.html
fix leak in heap used by codecs
A bug in the deallocation function resulted that memory was not actually
freed properly, causing eventual exhaustion of heap.
A bug in the deallocation function resulted that memory was not actually
freed properly, causing eventual exhaustion of heap.
add ducati and IVAHD power management
some trace cleanups, and update HDVICP
Updated HDVICP fixes some issues seen with realvideo codec.
Updated HDVICP fixes some issues seen with realvideo codec.
update realvdec codec
add missing Ipc_destroy() call at shutdown
updates for latest h264dec
Additional parameters to set, and fix for width/height that is not a
multiple of 16.
Additional parameters to set, and fix for width/height that is not a
multiple of 16.
update h264dec, realvideo, mpeg4dec
add mpeg2vdec
resource cleanup
Support to automatically cleanup after apps that don't exit properly,
crash, etc.
Support to automatically cleanup after apps that don't exit properly,
crash, etc.
fix crash in VIDDEC3_delete
fix typo
do initialization/cleanup on Engine_open()/Engine_close()
rather than via constructor/destructor functions
rather than via constructor/destructor functions
Adding sysm3 image too
This can be build from syslink public git tree, so it isn't strictly
required. But is provided for convenience in case you don't feel like
setting up the toolchain and dependencies to build syslink.
This can be build from syslink public git tree, so it isn't strictly
required. But is provided for convenience in case you don't feel like
setting up the toolchain and dependencies to build syslink.
Disable debug traces by default
updates for L24.11
TILER allocations that used to be proxied via syslink_daemon now are
handled by the linux side client by way of a memsrv RcmServer. This causes
any allocations done by codec to be recovered when the libdce process
exits.
It would be nicer if syslink provided a memsrv via a shared library so we
didn't have to copy all that code into this tree.. but until then, memsrv.c
is a hacked up variant of one of the syslink sample apps.
TILER allocations that used to be proxied via syslink_daemon now are
handled by the linux side client by way of a memsrv RcmServer. This causes
any allocations done by codec to be recovered when the libdce process
exits.
It would be nicer if syslink provided a memsrv via a shared library so we
didn't have to copy all that code into this tree.. but until then, memsrv.c
is a hacked up variant of one of the syslink sample apps.
add more codecs
add mpeg4dec codec
initial commit