| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
fix the build warning caused by 1c5de1eb22007a5816cd831745e3c1b169bf34d0
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
|
|
|
|
| |
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
|
|
|
|
|
|
| |
glibc could probably use `__GLIBC_PREREQ(2, 30)` but musl doesn't
expose version to consumers. Instead of dancing around various libc
always use local definition.
|
|
|
|
| |
Unlike Linux it uses Meson/Clang to improve coverage.
|
|
|
|
|
|
| |
../va/va.c:455:64: error: use of undeclared identifier 'RTLD_NODELETE'
handle = dlopen( driver_path, RTLD_NOW | RTLD_GLOBAL | RTLD_NODELETE );
^
|
|
|
|
| |
ld: error: unable to find library -ldl
|
|
|
|
| |
On DragonFly and FreeBSD config are installed to /usr/local/etc by default.
|
|
|
|
|
|
|
|
|
| |
BSD sed doesn't support inlining argument for `i` command, so embed
newlines as proposed in http://austingroupbugs.net/view.php?id=249
GEN wayland-drm-client-protocol.c
sed: 1: "1i#include "sysdeps.h"
": command i expects \ followed by text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
../va/va_trace.c:291:28: error: use of undeclared identifier '__NR_gettid'
pid_t thd_id = syscall(__NR_gettid);
^
../va/va_trace.c:669:28: error: use of undeclared identifier '__NR_gettid'
pid_t thd_id = syscall(__NR_gettid);
^
../va/va_trace.c:708:28: error: use of undeclared identifier '__NR_gettid'
pid_t thd_id = syscall(__NR_gettid);
^
../va/va_trace.c:1234:28: error: use of undeclared identifier '__NR_gettid'
pid_t thd_id = syscall(__NR_gettid);
^
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
| |
|
|
|
|
| |
Signed-off-by: He Junyan <junyan.he@intel.com>
|
|
|
|
| |
Signed-off-by: He Junyan <junyan.he@intel.com>
|
|
|
|
| |
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
|
|
|
|
| |
Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
|
|
|
|
|
|
| |
Fixes #411
Signed-off-by: Sherry Lin <sherry.lin@intel.com>
|
|
|
|
| |
Signed-off-by: FurongZhang <furong.zhang@intel.com>
|
|
|
|
|
|
|
| |
application could set priority for each VAContext
then HW will check the priority for the tasks related with the VAContext
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
|
|
| |
Added notes about different ways for synchronization.
Signed-off-by: Dmitry Ermilov <dmitry.ermilov@intel.com>
|
|
|
|
|
|
| |
Fixes #397
Signed-off-by: Dmitry Ermilov <dmitry.ermilov@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
|
|
| |
seperate the log and dump
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If code calls va_isDRI2Connected multiple times there is no protection
against calling open(device_name, O_RDWR) again and losing the reference
to the original file descriptor.
A recent change (bc8a12) calls the method twice each time libva is
initialized, thus causing a leak of device file descriptors.
This change moves the initializing of the fd and auth_type to the
creation of the dri_state and then checks to see if they have already
been initialized. If so, the method va_isDRIConnected returns true
and fills in driver_name.
|
|
|
|
| |
Signed-off-by: Xinpeng Sun <xinpeng.sun@intel.com>
|
|
|
|
|
|
|
|
|
|
| |
C source code and header files shouldn't be executable.
This is a problem that already occurred pretty often in the past and was
e.g. fixed in 5d8c5ae, be95415, and 8352ab9. Therefore it seems like a
good idea to add a CI check to prevent this issue from reappearing
(probably mostly accidents due to wrong Git configurations, etc.).
Signed-off-by: Michael Weiss <dev.primeos@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
C source code and header files shouldn't be executable. When building
libva via Meson the affected files would be installed with the execute
file mode bit set (with GNU Autotools this wasn't the case).
To check: find -name '*.[ch]' -executable
Signed-off-by: Michael Weiss <dev.primeos@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Meson allows to use absolute paths for libdir and includedir [0]. But if
they're absolute paths one must not append the prefix. The join_paths()
function [1] nicely supports both relative and absolute paths.
This is e.g. important for Nixpkgs (i.e. Nix(OS)) where binaries and
library files are usually installed under different prefixes.
[0]: https://github.com/mesonbuild/meson/commit/a513bcfde613f2a0403f7b0cd34d4bd62674c1d8
[1]: https://mesonbuild.com/Reference-manual.html#join_paths
Signed-off-by: Michael Weiss <dev.primeos@gmail.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
In function `vaCopySurfaceGLX_impl_libva`, if `pSurfaceGLX->gl_context`
points to `struct OpenGLContextState` memory object with all its members
set to 0, the first call of `gl_set_current_context` returns true
without setting the `old_cs`. Then, in the following next call of
`gl_set_current_context`, `&old_cs` will point to random data on the
stack.
Similarly for `vaDestroySurfaceGLX_impl_libva`.
|
|
|
|
|
|
|
| |
it is a suppliment of https://github.com/intel/libva/pull/220
to distinguish P frame and low delay B frame
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
|
|
| |
Original field names were copied from va_TraceVAPictureParameterBufferH264
Signed-off-by: Ruben Gonzalez <rgonzalez@fluendo.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
|
|
|
| |
Fixes #361
some backend driver does not implement vaPutSurface
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
|
|
| |
Introduce new CAPS flag, fix typos in flag descriptions.
Signed-off-by: Alexei Leonenko <alexei.leonenko@intel.com>
|
|
|
|
|
|
| |
Fixes #396
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
|
|
|
|
|
|
|
| |
Map the i965 DRI driver to iHD and i965 vaapi
driver candidates.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
|
|
|
|
|
|
| |
Enable/Disable IBC and palette mode
Signed-off-by: Tianhao Liu <tianhao.liu@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
MaxCLL and MaxFALL should be in units of 1 nit as
specified in CEA-861.3, Appendix A.
A uint16_t is not large enough to specify the full
range of possible MaxCLL/MaxFALL values in units of
0.0001 cd/m^2. So conclude that the documentation
was just a copy/paste error.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
|
|
|
|
| |
Signed-off-by: Tianhao Liu <tianhao.liu@intel.com>
|
|
|
|
|
|
| |
Bump VA-API version to 1.8.0 pre1 and libva to 2.8.0 pre1
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
|
|
|
| |
Signed-off-by: Tianhao Liu <tianhao.liu@intel.com>
|
|
|
|
| |
Signed-off-by: Carl Zhang <carl.zhang@intel.com>
|
| |
|