[Beignet] [PATCH] Fix build with clang++

2016-02-21 Thread Armin K
Dynamic length arrays don't seem to be supported. Also, default floating point type is double in C++, no need for "d" suffix, which isn't supported by clang++ --- backend/src/backend/gen8_context.cpp | 4 ++-- backend/src/llvm/llvm_gen_backend.cpp | 2 +- 2 files changed, 3 insertions(+), 3 dele

[Beignet] [PATCH] RFC: Fix building with clang++ -stdlib=libc++

2016-02-21 Thread Armin K
libc++ ships its own math.h file which includes ldexp definition that conflicts with one in the test. This works around the build problems. I can't verify if it works at runtime. I have llvm 3.8/clang 3.8 rc2 which is claimed not to be properly supported at the moment. --- utests/utest_math_gen.py

[Beignet] [PATCH] Fix building with clang++

2016-02-21 Thread Armin K
Don't include d suffix for double values. Default floating point values in C++ are doubles anyways unless specified otherwise Using d suffix works with g++ but not with clang++ --- utests/compiler_double_convert.cpp | 28 ++-- utests/compiler_double_div.cpp | 12 ++

[Beignet] [PATCH] Fix build with latest libdrm

2016-10-05 Thread Armin K
libdrm-2.4.71 contains drm_intel_get_pooled_eu and drm_intel_get_min_eu_in_pool with different signatures. Signed-off-by: Armin K --- src/intel/intel_driver.c | 9 ++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/src/intel/intel_driver.c b/src/intel/intel_driver.c index

[Beignet] [PATCH] buildsys: Use CMRT_LIBDIR instead of CMRT_LIBRARY_DIRS

2016-10-05 Thread Armin K
CMRT_LIBRARY_DIRS doesn't include any library paths when the library is installed system-wide, such as /usr. Also dlopen versioned library, as distros tend to split non-versioned sonames into -devel packages. Signed-off-by: Armin K --- src/CMakeLists.txt | 2 +- 1 file changed, 1 inse

Re: [Beignet] [PATCH] buildsys: Use CMRT_LIBDIR instead of CMRT_LIBRARY_DIRS

2016-10-08 Thread Armin K.
l, I just happened to have libcmrt installed because I have intel-hybrid-driver for Skylake VP9 hybrid decoder. Without this patch, the utest testing the cmrt code will fail. See also: https://bugs.freedesktop.org/show_bug.cgi?id=94636 -Original Message- From: Beignet [mailto:beignet-boun...

Re: [Beignet] [PATCH] Fix build with latest libdrm

2016-10-08 Thread Armin K.
On 8.10.2016 12:06, Rebecca N. Palmer wrote: + if (drm_intel_get_pooled_eu(driver->fd) >= 0) { Shouldn't this be >0 , not >=0? The libdrm commit message ( https://cgit.freedesktop.org/mesa/drm/commit/intel/intel_bufmgr_gem.c?id=98887140e343493f01be7a1dec721c024bcf72c7 ) says 0 means not suppor

[Beignet] [PATCH v2] Fix build with latest libdrm

2016-10-09 Thread Armin K
libdrm-2.4.71 contains drm_intel_get_pooled_eu and drm_intel_get_min_eu_in_pool with different signatures. v2: - Fix conditional not to include 0. Signed-off-by: Armin K --- src/intel/intel_driver.c | 9 ++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/src/intel

Re: [Beignet] [PATCH] buildsys: Use CMRT_LIBDIR instead of CMRT_LIBRARY_DIRS

2016-10-09 Thread Armin K.
On 09.10.2016 04:34, Guo, Yejun wrote: > thanks Armin, the patch looks good to me. > > could you also send a patch to fix CMRT_INCLUDE_DIRS? thanks. > > > thanks > yejun > Usage of CMRT_INCLUDE_DIRS is correct. Nothing to fix there. signature.asc Description: OpenPGP digital signature

Re: [Beignet] [PATCH v2] Fix build with latest libdrm

2016-10-09 Thread Armin K.
Hm, I didn't see it. I'm fine with whatever one gets merged, as long as it's fixed. On 10.10.2016 8:05, Igor Gnatenko wrote: I think I had more clear patch in bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98090 On Sun, Oct 9, 2016 at 6:54 PM Armin K mailto:kre...@em