On Thu, May 16, 2013 at 03:58:50PM +0800, He Junyan wrote: > On 05/16/2013 01:44 AM, Tom Stellard wrote: > >On Wed, May 15, 2013 at 02:57:53PM +0800, [email protected] wrote: > >>From: Junyan He <[email protected]> > >> > >>Some OpenCL lib's name is not libOpenCL.so. > >>Some thing like libCL.so, libcl.so may exist, > >>and in these cases the check failed. > >>Add the names to fix this problem. > >> > >>Signed-off-by: Junyan He <[email protected]> > >>--- > >> cmake/Modules/FindOpenCL.cmake | 3 ++- > >> 1 file changed, 2 insertions(+), 1 deletion(-) > >> > >>diff --git a/cmake/Modules/FindOpenCL.cmake b/cmake/Modules/FindOpenCL.cmake > >>index cc66714..121dcf2 100644 > >>--- a/cmake/Modules/FindOpenCL.cmake > >>+++ b/cmake/Modules/FindOpenCL.cmake > >>@@ -35,7 +35,8 @@ > >> find_path(OPENCL_INCLUDE_PATH CL/opencl.h) > >> endif() > >>- find_library(OPENCL_opencl_LIBRARY OpenCL) > >>+ find_library(OPENCL_opencl_LIBRARY > >>+ NAMES OpenCL CL cl) > >> > >Which OpenCL implementations create libCL.so or libcl.so ? Is the > >purpose of this to test implementations without going through the ICD > >loader? > yes, the intel's openCL open source code implement called beignet > use the name libcl.so. > And on my platform I do not want to install ICD every time, so I > want to loader my CL lib > directly. I think it will have no harm to the current behavior, > because it will check the > existence of libOpenCL.so firstly.
I've pushed this patch, thanks! -Tom > > > >I don't know much about CMake, but this patch is OK with me as long as > >it doesn't break the current behavior. > > > >-Tom > >> include(FindPackageHandleStandardArgs) > >> FIND_PACKAGE_HANDLE_STANDARD_ARGS(OpenCL > >>-- > >>1.7.9.5 > >> > >>_______________________________________________ > >>Piglit mailing list > >>[email protected] > >>http://lists.freedesktop.org/mailman/listinfo/piglit > _______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
