Message ID | 20250220210359.214628-3-a-christidis@ti.com |
---|---|
State | New |
Headers | show |
Series | Introducing Opencl-cts | expand |
On 2/20/2025 3:03 PM, a-christidis@ti.com wrote: > From: Antonios Christidis <a-christidis@ti.com> > > Within the cmake build system, the compiler -Werror flag is enabled. This > turns all warnings into errors and doesn't allow the system to compile > correctly. Change the system to ignore the 3 warning rules that show up when > compiling. > > Signed-off-by: Antonios Christidis <a-christidis@ti.com> > --- > v3: > - Changed from adress, it was wrong. > > v2: > - Fixed spelling mistakes > > .../files/0001-Ignore-Compiler-Warnings.patch | 18 ++++++++++++++++++ > 1 file changed, 18 insertions(+) > create mode 100644 meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch > > diff --git a/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch b/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch > new file mode 100644 > index 00000000..067cffa1 > --- /dev/null > +++ b/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch > @@ -0,0 +1,18 @@ > +diff --git a/CMakeLists.txt b/CMakeLists.txt > +index 5cfef6b3..71b6ddb4 100644 > +--- a/CMakeLists.txt > ++++ b/CMakeLists.txt > +@@ -109,7 +109,9 @@ if(CMAKE_COMPILER_IS_GNUCC OR "${CMAKE_CXX_COMPILER_ID}" MATCHES "(Apple)?Clang" > + add_cxx_flag_if_supported(-Wno-error=cpp) # Allow #warning directive > + add_cxx_flag_if_supported(-Wno-unknown-pragmas) # Issue #785 > + add_cxx_flag_if_supported(-Wno-error=asm-operand-widths) # Issue #784 > +- > ++ add_cxx_flag_if_supported(-Wno-maybe-uninitialized) > ++ add_cxx_flag_if_supported(-Wno-stringop-truncation) > ++ add_cxx_flag_if_supported(-Wno-strict-aliasing) > + # -msse -mfpmath=sse to force gcc to use sse for float math, > + # avoiding excess precision problems that cause tests like int2float > + # to falsely fail. -ffloat-store also works, but WG suggested > +-- > +2.34.1 > + The first patch in the series refers to this patch. The second patch (this one) provides the patch. You need to fix in one of two ways: 1) Squash the two patches together in a single patch. 2) Move the inclusion of this patch file in the SRC_URI into this patch. So that the first patch creates the recipe. And the second patch adds a patch file. I lean towards #1, but your call.
diff --git a/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch b/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch new file mode 100644 index 00000000..067cffa1 --- /dev/null +++ b/meta-arago-test/recipes-graphics/opencl/files/0001-Ignore-Compiler-Warnings.patch @@ -0,0 +1,18 @@ +diff --git a/CMakeLists.txt b/CMakeLists.txt +index 5cfef6b3..71b6ddb4 100644 +--- a/CMakeLists.txt ++++ b/CMakeLists.txt +@@ -109,7 +109,9 @@ if(CMAKE_COMPILER_IS_GNUCC OR "${CMAKE_CXX_COMPILER_ID}" MATCHES "(Apple)?Clang" + add_cxx_flag_if_supported(-Wno-error=cpp) # Allow #warning directive + add_cxx_flag_if_supported(-Wno-unknown-pragmas) # Issue #785 + add_cxx_flag_if_supported(-Wno-error=asm-operand-widths) # Issue #784 +- ++ add_cxx_flag_if_supported(-Wno-maybe-uninitialized) ++ add_cxx_flag_if_supported(-Wno-stringop-truncation) ++ add_cxx_flag_if_supported(-Wno-strict-aliasing) + # -msse -mfpmath=sse to force gcc to use sse for float math, + # avoiding excess precision problems that cause tests like int2float + # to falsely fail. -ffloat-store also works, but WG suggested +-- +2.34.1 +