Merge remote-tracking branch 'upstream/develop' into develop

.gitignore

@@ -5,6 +5,7 @@
 *.def
 *.o
 *.out
+*.tmp
 lapack-3.1.1
 lapack-3.1.1.tgz
 lapack-3.4.1

.travis.yml

@@ -1,4 +1,119 @@
+# XXX: Precise is already deprecated, new default is Trusty.
+# https://blog.travis-ci.com/2017-07-11-trusty-as-default-linux-is-coming
+dist: precise
+sudo: false
 language: c
+compiler: gcc
+
+jobs:
+  include:
+    - &test-ubuntu
+      stage: test
+      addons:
+        apt:
+          packages:
+            - gfortran
+      before_script: &common-before
+        - COMMON_FLAGS="DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32"
+      script:
+        - set -e
+        - make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+        - make -C test $COMMON_FLAGS $BTYPE
+        - make -C ctest $COMMON_FLAGS $BTYPE
+        - make -C utest $COMMON_FLAGS $BTYPE
+      env:
+        - TARGET_BOX=LINUX64
+        - BTYPE="BINARY=64"
+
+    - <<: *test-ubuntu
+      env:
+        - TARGET_BOX=LINUX64
+        - BTYPE="BINARY=64 USE_OPENMP=1"
+
+    - <<: *test-ubuntu
+      env:
+        - TARGET_BOX=LINUX64
+        - BTYPE="BINARY=64 INTERFACE64=1"
+
+    - <<: *test-ubuntu
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+            - gfortran-multilib
+      env:
+        - TARGET_BOX=LINUX32
+        - BTYPE="BINARY=32"
+
+    - stage: test
+      addons:
+        apt:
+          packages:
+            - binutils-mingw-w64-x86-64
+            - gcc-mingw-w64-x86-64
+            - gfortran-mingw-w64-x86-64
+      before_script: *common-before
+      script:
+        - make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+      env:
+        - TARGET_BOX=WIN64
+        - BTYPE="BINARY=64 HOSTCC=gcc CC=x86_64-w64-mingw32-gcc FC=x86_64-w64-mingw32-gfortran"
+
+    # Build & test on Alpine Linux inside chroot, i.e. on system with musl libc.
+    # These jobs needs sudo, so Travis runs them on VM-based infrastructure
+    # which is slower than container-based infrastructure used for jobs
+    # that don't require sudo.
+    - &test-alpine
+      stage: test
+      dist: trusty
+      sudo: true
+      language: minimal
+      before_install:
+        - "wget 'https://raw.githubusercontent.com/alpinelinux/alpine-chroot-install/v0.6.0/alpine-chroot-install' \
+              && echo 'a827a4ba3d0817e7c88bae17fe34e50204983d1e  alpine-chroot-install' | sha1sum -c || exit 1"
+        - alpine() { /alpine/enter-chroot -u "$USER" "$@"; }
+      install:
+        - sudo sh alpine-chroot-install -p 'build-base gfortran perl linux-headers'
+      before_script: *common-before
+      script:
+        - set -e
+        # XXX: Disable some warnings for now to avoid exceeding Travis limit for log size.
+        - alpine make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+              CFLAGS="-Wno-misleading-indentation -Wno-sign-conversion -Wno-incompatible-pointer-types"
+        - alpine make -C test $COMMON_FLAGS $BTYPE
+        - alpine make -C ctest $COMMON_FLAGS $BTYPE
+        - alpine make -C utest $COMMON_FLAGS $BTYPE
+      env:
+        - TARGET_BOX=LINUX64_MUSL
+        - BTYPE="BINARY=64"
+
+    # XXX: This job segfaults in TESTS OF THE COMPLEX LEVEL 3 BLAS,
+    # so it's "allowed to fail" for now (see allow_failures).
+    - &test-alpine-openmp
+      <<: *test-alpine
+      env:
+        - TARGET_BOX=LINUX64_MUSL
+        - BTYPE="BINARY=64 USE_OPENMP=1"
+
+    - <<: *test-alpine
+      env:
+        - TARGET_BOX=LINUX64_MUSL
+        - BTYPE="BINARY=64 INTERFACE64=1"
+
+    # Build with the same flags as Alpine do in OpenBLAS package.
+    - <<: *test-alpine
+      env:
+        - TARGET_BOX=LINUX64_MUSL
+        - BTYPE="BINARY=64 NO_AFFINITY=1 USE_OPENMP=0 NO_LAPACK=0 TARGET=core2"
+
+  allow_failures:
+    - <<: *test-alpine-openmp
+
+# whitelist
+branches:
+  only:
+    - master
+    - develop
 
 notifications:
   webhooks:
@@ -7,32 +122,3 @@ notifications:
     on_success: change  # options: [always|never|change] default: always
     on_failure: always  # options: [always|never|change] default: always
     on_start: never     # options: [always|never|change] default: always
-
-compiler:
-  - gcc
-
-env:
-  - TARGET_BOX=LINUX64 BTYPE="BINARY=64"
-  - TARGET_BOX=LINUX64 BTYPE="BINARY=64 USE_OPENMP=1"
-  - TARGET_BOX=LINUX64 BTYPE="BINARY=64 INTERFACE64=1"
-  - TARGET_BOX=LINUX32 BTYPE="BINARY=32"
-  - TARGET_BOX=WIN64 BTYPE="BINARY=64 HOSTCC=gcc CC=x86_64-w64-mingw32-gcc FC=x86_64-w64-mingw32-gfortran"
-
-before_install:
- - sudo apt-get update -qq
- - sudo apt-get install -qq gfortran
- - if [[ "$TARGET_BOX" == "WIN64" ]]; then sudo apt-get install -qq binutils-mingw-w64-x86-64 gcc-mingw-w64-x86-64 gfortran-mingw-w64-x86-64; fi
- - if [[ "$TARGET_BOX" == "LINUX32" ]]; then sudo apt-get install -qq  gcc-multilib gfortran-multilib; fi
-
-script: 
- - set -e
- - make QUIET_MAKE=1 DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32 $BTYPE
- - if [ "$TARGET_BOX" == "LINUX32" ] || [ "$TARGET_BOX" == "LINUX64" ]; then make -C test DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32 $BTYPE; fi
- - if [ "$TARGET_BOX" == "LINUX32" ] || [ "$TARGET_BOX" == "LINUX64" ]; then make -C ctest DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32 $BTYPE; fi
- - if [ "$TARGET_BOX" == "LINUX32" ] || [ "$TARGET_BOX" == "LINUX64" ]; then make -C utest DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32 $BTYPE; fi
-
-# whitelist
-branches:
-  only:
-    - master
-    - develop

CMakeLists.txt

@@ -231,43 +231,33 @@ install(TARGETS ${OpenBLAS_LIBNAME}
 
 # Install include files
 	set (GENCONFIG_BIN ${CMAKE_BINARY_DIR}/gen_config_h${CMAKE_EXECUTABLE_SUFFIX})
-	ADD_CUSTOM_COMMAND(
-	OUTPUT ${CMAKE_BINARY_DIR}/openblas_config.h
-	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/config.h
-	COMMAND ${GENCONFIG_BIN} ${CMAKE_CURRENT_SOURCE_DIR}/config.h ${CMAKE_CURRENT_SOURCE_DIR}/openblas_config_template.h > ${CMAKE_BINARY_DIR}/openblas_config.h
-	)
 
-	ADD_CUSTOM_TARGET(genconfig
-	ALL
-	DEPENDS openblas_config.h
-	)
-	add_dependencies(genconfig ${OpenBLAS_LIBNAME})
+	execute_process(COMMAND ${GENCONFIG_BIN}
+	                        ${CMAKE_CURRENT_SOURCE_DIR}/config.h
+	                        ${CMAKE_CURRENT_SOURCE_DIR}/openblas_config_template.h
+	                OUTPUT_VARIABLE OPENBLAS_CONFIG_H_CONTENTS)
+
+	file(WRITE ${CMAKE_BINARY_DIR}/openblas_config.tmp "${OPENBLAS_CONFIG_H_CONTENTS}")
+	configure_file(${CMAKE_BINARY_DIR}/openblas_config.tmp ${CMAKE_BINARY_DIR}/openblas_config.h COPYONLY)
 
 	install (FILES ${CMAKE_BINARY_DIR}/openblas_config.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 
 	message(STATUS "Generating f77blas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
-	ADD_CUSTOM_TARGET(genf77blas
-	ALL
-	COMMAND ${AWK} 'BEGIN{print \"\#ifndef OPENBLAS_F77BLAS_H\" \; print \"\#define OPENBLAS_F77BLAS_H\" \; print \"\#include \\"openblas_config.h\\" \"}; NF {print}; END{print \"\#endif\"}' ${CMAKE_CURRENT_SOURCE_DIR}/common_interface.h > ${CMAKE_BINARY_DIR}/f77blas.h
-	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/config.h
-	)
-	add_dependencies(genf77blas ${OpenBLAS_LIBNAME})
-
+	file(WRITE ${CMAKE_BINARY_DIR}/f77blas.h "")
+	file(APPEND ${CMAKE_BINARY_DIR}/f77blas.h "#ifndef OPENBLAS_F77BLAS_H\n#define OPENBLAS_F77BLAS_H\n#include \"openblas_config.h\"\n")
+	file(READ ${CMAKE_CURRENT_SOURCE_DIR}/common_interface.h COMMON_INTERFACE_H_CONTENTS)
+	file(APPEND ${CMAKE_BINARY_DIR}/f77blas.h "${COMMON_INTERFACE_H_CONTENTS}")
+	file(APPEND ${CMAKE_BINARY_DIR}/f77blas.h "#endif")
 	install (FILES ${CMAKE_BINARY_DIR}/f77blas.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 
 if(NOT NO_CBLAS)
 	message (STATUS "Generating cblas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
-	ADD_CUSTOM_TARGET(gencblas
-	ALL
-	COMMAND ${SED} 's/common/openblas_config/g' ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h > "${CMAKE_BINARY_DIR}/cblas.tmp"
-	COMMAND cp "${CMAKE_BINARY_DIR}/cblas.tmp" "${CMAKE_BINARY_DIR}/cblas.h"
-	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h
-	)
-	add_dependencies(gencblas ${OpenBLAS_LIBNAME})
-
-	install (FILES ${CMAKE_BINARY_DIR}/cblas.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+	file(READ ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h CBLAS_H_CONTENTS)
+	string(REPLACE "common" "openblas_config" CBLAS_H_CONTENTS_NEW "${CBLAS_H_CONTENTS}")
+	file(WRITE ${CMAKE_BINARY_DIR}/cblas.tmp "${CBLAS_H_CONTENTS_NEW}")
+	install (FILES ${CMAKE_BINARY_DIR}/cblas.tmp DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} RENAME cblas.h)
 endif()
 
 if(NOT NO_LAPACKE)
@@ -277,7 +267,7 @@ if(NOT NO_LAPACKE)
 	install (FILES ${INCLUDE_FILES} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
  
 	ADD_CUSTOM_TARGET(genlapacke
-	COMMAND cp ${CMAKE_CURRENT_SOURCE_DIR}/lapack-netlib/LAPACKE/include/lapacke_mangling_with_flags.h.in "${CMAKE_BINARY_DIR}/lapacke_mangling.h"
+	COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/lapack-netlib/LAPACKE/include/lapacke_mangling_with_flags.h.in "${CMAKE_BINARY_DIR}/lapacke_mangling.h"
 	)
 	install (FILES ${CMAKE_BINARY_DIR}/lapacke_mangling.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 endif()

cmake/c_check.cmake

@@ -87,7 +87,7 @@ endif ()
 
 string(TOUPPER ${ARCH} UC_ARCH)
 
-file(WRITE ${TARGET_CONF}
+file(WRITE ${TARGET_CONF_TEMP}
   "#define OS_${HOST_OS}\t1\n"
   "#define ARCH_${UC_ARCH}\t1\n"
   "#define C_${COMPILER_ID}\t1\n"
@@ -95,7 +95,7 @@ file(WRITE ${TARGET_CONF}
   "#define FUNDERSCORE\t${FU}\n")
 
 if (${HOST_OS} STREQUAL "WINDOWSSTORE")
-  file(APPEND ${TARGET_CONF}
+  file(APPEND ${TARGET_CONF_TEMP}
     "#define OS_WINNT\t1\n")
 endif ()
 

cmake/f_check.cmake

@@ -44,7 +44,7 @@ if (NOT ONLY_CBLAS)
   # TODO: set FEXTRALIB flags a la f_check?
 
   set(BU "_")
-  file(APPEND ${TARGET_CONF}
+  file(APPEND ${TARGET_CONF_TEMP}
     "#define BUNDERSCORE _\n"
     "#define NEEDBUNDERSCORE 1\n"
     "#define NEED2UNDERSCORES 0\n")
@@ -56,7 +56,7 @@ else ()
   set(NO_FBLAS 1)
   #set(F_COMPILER GFORTRAN) # CMake handles the fortran compiler
   set(BU "_")
-  file(APPEND ${TARGET_CONF}
+  file(APPEND ${TARGET_CONF_TEMP}
     "#define BUNDERSCORE _\n"
     "#define NEEDBUNDERSCORE 1\n")
 endif()

cmake/lapacke.cmake

@@ -2391,6 +2391,6 @@ foreach (Utils_FILE ${Utils_SRC})
 endforeach ()
 
 set(lapacke_include_dir "${NETLIB_LAPACK_DIR}/LAPACKE/include")
-execute_process(COMMAND ${CMAKE_COMMAND} -E copy "${lapacke_include_dir}/lapacke_mangling_with_flags.h.in" "${lapacke_include_dir}/lapacke_mangling.h")
+configure_file("${lapacke_include_dir}/lapacke_mangling_with_flags.h.in" "${lapacke_include_dir}/lapacke_mangling.h" COPYONLY)
 include_directories(${lapacke_include_dir})
 set_source_files_properties(${LAPACKE_SOURCES} PROPERTIES COMPILE_FLAGS "${LAPACK_CFLAGS}")

cmake/prebuild.cmake

@@ -51,6 +51,7 @@ else()
   set(TARGET_CONF "config.h")
 endif ()
 
+set(TARGET_CONF_TEMP "${PROJECT_BINARY_DIR}/${TARGET_CONF}.tmp")
 include("${PROJECT_SOURCE_DIR}/cmake/c_check.cmake")
 
 if (NOT NOFORTRAN)
@@ -79,10 +80,11 @@ endif ()
 set(GETARCH_DIR "${PROJECT_BINARY_DIR}/getarch_build")
 set(GETARCH_BIN "getarch${CMAKE_EXECUTABLE_SUFFIX}")
 file(MAKE_DIRECTORY ${GETARCH_DIR})
+configure_file(${TARGET_CONF_TEMP} ${GETARCH_DIR}/${TARGET_CONF} COPYONLY)
 if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
   try_compile(GETARCH_RESULT ${GETARCH_DIR}
     SOURCES ${GETARCH_SRC}
-    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} -I${PROJECT_SOURCE_DIR}
+    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} -I${GETARCH_DIR} -I${PROJECT_SOURCE_DIR} -I${PROJECT_BINARY_DIR}
     OUTPUT_VARIABLE GETARCH_LOG
     COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH_BIN}
   )
@@ -100,16 +102,17 @@ execute_process(COMMAND ${PROJECT_BINARY_DIR}/${GETARCH_BIN} 1 OUTPUT_VARIABLE G
 message(STATUS "GETARCH results:\n${GETARCH_MAKE_OUT}")
 
 # append config data from getarch to the TARGET file and read in CMake vars
-file(APPEND ${TARGET_CONF} ${GETARCH_CONF_OUT})
+file(APPEND ${TARGET_CONF_TEMP} ${GETARCH_CONF_OUT})
 ParseGetArchVars(${GETARCH_MAKE_OUT})
 
 set(GETARCH2_DIR "${PROJECT_BINARY_DIR}/getarch2_build")
 set(GETARCH2_BIN "getarch_2nd${CMAKE_EXECUTABLE_SUFFIX}")
 file(MAKE_DIRECTORY ${GETARCH2_DIR})
+configure_file(${TARGET_CONF_TEMP} ${GETARCH2_DIR}/${TARGET_CONF} COPYONLY)
 if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
   try_compile(GETARCH2_RESULT ${GETARCH2_DIR}
     SOURCES ${PROJECT_SOURCE_DIR}/getarch_2nd.c
-    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GETARCH2_FLAGS} -I${PROJECT_SOURCE_DIR}
+    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GETARCH2_FLAGS} -I${GETARCH2_DIR} -I${PROJECT_SOURCE_DIR} -I${PROJECT_BINARY_DIR}
     OUTPUT_VARIABLE GETARCH2_LOG
     COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH2_BIN}
   )
@@ -124,7 +127,8 @@ execute_process(COMMAND ${PROJECT_BINARY_DIR}/${GETARCH2_BIN} 0 OUTPUT_VARIABLE
 execute_process(COMMAND ${PROJECT_BINARY_DIR}/${GETARCH2_BIN} 1 OUTPUT_VARIABLE GETARCH2_CONF_OUT)
 
 # append config data from getarch_2nd to the TARGET file and read in CMake vars
-file(APPEND ${TARGET_CONF} ${GETARCH2_CONF_OUT})
+file(APPEND ${TARGET_CONF_TEMP} ${GETARCH2_CONF_OUT})
+configure_file(${TARGET_CONF_TEMP} ${PROJECT_BINARY_DIR}/${TARGET_CONF} COPYONLY)
 ParseGetArchVars(${GETARCH2_MAKE_OUT})
 
 # compile get_config_h
@@ -144,4 +148,4 @@ if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
   if (NOT ${GEN_CONFIG_H_RESULT})
     MESSAGE(FATAL_ERROR "Compiling gen_config_h failed ${GEN_CONFIG_H_LOG}")
   endif ()
-endif ()
+endif ()

cmake/utils.cmake

@@ -234,7 +234,9 @@ function(GenerateNamedObjects sources_in)
 
       string(REPLACE ";" "\n#define " define_source "${obj_defines}")
       string(REPLACE "=" " " define_source "${define_source}")
-      file(WRITE ${new_source_file} "#define ${define_source}\n#include \"${old_source_file}\"")
+      file(WRITE ${new_source_file}.tmp "#define ${define_source}\n#include \"${old_source_file}\"")
+      configure_file(${new_source_file}.tmp ${new_source_file} COPYONLY)
+      file(REMOVE ${new_source_file}.tmp)
       list(APPEND SRC_LIST_OUT ${new_source_file})
 
     endforeach ()

ctest/CMakeLists.txt

@@ -1,4 +1,5 @@
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 enable_language(Fortran)
 

driver/level2/CMakeLists.txt

@@ -1,5 +1,6 @@
 
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 # sources that need to be compiled twice, once with no flags and once with LOWER
 set(UL_SOURCES

driver/level3/CMakeLists.txt

@@ -1,4 +1,5 @@
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 # N.B. In the original makefile there was a BLOCKS define used in the compilation of these files but I don't see any evidence of it being set anywhere. -hpa
 

driver/others/CMakeLists.txt

@@ -1,4 +1,5 @@
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 if (${CORE} STREQUAL "PPC440")
   set(MEMORY memory_qalloc.c)

driver/others/memory.c

@@ -155,7 +155,6 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #ifdef DYNAMIC_ARCH
 gotoblas_t *gotoblas = NULL;
 #endif
-
 extern void openblas_warning(int verbose, const char * msg);
 
 #ifndef SMP
@@ -187,25 +186,24 @@ int i,n;
 
 #if !defined(__GLIBC_PREREQ)
    return nums;
-#endif
-#if !__GLIBC_PREREQ(2, 3)
+#else
+ #if !__GLIBC_PREREQ(2, 3)
    return nums;
-#endif
+ #endif
 
-#if !__GLIBC_PREREQ(2, 7)
+ #if !__GLIBC_PREREQ(2, 7)
   ret = sched_getaffinity(0,sizeof(cpu_set_t), cpusetp);
   if (ret!=0) return nums;
   n=0;
-#if !__GLIBC_PREREQ(2, 6)
+  #if !__GLIBC_PREREQ(2, 6)
   for (i=0;i<nums;i++)
      if (CPU_ISSET(i,cpusetp)) n++;
   nums=n;
-#else
+  #else
   nums = CPU_COUNT(sizeof(cpu_set_t),cpusetp);
-#endif
+  #endif
   return nums;
-#endif
-
+ #else
   cpusetp = CPU_ALLOC(nums);
   if (cpusetp == NULL) return nums;
   size = CPU_ALLOC_SIZE(nums);
@@ -214,6 +212,8 @@ int i,n;
   nums = CPU_COUNT_S(size,cpusetp);
   CPU_FREE(cpusetp);
   return nums;
+ #endif
+#endif
 }
 #endif
 #endif

interface/CMakeLists.txt

@@ -1,5 +1,6 @@
 
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 
 set(BLAS1_SOURCES

kernel/CMakeLists.txt

@@ -1,5 +1,6 @@
 
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 include("${PROJECT_SOURCE_DIR}/cmake/kernel.cmake")
 
 # Makefile

kernel/arm64/casum_thunderx2t99.c

@@ -147,57 +147,57 @@ static FLOAT casum_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	fmov	s6, "REG0"			\n"
 	"	fmov	s7, "REG0"			\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lasum_kernel_S_BEGIN		\n"
+	"	bne	5f //asum_kernel_S_BEGIN	\n"
 
-	".Lasum_kernel_F_BEGIN:				\n"
+	"1: //asum_kernel_F_BEGIN:			\n"
 	"	asr	"J", "N", #5			\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lasum_kernel_F1		\n"
+	"	beq	3f //asum_kernel_F1		\n"
 
-	".Lasum_kernel_F32:				\n"
+	"2: //asum_kernel_F32:				\n"
 	"	"KERNEL_F32"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F32		\n"
+	"	bne	2b //asum_kernel_F32		\n"
 	"	"KERNEL_F32_FINALIZE"			\n"
 
-	".Lasum_kernel_F1:				\n"
+	"3: //asum_kernel_F1:				\n"
 	"	ands	"J", "N", #31			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_F10:				\n"
+	"4: //asum_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs    "J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F10		\n"
-	"	b	.Lasum_kernel_L999		\n"
+	"	bne	4b //asum_kernel_F10		\n"
+	"	b	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S_BEGIN:				\n"
+	"5: //asum_kernel_S_BEGIN:			\n"
 	"	"INIT_S"				\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lasum_kernel_S1		\n"
+	"	ble	7f //asum_kernel_S1		\n"
 
-	".Lasum_kernel_S4:				\n"
+	"6: //asum_kernel_S4:				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S4		\n"
+	"	bne	6b //asum_kernel_S4		\n"
 
-	".Lasum_kernel_S1:				\n"
+	"7: //asum_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S10:				\n"
+	"8: //asum_kernel_S10:				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S10		\n"
+	"	bne	8b //asum_kernel_S10		\n"
 
-	".Lasum_kernel_L999:				\n"
+	"9: //asum_kernel_L999:				\n"
 	"	fmov	%[ASUM_], "SUMFD"		\n"
 
 	: [ASUM_] "=r" (asum)		//%0

kernel/arm64/copy_thunderx2t99.c

@@ -90,62 +90,62 @@ static int do_copy(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_
 	"	mov	"Y", %[Y_]			\n"
 	"	mov	"INC_Y", %[INCY_]		\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lcopy_kernel_L999		\n"
+	"	ble	8f //copy_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lcopy_kernel_S_BEGIN		\n"
+	"	bne	4f //copy_kernel_S_BEGIN	\n"
 	"	cmp	"INC_Y", #1			\n"
-	"	bne	.Lcopy_kernel_S_BEGIN		\n"
+	"	bne	4f //copy_kernel_S_BEGIN	\n"
 
-	".Lcopy_kernel_F_BEGIN:				\n"
+	"// .Lcopy_kernel_F_BEGIN:			\n"
 	"	"INIT"					\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"	\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lcopy_kernel_F1		\n"
+	"	beq	2f //copy_kernel_F1		\n"
 	"	.align 5				\n"
 
-	".Lcopy_kernel_F:				\n"
+	"1: //copy_kernel_F:				\n"
 	"	"KERNEL_F"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lcopy_kernel_F			\n"
+	"	bne	1b //copy_kernel_F		\n"
 
-	".Lcopy_kernel_F1:				\n"
+	"2: //copy_kernel_F1:				\n"
 #if defined(COMPLEX) && defined(DOUBLE)
-	"	b	.Lcopy_kernel_L999		\n"
+	"	b	8f //copy_kernel_L999		\n"
 #else
 	"	ands	"J", "N", #"N_REM_MASK"		\n"
-	"	ble	.Lcopy_kernel_L999		\n"
+	"	ble	8f //copy_kernel_L999		\n"
 #endif
 
-	".Lcopy_kernel_F10:				\n"
+	"3: //copy_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lcopy_kernel_F10		\n"
-	"	b	.Lcopy_kernel_L999		\n"
+	"	bne	3b //copy_kernel_F10		\n"
+	"	b	8f //copy_kernel_L999		\n"
 
-	".Lcopy_kernel_S_BEGIN:				\n"
+	"4: //copy_kernel_S_BEGIN:			\n"
 	"	"INIT"					\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lcopy_kernel_S1		\n"
+	"	ble	6f //copy_kernel_S1		\n"
 
-	".Lcopy_kernel_S4:				\n"
+	"5: //copy_kernel_S4:				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lcopy_kernel_S4		\n"
+	"	bne	5b //copy_kernel_S4		\n"
 
-	".Lcopy_kernel_S1:				\n"
+	"6: //copy_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lcopy_kernel_L999		\n"
+	"	ble	8f //copy_kernel_L999		\n"
 
-	".Lcopy_kernel_S10:				\n"
+	"7: //copy_kernel_S10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lcopy_kernel_S10		\n"
+	"	bne	7b //copy_kernel_S10		\n"
 
-	".Lcopy_kernel_L999:				\n"
+	"8: //copy_kernel_L999:				\n"
 
 	:
 	: [N_]    "r"  (n),		//%1

kernel/arm64/dasum_thunderx2t99.c

@@ -141,58 +141,58 @@ static FLOAT dasum_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	fmov	d6, "REG0"			\n"
 	"	fmov	d7, "REG0"			\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lasum_kernel_S_BEGIN		\n"
+	"	bne	5f //asum_kernel_S_BEGIN	\n"
 
-	".Lasum_kernel_F_BEGIN:				\n"
+	"1: //asum_kernel_F_BEGIN:			\n"
 	"	asr	"J", "N", #5			\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lasum_kernel_F1		\n"
+	"	beq	3f //asum_kernel_F1		\n"
 
 	".align 5					\n"
-	".Lasum_kernel_F32:				\n"
+	"2: //asum_kernel_F32:				\n"
 	"	"KERNEL_F32"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F32		\n"
+	"	bne	2b //asum_kernel_F32		\n"
 	"	"KERNEL_F32_FINALIZE"			\n"
 
-	".Lasum_kernel_F1:				\n"
+	"3: //asum_kernel_F1:				\n"
 	"	ands	"J", "N", #31			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_F10:				\n"
+	"4: //asum_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs    "J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F10		\n"
-	"	b	.Lasum_kernel_L999		\n"
+	"	bne	4b //asum_kernel_F10		\n"
+	"	b	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S_BEGIN:				\n"
+	"5: //asum_kernel_S_BEGIN:			\n"
 	"	"INIT_S"				\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lasum_kernel_S1		\n"
+	"	ble	7f //asum_kernel_S1		\n"
 
-	".Lasum_kernel_S4:				\n"
+	"6: //asum_kernel_S4:				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S4		\n"
+	"	bne	6b //asum_kernel_S4		\n"
 
-	".Lasum_kernel_S1:				\n"
+	"7: //asum_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S10:				\n"
+	"8: //asum_kernel_S10:				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S10		\n"
+	"	bne	8b //asum_kernel_S10		\n"
 
-	".Lasum_kernel_L999:				\n"
+	"9: //asum_kernel_L999:				\n"
 	"	fmov	%[ASUM_], "SUMF"		\n"
 
 	: [ASUM_] "=r" (asum)		//%0

kernel/arm64/dot_thunderx2t99.c

@@ -291,61 +291,61 @@ static RETURN_TYPE dot_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, B
 	"	fmov	d6, xzr				\n"
 	"	fmov	d7, xzr				\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Ldot_kernel_S_BEGIN		\n"
+	"	bne	5f //dot_kernel_S_BEGIN		\n"
 	"	cmp	"INC_Y", #1			\n"
-	"	bne	.Ldot_kernel_S_BEGIN		\n"
+	"	bne	5f //dot_kernel_S_BEGIN		\n"
 
-	".Ldot_kernel_F_BEGIN:				\n"
+	"1: //dot_kernel_F_BEGIN:			\n"
 	"	lsl	"INC_X", "INC_X", "INC_SHIFT"	\n"
 	"	lsl	"INC_Y", "INC_Y", "INC_SHIFT"	\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"	\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Ldot_kernel_F1			\n"
+	"	beq	3f //dot_kernel_F1		\n"
 
 	"	.align 5				\n"
-	".Ldot_kernel_F:				\n"
+	"2: //dot_kernel_F:				\n"
 	"	"KERNEL_F"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_F			\n"
+	"	bne	2b //dot_kernel_F		\n"
 	"	"KERNEL_F_FINALIZE"			\n"
 
-	".Ldot_kernel_F1:				\n"
+	"3: //dot_kernel_F1:				\n"
 	"	ands	"J", "N", #"N_REM_MASK"		\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_F10:				\n"
+	"4: //dot_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_F10		\n"
-	"	b	.Ldot_kernel_L999		\n"
+	"	bne	4b //dot_kernel_F10		\n"
+	"	b	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_S_BEGIN:				\n"
+	"5: //dot_kernel_S_BEGIN:			\n"
 	"	lsl	"INC_X", "INC_X", "INC_SHIFT"	\n"
 	"	lsl	"INC_Y", "INC_Y", "INC_SHIFT"	\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Ldot_kernel_S1			\n"
+	"	ble	7f //dot_kernel_S1		\n"
 
-	".Ldot_kernel_S4:				\n"
+	"6: //dot_kernel_S4:				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_S4			\n"
+	"	bne	6b //dot_kernel_S4		\n"
 
-	".Ldot_kernel_S1:				\n"
+	"7: //dot_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_S10:				\n"
+	"8: //dot_kernel_S10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_S10		\n"
+	"	bne	8b //dot_kernel_S10		\n"
 
-	".Ldot_kernel_L999:				\n"
+	"9: //dot_kernel_L999:				\n"
 	"	str	"DOTF", [%[DOT_]]		\n"
 
 	:

kernel/arm64/dznrm2_thunderx2t99.c

@@ -74,33 +74,33 @@ static void nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x,
 	"	fmov	"SCALE", xzr				\n"
 	"	fmov	"SSQ", #1.0				\n"
 	"	cmp	"N", xzr				\n"
-	"	ble	.Lnrm2_kernel_L999			\n"
+	"	ble	9f //nrm2_kernel_L999			\n"
 	"	cmp	"INC_X", xzr				\n"
-	"	ble	.Lnrm2_kernel_L999			\n"
+	"	ble	9f //nrm2_kernel_L999			\n"
 
-	".Lnrm2_kernel_F_BEGIN:					\n"
+	"1: //nrm2_kernel_F_BEGIN:				\n"
 	"	fmov	"REGZERO", xzr				\n"
 	"	fmov	"REGONE", #1.0				\n"
 	"	lsl	"INC_X", "INC_X", #"INC_SHIFT"		\n"
 	"	mov	"J", "N"				\n"
 	"	cmp	"J", xzr				\n"
-	"	beq	.Lnrm2_kernel_L999			\n"
+	"	beq	9f //nrm2_kernel_L999			\n"
 
-	".Lnrm2_kernel_F_ZERO_SKIP:				\n"
+	"2: //nrm2_kernel_F_ZERO_SKIP:				\n"
 	"	ldr	d4, ["X"]				\n"
 	"	fcmp	d4, "REGZERO"				\n"
-	"	bne	.Lnrm2_kernel_F_INIT			\n"
+	"	bne	3f //nrm2_kernel_F_INIT			\n"
 #if defined(COMPLEX)
 	"	ldr	d4, ["X", #8]				\n"
 	"	fcmp	d4, "REGZERO"				\n"
-	"	bne	.Lnrm2_kernel_F_INIT_I			\n"
+	"	bne	4f //nrm2_kernel_F_INIT_I		\n"
 #endif
 	"	add	"X", "X", "INC_X"			\n"
 	"	subs	"J", "J", #1				\n"
-	"	beq	.Lnrm2_kernel_L999			\n"
-	"	b	.Lnrm2_kernel_F_ZERO_SKIP		\n"
+	"	beq	9f //nrm2_kernel_L999			\n"
+	"	b	2b //nrm2_kernel_F_ZERO_SKIP		\n"
 
-	".Lnrm2_kernel_F_INIT:					\n"
+	"3: //nrm2_kernel_F_INIT:				\n"
 	"	ldr	d4, ["X"]				\n"
 	"	fabs	d4, d4					\n"
 	"	fmax	"CUR_MAX", "SCALE", d4			\n"
@@ -112,7 +112,7 @@ static void nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x,
 	"	fadd	"SSQ", "SSQ", d4			\n"
 	"	fmov	"SCALE", "CUR_MAX"			\n"
 #if defined(COMPLEX)
-	".Lnrm2_kernel_F_INIT_I:				\n"
+	"4: //nrm2_kernel_F_INIT_I:				\n"
 	"	ldr	d3, ["X", #8]				\n"
 	"	fabs	d3, d3					\n"
 	"	fmax	"CUR_MAX", "SCALE", d3			\n"
@@ -126,16 +126,16 @@ static void nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x,
 #endif
 	"	add	"X", "X", "INC_X"			\n"
 	"	subs	"J", "J", #1				\n"
-	"	beq	.Lnrm2_kernel_L999			\n"
+	"	beq	9f //nrm2_kernel_L999			\n"
 
-	".Lnrm2_kernel_F_START:					\n"
+	"5: //nrm2_kernel_F_START:				\n"
 	"	cmp	"INC_X", #"SZ"				\n"
-	"	bne	.Lnrm2_kernel_F1			\n"
+	"	bne	8f //nrm2_kernel_F1			\n"
 	"	asr	"K", "J", #4				\n"
 	"	cmp	"K", xzr				\n"
-	"	beq	.Lnrm2_kernel_F1			\n"
+	"	beq	8f //nrm2_kernel_F1			\n"
 
-	".Lnrm2_kernel_F:					\n"
+	"6: //nrm2_kernel_F:					\n"
 	"	ldp	q16, q17,  ["X"]			\n"
 	"	ldp	q18, q19,  ["X", #32]			\n"
 	"	ldp	q20, q21,  ["X", #64]			\n"
@@ -255,13 +255,13 @@ static void nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x,
 	"	fmov	"SCALE", "CUR_MAX"			\n"
 #endif
 	"	subs	"K", "K", #1				\n"
-	"	bne	.Lnrm2_kernel_F				\n"
+	"	bne	6b //nrm2_kernel_F			\n"
 
-	".Lnrm2_kernel_F_DONE:					\n"
+	"7: //nrm2_kernel_F_DONE:				\n"
 	"	ands	"J", "J", #15				\n"
-	"	beq	.Lnrm2_kernel_L999			\n"
+	"	beq	9f //nrm2_kernel_L999			\n"
 
-	".Lnrm2_kernel_F1:					\n"
+	"8: //nrm2_kernel_F1:					\n"
 	"	ldr	d4, ["X"]				\n"
 	"	fabs	d4, d4					\n"
 	"	fmax	"CUR_MAX", "SCALE", d4			\n"
@@ -286,9 +286,9 @@ static void nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x,
 #endif
 	"	add	"X", "X", "INC_X"			\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Lnrm2_kernel_F1			\n"
+	"	bne	8b //nrm2_kernel_F1			\n"
 
-	".Lnrm2_kernel_L999:					\n"
+	"9: //nrm2_kernel_L999:					\n"
 	"	str	"SSQ", [%[SSQ_]]			\n"
 	"	str	"SCALE", [%[SCALE_]]			\n"
 

kernel/arm64/iamax_thunderx2t99.c

@@ -208,7 +208,7 @@ extern int blas_level1_thread_with_return_value(int mode, BLASLONG m, BLASLONG n
 #endif
 
 
-static BLASLONG iamax_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
+static BLASLONG __attribute__((noinline)) iamax_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 {
 	BLASLONG index = 0;
 
@@ -220,72 +220,72 @@ static BLASLONG iamax_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	mov	"INC_X", %[INCX_]			\n"
 
 	"	cmp	"N", xzr				\n"
-	"	ble	.Liamax_kernel_zero			\n"
+	"	ble	10f //iamax_kernel_zero			\n"
 	"	cmp	"INC_X", xzr				\n"
-	"	ble	.Liamax_kernel_zero			\n"
+	"	ble	10f //iamax_kernel_zero			\n"
 	"	cmp	"INC_X", #1				\n"
-	"	bne	.Liamax_kernel_S_BEGIN			\n"
+	"	bne	5f //iamax_kernel_S_BEGIN		\n"
 	"	mov	x7, "X"					\n"
 
-	".Liamax_kernel_F_BEGIN:				\n"
+	"1: //iamax_kernel_F_BEGIN:				\n"
 	"	"INIT"						\n"
 	"	subs	"N", "N", #1				\n"
-	"	ble	.Liamax_kernel_L999			\n"
+	"	ble	9f //iamax_kernel_L999			\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"		\n"
 	"	cmp	"J", xzr				\n"
-	"	beq	.Liamax_kernel_F1			\n"
+	"	beq	3f //iamax_kernel_F1			\n"
 	"	add	"Z", "Z", #1				\n"
 
-	".Liamax_kernel_F:					\n"
+	"2: //iamax_kernel_F:					\n"
 	"	"KERNEL_F"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Liamax_kernel_F			\n"
+	"	bne	2b //iamax_kernel_F			\n"
 	"	"KERNEL_F_FINALIZE"				\n"
 	"	sub	"Z", "Z", #1				\n"
 
-	".Liamax_kernel_F1:					\n"
+	"3: //iamax_kernel_F1:					\n"
 	"	ands	"J", "N", #"N_REM_MASK"			\n"
-	"	ble	.Liamax_kernel_L999			\n"
+	"	ble	9f //iamax_kernel_L999			\n"
 
-	".Liamax_kernel_F10:					\n"
+	"4: //iamax_kernel_F10:					\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Liamax_kernel_F10			\n"
-	"	b	.Liamax_kernel_L999			\n"
+	"	bne	4b //iamax_kernel_F10			\n"
+	"	b	9f //iamax_kernel_L999			\n"
 
-	".Liamax_kernel_S_BEGIN:				\n"
+	"5: //iamax_kernel_S_BEGIN:				\n"
 	"	"INIT"						\n"
 	"	subs	"N", "N", #1				\n"
-	"	ble	.Liamax_kernel_L999			\n"
+	"	ble	9f //iamax_kernel_L999			\n"
 	"	asr	"J", "N", #2				\n"
 	"	cmp	"J", xzr				\n"
-	"	ble	.Liamax_kernel_S1			\n"
+	"	ble	7f //iamax_kernel_S1			\n"
 
-	".Liamax_kernel_S4:					\n"
+	"6: //iamax_kernel_S4:					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Liamax_kernel_S4			\n"
+	"	bne	6b //iamax_kernel_S4			\n"
 
-	".Liamax_kernel_S1:					\n"
+	"7: //iamax_kernel_S1:					\n"
 	"	ands	"J", "N", #3				\n"
-	"	ble	.Liamax_kernel_L999			\n"
+	"	ble	9f //iamax_kernel_L999			\n"
 
-	".Liamax_kernel_S10:					\n"
+	"8: //iamax_kernel_S10:					\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Liamax_kernel_S10			\n"
+	"	bne	8b //iamax_kernel_S10			\n"
 
-	".Liamax_kernel_L999:					\n"
+	"9: //iamax_kernel_L999:				\n"
 	"	mov	x0, "INDEX"				\n"
-	"	b	.Liamax_kernel_DONE			\n"
+	"	b	11f //iamax_kernel_DONE			\n"
 
-	".Liamax_kernel_zero:					\n"
+	"10: //iamax_kernel_zero:				\n"
 	"	mov	x0, xzr					\n"
 
-	".Liamax_kernel_DONE:					\n"
+	"11: //iamax_kernel_DONE:				\n"
 	"	mov	%[INDEX_], "INDEX"			\n"
 
 	: [INDEX_] "=r" (index)		//%0

kernel/arm64/izamax_thunderx2t99.c

@@ -229,72 +229,72 @@ static BLASLONG izamax_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	mov	"INC_X", %[INCX_]			\n"
 
 	"	cmp	"N", xzr				\n"
-	"	ble	.Lizamax_kernel_zero			\n"
+	"	ble	10f //izamax_kernel_zero		\n"
 	"	cmp	"INC_X", xzr				\n"
-	"	ble	.Lizamax_kernel_zero			\n"
+	"	ble	10f //izamax_kernel_zero		\n"
 	"	cmp	"INC_X", #1				\n"
-	"	bne	.Lizamax_kernel_S_BEGIN			\n"
+	"	bne	5f //izamax_kernel_S_BEGIN		\n"
 	"	mov	x7, "X"					\n"
 
-	".Lizamax_kernel_F_BEGIN:				\n"
+	"1: //izamax_kernel_F_BEGIN:				\n"
 	"	"INIT"						\n"
 	"	subs	"N", "N", #1				\n"
-	"	ble	.Lizamax_kernel_L999			\n"
+	"	ble	9f //izamax_kernel_L999			\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"		\n"
 	"	cmp	"J", xzr				\n"
-	"	beq	.Lizamax_kernel_F1			\n"
+	"	beq	3f //izamax_kernel_F1			\n"
 	"	add	"Z", "Z", #1				\n"
 
-	".Lizamax_kernel_F:					\n"
+	"2: //izamax_kernel_F:					\n"
 	"	"KERNEL_F"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Lizamax_kernel_F			\n"
+	"	bne	2b //izamax_kernel_F			\n"
 	"	"KERNEL_F_FINALIZE"				\n"
 	"	sub	"Z", "Z", #1				\n"
 
-	".Lizamax_kernel_F1:					\n"
+	"3: //izamax_kernel_F1:					\n"
 	"	ands	"J", "N", #"N_REM_MASK"			\n"
-	"	ble	.Lizamax_kernel_L999			\n"
+	"	ble	9f //izamax_kernel_L999			\n"
 
-	".Lizamax_kernel_F10:					\n"
+	"4: //izamax_kernel_F10:				\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Lizamax_kernel_F10			\n"
-	"	b	.Lizamax_kernel_L999			\n"
+	"	bne	4b //izamax_kernel_F10			\n"
+	"	b	9f //izamax_kernel_L999			\n"
 
-	".Lizamax_kernel_S_BEGIN:				\n"
+	"5: //izamax_kernel_S_BEGIN:				\n"
 	"	"INIT"						\n"
 	"	subs	"N", "N", #1				\n"
-	"	ble	.Lizamax_kernel_L999			\n"
+	"	ble	9f //izamax_kernel_L999			\n"
 	"	asr	"J", "N", #2				\n"
 	"	cmp	"J", xzr				\n"
-	"	ble	.Lizamax_kernel_S1			\n"
+	"	ble	7f //izamax_kernel_S1			\n"
 
-	".Lizamax_kernel_S4:					\n"
+	"6: //izamax_kernel_S4:					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Lizamax_kernel_S4			\n"
+	"	bne	6b //izamax_kernel_S4			\n"
 
-	".Lizamax_kernel_S1:					\n"
+	"7: //izamax_kernel_S1:					\n"
 	"	ands	"J", "N", #3				\n"
-	"	ble	.Lizamax_kernel_L999			\n"
+	"	ble	9f //izamax_kernel_L999			\n"
 
-	".Lizamax_kernel_S10:					\n"
+	"8: //izamax_kernel_S10:				\n"
 	"	"KERNEL_S1"					\n"
 	"	subs	"J", "J", #1				\n"
-	"	bne	.Lizamax_kernel_S10			\n"
+	"	bne	8b //izamax_kernel_S10			\n"
 
-	".Lizamax_kernel_L999:					\n"
+	"9: //izamax_kernel_L999:				\n"
 	"	mov	x0, "INDEX"				\n"
-	"	b	.Lizamax_kernel_DONE			\n"
+	"	b	11f //izamax_kernel_DONE		\n"
 
-	".Lizamax_kernel_zero:					\n"
+	"10: //izamax_kernel_zero:				\n"
 	"	mov	x0, xzr					\n"
 
-	".Lizamax_kernel_DONE:					\n"
+	"11: //izamax_kernel_DONE:				\n"
 	"	mov	%[INDEX_], "INDEX"			\n"
 
 	: [INDEX_] "=r" (index)		//%0

kernel/arm64/sasum_thunderx2t99.c

@@ -143,58 +143,58 @@ static FLOAT sasum_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	fmov	s6, "REG0"			\n"
 	"	fmov	s7, "REG0"			\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lasum_kernel_S_BEGIN		\n"
+	"	bne	5f //asum_kernel_S_BEGIN	\n"
 
-	".Lasum_kernel_F_BEGIN:				\n"
+	"1: //asum_kernel_F_BEGIN:			\n"
 	"	asr	"J", "N", #6			\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lasum_kernel_F1		\n"
+	"	beq	3f //asum_kernel_F1		\n"
 
 	".align 5					\n"
-	".Lasum_kernel_F64:				\n"
+	"2: //asum_kernel_F64:				\n"
 	"	"KERNEL_F64"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F64		\n"
+	"	bne	2b //asum_kernel_F64		\n"
 	"	"KERNEL_F64_FINALIZE"			\n"
 
-	".Lasum_kernel_F1:				\n"
+	"3: //asum_kernel_F1:				\n"
 	"	ands	"J", "N", #63			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_F10:				\n"
+	"4: //asum_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs    "J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F10		\n"
-	"	b	.Lasum_kernel_L999		\n"
+	"	bne	4b //asum_kernel_F10		\n"
+	"	b	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S_BEGIN:				\n"
+	"5: //asum_kernel_S_BEGIN:			\n"
 	"	"INIT_S"				\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lasum_kernel_S1		\n"
+	"	ble	7f //asum_kernel_S1		\n"
 
-	".Lasum_kernel_S4:				\n"
+	"6: //asum_kernel_S4:				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S4		\n"
+	"	bne	6b //asum_kernel_S4		\n"
 
-	".Lasum_kernel_S1:				\n"
+	"7: //asum_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S10:				\n"
+	"8: //asum_kernel_S10:				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S10		\n"
+	"	bne	8b //asum_kernel_S10		\n"
 
-	".Lasum_kernel_L999:				\n"
+	"9: //asum_kernel_L999:				\n"
 	"	fmov	%[ASUM_], "SUMFD"		\n"
 
 	: [ASUM_] "=r" (asum)		//%0

kernel/arm64/scnrm2_thunderx2t99.c

@@ -227,58 +227,58 @@ static double nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	fmov	d6, xzr				\n"
 	"	fmov	d7, xzr				\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lnrm2_kernel_L999		\n"
+	"	ble	9f //nrm2_kernel_L999		\n"
 	"	cmp	"INC_X", xzr			\n"
-	"	ble	.Lnrm2_kernel_L999		\n"
+	"	ble	9f //nrm2_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lnrm2_kernel_S_BEGIN		\n"
+	"	bne	5f //nrm2_kernel_S_BEGIN	\n"
 
-	".Lnrm2_kernel_F_BEGIN:				\n"
+	"1: //nrm2_kernel_F_BEGIN:			\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"	\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lnrm2_kernel_S_BEGIN		\n"
+	"	beq	5f //nrm2_kernel_S_BEGIN	\n"
 
 	"	.align 5				\n"
-	".Lnrm2_kernel_F:				\n"
+	"2: //nrm2_kernel_F:				\n"
 	"	"KERNEL_F"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lnrm2_kernel_F			\n"
+	"	bne	2b //nrm2_kernel_F		\n"
 	"	"KERNEL_F_FINALIZE"			\n"
 
-	".Lnrm2_kernel_F1:				\n"
+	"3: //nrm2_kernel_F1:				\n"
 	"	ands	"J", "N", #"N_REM_MASK"		\n"
-	"	ble	.Lnrm2_kernel_L999		\n"
+	"	ble	9f //nrm2_kernel_L999		\n"
 
-	".Lnrm2_kernel_F10:				\n"
+	"4: //nrm2_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lnrm2_kernel_F10		\n"
-	"	b	.Lnrm2_kernel_L999		\n"
+	"	bne	4b //nrm2_kernel_F10		\n"
+	"	b	9f //nrm2_kernel_L999		\n"
 
-	".Lnrm2_kernel_S_BEGIN:				\n"
+	"5: //nrm2_kernel_S_BEGIN:			\n"
 	"	lsl	"INC_X", "INC_X", #"INC_SHIFT"	\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lnrm2_kernel_S1		\n"
+	"	ble	7f //nrm2_kernel_S1		\n"
 
-	".Lnrm2_kernel_S4:				\n"
+	"6: //nrm2_kernel_S4:				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lnrm2_kernel_S4		\n"
+	"	bne	6b //nrm2_kernel_S4		\n"
 
-	".Lnrm2_kernel_S1:				\n"
+	"7: //nrm2_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lnrm2_kernel_L999		\n"
+	"	ble	9f //nrm2_kernel_L999		\n"
 
-	".Lnrm2_kernel_S10:				\n"
+	"8: //nrm2_kernel_S10:				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lnrm2_kernel_S10		\n"
+	"	bne	8b //nrm2_kernel_S10		\n"
 
-	".Lnrm2_kernel_L999:				\n"
+	"9: //nrm2_kernel_L999:				\n"
 	"	"KERNEL_FINALIZE"			\n"
 	"	fmov	%[RET_], "SSQD"			\n"
 

kernel/arm64/zasum_thunderx2t99.c

@@ -143,58 +143,58 @@ static FLOAT zasum_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
 	"	fmov	d6, "REG0"			\n"
 	"	fmov	d7, "REG0"			\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", xzr			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Lasum_kernel_S_BEGIN		\n"
+	"	bne	5f //asum_kernel_S_BEGIN	\n"
 
-	".Lasum_kernel_F_BEGIN:				\n"
+	"1: //asum_kernel_F_BEGIN:			\n"
 	"	asr	"J", "N", #4			\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Lasum_kernel_F1		\n"
+	"	beq	3f //asum_kernel_F1		\n"
 
 	".align 5					\n"
-	".Lasum_kernel_F16:				\n"
+	"2: //asum_kernel_F16:				\n"
 	"	"KERNEL_F16"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F16		\n"
+	"	bne	2b //asum_kernel_F16		\n"
 	"	"KERNEL_F16_FINALIZE"			\n"
 
-	".Lasum_kernel_F1:				\n"
+	"3: //asum_kernel_F1:				\n"
 	"	ands	"J", "N", #15			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_F10:				\n"
+	"4: //asum_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs    "J", "J", #1			\n"
-	"	bne	.Lasum_kernel_F10		\n"
-	"	b	.Lasum_kernel_L999		\n"
+	"	bne	4b //asum_kernel_F10		\n"
+	"	b	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S_BEGIN:				\n"
+	"5: //asum_kernel_S_BEGIN:			\n"
 	"	"INIT_S"				\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Lasum_kernel_S1		\n"
+	"	ble	7f //asum_kernel_S1		\n"
 
-	".Lasum_kernel_S4:				\n"
+	"6: //asum_kernel_S4:				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S4		\n"
+	"	bne	6b //asum_kernel_S4		\n"
 
-	".Lasum_kernel_S1:				\n"
+	"7: //asum_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Lasum_kernel_L999		\n"
+	"	ble	9f //asum_kernel_L999		\n"
 
-	".Lasum_kernel_S10:				\n"
+	"8: //asum_kernel_S10:				\n"
 	"	"KERNEL_S1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Lasum_kernel_S10		\n"
+	"	bne	8b //asum_kernel_S10		\n"
 
-	".Lasum_kernel_L999:				\n"
+	"9: //asum_kernel_L999:				\n"
 	"	fmov	%[ASUM_], "SUMF"		\n"
 
 	: [ASUM_] "=r" (asum)		//%0

kernel/arm64/zdot_thunderx2t99.c

@@ -218,61 +218,61 @@ static void zdot_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLON
 	"	fmov	d6, xzr				\n"
 	"	fmov	d7, xzr				\n"
 	"	cmp	"N", xzr			\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 	"	cmp	"INC_X", #1			\n"
-	"	bne	.Ldot_kernel_S_BEGIN		\n"
+	"	bne	5f //dot_kernel_S_BEGIN		\n"
 	"	cmp	"INC_Y", #1			\n"
-	"	bne	.Ldot_kernel_S_BEGIN		\n"
+	"	bne	5f //dot_kernel_S_BEGIN		\n"
 
-	".Ldot_kernel_F_BEGIN:				\n"
+	"1: //dot_kernel_F_BEGIN:			\n"
 	"	lsl	"INC_X", "INC_X", "INC_SHIFT"	\n"
 	"	lsl	"INC_Y", "INC_Y", "INC_SHIFT"	\n"
 	"	asr	"J", "N", #"N_DIV_SHIFT"	\n"
 	"	cmp	"J", xzr			\n"
-	"	beq	.Ldot_kernel_F1			\n"
+	"	beq	3f //dot_kernel_F1		\n"
 
 	"	.align 5				\n"
-	".Ldot_kernel_F:				\n"
+	"2: //dot_kernel_F:				\n"
 	"	"KERNEL_F"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_F			\n"
+	"	bne	2b //dot_kernel_F		\n"
 	"	"KERNEL_F_FINALIZE"			\n"
 
-	".Ldot_kernel_F1:				\n"
+	"3: //dot_kernel_F1:				\n"
 	"	ands	"J", "N", #"N_REM_MASK"		\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_F10:				\n"
+	"4: //dot_kernel_F10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_F10		\n"
-	"	b	.Ldot_kernel_L999		\n"
+	"	bne	4b //dot_kernel_F10		\n"
+	"	b	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_S_BEGIN:				\n"
+	"5: //dot_kernel_S_BEGIN:			\n"
 	"	lsl	"INC_X", "INC_X", "INC_SHIFT"	\n"
 	"	lsl	"INC_Y", "INC_Y", "INC_SHIFT"	\n"
 	"	asr	"J", "N", #2			\n"
 	"	cmp	"J", xzr			\n"
-	"	ble	.Ldot_kernel_S1			\n"
+	"	ble	7f //dot_kernel_S1		\n"
 
-	".Ldot_kernel_S4:				\n"
+	"6: //dot_kernel_S4:				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_S4			\n"
+	"	bne	6b //dot_kernel_S4		\n"
 
-	".Ldot_kernel_S1:				\n"
+	"7: //dot_kernel_S1:				\n"
 	"	ands	"J", "N", #3			\n"
-	"	ble	.Ldot_kernel_L999		\n"
+	"	ble	9f //dot_kernel_L999		\n"
 
-	".Ldot_kernel_S10:				\n"
+	"8: //dot_kernel_S10:				\n"
 	"	"KERNEL_F1"				\n"
 	"	subs	"J", "J", #1			\n"
-	"	bne	.Ldot_kernel_S10		\n"
+	"	bne	8b //dot_kernel_S10		\n"
 
-	".Ldot_kernel_L999:				\n"
+	"9: //dot_kernel_L999:				\n"
 	"	str	"DOTF", [%[DOTR_]]		\n"
 	"	str	"DOTI", [%[DOTI_]]		\n"
 

lapack/CMakeLists.txt

@@ -1,5 +1,6 @@
 
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 
 set(LAPACK_SOURCES

test/CMakeLists.txt

@@ -1,4 +1,5 @@
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 enable_language(Fortran)
 
@@ -35,4 +36,4 @@ add_test(NAME "${float_type}blas2"
   COMMAND sh "${CMAKE_CURRENT_BINARY_DIR}/test_helper.sh" "${CMAKE_CURRENT_BINARY_DIR}/${float_type}blat2" "${PROJECT_SOURCE_DIR}/test/${float_type}blat2.dat" ${float_type_upper}BLAT2.SUMM)
 add_test(NAME "${float_type}blas3" 
   COMMAND sh "${CMAKE_CURRENT_BINARY_DIR}/test_helper.sh" "${CMAKE_CURRENT_BINARY_DIR}/${float_type}blat3" "${PROJECT_SOURCE_DIR}/test/${float_type}blat3.dat" ${float_type_upper}BLAT3.SUMM)
-endforeach()
+endforeach()

utest/CMakeLists.txt

@@ -1,4 +1,5 @@
 include_directories(${PROJECT_SOURCE_DIR})
+include_directories(${PROJECT_BINARY_DIR})
 
 set(OpenBLAS_utest_src
   utest_main.c
@@ -39,4 +40,4 @@ add_custom_command(TARGET ${OpenBLAS_utest_bin}
 )
 endif()
 
-add_test(${OpenBLAS_utest_bin} ${CMAKE_CURRENT_BINARY_DIR}/${OpenBLAS_utest_bin})
+add_test(${OpenBLAS_utest_bin} ${CMAKE_CURRENT_BINARY_DIR}/${OpenBLAS_utest_bin})