data2/text/range/10001-15000/777773.txt

Ticket Name: Linux/TDA2: VISION SDK - Build tidl_model_import for Linux

Query Text:
Part Number: TDA2 Tool/software: Linux Is there a way to get tidl model input converter working on Linux OS? In PROCESSOR_SDK_VISION/ti_components/algorithms/REL.TIDL.01.01.01.00/modules/ti_dl/utils/tidlModelImport there is make file, but make does not work, its full of errors I can't fix. Can anyone help me with this? Regards.

Responses:
Hi, can you share the build log? Regards, Yordan

Hi, here it is: compiling tidl_import_common.cpp In file included from ti_dl.h:76:0, from tidl_import_common.cpp:88: ../../inc/itidl_ti.h:81:0: warning: "USE_IVISION" redefined #define USE_IVISION (0) ^ ../../inc/itidl_ti.h:78:0: note: this is the location of the previous definition #define USE_IVISION (1) ^ tidl_import_common.cpp: In function ‘int32_t TIDL_QuantizeUnsignedMax(uint8_t*, float*, int32_t, float, float, int32_t, int32_t, int32_t*)’: tidl_import_common.cpp:109:33: error: call of overloaded ‘abs(float)’ is ambiguous float absRange = abs(max - min); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ tidl_import_common.cpp: In function ‘int32_t TIDL_normalize(float, float, float)’: tidl_import_common.cpp:186:33: error: call of overloaded ‘abs(float)’ is ambiguous float absRange = abs(max - min); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ tidl_import_common.cpp: In function ‘bool TIDL_readProtoFromTextFile(const char*, google::protobuf::Message*)’: tidl_import_common.cpp:212:12: warning: NULL used in arithmetic [-Wpointer-arith] if(fd == NULL) ^ tidl_import_common.cpp: In function ‘bool TIDL_readProtoFromBinaryFile(const char*, google::protobuf::Message*)’: tidl_import_common.cpp:238:12: warning: NULL used in arithmetic [-Wpointer-arith] if(fd == NULL) ^ tidl_import_common.cpp: In function ‘int32_t TIDL_findSymQ(float, float)’: tidl_import_common.cpp:368:25: error: call of overloaded ‘abs(float&)’ is ambiguous float absMax = (abs(min) > abs(max)) ? abs(min) : abs(max); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ tidl_import_common.cpp:368:36: error: call of overloaded ‘abs(float&)’ is ambiguous float absMax = (abs(min) > abs(max)) ? abs(min) : abs(max); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ tidl_import_common.cpp:368:48: error: call of overloaded ‘abs(float&)’ is ambiguous float absMax = (abs(min) > abs(max)) ? abs(min) : abs(max); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ tidl_import_common.cpp:368:59: error: call of overloaded ‘abs(float&)’ is ambiguous float absMax = (abs(min) > abs(max)) ? abs(min) : abs(max); ^ In file included from /usr/local/include/google/protobuf/stubs/port.h:35:0, from /usr/local/include/google/protobuf/stubs/common.h:40, from /usr/local/include/google/protobuf/io/coded_stream.h:134, from tidl_import_common.cpp:63: /usr/include/stdlib.h:774:12: note: candidate: int abs(int) extern int abs (int __x) __THROW __attribute__ ((__const__)) __wur; ^ In file included from /usr/include/c++/5/bits/stl_algo.h:59:0, from /usr/include/c++/5/algorithm:62, from /usr/local/include/google/protobuf/stubs/shared_ptr.h:38, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl_lite.h:49, from /usr/local/include/google/protobuf/io/zero_copy_stream_impl.h:46, from tidl_import_common.cpp:64: /usr/include/c++/5/cstdlib:174:3: note: candidate: long long int std::abs(long long int) abs(long long __x) { return __builtin_llabs (__x); } ^ /usr/include/c++/5/cstdlib:166:3: note: candidate: long int std::abs(long int) abs(long __i) { return __builtin_labs(__i); } ^ /home/rtrk/PROCESSOR_SDK_VISION_03_04_00_00/ti_components/algorithms/REL.TIDL.01.01.01.00/makerules/rules.mk:450: recipe for target '/home/rtrk/PROCESSOR_SDK_VISION_03_04_00_00/ti_components/algorithms/REL.TIDL.01.01.01.00/out/PC/dsp/debug/modules/ti_dl/utils/tidlModelImport/tidl_import_common.obj' failed make: *** [/home/rtrk/PROCESSOR_SDK_VISION_03_04_00_00/ti_components/algorithms/REL.TIDL.01.01.01.00/out/PC/dsp/debug/modules/ti_dl/utils/tidlModelImport/tidl_import_common.obj] Error 1 Regards.

Hi Yordan, I'll add more details about my software: - Ubuntu 16.04 -PROCESSOR_SDK_VISION_03_04_00_00 And more details on what am I doing and what i did: -I'm trying to use tidlModelImport as a module in project that uses TIDL in VISION SDK. --Project is designed to work on Ubuntu 16.04 -First when i looked at build log I added "-std=gnu++11" to CFLAGS in `makefile` --and then changed include <math.h> to <cmath> in `tidl_import_common.cpp` -After that passed, got new compile arrors about protobuf library: --Forums say that libprotobuf-dev is outdated on official Ubuntu ---then I manualy build protobuf v3.2.0 ---updated makefile to use those libs -After that passed, got new errors where: --"compiling ../caffeImport/caffe.pb.cc" complained about: ---"/protobuf/src/google/protobuf/arena.h:xxx: undefined reference to ..." many things ---You will find whole output in `build_log` That confused me hence every file in "../tfImport/proto_cc/tensorflow/core/framework/" includes that `arena.h` file but compiling them passes with no errors. The clean build log output (where I did not modify anything) can be found in `build_log_clean`. Regards, Uros.makefile.doctidl_import_common.cppbuild_log.docbuild_log_clean.doc

Hi Uros, in your makefile can you add to LDFLAGS also "-lstdc++ -m32 -static" flags and give a try? They should be added if PLATFORM_BUILD == 'x86', but I am not sure it happens. Regards, Yordan

Hi Yordan, As you suggested i modified makefile: CFLAGS+=-DTIDL_IMPORT_TOOL
LDFLAGS= -L $(PROTOBUF_LIB_DIR)
LDFLAGS+= -lprotoc 
LDFLAGS+= -lprotobuf
LDFLAGS+= -lpthread

CFLAGS+= -m32
LDFLAGS+= -lstdc++ -m32 -static

# ifeq ($(PLATFORM_BUILD), x86)
# CFLAGS+= -m32
# LDFLAGS+= -lstdc++ -m32 -static
# else
# LDFLAGS+= -lstdc++ -lz
# endif Build still fails, but now when compiling caffeImport occurs I get this: compiling ../caffeImport/caffe.pb.cc /usr/bin/ld: skipping incompatible /home/rtrk/protobuf/src/.libs/libprotoc.a when searching for -lprotoc /usr/bin/ld: skipping incompatible //usr/local/lib/libprotoc.a when searching for -lprotoc /usr/bin/ld: cannot find -lprotoc /usr/bin/ld: skipping incompatible /home/rtrk/protobuf/src/.libs/libprotobuf.a when searching for -lprotobuf /usr/bin/ld: skipping incompatible //usr/local/lib/libprotobuf.a when searching for -lprotobuf /usr/bin/ld: cannot find -lprotobuf collect2: error: ld returned 1 exit status /home/rtrk/PROCESSOR_SDK_VISION_03_04_00_00/ti_components/algorithms/REL.TIDL.01.01.01.00/makerules/rules.mk:603: recipe for target 'outfile' failed make: *** [outfile] Error 1 You can find whole output in `build.log.doc` attachment. Regards, Uros. build.log.doc

Hi Uros, can you try to rebuild protobuf and protoc libraries with '-m32' flag added to their CFLAGS and LDFLAGS? Regards, Yordan

Hi, Yordan, I tried that, and everything is still the same. Regards, Uros

Hi Uros, I do not have more ideas and pinged TIDL experts to provide help. Regards, Yordan

Hi Yordan, I want to inform You that I have successfully build tidl_model_converer for Linux OS. Make flags for protobuf and protoc libs needed more additions: build=i686-pc-linux-gnu CFLAGS=-m32 -DNDEBUG CXXFLAGS=-m32 -DNDEBUG LDFLAGS=-m32 Like this library is sure to be 32-bit. Regards, Uros

Hi Uros, Thanks for updating the thread. Regards, Yordan