Build issue on mips64el

[eddelbuettel]

Season's Greetings! I should have filed this here earlier but we apparently have issues in the current and previous releases with the platform Debian identifies as mips64el.

From the package tracker to build logs (on the right) we get to the fail log on mips64el showing

running build_py
creating build/lib.linux-mips64-cpython-313/QuantLib
copying src/QuantLib/__init__.py -> build/lib.linux-mips64-cpython-313/QuantLib
copying src/QuantLib/QuantLib.py -> build/lib.linux-mips64-cpython-313/QuantLib
running build_ext
building 'QuantLib._QuantLib' extension
creating build/temp.linux-mips64-cpython-313/src/QuantLib
g++ -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR -fPIC -DPy_LIMITED_API=0x03080000 -DNDEBUG -I/usr/include -I/usr/include/python3.13 -c src/QuantLib/quantlib_wrap.cpp -o build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o -fopenmp -std=c++17 -Wno-unused -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR
src/QuantLib/quantlib_wrap.cpp: In function ‘std::vector<std::pair<double, double> >::value_type std_vector_Sl_std_pair_Sl_double_Sc_double_Sg__Sg__pop(std::vector<std::pair<double, double> >*)’:
src/QuantLib/quantlib_wrap.cpp:8999:168: note: the ABI for returning a value with C++17 empty bases but otherwise an aggregate with only one or two floating-point fields was changed in GCC 12.1
 8999 | SWIGINTERN std::vector< std::pair< double,double > >::value_type std_vector_Sl_std_pair_Sl_double_Sc_double_Sg__Sg__pop(std::vector< std::pair< double,double > > *self){
      |                                                                                                                                                                        ^
src/QuantLib/quantlib_wrap.cpp: In function ‘MatrixRow Matrix___getitem__(QuantLib::Matrix*, QuantLib::Integer)’:
src/QuantLib/quantlib_wrap.cpp:9804:29: warning: comparison of integer expressions of different signedness: ‘QuantLib::Integer’ {aka ‘int’} and ‘QuantLib::Size’ {aka ‘long unsigned int’} [-Wsign-compare]
 9804 |             if (i >= 0 && i < self->rows())
      |                           ~~^~~~~~~~~~~~~~
src/QuantLib/quantlib_wrap.cpp: In function ‘PyObject* _wrap_new_FdBlackScholesVanillaEngine__SWIG_19(PyObject*, Py_ssize_t, PyObject**)’:
src/QuantLib/quantlib_wrap.cpp:339384: note: ‘-Wmisleading-indentation’ is disabled from this point onwards, since column-tracking was disabled due to the size of the code/headers
339384 |       if (argp1) tempshared1 = *reinterpret_cast< ext::shared_ptr< GeneralizedBlackScholesProcess > * >(argp1);
src/QuantLib/quantlib_wrap.cpp:339384: note: adding ‘-flarge-source-files’ will allow for more column-tracking support, at the expense of compilation time and memory
g++ -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-z,relro -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o -L/usr/lib -L/usr/lib/mips64el-linux-gnuabi64 -lQuantLib -o build/lib.linux-mips64-cpython-313/QuantLib/_QuantLib.abi3.so -fopenmp
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `PyObject_TypeCheck':
quantlib_wrap.cpp:(.text+0x108): relocation truncated to fit: R_MIPS_CALL16 against `PyType_IsSubtype'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `Py_XDECREF':
quantlib_wrap.cpp:(.text+0x258): relocation truncated to fit: R_MIPS_CALL16 against `_Py_Dealloc'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `PyType_HasFeature':
quantlib_wrap.cpp:(.text+0x2c0): relocation truncated to fit: R_MIPS_CALL16 against `PyType_GetFlags'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_TypeCmp':
quantlib_wrap.cpp:(.text+0x484): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_TypeCheck':
quantlib_wrap.cpp:(.text+0x654): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_MangledTypeQueryModule':
quantlib_wrap.cpp:(.text+0xc2c): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_PackVoidPtr':
quantlib_wrap.cpp:(.text+0x11b4): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
quantlib_wrap.cpp:(.text+0x1200): relocation truncated to fit: R_MIPS_CALL16 against `strcpy@@GLIBC_2.0'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_UnpackVoidPtr':
quantlib_wrap.cpp:(.text+0x127c): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-313/src/QuantLib/quantlib_wrap.o: in function `SWIG_PackDataName':
quantlib_wrap.cpp:(.text+0x134c): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
quantlib_wrap.cpp:(.text+0x13f4): additional relocation overflows omitted from the output
collect2: error: ld returned 1 exit status
error: command '/usr/bin/g++' failed with exit code 1
running build
running build_py
creating build/lib.linux-mips64-cpython-312/QuantLib
copying src/QuantLib/__init__.py -> build/lib.linux-mips64-cpython-312/QuantLib
copying src/QuantLib/QuantLib.py -> build/lib.linux-mips64-cpython-312/QuantLib
running build_ext
building 'QuantLib._QuantLib' extension
creating build/temp.linux-mips64-cpython-312/src/QuantLib
g++ -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR -fPIC -DPy_LIMITED_API=0x03080000 -DNDEBUG -I/usr/include -I/usr/include/python3.12 -c src/QuantLib/quantlib_wrap.cpp -o build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o -fopenmp -std=c++17 -Wno-unused -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR
src/QuantLib/quantlib_wrap.cpp: In function ‘std::vector<std::pair<double, double> >::value_type std_vector_Sl_std_pair_Sl_double_Sc_double_Sg__Sg__pop(std::vector<std::pair<double, double> >*)’:
src/QuantLib/quantlib_wrap.cpp:8999:168: note: the ABI for returning a value with C++17 empty bases but otherwise an aggregate with only one or two floating-point fields was changed in GCC 12.1
 8999 | SWIGINTERN std::vector< std::pair< double,double > >::value_type std_vector_Sl_std_pair_Sl_double_Sc_double_Sg__Sg__pop(std::vector< std::pair< double,double > > *self){
      |                                                                                                                                                                        ^
src/QuantLib/quantlib_wrap.cpp: In function ‘MatrixRow Matrix___getitem__(QuantLib::Matrix*, QuantLib::Integer)’:
src/QuantLib/quantlib_wrap.cpp:9804:29: warning: comparison of integer expressions of different signedness: ‘QuantLib::Integer’ {aka ‘int’} and ‘QuantLib::Size’ {aka ‘long unsigned int’} [-Wsign-compare]
 9804 |             if (i >= 0 && i < self->rows())
      |                           ~~^~~~~~~~~~~~~~
src/QuantLib/quantlib_wrap.cpp: In function ‘PyObject* _wrap_new_FdBlackScholesVanillaEngine__SWIG_18(PyObject*, Py_ssize_t, PyObject**)’:
src/QuantLib/quantlib_wrap.cpp:339281: note: ‘-Wmisleading-indentation’ is disabled from this point onwards, since column-tracking was disabled due to the size of the code/headers
339281 |   if ((nobjs < 7) || (nobjs > 7)) SWIG_fail;
src/QuantLib/quantlib_wrap.cpp:339281: note: adding ‘-flarge-source-files’ will allow for more column-tracking support, at the expense of compilation time and memory
g++ -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-z,relro -g0 -Wall -Wno-strict-aliasing -DBOOST_NO_AUTO_PTR build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o -L/usr/lib -L/usr/lib/mips64el-linux-gnuabi64 -lQuantLib -o build/lib.linux-mips64-cpython-312/QuantLib/_QuantLib.abi3.so -fopenmp
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `PyObject_TypeCheck':
quantlib_wrap.cpp:(.text+0x108): relocation truncated to fit: R_MIPS_CALL16 against `PyType_IsSubtype'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `Py_XDECREF':
quantlib_wrap.cpp:(.text+0x258): relocation truncated to fit: R_MIPS_CALL16 against `_Py_Dealloc'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `PyType_HasFeature':
quantlib_wrap.cpp:(.text+0x2c0): relocation truncated to fit: R_MIPS_CALL16 against `PyType_GetFlags'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_TypeCmp':
quantlib_wrap.cpp:(.text+0x484): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_TypeCheck':
quantlib_wrap.cpp:(.text+0x654): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_MangledTypeQueryModule':
quantlib_wrap.cpp:(.text+0xc2c): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_PackVoidPtr':
quantlib_wrap.cpp:(.text+0x11b4): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
quantlib_wrap.cpp:(.text+0x1200): relocation truncated to fit: R_MIPS_CALL16 against `strcpy@@GLIBC_2.0'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_UnpackVoidPtr':
quantlib_wrap.cpp:(.text+0x127c): relocation truncated to fit: R_MIPS_CALL16 against `strcmp@@GLIBC_2.0'
build/temp.linux-mips64-cpython-312/src/QuantLib/quantlib_wrap.o: in function `SWIG_PackDataName':
quantlib_wrap.cpp:(.text+0x134c): relocation truncated to fit: R_MIPS_CALL16 against `strlen@@GLIBC_2.0'
quantlib_wrap.cpp:(.text+0x13f4): additional relocation overflows omitted from the output
collect2: error: ld returned 1 exit status
error: command '/usr/bin/g++' failed with exit code 1
make: *** [debian/rules:104: build-stamp] Error 1
dpkg-buildpackage: error: debian/rules build-arch subprocess returned exit status 2

Any idea?

[boring-cyborg]

Thanks for posting! It might take a while before we look at your issue, so don't worry if there seems to be no feedback. We'll get to it.

[lballabio]

I won't say I understand the error, but https://stackoverflow.com/questions/10486116/what-does-this-gcc-error-relocation-truncated-to-fit-mean may contain some suggestions.

[eddelbuettel]

Good pointer. Scanning this now. We do use -fPIC (and always have) but maybe the 'large program size' is at play. I will also try to reach out to the porters list.

[eddelbuettel]

No word from the mips porters, sadly. And following helpful-as-usual email with someone from the release team, I filed a request to remove to old previous mips64el build (of 1.35) which was blocking progress as the current 1.37 builds did not match all prior build architectures. With the removal that is no longer an issue and 1.37 can and should migrate to testing.

Hopefully someone who knows the platform will have a tip to reenable is eventually. When it builds it will join the other platforms automagically. So nothing to do from here and closing....