Hi! C++20, in particular https://wg21.link/P1120R0 paper voted into it, deprecates various operations between enumerators from different enumeration types etc., and as we've switched to -std=gnu++20 by default, this now results in warnings or errors during stage2 and onwards.
The following patch should fix loongarch build. Tested on x86_64 with a cross to powerpc, by doing make clean; make -j32 CXXFLAGS='-g -std=gnu++20' in the gcc/ directory. 2025-11-28 Jakub Jelinek <[email protected]> * config/rs6000/rs6000.cc (complex_multiply_builtin_code): Avoid arithmetics between enumerators from different enum types. (complex_divide_builtin_code): Likewise. --- gcc/config/rs6000/rs6000.cc.jj 2025-10-09 17:50:03.359651422 +0200 +++ gcc/config/rs6000/rs6000.cc 2025-11-28 21:56:27.230568674 +0100 @@ -28490,7 +28490,7 @@ static inline built_in_function complex_multiply_builtin_code (machine_mode mode) { gcc_assert (IN_RANGE (mode, MIN_MODE_COMPLEX_FLOAT, MAX_MODE_COMPLEX_FLOAT)); - int func = BUILT_IN_COMPLEX_MUL_MIN + mode - MIN_MODE_COMPLEX_FLOAT; + int func = BUILT_IN_COMPLEX_MUL_MIN + (mode - MIN_MODE_COMPLEX_FLOAT); return (built_in_function) func; } @@ -28501,7 +28501,7 @@ static inline built_in_function complex_divide_builtin_code (machine_mode mode) { gcc_assert (IN_RANGE (mode, MIN_MODE_COMPLEX_FLOAT, MAX_MODE_COMPLEX_FLOAT)); - int func = BUILT_IN_COMPLEX_DIV_MIN + mode - MIN_MODE_COMPLEX_FLOAT; + int func = BUILT_IN_COMPLEX_DIV_MIN + (mode - MIN_MODE_COMPLEX_FLOAT); return (built_in_function) func; } Jakub
