Hi! To my surprise the C FE marks as builtin even a declaration which has incorrect return type. Normally gimple_builtin_call_types_compatible_p etc. will just punt in cases where the return type is wrong, but builtins.cc doesn't use that. For e.g. the mathfn builtins like sqrt and many others, it will punt on weird return types, but for fold_builtin_abs it doesn't and happily tests TYPE_UNSIGNED on it and fold_convert the integral operand to it etc., which ICEs if the return type is aggregate.
The following patch fixes it by punting if type is not integral. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? 2026-01-23 Jakub Jelinek <[email protected]> PR middle-end/123703 * builtins.cc (fold_builtin_abs): Return NULL_TREE if type is not integral. * gcc.c-torture/compile/pr123703.c: New test. --- gcc/builtins.cc.jj 2026-01-02 09:56:09.907340908 +0100 +++ gcc/builtins.cc 2026-01-22 11:20:57.607868166 +0100 @@ -9538,7 +9538,7 @@ fold_builtin_fabs (location_t loc, tree static tree fold_builtin_abs (location_t loc, tree arg, tree type) { - if (!validate_arg (arg, INTEGER_TYPE)) + if (!validate_arg (arg, INTEGER_TYPE) || !INTEGRAL_TYPE_P (type)) return NULL_TREE; if (TYPE_UNSIGNED (type)) --- gcc/testsuite/gcc.c-torture/compile/pr123703.c.jj 2026-01-22 11:22:20.733473067 +0100 +++ gcc/testsuite/gcc.c-torture/compile/pr123703.c 2026-01-22 11:23:34.705231596 +0100 @@ -0,0 +1,10 @@ +/* PR middle-end/123703 */ + +struct S { int a; }; +struct S abs (int); + +struct S +bar (int j) +{ + return abs (j); +} Jakub
