[Bug tree-optimization/104547] std::vector::resize(v.size() - n) produces poor code

[redi at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104547

--- Comment #3 from Jonathan Wakely <redi at gcc dot gnu.org> ---
For this TU with three functions, clang generates good code:


#include <vector>

void shrink_pop(std::vector<int>& v, std::size_t n) {
    while (n--)
      v.pop_back();
}

void shrink_assume(std::vector<int>& v, std::size_t n) {
    if (v.size() < n)
      __builtin_unreachable();
    v.resize(v.size() - n);
}

void shrink_min(std::vector<int>& v, std::size_t n) {
    v.resize(std::min<std::size_t>(v.size(), n));
}


Clang removes the unreachable __throw_length_error, doesn't explode if the type
is size_t, and doesn't keep the unused _M_default_append instantiation.