Possible issue when importing 8-bit chunks into cpp_int

[ckormanyos]

Well, I ran into the following phenomenon today.

While doing some in-depth testing of cpp_int's facility for import_bits(), I found what seems to be a potential bug. It is only in GCC, not in MSVC (did not try clang yet).

Anyway, when importing a chunk_size or 8 from an array of 16-bit unsigned integers, and only for msv_first being false, the import function seems to be optimized and imports too large of chunks.

I would like to retain the optimization, but maybe it needs improved logic. It seems to be independent of singed/unsigned versions of char. Yet I can't quite figure out why MSVC gets it OK, but GCC does not.

The entire code is below. The link to Wandbox is here.

#include <array>
#include <cstdint>
#include <iomanip>
#include <iostream>

#include <boost/multiprecision/cpp_int.hpp>

// cd /mnt/c/Users/User/Documents/Ks/PC_Software/Test
// g++-10 -march=native -mtune=native -O3 -Wall -Wextra -std=c++20 -I/mnt/c/boost/boost_1_80_0 test.cpp -o test.exe

auto main() -> int
{
  using local_uint265_type = boost::multiprecision::uint256_t;

  using bits_array_type = std::array<std::uint16_t, static_cast<std::size_t>(UINT8_C(16))>;

  bits_array_type bits { };

  bits.fill(static_cast<std::uint16_t>(UINT16_C(0x5678)));

  local_uint265_type u { };

  const std::array<bool, static_cast<std::size_t>(UINT8_C(2))> msv_values { true, false };

  const auto flg = std::cout.flags();

  for(const auto msv_first : msv_values)
  {
    static_cast<void>(import_bits(u, bits.cbegin(), bits.cend(), static_cast<unsigned>(UINT8_C(8)), msv_first));

    std::cout << std::hex << std::uppercase << u << std::endl;
  }

  std::cout.flags(flg);

}

[ckormanyos]

It is present in 1.80 and on develop as well. Seems also like the code has been around for a while.

[jzmaddock]

I have a trivial fix, but need to devise a portable test case (which is somewhat harder).

The issue is triggered by std::array<>::const_iterator being a pointer in gcc, and the fact that you're only importing the first byte of each value and not the whole thing, it's a check for the latter which is missing in the gcc case.

[ckormanyos]

Thank you John!