Problem With Comparison Operator <=> in G++

[Oralloy]

Hello folks,
I am unable to find appropriate documentation on the type promotion of bit-fields when using the generalised comparison operator "<=>".
The problem is that using the GNU compilers, it seems that the internal comparison operator "<=>" tries to promote arguments from unsigned to signed.
This is as boiled down as I can make it.
Here is my compilation command:

g++-12 -std=c++20 -Wnarrowing bit_field.cpp

Here is the code in bit_field.cpp:

//
//  bit_field.cpp
//  - demonstrate problem with bit field comparisons
//
 
#include <cstdint>
#include <iostream>
 
struct thing_t
{
  ::std::uint32_t value : 4;
  auto operator<=>(thing_t const &that) const
  {
    return  (this->value <=> that.value);
  }
};
 
int main()
{
  thing_t s0{1};
  thing_t s1{2};
 
  ::std::cout
    << ::std::boolalpha
    << "s0 -- s1 ? " << ((s0 <=> s1) == 0) << "\n";
 
  return 0;
}
 

And, here are the errors which I see:

bit_field.cpp: In member function ‘auto thing_t::operator<=>(const thing_t&) const’:
bit_field.cpp:14:20: warning: narrowing conversion of ‘((const thing_t*)this)->thing_t::value’ from ‘const uint32_t’ {aka ‘const unsigned int’} to ‘int’ [-Wnarrowing]
   14 |     return  (this->value <=> that.value);
      |              ~~~~~~^~~~~
bit_field.cpp:14:35: warning: narrowing conversion of ‘that.thing_t::value’ from ‘const uint32_t’ {aka ‘const unsigned int’} to ‘int’ [-Wnarrowing]
   14 |     return  (this->value <=> that.value);
      |                              ~~~~~^~~~~
 

Needless to say, I am confused.
My understanding from what I have read is that "value" would be promoted to a fully unsigned int32_t value, which ought be comparable to another of the same type.
My gut reaction is that I have encountered a compiler bug, however I have found only a few in my career, so likely not.
Obviously I have done something stupid, as I can usually read multiple sources of documentation and ascertain where I encroached into undefined behaviour.
Can any of you provide explanatory links or enlighten me?
Thank You,
Oralloy