Opinion about -Wtautological-compare

[Simon Marchi <simon.marchi@ericsson.com>]

Hi all,

When building with Clang, I see warnings like these in multiple files:

/home/emaisin/src/binutils-gdb/gdb/arm-tdep.c:10013:13: error: comparison of 0 <= unsigned expression is always true [-Werror,-Wtautological-compare]
      if (0 <= insn_op1 && 3 >= insn_op1)
          ~ ^  ~~~~~~~~
/home/emaisin/src/binutils-gdb/gdb/arm-tdep.c:11722:20: error: comparison of unsigned expression >= 0 is always true [-Werror,-Wtautological-compare]
      else if (opB >= 0 && opB <= 2)
               ~~~ ^  ~

They are saying that because the variable is unsigned, the comparison
"variable >= 0" will always be true.  What do you think we should do with
this?

1) Remove the comparisons and keep the warning:

  - if (0 <= insn_op1 && 3 >= insn_op1)
  + if (3 >= insn_op1)

2) Leave the comparisons, but put them in comments and keep the warning:

  - if (0 <= insn_op1 && 3 >= insn_op1)
  + if (/* 0 <= insn_op1 && */ 3 >= insn_op1)

3) Make a util function "in_inclusive_range" and use it (and keep the warning).  Something like:

static bool
in_inclusive_range (unsigned int value, unsigned int low, unsigned int high)
{
  return value >= low && value <= high;
}

and

  - if (0 <= insn_op1 && 3 >= insn_op1)
  + if (in_inclusive_range (insn_op1, 0, 3))


4) Leave the code as-is and remove the warning (add -Wno-tautological-compare to the build).

Personally, I think the warning is useful and can reveal bugs, so I'd like to keep
it.  I lean towards 2 or 3, because they help convey the idea that we check if the
value is within a range.  If you are following with the architecture manual on the
side, it will probably show the same range (0-3) for those bits, so it helps if the
code does the same.

Simon