PATCH: remove impossible case from gas/expr.c:expr

PATCH: remove impossible case from gas/expr.c:expr

[Ben Elliston]

A static analysis tool that I ran over gas picked this up.  It is
impossible for op_left to be O_add in the switch (op_left), as it is
within the body of this if:

1762:  else if (resultP->X_op == O_constant && right.X_op == O_constant)

However, earlier, we have another if condition that handles the
possibility of an O_add operator and resultP->X_op being a constant:

1753:  else if (op_left == O_add && resultP->X_op == O_constant)

Anyone disagree? :-)

Cheers, Ben


2006-10-24  Ben Elliston  <bje@au.ibm.com>

        * expr.c (expr): Don't handle impossible case O_add when switching
        on op_left.

Index: expr.c
===================================================================
RCS file: /cvs/src/src/gas/expr.c,v
retrieving revision 1.67
diff -u -p -r1.67 expr.c
--- expr.c      7 Jun 2006 11:27:57 -0000       1.67
+++ expr.c      24 Oct 2006 05:07:15 -0000
@@ -1785,7 +1785,6 @@ expr (int rankarg,                /* Larger # is highe
            case O_bit_or_not:          resultP->X_add_number |= ~v; break;
            case O_bit_exclusive_or:    resultP->X_add_number ^= v; break;
            case O_bit_and:             resultP->X_add_number &= v; break;
-           case O_add:                 resultP->X_add_number += v; break;
            case O_subtract:            resultP->X_add_number -= v; break;
            case O_eq:
              resultP->X_add_number =

Re: PATCH: remove impossible case from gas/expr.c:expr

[Ian Lance Taylor]

Ben Elliston <bje@au1.ibm.com> writes:

> A static analysis tool that I ran over gas picked this up.  It is
> impossible for op_left to be O_add in the switch (op_left), as it is
> within the body of this if:

That's a nice bit of analysis.

> 2006-10-24  Ben Elliston  <bje@au.ibm.com>
> 
>         * expr.c (expr): Don't handle impossible case O_add when switching
>         on op_left.

Please leave a comment behind, so that a couple of years from now
somebody comparing the list in expr.h with the list in expr.c won't
get confused.

OK with that change.

Thanks.

Ian