[coldfire-gnu-discuss] Toolchain, asm parameters numbering and optimisation flags

[fred at aidounix.com]

> Here's your problem.  The inline asm semantics are that all the input
> operands
> are read before any of the output operands are written.  thus it is safe
> to
> allocate the same register(s) for inputs and outputs.  All gcc targets are
> the
> same in this regard.  Your assembly breaks that.  you can fix this by
> telling
> gcc that 'sum' is an early clobber
> 	:"=&d" (sum)
>
Hi Nathan, your knowledge is some order of magnitude beyound mine! :)

I've corrected those points, now I can compile with -O0.
But GCC still scramble the registers (assigning D0 to %2 and %3 for
example).

for example :
C source :
move.l (%[X])+, %[tmp1];
move.l (%[Y])+, %[tmp2];
move.w %[tmp1], %[R1];
move.w %[tmp2], %[R2];

generates, with -O2 :
move.l (a0)+, d1;
move.l (a1)+, d1; ouch!
move.w d1, d2;
move.w d1, d3;

I really don't understand why it nicelly works with -O0 and not
with -O2.

Any idea?

Thanx again