[C,C++,Ada,Java,F77] opt/5076 and opt/2847

[Mark Mitchell]

--On Wednesday, January 30, 2002 10:40:07 AM -0800 Richard Henderson 
<rth@redhat.com> wrote:

> On Wed, Jan 30, 2002 at 10:28:34AM -0800, Mark Mitchell wrote:
>> Why are we getting the wrong answer in expand_end_loop?
>
>   while (test == 0)
>     {
>       inc ();
>       if (iNbr == 0)
>         break;
>       else

OK, that makes sense.  What if we ran the code in expand_end_loop
in the reverse order so that we found the last test in the body
of the loop?  I guess that really makes no difference; without
knowledge of control-flow the lexical order these things appear
in doesn't really mean much.

OK, I agree; this is broken.  I think your marker idea is the
best bet at this point.

Thanks for working this through.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com