minor code-quality regression vs. 2.95

[Jeffrey A Law]

  In message < 20000413164849.B20251@cpopetz.com >you write:
  > Thu Apr 13 15:05:04 CDT 2000  Clinton Popetz  <cpopetz@cygnus.com>
  > 
  > 	* i386.c (ix86_expand_branch): Switch the order of the 
  > 	first two branches for a DImode branch.
We'd need some code to benchmark this.  Presumably something that does
a lot of long long conditionals.

I'd hoped the linux kernel would be a good test, but only 11 files were
affected by the change and only one of those files had a change in the
size of the text segment (improved by a half-dozen or so bytes).

Maybe the gnu-mp or glibc would be a reasonable test?!?

jeff