[Patch] fix for loop optimizer (strength reduction)

[Jose Luu]

Hi Jeff,

thanks for your review, I will take the remarks into account for the next
patch I submit. Now, for this one, I did not see a commit of yours (besides
the test case), do you mean that I need to come up and resubmit it with a
more efficient loop scan before you integrate it ?

Can it be integrated in the 2.95 branch ?
Is there a 2.95.3 planned ?
>
>Your code is also rather expensive in that it scans much of the loop and
>has to call reg_mentioned_p on the pattern for each insn it scans.  Ouch!
>I would be rather surprised if there aren't some variables you can check
>to determine the relative positions of the first use and the first
>set of a potential GIV.
>
>jeff
>
>