[PATCH] Fix PR optimization/11637 (x86)

[Eric Botcazou]

> Then I think the code that adjusts notes should be modified to deal with
> this or perhaps (if that's too much work),

I don't think so.

> the optimization should only be suppressed if there *is* a REG_EQUAL note.

I am under the impression that the LOG_LINKS are botched by the optimization 
too. And, indeed, there is another codepath that can change the destination 
of i3 (see combine.c:2329) and that adjusts the LOG_LINKS afterwards in that 
case (it may or may not be affected by the REG_EQUAL note problem too).

-- 
Eric Botcazou