This is the mail archive of the
mailing list for the GCC project.
Re: [patch] for PR 18687
Mark Mitchell wrote:
Zdenek, from your original posting it sounded like the two improvements
that you made should not affect the generated code.
Did you expect your change to have other side-effects?
This same patch also fixed PR 18977 as a side effect. I didn't bother
fully analyzing the problem, but my suspicion is that it was fixed by
the addition of per_type to may_eliminate_iv. The new code is clearly
doing a type conversion that wasn't performed by the original code, and
hence this is clearly not just a compile-time only change. In this
case, however, the patch did something good.
The code is also a little sloppy here. There is a redundant assignment
to wider_type that could be deleted.
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com