This is the mail archive of the
mailing list for the GCC project.
Re: Aliasing patch for tree-optimization/28778
On 8/27/06, Mark Mitchell <firstname.lastname@example.org> wrote:
Daniel Berlin wrote:
> I don't disagree, I just don't believe the size of the change it will
> take to fix this (probably about 500-1000 lines), and how it is going to
> affect core infrastructure, is really something that should be in stage3.
> But you are the release manager, so if you want me to write up something
> to fix this correctly, i will do it for 4.2. It will take about 2 weeks
> though, as i can't do it during work hours right now (google assignment
I would certainly like you to try to fix the bug. Then, we can decide
whether to take the risk of putting it in during Stage 3, or, at worst,
have it ready to go for 4.3. We might even consider branching, putting
it into 4.3, and then backporting it. But, if we don't have the code,
we don't have any options.
I will code up a non-pessimizing correct fix.
The pessimizing version of the conservative correct fix (IE mark
everything clobbered in any non-trivial function) was attached to
28778 by Richard Guenther if you want to look at it.
We both believe we can't make the decision as to whether this makes
sense to put in 4.2 is yours.
Richard also pointed out that we could disable strict aliasing for
4.2, which would fix this bug and others with the same root cause
(TBAA pruning alias sets).