On Tue, Apr 24, 2012 at 9:55 AM, Chandler Carruth <[email protected]>wrote:

> On Mon, Apr 23, 2012 at 1:09 AM, Kostya Serebryany <[email protected]> wrote:
>
>> Hello,
>> Please review the following updated patch.
>> TBAA should be generated for ThreadSanitizer regardless of opt level
>> and relaxed-aliasing.
>>
>
> Since Chris has goon a bit silent, I'll pick this up. I think you've
> addressed his feedback already.
>
> Couple of tiny nits, and with these fixes, please commit.
>
> +// when -fthread-sanitizer is used TBAA should be generated at all opt
> levels
> +// even if -relaxed-aliasing is present.
>
> s/when/When/
>
Done.

>
> Can you add a negative test that demonstrates we *don't* generate this
> TBAA in -O0 or -O2 -relaxed-aliasing?
>
Done.

>
> I wonder if -relaxed-aliasing should trump -fthread-sanitizer, and fully
> suppress the TBAA...
>

Then those using -relaxed-aliasing and -fthread-sanitizer will be getting
undesirable warnings.
BTW, we *are* using -relaxed-aliasing :)

Thanks!

--kcc


> Not clear, but not a big deal to get addressed before you commit as it
> only impacts -fthread-sanitizer.
>
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to