[PATCH] Fix Darwin bootstrap (middle-end/37293)

[Richard Guenther]

On Wed, Sep 3, 2008 at 3:21 AM, Andrew Pinski
<Andrew_Pinski@playstation.sony.com> wrote:
> Hi,
>  With turning on IPA-CP, we were hitting the case where a function is
> cloned the function was still being marked as weak.  After fixing that
> so it is no longer being marked as weak, the function calls need to be
> updated as we changed the call from being weak to non weak so the tree
> optimizers can now see that the function is non throwing.
>
> This fixes the two issues that were found by the darwin bootstrap failure.
> OK? Bootstrapped and tested on i386-darwin with no regressions that
> are related to the patch, some of them are related to the IRA merge
> and others are related to optimize for size changes.

Do we need to purge dead EH edges at this point?

Otherwise the patch is ok.

Thanks,
Richard.

> Thanks,
> Andrew Pinski
>
> ChangeLog:
> * cgraphunit.c (update_call_expr): Remove eh regions from statements
> which become non throw.
> (cgraph_function_versioning): Also clear DECL_WEAK.  Call
> update_call_expr after updating the flags on the decl.
>