[llvm-commits] [llvm] r150885 - /llvm/trunk/lib/Analysis/ScalarEvolutionExpander.cpp

Rafael Ávila de Espíndola rafael.espindola at gmail.com
Sat Feb 18 14:03:05 PST 2012


On 18/02/12 03:04 PM, Rafael Ávila de Espíndola wrote:
> Looks like the problem was not the assert, but the dominate function!
> 
> When A is an invoke, the value is only available on the normal dest BB,
> but the instruction is certainly not in that BB. That makes us conclude
> that an invoke never dominates anything.
> 
> The attached patch fixes it. Is it the right fix? Should the value
> produced by the invoke dominate the phis on the dest block (I think so,
> but it is a corner case)..

Extra testing found that we were still not getting the case of one
invoke dominating itself. The attached patch fixes it and adds a testcase.

Cheers,
Rafael
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: dominate.patch
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120218/905bb6e0/attachment.ksh>


More information about the llvm-commits mailing list