[llvm] r259675 - Revert r259662, which caused regressions on polly tests.

Daniel Berlin via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 3 23:18:52 PST 2016

> In SCEVExpander::expand, there is a logic to check the cached value to
> be used should dominate the insertion point. However, for the above
> case, the check always passes. That is because the insertion point is
> in a new function, which is unreachable from the old function. However
> for unreachable node, DominatorTreeBase::dominates thinks it will be
> dominated by any other node.

Sigh. Just about 6 months ago, we went through a discussion on this mailing
list where i pointed out this behavior by dominates was bad, and the
general consensus was "hey, how often is it really going to cause a
problem" vs "simplify life".

Chalk one up to the former column.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160203/43147ef9/attachment.html>

More information about the llvm-commits mailing list