[PATCH] MemoryDependenceAnalysis always depends on DominatorTree

Daniel Berlin dberlin at dberlin.org
Tue Apr 16 14:06:37 PDT 2013


I didn't check to see if true was an acceptable result :)

MemoryDependenceAnalysis actually uses fake results in some cases to
signal failure (search around for getUnknown)


On Tue, Apr 16, 2013 at 2:01 PM, Matt Arsenault
<Matthew.Arsenault at amd.com> wrote:
>
>   In that case, why bother figuring out some fake result to add, and just return true for failure? Wouldn't the result be redundant?
>
> http://llvm-reviews.chandlerc.com/D583



More information about the llvm-commits mailing list