[PATCH] Optionally extend alloca lifetimes to allow for callslot optimization to happen

[Philip Reames]

I'm reasonable sure this code is correct, but I'm not sure it's as well structured as it could be.  I reserve the right to be wrong here; you may just need to explain a few things.  :)


================
Comment at: lib/Transforms/Scalar/MemCpyOptimizer.cpp:891
@@ +890,3 @@
+  AliasAnalysis::Location DstLoc = AliasAnalysis::getLocationForDest(M);
+  MemDepResult DstDepInfo = MD->getPointerDependencyFrom(DstLoc, true,
+                                                         M, M->getParent());
----------------
I'm a bit concerned about running these queries unconditionally.  MDA is rather expensive.  

================
Comment at: lib/Transforms/Scalar/MemCpyOptimizer.cpp:894
@@ +893,3 @@
+
+  // If the destination just started its lifetime, pretend that the lifetime
+  // start isn't present, try the callslot optimization and if it succeeds,
----------------
I'm not clear I understand the problem here.  Is the issue that MD->getDependency is returning the lifetime_start as the dependency but MD->getPointerDependency(Src) is not?  

Also, would it be *correct* to always use the source pointer dependency here?  Or do you need to consider both source and dest dependencies when considering the callSlotOpt?  Depending on the implementation of the callSlotOpt, I could see that going either way.

http://reviews.llvm.org/D7984

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/