Optimize ModuleManager::visit() by precomputing the visitation order
and limiting ourselves to two memory allocations. 10% speedup in
-fsyntax-only time for modules.

With this change, we can actually see some performance different from
the global module index, but it's still about 1%.



git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@173512 91177308-0d34-0410-b5e6-96231b3b80d8
2 files changed