From: Chris Lattner <sabre@nondot.org>
Date: Mon, 15 Jan 2007 07:41:51 +0000 (+0000)
Subject: Run an instcombine pass after inlining but before scalarrepl.  This allows
X-Git-Url: http://plrg.eecs.uci.edu/git/?p=oota-llvm.git;a=commitdiff_plain;h=ac98024e5d23d07b085c8d6b035b785c16689f10

Run an instcombine pass after inlining but before scalarrepl.  This allows
instcombine to clean up the code, which makes more code suitable for SRoA.
This helps C++ code in particular, e.g. speeding up tramp3d by 31%.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33235 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/tools/gccas/gccas.cpp b/tools/gccas/gccas.cpp
index 2d08f36cc83..587a6a850d8 100644
--- a/tools/gccas/gccas.cpp
+++ b/tools/gccas/gccas.cpp
@@ -102,6 +102,7 @@ void AddConfiguredTransformationPasses(PassManager &PM) {
 
   addPass(PM, createRaisePointerReferencesPass());// Recover type information
   addPass(PM, createTailDuplicationPass());      // Simplify cfg by copying code
+  addPass(PM, createInstructionCombiningPass()); // Cleanup for scalarrepl.
   addPass(PM, createCFGSimplificationPass());    // Merge & remove BBs
   addPass(PM, createScalarReplAggregatesPass()); // Break up aggregate allocas
   addPass(PM, createInstructionCombiningPass()); // Combine silly seq's