llvm-6502/lib/Analysis
Duncan Sands d0c6f3dafd Factor code for testing whether replacing one value with another
preserves LCSSA form out of ScalarEvolution and into the LoopInfo
class.  Use it to check that SimplifyInstruction simplifications
are not breaking LCSSA form.  Fixes PR8622.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119727 91177308-0d34-0410-b5e6-96231b3b80d8
2010-11-18 19:59:41 +00:00
..
IPA Make ModRefBehavior a lattice. Use this to clean up AliasAnalysis 2010-11-10 01:02:18 +00:00
AliasAnalysis.cpp Add helper functions for computing the Location of load, store, 2010-11-11 21:50:19 +00:00
AliasAnalysisCounter.cpp Extend the AliasAnalysis::pointsToConstantMemory interface to allow it 2010-11-08 16:45:26 +00:00
AliasAnalysisEvaluator.cpp Reapply r116831 and r116839, converting AliasAnalysis to use 2010-10-19 22:54:46 +00:00
AliasDebugger.cpp Extend the AliasAnalysis::pointsToConstantMemory interface to allow it 2010-11-08 16:45:26 +00:00
AliasSetTracker.cpp Don't forget the TBAA info, if available. 2010-11-11 21:27:26 +00:00
Analysis.cpp
BasicAliasAnalysis.cpp Make ModRefBehavior a lattice. Use this to clean up AliasAnalysis 2010-11-10 01:02:18 +00:00
CaptureTracking.cpp VAArg doesn't capture its operand. 2010-11-09 20:09:35 +00:00
CFGPrinter.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
CMakeLists.txt Introduce DIBuilder. It is intended to be a front-end friendly interface to emit debuggging information entries in LLVM IR. 2010-11-04 15:01:38 +00:00
ConstantFolding.cpp Boost the power of phi node constant folding slightly: if all 2010-11-14 12:53:18 +00:00
DbgInfoPrinter.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
DebugInfo.cpp Take care of special characters while creating named MDNode name to hold function specific local variable's info. 2010-11-10 22:19:21 +00:00
DIBuilder.cpp Use arrays instead of constant-sized SmallVectors. 2010-11-04 18:45:27 +00:00
DomPrinter.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
InlineCost.cpp
InstCount.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
InstructionSimplify.cpp Move some those Xor simplifications which don't require creating new 2010-11-17 18:52:15 +00:00
Interval.cpp
IntervalPartition.cpp
IVUsers.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
LazyValueInfo.cpp
LibCallAliasAnalysis.cpp
LibCallSemantics.cpp
Lint.cpp Now that hasConstantValue has been made simpler, it may return the 2010-11-17 10:23:23 +00:00
LiveValues.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
Loads.cpp Reapply r116831 and r116839, converting AliasAnalysis to use 2010-10-19 22:54:46 +00:00
LoopDependenceAnalysis.cpp Move SCEV::isLoopInvariant and hasComputableLoopEvolution to be member 2010-11-17 21:23:15 +00:00
LoopInfo.cpp Fix typo pointed out by Trevor Harmon. 2010-11-13 12:16:27 +00:00
LoopPass.cpp
Makefile
MemDepPrinter.cpp Add helper functions for computing the Location of load, store, 2010-11-11 21:50:19 +00:00
MemoryBuiltins.cpp
MemoryDependenceAnalysis.cpp Add helper functions for computing the Location of load, store, 2010-11-11 21:50:19 +00:00
ModuleDebugInfoPrinter.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
NoAliasAnalysis.cpp Extend the AliasAnalysis::pointsToConstantMemory interface to allow it 2010-11-08 16:45:26 +00:00
PHITransAddr.cpp Strip trailing whitespace. 2010-11-18 17:06:31 +00:00
PostDominators.cpp
ProfileEstimatorPass.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
ProfileInfo.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
ProfileInfoLoader.cpp
ProfileInfoLoaderPass.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
ProfileVerifierPass.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
README.txt
RegionInfo.cpp Make some symbols static, move classes into anonymous namespaces. 2010-10-22 17:35:07 +00:00
RegionPass.cpp Add RegionPass support. 2010-10-20 01:54:44 +00:00
RegionPrinter.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
ScalarEvolution.cpp Factor code for testing whether replacing one value with another 2010-11-18 19:59:41 +00:00
ScalarEvolutionAliasAnalysis.cpp Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which 2010-10-19 17:21:58 +00:00
ScalarEvolutionExpander.cpp Introduce memoization for ScalarEvolution dominates and properlyDominates 2010-11-18 00:34:22 +00:00
ScalarEvolutionNormalization.cpp
SparsePropagation.cpp
Trace.cpp
TypeBasedAliasAnalysis.cpp Re-disable TBAA for now; it broke MultiSource/Applications/JM/lencod, 2010-11-12 11:21:08 +00:00
ValueTracking.cpp

Analysis Opportunities:

//===---------------------------------------------------------------------===//

In test/Transforms/LoopStrengthReduce/quadradic-exit-value.ll, the
ScalarEvolution expression for %r is this:

  {1,+,3,+,2}<loop>

Outside the loop, this could be evaluated simply as (%n * %n), however
ScalarEvolution currently evaluates it as

  (-2 + (2 * (trunc i65 (((zext i64 (-2 + %n) to i65) * (zext i64 (-1 + %n) to i65)) /u 2) to i64)) + (3 * %n))

In addition to being much more complicated, it involves i65 arithmetic,
which is very inefficient when expanded into code.

//===---------------------------------------------------------------------===//

In formatValue in test/CodeGen/X86/lsr-delayed-fold.ll,

ScalarEvolution is forming this expression:

((trunc i64 (-1 * %arg5) to i32) + (trunc i64 %arg5 to i32) + (-1 * (trunc i64 undef to i32)))

This could be folded to

(-1 * (trunc i64 undef to i32))

//===---------------------------------------------------------------------===//