llvm-mirror/lib/Analysis
Chandler Carruth d1c1c98162 Teach instsimplify how to constant fold pointer differences.
Typically instcombine has handled this, but pointer differences show up
in several contexts where we would like to get constant folding, and
cannot afford to run instcombine. Specifically, I'm working on improving
the constant folding of arguments used in inline cost analysis with
instsimplify.

Doing this in instsimplify implies some algorithm changes. We have to
handle multiple layers of all-constant GEPs because instsimplify cannot
fold them into a single GEP the way instcombine can. Also, we're only
interested in all-constant GEPs. The result is that this doesn't really
replace the instcombine logic, it's just complimentary and focused on
constant folding.

Reviewed on IRC by Benjamin Kramer.

llvm-svn: 152555
2012-03-12 11:19:31 +00:00
..
IPA LLVMBuild: Remove trailing newline, which irked me. 2011-12-12 19:48:00 +00:00
AliasAnalysis.cpp Move isKnownNonNull from private implementation detail of BasicAA to a public 2012-02-25 10:56:28 +00:00
AliasAnalysisCounter.cpp Persuade GCC that there is nothing worth warning about here (there isn't). 2012-02-05 14:20:11 +00:00
AliasAnalysisEvaluator.cpp Remove unnecessary default cases in switches that cover all enum values. 2012-01-10 16:47:17 +00:00
AliasDebugger.cpp Extend the AliasAnalysis::pointsToConstantMemory interface to allow it 2010-11-08 16:45:26 +00:00
AliasSetTracker.cpp Have AliasSet::aliasesUnknownInst use pointer TBAA info when available 2012-02-10 15:52:39 +00:00
Analysis.cpp C API functions must be able to see their extern "C" definitions, or it will be impossible to call them from C. 2011-08-19 01:36:54 +00:00
BasicAliasAnalysis.cpp Duncan pointed out that if the alignment isn't explicitly specified, it defaults to the ABI alignment. Given that, make this code a bit more aggressive in such cases. 2012-02-27 23:16:46 +00:00
BlockFrequencyInfo.cpp Add some constantness to BranchProbabilityInfo and BlockFrequnencyInfo. 2011-12-20 20:03:10 +00:00
BranchProbabilityInfo.cpp Make the unreachable probability much much heavier. The previous 2011-12-22 09:26:37 +00:00
CaptureTracking.cpp Move includes to the .cpp file. 2012-01-17 22:16:31 +00:00
CFGPrinter.cpp Remove all remaining uses of Value::getNameStr(). 2011-11-15 16:27:03 +00:00
CMakeLists.txt build/CMake: Finish removal of add_llvm_library_dependencies. 2011-11-29 19:25:30 +00:00
ConstantFolding.cpp Convert assert(0) to llvm_unreachable 2012-02-07 05:05:23 +00:00
DbgInfoPrinter.cpp Fix for DbgInfoPrinter.cpp:174:12: warning: ‘LineNo’ may be used uninitialized in this function. 2011-09-21 23:34:23 +00:00
DebugInfo.cpp Make this a bit prettier and more obvious when a derived type isn't 2012-02-20 18:04:39 +00:00
DIBuilder.cpp Add support for runtime languages on our forward declarations. 2012-02-20 18:04:14 +00:00
DominanceFrontier.cpp Unweaken vtables as per http://llvm.org/docs/CodingStandards.html#ll_virtual_anch 2011-12-20 02:50:00 +00:00
DomPrinter.cpp remove the blank line from previous ci. 2012-02-04 03:18:47 +00:00
InlineCost.cpp Make helper static, so it can be inlined into its sole caller. 2012-03-10 22:41:06 +00:00
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 Teach instsimplify how to constant fold pointer differences. 2012-03-12 11:19:31 +00:00
Interval.cpp
IntervalPartition.cpp Now with fewer extraneous semicolons! 2010-10-07 22:25:06 +00:00
IVUsers.cpp Put all IVUsers in the processed set. Allow querying IVUsers with isIVUserOrOperand. 2012-01-06 21:41:55 +00:00
LazyValueInfo.cpp llvm::SwitchInst 2012-03-11 06:09:17 +00:00
LibCallAliasAnalysis.cpp Now with fewer extraneous semicolons! 2010-10-07 22:25:06 +00:00
LibCallSemantics.cpp
Lint.cpp Fix a few more places where TargetData/TargetLibraryInfo is not being passed. 2011-12-02 01:26:24 +00:00
LLVMBuild.txt LLVMBuild: Introduce a common section which currently has a list of the 2011-12-12 22:45:54 +00:00
Loads.cpp Add some comments here because the lack of a check for volatile/atomic here is a bit unusual. 2011-08-15 21:56:39 +00:00
LoopDependenceAnalysis.cpp More dead code removal (using -Wunreachable-code) 2012-01-20 21:51:11 +00:00
LoopInfo.cpp Move Instruction::isSafeToSpeculativelyExecute out of VMCore and 2011-12-14 23:49:11 +00:00
LoopPass.cpp Reapply r138695. Fix PassManager stack depths. 2011-08-29 17:07:00 +00:00
Makefile
MemDepPrinter.cpp Fix a stupid typo in MemDepPrinter. 2011-12-14 02:54:39 +00:00
MemoryBuiltins.cpp Validate the return type when checking if a function is malloc. 2011-11-23 17:58:47 +00:00
MemoryDependenceAnalysis.cpp Don't call dominates on unreachable instructions. Should fix the dragonegg 2012-02-26 05:30:08 +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 Use the new addEscapingValue callback to update GlobalsModRef when GVN adds PHIs of GEPs. For the moment, 2011-01-03 23:51:43 +00:00
PathNumbering.cpp [unwind removal] We no longer have 'unwind' instructions being generated, so 2012-02-06 21:16:41 +00:00
PathProfileInfo.cpp Implementation of path profiling. 2011-01-29 01:09:53 +00:00
PathProfileVerifier.cpp Remove all remaining uses of Value::getNameStr(). 2011-11-15 16:27:03 +00:00
PHITransAddr.cpp More dead code removal (using -Wunreachable-code) 2012-01-20 21:51:11 +00:00
PostDominators.cpp remove postdom frontiers, because it is dead. Forward dom frontiers are 2011-04-05 21:57:17 +00:00
ProfileEstimatorPass.cpp Remove all remaining uses of Value::getNameStr(). 2011-11-15 16:27:03 +00:00
ProfileInfo.cpp Fix a ton of comment typos found by codespell. Patch by 2011-04-15 05:18:47 +00:00
ProfileInfoLoader.cpp Remove unused STL header includes. 2011-04-23 19:53:52 +00:00
ProfileInfoLoaderPass.cpp Remove all remaining uses of Value::getNameStr(). 2011-11-15 16:27:03 +00:00
ProfileVerifierPass.cpp Move code into anonymous namespaces. 2011-11-26 23:01:57 +00:00
README.txt
RegionInfo.cpp Remove extra semi-colons. 2012-02-22 17:25:00 +00:00
RegionPass.cpp Reapply r138695. Fix PassManager stack depths. 2011-08-29 17:07:00 +00:00
RegionPrinter.cpp RegionPrinter: Ignore back edges when layouting the graph 2011-02-27 04:11:07 +00:00
ScalarEvolution.cpp switch SCEV to use the new ConstantFoldLoadThroughGEPIndices function 2012-01-24 05:49:24 +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 Fix this assert. IP can point to an instruction with strange dominance 2012-02-27 02:13:03 +00:00
ScalarEvolutionNormalization.cpp More dead code removal (using -Wunreachable-code) 2012-01-20 21:51:11 +00:00
SparsePropagation.cpp Taken into account Duncan's comments for r149481 dated by 2nd Feb 2012: 2012-03-08 07:06:20 +00:00
Trace.cpp Remove all remaining uses of Value::getNameStr(). 2011-11-15 16:27:03 +00:00
TypeBasedAliasAnalysis.cpp Fix a ton of comment typos found by codespell. Patch by 2011-04-15 05:18:47 +00:00
ValueTracking.cpp Factor out the analysis of addition and subtraction in ComputeMaskedBits. Reuse 2012-03-09 09:23:50 +00:00

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))

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