aboutsummaryrefslogtreecommitdiff
path: root/lib/Transforms/Scalar/LoopStrengthReduce.cpp
AgeCommit message (Collapse)Author
2009-06-22SCEVHandle is no more!Owen Anderson
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73906 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-21Rename a variable for consistency with the ExitBlock vs ExitingBlockDan Gohman
terminology that LoopInfo uses. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73863 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-19Fix a typo in a comment that Frits von Bommel noticed.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73796 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-19Re-apply r73718, now that the fix in r73787 is in, and add aDan Gohman
hand-crafted testcase which demonstrates the bug that was exposed in 254.gap. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73793 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-19Fix LSR's OptimizeSMax to ignore max operators with more than 2 operands,Dan Gohman
which it isn't prepared to handle. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73787 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-19Revert 73718. It's breaking 254.gap.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73783 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-18Generalize LSR's OptimizeSMax to handle unsigned max tests as wellDan Gohman
as signed max tests. Along with r73717, this helps CodeGen avoid emitting code for a maximum operation for this class of loop. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73718 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-18Remove the code from IVUsers that attempted to handleDan Gohman
casted induction variables in cases where the cast isn't foldable. It ended up being a pessimization in many cases. This could be fixed, but it would require a bunch of complicated code in IVUsers' clients. The advantages of this approach aren't visible enough to justify it at this time. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73706 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-17Update comments to use doxygen syntax.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73621 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-15Support vector casts in more places, fixing a variety of assertionDan Gohman
failures. To support this, add some utility functions to Type to help support vector/scalar-independent code. Change ConstantInt::get and ConstantFP::get to support vector types, and add an overload to ConstantInt::get that uses a static IntegerType type, for convenience. Introduce a new getConstant method for ScalarEvolution, to simplify common use cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73431 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-14Convert several parts of the ScalarEvolution framework to useDan Gohman
SmallVector instead of std::vector. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73357 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-05Simplify.Devang Patel
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72965 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-04Split the Add, Sub, and Mul instruction opcodes into separateDan Gohman
integer and floating-point opcodes, introducing FAdd, FSub, and FMul. For now, the AsmParser, BitcodeReader, and IRBuilder all preserve backwards compatability, and the Core LLVM APIs preserve backwards compatibility for IR producers. Most front-ends won't need to change immediately. This implements the first step of the plan outlined here: http://nondot.org/sabre/LLVMNotes/IntegerOverflow.txt git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72897 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-27Revert 72493 and replace it with a more conservative fix, for now: don'tDan Gohman
rewrite the comparison if there is any implicit extension or truncation on the induction variable. I'm planning for IVUsers to eventually take over some of the work of this code, and for it to be generalized. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72496 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-27In ChangeCompareStride, when the stride to be reused is truncated toDan Gohman
a smaller type, promoted its offset back up to the type of the new comparison. This fixes PR4222. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72493 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-20Suppress the IV reversal transformation in the case that the RHSDan Gohman
of the comparison is defined inside the loop. This fixes a use-before-def problem, because the transformation puts a use of the RHS outside the loop. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72149 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-19Add some more comments to the top of this file.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72131 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-19Trim unneeded #includes.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72130 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-19Teach SCEVExpander to expand arithmetic involving pointers into GEPDan Gohman
instructions. It attempts to create high-level multi-operand GEPs, though in cases where this isn't possible it falls back to casting the pointer to i8* and emitting a GEP with that. Using GEP instructions instead of ptrtoint+arithmetic+inttoptr helps pointer analyses that don't use ScalarEvolution, such as BasicAliasAnalysis. Also, make the AddrModeMatcher more aggressive in handling GEPs. Previously it assumed that operand 0 of a GEP would require a register in almost all cases. It now does extra checking and can do more matching if operand 0 of the GEP is foldable. This fixes a problem that was exposed by SCEVExpander using GEPs. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72093 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-18Rename UseTy to AccessTy, for consistency with getAccessType, and toDan Gohman
avoid ambiguity with the word "use" in IVStrideUse. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72012 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-13Add an int64_t variant of abs, for host environmentsDale Johannesen
without one. Use it where we were using abs on int64_t objects. (I strongly suspect the casts to unsigned in the fragments in LoopStrengthReduce are not doing whatever the original intent was, but the obvious change to uint64_t doesn't work. Maybe later.) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71612 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-12Factor the code for collecting IV users out of LSR into an IVUsers class,Dan Gohman
and generalize it so that it can be used by IndVarSimplify. Implement the base IndVarSimplify transformation code using IVUsers. This removes TestOrigIVForWrap and associated code, as ScalarEvolution now has enough builtin overflow detection and folding logic to handle all the same cases, and more. Run "opt -iv-users -analyze -disable-output" on your favorite loop for an example of what IVUsers does. This lets IndVarSimplify eliminate IV casts and compute trip counts in more cases. Also, this happens to finally fix the remaining testcases in PR1301. Now that IndVarSimplify is being more aggressive, it occasionally runs into the problem where ScalarEvolutionExpander's code for avoiding duplicate expansions makes it difficult to ensure that all expanded instructions dominate all the instructions that will use them. As a temporary measure, IndVarSimplify now uses a FixUsesBeforeDefs function to fix up instructions inserted by SCEVExpander. Fortunately, this code is contained, and can be easily removed once a more comprehensive solution is available. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71535 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-11Teach LSR to optimize more loop exit compares, i.e. change them to use ↵Evan Cheng
postinc iv value. Previously LSR would only optimize those which are in the loop latch block. However, if LSR can prove it is safe (and profitable), it's now possible to change those not in the latch blocks to use postinc values. Also, if the compare is the only use, LSR would place the iv increment instruction before the compare instead in the latch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71485 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-11Reverse a loop that is counting up to a maximum toDale Johannesen
count down to 0 instead, under very restricted circumstances. Adjust 4 testcases in which this optimization fires. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71439 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-09Factor out code that optimize loop terminating condition.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71305 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-06Unbreak the build.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71091 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-06Make sure to use signed arithmetic in APInt to fix a regression.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71090 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-05Fix a copy+pasto in a comment.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71035 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-05Delete a FIXME which is no longer relevant, and add a FIXME that is.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71033 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-05Temporarily reverting r71008. It was causing this failure:Bill Wendling
Running /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.src/test/ CodeGen/X86/dg.exp ... FAIL: /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.src/test/ CodeGen/X86/change-compare-stride-1.ll Failed with exit(1) at line 2 while running: grep {cmpq $-478,} change-compare-stride-1.ll.tmp child process exited abnormally git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71013 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-05Handle overflow of 64-bit loop conditions.David Greene
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71008 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-04Re-apply 70645, converting ScalarEvolution to useDan Gohman
CallbackVH, with fixes. allUsesReplacedWith need to walk the def-use chains and invalidate all users of a value that is replaced. SCEVs of users need to be recalcualted even if the new value is equivalent. Also, make forgetLoopPHIs walk def-use chains, since any SCEV that depends on a PHI should be recalculated when more information about that PHI becomes available. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70927 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-04Constify a bunch of SCEV-using code.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70919 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-03Revert r70645 for now; it's causing a variety of regressions.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70661 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-02Convert ScalarEvolution to use CallbackVH for its internal map. ThisDan Gohman
makes ScalarEvolution::deleteValueFromRecords, and it's code that subtly needed to be called before ReplaceAllUsesWith, unnecessary. It also makes ValueDeletionListener unnecessary. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70645 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-02Previously, RecursivelyDeleteDeadInstructions provided an optionDan Gohman
of returning a list of pointers to Values that are deleted. This was unsafe, because the pointers in the list are, by nature of what RecursivelyDeleteDeadInstructions does, always dangling. Replace this with a simple callback mechanism. This may eventually be removed if all clients can reasonably be expected to use CallbackVH. Use this to factor out the dead-phi-cycle-elimination code from LSR utility function, and generalize it to use the RecursivelyDeleteTriviallyDeadInstructions utility function. This makes LSR more aggressive about eliminating dead PHI cycles; adjust tests to either be less trivial or to simply expect fewer instructions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70636 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-02Don't split critical edges during the AddUsersIfInteresting phaseDan Gohman
of LSR. This makes the AddUsersIfInteresting phase of LSR a pure analysis instead of a phase that potentially does CFG modifications. The conditions where this code would actually perform a split are rare, and in the cases where it actually would do a split the split is usually undone by CodeGenPrepare, and in cases where splits actually survive into codegen, they appear to hurt more often than they help. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70625 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-01Make RequiresTypeConversion canonicalize the types before calling theDan Gohman
target hooks canLosslesslyBitCastTo and isTruncateFree. This allows targets to avoid worrying about handling all combinations of integer and pointer types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70555 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-01Minor whitespace fix.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70551 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-01Fix some code to work if TargetLowering is not available.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70546 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-29Print correct instruction in dump.Dale Johannesen
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70427 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-27Permit ChangeCompareStride to rewrite a comparison when the factorDan Gohman
between the comparison's iv stride and the candidate stride is exactly -1. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70244 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-21Factor out a common base class from SCEVTruncateExpr, SCEVZeroExtendExpr,Dan Gohman
and SCEVSignExtendExpr. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69649 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-21Introduce encapsulation for ScalarEvolution's TargetData object, and refactorDan Gohman
the code to minimize dependencies on TargetData. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69644 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-18Use more const qualifiers with SCEV interfaces.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69450 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-17Don't create ConstantInts with pointer type. This fixes aDan Gohman
regression in 403.gcc in PIC_CODEGEN=1 and DISABLE_LTO=1 mode. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69344 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-16Use TargetData::getTypeSizeInBits instead of getPrimitiveSizeInBits()Dan Gohman
to get the correct answer for pointer types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69321 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-16Minor code simplifications. Don't attempt LSR on theoreticalDan Gohman
targets with pointers larger than 64 bits, due to the code not yet being APInt clean. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69296 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-16LSR is no longer a GEP optimizer. It is now an IV expressionDan Gohman
optimizer, which just happen to frequently involve optimizing GEPs. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69295 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-16Use ConstantExpr::getIntToPtr instead of SCEVExpander::InsertCastOfTo,Dan Gohman
since the operand is always a constant. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69291 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.2.3-18-g5258 (git 2.32.0) at 2025-05-08 05:50:46 +0000