| Commit message (Expand) | Author | Age | Files | Lines |
* | Generalize the FoldOrWithConstant method to fold for any two constants which | Bill Wendling | 2008-12-01 | 1 | -23/+22 |
* | Reduce copy-and-paste code by splitting out the code into its own function. | Bill Wendling | 2008-12-01 | 1 | -58/+50 |
* | Use m_Specific() instead of double matching. | Bill Wendling | 2008-12-01 | 1 | -18/+12 |
* | Move pattern check outside of the if-then statement. This prevents us from fi... | Bill Wendling | 2008-12-01 | 1 | -10/+12 |
* | Rename some variables, only increment BI once at the start of the loop instea... | Chris Lattner | 2008-12-01 | 1 | -38/+30 |
* | pull the predMap densemap out of the inner loop of performPRE, so | Chris Lattner | 2008-12-01 | 1 | -2/+4 |
* | switch a couple more calls to use array_pod_sort. | Chris Lattner | 2008-12-01 | 2 | -3/+5 |
* | Introduce a new array_pod_sort function and switch LSR to use it | Chris Lattner | 2008-12-01 | 1 | -1/+1 |
* | Eliminate use of setvector for the DeadInsts set, just use a smallvector. | Chris Lattner | 2008-12-01 | 1 | -17/+31 |
* | DeleteTriviallyDeadInstructions is always passed the | Chris Lattner | 2008-12-01 | 1 | -10/+9 |
* | simplify DeleteTriviallyDeadInstructions again, unlike my previous | Chris Lattner | 2008-12-01 | 1 | -20/+13 |
* | simplify these patterns using m_Specific. No need to grep for | Chris Lattner | 2008-12-01 | 1 | -16/+6 |
* | Teach jump threading to clean up after itself, DCE and constfolding the | Chris Lattner | 2008-12-01 | 1 | -1/+24 |
* | Change instcombine to use FoldPHIArgGEPIntoPHI to fold two operand PHIs | Chris Lattner | 2008-12-01 | 1 | -17/+5 |
* | Teach inst combine to merge GEPs through PHIs. This is really | Chris Lattner | 2008-12-01 | 1 | -16/+95 |
* | Make GVN be more intelligent about redundant load | Chris Lattner | 2008-12-01 | 1 | -2/+19 |
* | Reimplement the non-local dependency data structure in terms of a sorted | Chris Lattner | 2008-12-01 | 1 | -22/+34 |
* | Implement ((A|B)&1)|(B&-2) -> (A&1) | B transformation. This also takes care of | Bill Wendling | 2008-12-01 | 1 | -0/+67 |
* | Cache analyses in ivars and add some useful DEBUG output. | Chris Lattner | 2008-12-01 | 1 | -37/+30 |
* | improve indentation, do cheap checks before expensive ones, | Chris Lattner | 2008-11-30 | 1 | -52/+51 |
* | Minor cleanup: use getTrue and getFalse where appropriate. No | Eli Friedman | 2008-11-30 | 1 | -20/+20 |
* | Some minor cleanups to instcombine; no functionality change. | Eli Friedman | 2008-11-30 | 1 | -56/+17 |
* | Add instruction combining for ((A&~B)|(~A&B)) -> A^B and all permutations. | Bill Wendling | 2008-11-30 | 1 | -0/+23 |
* | Implement (A&((~A)|B)) -> A&B transformation in the instruction combiner. This | Bill Wendling | 2008-11-30 | 1 | -0/+19 |
* | Forgot one remaining call to getSExtValue(). | Bill Wendling | 2008-11-30 | 1 | -1/+1 |
* | getSExtValue() doesn't work for ConstantInts with bitwidth > 64 bits. Use all | Bill Wendling | 2008-11-30 | 1 | -5/+5 |
* | Optimize memmove and memset into the LLVM builtins. Note that these | Eli Friedman | 2008-11-30 | 1 | -4/+57 |
* | Don't make TwoToExp signed by default. | Bill Wendling | 2008-11-30 | 1 | -2/+1 |
* | From Hacker's Delight: | Bill Wendling | 2008-11-30 | 1 | -8/+10 |
* | Instcombine was illegally transforming -X/C into X/-C when either X or C | Bill Wendling | 2008-11-30 | 1 | -3/+20 |
* | Fix a fixme by making memdep's handling of allocations more logical. | Chris Lattner | 2008-11-30 | 1 | -15/+12 |
* | Eliminate the dropInstruction method, which is not needed any more. | Chris Lattner | 2008-11-29 | 1 | -5/+2 |
* | Change MemDep::getNonLocalDependency to return its results as | Chris Lattner | 2008-11-29 | 1 | -5/+5 |
* | reimplement getNonLocalDependency with a simpler worklist | Chris Lattner | 2008-11-29 | 1 | -4/+21 |
* | Fix a thinko that manifested as a crash on clamav last night. | Chris Lattner | 2008-11-29 | 1 | -2/+2 |
* | Split getDependency into getDependency and getDependencyFrom, the | Chris Lattner | 2008-11-29 | 2 | -2/+2 |
* | Temporarily revert r60195. It's causing an optimized bootstrap of llvm-gcc to... | Bill Wendling | 2008-11-29 | 1 | -20/+21 |
* | Introduce and use a new MemDepResult class to hold the results of a memdep | Chris Lattner | 2008-11-29 | 3 | -54/+44 |
* | Reimplement the internal abstraction used by MemDep in terms | Chris Lattner | 2008-11-29 | 3 | -52/+56 |
* | don't revisit instructions off the beginning of the block. | Chris Lattner | 2008-11-28 | 1 | -2/+4 |
* | simplify some code, remove escaped newline. | Chris Lattner | 2008-11-28 | 1 | -22/+28 |
* | don't call MergeBasicBlockIntoOnlyPred on a block whose only | Chris Lattner | 2008-11-28 | 2 | -11/+14 |
* | rewrite RecursivelyDeleteTriviallyDeadInstructions to use a more efficient | Chris Lattner | 2008-11-28 | 1 | -16/+26 |
* | remove some weirdness that came from the LSR code that has | Chris Lattner | 2008-11-28 | 1 | -8/+1 |
* | rewrite a big chunk of how DSE does recursive dead operand | Chris Lattner | 2008-11-28 | 1 | -181/+105 |
* | delete ErasePossiblyDeadInstructionTree, replacing uses of it with | Chris Lattner | 2008-11-27 | 1 | -41/+9 |
* | Simplify LoopStrengthReduce::DeleteTriviallyDeadInstructions by | Chris Lattner | 2008-11-27 | 1 | -21/+20 |
* | enhance RecursivelyDeleteTriviallyDeadInstructions to make | Chris Lattner | 2008-11-27 | 1 | -0/+9 |
* | Enhance RecursivelyDeleteTriviallyDeadInstructions to optionally | Chris Lattner | 2008-11-27 | 1 | -7/+15 |
* | use continue to reduce indentation | Chris Lattner | 2008-11-27 | 1 | -18/+19 |