- 0ac5d2d Revert turning copysignl into a COPYSIGN node for the moment: by Duncan Sands · 15 years ago
- bb2f0c5 Now that it's supported, turn copysignl into a COPYSIGN node. by Duncan Sands · 15 years ago
- 5d54b41 Fix PR6522: implement copysign expansion for x86 long double by Duncan Sands · 15 years ago
- b3e87b2 fix PR6577, a bug in sdbuilder lowering select instructions by Chris Lattner · 15 years ago
- a257095 Remove getWidenVectorType, which is no longer used. by Dan Gohman · 15 years ago
- af1871f In case of tail call size of Ins and InVals may not match. by Evan Cheng · 15 years ago
- 0193c09 Remove dead include. by Daniel Dunbar · 15 years ago
- f2f64e9 fix PR6533 by updating the br(xor) code to remember the case by Chris Lattner · 15 years ago
- e3b8533 Cosmetic: lengthen names and improve comments. by Dale Johannesen · 15 years ago
- bfdf7f3 Progress towards shepherding debug info through SelectionDAG. by Dale Johannesen · 15 years ago
- 8f78e3c Fix another bitwidth calculation to handle vector types; based on a by Dan Gohman · 15 years ago
- 19b3826 Attempt to make this debug output meaningful, both in the case of by Dan Gohman · 15 years ago
- 8f03d75 Print the correct index in the "match failed at index" message. by Dan Gohman · 15 years ago
- fab4a25 Add Order to SDDbgValue by Dale Johannesen · 15 years ago
- acee647 Use Other as a sentinel instead of iAny. by Chris Lattner · 15 years ago
- 06a2663 Add some new bits of debug info handling. No functional change yet. by Dale Johannesen · 15 years ago
- 0fa9d1d Reapply r97778 and r97779, enabled only for unsigned i64 to f64 conversions. by Dan Gohman · 15 years ago
- 580bba2 Avoid creating bad PHI instructions when BR is being const-folded. by Jakob Stoklund Olesen · 15 years ago
- 18fdaba Fix PR6497, a bug where we'd fold a load into an addc by Chris Lattner · 15 years ago
- 9bbcd5e inline a small function with one call site. by Chris Lattner · 15 years ago
- b6b343d Revert r97778 and r97779. They're somehow breaking llvm-gcc builds. by Dan Gohman · 15 years ago
- 808ab34 Fix these constants to be more portable. by Dan Gohman · 15 years ago
- cfa7ca1 Rewrite i64-to-f64 conversion using an algorithm which handles by Dan Gohman · 15 years ago
- c19ae9d add a statistic for # times fastisel fails. by Chris Lattner · 15 years ago
- ba72b0c Fix a typo Duncan noticed. by Dan Gohman · 15 years ago
- 57bf8a4 change the new isel matcher to emit ComplexPattern matches by Chris Lattner · 15 years ago
- 6900a39 Fix more code to work properly with vector operands. Based on by Dan Gohman · 15 years ago
- 2c4afd1 inline CannotYetSelectIntrinsic into CannotYetSelect and simplify. by Chris Lattner · 15 years ago
- a72d2a2 Fix a bug in SelectionDAG's ReplaceAllUsesWith in the case where by Dan Gohman · 15 years ago
- da828e3 add some of the more obscure predicate types to the Scope accelerator. by Chris Lattner · 15 years ago
- f1b7c7d speed up scope node processing: if the first element of a scope by Chris Lattner · 15 years ago
- cfe2eab introduce a new SwitchTypeMatcher node (which is analogous to by Chris Lattner · 15 years ago
- 32f9eb2 Use APInt instead of zext value. by Bill Wendling · 15 years ago
- 7d9f2b9 This test case: by Bill Wendling · 15 years ago
- d1b7382 Fix some issues in WalkChainUsers dealing with by Chris Lattner · 15 years ago
- 00592ec run HandleMergeInputChains even if we only have one input chain. by Chris Lattner · 15 years ago
- 82dd3d3 Fix the xfail I added a couple of patches back. The issue by Chris Lattner · 15 years ago
- 61c97f6 I was confused about this, it turns out that MorphNodeTo by Chris Lattner · 15 years ago
- 3ff1e4d factor node morphing out to its own helper method. by Chris Lattner · 15 years ago
- 7c306da Sink InstructionSelect() out of each target into SDISel, and rename it by Chris Lattner · 15 years ago
- 6183fbd Use the right induction variable. by Chris Lattner · 15 years ago
- c6d7ad3 Rewrite chain handling validation and input TokenFactor handling by Chris Lattner · 15 years ago
- 5b870af Fix several places to handle vector operands properly. by Dan Gohman · 15 years ago
- 46ada19 Remove dead parameter passing. by Bill Wendling · 15 years ago
- 319ef94 remove dead code. by Chris Lattner · 15 years ago
- 6b30792 refactor some code out of OPC_EmitMergeInputChains into a new helper function. by Chris Lattner · 15 years ago
- df51247 remove all but one version of SelectionDAG::MorphNodeTo by Chris Lattner · 15 years ago
- 7390eeb Accelerate isel dispatch for tables that start with a top-level by Chris Lattner · 15 years ago
- 042919c Fix optimization of ISD::TRUNCATE on vector operands. Based on a patch by Dan Gohman · 15 years ago
- 7d892d6 some trivial microoptimizations. by Chris Lattner · 15 years ago
- fa342fa eliminate the CheckMultiOpcodeMatcher code and have each by Chris Lattner · 15 years ago
- eb66921 add a new OPC_SwitchOpcode which is semantically equivalent by Chris Lattner · 15 years ago
- 14df8dc eliminate GetInt1/2 by Chris Lattner · 15 years ago
- 2a49d57 hoist the new isel interpreter out of DAGISelHeader.h by Chris Lattner · 15 years ago
- 21221e3 enhance the new isel to handle the 'node already exists' by Chris Lattner · 15 years ago
- 25d5837 simplify this code, return only ever has zero or one operands. by Chris Lattner · 15 years ago
- 2c755ba Re-apply 97040 with fix. This survives a ppc self-host llvm-gcc bootstrap. by Evan Cheng · 15 years ago
- 45df761 Move dbg_value generation to target-independent FastISel, by Dale Johannesen · 15 years ago
- 9949dd6 Fix ExpandVectorBuildThroughStack for the case where the by Dan Gohman · 15 years ago
- aa9d854 Revert r97064. Duncan pointed out that bitcasts are defined in by Dan Gohman · 15 years ago
- 5cf0b6e clean up various VT manipulations, patch by Micah Villmow! PR6337 by Chris Lattner · 15 years ago
- c382bc3 Make getTypeSizeInBits work correctly for array types; it should return by Dan Gohman · 15 years ago
- 46ca5ef convert cycle checker to smallptrset, add comments and make it more elegant. by Chris Lattner · 15 years ago
- a1461cc revert david's patch which does not even build. by Chris Lattner · 15 years ago
- 2f9b685 by David Greene · 15 years ago
- cfe30ef Speculatively revert r97011, "Re-apply 96540 and 96556 with fixes.", again in by Daniel Dunbar · 15 years ago
- e832693 When forming SSE min and max nodes for UGE and ULE comparisons, it's by Dan Gohman · 15 years ago
- 736a6ea Change the scheduler from adding nodes in allnodes order by Chris Lattner · 15 years ago
- 91ff7f7 add node #'s to debug dumps. by Chris Lattner · 15 years ago
- 97e6b83 Re-apply 96540 and 96556 with fixes. by Evan Cheng · 15 years ago
- 5857e0a make selectnodeto set the nodeid to -1. This makes it more akin to by Chris Lattner · 15 years ago
- da244a0 fix a bug in findNonImmUse (used by IsLegalToFold) where nodes with by Chris Lattner · 15 years ago
- 4548e02 Print node ID's in dumps and views if set. by Chris Lattner · 15 years ago
- e3d97c7 by David Greene · 15 years ago
- 3eba667 Revert commits 96556 and 96640, because commit 96556 breaks the by Duncan Sands · 15 years ago
- 97a35fc Transform (xor (setcc), (setcc)) == / != 1 to (xor (setcc), (setcc)) != / == 1. by Evan Cheng · 15 years ago
- 89bb7b5 Some dag combiner goodness: by Evan Cheng · 15 years ago
- 1157f79 by David Greene · 15 years ago
- 409ac58 sink special case "cannotyetselect" for intrinsics out of the by Chris Lattner · 15 years ago
- 1df9859 There are two ways of checking for a given type, for example isa<PointerType>(T) by Duncan Sands · 15 years ago
- f76de00 Fix a memory leak. Patch by Nicolas Geoffray. by Evan Cheng · 15 years ago
- 014bf21 Split SelectionDAGISel::IsLegalAndProfitableToFold to by Evan Cheng · 15 years ago
- 1e55944 by David Greene · 15 years ago
- b0bc6c3 Uniformize the names of type predicates: rather than having isFloatTy and by Duncan Sands · 15 years ago
- b5b90ed Use array_pod_sort instead of std::sort for improved code size. by Jakob Stoklund Olesen · 16 years ago
- 598b24c Remove duplicate successors from indirectbr instructions before building the machine CFG. by Jakob Stoklund Olesen · 16 years ago
- f7ea6c3 The previous fix of widening divides that trap was too fragile as it depends on custom by Mon P Wang · 16 years ago
- f451cb8 Fix "the the" and similar typos. by Dan Gohman · 16 years ago
- 046632f Now that ShrinkDemandedOps() is separated out from DAG combine. It sometimes leave some obvious nops which dag combine used to clean up afterwards e.g. (trunk (ext n)) -> n. Look for them and squash them. by Evan Cheng · 16 years ago
- 8112b53 Emit an error for illegal inline asm constraint (which uses illegal type) rather than asserting. by Evan Cheng · 16 years ago
- 10fedd2 Fix comments to reflect renaming elsewhere. by Dale Johannesen · 16 years ago
- 3914370 by David Greene · 16 years ago
- 518bb53 move target-independent opcodes out of TargetInstrInfo by Chris Lattner · 16 years ago
- 8ac38f2 Apply the 95471 fix to SelectionDAGBuilder as well; by Dale Johannesen · 16 years ago
- c2e93b2 In guaranteed tailcall mode, don't decline the tailcall optimization by Dan Gohman · 16 years ago
- dc91856 After Victor's latest commits I am seeing null by Dale Johannesen · 16 years ago
- 5ba8bf6 When the scheduler unfold a load folding instruction it move some of the predecessors to the unfolded load. It decides what gets moved to the load by checking whether the new load is using the predecessor as an operand. The check neglects the cases whether the predecessor is a flagged scheduling unit. by Evan Cheng · 16 years ago
- 6fdce65 Fix typo Duncan noticed. by Evan Cheng · 16 years ago
- 446bc10 It's too risky to eliminate sext / zext of call results for tail call optimization even if the caller / callee attributes completely match. The callee may have been bitcast'ed (or otherwise lied about what it's doing). by Evan Cheng · 16 years ago
- febc816 Revert 94937 and move the noreturn check to codegen. by Evan Cheng · 16 years ago