- db89809 Teach ProcessImplicitDefs to transform more COPY instructions into IMPLICIT_DEF (and subsequently eliminate them). This allows machine LICM to hoist IMPLICIT_DEF's. PR7620. by Evan Cheng · 15 years ago
- 7791080 change machinelicm to use MachineInstr::isSafeToMove. No by Chris Lattner · 15 years ago
- aa74260 Fix MachineLICM to actually visit inner loops. by Dan Gohman · 15 years ago
- 853d3fb Move PHIElimination's SplitCriticalEdge for MachineBasicBlocks out by Dan Gohman · 15 years ago
- 82e0a1a Doh. Machine LICM is re-initializing the CSE map over and over. Patch by Anna Zaks. rdar://8037934. by Evan Cheng · 15 years ago
- e6cd757 Teach MachineLICM and MachineSink how to clear kill flags conservatively by Dan Gohman · 15 years ago
- 9196ab6 When MachineLICM is hoisting a physical register after regalloc, make sure the by Jakob Stoklund Olesen · 15 years ago
- 94d1d9c Postra machine licm must add registers defined by loop invariants to *all* of by Evan Cheng · 15 years ago
- 6327537 Fast path implicit_def check. by Evan Cheng · 15 years ago
- c15d913 Avoid variable shadowing. by Evan Cheng · 15 years ago
- aeb2f4a Expand postra machine licm's capability a little more. If an instruction's register operands are all loop invariants, then it's safe to hoist it. by Evan Cheng · 15 years ago
- 5dc57ce Teach postra machine licm to hoist more obvious invariants, e.g. instructions with no source operands. by Evan Cheng · 15 years ago
- 81bf03e Eliminate MachineBasicBlock::const_livein_iterator and make by Dan Gohman · 15 years ago
- 678d9b7 Plug trivial leak. by Benjamin Kramer · 15 years ago
- 7d597b4 Delete this obsolete comment. by Dan Gohman · 15 years ago
- 4038f9c Make post regalloc machine licm functional. It now passes all of MultiSource. by Evan Cheng · 15 years ago
- 5b46390 Add comments for missed opportunities. by Evan Cheng · 15 years ago
- 348856e Fix typo. by Evan Cheng · 15 years ago
- d94671a Post regalloc LICM. Work in progress. by Evan Cheng · 15 years ago
- 506049f - Change MachineInstr::isIdenticalTo to take a new option that determines whether it should skip checking defs or at least virtual register defs. This subsumes part of the TargetInstrInfo::isIdentical functionality. by Evan Cheng · 15 years ago
- 6ac33b4 Don't try to replace physical registers when doing CSE. by Dan Gohman · 15 years ago
- a363a9b Don't unconditionally suppress hoisting of instructions with implicit by Dan Gohman · 15 years ago
- 518bb53 move target-independent opcodes out of TargetInstrInfo by Chris Lattner · 16 years ago
- 65a41eb by David Greene · 16 years ago
- 92329c7 Add Loop contains utility methods for testing whether a loop by Dan Gohman · 16 years ago
- c26abd9 Enable hoisting load from constant memories. by Evan Cheng · 16 years ago
- 87b75ba Add option -licm-const-load to hoist all loads from constant memory. by Evan Cheng · 16 years ago
- 324da76 Add MachineBasicBlock::getName, and use it in place of getBasicBlock()->getName. by Jakob Stoklund Olesen · 16 years ago
- 7adcdc3 Fix comment. by Evan Cheng · 16 years ago
- 78e5c11 - Add TargetInstrInfo::isIdentical(). It's similar to MachineInstr::isIdentical by Evan Cheng · 16 years ago
- 9fb744e Code refactoring. by Evan Cheng · 16 years ago
- 777c6b7 Re-apply 85799. It turns out my code isn't buggy. by Evan Cheng · 16 years ago
- 454ac89 Revert 85799 for now. It might be breaking llvm-gcc driver. by Evan Cheng · 16 years ago
- 72ed88f Initilize the machine LICM CSE map upon the first time an instruction is hoisted to by Evan Cheng · 16 years ago
- 0115e16 Fix MachineLICM to use the correct virtual register class when by Dan Gohman · 16 years ago
- 5c95230 Refactor the code for unfolding a load into a separate function. by Dan Gohman · 16 years ago
- b8457a1 Simplify this code: if the unfolded load can't be hoisted, just delete by Dan Gohman · 16 years ago
- 589f1f5 Teach MachineLICM to unfold loads from constant memory from by Dan Gohman · 16 years ago
- f5a86f4 Remove includes of Support/Compiler.h that are no longer needed after the by Nick Lewycky · 16 years ago
- 6726b6d Remove VISIBILITY_HIDDEN from class/struct found inside anonymous namespaces. by Nick Lewycky · 16 years ago
- a70dca1 Factor out LiveIntervalAnalysis' code to determine whether an instruction by Dan Gohman · 16 years ago
- 599a87a isTriviallyReMaterializable checks the by Dan Gohman · 16 years ago
- 64efb55 Give Dan and my recent changes, machine LICM is now code size neutral. by Evan Cheng · 16 years ago
- 3dc326b Fix a logic error that caused non-rematable loop invariants loads to be licm'ed out of loop. by Evan Cheng · 16 years ago
- e33f44c Replace TargetInstrInfo::isInvariantLoad and its target-specific by Dan Gohman · 16 years ago
- 45094e3 Don't hoist or sink instructions with physreg uses if the physreg is by Dan Gohman · 16 years ago
- a8fb336 Unbreak MachineLICM for instructions that reference RIP on x86-64 too. by Dan Gohman · 16 years ago
- b7a8992 Convert DOUT to DEBUG(errs()...). by Bill Wendling · 16 years ago
- ffef8ac Tidy #includes. by Dan Gohman · 16 years ago
- ce63ffb More migration to raw_ostream, the water has dried up around the iostream hole. by Daniel Dunbar · 16 years ago
- efc7839 MachineLICM CSE should match destination register classes; avoid hoisting implicit_def's. by Evan Cheng · 16 years ago
- 740854b Machine LICM increases register pressure and it almost always increase code size. For now, disable it for optimizing for size. by Evan Cheng · 17 years ago
- af6949d Teach machine licm to CSE hoisted instructions. by Evan Cheng · 17 years ago
- 5caa883 Don't call isInvariantLoad twice. by Evan Cheng · 17 years ago
- 45e94d6 Don't bother hoisting out a "cheap" instruction if all of its uses are PHIs. LICM "cheap" instructions are not particularly beneficial to start with. This will just end up making the copies harder to coalesce. by Evan Cheng · 17 years ago
- 9b61f33 For now, only hoist re-materilizable instructions. LICM will increase register pressure. We want to avoid spilling more instructions if it's possible. by Evan Cheng · 17 years ago
- c475c36 Simplify the MachineLICM pass by having it only traverse outer by Dan Gohman · 17 years ago
- d68a076 Tidy up #includes, deleting a bunch of unnecessary #includes. by Dan Gohman · 17 years ago
- 237dee1 Use isTerminator() instead of isBranch()||isReturn() in by Dan Gohman · 17 years ago
- 7224170 Give MachineLICM a name, for -time-passes etc. by Dan Gohman · 17 years ago
- d735b80 Switch the MachineOperand accessors back to the short names like by Dan Gohman · 17 years ago
- ae73dc1 Tidy up several unbeseeming casts from pointer to intptr_t. by Dan Gohman · 17 years ago
- acb04ec Cosmetic changes to Machine LICM. No functionality change. by Bill Wendling · 17 years ago
- fb018d0 Don't hoist instructions that define a physical register. by Bill Wendling · 17 years ago
- d3361e9 Don't hoist instructions that have implicit uses or defines. Before, it was by Bill Wendling · 17 years ago
- 8870ce9 Prevent option name conflict. by Bill Wendling · 17 years ago
- 844731a Clean up the use of static and anonymous namespaces. This turned up by Dan Gohman · 17 years ago
- c1b7f52 Revert the previous commit. Go ahead and hoist rematerializable instructions. by Bill Wendling · 17 years ago
- e4fc1cc One real change - don't hoist something that's trivially rematerializable. It's by Bill Wendling · 17 years ago
- 074223a Minor cleanup. No functionality change. by Bill Wendling · 17 years ago
- e6d088a Rename PrintableName to Name. by Bill Wendling · 17 years ago
- 74ab84c Change "Name" to "AsmName" in the target register info. Gee, a refactoring tool by Bill Wendling · 17 years ago
- 6f0d024 Rename MRegisterInfo to TargetRegisterInfo. by Dan Gohman · 18 years ago
- a22edc8 Simplify the side effect stuff a bit more and make licm/sinking by Chris Lattner · 18 years ago
- 749c6f6 rename TargetInstrDescriptor -> TargetInstrDesc. by Chris Lattner · 18 years ago
- 6924430 Rename MachineInstr::getInstrDescriptor -> getDesc(), which reflects by Chris Lattner · 18 years ago
- d5da704 Add that this preserves some analyses. by Bill Wendling · 18 years ago
- cc8f603 Move option to enable machine LICM into LLVMTargetMachine.cpp. by Bill Wendling · 18 years ago
- b082c6f Call the parent's getAnalysisUsage. by Bill Wendling · 18 years ago
- ac69582 remove dead #includes and reorder the rest. by Chris Lattner · 18 years ago
- dde059a Use the correct MachineRegisterInfo object. by Bill Wendling · 18 years ago
- a4b6622 Remove dead code. by Bill Wendling · 18 years ago
- 9258cd3 Use the new architecture to get the containing machine basic block for a machine by Bill Wendling · 18 years ago
- 4ee451d Remove attribution from file headers, per discussion on llvmdev. by Chris Lattner · 18 years ago
- 60ff1a3 Updated comments to reflect what "side effects" means in this situation. by Bill Wendling · 18 years ago
- 280f456 Add debugging info. Use the newly created "hasUnmodelledSideEffects" method. by Bill Wendling · 18 years ago
- efe2be7 Need to grow the indexed map. Added debug statements. by Bill Wendling · 18 years ago
- a17ad59 Simplify slightly. by Bill Wendling · 18 years ago
- 12ebf14 Blark! How in the world did this work without this?! by Bill Wendling · 18 years ago
- 28bd5f0 - Update the virtual reg to machine instruction map when hoisting. by Bill Wendling · 18 years ago
- 650b052 Checking for "zero operands" during the "CanHoistInst()" method isn't necessary by Bill Wendling · 18 years ago
- 041b3f8 Reverting 44702. It wasn't correct to rename them. by Bill Wendling · 18 years ago
- 320c630 Renaming: by Bill Wendling · 18 years ago
- b48519c Incorporated comments from Evan and Chris: by Bill Wendling · 18 years ago
- 0f940c9 Initial commit of the machine code LICM pass. It successfully hoists this: by Bill Wendling · 18 years ago