Programming Languages Research Group: Git

2009-12-07	Evan Cheng	If BB is empty, insert PHI before end() instead of...	commit \| commitdiff \| tree
2009-12-04	Evan Cheng	Handle recursive PHI's.	commit \| commitdiff \| tree
2009-12-04	Evan Cheng	Add a pre-regalloc tail duplication pass.	commit \| commitdiff \| tree
2009-12-04	Evan Cheng	Don't try to be cute with undef optimization here....	commit \| commitdiff \| tree
2009-12-04	Evan Cheng	- If the reaching definition is an undef and the use...	commit \| commitdiff \| tree
2009-12-03	Evan Cheng	Handle undef values properly.	commit \| commitdiff \| tree
2009-12-03	Evan Cheng	Watch out for PHI instruction with no source operands.	commit \| commitdiff \| tree
2009-12-03	Evan Cheng	Teach tail duplication to update SSA form. Work in...	commit \| commitdiff \| tree
2009-12-03	Evan Cheng	Fill out codegen SSA updater. It's not yet tested.	commit \| commitdiff \| tree
2009-12-02	Evan Cheng	Skeleton for MachineInstr level SSA updater.	commit \| commitdiff \| tree
2009-12-01	Evan Cheng	Fix PR5391: support early clobber physical register...	commit \| commitdiff \| tree
2009-12-01	Evan Cheng	Fix PR5614: parts of a physical register def may be...	commit \| commitdiff \| tree
2009-11-26	Evan Cheng	Test for 89905.	commit \| commitdiff \| tree
2009-11-26	Evan Cheng	When all defs of a vr are implicit_def, delete all...	commit \| commitdiff \| tree
2009-11-25	Evan Cheng	ProcessImplicitDefs should watch out for invalidated...	commit \| commitdiff \| tree
2009-11-24	Evan Cheng	Enable predication of NEON instructions in Thumb2 mode.	commit \| commitdiff \| tree
2009-11-24	Evan Cheng	Data type suffix must come after predicate.	commit \| commitdiff \| tree
2009-11-23	Evan Cheng	Massive refactoring of NEON instructions. Separate...	commit \| commitdiff \| tree
2009-11-21	Evan Cheng	Add predicate operand to NEON instructions. Fix lots...	commit \| commitdiff \| tree
2009-11-21	Evan Cheng	Allow target to disable if-converting predicable instruction...	commit \| commitdiff \| tree
2009-11-21	Evan Cheng	Maintain stylistic consistency.	commit \| commitdiff \| tree
2009-11-20	Evan Cheng	Enable hoisting load from constant memories.	commit \| commitdiff \| tree
2009-11-20	Evan Cheng	Remat VLDRD from constpool. Clean up some instruction...	commit \| commitdiff \| tree
2009-11-20	Evan Cheng	Add option -licm-const-load to hoist all loads from...	commit \| commitdiff \| tree
2009-11-20	Evan Cheng	Also CSE non-pic load from constant pools.	commit \| commitdiff \| tree
2009-11-20	Evan Cheng	Fix codegen of conditional move of immediates. We were...	commit \| commitdiff \| tree
2009-11-19	Evan Cheng	Refactor cmov selection code out to a separate function...	commit \| commitdiff \| tree
2009-11-19	Evan Cheng	80 col violation.	commit \| commitdiff \| tree
2009-11-19	Evan Cheng	More consistent thumb1 asm printing.	commit \| commitdiff \| tree
2009-11-19	Evan Cheng	Shrink ldr / str [sp, imm0-1024] to 16-bit instructions.	commit \| commitdiff \| tree
2009-11-19	Evan Cheng	Eliminate more * 4 in Thumb1 asm printing for consistency...	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Add a WriteAsOperand for MachineBasicBlock so MachineLoopInf...	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Fix comment.	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Generalize OptimizeLoopTermCond to optimize more loop...	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Revert 89021. It's miscompiling llvm-gcc driver driver...	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Re-apply 89011. It's not to be blamed.	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	Revert 89011. Buildbot thinks it might be breaking...	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	MOV64rm should be marked isReMaterializable.	commit \| commitdiff \| tree
2009-11-17	Evan Cheng	A few more instructions that should be marked re-materializable.	commit \| commitdiff \| tree
2009-11-16	Evan Cheng	- Check memoperand alignment instead of checking stack...	commit \| commitdiff \| tree
2009-11-16	Evan Cheng	Special case FixedStackPseudoSourceValueVal as well...	commit \| commitdiff \| tree
2009-11-16	Evan Cheng	Check if subreg index is zero.	commit \| commitdiff \| tree
2009-11-16	Evan Cheng	For some targets, a copy can use a register multiple...	commit \| commitdiff \| tree
2009-11-16	Evan Cheng	xfail for now. It has been failing.	commit \| commitdiff \| tree
2009-11-14	Evan Cheng	Added getSubRegIndex(A,B) that returns subreg index...	commit \| commitdiff \| tree
2009-11-14	Evan Cheng	- Change TargetInstrInfo::reMaterialize to pass in...	commit \| commitdiff \| tree
2009-11-14	Evan Cheng	Add radar number.	commit \| commitdiff \| tree
2009-11-14	Evan Cheng	Fix PR5412: Fix an inverted check and another missing...	commit \| commitdiff \| tree
2009-11-14	Evan Cheng	When expanding t2STRDi8 r, r to two stores, add kill...	commit \| commitdiff \| tree
2009-11-13	Evan Cheng	Fix PR5411. Bug in UpdateKills. A reg def partially...	commit \| commitdiff \| tree
2009-11-13	Evan Cheng	Fix PR5410: LiveVariables lost subreg def:	commit \| commitdiff \| tree
2009-11-12	Evan Cheng	RegScavenger::enterBasicBlock should always reset register...	commit \| commitdiff \| tree
2009-11-12	Evan Cheng	- Teach LSR to avoid changing cmp iv stride if it will...	commit \| commitdiff \| tree
2009-11-12	Evan Cheng	Use table to separate opcode from operands.	commit \| commitdiff \| tree
2009-11-12	Evan Cheng	isLegalICmpImmediate should take a signed integer;...	commit \| commitdiff \| tree
2009-11-11	Evan Cheng	Add TargetLowering::isLegalICmpImmediate. It tells...	commit \| commitdiff \| tree
2009-11-11	Evan Cheng	Add nounwind.	commit \| commitdiff \| tree
2009-11-11	Evan Cheng	Block terminator may be a switch.	commit \| commitdiff \| tree
2009-11-10	Evan Cheng	Generalize lsr code that optimize loop to count down...	commit \| commitdiff \| tree
2009-11-10	Evan Cheng	Change Thumb1 address mode printing, instead of	commit \| commitdiff \| tree
2009-11-10	Evan Cheng	Add a comment.	commit \| commitdiff \| tree
2009-11-09	Evan Cheng	Hide a couple of options.	commit \| commitdiff \| tree
2009-11-09	Evan Cheng	80 col.	commit \| commitdiff \| tree
2009-11-08	Evan Cheng	Refactor code.	commit \| commitdiff \| tree
2009-11-07	Evan Cheng	t2ldrpci_pic can be used for blockaddress as well.	commit \| commitdiff \| tree
2009-11-07	Evan Cheng	Missed this.	commit \| commitdiff \| tree
2009-11-07	Evan Cheng	Refactor code. Fix a potential missing check. Teach...	commit \| commitdiff \| tree
2009-11-07	Evan Cheng	- Add TargetInstrInfo::isIdentical(). It's similar...	commit \| commitdiff \| tree
2009-11-06	Evan Cheng	- Add pseudo instructions tLDRpci_pic and t2LDRpci_pic...	commit \| commitdiff \| tree
2009-11-06	Evan Cheng	Remove ARMPCLabelIndex from ARMISelLowering. Use ARMFunction...	commit \| commitdiff \| tree
2009-11-05	Evan Cheng	Now that code placement optimization pass is run for...	commit \| commitdiff \| tree
2009-11-05	Evan Cheng	Code refactoring.	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	Look for llvm-gcc under /Developer/usr/bin first.	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	RangeIsDefinedByCopyFromReg() should check for subreg_to_reg...	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	The .n suffix must go after the predicate.	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	Silence implicit conversion warnings.	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	Fix test.	commit \| commitdiff \| tree
2009-11-04	Evan Cheng	Use ldr.n to workaround a darwin assembler bug.	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Fix t2Int_eh_sjlj_setjmp. Immediate form of orr is...	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	fconsts / fconstd immediate should be proceeded with #.	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Re-apply 85799. It turns out my code isn't buggy.	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Trim unnecessary include.	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Fix PR5367. QPR_8 is the super regclass of DPR_8 and...	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Clean up copyRegToReg.	commit \| commitdiff \| tree
2009-11-03	Evan Cheng	Add QPR_8 as a superreg class of SPR_8 and DPR_8.	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	Revert 85799 for now. It might be breaking llvm-gcc...	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	Initilize the machine LICM CSE map upon the first time...	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	These are done / no longer care.	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	Add an entry.	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	Remove an irrelevant and poorly reduced test case.	commit \| commitdiff \| tree
2009-11-02	Evan Cheng	Unbreak ARMBaseRegisterInfo::copyRegToReg.	commit \| commitdiff \| tree
2009-11-01	Evan Cheng	Add PseudoSourceValue::mayAlias. It returns true if...	commit \| commitdiff \| tree
2009-11-01	Evan Cheng	Fix a couple more places where we are creating ld ...	commit \| commitdiff \| tree
2009-11-01	Evan Cheng	Make use of imm12 version of Thumb2 ldr / str instructions...	commit \| commitdiff \| tree
2009-11-01	Evan Cheng	Fix tests.	commit \| commitdiff \| tree
2009-10-31	Evan Cheng	Use cbz and cbnz instructions.	commit \| commitdiff \| tree
2009-10-31	Evan Cheng	It's safe to remat t2LDRpci; Add PseudoSourceValue...	commit \| commitdiff \| tree
2009-10-31	Evan Cheng	Revert 85634. It's breaking consumer-typeset (and others).	commit \| commitdiff \| tree
2009-10-30	Evan Cheng	Add option to createGVNPass to disable PRE.	commit \| commitdiff \| tree
2009-10-30	Evan Cheng	I forgot to commit this test.	commit \| commitdiff \| tree
next

C/C++ LLVM-based compilers that forbids OOTA behaviors

RSS Atom