- d7f0810 Enable predication of NEON instructions in Thumb2 mode. by Evan Cheng · 15 years ago
- ac0869d Add predicate operand to NEON instructions. Fix lots (but not all) 80 col violations in ARMInstrNEON.td. by Evan Cheng · 15 years ago
- 9b82425 Also CSE non-pic load from constant pools. by Evan Cheng · 15 years ago
- 834b08a Add a target hook to allow changing the tail duplication limit based on the by Bob Wilson · 15 years ago
- a443217 Detect need for autoalignment of the stack earlier to catch spills more by Jim Grosbach · 15 years ago
- 6cb6788 set the def of the VLD1q64 properly by Jim Grosbach · 15 years ago
- d57cdd5 - Change TargetInstrInfo::reMaterialize to pass in TargetRegisterInfo. by Evan Cheng · 15 years ago
- e516549 Use Unified Assembly Syntax for the ARM backend. by Jim Grosbach · 15 years ago
- 31bc849 Use aligned load/store instructions for spilling Q registers when we know the stack slot is 128 bit aligned by Jim Grosbach · 15 years ago
- fdc8340 Refactor code. by Evan Cheng · 15 years ago
- 31c24bf 80-column cleanup of file header comments by Jim Grosbach · 15 years ago
- d457e6e Refactor code. Fix a potential missing check. Teach isIdentical() about tLDRpci_pic. by Evan Cheng · 15 years ago
- 5a1cd36 Fix t2Int_eh_sjlj_setjmp. Immediate form of orr is a 32-bit instruction. So it should be 22 bytes instead of 20 bytes long. by Evan Cheng · 15 years ago
- f6c0bff Trim unnecessary include. by Evan Cheng · 15 years ago
- b4db6a4 Clean up copyRegToReg. by Evan Cheng · 15 years ago
- 7aaf94b Turn neon reg-reg moves fixup code into separate pass. This should reduce the compile time. by Anton Korobeynikov · 15 years ago
- 7baae87 Unbreak ARMBaseRegisterInfo::copyRegToReg. by Evan Cheng · 15 years ago
- f95215f Use NEON reg-reg moves, where profitable. This reduces "domain-cross" stalls, when we used to mix vfp and neon code (the former were used for reg-reg moves) by Anton Korobeynikov · 15 years ago
- 8d4de5a Add a Thumb BRIND pattern. Change the ARM BRIND assembly to separate the by Bob Wilson · 15 years ago
- ed3ad21 Don't forget subreg indices when folding load / store. by Evan Cheng · 15 years ago
- 5a850be 80 col violation. by Evan Cheng · 15 years ago
- ff89dcb -Revert parts of 84326 and 84411. Distinquishing between fixed and non-fixed by Evan Cheng · 15 years ago
- 491f54f Distinquish stack slots from other stack objects. They (and fixed objects) get FixedStack PseudoSourceValues. by Evan Cheng · 15 years ago
- 6553155 Revert 84315 for now. Re-thinking the patch. by Evan Cheng · 15 years ago
- bf12558 Rename getFixedStack to getStackObject. The stack objects represented are not by Evan Cheng · 15 years ago
- 249fb33 Add PseudoSourceValues for constpool stuff on ELF (Darwin should use something similar) by Anton Korobeynikov · 15 years ago
- 26207e5 Introduce the TargetInstrInfo::KILL machine instruction and get rid of the by Jakob Stoklund Olesen · 15 years ago
- 5adb66a Make ARM and Thumb2 32-bit immediate materialization into a single 32-bit pseudo by Evan Cheng · 15 years ago
- e56f908 Add QPR_VFP2 regclass and add copy_to_regclass nodes, where needed to by Anton Korobeynikov · 15 years ago
- 6ca0b9e Add NEON 'laned' operations. This fixes another bunch of gcc testsuite fails and by Anton Korobeynikov · 15 years ago
- cdbb3f5 Fix PR4789. Teach eliminateFrameIndex how to handle VLDRQ and VSTRQ which cannot fold any immediate offset. by Evan Cheng · 15 years ago
- 33adcfb rename TAI -> MAI, being careful not to make MAILJMP instructions :) by Chris Lattner · 15 years ago
- af76e59 Rename TargetAsmInfo (and its subclasses) to MCAsmInfo. by Chris Lattner · 15 years ago
- 24f20e0 Record variable debug info at ISel time directly. by Devang Patel · 15 years ago
- 5aa1684 Add Thumb2 eh_sjlj_setjmp implementation by Jim Grosbach · 15 years ago
- cdc17eb fix GetInstSizeInBytes for eh_sjlj_setjmp by Jim Grosbach · 15 years ago
- 764ab52 Whitespace cleanup. Remove trailing whitespace. by Jim Grosbach · 15 years ago
- 19068ba Add support for folding loads / stores into 16-bit moves used by Thumb2. by Evan Cheng · 15 years ago
- 2294645 80 col violation. by Evan Cheng · 15 years ago
- baf3108 Use VLDM / VSTM to spill/reload 128-bit Neon registers by Anton Korobeynikov · 15 years ago
- 8fb9036 Code refactoring. No functionality change. by Evan Cheng · 15 years ago
- 1d2426c Fix support to use NEON for single precision fp math. by Evan Cheng · 15 years ago
- 8619864 It turns out most of the thumb2 instructions are not allowed to touch SP. The semantics of such instructions are unpredictable. We have just been lucky that tests have been passing. by Evan Cheng · 15 years ago
- 7bfdca0 When using NEON for single-precision FP, the NEON result must be placed in D0-D15 as these are the only D registers with S subregs. Introduce a new regclass to represent D0-D15 and use it in the NEON single-precision FP patterns. by David Goodwin · 15 years ago
- d90183d Move the getInlineAsmLength virtual method from TAI to TII, where by Chris Lattner · 15 years ago
- 25f7cfc Workaround a couple of Darwin assembler bugs. by Evan Cheng · 15 years ago
- a0ee862 t2BR_JT is mov pc, it's 2 byte long, not 4. by Evan Cheng · 15 years ago
- d26b14c - Teach TBB / TBH offset limits are 510 and 131070 respectively since the offset by Evan Cheng · 15 years ago
- 6495f63 - More refactoring. This gets rid of all of the getOpcode calls. by Evan Cheng · 15 years ago
- 78703dd convertToThreeAddress can't handle Thumb2 instructions (which don't have same address mode as ARM instructions). by Evan Cheng · 15 years ago
- 83e0e36 Clean up. by Evan Cheng · 15 years ago
- 5ca53a7 Get rid of some more getOpcode calls. by Evan Cheng · 15 years ago
- 1f5c988 If CPSR is modified but the def is dead, then it's ok to fold the load / store. by Evan Cheng · 15 years ago
- 5732ca0 Use t2LDRi12 and t2STRi12 to load / store to / from stack frames. Eliminate more getOpcode calls. by Evan Cheng · 15 years ago
- 08b93c6 Use the right instructions to copy between GPR and the more strictive tGPR classes. t2MOV does not match the RC requirements. by Evan Cheng · 15 years ago
- dced03f Merge isLoadFromStackSlot into one since it behaves the same regardless of sub-target. by Evan Cheng · 15 years ago
- 68e3c6a Just use a single isMoveInstr to catch all the cases. by Evan Cheng · 15 years ago
- 66ac531 Change Thumb2 jumptable codegen to one that uses two level jumps: by Evan Cheng · 15 years ago
- 7894762 Make sure thumb2 jumptable entries are aligned. by Evan Cheng · 15 years ago
- 23ed527 Remove unused member functions. by Eli Friedman · 15 years ago
- b74bb1a FLDD, FLDS, FCPYD, FCPYS, FSTD, FSTS, VMOVD, VMOVQ maps to the same instructions on all sub-targets. by Evan Cheng · 15 years ago
- 5ff58b5 Correctly handle the Thumb-2 imm8 addrmode. Specialize frame index elimination more exactly for Thumb-2 to get better code gen. by David Goodwin · 15 years ago
- b8e9ac8 Emit cross regclass register moves for thumb2. Minor code duplication cleanup. by Anton Korobeynikov · 15 years ago
- c23197a llvm_unreachable->llvm_unreachable(0), LLVM_UNREACHABLE->llvm_unreachable. by Torok Edwin · 15 years ago
- c25e758 assert(0) -> LLVM_UNREACHABLE. by Torok Edwin · 15 years ago
- dd6f632 80 col violation. by Evan Cheng · 15 years ago
- ab33150 Move isPredicated from .cpp to .h by Evan Cheng · 15 years ago
- e7cbe41 Change how so_imm and t2_so_imm are handled. At instruction selection time, the immediates are no longer encoded in the imm8 + rot format, that are left as it is. The encoding is now done in ams printing and code emission time instead. by Evan Cheng · 15 years ago
- 334c264 Checkpoint Thumb2 Instr info work. Generalized base code so that it can be shared between ARM and Thumb2. Not yet activated because register information must be generalized first. by David Goodwin · 15 years ago