Only use clairvoyance when defining a register, and then only if it has one use.
[oota-llvm.git] / lib /
2010-05-17 Jakob Stoklund OlesenOnly use clairvoyance when defining a register, and...
2010-05-17 Jakob Stoklund OlesenEliminate a hash table probe when killing virtual regis...
2010-05-17 Jakob Stoklund OlesenExecute virtreg kills immediately instead of after...
2010-05-17 Jakob Stoklund OlesenSprinkle superregister <imp-def> and <imp-kill> operand...
2010-05-17 Jakob Stoklund OlesenNow that we don't keep live registers across calls...
2010-05-17 Jakob Stoklund OlesenBoldly attempt consistent capitalization. Functional...
2010-05-17 Eric ChristopherAssume that we'll handle mangling the symbols earlier...
2010-05-17 Jakob Stoklund OlesenSpill and kill all virtual registers across a call.
2010-05-17 Jakob Stoklund OlesenReduce hashtable probes by using DenseMap::insert(...
2010-05-17 Jakob Stoklund OlesenMake MBB a class member instead of passing it around...
2010-05-17 Evan ChengYes, if the redef is a copy, update the old val# with...
2010-05-17 Evan ChengNo reason not to run the NEON domain croassing fix...
2010-05-16 Dale JohannesenRevert 103911; it broke a test that expects bitconvert
2010-05-16 Dale JohannesenMake x86-64 64-bit bitconvert work when SSE is not...
2010-05-16 Anton KorobeynikovChris said that the comment char should be escaped...
2010-05-16 Anton KorobeynikovAdd support for thiscall calling convention.
2010-05-16 Anton KorobeynikovGeneralize the ARM DAG combiner of mul with constants...
2010-05-16 Evan ChengModel vst lane instructions with REG_SEQUENCE.
2010-05-15 Dale JohannesenFix uint64->{float, double} conversion to do rounding...
2010-05-15 Dale JohannesenImprove assertion messages.
2010-05-15 Anton KorobeynikovSome cheap DAG combine goodness for multiplication...
2010-05-15 Anton Korobeynikov"trap" pseudo-op turned out to be apple-local.
2010-05-15 Chris Lattnerimprove portability to systems that don't have round...
2010-05-15 Chris Lattnerimprove portability to systems that don't have powf...
2010-05-15 Chandler CarruthFix an GCC warning that seems to have actually caught...
2010-05-15 Evan ChengModel 128-bit vld lane with REG_SEQUENCE.
2010-05-15 Jakob Stoklund OlesenCalculate liveness on the fly for local registers.
2010-05-15 Nick LewyckyTeach the always inliner to release its inline cost...
2010-05-15 Nick LewyckyClean up, no functional change.
2010-05-15 Evan Chengv4i64 and v8i64 are only synthesizable when NEON is...
2010-05-15 Evan ChengAllow TargetLowering::getRegClassFor() to be called...
2010-05-15 Evan ChengModel 64-bit lane vld with REG_SEQUENCE.
2010-05-15 Evan ChengA partial re-def instruction may be a copy.
2010-05-14 Evan ChengTeach two-address pass to do some coalescing while...
2010-05-14 Evan ChengModel VST*_UPD and VST*oddUPD pair with REG_SEQUENCE.
2010-05-14 Dan GohmanFast ISel trivially coalesces away no-op casts, so...
2010-05-14 Jakob Stoklund OlesenDon't bother spilling before a return
2010-05-14 Jakob Stoklund OlesenRegAllocLocal can count copies too
2010-05-14 Bill WendlingSystemZ really does mean "has calls" and not just ...
2010-05-14 Jakob Stoklund OlesenTrack allocatable instead of reserved regs, and never...
2010-05-14 Dan GohmanDon't set kill flags for instructions which the schedul...
2010-05-14 Dan GohmanBR is a barrier.
2010-05-14 Bill WendlingSeveral tail call tests apparently rely upon this being...
2010-05-14 Jakob Stoklund OlesenAvoid scanning the long tail of physreg operands on...
2010-05-14 Devang PatelDo not forget to mark prcessed arguments.
2010-05-14 Jakob Stoklund OlesenCount coalesced copies
2010-05-14 Jakob Stoklund OlesenAllow virtreg redefines when verifying for RegAllocFast
2010-05-14 Bill WendlingThis should happen if there are no calls, not if it...
2010-05-14 Bill WendlingRevert r103804. The comment is correct.
2010-05-14 Jim GrosbachRemove trailing whitespace
2010-05-14 Jim Grosbach80 column and trailing whitespace cleanup
2010-05-14 Jim Grosbachadd cmd line option to leave dbgvalues in during post...
2010-05-14 Bill WendlingFix comment.
2010-05-14 Bill WendlingRename "HasCalls" in MachineFrameInfo to "AdjustsStack...
2010-05-14 Dan GohmanLowering of atomic instructions can result in operands...
2010-05-14 Devang PatelAdd support to preserve type info for the variables...
2010-05-14 Jakob Stoklund OlesenWhen verifying two-address instructions, check the...
2010-05-14 Kevin EnderbyFix so "int3" is correctly accepted, added "into" and...
2010-05-14 Evan ChengModel VLD*_UPD and VLD*odd_UPD pair with REG_SEQUENCE.
2010-05-14 Daniel DunbarMC/Mach-O/x86_64: Darwin's special "signed_N" relocatio...
2010-05-14 Jakob Stoklund OlesenSimplify the handling of physreg defs and uses in RegAl...
2010-05-14 Dan GohmanSet isTerminator on TRAP instructions.
2010-05-14 Dan GohmanDon't use isBarrier for the PowerPC sync instruction...
2010-05-14 Dan GohmanAdd mayLoad and mayStore flags to instructions which...
2010-05-14 Daniel DunbarXFAIL the test I added with vg_leak, apparently it...
2010-05-14 Daniel DunbarInline Asm: Ensure buffer is newline terminated to...
2010-05-14 Jakob Stoklund OlesenEnable opportunistic coalescing
2010-05-14 Evan ChengAdded a QQQQ register file to model 4-consecutive Q...
2010-05-14 Eric ChristopherAdd AsmParser support for darwin tbss directive.
2010-05-14 Eric ChristopherFix a couple of typos.
2010-05-14 Daniel DunbarMC: Switch to completely lazy layout.
2010-05-14 Daniel DunbarMC: Extend MCAsmLayout to explicitly track which fragme...
2010-05-14 Daniel DunbarMC: Implicitly assign section addresses when the previo...
2010-05-14 Daniel DunbarMC: Switch MCFragment to storing the layout order index...
2010-05-14 Daniel DunbarMC: Change LayoutSection() to only do the section initi...
2010-05-14 Evan ChengFix comments.
2010-05-14 Jakob Stoklund OlesenTrust kill flags from isel and later passes.
2010-05-14 Jakob Stoklund OlesenFix an embarrassing runtime regression for RegAllocFast.
2010-05-13 Evan ChengAdding a v8i64 512-bit vector type. This will be used...
2010-05-13 Dale JohannesenImplement a correct ui64->f32 conversion. The old
2010-05-13 Jakob Stoklund OlesenClean up RegAllocFast debug output
2010-05-13 Daniel DunbarMC: Move Layout{Fragment,Section} into MCAsmLayout...
2010-05-13 Dan GohmanTeach MachineLICM and MachineSink how to clear kill...
2010-05-13 Evan ChengAdd comment about the pseudo registers QQ, each of...
2010-05-13 Bob WilsonFix pr7110: For non-Darwin targets UnspilledCS1GPRs...
2010-05-13 Oscar FuentesCMake: fixes 64 bit Visual Studio IDE build. Fixes...
2010-05-13 Dan GohmanAdd a utility function for conservatively clearing...
2010-05-13 Dan GohmanAn Instruction has a trivial kill only if its use is...
2010-05-13 Daniel DunbarMC: Factor out MCAssembler::ComputeFragmentSize.
2010-05-13 Daniel DunbarMC: Add section layout order indices to MCSectionData.
2010-05-13 Daniel DunbarMC: Move ordinal calculation, to make sure fragments...
2010-05-13 Daniel DunbarMC: Create dummy fragments to avoid ever having empty...
2010-05-13 Daniel DunbarMC: Add MCAsmLayout::FragmentReplaced() helper function.
2010-05-13 Anton KorobeynikovProperly set thread-local flag on globals during cpp...
2010-05-13 Nick LewyckyRemove heinous tabs.
2010-05-13 Nick LewyckyReplace the core comparison login in merge functions...
2010-05-13 Daniel DunbarFix -Asserts warning.
2010-05-13 Daniel DunbarMC: Eliminate MCSectionData::{,Address,File}Size, which...
2010-05-13 Daniel DunbarFix -Asserts warning.
2010-05-13 Daniel DunbarMC: Switch to using explicit MCAlignFragments with...
next