- 018f9b0 - Mark last use of a split interval as kill instead of letting spiller track it. by Evan Cheng · 18 years ago
- d64b5c8 Add a argument to storeRegToStackSlot and storeRegToAddr to specify whether by Evan Cheng · 18 years ago
- dd86ce3 Remove a unsafe optimization. This fixes 401.bzip2. by Evan Cheng · 18 years ago
- 35a3e4a Spiller unfold optimization bug: do not clobber a reusable stack slot value unless it can be modified. by Evan Cheng · 18 years ago
- e4b3900 Bug fixes. by Evan Cheng · 18 years ago
- 70306f8 Update kill info for uses of split intervals. by Evan Cheng · 18 years ago
- aee4af6 Remove redundant foldMemoryOperand variants and other code clean up. by Evan Cheng · 18 years ago
- 0cbb116 Fixed various live interval splitting bugs / compile time issues. by Evan Cheng · 18 years ago
- cada245 Recover compile time regression. by Evan Cheng · 18 years ago
- 81a0382 Live interval splitting: by Evan Cheng · 18 years ago
- c498b02 Clean up sub-register implementation by moving subReg information back to by Evan Cheng · 18 years ago
- 7277a7d One more extract_subreg coalescing bug. by Evan Cheng · 18 years ago
- cbfb9b2 - Only perform the unfolding optimization when the folding in question is modref. by Evan Cheng · 18 years ago
- 66f7163 Local spiller optimization: by Evan Cheng · 18 years ago
- 7f56625 Local spiller optimization: by Evan Cheng · 18 years ago
- 32dfbea EXTRACT_SUBREG coalescing support. The coalescer now treats EXTRACT_SUBREG like by Evan Cheng · 18 years ago
- 9efce63 Allow copyRegToReg to emit cross register classes copies. by Evan Cheng · 18 years ago
- 92dfe20 Remove isReg, isImm, and isMBB, and change all their users to use by Dan Gohman · 18 years ago
- 04fa32f by David Greene · 18 years ago
- 90a43c3 If the source of a move is in spill slot, the reload may be folded to essentially a load from stack slot. It's ok to mark the stack slot value as available for reuse. But it should not be clobbered since the destination of the move is live. by Evan Cheng · 18 years ago
- b6ca4b3 - If a def is dead, do not spill it. by Evan Cheng · 18 years ago
- c91f0b8 If a MI's def is remat as well as spilled, and the store is later deemed dead, mark the def operand as isDead. by Evan Cheng · 18 years ago
- fff3e19 If a spilled value is being reused and the use is a kill, that means there are by Evan Cheng · 18 years ago
- dc6be19 If a rematerializable def is not deleted, i.e. it is also spilled, check if the by Evan Cheng · 18 years ago
- 549f27d3 Re-implement trivial rematerialization. This allows def MIs whose live intervals that are coalesced to be rematerialized. by Evan Cheng · 18 years ago
- 28bb462 Missed a couple of places where new instructions are added due to spill / restore. by Evan Cheng · 18 years ago
- b9591c6 No longer need to track last def / use. by Evan Cheng · 18 years ago
- 0c40d72 Fix for PR1545: Revamp code that update kill information due to register reuse. by Evan Cheng · 18 years ago
- 82a87a0 Replace M_REMATERIALIZIBLE and the newly-added isOtherReMaterializableLoad by Dan Gohman · 18 years ago
- c101e95 Add a target hook to allow loads from constant pools to be rematerialized, and an by Dan Gohman · 18 years ago
- faa5107 Rename findRegisterUseOperand to findRegisterUseOperandIdx to avoid confusion. by Evan Cheng · 18 years ago
- 6c087e5 Match MachineFunction::UsedPhysRegs changes. by Evan Cheng · 18 years ago
- 9193514 Re-materialize all loads from fixed stack slots. by Evan Cheng · 18 years ago
- 7cb33c8 Don't add the same MI to register reuse "last def/use" twice if it reads the by Evan Cheng · 18 years ago
- 018d6e1 Don't call getOperandConstraint() if operand index is greater than by Evan Cheng · 18 years ago
- ad7ccf3 Fix for PR1266. Don't mark a two address operand IsKill. by Evan Cheng · 18 years ago
- 2638e1a First cut trivial re-materialization support. by Evan Cheng · 18 years ago
- a7288df Only propagate IsKill if the last use is a kill. by Evan Cheng · 18 years ago
- 6b44809 - Keep track all def and uses of stack slot available in register. by Evan Cheng · 18 years ago
- c3dd314 Invalidate last use of a reused register if the use is a deleted noop copy. by Evan Cheng · 18 years ago
- c0ba1bc A restore is promoted to copy (or deleted entirely), remove the kill from the last use of the targetted register. by Evan Cheng · 18 years ago
- de4e942 A couple of more places where a register liveness has been extended and its last kill should be updated accordingly. by Evan Cheng · 18 years ago
- 50d25d7 Reuse extends the liveness of a register. Transfer the kill to the operand that reuse it. by Evan Cheng · 18 years ago
- 91e2390 A spill kills the register being stored. But it is later being reused by spiller, its live range has to be extended. by Evan Cheng · 18 years ago
- 957840b Use BitVector instead. No functionality change. by Evan Cheng · 18 years ago
- 667089d Dead code. by Evan Cheng · 18 years ago
- f50d09a Fixed a long standing spiller bug that's exposed by Thumb: by Evan Cheng · 19 years ago
- 08a4d5a Switch this to use SmallSet to avoid mallocs in the common case. by Chris Lattner · 19 years ago
- 3c82cab GetRegForReload() now keeps track which registers have been considered and rejected during its quest to find a suitable reload register. This avoids an infinite loop in case like this: by Evan Cheng · 19 years ago
- cd3245a Eliminate static ctors from Statistics by Chris Lattner · 19 years ago
- 5c7e326 Added an automatic cast to "std::ostream*" etc. from OStream. We then can by Bill Wendling · 19 years ago
- 86facc2 Minor clean up. by Evan Cheng · 19 years ago
- 7a0d51c Fix a long-standing spiller bug: by Evan Cheng · 19 years ago
- cc22a7a Move findTiedToSrcOperand to TargetInstrDescriptor. by Evan Cheng · 19 years ago
- 5c2a460 Proper fix for PR1037: to determine is a VR is a modref, check 1) whether it is by Evan Cheng · 19 years ago
- fe46361 Revision 1.83 causes PR1037. Reverted. by Reid Spencer · 19 years ago
- e815619 Changed llvm_ostream et all to OStream. llvm_cerr, llvm_cout, llvm_null, are by Bill Wendling · 19 years ago
- 51cdcd1 MI keeps a ptr of TargetInstrDescriptor, use it. by Evan Cheng · 19 years ago
- 6065f66 getOperandConstraint returns -1 if the operand does have the specific constraint. This bug was causing excessive spills. by Evan Cheng · 19 years ago
- ac0b6ae Detemplatize the Statistic class. The only type it is instantiated with by Chris Lattner · 19 years ago
- ba59a1e Match TargetInstrInfo changes. by Evan Cheng · 19 years ago
- b2b9c20 More removal of std::cerr and DEBUG, replacing with DOUT instead. by Bill Wendling · 19 years ago
- e077ef6 Fixed some spiller bugs exposed by the recent two-address code changes. Now by Evan Cheng · 19 years ago
- e6ae14e Rename by Evan Cheng · 19 years ago
- 360c2dd Two-address instructions no longer have to be A := A op C. Now any pair of dest / src operands can be tied together. by Evan Cheng · 19 years ago
- 6ec3626 restore my previous patch, now that the X86 backend bug has been fixed: by Chris Lattner · 19 years ago
- b870100 Backing out Chris' last commit. It's breaking llvm-gcc bootstrapping. by Evan Cheng · 19 years ago
- f183cab If we see a load from a stack slot into a physreg, consider it as providing by Chris Lattner · 19 years ago
- 2926869 Fix a long-standing wart in the code generator: two-address instruction lowering by Chris Lattner · 19 years ago
- a4f0b3a s|llvm/Support/Visibility.h|llvm/Support/Compiler.h| by Chris Lattner · 19 years ago
- f7da2c7 Take advantage of the recent improvements to the liveintervals set (tracking by Chris Lattner · 19 years ago
- d36d03b Added a check so that if we have two machine instructions in this form by Bill Wendling · 19 years ago
- cd4317e Eliminate data relocations by using NULL instead of global empty list. by Jim Laskey · 19 years ago
- ed41f1b Reduce number of exported symbols by Andrew Lenharth · 19 years ago
- f8c68f6 Shave another 27K off libllvmgcc.dylib with visibility hidden by Chris Lattner · 19 years ago
- e53f4a0 Move some methods out of MachineInstr into MachineOperand by Chris Lattner · 19 years ago
- 229924a Fix a latent bug that my spiller patch last week exposed: we were leaving by Chris Lattner · 19 years ago
- 8a18c13 When promoting a load to a reg-reg copy, where the load was a previous by Chris Lattner · 19 years ago
- 35f2705 Remove previous patch, which wasn't quite right. by Chris Lattner · 19 years ago
- 10dbd3e Remove temp. option -spiller-check-liveout, it didn't cause any failure nor performance regressions. by Evan Cheng · 19 years ago
- 200370f Local spiller kills a store if the folded restore is turned into a copy. by Evan Cheng · 19 years ago
- ba1fc3d Mapping of physregs can make it so that the designated and input physregs are by Chris Lattner · 19 years ago
- addc55a When we have a two-address instruction where the input cannot be clobbered by Chris Lattner · 19 years ago
- 28bad08 Fix a bug that Evan exposed with some changes he's making, and that was by Chris Lattner · 19 years ago
- 47cb717 Remove debugging printout :) by Chris Lattner · 19 years ago
- 540fec6 Refactor some code from being inline to being out in a new class with methods. by Chris Lattner · 19 years ago
- 003cecb Fix VC++ warning. by Jeff Cohen · 20 years ago
- 8666249 Handle another case exposed on X86. by Chris Lattner · 20 years ago
- 593c958 Fix a nasty problem on two-address machines in the following situation: by Chris Lattner · 20 years ago
- 66cf80f significantly simplify the VirtRegMap code by pulling the SpillSlotsAvailable by Chris Lattner · 20 years ago
- 2ba0b02 Fix VC++ compilation error caused by using a std::map iterator variable to receive by Jeff Cohen · 20 years ago
- 109afed Remove move copies and dead stuff by not clobbering the result reg of a noop copy. by Chris Lattner · 20 years ago
- 84e752a Simplify some code by Chris Lattner · 20 years ago
- 1118d25 Add code that checks for noop copies, which triggers when either: by Chris Lattner · 20 years ago
- 07cf141 Physregs may hold multiple stack slot values at the same time. Keep track by Chris Lattner · 20 years ago
- cd81639 Fix a deficiency in the spiller that Evan noticed. In particular, consider by Chris Lattner · 20 years ago
- 4083960 Move isLoadFrom/StoreToStackSlot from MRegisterInfo to TargetInstrInfo,a far more logical place. Other methods should also be moved if anyoneis interested. :) by Chris Lattner · 20 years ago
- 2c2c6c6 Add explicit #includes of <iostream> by Chris Lattner · 20 years ago
- 172c362 Add an assertion, update DefInst even though no one uses it (dangling pointers by Chris Lattner · 20 years ago
- 8a61a75 Fix the LLC regressions on X86 last night. In particular, when undoing by Chris Lattner · 20 years ago