[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.43 - 1.44 --- Log message: Add a register allocation preference field; add a method to compute size of a live interval. --- Diffs of the changes: (+9 -0) LiveInterval.cpp |9 + 1 files changed, 9 insertions(+)

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

On Dec 15, 2006, at 6:11 PM, Jeff Cohen wrote: I got it to work... I'll be committing shortly. I'm not sure what the problem was, but it seems to have something to do with resolving the overload of operator and almost certainly is a VC++ bug. It won't accept a forward declared type

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.40 - 1.41 MachineBasicBlock.cpp updated: 1.36 - 1.37 MachineInstr.cpp updated: 1.138 - 1.139 --- Log message: An even better unbreakage... --- Diffs of the changes: (+24 -0) LiveInterval.cpp |7 +++

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

Hold on. If you make these non-inlined, then the only chance for the compiler to remove potentially dead code (when o.stream() == 0) is through link time optimizations. Because this is still compiled by GCC right now, this won't be done and debug outputs will still be in the code. That is,

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

Chris asked me to not include ostream from header files, and to move stuff out of the header files if possible to make it so. Bill Wendling wrote: Hold on. If you make these non-inlined, then the only chance for the compiler to remove potentially dead code (when o.stream() == 0) is through

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

That's reasonable. So then a better solution would be to keep everything in the .h file, but change operator(OStream) to call the print() function directly. So something like this: #include iosfwd // ... OStream operator(OStream OS, const MachineBasicBlock MBB) { if (*OS.stream())

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

On Dec 15, 2006, at 3:31 PM, Bill Wendling wrote: That's reasonable. So then a better solution would be to keep everything in the .h file, but change operator(OStream) to call the print() function directly. So something like this: #include iosfwd // ... OStream operator(OStream OS,

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

I'll look into making this change. Part of the problem is that VC++ doesn't like *OS.stream() in a header file when ostream isn't included, even if the declaration of std::basic_ostream shouldn't be needed (but doesn't have a problem with it in a .cpp file, probably because it isn't inlined

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

On Dec 15, 2006, at 3:58 PM, Jeff Cohen wrote: I'll look into making this change. Part of the problem is that VC++ doesn't like *OS.stream() in a header file when ostream isn't included, even if the declaration of std::basic_ostream shouldn't be needed (but doesn't have a problem with

Re: [llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp MachineBasicBlock.cpp MachineInstr.cpp

I got it to work... I'll be committing shortly. I'm not sure what the problem was, but it seems to have something to do with resolving the overload of operator and almost certainly is a VC++ bug. It won't accept a forward declared type when it ought to. Chris Lattner wrote: On Dec 15,

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.37 - 1.38 --- Log message: Put the #include for a module first. --- Diffs of the changes: (+1 -1) LiveInterval.cpp |2 +- 1 files changed, 1 insertion(+), 1 deletion(-) Index: llvm/lib/CodeGen/LiveInterval.cpp diff

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp LiveIntervalAnalysis.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.34 - 1.35 LiveIntervalAnalysis.cpp updated: 1.179 - 1.180 --- Log message: When joining two intervals where the RHS is really simple, use a light-weight method for joining the live ranges instead of the fully-general one. ---

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp LiveIntervalAnalysis.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.33 - 1.34 LiveIntervalAnalysis.cpp updated: 1.173 - 1.174 --- Log message: avoid calling the virtual isMoveInstr method endlessly by caching its results. --- Diffs of the changes: (+73 -75) LiveInterval.cpp | 24

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp LiveIntervalAnalysis.cpp VirtRegMap.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.28 - 1.29 LiveIntervalAnalysis.cpp updated: 1.166 - 1.167 VirtRegMap.cpp updated: 1.70 - 1.71 --- Log message: Take advantage of the recent improvements to the liveintervals set (tracking instructions which define each value#)

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.29 - 1.30 --- Log message: When replacing value numbers, make sure to compactify the value # space. --- Diffs of the changes: (+12 -0) LiveInterval.cpp | 12 1 files changed, 12 insertions(+) Index:

[llvm-commits] CVS: llvm/lib/CodeGen/LiveInterval.cpp LiveIntervalAnalysis.cpp

Changes in directory llvm/lib/CodeGen: LiveInterval.cpp updated: 1.27 - 1.28 LiveIntervalAnalysis.cpp updated: 1.165 - 1.166 --- Log message: Improve the LiveInterval class to keep track of which machine instruction defines each value# tracked by the interval. This will be used to improve