[llvm-commits] [llvm] r40965 - in /llvm/trunk: include/llvm/Target/TargetLowering.h lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

2007-08-09 Thread Dale Johannesen
Author: johannes Date: Thu Aug 9 12:27:48 2007 New Revision: 40965 URL: http://llvm.org/viewvc/llvm-project?rev=40965&view=rev Log: Update per review comments. Modified: llvm/trunk/include/llvm/Target/TargetLowering.h llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Modified: llvm/tr

[llvm-commits] [llvm] r40970 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.td

2007-08-09 Thread Evan Cheng
Author: evancheng Date: Thu Aug 9 13:05:17 2007 New Revision: 40970 URL: http://llvm.org/viewvc/llvm-project?rev=40970&view=rev Log: GR16_ sub-register class should be GR8_, not GR8. That is, it should only be 8-bit registers in 32-bit mode. Ditto for GR32_. Modified: llvm/trunk/lib/Target/

[llvm-commits] [llvm] r40971 - in /llvm/trunk: lib/Transforms/Scalar/InstructionCombining.cpp test/Transforms/InstCombine/align-inc.ll

2007-08-09 Thread Chris Lattner
Author: lattner Date: Thu Aug 9 14:05:49 2007 New Revision: 40971 URL: http://llvm.org/viewvc/llvm-project?rev=40971&view=rev Log: when we see a unaligned load from an insufficiently aligned global or alloca, increase the alignment of the load, turning it into an aligned load. This allows us to

Re: [llvm-commits] [llvm] r40970 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.td

2007-08-09 Thread Christopher Lamb
Evan, I don't think that this change is working quite correctly, and may not be needed, given the way that the subreg code uses those values. This is currently causing regressions in dejagnu. The register classes in the SubRegClassList are used when new subreg vregs are created. The key i

[llvm-commits] [llvm] r40972 - in /llvm/trunk/lib/Target/X86: README-X86-64.txt X86ISelDAGToDAG.cpp

2007-08-09 Thread Evan Cheng
Author: evancheng Date: Thu Aug 9 16:59:35 2007 New Revision: 40972 URL: http://llvm.org/viewvc/llvm-project?rev=40972&view=rev Log: divb / mulb outputs to ah. Under x86-64 it's not legal to read ah if the instruction requires a rex prefix (i.e. outputs to r8b, etc.). So issue shift right by 8

Re: [llvm-commits] [llvm] r40970 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.td

2007-08-09 Thread Evan Cheng
On Aug 9, 2007, at 2:45 PM, Christopher Lamb wrote: Evan, I don't think that this change is working quite correctly, and may not be needed, given the way that the subreg code uses those values. This is currently causing regressions in dejagnu. The register classes in the SubRegClassList

[llvm-commits] [llvm] r40973 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.td

2007-08-09 Thread Evan Cheng
Author: evancheng Date: Thu Aug 9 17:25:35 2007 New Revision: 40973 URL: http://llvm.org/viewvc/llvm-project?rev=40973&view=rev Log: Temporarily backing out this change until we know why some dejagnu tests are failing. Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.td Modified: llvm/t

Re: [llvm-commits] [llvm] r40970 - /llvm/trunk/lib/Target/X86/X86RegisterInfo.td

2007-08-09 Thread Christopher Lamb
On Aug 9, 2007, at 3:20 PM, Evan Cheng wrote: On Aug 9, 2007, at 2:45 PM, Christopher Lamb wrote: Evan, I don't think that this change is working quite correctly, and may not be needed, given the way that the subreg code uses those values. This is currently causing regressions in dejagn

[llvm-commits] [llvm] r40974 - in /llvm/trunk/lib: Bitcode/Reader/BitcodeReader.cpp Bitcode/Writer/BitcodeWriter.cpp VMCore/Constants.cpp

2007-08-09 Thread Dale Johannesen
Author: johannes Date: Thu Aug 9 17:51:36 2007 New Revision: 40974 URL: http://llvm.org/viewvc/llvm-project?rev=40974&view=rev Log: Patch 10 for long double. Doing constants right needs expanding ConstantFP to handle values bigger than double. If we assume host==target and host long double work

Re: [llvm-commits] [llvm] r40974 - in /llvm/trunk/lib: Bitcode/Reader/BitcodeReader.cpp Bitcode/Writer/BitcodeWriter.cpp VMCore/Constants.cpp

2007-08-09 Thread Chris Lattner
> Patch 10 for long double. Doing constants right needs expanding > ConstantFP > to handle values bigger than double. If we assume host==target and > host > long double works correctly, this is not too bad, but we don't want to > have that limitation longterm. I could implement accepting dou

[llvm-commits] [llvm] r40975 - /llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

2007-08-09 Thread Evan Cheng
Author: evancheng Date: Thu Aug 9 18:14:39 2007 New Revision: 40975 URL: http://llvm.org/viewvc/llvm-project?rev=40975&view=rev Log: Bug fix. ~1U marks the val# dead. Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

Re: [llvm-commits] [llvm] r40975 - /llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

2007-08-09 Thread Dale Johannesen
> -ValueNumberInfo[VN].def = RHSValNoInfo.def; > +ValueNumberInfo[VN]->def = ~1U; // Now this val# is > dead. Is .def equivalent to ->def here, or is this another change? ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu

Re: [llvm-commits] [llvm] r40975 - /llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

2007-08-09 Thread Evan Cheng
Something else changed. But not checked in. :-) Will fix. Thx. Evan On Aug 9, 2007, at 4:19 PM, Dale Johannesen wrote: > >> -ValueNumberInfo[VN].def = RHSValNoInfo.def; >> +ValueNumberInfo[VN]->def = ~1U; // Now this val# is >> dead. > > Is .def equivalent to ->def here,

[llvm-commits] [llvm] r40976 - /llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

2007-08-09 Thread Chris Lattner
Author: lattner Date: Thu Aug 9 18:55:17 2007 New Revision: 40976 URL: http://llvm.org/viewvc/llvm-project?rev=40976&view=rev Log: unbreak the build Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp URL: http://llvm.

Re: [llvm-commits] [llvm] r40974 - in /llvm/trunk/lib: Bitcode/Reader/BitcodeReader.cpp Bitcode/Writer/BitcodeWriter.cpp VMCore/Constants.cpp

2007-08-09 Thread Dale Johannesen
On Aug 9, 2007, at 4:58 PM, Chris Lattner wrote: > Using host long double is an acceptable hack for now. Alternatively: OK, I'll do that. ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

[llvm-commits] [llvm] r40979 - /llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp

2007-08-09 Thread Devang Patel
Author: dpatel Date: Thu Aug 9 19:59:03 2007 New Revision: 40979 URL: http://llvm.org/viewvc/llvm-project?rev=40979&view=rev Log: Remove unncessary duplication. Modified: llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp Modified: llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp URL:

[llvm-commits] [llvm] r40977 - /llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp

2007-08-09 Thread Devang Patel
Author: dpatel Date: Thu Aug 9 19:33:50 2007 New Revision: 40977 URL: http://llvm.org/viewvc/llvm-project?rev=40977&view=rev Log: ExitCondition and Induction variable are loop constraints not split condition constraints. Modified: llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp Modifie

[llvm-commits] [llvm] r40978 - /llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp

2007-08-09 Thread Devang Patel
Author: dpatel Date: Thu Aug 9 19:53:35 2007 New Revision: 40978 URL: http://llvm.org/viewvc/llvm-project?rev=40978&view=rev Log: Calculate exit and start value of true loop and false loop respectively. Modified: llvm/trunk/lib/Transforms/Scalar/LoopIndexSplit.cpp Modified: llvm/trunk/lib/T

[llvm-commits] [llvm] r40982 - in /llvm/trunk: include/llvm/IntrinsicsX86.td lib/Target/X86/X86InstrSSE.td

2007-08-09 Thread Bill Wendling
Author: void Date: Fri Aug 10 01:22:27 2007 New Revision: 40982 URL: http://llvm.org/viewvc/llvm-project?rev=40982&view=rev Log: Adding SSSE3 intrinsics. Modified: llvm/trunk/include/llvm/IntrinsicsX86.td llvm/trunk/lib/Target/X86/X86InstrSSE.td Modified: llvm/trunk/include/llvm/Intrinsi

[llvm-commits] [llvm] r40981 - /llvm/trunk/lib/VMCore/PassManager.cpp

2007-08-09 Thread Chris Lattner
Author: lattner Date: Fri Aug 10 01:22:25 2007 New Revision: 40981 URL: http://llvm.org/viewvc/llvm-project?rev=40981&view=rev Log: minor simplifications. Modified: llvm/trunk/lib/VMCore/PassManager.cpp Modified: llvm/trunk/lib/VMCore/PassManager.cpp URL: http://llvm.org/viewvc/llvm-project

[llvm-commits] [llvm] r40983 - /llvm/trunk/include/llvm/ADT/SmallVector.h

2007-08-09 Thread Chris Lattner
Author: lattner Date: Fri Aug 10 01:54:38 2007 New Revision: 40983 URL: http://llvm.org/viewvc/llvm-project?rev=40983&view=rev Log: small speedup in the case where a smallvector is default ctor'd from an empty vector. This speeds up llc slightly. Modified: llvm/trunk/include/llvm/ADT/SmallVe

[llvm-commits] [llvm] r40980 - in /llvm/trunk: include/llvm/PassManagers.h lib/VMCore/PassManager.cpp

2007-08-09 Thread Chris Lattner
Author: lattner Date: Fri Aug 10 01:17:04 2007 New Revision: 40980 URL: http://llvm.org/viewvc/llvm-project?rev=40980&view=rev Log: avoid copying strings. Modified: llvm/trunk/include/llvm/PassManagers.h llvm/trunk/lib/VMCore/PassManager.cpp Modified: llvm/trunk/include/llvm/PassManagers