summaryrefslogtreecommitdiffstats
path: root/llvm
Commit message (Collapse)AuthorAgeFilesLines
* Add more reset methods to make all objects that the backend may use for ↵Pedro Artigas2012-12-147-1/+40
| | | | | | outputting code have a reset, some are not used but were declared for completeness llvm-svn: 170227
* rdar://12753946Shuxin Yang2012-12-142-6/+66
| | | | | | Implement rule : "x * (select cond 1.0, 0.0) -> select cond x, 0.0" llvm-svn: 170226
* [CMake] Move libxml2 stuff from clang to llvm/cmake.NAKAMURA Takumi2012-12-142-0/+8
| | | | llvm-svn: 170225
* This patch improves the 64-bit PowerPC InitialExec TLS support by providingBill Schmidt2012-12-1413-48/+67
| | | | | | | | | | | | | | | | | | | | | | | for a wider range of GOT entries that can hold thread-relative offsets. This matches the behavior of GCC, which was not documented in the PPC64 TLS ABI. The ABI will be updated with the new code sequence. Former sequence: ld 9,x@got@tprel(2) add 9,9,x@tls New sequence: addis 9,2,x@got@tprel@ha ld 9,x@got@tprel@l(9) add 9,9,x@tls Note that a linker optimization exists to transform the new sequence into the shorter sequence when appropriate, by replacing the addis with a nop and modifying the base register and relocation type of the ld. llvm-svn: 170209
* Fix lint warnings in MemorySanitizer.cpp.Evgeniy Stepanov2012-12-141-9/+12
| | | | llvm-svn: 170203
* The CellSPU backend is more than experimental, it's dead now...Chandler Carruth2012-12-141-1/+1
| | | | llvm-svn: 170201
* Remove a section that was in 3.1's notes.Chandler Carruth2012-12-141-15/+0
| | | | | | Just give a shout if this was actually still important.... llvm-svn: 170200
* [msan] Origin stores and loads do not need explicit alignment.Evgeniy Stepanov2012-12-142-9/+38
| | | | | | Origin address is always 4 byte aligned, and the access type is always i32. llvm-svn: 170199
* Remove several entries from the 3.2 release notes that were in the 3.1Chandler Carruth2012-12-141-16/+0
| | | | | | release notes already. =] llvm-svn: 170198
* Fix the order of these sections of the release notes.Chandler Carruth2012-12-141-8/+8
| | | | llvm-svn: 170197
* Add a rough draft of some content about the new SROA. I'll try to proofChandler Carruth2012-12-141-1/+15
| | | | | | | read this and clean it up tomorrow, but hopefully it's a good placeholder. llvm-svn: 170196
* Delete a long-stale "if its ready" comment. All indications is that thisChandler Carruth2012-12-141-1/+0
| | | | | | | will look a bit different when we have time to get it ready to turn on, and we won't likely need this reminder. llvm-svn: 170195
* Provide some actual highlight bullets for Clang.Chandler Carruth2012-12-141-1/+3
| | | | | | | If anyone has better highlights (I'm obviously biased by the things that I'm excited about) jump in and add them! llvm-svn: 170194
* [msan] Refactor default shadow propagation and origin tracking.Evgeniy Stepanov2012-12-141-52/+117
| | | | | | | | This change moves the code for default shadow propagaition (handleShadowOr) and origin tracking (setOriginForNaryOp) into a new builder-like class. Also gets rid of handleShadowOrBinary. llvm-svn: 170192
* Change TargetLowering::getLoadExtAction to take an MVT, instead ofPatrik Hagglund2012-12-142-6/+6
| | | | | | EVT. llvm-svn: 170183
* revert r170166 - disable the loop vectorizer.Nadav Rotem2012-12-141-1/+1
| | | | llvm-svn: 170172
* Enable the loop vectorizer.Nadav Rotem2012-12-141-1/+1
| | | | llvm-svn: 170166
* Disable the loop vectorizer.Nadav Rotem2012-12-141-1/+1
| | | | llvm-svn: 170162
* Use the new MI bundling API in MachineInstrBundle itself.Jakob Stoklund Olesen2012-12-131-4/+5
| | | | | | | The new API is higher level than just manipulating the bundle flags directly, and the setIsInsideBundle() function will disappear soon. llvm-svn: 170159
* Remove two popcount patterns which we are already able to recognize.Shuxin Yang2012-12-131-15/+0
| | | | llvm-svn: 170158
* Enable the Loop Vectorizer by default for O2 and O3. Disable if-conversion ↵Nadav Rotem2012-12-132-2/+2
| | | | | | by default. I plan to revert this patch later today. llvm-svn: 170157
* Debug Info: add support to mark member variables as artificialDavid Blaikie2012-12-132-0/+4
| | | | | | This is the LLVM portion of r170154. llvm-svn: 170156
* fix comment.Chris Lattner2012-12-131-1/+1
| | | | llvm-svn: 170155
* This is another cleanup patch for 64-bit PowerPC TLS processing. I hadBill Schmidt2012-12-133-57/+11
| | | | | | | some hackery in place that hid my poor use of TblGen, which I've now sorted out and cleaned up. No change in observable behavior, so no new test cases. llvm-svn: 170149
* Change TargetLowering::setTypeAction to take an MVT, instead fo EVT.Patrik Hagglund2012-12-132-3/+3
| | | | llvm-svn: 170148
* docs: Improve discussion of syntax highlighting.Sean Silva2012-12-131-6/+19
| | | | llvm-svn: 170145
* Documentation: CompilerWriterInfo.rst: update link to Intel documentationDmitri Gribenko2012-12-131-1/+1
| | | | | | | | Replaces old Pentium 4 documentation link with generic current documentation link. Patch by Kevin Schoedel. llvm-svn: 170144
* Fix warnings with -DNDEBUGTom Stellard2012-12-133-2/+3
| | | | | | Patch by: NAKAMURA Takumi llvm-svn: 170142
* This is just a clean-up patch that simplifies the initial-exec TLS logic byBill Schmidt2012-12-134-19/+7
| | | | | | | avoiding use of machine operand flags. No change in observable behavior, so no new test cases. llvm-svn: 170141
* Change TargetLowering::getRepRegClassFor to take an MVT, instead ofPatrik Hagglund2012-12-135-15/+14
| | | | | | | | EVT. Accordingly, change RegDefIter to contain MVTs instead of EVTs. llvm-svn: 170140
* Add options to disable building of ARCMT, Rewriter and Static AnalyzerRoman Divacky2012-12-133-7/+187
| | | | | | in clang. The default remains to build those. llvm-svn: 170134
* Fix spellingJoel Jones2012-12-131-2/+2
| | | | llvm-svn: 170130
* JITEventListener.h: Use llvm-config.h instead of config.h.NAKAMURA Takumi2012-12-133-1/+13
| | | | llvm-svn: 170129
* Revert r170020, "Simplify negated bit test", for now.NAKAMURA Takumi2012-12-132-44/+0
| | | | | | | | | This assumes (1 << n) is always not zero. Consider n is greater than word size. Although I know it is undefined, this transforms undefined behavior hidden. This led clang unexpected behavior with some failures. I will investigate to fix undefined shl in clang. llvm-svn: 170128
* Add a dummy documentation file to unbreak 'make install'. We need to find howDmitri Gribenko2012-12-131-0/+0
| | | | | | to package reST-formatted documentation. llvm-svn: 170126
* Revert "Restore the PHI optimization I accidently removed" temporarily sinceEric Christopher2012-12-131-4/+0
| | | | | | | | it seems to be breaking self-host for a few people and is PR14592. This reverts commit r170024. llvm-svn: 170106
* Revert "Add a funciton to get the segment name of a section."Eric Christopher2012-12-134-48/+14
| | | | | | This reverts commit r170095 since it appears to be breaking the bots. llvm-svn: 170105
* Change TargetLowering::getRegClassFor to take an MVT, instead of EVT.Patrik Hagglund2012-12-1313-81/+89
| | | | | | | | | | | | Accordingly, add helper funtions getSimpleValueType (in parallel to getValueType) in SDValue, SDNode, and TargetLowering. This is the first, in a series of patches. This is the second attempt. In the first attempt (r169837), a few getSimpleVT() were hoisted too far, detected by bootstrap failures. llvm-svn: 170104
* Add a funciton to get the segment name of a section.Rafael Espindola2012-12-134-14/+48
| | | | | | | | | | | | | | | On MachO, sections also have segment names. When a tool looking at a .o file prints a segment name, this is what they mean. In reality, a .o has only one, anonymous, segment. This patch adds a MachO only function to fetch that segment name. I named it getSectionFinalSegmentName since the main use for the name seems to be informing the linker with segment this section should go to. The patch also changes MachOObjectFile::getSectionName to return just the section name instead of computing SegmentName,SectionName. llvm-svn: 170095
* Missed these calls from the previous rename somehow.Rafael Espindola2012-12-131-2/+2
| | | | llvm-svn: 170094
* Rename isPowerOfTwo to isKnownToBeAPowerOfTwo.Rafael Espindola2012-12-134-21/+21
| | | | | | | | In a previous thread it was pointed out that isPowerOfTwo is not a very precise name since it can return false for powers of two if it is unable to show that they are powers of two. llvm-svn: 170093
* [mips] Do not copy GOT address to register $gp if the function being called hasAkira Hatanaka2012-12-132-4/+36
| | | | | | internal linkage. llvm-svn: 170092
* Pattern matching code for intrinsics.Michael Ilseman2012-12-132-15/+115
| | | | | | Provides m_Argument that allows matching against a CallSite's specified argument. Provides m_Intrinsic pattern that can be templatized over the intrinsic id and bind/match arguments similarly to other pattern matchers. Implementations provided for 0 to 4 arguments, though it's very simple to extend for more. Also provides example template specialization for bswap (m_BSwap) and example of code cleanup for its use. llvm-svn: 170091
* Remove extraneous debugging code.Eric Christopher2012-12-131-2/+0
| | | | llvm-svn: 170090
* Use default label name for a section in emitting abbreviationEric Christopher2012-12-131-5/+8
| | | | | | section to help prep some code to be split about. llvm-svn: 170088
* Add a way of printing out an arbitrary label name for a sectionEric Christopher2012-12-135-0/+31
| | | | | | given the section. llvm-svn: 170087
* m_CombineOr and m_CombineAnd pattern combinatorsMichael Ilseman2012-12-131-10/+49
| | | | llvm-svn: 170086
* Define getHostCPUFeatures for ARM Linux platformHao Liu2012-12-131-0/+58
| | | | llvm-svn: 170085
* [mips] Delete all floating point instruction classes that are no longer used.Akira Hatanaka2012-12-133-279/+2
| | | | | | No functionality change. llvm-svn: 170084
* Make this Lit config file a bit slimmerEli Bendersky2012-12-131-8/+1
| | | | llvm-svn: 170083
OpenPOWER on IntegriCloud