summaryrefslogtreecommitdiffstats
path: root/llvm/lib
Commit message (Collapse)AuthorAgeFilesLines
* If pass reserves all analysis info then each info is not separatelyDevang Patel2006-12-071-3/+6
| | | | | | included in PreservedSet. So check getPreservesAll() first. llvm-svn: 32319
* Pass Managers themselves do not invalidate any analysis info.Devang Patel2006-12-071-1/+22
| | | | llvm-svn: 32318
* Add a handle to the top level pass manager in PMDataManager.Devang Patel2006-12-071-1/+14
| | | | llvm-svn: 32317
* Cosmetic markers to divide code in separate chunks.Devang Patel2006-12-071-4/+13
| | | | llvm-svn: 32316
* s/noteDownAvailableAnalysis/recordAvailableAnalysisDevang Patel2006-12-071-13/+12
| | | | | | While recording available analysis, include interfaces implemented. llvm-svn: 32315
* Add PMTopLevelManager. It is not used yet.Devang Patel2006-12-071-0/+76
| | | | llvm-svn: 32314
* RequiredAnalysis support is buggy and not used at the moment so removeDevang Patel2006-12-071-41/+10
| | | | | | the code. Add TODO note. llvm-svn: 32313
* s/clearAnalysis/initializeAnalysisInfo/gDevang Patel2006-12-071-7/+6
| | | | llvm-svn: 32312
* s/CommonPassManagerImpl/PMDataManager/gDevang Patel2006-12-071-16/+15
| | | | llvm-svn: 32311
* make sure that we don't use a common symbol if a section was specifiedRafael Espindola2006-12-071-0/+1
| | | | llvm-svn: 32310
* Add overview of pass manager.Devang Patel2006-12-071-0/+64
| | | | llvm-svn: 32309
* Be sure to grab weak functions too, and make implicit defs commentsAndrew Lenharth2006-12-072-3/+10
| | | | llvm-svn: 32308
* Revision 1.83 causes PR1037.Reid Spencer2006-12-071-2/+1
| | | | | | Reverted. llvm-svn: 32305
* Provide a getOpcode() method on CmpInst to ensure the opcode is returnedReid Spencer2006-12-072-4/+2
| | | | | | as the right type. Use this to shorten some code. llvm-svn: 32300
* Initialize {Min|Max}CSFrameIndex properly.Evan Cheng2006-12-071-2/+4
| | | | llvm-svn: 32299
* Changed llvm_ostream et all to OStream. llvm_cerr, llvm_cout, llvm_null, areBill Wendling2006-12-0770-1409/+1563
| | | | | | now cerr, cout, and NullStream resp. llvm-svn: 32298
* Fix i64 uint_to_fp on ppc64Chris Lattner2006-12-071-1/+2
| | | | llvm-svn: 32297
* MI keeps a ptr of TargetInstrDescriptor, use it.Evan Cheng2006-12-073-4/+7
| | | | llvm-svn: 32296
* getOperandConstraint returns -1 if the operand does have the specific ↵Evan Cheng2006-12-071-1/+1
| | | | | | constraint. This bug was causing excessive spills. llvm-svn: 32295
* implement sextinreg i8->i64 and i16->i64Chris Lattner2006-12-061-0/+8
| | | | llvm-svn: 32293
* fix another sradi encoding bug. This fixes Olden/health with the ppc64 jit.Chris Lattner2006-12-061-1/+1
| | | | llvm-svn: 32291
* Always pass "true" to isMaxValue(bool) because we know the type is LongTy.Reid Spencer2006-12-061-1/+1
| | | | llvm-svn: 32290
* Adjust to new ConstantIntegral interface for Max/Min tests.Reid Spencer2006-12-062-11/+57
| | | | llvm-svn: 32289
* Update ConstantIntegral Max/Min tests for new interface.Reid Spencer2006-12-061-4/+4
| | | | llvm-svn: 32288
* For PR950:Reid Spencer2006-12-061-47/+0
| | | | | | | | | Remove the getMaxValue and getMinValue functions from ConstantIntegral. They don't make sense for a signless type. Also, for isMaxValue and isMinValue, have the caller provided the signedness rather than obtaining it from the constant's type. llvm-svn: 32287
* fix the jit encoding of sradi, simplify the MDForm1 description.Chris Lattner2006-12-061-5/+5
| | | | llvm-svn: 32285
* add relocation support for ppc64 branches.Chris Lattner2006-12-061-2/+2
| | | | llvm-svn: 32284
* merge the Statistic and StatisticBase classes, eliminating virtual methodsChris Lattner2006-12-061-10/+5
| | | | | | and eliminating #includes from the Statistic.h file. llvm-svn: 32282
* add #includeChris Lattner2006-12-061-0/+1
| | | | llvm-svn: 32281
* add missing #includeChris Lattner2006-12-062-0/+2
| | | | llvm-svn: 32280
* Detemplatize the Statistic class. The only type it is instantiated withChris Lattner2006-12-0696-235/+233
| | | | | | is 'unsigned'. llvm-svn: 32279
* Some addresssed should be 64-bit and some shouldn't.Jim Laskey2006-12-061-19/+25
| | | | llvm-svn: 32278
* Make it easier for gdb to find the return address.Jim Laskey2006-12-063-58/+98
| | | | llvm-svn: 32277
* print weak referencesRafael Espindola2006-12-062-1/+14
| | | | llvm-svn: 32276
* Move copyKillDeadInfo out-of-line. Add findRegisterUseOperand().Evan Cheng2006-12-061-0/+31
| | | | llvm-svn: 32273
* Remove the dead CachedWriter class.Chris Lattner2006-12-061-57/+11
| | | | llvm-svn: 32271
* Simplify codeChris Lattner2006-12-061-10/+4
| | | | llvm-svn: 32270
* printName is almost always true. In the cases that mattered where it was false,Chris Lattner2006-12-061-35/+25
| | | | | | | | | | | it was effectively set to true by this: - if ((PrintName || isa<GlobalValue>(V)) && V->hasName()) + if (V->hasName()) Delete printname entirely. llvm-svn: 32265
* Remove the 'printname' argument to WriteAsOperand. It is always true, andChris Lattner2006-12-067-13/+13
| | | | | | passing false would make the asmprinter fail anyway. llvm-svn: 32264
* The hasSlot methods are gone.Chris Lattner2006-12-061-25/+20
| | | | | | | Remove the 'PrintName' argument to WriteAsOperand, as it is always true. Only call getOrCreateSlot on things that are valid. llvm-svn: 32263
* These asm printers shouldn't use assembly/writer.hChris Lattner2006-12-063-20/+4
| | | | llvm-svn: 32262
* remove unused api, simplify some codeChris Lattner2006-12-061-20/+2
| | | | llvm-svn: 32260
* remove more code that was only used by the bc writerChris Lattner2006-12-061-90/+2
| | | | llvm-svn: 32259
* remove dead code left over from when this functionality was shared with theChris Lattner2006-12-061-68/+5
| | | | | | bcwriter. llvm-svn: 32258
* rename createSlot -> getOrCreateSlot.Chris Lattner2006-12-061-21/+19
| | | | llvm-svn: 32256
* clean up some sloppy and inconsistent spacingChris Lattner2006-12-061-46/+46
| | | | llvm-svn: 32255
* wrap long linesChris Lattner2006-12-061-4/+6
| | | | llvm-svn: 32254
* Fix a CmpInst writing bug by removing merge cruft that I *know* I've removedReid Spencer2006-12-061-9/+4
| | | | | | | before. Also, make sure we write the predicate value for Cmp instructions using instruction format 0. llvm-svn: 32253
* counter should be unsigned.Chris Lattner2006-12-061-1/+1
| | | | llvm-svn: 32252
* eliminate fp statisticChris Lattner2006-12-061-3/+3
| | | | llvm-svn: 32251
OpenPOWER on IntegriCloud