summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Tidy up PMStack. Add a bunch of consts, use std::vector instead ofDan Gohman2010-08-072-14/+13
| | | | | | | std::deque, since this is a stack and only supports push/pop on one end, and remove an unimplemented declaration. llvm-svn: 110495
* Tidy some #includes and forward-declarations, and move the C binding codeDan Gohman2010-08-077-44/+39
| | | | | | out of PassManager.cpp and into Core.cpp with the rest of the C binding code. llvm-svn: 110494
* Add a convenience constructor.Owen Anderson2010-08-072-0/+6
| | | | llvm-svn: 110493
* Make AnalysisImpls private.Dan Gohman2010-08-071-1/+1
| | | | llvm-svn: 110492
* Use sdmem and sse_load_f64 (etc.) for the vectorDale Johannesen2010-08-072-8/+40
| | | | | | | | form of CMPSD (etc.) Matching a 128-bit memory operand is wrong, the instruction uses only 64 bits (same as ADDSD etc.) 8193553. llvm-svn: 110491
* Correct -ftrapv to trap on errors, instead of calling theChris Lattner2010-08-072-67/+13
| | | | | | | | | | | | | | | | __overflow_handler entrypoint that David Chisnall made up. Calling __overflow_handler is not part of the contract of -ftrapv provided by GCC, and should never have been checked in in the first place. According to: http://permalink.gmane.org/gmane.comp.compilers.clang.devel/8699 David is using this for some of arbitrary precision integer stuff or something, which is not an appropriate thing to implement on this. llvm-svn: 110490
* Don't attempt the PRE inline asm calls, since we don't value number them ↵Owen Anderson2010-08-071-0/+5
| | | | | | yet. Fixes PR7835. llvm-svn: 110489
* Add a predicate to determine if a call is an inline asm statement.Owen Anderson2010-08-071-0/+5
| | | | llvm-svn: 110488
* Putting out messages about the number of test cases to be run before running theJohnny Chen2010-08-071-0/+25
| | | | | | whole test suite. llvm-svn: 110487
* PR7837: For qualified id's, make sure the decl context is complete if notEli Friedman2010-08-062-20/+29
| | | | | | | dependent in ActOnIdExpression. (This issue only shows up with member operators because an operator is never a type.) llvm-svn: 110486
* Some cleanup. Use a class (OptionInfo) instead of a pair of a pair and removeRafael Espindola2010-08-063-13/+18
| | | | | | some default values that are not used. llvm-svn: 110485
* Removed IdempotentOperationChecker from default analysis and returned back ↵Tom Care2010-08-0610-5/+17
| | | | | | | | | to a flag (-analyzer-check-idempotent-operations) - Added IdempotentOperationChecker to experimental analyses for testing purposes - Updated test cases to explictly call the checker llvm-svn: 110482
* Lazily defer duplicating the live interval we are splitting until we know it isJakob Stoklund Olesen2010-08-063-53/+70
| | | | | | | | | | necessary. Sometimes, live range splitting doesn't shrink the current interval, but simply changes some instructions to use a new interval. That makes the original more suitable for spilling. In this case, we don't need to duplicate the original. llvm-svn: 110481
* Patterns to match AVX 256-bit vzero intrinsicsBruno Cardoso Lopes2010-08-062-6/+6
| | | | llvm-svn: 110480
* Eliminate PromoteMemoryToRegisterID; just use addPreserved("mem2reg")Dan Gohman2010-08-065-6/+3
| | | | | | instead, as an example of what this looks like. llvm-svn: 110478
* llc: Clarify -mc-relax-all description.Michael J. Spencer2010-08-061-1/+3
| | | | llvm-svn: 110477
* tidy upJim Grosbach2010-08-061-13/+13
| | | | llvm-svn: 110476
* MC: Add default value for AddrSpace argument to EmitValue.Daniel Dunbar2010-08-061-3/+4
| | | | llvm-svn: 110475
* Fix leaks of ExplicitSpecializationInfo objects by allocating them with 'new ↵Ted Kremenek2010-08-061-3/+6
| | | | | | (ASTContext)' instead of 'new'. llvm-svn: 110474
* Fix 80 col. violations.Ted Kremenek2010-08-061-6/+14
| | | | llvm-svn: 110473
* Nest variable declaration into into 'if' condition, thus restricting the ↵Ted Kremenek2010-08-061-7/+3
| | | | | | scope of the variable and condensing the code. llvm-svn: 110472
* Use 'GenerateNode()' instead of 'GenerateSink()' when reporting a leak. A ↵Ted Kremenek2010-08-061-3/+3
| | | | | | leak is not a hard enough bug to stop analyzing a path. llvm-svn: 110471
* Added more comments about unittest2.Johnny Chen2010-08-061-0/+6
| | | | llvm-svn: 110470
* Patterns to match AVX 256-bit permutation intrinsicsBruno Cardoso Lopes2010-08-062-13/+48
| | | | llvm-svn: 110468
* Test case for r110459. Radar 8264751. Test case by Fariborz Jahanian!Stuart Hastings2010-08-061-0/+16
| | | | llvm-svn: 110467
* Cleanup comment wordingJim Grosbach2010-08-061-3/+3
| | | | llvm-svn: 110466
* Remove empty processFunctionBeforeFrameFinalized(). The defaultJim Grosbach2010-08-062-6/+0
| | | | | | | implementation of the function is equivalent, so no need to provide the target-specific version until/unless it needs to do something. llvm-svn: 110465
* Keep the MachiuneFunctionPass pointer around. It is useful for verification.Jakob Stoklund Olesen2010-08-061-1/+3
| | | | llvm-svn: 110464
* Add LiveInterval::RenumberValues - Garbage collection for VNInfos.Jakob Stoklund Olesen2010-08-063-1/+22
| | | | | | | After heavy editing of a live interval, it is much easier to simply renumber the live values instead of trying to keep track of the unused ones. llvm-svn: 110463
* Reapply r110396, with fixes to appease the Linux buildbot gods.Owen Anderson2010-08-06213-490/+496
| | | | llvm-svn: 110460
* Implement a proper getModRefInfo for va_arg.Dan Gohman2010-08-063-4/+29
| | | | llvm-svn: 110458
* spellingJim Grosbach2010-08-061-1/+1
| | | | llvm-svn: 110457
* Be more conservative in the face of volatile.Dan Gohman2010-08-061-8/+8
| | | | llvm-svn: 110456
* Fix a comment.Dan Gohman2010-08-061-2/+2
| | | | llvm-svn: 110455
* Add more verification of LiveIntervals.Jakob Stoklund Olesen2010-08-061-4/+58
| | | | llvm-svn: 110454
* Fix swapped COPY operands.Jakob Stoklund Olesen2010-08-061-2/+2
| | | | llvm-svn: 110453
* Don't try to verify LiveIntervals for physical registers.Jakob Stoklund Olesen2010-08-061-8/+4
| | | | | | | | | When a physical register is in use, some alias of that register has a live interval with a relevant live range. That is the sad state of intervals after physreg coalescing of subregs, and it is good enough for correct register allocation. llvm-svn: 110452
* Initialize variable to work around warning; unfortunately, there isn't anyEli Friedman2010-08-061-3/+3
| | | | | | way to tell gcc "really, values outside the enum aren't valid". llvm-svn: 110450
* Finishing up block variable layout API by supporting Fariborz Jahanian2010-08-062-8/+48
| | | | | | | union type variables and their nesting inside other aggregate types. llvm-svn: 110448
* Fix eabi calling convention when a 64 bit value shadows r3.Rafael Espindola2010-08-065-1/+30
| | | | | | | | | | | Without this what was happening was: * R3 is not marked as "used" * ARM backend thinks it has to save it to the stack because of vaarg * Offset computation correctly ignores it * Offsets are wrong llvm-svn: 110446
* The pre-increment/pre-decrement grammar in C++ differs from that in C,Douglas Gregor2010-08-062-4/+13
| | | | | | | but we were parsing the C grammar. Handle the C++ grammar appropriately. Fixes PR7794. llvm-svn: 110445
* Remove some incorrect assertions when deduction template arguments inDouglas Gregor2010-08-062-2/+18
| | | | | | | | a template-argument-list. When template template parameters are involved, we won't already have checked the template-argument-list (it may not be known yet!). Fixes PR7807. llvm-svn: 110444
* Make sure that we diagnose attribute((overloadable)) functions withoutDouglas Gregor2010-08-062-20/+25
| | | | | | prototypes. Fixes PR7738. llvm-svn: 110443
* Define _INTEGRAL_MAX_BITS for the win32 and win64 targets, from Per Lindén!Douglas Gregor2010-08-062-1/+4
| | | | llvm-svn: 110442
* Template keyword should not be ignored building a QualifiedTemplateName.Abramo Bagnara2010-08-067-12/+26
| | | | llvm-svn: 110441
* Fix the #include search path when reading from stdin, from Jon Simons!Douglas Gregor2010-08-064-32/+66
| | | | | | Fixes PR4897. llvm-svn: 110440
* Diagnose the use of "inline" on block-scope function declarations inDouglas Gregor2010-08-063-4/+19
| | | | | | C++, from Andrea Nall! llvm-svn: 110439
* Don't pass -avoid-version to Darwin linkerDouglas Gregor2010-08-061-1/+1
| | | | llvm-svn: 110438
* Introduce implicit conversions between AltiVec vectors and GCCDouglas Gregor2010-08-066-19/+96
| | | | | | vectors, from Anton Yartsev! llvm-svn: 110437
* Introduce a new token kind 'cxx_defaultarg_end' to mark the end of C++ ↵Argyrios Kyrtzidis2010-08-064-2/+18
| | | | | | | | | | | | | | | | default arguments that were part of lexed method declarations. This avoid interference with tokens coming after the point where the default arg tokens were 'injected', e.g. for typedef struct Inst { void m(int x=0); } *InstPtr; when parsing '0' the next token would be '*' and things would be messed up. llvm-svn: 110436
OpenPOWER on IntegriCloud