summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Target
Commit message (Collapse)AuthorAgeFilesLines
* Change errs() to dbgs().David Greene2010-01-051-1/+2
| | | | llvm-svn: 92655
* Change errs() to dbgs().David Greene2010-01-051-8/+8
| | | | llvm-svn: 92654
* Change errs() to dbgs().David Greene2010-01-051-1/+2
| | | | llvm-svn: 92653
* Change errs() to dbgs().David Greene2010-01-051-1/+1
| | | | llvm-svn: 92651
* Change errs() to dbgs().David Greene2010-01-051-1/+1
| | | | llvm-svn: 92648
* Change errs() to dbgs().David Greene2010-01-051-30/+30
| | | | llvm-svn: 92647
* Change errs() to dbgs().David Greene2010-01-051-10/+10
| | | | llvm-svn: 92644
* Change SelectCode's argument from SDValue to SDNode *, to make it moreDan Gohman2010-01-0514-458/+441
| | | | | | | | | clear what information these functions are actually using. This is also a micro-optimization, as passing a SDNode * around is simpler than passing a { SDNode *, int } by value or reference. llvm-svn: 92564
* Remove dead debug info intrinsics.Devang Patel2010-01-051-15/+0
| | | | | | | | | | Intrinsic::dbg_stoppoint Intrinsic::dbg_region_start Intrinsic::dbg_region_end Intrinsic::dbg_func_start AutoUpgrade simply ignores these intrinsics now. llvm-svn: 92557
* Remove the SDNPAssociative properties for the flags-producingDan Gohman2010-01-051-6/+6
| | | | | | | operators. Eli pointed out that it's not obvious what that would mean. llvm-svn: 92555
* Perform this folding as a target specific dag combine:Evan Cheng2010-01-042-36/+61
| | | | | | | | (or (x << c) | (y >> (64 - c))) ==> (shld64 x, y, c) The isel patterns may not catch all the cases if general dag combine has reduced width of source operands. llvm-svn: 92513
* Remove some README.txt entries which are now implemented.Dan Gohman2010-01-041-80/+3
| | | | llvm-svn: 92511
* A use by operand 1 or 2 of a SELECT is not a FLAGS use. ThisDan Gohman2010-01-041-1/+1
| | | | | | lets the test-elimination work in more conditional-move cases. llvm-svn: 92508
* Flags-producing add, and, or, etc. have the same profibilityDan Gohman2010-01-041-0/+5
| | | | | | rules as normal add, and, or, etc. llvm-svn: 92507
* Add SDNPCommutative and SDNPAssociative to several X86 target nodes.Dan Gohman2010-01-041-6/+12
| | | | | | This lets isel fold loads into them in more cases. llvm-svn: 92506
* Fix invalid chain folding for memory variant of sdiv / udivAnton Korobeynikov2010-01-041-26/+26
| | | | llvm-svn: 92472
* implement an instcombine xform needed by clang's codegenChris Lattner2010-01-041-13/+0
| | | | | | | | on the example in PR4216. This doesn't trigger in the testsuite, so I'd really appreciate someone scrutinizing the logic for correctness. llvm-svn: 92458
* Teach codegen to lower llvm.powi to an efficient (but not optimal) Chris Lattner2010-01-011-30/+0
| | | | | | | | | | multiply sequence when the power is a constant integer. Before, our codegen for std::pow(.., int) always turned into a libcall, which was really inefficient. This should also make many gfortran programs happier I'd imagine. llvm-svn: 92388
* update this. To take the next step, llvm.powi should be generalized to workChris Lattner2010-01-011-13/+37
| | | | | | on integers as well and codegen should lower them to branch trees. llvm-svn: 92382
* Replace a few more SmallVectors with arrays.Benjamin Kramer2009-12-291-45/+29
| | | | llvm-svn: 92265
* Extern declaration for unordered.f32 libcall was not being emitted. Fixed that.Sanjiv Gupta2009-12-291-0/+1
| | | | llvm-svn: 92242
* move debug info stuff out of line, allowing two #includesChris Lattner2009-12-281-0/+1
| | | | | | to go away from IRBuilder.h llvm-svn: 92230
* Add missing include (for inline PATypeHolder::get).Benjamin Kramer2009-12-281-0/+1
| | | | llvm-svn: 92222
* Fixed llc crash for zext (i1 -> i8) loads.Sanjiv Gupta2009-12-281-2/+2
| | | | llvm-svn: 92201
* Allow targets to specify the return type of libcalls that are generated for ↵Sanjiv Gupta2009-12-282-0/+6
| | | | | | floating point comparisons, rather than hard-coding them as i32. llvm-svn: 92199
* Mark variable used by 'assert' as 'unused'.Bill Wendling2009-12-281-2/+2
| | | | llvm-svn: 92198
* Remove dead variable.Bill Wendling2009-12-281-1/+0
| | | | llvm-svn: 92197
* Remove dead variable.Bill Wendling2009-12-281-5/+1
| | | | llvm-svn: 92196
* Remove dead variable.Bill Wendling2009-12-281-1/+0
| | | | llvm-svn: 92195
* Remove dead variable.Bill Wendling2009-12-281-1/+0
| | | | llvm-svn: 92194
* Remove dead variable.Bill Wendling2009-12-281-2/+0
| | | | llvm-svn: 92193
* Remove dead variable.Bill Wendling2009-12-281-2/+1
| | | | llvm-svn: 92186
* Remove dead variable.Bill Wendling2009-12-281-1/+0
| | | | llvm-svn: 92185
* Remove dead variable.Bill Wendling2009-12-281-1/+0
| | | | llvm-svn: 92184
* Remove dead store. The initial value was never used, but always overridden.Bill Wendling2009-12-281-7/+6
| | | | llvm-svn: 92182
* Add an "ATTRIBUTE_UNUSED" macro (and use it). It's for variables which areBill Wendling2009-12-281-1/+2
| | | | | | | mainly used in debugging and/or assert situations. It should make the compiler and the static analyzer stop nagging us about them. llvm-svn: 92181
* PR5886: Make sure IMUL32m is marked as setting EFLAGS, so scheduling doesn'tEli Friedman2009-12-261-1/+1
| | | | | | do illegal stuff around it. No testcase because the issue is very fragile. llvm-svn: 92167
* Move kill flags when the same register occurs more than once in a sequence.Jakob Stoklund Olesen2009-12-231-1/+22
| | | | llvm-svn: 92058
* Handle undef operands properly.Jakob Stoklund Olesen2009-12-231-4/+8
| | | | llvm-svn: 92054
* Make insert position available to MergeOpsUpdate.Jakob Stoklund Olesen2009-12-231-24/+25
| | | | | | | Rearrange arguments. No functional changes llvm-svn: 92053
* Perform kill flag calculations in new method. No functional changes.Jakob Stoklund Olesen2009-12-231-12/+15
| | | | llvm-svn: 92052
* Move repeated code to a new method. No functional change.Jakob Stoklund Olesen2009-12-231-17/+47
| | | | llvm-svn: 92051
* Reapply 91904.Sanjiv Gupta2009-12-233-10/+44
| | | | llvm-svn: 91996
* Added missing patterns for subtract instruction.Sanjiv Gupta2009-12-232-34/+64
| | | | llvm-svn: 91995
* Reverting back 91904.Sanjiv Gupta2009-12-233-45/+10
| | | | llvm-svn: 91993
* really remove the instruction, don't just comment it outChris Lattner2009-12-231-7/+4
| | | | llvm-svn: 91976
* completely eliminate the MOV16r0 'instruction'. The onlyChris Lattner2009-12-234-19/+11
| | | | | | | interesting part of this is the divrem changes, which are already tested by CodeGen/X86/divrem.ll. llvm-svn: 91975
* More fixes for Visual C++. Replaced several very smallSean Callanan2009-12-231-10/+10
| | | | | | static inline functions with macros. llvm-svn: 91973
* stop pattern matching 16-bit zero's of a register to MOV16r0,Chris Lattner2009-12-231-4/+10
| | | | | | | | | instead use the appropriate subreggy thing. This generates identical code on some large apps (thanks to Evan's cross class coalescing stuff he did back in july). This means that MOV16r0 can go away completely in the future soon. llvm-svn: 91972
* Add a SPR register class to the ARM target.Jakob Stoklund Olesen2009-12-222-1/+14
| | | | | | Certain Thumb instructions require only SP (e.g. tSTRspi). llvm-svn: 91944
OpenPOWER on IntegriCloud