summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Transforms/Scalar
Commit message (Collapse)AuthorAgeFilesLines
* Check for !isa<Constant> instead of isa<Instruction>. ThisDan Gohman2009-08-041-1/+1
| | | | | | | matches what the comment says, and it avoids spurious BitCast instructions for Argument values. llvm-svn: 78121
* Add a new Constant::getIntegerValue helper function, and convert aDan Gohman2009-08-031-8/+4
| | | | | | | few places in InstCombine to use it, to fix problems handling pointer types. This fixes the recent llvm-gcc bootstrap error. llvm-svn: 78005
* Make SimplifyDemandedUseBits generate vector constants where Eli Friedman2009-08-031-2/+2
| | | | | | | | appropriate. Patch per report on llvmdev. No testcase because the original report didn't come with a testcase, and I can't come up with a case that actually fails. llvm-svn: 77986
* Move a few more APIs back to 2.5 forms. The only remaining ones left to ↵Owen Anderson2009-07-319-137/+124
| | | | | | | | change back are metadata related, which I'm waiting on to avoid conflicting with Devang. llvm-svn: 77721
* Fix GVN's debug output, now that operator<< on Value* doesn't printDan Gohman2009-07-311-11/+12
| | | | | | a trailing newline. llvm-svn: 77719
* PR4662: Fix a crash introduced by the recent LLVMContext changes.Eli Friedman2009-07-311-1/+2
| | | | llvm-svn: 77716
* Move getTrue() and getFalse() to 2.5-like APIs.Owen Anderson2009-07-316-103/+106
| | | | llvm-svn: 77685
* Move more code back to 2.5 APIs.Owen Anderson2009-07-3010-77/+72
| | | | llvm-svn: 77635
* Twines: Don't allow implicit conversion from integers, this is too tricky.Daniel Dunbar2009-07-301-3/+3
| | | | llvm-svn: 77605
* Switch obvious clients to Twine instead of utostr (when they were already usingDaniel Dunbar2009-07-301-4/+3
| | | | | | | | | a Twine, e.g., for names). - I am a little ambivalent about this; we don't want the string conversion of utostr, but using overload '+' mixed with string and integer arguments is sketchy. On the other hand, this particular usage is something of an idiom. llvm-svn: 77579
* Eliminate a few unused-variable warningsDouglas Gregor2009-07-291-2/+0
| | | | llvm-svn: 77519
* Move types back to the 2.5 API.Owen Anderson2009-07-295-55/+52
| | | | llvm-svn: 77516
* Fix PR4645 which was fallout from the fix for PR4641.Daniel Dunbar2009-07-291-8/+15
| | | | | | - Call RAUW to delete all instructions (this is a patch from Nick Lewycky). llvm-svn: 77512
* Move ConstantExpr to 2.5 API.Owen Anderson2009-07-299-139/+138
| | | | llvm-svn: 77494
* Bulk erasing instructions without RAUWing them is unsafe. Instead, break themNick Lewycky2009-07-291-7/+9
| | | | | | into a new BB that has no predecessors. llvm-svn: 77433
* Return ConstantVector to 2.5 API.Owen Anderson2009-07-282-10/+10
| | | | llvm-svn: 77366
* Change ConstantArray to 2.5 API.Owen Anderson2009-07-281-1/+1
| | | | llvm-svn: 77347
* Teach instcombine to respect and preserve inbounds. Add inboundsDan Gohman2009-07-281-21/+53
| | | | | | to a few tests where it is required for the expected transformation. llvm-svn: 77290
* Replace dyn_castGetElementPtr with dyn_cast<GEPOperator>.Dan Gohman2009-07-281-15/+5
| | | | llvm-svn: 77286
* Grab the LLVMContext and parent Module of SI ahead of theDan Gohman2009-07-281-3/+4
| | | | | | | point where SI can get deleted. This fixes a use of free'd memory. This fixes Externals/Povray. llvm-svn: 77285
* Avoid build warnings.Mike Stump2009-07-271-0/+1
| | | | llvm-svn: 77271
* Move ConstantFP construction back to the 2.5-ish API.Owen Anderson2009-07-273-9/+7
| | | | llvm-svn: 77247
* Remove Value::getName{Start,End}, the last of the old Name APIs.Daniel Dunbar2009-07-266-14/+12
| | | | llvm-svn: 77152
* Remove Value::getNameLenDaniel Dunbar2009-07-261-7/+6
| | | | llvm-svn: 77148
* Eliminate some uses of DOUT, cerr, and getNameStart().Daniel Dunbar2009-07-263-217/+217
| | | | llvm-svn: 77145
* Remove Value::{isName, getNameRef}.Daniel Dunbar2009-07-251-1/+1
| | | | | | Also, change MDString to use a StringRef. llvm-svn: 77098
* Initial update to VMCore to use Twines for string arguments.Daniel Dunbar2009-07-252-3/+6
| | | | | | | - The only meat here is in Value.{h,cpp} the rest is essential 'const std::string &' -> 'const Twine &'. llvm-svn: 77048
* Fix 80-col violations.Eric Christopher2009-07-251-5/+8
| | | | llvm-svn: 77045
* Move ExtractElementInst to ::Create instead of new. Update all uses.Eric Christopher2009-07-252-9/+9
| | | | llvm-svn: 77044
* Convert a few more things to use raw_ostream.Dan Gohman2009-07-251-14/+14
| | | | llvm-svn: 77039
* Convert a few more uses of llvm/Support/Streams.h to raw_ostream.Dan Gohman2009-07-251-5/+6
| | | | llvm-svn: 77033
* Make AliasAnalysis and related classes useDan Gohman2009-07-251-1/+1
| | | | | | getAnalysisIfAvailable<TargetData>(). llvm-svn: 77028
* More migration to raw_ostream, the water has dried up around the iostream hole.Daniel Dunbar2009-07-258-51/+65
| | | | | | | | | | - Some clients which used DOUT have moved to DEBUG. We are deprecating the "magic" DOUT behavior which avoided calling printing functions when the statement was disabled. In addition to being unnecessary magic, it had the downside of leaving code in -Asserts builds, and of hiding potentially unnecessary computations. llvm-svn: 77019
* Revert the ConstantInt constructors back to their 2.5 forms where possible, ↵Owen Anderson2009-07-2412-217/+221
| | | | | | thanks to contexts-on-types. More to come. llvm-svn: 77011
* Fix this condition I accidentally inverted.Dan Gohman2009-07-241-1/+1
| | | | llvm-svn: 76988
* Convert several more passes to use getAnalysisIfAvailable<TargetData>()Dan Gohman2009-07-243-26/+28
| | | | | | instead of getAnalysis<TargetData>(). llvm-svn: 76982
* Convert StringMap to using StringRef for its APIs.Daniel Dunbar2009-07-231-1/+1
| | | | | | | | | | | | - Yay for '-'s and simplifications! - I kept StringMap::GetOrCreateValue for compatibility purposes, this can eventually go away. Likewise the StringMapEntry Create functions still follow the old style. - NIFC. llvm-svn: 76888
* refactor a blob of code out to a new 'FoldOrOfFCmps' function andChris Lattner2009-07-231-67/+70
| | | | | | simplify it. llvm-svn: 76866
* Make some existing optimizations that would only trigger on scalarsChris Lattner2009-07-231-2/+20
| | | | | | | | | | | | | | | | | | | | | also apply to vectors. This allows us to compile this: #include <emmintrin.h> __m128i a(__m128 a, __m128 b) { return a==a & b==b; } __m128i b(__m128 a, __m128 b) { return a!=a | b!=b; } to: _a: cmpordps %xmm1, %xmm0 ret _b: cmpunordps %xmm1, %xmm0 ret with clang instead of to a ton of horrible code. llvm-svn: 76863
* refactor a bunch of code out into a helper function,Chris Lattner2009-07-231-63/+73
| | | | | | no functionality change. llvm-svn: 76859
* Get rid of the Pass+Context magic.Owen Anderson2009-07-2221-158/+210
| | | | llvm-svn: 76702
* Convert instcombine from using using getAnalysis<TargetData> toDan Gohman2009-07-211-46/+62
| | | | | | getAnalysisIfAvailable<TargetData>. llvm-svn: 76676
* Rename getConstantInt{True|False} to get{True|False} at Chris' behest.Owen Anderson2009-07-216-103/+103
| | | | llvm-svn: 76598
* Move a bit more state over to the LLVMContext.Owen Anderson2009-07-211-14/+17
| | | | llvm-svn: 76533
* use ExpandInlineAsm on TargetLowering instead of TargetAsmInfo.Chris Lattner2009-07-201-14/+10
| | | | llvm-svn: 76442
* Revert the addition of hasNoPointerOverflow to GEPOperator.Dan Gohman2009-07-201-84/+1
| | | | | | | | | | | | Getelementptrs that are defined to wrap are virtually useless to optimization, and getelementptrs that are undefined on any kind of overflow are too restrictive -- it's difficult to ensure that all intermediate addresses are within bounds. I'm going to take a different approach. Remove a few optimizations that depended on this flag. llvm-svn: 76437
* Canonicalize bitcasts between types like <1 x i64> and i64 to Eli Friedman2009-07-181-0/+24
| | | | | | | | | | insertelement/extractelement. I'm not entirely sure this is precisely what we want to do: should we prefer bitcast(insertelement) or insertelement(bitcast)? Similarly. should we prefer extractelement(bitcast) or bitcast(extractelement)? llvm-svn: 76345
* Back out 76300; apparently the preference is to canonicalize the other Eli Friedman2009-07-181-16/+6
| | | | | | way (bitcast -> insert/extractelement). llvm-svn: 76325
* Add combine: X sdiv (1 << Y) -> X udiv (1 << Y) when X doesn't have the Eli Friedman2009-07-181-4/+15
| | | | | | sign bit set. llvm-svn: 76304
* Remove no-op check.Eli Friedman2009-07-181-2/+1
| | | | llvm-svn: 76302
OpenPOWER on IntegriCloud