summaryrefslogtreecommitdiffstats
path: root/llvm
Commit message (Collapse)AuthorAgeFilesLines
...
* Correct the form of the atomic opcode names in a comment.Dan Gohman2009-01-081-2/+2
| | | | llvm-svn: 61947
* Do not inline functions with (dynamic) alloca intoDale Johannesen2009-01-085-4/+70
| | | | | | | | | | | functions that don't already have a (dynamic) alloca. Dynamic allocas cause inefficient codegen and we shouldn't propagate this (behavior follows gcc). Two existing tests assumed such inlining would be done; they are hacked by adding an alloca in the caller, preserving the point of the tests. llvm-svn: 61946
* Use mayBeOverridden here, in anticipation of theDuncan Sands2009-01-081-2/+2
| | | | | | day when more linkage types will be handled. llvm-svn: 61944
* ValueTracker can't assume that an alloca with no specified alignment Chris Lattner2009-01-082-1/+27
| | | | | | | | will get its preferred alignment. It has to be careful and cautiously assume it will just get the ABI alignment. This prevents instcombine from rounding up the alignment of a load/store without adjusting the alignment of the alloca. llvm-svn: 61934
* one more crash from PR3281, we now diagnose:Chris Lattner2009-01-081-2/+9
| | | | | | | llvm-as: t.ll:2:39: function may not return opaque type %"bwmoyl" = tail call coldcc opaque @g() ^ llvm-svn: 61933
* remove some exclusions that don't exist anymore.Chris Lattner2009-01-081-3/+0
| | | | llvm-svn: 61932
* this testcase is huge and hasn't regressed ever, I don't think it is worth ↵Chris Lattner2009-01-081-24452/+0
| | | | | | keeping. llvm-svn: 61931
* Add DebugInfo based APIs to record source line info.Devang Patel2009-01-081-4/+43
| | | | llvm-svn: 61928
* * Moved author attribution to CREDITS.TXTMisha Brukman2009-01-082-10/+11
| | | | | | * Removed trailing whitespace llvm-svn: 61927
* * Alphabetized #includesMisha Brukman2009-01-081-56/+56
| | | | | | * Removed trailing whitespace llvm-svn: 61926
* Some generic clean-ups. Also make the StringMapEntryInitializer ↵Bill Wendling2009-01-081-45/+44
| | | | | | specialization apply only to the tests that are actually testing it. llvm-svn: 61923
* * Don't explicitly cast "0" to "void*". This doesn't work well with specializedBill Wendling2009-01-082-6/+20
| | | | | | | | | StringMapEntryInitializer classes. Leave it for the compiler to figure out what the type is and what "0" should be transformed into. * Un-disable the unit tests which test the StringMapEntryInitializer class. llvm-svn: 61922
* the new scalarrepl changes are optimizing away a temporary alloca in Chris Lattner2009-01-081-41/+0
| | | | | | | | check242, which invalidates this test. This test is an x86-32 ABI test that is trying to be run in a target-independent way, which is not going to work very well. Just remove the test. llvm-svn: 61921
* 80-column violation fix.Bill Wendling2009-01-081-1/+1
| | | | llvm-svn: 61919
* add some more crazy strlen and memcpy stuff I noticed in spec.Chris Lattner2009-01-081-0/+99
| | | | llvm-svn: 61918
* add some notes about strlen craziness in eon.Chris Lattner2009-01-081-0/+57
| | | | llvm-svn: 61917
* Remove extra blank line and space.Misha Brukman2009-01-081-2/+1
| | | | llvm-svn: 61916
* This implements the second half of the fix for PR3290, handlingChris Lattner2009-01-082-2/+125
| | | | | | | | | loads from allocas that cover the entire aggregate. This handles some memcpy/byval cases that are produced by llvm-gcc. This triggers a few times in kc++ (with std::pair<std::_Rb_tree_const_iterator <kc::impl_abstract_phylum*>,bool>) and once in 176.gcc (with %struct..0anon). llvm-svn: 61915
* * Added unittests for StringMapMisha Brukman2009-01-082-2/+193
| | | | | | | | | * Fixed but in StringMap::clear() * Removed trailing whitespace Original patch by Talin. llvm-svn: 61914
* Add APIs to record regions and variables.Devang Patel2009-01-081-0/+27
| | | | | | Again, shamelessly copied from MMI. llvm-svn: 61912
* Add APIs to manage scope using DebugInfo interface.Devang Patel2009-01-081-0/+297
| | | | | | This is a shameless copy of similar APIs from MachineModuleInfo. The copy from MMI will be deleted in near future. llvm-svn: 61908
* * Fixed spelling of "sentinel"Misha Brukman2009-01-081-23/+23
| | | | | | * Removed trailing whitespace llvm-svn: 61907
* Use VIM's built-in shorthand for whitespace in regex.Misha Brukman2009-01-081-1/+1
| | | | llvm-svn: 61906
* Be sure to ignore the end-of-line character when considering trailingMisha Brukman2009-01-081-2/+2
| | | | | | whitespace. llvm-svn: 61905
* Removed trailing whitespace.Misha Brukman2009-01-082-48/+48
| | | | llvm-svn: 61904
* Fix failure messages in Verifier::PerformTypeCheck. The argument numbersBob Wilson2009-01-081-47/+42
| | | | | | | | | | | passed in to this function changed to support multiple return values, leading to some incorrect argument numbers in the failure messages. With this change, the ArgNo values used for return values and parameters are disjoint, and the new IntrinsicParam function translates those ArgNo values to strings that can be used in the messages. This also fixes a few places where PerformTypeCheck did not return false following calls to CheckFailed. llvm-svn: 61903
* CMake: removed lib/VMCore/DebugInfoBuilder.cpp.Oscar Fuentes2009-01-081-1/+0
| | | | llvm-svn: 61900
* Assert that VectorType::getTruncatedElementVectorType is not used withBob Wilson2009-01-072-4/+12
| | | | | | odd bit-width vector elements. Add a check in the verifier for this also. llvm-svn: 61899
* remove DebugInfoBuilder, it has been subsumed by Analysis/DebugInfo.hChris Lattner2009-01-072-425/+0
| | | | llvm-svn: 61898
* Fix off-by-one error in traversing an array; this fixes a test.Misha Brukman2009-01-072-3/+1
| | | | | | The error was reported by gcc-4.3.0 during compilation. llvm-svn: 61896
* revert to functionally equivalent formulationGabor Greif2009-01-071-1/+1
| | | | llvm-svn: 61895
* use the obvious gettersGabor Greif2009-01-071-3/+3
| | | | llvm-svn: 61893
* Remove redundant 'else's. No functionality change.Dan Gohman2009-01-071-7/+6
| | | | llvm-svn: 61891
* Add empty() methods for register def lists.Dan Gohman2009-01-071-0/+8
| | | | llvm-svn: 61890
* Minor cleanup for unittest:Misha Brukman2009-01-071-3/+3
| | | | | | | | | * Fixed {copy,assignment} constructor test names * s/EXPECT_EQ(true, ...)/ASSERT_TRUE(...)/ Patch by Talin. llvm-svn: 61883
* Whitespace - correct formatting.Duncan Sands2009-01-071-2/+2
| | | | llvm-svn: 61879
* Remove alloca tracking from nocapture analysis. Not onlyDuncan Sands2009-01-072-87/+25
| | | | | | | | | | was it not very helpful, it was also wrong! The problem is shown in the testcase: the alloca might be passed to a nocapture callee which dereferences it and returns the original pointer. But because it was a nocapture call we think we don't need to track its uses, but we do. llvm-svn: 61876
* CMake: replace `rm' with portable invocations of cmake.Oscar Fuentes2009-01-072-5/+5
| | | | | | Based on a bug report by Yonggang Luo. llvm-svn: 61875
* Reorder these.Duncan Sands2009-01-071-24/+24
| | | | llvm-svn: 61873
* Use a switch rather than a sequence of "isa" tests.Duncan Sands2009-01-071-16/+32
| | | | llvm-svn: 61872
* The verifier checks that the aliasee is not null.Duncan Sands2009-01-071-2/+1
| | | | llvm-svn: 61870
* Implement the first half of PR3290: if there is a store of an Chris Lattner2009-01-072-5/+164
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | integer to a (transitive) bitcast the alloca and if that integer has the full size of the alloca, then it clobbers the whole thing. Handle this by extracting pieces out of the stored integer and filing them away in the SROA'd elements. This triggers fairly frequently because the CFE uses integers to pass small structs by value and the inliner exposes these. For example, in kimwitu++, I see a bunch of these with i64 stores to "%struct.std::pair<std::_Rb_tree_const_iterator<kc::impl_abstract_phylum*>,bool>" In 176.gcc I see a few i32 stores to "%struct..0anon". In the testcase, this is a difference between compiling test1 to: _test1: subl $12, %esp movl 20(%esp), %eax movl %eax, 4(%esp) movl 16(%esp), %eax movl %eax, (%esp) movl (%esp), %eax addl 4(%esp), %eax addl $12, %esp ret vs: _test1: movl 8(%esp), %eax addl 4(%esp), %eax ret The second half of this will be to handle loads of the same form. llvm-svn: 61853
* Factor a bunch of code out into a helper method.Chris Lattner2009-01-071-148/+156
| | | | llvm-svn: 61852
* use continue to simplify code and reduce nesting, no functionalityChris Lattner2009-01-071-38/+58
| | | | | | change. llvm-svn: 61851
* Get TargetData once up front and cache as an ivar instead ofChris Lattner2009-01-071-45/+38
| | | | | | requerying it all over the place. llvm-svn: 61850
* Use the hasAllZeroIndices predicate to simplify some Chris Lattner2009-01-071-28/+6
| | | | | | code, no functionality change. llvm-svn: 61849
* The coalescer does not coalesce a virtual register to a physical register if ↵Evan Cheng2009-01-074-20/+115
| | | | | | | | | | | | | | any of the physical register's sub-register live intervals overlaps with the virtual register. This is overly conservative. It prevents a extract_subreg from being coalesced away: v1024 = EDI // not killed = = EDI One possible solution is for the coalescer to examine the sub-register live intervals in the same manner as the physical register. Another possibility is to examine defs and uses (when needed) of sub-registers. Both solutions are too expensive. For now, look for "short virtual intervals" and scan instructions to look for conflict instead. This is a small win on x86-64. e.g. It shaves 403.gcc by ~80 instructions. llvm-svn: 61847
* add a testcase.Chris Lattner2009-01-071-0/+56
| | | | llvm-svn: 61845
* Add patterns to match conditional moves with loads foldedDan Gohman2009-01-073-1/+111
| | | | | | | into their left operand, rather than their right. Do this by commuting the operands and inverting the condition. llvm-svn: 61842
* Add load-folding table entries for cmovno too.Dan Gohman2009-01-071-0/+3
| | | | llvm-svn: 61841
OpenPOWER on IntegriCloud