| 2008-10-06 |
Duncan Sands | Actually run Obj-C++ tests if llvm-gcc supports.
|
commit | commitdiff | tree |
| 2008-10-06 |
Duncan Sands | Clarify the relationship between byval and readonly/
|
commit | commitdiff | tree |
| 2008-10-04 |
Duncan Sands | Ignore loads from and stores to local memory (i.e....
|
commit | commitdiff | tree |
| 2008-10-03 |
Duncan Sands | The result of getSetCCResultType (eg: i32) may be larger
|
commit | commitdiff | tree |
| 2008-10-03 |
Duncan Sands | Teach internalize to preserve the callgraph.
|
commit | commitdiff | tree |
| 2008-10-01 |
Duncan Sands | Factorize code: remove variants of "strip off
|
commit | commitdiff | tree |
| 2008-09-30 |
Duncan Sands | Stacker is dead.
|
commit | commitdiff | tree |
| 2008-09-30 |
Duncan Sands | Revert commit 56835 since it breaks the build.
|
commit | commitdiff | tree |
| 2008-09-29 |
Duncan Sands | Speed up these passes when the callgraph has
|
commit | commitdiff | tree |
| 2008-09-29 |
Duncan Sands | Tweak some comments.
|
commit | commitdiff | tree |
| 2008-09-29 |
Duncan Sands | Rename isWeakForLinker to mayBeOverridden. Use it
|
commit | commitdiff | tree |
| 2008-09-23 |
Duncan Sands | Rationalize the names of passes that print information:
|
commit | commitdiff | tree |
| 2008-09-20 |
Duncan Sands | Implement review feedback from Devang: make use
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Turn on the AddReadAttrs pass.
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Add test for improvement of readonly to readnone,
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Turn on these tests!
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Remove the MarkModRef pass (use AddReadAttrs instead).
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Add a new pass AddReadAttrs which works out which functions
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Test the callgraph directly for the missing edge.
|
commit | commitdiff | tree |
| 2008-09-19 |
Duncan Sands | Teach -callgraph to always print the callgraph (as the
|
commit | commitdiff | tree |
| 2008-09-15 |
Duncan Sands | End of the GlobalsModRef experiment.
|
commit | commitdiff | tree |
| 2008-09-13 |
Duncan Sands | Fix PR2792: treat volatile loads as writing memory...
|
commit | commitdiff | tree |
| 2008-09-12 |
Duncan Sands | The "alias" keyword comes first.
|
commit | commitdiff | tree |
| 2008-09-12 |
Duncan Sands | Give GlobalsModRef a whirl in the nightly testers.
|
commit | commitdiff | tree |
| 2008-09-12 |
Duncan Sands | Rather than marking all internal globals "Ref"
|
commit | commitdiff | tree |
| 2008-09-11 |
Duncan Sands | Fix comment typo.
|
commit | commitdiff | tree |
| 2008-09-11 |
Duncan Sands | Intrinsics don't touch internal global variables
|
commit | commitdiff | tree |
| 2008-09-11 |
Duncan Sands | Intrinsics don't read these kinds of global
|
commit | commitdiff | tree |
| 2008-09-10 |
Duncan Sands | Fix name.
|
commit | commitdiff | tree |
| 2008-09-10 |
Duncan Sands | Add trampoline support for the new FastCC calling
|
commit | commitdiff | tree |
| 2008-09-10 |
Duncan Sands | Turn off the new FastCC for the moment. It causes
|
commit | commitdiff | tree |
| 2008-09-09 |
Duncan Sands | Simplify this some more. No functionality change.
|
commit | commitdiff | tree |
| 2008-09-09 |
Duncan Sands | Optimization suggested by Matthijs Kooijman.
|
commit | commitdiff | tree |
| 2008-09-09 |
Duncan Sands | Correct callgraph construction. It has two problems:
|
commit | commitdiff | tree |
| 2008-09-08 |
Duncan Sands | Didn't mean to commit this change to how the
|
commit | commitdiff | tree |
| 2008-09-08 |
Duncan Sands | Update the callgraph correctly.
|
commit | commitdiff | tree |
| 2008-09-08 |
Duncan Sands | Update the callgraph correctly in ArgumentPromotion.
|
commit | commitdiff | tree |
| 2008-09-08 |
Duncan Sands | Reapply 55859. This doesn't change anything as
|
commit | commitdiff | tree |
| 2008-09-07 |
Duncan Sands | Correct a comment and strip trailing whitespace.
|
commit | commitdiff | tree |
| 2008-09-06 |
Duncan Sands | When PruneEH turned an invoke into an ordinary
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Delete the removeCallEdgeTo callgraph method,
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Prevent our own passes from promoting this to
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Use removeAllCalledFunctions rather than removing
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Remove trailing whitespace.
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Strip trailing whitespace.
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | Make this pass return that it made a change if
|
commit | commitdiff | tree |
| 2008-09-05 |
Duncan Sands | "Fix" PR2762. The testcase now crashes codegen
|
commit | commitdiff | tree |
| 2008-09-04 |
Duncan Sands | Neaten this up a bit. No functionality change.
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | Testcase for commits 55700 and 55714.
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | If a SCC has a node without a function, then the SCC
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | Fix maxo bado thinko.
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | Since onlyReadsMemory returns true if in fact
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | Cleanup GlobalsModRef a bit. When analysing the
|
commit | commitdiff | tree |
| 2008-09-03 |
Duncan Sands | Testcase for PR1678.
|
commit | commitdiff | tree |
| 2008-09-01 |
Duncan Sands | Even though no caller actually uses the new value
|
commit | commitdiff | tree |
| 2008-09-01 |
Duncan Sands | Turn this legalize types test on.
|
commit | commitdiff | tree |
| 2008-09-01 |
Duncan Sands | Add a small pass that sets the readnone/readonly
|
commit | commitdiff | tree |
| 2008-08-13 |
Duncan Sands | Teach constant folding that an inttoptr of a
|
commit | commitdiff | tree |
| 2008-08-13 |
Duncan Sands | Rename this, in case people think that NullFolder
|
commit | commitdiff | tree |
| 2008-08-12 |
Duncan Sands | Add a NullFolder class that doesn't fold constants.
|
commit | commitdiff | tree |
| 2008-08-12 |
Duncan Sands | Point people to ConstantExpr and ConstantFolding,
|
commit | commitdiff | tree |
| 2008-08-11 |
Duncan Sands | Make it possible to use different constant
|
commit | commitdiff | tree |
| 2008-08-09 |
Duncan Sands | Unbreak the llvm-gcc build: recent changes to
|
commit | commitdiff | tree |
| 2008-08-07 |
Duncan Sands | Use correct compiler component names for objc
|
commit | commitdiff | tree |
| 2008-08-07 |
Duncan Sands | Move tests of the ObjC++ front-end to their own
|
commit | commitdiff | tree |
| 2008-08-01 |
Duncan Sands | Fix comment typos.
|
commit | commitdiff | tree |
| 2008-07-29 |
Duncan Sands | Fix PR2609. If a label is deleted, then it needs
|
commit | commitdiff | tree |
| 2008-07-29 |
Duncan Sands | Set the executable bit - means it can actually
|
commit | commitdiff | tree |
| 2008-07-28 |
Duncan Sands | Since build_vector is a variadic node, the number
|
commit | commitdiff | tree |
| 2008-07-28 |
Duncan Sands | This is not a binary file.
|
commit | commitdiff | tree |
| 2008-07-28 |
Duncan Sands | Test this differently: I saw this test fail
|
commit | commitdiff | tree |
| 2008-07-27 |
Duncan Sands | Some binary operations were being treated as
|
commit | commitdiff | tree |
| 2008-07-25 |
Duncan Sands | Fix error: ‘std::ostream’ has not been declared
|
commit | commitdiff | tree |
| 2008-07-22 |
Duncan Sands | LegalizeTypes support for VSETCC. Fixes PR2575.
|
commit | commitdiff | tree |
| 2008-07-21 |
Duncan Sands | Add VerifyNode, a place to put sanity checks on
|
commit | commitdiff | tree |
| 2008-07-19 |
Duncan Sands | Make sure custom lowering for LegalizeTypes
|
commit | commitdiff | tree |
| 2008-07-18 |
Duncan Sands | Softfloat support for FDIV. Patch by
|
commit | commitdiff | tree |
| 2008-07-18 |
Duncan Sands | Eliminate unused variable.
|
commit | commitdiff | tree |
| 2008-07-18 |
Duncan Sands | Supress a gcc-4.3 warning.
|
commit | commitdiff | tree |
| 2008-07-18 |
Duncan Sands | Revert 53729, after waking up in the middle of
|
commit | commitdiff | tree |
| 2008-07-17 |
Duncan Sands | Use a legal type for elements of the vector_shuffle
|
commit | commitdiff | tree |
| 2008-07-17 |
Duncan Sands | LegalizeTypes support for what seems to be the
|
commit | commitdiff | tree |
| 2008-07-17 |
Duncan Sands | Turn LegalizeTypes back off again for the moment:
|
commit | commitdiff | tree |
| 2008-07-17 |
Duncan Sands | Factorize some code for determining which libcall to...
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | Add support for promoting and expanding AssertZext
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | Test passing of integer parameters for integers
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | Test codegen of loads and stores of all integer
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | Reorder methods alphabetically. No functionality change.
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | Turn on LegalizeTypes by default.
|
commit | commitdiff | tree |
| 2008-07-16 |
Duncan Sands | The atomic.cmp.swap promotion logic is wrong: it
|
commit | commitdiff | tree |
| 2008-07-15 |
Duncan Sands | LegalizeTypes support for fabs on ppc long double.
|
commit | commitdiff | tree |
| 2008-07-15 |
Duncan Sands | LegalizeTypes support for promotion of bswap.
|
commit | commitdiff | tree |
| 2008-07-15 |
Duncan Sands | LegalizeTypes support for promotion of SIGN_EXTEND_INREG.
|
commit | commitdiff | tree |
| 2008-07-15 |
Duncan Sands | Reorder the integer promotion methods alphabetically.
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | I don't think BUILD_PAIR can have a vector result.
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | Tighten up some checks. Fix FPOWI splitting for
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | An INSERT_VECTOR_ELT can insert a larger value
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | According to the docs, it is possible to have an
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | There should be no extending loads or truncating
|
commit | commitdiff | tree |
| 2008-07-14 |
Duncan Sands | Ignore TargetConstant with an illegal type. These
|
commit | commitdiff | tree |
| next |
projects / oota-llvm.git / search