| 2009-02-25 |
Duncan Sands | Check that records with a known constant size are not
|
commit | commitdiff | tree |
| 2009-02-25 |
Duncan Sands | Flesh out the Ada section a little.
|
commit | commitdiff | tree |
| 2009-02-25 |
Duncan Sands | The C backend does not support arbitrary precision...
|
commit | commitdiff | tree |
| 2009-02-25 |
Duncan Sands | Note on new type legalizer.
|
commit | commitdiff | tree |
| 2009-02-25 |
Duncan Sands | Add note on support for arbitrary precision integers.
|
commit | commitdiff | tree |
| 2009-02-18 |
Duncan Sands | In theory the aliasee may have dead constant users
|
commit | commitdiff | tree |
| 2009-02-17 |
Duncan Sands | If an alias is dead and so is its aliasee, then globaldce...
|
commit | commitdiff | tree |
| 2009-02-17 |
Duncan Sands | This transform also applies to private linkage.
|
commit | commitdiff | tree |
| 2009-02-15 |
Duncan Sands | Make this more useful for cleaning up after the
|
commit | commitdiff | tree |
| 2009-02-15 |
Duncan Sands | If the target of an alias has internal linkage, then the
|
commit | commitdiff | tree |
| 2009-02-14 |
Duncan Sands | Generalize some alias analysis logic from atomic
|
commit | commitdiff | tree |
| 2009-02-14 |
Duncan Sands | The trampoline intrinsic can be marked IntrWriteArgMem,
|
commit | commitdiff | tree |
| 2009-02-14 |
Duncan Sands | IntrWriteArgMem implies nocapture, but this wasn't
|
commit | commitdiff | tree |
| 2009-02-13 |
Duncan Sands | Correct a comment: since AccessArguments is now
|
commit | commitdiff | tree |
| 2009-02-13 |
Duncan Sands | If a function only reads memory, then we know that...
|
commit | commitdiff | tree |
| 2009-02-13 |
Duncan Sands | Reapply r64301. These uses of "inline" can cause strange
|
commit | commitdiff | tree |
| 2009-02-11 |
Duncan Sands | Revert r64299: it breaks the build when configured
|
commit | commitdiff | tree |
| 2009-02-11 |
Duncan Sands | These uses of "inline" can cause strange link-time
|
commit | commitdiff | tree |
| 2009-02-11 |
Duncan Sands | Make sure the SCC pass manager initializes any contained
|
commit | commitdiff | tree |
| 2009-02-10 |
Duncan Sands | This is now done using a real i33, rather than
|
commit | commitdiff | tree |
| 2009-02-10 |
Duncan Sands | Pacify gcc-4.3 (thinks IsConstant may be used
|
commit | commitdiff | tree |
| 2009-02-09 |
Duncan Sands | Testcase for PR2437.
|
commit | commitdiff | tree |
| 2009-02-05 |
Duncan Sands | Revert the previous commit. It seems it didn't
|
commit | commitdiff | tree |
| 2009-02-05 |
Duncan Sands | Stop llvm-config thinking that there are circular
|
commit | commitdiff | tree |
| 2009-02-04 |
Duncan Sands | Allow the inverse transform x86_fp80 -> i80 (also
|
commit | commitdiff | tree |
| 2009-02-04 |
Duncan Sands | Fix PR3468: a crash when constant folding a bitcast of
|
commit | commitdiff | tree |
| 2009-02-03 |
Duncan Sands | Fix PR3411. When replacing values, nodes are analyzed
|
commit | commitdiff | tree |
| 2009-02-02 |
Duncan Sands | Remove trailing spaces.
|
commit | commitdiff | tree |
| 2009-02-02 |
Duncan Sands | Fix a comment (bytes -> bits), reformat a comment
|
commit | commitdiff | tree |
| 2009-02-02 |
Duncan Sands | Fix an obvious thinko.
|
commit | commitdiff | tree |
| 2009-02-02 |
Duncan Sands | This passes on x86-32 linux at least.
|
commit | commitdiff | tree |
| 2009-02-02 |
Duncan Sands | Make the XFAIL line actually match x86-32 targets.
|
commit | commitdiff | tree |
| 2009-02-01 |
Duncan Sands | Fix PR3453 and probably a bunch of other potential
|
commit | commitdiff | tree |
| 2009-01-31 |
Duncan Sands | Fix PR3401: when using large integers, the type
|
commit | commitdiff | tree |
| 2009-01-28 |
Duncan Sands | Formatting.
|
commit | commitdiff | tree |
| 2009-01-28 |
Duncan Sands | Rename getAnalysisToUpdate to getAnalysisIfAvailable.
|
commit | commitdiff | tree |
| 2009-01-28 |
Duncan Sands | Fix PR3415 (infinite loop in EscapeAnalysis) by
|
commit | commitdiff | tree |
| 2009-01-26 |
Duncan Sands | Fix PR3393, which amounts to a bug in the expensive
|
commit | commitdiff | tree |
| 2009-01-23 |
Duncan Sands | Patch to build llvm with srcdir != objdir from
|
commit | commitdiff | tree |
| 2009-01-22 |
Duncan Sands | Remove no-longer relevant comment. Pointed out
|
commit | commitdiff | tree |
| 2009-01-22 |
Duncan Sands | This passes on linux.
|
commit | commitdiff | tree |
| 2009-01-21 |
Duncan Sands | This was causing invalid memory accesses when
|
commit | commitdiff | tree |
| 2009-01-21 |
Duncan Sands | Let's try to have our cake and eat it to: move
|
commit | commitdiff | tree |
| 2009-01-21 |
Duncan Sands | Don't rely on grep -w working.
|
commit | commitdiff | tree |
| 2009-01-21 |
Duncan Sands | Cleanup whitespace and comments, and tweak some
|
commit | commitdiff | tree |
| 2009-01-20 |
Duncan Sands | Fix typo. Patch by Alexei Svitkine.
|
commit | commitdiff | tree |
| 2009-01-20 |
Duncan Sands | Check that the "don't barf on k8" fix is not
|
commit | commitdiff | tree |
| 2009-01-20 |
Duncan Sands | If a vector is empty, you're not allowed to access any
|
commit | commitdiff | tree |
| 2009-01-18 |
Duncan Sands | BasicAliasAnalysis and FunctionAttrs were both
|
commit | commitdiff | tree |
| 2009-01-16 |
Duncan Sands | Get this building with gcc-4.4.
|
commit | commitdiff | tree |
| 2009-01-16 |
Duncan Sands | Grammar fix.
|
commit | commitdiff | tree |
| 2009-01-13 |
Duncan Sands | When replacing uses and the same node is reached
|
commit | commitdiff | tree |
| 2009-01-13 |
Duncan Sands | Mark this XFAIL for the moment.
|
commit | commitdiff | tree |
| 2009-01-13 |
Duncan Sands | Fix some typos. Also, the WidenedVectors map
|
commit | commitdiff | tree |
| 2009-01-13 |
Duncan Sands | Correct a comment - this is not a sign extension.
|
commit | commitdiff | tree |
| 2009-01-13 |
Duncan Sands | Correct a comment.
|
commit | commitdiff | tree |
| 2009-01-12 |
Duncan Sands | Rename getABITypeSize to getTypePaddedSize, as
|
commit | commitdiff | tree |
| 2009-01-08 |
Duncan Sands | Use mayBeOverridden here, in anticipation of the
|
commit | commitdiff | tree |
| 2009-01-07 |
Duncan Sands | Whitespace - correct formatting.
|
commit | commitdiff | tree |
| 2009-01-07 |
Duncan Sands | Remove alloca tracking from nocapture analysis. Not...
|
commit | commitdiff | tree |
| 2009-01-07 |
Duncan Sands | Reorder these.
|
commit | commitdiff | tree |
| 2009-01-07 |
Duncan Sands | Use a switch rather than a sequence of "isa" tests.
|
commit | commitdiff | tree |
| 2009-01-07 |
Duncan Sands | The verifier checks that the aliasee is not null.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Add a note about passing MVT::Other to getSetCCResultType.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Teach the internalize pass to also internalize
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | When checking if an Argument escapes, check if
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Not having an aliasee is a theoretical possibility.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Format more neatly.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Remove trailing spaces.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Delete unused global aliases with internal linkage.
|
commit | commitdiff | tree |
| 2009-01-05 |
Duncan Sands | Don't spew bitcode to standard out if this test
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | Load tracking means that the value analyzed may
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | When calculating 'nocapture' argument attributes, allow
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | Improve comments and reorganize a bit - no functionality
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | Deleted too many properties: restore some.
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | Remove the "ignore" property on this directory.
|
commit | commitdiff | tree |
| 2009-01-02 |
Duncan Sands | Add tests for two types of traps that escape analysis
|
commit | commitdiff | tree |
| 2009-01-01 |
Duncan Sands | Mention that this pass does escape analysis in the
|
commit | commitdiff | tree |
| 2009-01-01 |
Duncan Sands | Factorize (and generalize) the code promoting SELECT
|
commit | commitdiff | tree |
| 2009-01-01 |
Duncan Sands | Remove trailing spaces.
|
commit | commitdiff | tree |
| 2009-01-01 |
Duncan Sands | Fix PR3274: when promoting the condition of a BRCOND...
|
commit | commitdiff | tree |
| 2008-12-31 |
Duncan Sands | Look through phi nodes and select instructions when
|
commit | commitdiff | tree |
| 2008-12-31 |
Duncan Sands | Don't analyze arguments already marked 'nocapture'.
|
commit | commitdiff | tree |
| 2008-12-31 |
Duncan Sands | Rename AddReadAttrs to FunctionAttrs, and teach it how
|
commit | commitdiff | tree |
| 2008-12-29 |
Duncan Sands | Make stripPointerCasts and getUnderlyingObject
|
commit | commitdiff | tree |
| 2008-12-29 |
Duncan Sands | Experiments show that looking through phi nodes
|
commit | commitdiff | tree |
| 2008-12-29 |
Duncan Sands | Clarify a bit. Based on feedback by Talin.
|
commit | commitdiff | tree |
| 2008-12-29 |
Duncan Sands | Allow readnone functions to read (and write!) global
|
commit | commitdiff | tree |
| 2008-12-29 |
Duncan Sands | Add braces, as suggested by a gcc warning.
|
commit | commitdiff | tree |
| 2008-12-14 |
Duncan Sands | Reapply r60997, this time without forgetting that
|
commit | commitdiff | tree |
| 2008-12-13 |
Duncan Sands | LegalizeDAG is not supposed to introduce illegal
|
commit | commitdiff | tree |
| 2008-12-13 |
Duncan Sands | On big-endian machines it is wrong to do a full
|
commit | commitdiff | tree |
| 2008-12-12 |
Duncan Sands | When expanding unaligned loads and stores do not make
|
commit | commitdiff | tree |
| 2008-12-12 |
Duncan Sands | When using a 4 byte jump table on a 64 bit machine,
|
commit | commitdiff | tree |
| 2008-12-12 |
Duncan Sands | Don't make use of an illegal type (i64) when
|
commit | commitdiff | tree |
| 2008-12-10 |
Duncan Sands | Update these instructions to the 2.4 release.
|
commit | commitdiff | tree |
| 2008-12-10 |
Duncan Sands | For amusement, implement SADDO, SSUBO, UADDO, USUBO
|
commit | commitdiff | tree |
| 2008-12-10 |
Duncan Sands | Mention the Ada and Fortran front-ends.
|
commit | commitdiff | tree |
| 2008-12-10 |
Duncan Sands | Don't dereference the end() iterator. This was
|
commit | commitdiff | tree |
| 2008-12-09 |
Duncan Sands | Fix PR3117: not all nodes being legalized. The
|
commit | commitdiff | tree |
| next |
projects / oota-llvm.git / search